url_canon_unittest.cc revision 7d4cd473f85ac64c3747c96c277f9e506a0d2246
1c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// Copyright 2013 The Chromium Authors. All rights reserved. 2c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// Use of this source code is governed by a BSD-style license that can be 3c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// found in the LICENSE file. 4c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 5c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#include <errno.h> 6c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 7c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#include "testing/gtest/include/gtest/gtest.h" 8c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#include "third_party/icu/public/common/unicode/ucnv.h" 9c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#include "url/url_canon.h" 10c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#include "url/url_canon_icu.h" 11c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#include "url/url_canon_internal.h" 12c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#include "url/url_canon_stdstring.h" 13c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#include "url/url_parse.h" 14c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#include "url/url_test_utils.h" 15c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 16c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// Some implementations of base/basictypes.h may define ARRAYSIZE. 17c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// If it's not defined, we define it to the ARRAYSIZE_UNSAFE macro 18c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// which is in our version of basictypes.h. 19c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#ifndef ARRAYSIZE 20c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#define ARRAYSIZE ARRAYSIZE_UNSAFE 21c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#endif 22c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 23c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)using url_test_utils::WStringToUTF16; 24c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)using url_test_utils::ConvertUTF8ToUTF16; 25c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)using url_test_utils::ConvertUTF16ToUTF8; 26c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)using url_canon::CanonHostInfo; 27c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 28c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)namespace { 29c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 30c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)struct ComponentCase { 31c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* input; 32c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* expected; 33c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component expected_component; 34c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool expected_success; 35c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)}; 36c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 37c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// ComponentCase but with dual 8-bit/16-bit input. Generally, the unit tests 38c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// treat each input as optional, and will only try processing if non-NULL. 39c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// The output is always 8-bit. 40c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)struct DualComponentCase { 41c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* input8; 42c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const wchar_t* input16; 43c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* expected; 44c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component expected_component; 45c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool expected_success; 46c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)}; 47c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 48c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// Test cases for CanonicalizeIPAddress(). The inputs are identical to 49c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// DualComponentCase, but the output has extra CanonHostInfo fields. 50c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)struct IPAddressCase { 51c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* input8; 52c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const wchar_t* input16; 53c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* expected; 54c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component expected_component; 55c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 56c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // CanonHostInfo fields, for verbose output. 57c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) CanonHostInfo::Family expected_family; 58c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int expected_num_ipv4_components; 59c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* expected_address_hex; // Two hex chars per IP address byte. 60c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)}; 61c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 62c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)std::string BytesToHexString(unsigned char bytes[16], int length) { 63c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_TRUE(length == 0 || length == 4 || length == 16) 64c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) << "Bad IP address length: " << length; 65c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string result; 66c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (int i = 0; i < length; ++i) { 67c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) result.push_back(url_canon::kHexCharLookup[(bytes[i] >> 4) & 0xf]); 68c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) result.push_back(url_canon::kHexCharLookup[bytes[i] & 0xf]); 69c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 70c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return result; 71c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 72c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 73c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)struct ReplaceCase { 74c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* base; 75c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* scheme; 76c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* username; 77c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* password; 78c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* host; 79c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* port; 80c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* path; 81c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* query; 82c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* ref; 83c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* expected; 84c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)}; 85c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 86c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// Wrapper around a UConverter object that managers creation and destruction. 87c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)class UConvScoper { 88c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) public: 89c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) explicit UConvScoper(const char* charset_name) { 90c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) UErrorCode err = U_ZERO_ERROR; 91c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) converter_ = ucnv_open(charset_name, &err); 92c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 93c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 94c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) ~UConvScoper() { 95c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (converter_) 96c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) ucnv_close(converter_); 97c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 98c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 99c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Returns the converter object, may be NULL. 100c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) UConverter* converter() const { return converter_; } 101c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 102c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) private: 103c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) UConverter* converter_; 104c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)}; 105c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 106c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// Magic string used in the replacements code that tells SetupReplComp to 107c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// call the clear function. 108c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)const char kDeleteComp[] = "|"; 109c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 110c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// Sets up a replacement for a single component. This is given pointers to 111c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// the set and clear function for the component being replaced, and will 112c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// either set the component (if it exists) or clear it (if the replacement 113c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// string matches kDeleteComp). 114c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// 115c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// This template is currently used only for the 8-bit case, and the strlen 116c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// causes it to fail in other cases. It is left a template in case we have 117c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// tests for wide replacements. 118c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)template<typename CHAR> 119c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)void SetupReplComp( 120c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) void (url_canon::Replacements<CHAR>::*set)(const CHAR*, 121c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const url_parse::Component&), 122c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) void (url_canon::Replacements<CHAR>::*clear)(), 123c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::Replacements<CHAR>* rep, 124c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const CHAR* str) { 125c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (str && str[0] == kDeleteComp[0]) { 126c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) (rep->*clear)(); 127c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } else if (str) { 128c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) (rep->*set)(str, url_parse::Component(0, static_cast<int>(strlen(str)))); 129c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 130c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 131c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 132c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} // namespace 133c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 134c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)TEST(URLCanonTest, DoAppendUTF8) { 135c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) struct UTF8Case { 136c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) unsigned input; 137c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* output; 138c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } utf_cases[] = { 139c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Valid code points. 140c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {0x24, "\x24"}, 141c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {0xA2, "\xC2\xA2"}, 142c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {0x20AC, "\xE2\x82\xAC"}, 143c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {0x24B62, "\xF0\xA4\xAD\xA2"}, 144c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {0x10FFFF, "\xF4\x8F\xBF\xBF"}, 145c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) }; 146c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string out_str; 147c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (size_t i = 0; i < ARRAYSIZE(utf_cases); i++) { 148c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) out_str.clear(); 149c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output(&out_str); 150c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::AppendUTF8Value(utf_cases[i].input, &output); 151c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output.Complete(); 152c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(utf_cases[i].output, out_str); 153c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 154c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 155c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 156c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// TODO(mattm): Can't run this in debug mode for now, since the DCHECK will 157c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// cause the Chromium stacktrace dialog to appear and hang the test. 158c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// See http://crbug.com/49580. 159c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#if defined(GTEST_HAS_DEATH_TEST) && defined(NDEBUG) 160c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)TEST(URLCanonTest, DoAppendUTF8Invalid) { 161c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string out_str; 162c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output(&out_str); 163c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Invalid code point (too large). 164c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) ASSERT_DEBUG_DEATH({ 165c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::AppendUTF8Value(0x110000, &output); 166c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output.Complete(); 167c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ("", out_str); 168c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) }, ""); 169c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 170c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#endif 171c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 172c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)TEST(URLCanonTest, UTF) { 173c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Low-level test that we handle reading, canonicalization, and writing 174c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // UTF-8/UTF-16 strings properly. 175c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) struct UTFCase { 176c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* input8; 177c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const wchar_t* input16; 178c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool expected_success; 179c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* output; 180c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } utf_cases[] = { 181c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Valid canonical input should get passed through & escaped. 182c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"\xe4\xbd\xa0\xe5\xa5\xbd", L"\x4f60\x597d", true, "%E4%BD%A0%E5%A5%BD"}, 183c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Test a characer that takes > 16 bits (U+10300 = old italic letter A) 184c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"\xF0\x90\x8C\x80", L"\xd800\xdf00", true, "%F0%90%8C%80"}, 185c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Non-shortest-form UTF-8 are invalid. The bad char should be replaced 186c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // with the invalid character (EF BF DB in UTF-8). 187c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"\xf0\x84\xbd\xa0\xe5\xa5\xbd", NULL, false, "%EF%BF%BD%E5%A5%BD"}, 188c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Invalid UTF-8 sequences should be marked as invalid (the first 189c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // sequence is truncated). 190c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"\xe4\xa0\xe5\xa5\xbd", L"\xd800\x597d", false, "%EF%BF%BD%E5%A5%BD"}, 191c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Character going off the end. 192c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"\xe4\xbd\xa0\xe5\xa5", L"\x4f60\xd800", false, "%E4%BD%A0%EF%BF%BD"}, 193c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // ...same with low surrogates with no high surrogate. 194c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"\xed\xb0\x80", L"\xdc00", false, "%EF%BF%BD"}, 195c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Test a UTF-8 encoded surrogate value is marked as invalid. 196c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // ED A0 80 = U+D800 197c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"\xed\xa0\x80", NULL, false, "%EF%BF%BD"}, 198c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) }; 199c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 200c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string out_str; 201c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (size_t i = 0; i < ARRAYSIZE(utf_cases); i++) { 202c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (utf_cases[i].input8) { 203c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) out_str.clear(); 204c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output(&out_str); 205c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 206c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int input_len = static_cast<int>(strlen(utf_cases[i].input8)); 207c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool success = true; 208c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (int ch = 0; ch < input_len; ch++) { 209c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) success &= AppendUTF8EscapedChar(utf_cases[i].input8, &ch, input_len, 210c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) &output); 211c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 212c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output.Complete(); 213c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(utf_cases[i].expected_success, success); 214c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(std::string(utf_cases[i].output), out_str); 215c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 216c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (utf_cases[i].input16) { 217c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) out_str.clear(); 218c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output(&out_str); 219c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 2207d4cd473f85ac64c3747c96c277f9e506a0d2246Torne (Richard Coles) base::string16 input_str(WStringToUTF16(utf_cases[i].input16)); 221c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int input_len = static_cast<int>(input_str.length()); 222c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool success = true; 223c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (int ch = 0; ch < input_len; ch++) { 224c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) success &= AppendUTF8EscapedChar(input_str.c_str(), &ch, input_len, 225c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) &output); 226c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 227c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output.Complete(); 228c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(utf_cases[i].expected_success, success); 229c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(std::string(utf_cases[i].output), out_str); 230c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 231c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 232c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (utf_cases[i].input8 && utf_cases[i].input16 && 233c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) utf_cases[i].expected_success) { 234c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Check that the UTF-8 and UTF-16 inputs are equivalent. 235c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 236c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // UTF-16 -> UTF-8 237c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string input8_str(utf_cases[i].input8); 2387d4cd473f85ac64c3747c96c277f9e506a0d2246Torne (Richard Coles) base::string16 input16_str(WStringToUTF16(utf_cases[i].input16)); 239c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(input8_str, ConvertUTF16ToUTF8(input16_str)); 240c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 241c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // UTF-8 -> UTF-16 242c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(input16_str, ConvertUTF8ToUTF16(input8_str)); 243c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 244c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 245c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 246c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 247c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)TEST(URLCanonTest, ICUCharsetConverter) { 248c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) struct ICUCase { 249c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const wchar_t* input; 250c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* encoding; 251c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* expected; 252c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } icu_cases[] = { 253c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // UTF-8. 254c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {L"Hello, world", "utf-8", "Hello, world"}, 255c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {L"\x4f60\x597d", "utf-8", "\xe4\xbd\xa0\xe5\xa5\xbd"}, 256c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Non-BMP UTF-8. 257c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {L"!\xd800\xdf00!", "utf-8", "!\xf0\x90\x8c\x80!"}, 258c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Big5 259c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {L"\x4f60\x597d", "big5", "\xa7\x41\xa6\x6e"}, 260c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Unrepresentable character in the destination set. 261c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {L"hello\x4f60\x06de\x597dworld", "big5", "hello\xa7\x41%26%231758%3B\xa6\x6eworld"}, 262c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) }; 263c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 264c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (size_t i = 0; i < ARRAYSIZE(icu_cases); i++) { 265c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) UConvScoper conv(icu_cases[i].encoding); 266c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) ASSERT_TRUE(conv.converter() != NULL); 267c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::ICUCharsetConverter converter(conv.converter()); 268c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 269c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string str; 270c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output(&str); 271c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 2727d4cd473f85ac64c3747c96c277f9e506a0d2246Torne (Richard Coles) base::string16 input_str(WStringToUTF16(icu_cases[i].input)); 273c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int input_len = static_cast<int>(input_str.length()); 274c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) converter.ConvertFromUTF16(input_str.c_str(), input_len, &output); 275c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output.Complete(); 276c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 277c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_STREQ(icu_cases[i].expected, str.c_str()); 278c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 279c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 280c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Test string sizes around the resize boundary for the output to make sure 281c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // the converter resizes as needed. 282c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const int static_size = 16; 283c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) UConvScoper conv("utf-8"); 284c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) ASSERT_TRUE(conv.converter()); 285c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::ICUCharsetConverter converter(conv.converter()); 286c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (int i = static_size - 2; i <= static_size + 2; i++) { 287c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Make a string with the appropriate length. 2887d4cd473f85ac64c3747c96c277f9e506a0d2246Torne (Richard Coles) base::string16 input; 289c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (int ch = 0; ch < i; ch++) 290c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) input.push_back('a'); 291c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 292c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::RawCanonOutput<static_size> output; 293c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) converter.ConvertFromUTF16(input.c_str(), static_cast<int>(input.length()), 294c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) &output); 295c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(input.length(), static_cast<size_t>(output.length())); 296c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 297c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 298c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 299c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)TEST(URLCanonTest, Scheme) { 300c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Here, we're mostly testing that unusual characters are handled properly. 301c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // The canonicalizer doesn't do any parsing or whitespace detection. It will 302c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // also do its best on error, and will escape funny sequences (these won't be 303c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // valid schemes and it will return error). 304c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // 305c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Note that the canonicalizer will append a colon to the output to separate 306c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // out the rest of the URL, which is not present in the input. We check, 307c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // however, that the output range includes everything but the colon. 308c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) ComponentCase scheme_cases[] = { 309c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http", "http:", url_parse::Component(0, 4), true}, 310c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"HTTP", "http:", url_parse::Component(0, 4), true}, 311c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {" HTTP ", "%20http%20:", url_parse::Component(0, 10), false}, 312c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"htt: ", "htt%3A%20:", url_parse::Component(0, 9), false}, 313c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"\xe4\xbd\xa0\xe5\xa5\xbdhttp", "%E4%BD%A0%E5%A5%BDhttp:", url_parse::Component(0, 22), false}, 314c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Don't re-escape something already escaped. Note that it will 315c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // "canonicalize" the 'A' to 'a', but that's OK. 316c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"ht%3Atp", "ht%3atp:", url_parse::Component(0, 7), false}, 317c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) }; 318c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 319c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string out_str; 320c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 321c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (size_t i = 0; i < arraysize(scheme_cases); i++) { 322c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int url_len = static_cast<int>(strlen(scheme_cases[i].input)); 323c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component in_comp(0, url_len); 324c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component out_comp; 325c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 326c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) out_str.clear(); 327c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output1(&out_str); 328c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool success = url_canon::CanonicalizeScheme(scheme_cases[i].input, 329c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) in_comp, &output1, &out_comp); 330c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output1.Complete(); 331c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 332c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(scheme_cases[i].expected_success, success); 333c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(std::string(scheme_cases[i].expected), out_str); 334c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(scheme_cases[i].expected_component.begin, out_comp.begin); 335c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(scheme_cases[i].expected_component.len, out_comp.len); 336c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 337c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Now try the wide version 338c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) out_str.clear(); 339c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output2(&out_str); 340c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 3417d4cd473f85ac64c3747c96c277f9e506a0d2246Torne (Richard Coles) base::string16 wide_input(ConvertUTF8ToUTF16(scheme_cases[i].input)); 342c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) in_comp.len = static_cast<int>(wide_input.length()); 343c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) success = url_canon::CanonicalizeScheme(wide_input.c_str(), in_comp, 344c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) &output2, &out_comp); 345c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output2.Complete(); 346c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 347c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(scheme_cases[i].expected_success, success); 348c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(std::string(scheme_cases[i].expected), out_str); 349c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(scheme_cases[i].expected_component.begin, out_comp.begin); 350c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(scheme_cases[i].expected_component.len, out_comp.len); 351c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 352c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 353c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Test the case where the scheme is declared nonexistant, it should be 354c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // converted into an empty scheme. 355c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component out_comp; 356c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) out_str.clear(); 357c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output(&out_str); 358c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 359c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_TRUE(url_canon::CanonicalizeScheme("", url_parse::Component(0, -1), 360c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) &output, &out_comp)); 361c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output.Complete(); 362c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 363c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(std::string(":"), out_str); 364c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(0, out_comp.begin); 365c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(0, out_comp.len); 366c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 367c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 368c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)TEST(URLCanonTest, Host) { 369c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) IPAddressCase host_cases[] = { 370c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Basic canonicalization, uppercase should be converted to lowercase. 371c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"GoOgLe.CoM", L"GoOgLe.CoM", "google.com", url_parse::Component(0, 10), CanonHostInfo::NEUTRAL, -1, ""}, 372c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Spaces and some other characters should be escaped. 373c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"Goo%20 goo%7C|.com", L"Goo%20 goo%7C|.com", "goo%20%20goo%7C%7C.com", url_parse::Component(0, 22), CanonHostInfo::NEUTRAL, -1, ""}, 374c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Exciting different types of spaces! 375c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {NULL, L"GOO\x00a0\x3000goo.com", "goo%20%20goo.com", url_parse::Component(0, 16), CanonHostInfo::NEUTRAL, -1, ""}, 376c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Other types of space (no-break, zero-width, zero-width-no-break) are 377c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // name-prepped away to nothing. 378c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {NULL, L"GOO\x200b\x2060\xfeffgoo.com", "googoo.com", url_parse::Component(0, 10), CanonHostInfo::NEUTRAL, -1, ""}, 379c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Ideographic full stop (full-width period for Chinese, etc.) should be 380c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // treated as a dot. 381c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {NULL, L"www.foo\x3002"L"bar.com", "www.foo.bar.com", url_parse::Component(0, 15), CanonHostInfo::NEUTRAL, -1, ""}, 382c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Invalid unicode characters should fail... 383c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // ...In wide input, ICU will barf and we'll end up with the input as 384c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // escaped UTF-8 (the invalid character should be replaced with the 385c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // replacement character). 386c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"\xef\xb7\x90zyx.com", L"\xfdd0zyx.com", "%EF%BF%BDzyx.com", url_parse::Component(0, 16), CanonHostInfo::BROKEN, -1, ""}, 387c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // ...This is the same as previous but with with escaped. 388c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"%ef%b7%90zyx.com", L"%ef%b7%90zyx.com", "%EF%BF%BDzyx.com", url_parse::Component(0, 16), CanonHostInfo::BROKEN, -1, ""}, 389c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Test name prepping, fullwidth input should be converted to ASCII and NOT 390c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // IDN-ized. This is "Go" in fullwidth UTF-8/UTF-16. 391c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"\xef\xbc\xa7\xef\xbd\x8f.com", L"\xff27\xff4f.com", "go.com", url_parse::Component(0, 6), CanonHostInfo::NEUTRAL, -1, ""}, 392c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Test that fullwidth escaped values are properly name-prepped, 393c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // then converted or rejected. 394c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // ...%41 in fullwidth = 'A' (also as escaped UTF-8 input) 395c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"\xef\xbc\x85\xef\xbc\x94\xef\xbc\x91.com", L"\xff05\xff14\xff11.com", "a.com", url_parse::Component(0, 5), CanonHostInfo::NEUTRAL, -1, ""}, 396c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"%ef%bc%85%ef%bc%94%ef%bc%91.com", L"%ef%bc%85%ef%bc%94%ef%bc%91.com", "a.com", url_parse::Component(0, 5), CanonHostInfo::NEUTRAL, -1, ""}, 397c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // ...%00 in fullwidth should fail (also as escaped UTF-8 input) 398c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"\xef\xbc\x85\xef\xbc\x90\xef\xbc\x90.com", L"\xff05\xff10\xff10.com", "%00.com", url_parse::Component(0, 7), CanonHostInfo::BROKEN, -1, ""}, 399c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"%ef%bc%85%ef%bc%90%ef%bc%90.com", L"%ef%bc%85%ef%bc%90%ef%bc%90.com", "%00.com", url_parse::Component(0, 7), CanonHostInfo::BROKEN, -1, ""}, 400c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Basic IDN support, UTF-8 and UTF-16 input should be converted to IDN 401c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"\xe4\xbd\xa0\xe5\xa5\xbd\xe4\xbd\xa0\xe5\xa5\xbd", L"\x4f60\x597d\x4f60\x597d", "xn--6qqa088eba", url_parse::Component(0, 14), CanonHostInfo::NEUTRAL, -1, ""}, 402c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Mixed UTF-8 and escaped UTF-8 (narrow case) and UTF-16 and escaped 403c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // UTF-8 (wide case). The output should be equivalent to the true wide 404c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // character input above). 405c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"%E4%BD%A0%E5%A5%BD\xe4\xbd\xa0\xe5\xa5\xbd", L"%E4%BD%A0%E5%A5%BD\x4f60\x597d", "xn--6qqa088eba", url_parse::Component(0, 14), CanonHostInfo::NEUTRAL, -1, ""}, 406c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Invalid escaped characters should fail and the percents should be 407c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // escaped. 408c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"%zz%66%a", L"%zz%66%a", "%25zzf%25a", url_parse::Component(0, 10), CanonHostInfo::BROKEN, -1, ""}, 409c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // If we get an invalid character that has been escaped. 410c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"%25", L"%25", "%25", url_parse::Component(0, 3), CanonHostInfo::BROKEN, -1, ""}, 411c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"hello%00", L"hello%00", "hello%00", url_parse::Component(0, 8), CanonHostInfo::BROKEN, -1, ""}, 412c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Escaped numbers should be treated like IP addresses if they are. 413c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"%30%78%63%30%2e%30%32%35%30.01", L"%30%78%63%30%2e%30%32%35%30.01", "192.168.0.1", url_parse::Component(0, 11), CanonHostInfo::IPV4, 3, "C0A80001"}, 414c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"%30%78%63%30%2e%30%32%35%30.01%2e", L"%30%78%63%30%2e%30%32%35%30.01%2e", "192.168.0.1", url_parse::Component(0, 11), CanonHostInfo::IPV4, 3, "C0A80001"}, 415c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Invalid escaping should trigger the regular host error handling. 416c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"%3g%78%63%30%2e%30%32%35%30%2E.01", L"%3g%78%63%30%2e%30%32%35%30%2E.01", "%253gxc0.0250..01", url_parse::Component(0, 17), CanonHostInfo::BROKEN, -1, ""}, 417c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Something that isn't exactly an IP should get treated as a host and 418c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // spaces escaped. 419c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"192.168.0.1 hello", L"192.168.0.1 hello", "192.168.0.1%20hello", url_parse::Component(0, 19), CanonHostInfo::NEUTRAL, -1, ""}, 420c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Fullwidth and escaped UTF-8 fullwidth should still be treated as IP. 421c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // These are "0Xc0.0250.01" in fullwidth. 422c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"\xef\xbc\x90%Ef%bc\xb8%ef%Bd%83\xef\xbc\x90%EF%BC%8E\xef\xbc\x90\xef\xbc\x92\xef\xbc\x95\xef\xbc\x90\xef\xbc%8E\xef\xbc\x90\xef\xbc\x91", L"\xff10\xff38\xff43\xff10\xff0e\xff10\xff12\xff15\xff10\xff0e\xff10\xff11", "192.168.0.1", url_parse::Component(0, 11), CanonHostInfo::IPV4, 3, "C0A80001"}, 423c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Broken IP addresses get marked as such. 424c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"192.168.0.257", L"192.168.0.257", "192.168.0.257", url_parse::Component(0, 13), CanonHostInfo::BROKEN, -1, ""}, 425c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[google.com]", L"[google.com]", "[google.com]", url_parse::Component(0, 12), CanonHostInfo::BROKEN, -1, ""}, 426c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Cyrillic letter followed buy ( should return punicode for ( escaped before punicode string was created. I.e. 427c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // if ( is escaped after punicode is created we would get xn--%28-8tb (incorrect). 428c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"\xd1\x82(", L"\x0442(", "xn--%28-7ed", url_parse::Component(0, 11), CanonHostInfo::NEUTRAL, -1, ""}, 429c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Address with all hexidecimal characters with leading number of 1<<32 430c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // or greater and should return NEUTRAL rather than BROKEN if not all 431c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // components are numbers. 432c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"12345678912345.de", L"12345678912345.de", "12345678912345.de", url_parse::Component(0, 17), CanonHostInfo::NEUTRAL, -1, ""}, 433c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"1.12345678912345.de", L"1.12345678912345.de", "1.12345678912345.de", url_parse::Component(0, 19), CanonHostInfo::NEUTRAL, -1, ""}, 434c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"12345678912345.12345678912345.de", L"12345678912345.12345678912345.de", "12345678912345.12345678912345.de", url_parse::Component(0, 32), CanonHostInfo::NEUTRAL, -1, ""}, 435c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"1.2.0xB3A73CE5B59.de", L"1.2.0xB3A73CE5B59.de", "1.2.0xb3a73ce5b59.de", url_parse::Component(0, 20), CanonHostInfo::NEUTRAL, -1, ""}, 436c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"12345678912345.0xde", L"12345678912345.0xde", "12345678912345.0xde", url_parse::Component(0, 19), CanonHostInfo::BROKEN, -1, ""}, 437c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) }; 438c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 439c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // CanonicalizeHost() non-verbose. 440c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string out_str; 441c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (size_t i = 0; i < arraysize(host_cases); i++) { 442c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Narrow version. 443c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (host_cases[i].input8) { 444c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int host_len = static_cast<int>(strlen(host_cases[i].input8)); 445c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component in_comp(0, host_len); 446c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component out_comp; 447c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 448c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) out_str.clear(); 449c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output(&out_str); 450c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 451c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool success = url_canon::CanonicalizeHost(host_cases[i].input8, in_comp, 452c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) &output, &out_comp); 453c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output.Complete(); 454c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 455c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(host_cases[i].expected_family != CanonHostInfo::BROKEN, 456c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) success); 457c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(std::string(host_cases[i].expected), out_str); 458c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(host_cases[i].expected_component.begin, out_comp.begin); 459c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(host_cases[i].expected_component.len, out_comp.len); 460c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 461c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 462c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Wide version. 463c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (host_cases[i].input16) { 4647d4cd473f85ac64c3747c96c277f9e506a0d2246Torne (Richard Coles) base::string16 input16(WStringToUTF16(host_cases[i].input16)); 465c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int host_len = static_cast<int>(input16.length()); 466c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component in_comp(0, host_len); 467c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component out_comp; 468c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 469c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) out_str.clear(); 470c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output(&out_str); 471c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 472c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool success = url_canon::CanonicalizeHost(input16.c_str(), in_comp, 473c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) &output, &out_comp); 474c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output.Complete(); 475c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 476c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(host_cases[i].expected_family != CanonHostInfo::BROKEN, 477c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) success); 478c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(std::string(host_cases[i].expected), out_str); 479c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(host_cases[i].expected_component.begin, out_comp.begin); 480c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(host_cases[i].expected_component.len, out_comp.len); 481c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 482c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 483c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 484c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // CanonicalizeHostVerbose() 485c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (size_t i = 0; i < arraysize(host_cases); i++) { 486c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Narrow version. 487c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (host_cases[i].input8) { 488c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int host_len = static_cast<int>(strlen(host_cases[i].input8)); 489c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component in_comp(0, host_len); 490c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 491c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) out_str.clear(); 492c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output(&out_str); 493c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) CanonHostInfo host_info; 494c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 495c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::CanonicalizeHostVerbose(host_cases[i].input8, in_comp, 496c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) &output, &host_info); 497c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output.Complete(); 498c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 499c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(host_cases[i].expected_family, host_info.family); 500c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(std::string(host_cases[i].expected), out_str); 501c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(host_cases[i].expected_component.begin, 502c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) host_info.out_host.begin); 503c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(host_cases[i].expected_component.len, host_info.out_host.len); 504c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(std::string(host_cases[i].expected_address_hex), 505c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) BytesToHexString(host_info.address, host_info.AddressLength())); 506c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (host_cases[i].expected_family == CanonHostInfo::IPV4) { 507c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(host_cases[i].expected_num_ipv4_components, 508c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) host_info.num_ipv4_components); 509c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 510c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 511c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 512c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Wide version. 513c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (host_cases[i].input16) { 5147d4cd473f85ac64c3747c96c277f9e506a0d2246Torne (Richard Coles) base::string16 input16(WStringToUTF16(host_cases[i].input16)); 515c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int host_len = static_cast<int>(input16.length()); 516c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component in_comp(0, host_len); 517c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 518c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) out_str.clear(); 519c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output(&out_str); 520c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) CanonHostInfo host_info; 521c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 522c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::CanonicalizeHostVerbose(input16.c_str(), in_comp, 523c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) &output, &host_info); 524c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output.Complete(); 525c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 526c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(host_cases[i].expected_family, host_info.family); 527c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(std::string(host_cases[i].expected), out_str); 528c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(host_cases[i].expected_component.begin, 529c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) host_info.out_host.begin); 530c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(host_cases[i].expected_component.len, host_info.out_host.len); 531c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(std::string(host_cases[i].expected_address_hex), 532c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) BytesToHexString(host_info.address, host_info.AddressLength())); 533c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (host_cases[i].expected_family == CanonHostInfo::IPV4) { 534c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(host_cases[i].expected_num_ipv4_components, 535c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) host_info.num_ipv4_components); 536c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 537c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 538c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 539c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 540c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 541c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)TEST(URLCanonTest, IPv4) { 542c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) IPAddressCase cases[] = { 543c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Empty is not an IP address. 544c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"", L"", "", url_parse::Component(), CanonHostInfo::NEUTRAL, -1, ""}, 545c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {".", L".", "", url_parse::Component(), CanonHostInfo::NEUTRAL, -1, ""}, 546c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Regular IP addresses in different bases. 547c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"192.168.0.1", L"192.168.0.1", "192.168.0.1", url_parse::Component(0, 11), CanonHostInfo::IPV4, 4, "C0A80001"}, 548c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"0300.0250.00.01", L"0300.0250.00.01", "192.168.0.1", url_parse::Component(0, 11), CanonHostInfo::IPV4, 4, "C0A80001"}, 549c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"0xC0.0Xa8.0x0.0x1", L"0xC0.0Xa8.0x0.0x1", "192.168.0.1", url_parse::Component(0, 11), CanonHostInfo::IPV4, 4, "C0A80001"}, 550c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Non-IP addresses due to invalid characters. 551c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"192.168.9.com", L"192.168.9.com", "", url_parse::Component(), CanonHostInfo::NEUTRAL, -1, ""}, 552c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Invalid characters for the base should be rejected. 553c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"19a.168.0.1", L"19a.168.0.1", "", url_parse::Component(), CanonHostInfo::NEUTRAL, -1, ""}, 554c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"0308.0250.00.01", L"0308.0250.00.01", "", url_parse::Component(), CanonHostInfo::NEUTRAL, -1, ""}, 555c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"0xCG.0xA8.0x0.0x1", L"0xCG.0xA8.0x0.0x1", "", url_parse::Component(), CanonHostInfo::NEUTRAL, -1, ""}, 556c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // If there are not enough components, the last one should fill them out. 557c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"192", L"192", "0.0.0.192", url_parse::Component(0, 9), CanonHostInfo::IPV4, 1, "000000C0"}, 558c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"0xC0a80001", L"0xC0a80001", "192.168.0.1", url_parse::Component(0, 11), CanonHostInfo::IPV4, 1, "C0A80001"}, 559c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"030052000001", L"030052000001", "192.168.0.1", url_parse::Component(0, 11), CanonHostInfo::IPV4, 1, "C0A80001"}, 560c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"000030052000001", L"000030052000001", "192.168.0.1", url_parse::Component(0, 11), CanonHostInfo::IPV4, 1, "C0A80001"}, 561c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"192.168", L"192.168", "192.0.0.168", url_parse::Component(0, 11), CanonHostInfo::IPV4, 2, "C00000A8"}, 562c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"192.0x00A80001", L"192.0x000A80001", "192.168.0.1", url_parse::Component(0, 11), CanonHostInfo::IPV4, 2, "C0A80001"}, 563c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"0xc0.052000001", L"0xc0.052000001", "192.168.0.1", url_parse::Component(0, 11), CanonHostInfo::IPV4, 2, "C0A80001"}, 564c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"192.168.1", L"192.168.1", "192.168.0.1", url_parse::Component(0, 11), CanonHostInfo::IPV4, 3, "C0A80001"}, 565c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Too many components means not an IP address. 566c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"192.168.0.0.1", L"192.168.0.0.1", "", url_parse::Component(), CanonHostInfo::NEUTRAL, -1, ""}, 567c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // We allow a single trailing dot. 568c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"192.168.0.1.", L"192.168.0.1.", "192.168.0.1", url_parse::Component(0, 11), CanonHostInfo::IPV4, 4, "C0A80001"}, 569c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"192.168.0.1. hello", L"192.168.0.1. hello", "", url_parse::Component(), CanonHostInfo::NEUTRAL, -1, ""}, 570c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"192.168.0.1..", L"192.168.0.1..", "", url_parse::Component(), CanonHostInfo::NEUTRAL, -1, ""}, 571c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Two dots in a row means not an IP address. 572c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"192.168..1", L"192.168..1", "", url_parse::Component(), CanonHostInfo::NEUTRAL, -1, ""}, 573c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Any numerical overflow should be marked as BROKEN. 574c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"0x100.0", L"0x100.0", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 575c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"0x100.0.0", L"0x100.0.0", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 576c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"0x100.0.0.0", L"0x100.0.0.0", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 577c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"0.0x100.0.0", L"0.0x100.0.0", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 578c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"0.0.0x100.0", L"0.0.0x100.0", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 579c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"0.0.0.0x100", L"0.0.0.0x100", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 580c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"0.0.0x10000", L"0.0.0x10000", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 581c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"0.0x1000000", L"0.0x1000000", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 582c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"0x100000000", L"0x100000000", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 583c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Repeat the previous tests, minus 1, to verify boundaries. 584c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"0xFF.0", L"0xFF.0", "255.0.0.0", url_parse::Component(0, 9), CanonHostInfo::IPV4, 2, "FF000000"}, 585c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"0xFF.0.0", L"0xFF.0.0", "255.0.0.0", url_parse::Component(0, 9), CanonHostInfo::IPV4, 3, "FF000000"}, 586c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"0xFF.0.0.0", L"0xFF.0.0.0", "255.0.0.0", url_parse::Component(0, 9), CanonHostInfo::IPV4, 4, "FF000000"}, 587c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"0.0xFF.0.0", L"0.0xFF.0.0", "0.255.0.0", url_parse::Component(0, 9), CanonHostInfo::IPV4, 4, "00FF0000"}, 588c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"0.0.0xFF.0", L"0.0.0xFF.0", "0.0.255.0", url_parse::Component(0, 9), CanonHostInfo::IPV4, 4, "0000FF00"}, 589c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"0.0.0.0xFF", L"0.0.0.0xFF", "0.0.0.255", url_parse::Component(0, 9), CanonHostInfo::IPV4, 4, "000000FF"}, 590c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"0.0.0xFFFF", L"0.0.0xFFFF", "0.0.255.255", url_parse::Component(0, 11), CanonHostInfo::IPV4, 3, "0000FFFF"}, 591c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"0.0xFFFFFF", L"0.0xFFFFFF", "0.255.255.255", url_parse::Component(0, 13), CanonHostInfo::IPV4, 2, "00FFFFFF"}, 592c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"0xFFFFFFFF", L"0xFFFFFFFF", "255.255.255.255", url_parse::Component(0, 15), CanonHostInfo::IPV4, 1, "FFFFFFFF"}, 593c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Old trunctations tests. They're all "BROKEN" now. 594c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"276.256.0xf1a2.077777", L"276.256.0xf1a2.077777", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 595c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"192.168.0.257", L"192.168.0.257", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 596c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"192.168.0xa20001", L"192.168.0xa20001", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 597c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"192.015052000001", L"192.015052000001", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 598c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"0X12C0a80001", L"0X12C0a80001", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 599c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"276.1.2", L"276.1.2", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 600c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Spaces should be rejected. 601c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"192.168.0.1 hello", L"192.168.0.1 hello", "", url_parse::Component(), CanonHostInfo::NEUTRAL, -1, ""}, 602c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Very large numbers. 603c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"0000000000000300.0x00000000000000fF.00000000000000001", L"0000000000000300.0x00000000000000fF.00000000000000001", "192.255.0.1", url_parse::Component(0, 11), CanonHostInfo::IPV4, 3, "C0FF0001"}, 604c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"0000000000000300.0xffffffffFFFFFFFF.3022415481470977", L"0000000000000300.0xffffffffFFFFFFFF.3022415481470977", "", url_parse::Component(0, 11), CanonHostInfo::BROKEN, -1, ""}, 605c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // A number has no length limit, but long numbers can still overflow. 606c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"00000000000000000001", L"00000000000000000001", "0.0.0.1", url_parse::Component(0, 7), CanonHostInfo::IPV4, 1, "00000001"}, 607c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"0000000000000000100000000000000001", L"0000000000000000100000000000000001", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 608c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // If a long component is non-numeric, it's a hostname, *not* a broken IP. 609c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"0.0.0.000000000000000000z", L"0.0.0.000000000000000000z", "", url_parse::Component(), CanonHostInfo::NEUTRAL, -1, ""}, 610c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"0.0.0.100000000000000000z", L"0.0.0.100000000000000000z", "", url_parse::Component(), CanonHostInfo::NEUTRAL, -1, ""}, 611c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Truncation of all zeros should still result in 0. 612c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"0.00.0x.0x0", L"0.00.0x.0x0", "0.0.0.0", url_parse::Component(0, 7), CanonHostInfo::IPV4, 4, "00000000"}, 613c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) }; 614c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 615c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (size_t i = 0; i < arraysize(cases); i++) { 616c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // 8-bit version. 617c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component component(0, 618c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) static_cast<int>(strlen(cases[i].input8))); 619c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 620c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string out_str1; 621c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output1(&out_str1); 622c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::CanonHostInfo host_info; 623c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::CanonicalizeIPAddress(cases[i].input8, component, &output1, 624c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) &host_info); 625c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output1.Complete(); 626c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 627c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(cases[i].expected_family, host_info.family); 628c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(std::string(cases[i].expected_address_hex), 629c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) BytesToHexString(host_info.address, host_info.AddressLength())); 630c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (host_info.family == CanonHostInfo::IPV4) { 631c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_STREQ(cases[i].expected, out_str1.c_str()); 632c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(cases[i].expected_component.begin, host_info.out_host.begin); 633c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(cases[i].expected_component.len, host_info.out_host.len); 634c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(cases[i].expected_num_ipv4_components, 635c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) host_info.num_ipv4_components); 636c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 637c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 638c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // 16-bit version. 6397d4cd473f85ac64c3747c96c277f9e506a0d2246Torne (Richard Coles) base::string16 input16(WStringToUTF16(cases[i].input16)); 640c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) component = url_parse::Component(0, static_cast<int>(input16.length())); 641c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 642c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string out_str2; 643c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output2(&out_str2); 644c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::CanonicalizeIPAddress(input16.c_str(), component, &output2, 645c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) &host_info); 646c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output2.Complete(); 647c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 648c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(cases[i].expected_family, host_info.family); 649c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(std::string(cases[i].expected_address_hex), 650c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) BytesToHexString(host_info.address, host_info.AddressLength())); 651c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (host_info.family == CanonHostInfo::IPV4) { 652c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_STREQ(cases[i].expected, out_str2.c_str()); 653c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(cases[i].expected_component.begin, host_info.out_host.begin); 654c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(cases[i].expected_component.len, host_info.out_host.len); 655c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(cases[i].expected_num_ipv4_components, 656c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) host_info.num_ipv4_components); 657c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 658c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 659c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 660c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 661c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)TEST(URLCanonTest, IPv6) { 662c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) IPAddressCase cases[] = { 663c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Empty is not an IP address. 664c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"", L"", "", url_parse::Component(), CanonHostInfo::NEUTRAL, -1, ""}, 665c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Non-IPs with [:] characters are marked BROKEN. 666c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {":", L":", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 667c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[", L"[", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 668c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[:", L"[:", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 669c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"]", L"]", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 670c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {":]", L":]", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 671c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[]", L"[]", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 672c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[:]", L"[:]", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 673c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Regular IP address is invalid without bounding '[' and ']'. 674c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"2001:db8::1", L"2001:db8::1", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 675c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[2001:db8::1", L"[2001:db8::1", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 676c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"2001:db8::1]", L"2001:db8::1]", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 677c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Regular IP addresses. 678c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[::]", L"[::]", "[::]", url_parse::Component(0,4), CanonHostInfo::IPV6, -1, "00000000000000000000000000000000"}, 679c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[::1]", L"[::1]", "[::1]", url_parse::Component(0,5), CanonHostInfo::IPV6, -1, "00000000000000000000000000000001"}, 680c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[1::]", L"[1::]", "[1::]", url_parse::Component(0,5), CanonHostInfo::IPV6, -1, "00010000000000000000000000000000"}, 681c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 682c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Leading zeros should be stripped. 683c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[000:01:02:003:004:5:6:007]", L"[000:01:02:003:004:5:6:007]", "[0:1:2:3:4:5:6:7]", url_parse::Component(0,17), CanonHostInfo::IPV6, -1, "00000001000200030004000500060007"}, 684c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 685c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Upper case letters should be lowercased. 686c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[A:b:c:DE:fF:0:1:aC]", L"[A:b:c:DE:fF:0:1:aC]", "[a:b:c:de:ff:0:1:ac]", url_parse::Component(0,20), CanonHostInfo::IPV6, -1, "000A000B000C00DE00FF0000000100AC"}, 687c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 688c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // The same address can be written with different contractions, but should 689c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // get canonicalized to the same thing. 690c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[1:0:0:2::3:0]", L"[1:0:0:2::3:0]", "[1::2:0:0:3:0]", url_parse::Component(0,14), CanonHostInfo::IPV6, -1, "00010000000000020000000000030000"}, 691c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[1::2:0:0:3:0]", L"[1::2:0:0:3:0]", "[1::2:0:0:3:0]", url_parse::Component(0,14), CanonHostInfo::IPV6, -1, "00010000000000020000000000030000"}, 692c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 693c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Addresses with embedded IPv4. 694c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[::192.168.0.1]", L"[::192.168.0.1]", "[::c0a8:1]", url_parse::Component(0,10), CanonHostInfo::IPV6, -1, "000000000000000000000000C0A80001"}, 695c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[::ffff:192.168.0.1]", L"[::ffff:192.168.0.1]", "[::ffff:c0a8:1]", url_parse::Component(0,15), CanonHostInfo::IPV6, -1, "00000000000000000000FFFFC0A80001"}, 696c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[::eeee:192.168.0.1]", L"[::eeee:192.168.0.1]", "[::eeee:c0a8:1]", url_parse::Component(0, 15), CanonHostInfo::IPV6, -1, "00000000000000000000EEEEC0A80001"}, 697c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[2001::192.168.0.1]", L"[2001::192.168.0.1]", "[2001::c0a8:1]", url_parse::Component(0, 14), CanonHostInfo::IPV6, -1, "200100000000000000000000C0A80001"}, 698c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[1:2:192.168.0.1:5:6]", L"[1:2:192.168.0.1:5:6]", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 699c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 700c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // IPv4 with last component missing. 701c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[::ffff:192.1.2]", L"[::ffff:192.1.2]", "[::ffff:c001:2]", url_parse::Component(0,15), CanonHostInfo::IPV6, -1, "00000000000000000000FFFFC0010002"}, 702c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 703c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // IPv4 using hex. 704c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // TODO(eroman): Should this format be disallowed? 705c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[::ffff:0xC0.0Xa8.0x0.0x1]", L"[::ffff:0xC0.0Xa8.0x0.0x1]", "[::ffff:c0a8:1]", url_parse::Component(0,15), CanonHostInfo::IPV6, -1, "00000000000000000000FFFFC0A80001"}, 706c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 707c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // There may be zeros surrounding the "::" contraction. 708c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[0:0::0:0:8]", L"[0:0::0:0:8]", "[::8]", url_parse::Component(0,5), CanonHostInfo::IPV6, -1, "00000000000000000000000000000008"}, 709c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 710c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[2001:db8::1]", L"[2001:db8::1]", "[2001:db8::1]", url_parse::Component(0,13), CanonHostInfo::IPV6, -1, "20010DB8000000000000000000000001"}, 711c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 712c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Can only have one "::" contraction in an IPv6 string literal. 713c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[2001::db8::1]", L"[2001::db8::1]", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 714c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // No more than 2 consecutive ':'s. 715c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[2001:db8:::1]", L"[2001:db8:::1]", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 716c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[:::]", L"[:::]", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 717c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Non-IP addresses due to invalid characters. 718c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[2001::.com]", L"[2001::.com]", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 719c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // If there are not enough components, the last one should fill them out. 720c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // ... omitted at this time ... 721c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Too many components means not an IP address. Similarly with too few if using IPv4 compat or mapped addresses. 722c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[::192.168.0.0.1]", L"[::192.168.0.0.1]", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 723c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[::ffff:192.168.0.0.1]", L"[::ffff:192.168.0.0.1]", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 724c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[1:2:3:4:5:6:7:8:9]", L"[1:2:3:4:5:6:7:8:9]", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 725c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Too many bits (even though 8 comonents, the last one holds 32 bits). 726c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[0:0:0:0:0:0:0:192.168.0.1]", L"[0:0:0:0:0:0:0:192.168.0.1]", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 727c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 728c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Too many bits specified -- the contraction would have to be zero-length 729c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // to not exceed 128 bits. 730c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[1:2:3:4:5:6::192.168.0.1]", L"[1:2:3:4:5:6::192.168.0.1]", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 731c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 732c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // The contraction is for 16 bits of zero. 733c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[1:2:3:4:5:6::8]", L"[1:2:3:4:5:6::8]", "[1:2:3:4:5:6:0:8]", url_parse::Component(0,17), CanonHostInfo::IPV6, -1, "00010002000300040005000600000008"}, 734c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 735c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Cannot have a trailing colon. 736c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[1:2:3:4:5:6:7:8:]", L"[1:2:3:4:5:6:7:8:]", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 737c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[1:2:3:4:5:6:192.168.0.1:]", L"[1:2:3:4:5:6:192.168.0.1:]", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 738c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 739c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Cannot have negative numbers. 740c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[-1:2:3:4:5:6:7:8]", L"[-1:2:3:4:5:6:7:8]", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 741c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 742c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Scope ID -- the URL may contain an optional ["%" <scope_id>] section. 743c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // The scope_id should be included in the canonicalized URL, and is an 744c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // unsigned decimal number. 745c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 746c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Invalid because no ID was given after the percent. 747c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 748c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Don't allow scope-id 749c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[1::%1]", L"[1::%1]", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 750c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[1::%eth0]", L"[1::%eth0]", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 751c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[1::%]", L"[1::%]", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 752c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[%]", L"[%]", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 753c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[::%:]", L"[::%:]", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 754c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 755c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Don't allow leading or trailing colons. 756c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[:0:0::0:0:8]", L"[:0:0::0:0:8]", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 757c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[0:0::0:0:8:]", L"[0:0::0:0:8:]", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 758c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[:0:0::0:0:8:]", L"[:0:0::0:0:8:]", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 759c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 760c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // We allow a single trailing dot. 761c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // ... omitted at this time ... 762c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Two dots in a row means not an IP address. 763c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[::192.168..1]", L"[::192.168..1]", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 764c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Any non-first components get truncated to one byte. 765c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // ... omitted at this time ... 766c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Spaces should be rejected. 767c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"[::1 hello]", L"[::1 hello]", "", url_parse::Component(), CanonHostInfo::BROKEN, -1, ""}, 768c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) }; 769c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 770c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (size_t i = 0; i < arraysize(cases); i++) { 771c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // 8-bit version. 772c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component component(0, 773c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) static_cast<int>(strlen(cases[i].input8))); 774c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 775c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string out_str1; 776c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output1(&out_str1); 777c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::CanonHostInfo host_info; 778c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::CanonicalizeIPAddress(cases[i].input8, component, &output1, 779c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) &host_info); 780c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output1.Complete(); 781c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 782c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(cases[i].expected_family, host_info.family); 783c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(std::string(cases[i].expected_address_hex), 784c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) BytesToHexString(host_info.address, host_info.AddressLength())) << "iter " << i << " host " << cases[i].input8; 785c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (host_info.family == CanonHostInfo::IPV6) { 786c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_STREQ(cases[i].expected, out_str1.c_str()); 787c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(cases[i].expected_component.begin, 788c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) host_info.out_host.begin); 789c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(cases[i].expected_component.len, host_info.out_host.len); 790c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 791c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 792c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // 16-bit version. 7937d4cd473f85ac64c3747c96c277f9e506a0d2246Torne (Richard Coles) base::string16 input16(WStringToUTF16(cases[i].input16)); 794c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) component = url_parse::Component(0, static_cast<int>(input16.length())); 795c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 796c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string out_str2; 797c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output2(&out_str2); 798c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::CanonicalizeIPAddress(input16.c_str(), component, &output2, 799c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) &host_info); 800c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output2.Complete(); 801c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 802c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(cases[i].expected_family, host_info.family); 803c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(std::string(cases[i].expected_address_hex), 804c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) BytesToHexString(host_info.address, host_info.AddressLength())); 805c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (host_info.family == CanonHostInfo::IPV6) { 806c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_STREQ(cases[i].expected, out_str2.c_str()); 807c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(cases[i].expected_component.begin, host_info.out_host.begin); 808c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(cases[i].expected_component.len, host_info.out_host.len); 809c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 810c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 811c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 812c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 813c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)TEST(URLCanonTest, IPEmpty) { 814c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string out_str1; 815c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output1(&out_str1); 816c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::CanonHostInfo host_info; 817c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 818c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // This tests tests. 819c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char spec[] = "192.168.0.1"; 820c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::CanonicalizeIPAddress(spec, url_parse::Component(), 821c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) &output1, &host_info); 822c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_FALSE(host_info.IsIPAddress()); 823c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 824c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::CanonicalizeIPAddress(spec, url_parse::Component(0, 0), 825c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) &output1, &host_info); 826c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_FALSE(host_info.IsIPAddress()); 827c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 828c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 829c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)TEST(URLCanonTest, UserInfo) { 830c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Note that the canonicalizer should escape and treat empty components as 831c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // not being there. 832c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 833c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // We actually parse a full input URL so we can get the initial components. 834c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) struct UserComponentCase { 835c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* input; 836c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* expected; 837c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component expected_username; 838c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component expected_password; 839c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool expected_success; 840c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } user_info_cases[] = { 841c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://user:pass@host.com/", "user:pass@", url_parse::Component(0, 4), url_parse::Component(5, 4), true}, 842c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://@host.com/", "", url_parse::Component(0, -1), url_parse::Component(0, -1), true}, 843c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://:@host.com/", "", url_parse::Component(0, -1), url_parse::Component(0, -1), true}, 844c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://foo:@host.com/", "foo@", url_parse::Component(0, 3), url_parse::Component(0, -1), true}, 845c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://:foo@host.com/", ":foo@", url_parse::Component(0, 0), url_parse::Component(1, 3), true}, 846c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://^ :$\t@host.com/", "%5E%20:$%09@", url_parse::Component(0, 6), url_parse::Component(7, 4), true}, 847c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://user:pass@/", "user:pass@", url_parse::Component(0, 4), url_parse::Component(5, 4), true}, 848c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://%2540:bar@domain.com/", "%2540:bar@", url_parse::Component(0, 5), url_parse::Component(6, 3), true }, 849c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 850c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // IE7 compatability: old versions allowed backslashes in usernames, but 851c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // IE7 does not. We disallow it as well. 852c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"ftp://me\\mydomain:pass@foo.com/", "", url_parse::Component(0, -1), url_parse::Component(0, -1), true}, 853c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) }; 854c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 855c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (size_t i = 0; i < ARRAYSIZE(user_info_cases); i++) { 856c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int url_len = static_cast<int>(strlen(user_info_cases[i].input)); 857c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Parsed parsed; 858c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::ParseStandardURL(user_info_cases[i].input, url_len, &parsed); 859c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component out_user, out_pass; 860c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string out_str; 861c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output1(&out_str); 862c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 863c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool success = url_canon::CanonicalizeUserInfo(user_info_cases[i].input, 864c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) parsed.username, 865c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) user_info_cases[i].input, 866c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) parsed.password, 867c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) &output1, &out_user, 868c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) &out_pass); 869c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output1.Complete(); 870c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 871c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(user_info_cases[i].expected_success, success); 872c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(std::string(user_info_cases[i].expected), out_str); 873c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(user_info_cases[i].expected_username.begin, out_user.begin); 874c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(user_info_cases[i].expected_username.len, out_user.len); 875c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(user_info_cases[i].expected_password.begin, out_pass.begin); 876c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(user_info_cases[i].expected_password.len, out_pass.len); 877c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 878c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Now try the wide version 879c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) out_str.clear(); 880c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output2(&out_str); 8817d4cd473f85ac64c3747c96c277f9e506a0d2246Torne (Richard Coles) base::string16 wide_input(ConvertUTF8ToUTF16(user_info_cases[i].input)); 882c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) success = url_canon::CanonicalizeUserInfo(wide_input.c_str(), 883c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) parsed.username, 884c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) wide_input.c_str(), 885c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) parsed.password, 886c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) &output2, &out_user, &out_pass); 887c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output2.Complete(); 888c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 889c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(user_info_cases[i].expected_success, success); 890c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(std::string(user_info_cases[i].expected), out_str); 891c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(user_info_cases[i].expected_username.begin, out_user.begin); 892c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(user_info_cases[i].expected_username.len, out_user.len); 893c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(user_info_cases[i].expected_password.begin, out_pass.begin); 894c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(user_info_cases[i].expected_password.len, out_pass.len); 895c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 896c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 897c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 898c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)TEST(URLCanonTest, Port) { 899c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // We only need to test that the number gets properly put into the output 900c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // buffer. The parser unit tests will test scanning the number correctly. 901c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // 902c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Note that the CanonicalizePort will always prepend a colon to the output 903c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // to separate it from the colon that it assumes preceeds it. 904c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) struct PortCase { 905c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* input; 906c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int default_port; 907c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* expected; 908c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component expected_component; 909c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool expected_success; 910c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } port_cases[] = { 911c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Invalid input should be copied w/ failure. 912c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"as df", 80, ":as%20df", url_parse::Component(1, 7), false}, 913c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"-2", 80, ":-2", url_parse::Component(1, 2), false}, 914c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Default port should be omitted. 915c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"80", 80, "", url_parse::Component(0, -1), true}, 916c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"8080", 80, ":8080", url_parse::Component(1, 4), true}, 917c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // PORT_UNSPECIFIED should mean always keep the port. 918c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"80", url_parse::PORT_UNSPECIFIED, ":80", url_parse::Component(1, 2), true}, 919c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) }; 920c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 921c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (size_t i = 0; i < ARRAYSIZE(port_cases); i++) { 922c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int url_len = static_cast<int>(strlen(port_cases[i].input)); 923c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component in_comp(0, url_len); 924c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component out_comp; 925c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string out_str; 926c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output1(&out_str); 927c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool success = url_canon::CanonicalizePort(port_cases[i].input, in_comp, 928c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) port_cases[i].default_port, 929c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) &output1, &out_comp); 930c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output1.Complete(); 931c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 932c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(port_cases[i].expected_success, success); 933c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(std::string(port_cases[i].expected), out_str); 934c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(port_cases[i].expected_component.begin, out_comp.begin); 935c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(port_cases[i].expected_component.len, out_comp.len); 936c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 937c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Now try the wide version 938c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) out_str.clear(); 939c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output2(&out_str); 9407d4cd473f85ac64c3747c96c277f9e506a0d2246Torne (Richard Coles) base::string16 wide_input(ConvertUTF8ToUTF16(port_cases[i].input)); 941c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) success = url_canon::CanonicalizePort(wide_input.c_str(), in_comp, 942c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) port_cases[i].default_port, 943c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) &output2, &out_comp); 944c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output2.Complete(); 945c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 946c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(port_cases[i].expected_success, success); 947c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(std::string(port_cases[i].expected), out_str); 948c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(port_cases[i].expected_component.begin, out_comp.begin); 949c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(port_cases[i].expected_component.len, out_comp.len); 950c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 951c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 952c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 953c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)TEST(URLCanonTest, Path) { 954c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) DualComponentCase path_cases[] = { 955c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // ----- path collapsing tests ----- 956c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"/././foo", L"/././foo", "/foo", url_parse::Component(0, 4), true}, 957c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"/./.foo", L"/./.foo", "/.foo", url_parse::Component(0, 5), true}, 958c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"/foo/.", L"/foo/.", "/foo/", url_parse::Component(0, 5), true}, 959c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"/foo/./", L"/foo/./", "/foo/", url_parse::Component(0, 5), true}, 960c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // double dots followed by a slash or the end of the string count 961c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"/foo/bar/..", L"/foo/bar/..", "/foo/", url_parse::Component(0, 5), true}, 962c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"/foo/bar/../", L"/foo/bar/../", "/foo/", url_parse::Component(0, 5), true}, 963c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // don't count double dots when they aren't followed by a slash 964c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"/foo/..bar", L"/foo/..bar", "/foo/..bar", url_parse::Component(0, 10), true}, 965c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // some in the middle 966c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"/foo/bar/../ton", L"/foo/bar/../ton", "/foo/ton", url_parse::Component(0, 8), true}, 967c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"/foo/bar/../ton/../../a", L"/foo/bar/../ton/../../a", "/a", url_parse::Component(0, 2), true}, 968c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // we should not be able to go above the root 969c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"/foo/../../..", L"/foo/../../..", "/", url_parse::Component(0, 1), true}, 970c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"/foo/../../../ton", L"/foo/../../../ton", "/ton", url_parse::Component(0, 4), true}, 971c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // escaped dots should be unescaped and treated the same as dots 972c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"/foo/%2e", L"/foo/%2e", "/foo/", url_parse::Component(0, 5), true}, 973c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"/foo/%2e%2", L"/foo/%2e%2", "/foo/.%2", url_parse::Component(0, 8), true}, 974c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"/foo/%2e./%2e%2e/.%2e/%2e.bar", L"/foo/%2e./%2e%2e/.%2e/%2e.bar", "/..bar", url_parse::Component(0, 6), true}, 975c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Multiple slashes in a row should be preserved and treated like empty 976c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // directory names. 977c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"////../..", L"////../..", "//", url_parse::Component(0, 2), true}, 978c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 979c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // ----- escaping tests ----- 980c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"/foo", L"/foo", "/foo", url_parse::Component(0, 4), true}, 981c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Valid escape sequence 982c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"/%20foo", L"/%20foo", "/%20foo", url_parse::Component(0, 7), true}, 983c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Invalid escape sequence we should pass through unchanged. 984c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"/foo%", L"/foo%", "/foo%", url_parse::Component(0, 5), true}, 985c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"/foo%2", L"/foo%2", "/foo%2", url_parse::Component(0, 6), true}, 986c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Invalid escape sequence: bad characters should be treated the same as 987c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // the sourrounding text, not as escaped (in this case, UTF-8). 988c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"/foo%2zbar", L"/foo%2zbar", "/foo%2zbar", url_parse::Component(0, 10), true}, 989c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"/foo%2\xc2\xa9zbar", NULL, "/foo%2%C2%A9zbar", url_parse::Component(0, 16), true}, 990c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {NULL, L"/foo%2\xc2\xa9zbar", "/foo%2%C3%82%C2%A9zbar", url_parse::Component(0, 22), true}, 991c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Regular characters that are escaped should be unescaped 992c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"/foo%41%7a", L"/foo%41%7a", "/fooAz", url_parse::Component(0, 6), true}, 993c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Funny characters that are unescaped should be escaped 994c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"/foo\x09\x91%91", NULL, "/foo%09%91%91", url_parse::Component(0, 13), true}, 995c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {NULL, L"/foo\x09\x91%91", "/foo%09%C2%91%91", url_parse::Component(0, 16), true}, 996c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Invalid characters that are escaped should cause a failure. 997c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"/foo%00%51", L"/foo%00%51", "/foo%00Q", url_parse::Component(0, 8), false}, 998c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Some characters should be passed through unchanged regardless of esc. 999c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"/(%28:%3A%29)", L"/(%28:%3A%29)", "/(%28:%3A%29)", url_parse::Component(0, 13), true}, 1000c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Characters that are properly escaped should not have the case changed 1001c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // of hex letters. 1002c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"/%3A%3a%3C%3c", L"/%3A%3a%3C%3c", "/%3A%3a%3C%3c", url_parse::Component(0, 13), true}, 1003c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Funny characters that are unescaped should be escaped 1004c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"/foo\tbar", L"/foo\tbar", "/foo%09bar", url_parse::Component(0, 10), true}, 1005c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Backslashes should get converted to forward slashes 1006c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"\\foo\\bar", L"\\foo\\bar", "/foo/bar", url_parse::Component(0, 8), true}, 1007c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Hashes found in paths (possibly only when the caller explicitly sets 1008c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // the path on an already-parsed URL) should be escaped. 1009c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"/foo#bar", L"/foo#bar", "/foo%23bar", url_parse::Component(0, 10), true}, 1010c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // %7f should be allowed and %3D should not be unescaped (these were wrong 1011c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // in a previous version). 1012c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"/%7Ffp3%3Eju%3Dduvgw%3Dd", L"/%7Ffp3%3Eju%3Dduvgw%3Dd", "/%7Ffp3%3Eju%3Dduvgw%3Dd", url_parse::Component(0, 24), true}, 1013c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // @ should be passed through unchanged (escaped or unescaped). 1014c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"/@asdf%40", L"/@asdf%40", "/@asdf%40", url_parse::Component(0, 9), true}, 1015c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1016c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // ----- encoding tests ----- 1017c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Basic conversions 1018c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"/\xe4\xbd\xa0\xe5\xa5\xbd\xe4\xbd\xa0\xe5\xa5\xbd", L"/\x4f60\x597d\x4f60\x597d", "/%E4%BD%A0%E5%A5%BD%E4%BD%A0%E5%A5%BD", url_parse::Component(0, 37), true}, 1019c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Invalid unicode characters should fail. We only do validation on 1020c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // UTF-16 input, so this doesn't happen on 8-bit. 1021c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"/\xef\xb7\x90zyx", NULL, "/%EF%B7%90zyx", url_parse::Component(0, 13), true}, 1022c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {NULL, L"/\xfdd0zyx", "/%EF%BF%BDzyx", url_parse::Component(0, 13), false}, 1023c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) }; 1024c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1025c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (size_t i = 0; i < arraysize(path_cases); i++) { 1026c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (path_cases[i].input8) { 1027c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int len = static_cast<int>(strlen(path_cases[i].input8)); 1028c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component in_comp(0, len); 1029c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component out_comp; 1030c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string out_str; 1031c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output(&out_str); 1032c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool success = url_canon::CanonicalizePath(path_cases[i].input8, in_comp, 1033c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) &output, &out_comp); 1034c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output.Complete(); 1035c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1036c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(path_cases[i].expected_success, success); 1037c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(path_cases[i].expected_component.begin, out_comp.begin); 1038c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(path_cases[i].expected_component.len, out_comp.len); 1039c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(path_cases[i].expected, out_str); 1040c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 1041c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1042c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (path_cases[i].input16) { 10437d4cd473f85ac64c3747c96c277f9e506a0d2246Torne (Richard Coles) base::string16 input16(WStringToUTF16(path_cases[i].input16)); 1044c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int len = static_cast<int>(input16.length()); 1045c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component in_comp(0, len); 1046c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component out_comp; 1047c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string out_str; 1048c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output(&out_str); 1049c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1050c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool success = url_canon::CanonicalizePath(input16.c_str(), in_comp, 1051c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) &output, &out_comp); 1052c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output.Complete(); 1053c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1054c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(path_cases[i].expected_success, success); 1055c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(path_cases[i].expected_component.begin, out_comp.begin); 1056c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(path_cases[i].expected_component.len, out_comp.len); 1057c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(path_cases[i].expected, out_str); 1058c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 1059c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 1060c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1061c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Manual test: embedded NULLs should be escaped and the URL should be marked 1062c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // as invalid. 1063c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char path_with_null[] = "/ab\0c"; 1064c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component in_comp(0, 5); 1065c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component out_comp; 1066c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1067c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string out_str; 1068c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output(&out_str); 1069c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool success = url_canon::CanonicalizePath(path_with_null, in_comp, 1070c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) &output, &out_comp); 1071c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output.Complete(); 1072c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_FALSE(success); 1073c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ("/ab%00c", out_str); 1074c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 1075c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1076c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)TEST(URLCanonTest, Query) { 1077c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) struct QueryCase { 1078c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* input8; 1079c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const wchar_t* input16; 1080c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* encoding; 1081c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* expected; 1082c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } query_cases[] = { 1083c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Regular ASCII case in some different encodings. 1084c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"foo=bar", L"foo=bar", NULL, "?foo=bar"}, 1085c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"foo=bar", L"foo=bar", "utf-8", "?foo=bar"}, 1086c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"foo=bar", L"foo=bar", "shift_jis", "?foo=bar"}, 1087c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"foo=bar", L"foo=bar", "gb2312", "?foo=bar"}, 1088c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Allow question marks in the query without escaping 1089c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"as?df", L"as?df", NULL, "?as?df"}, 1090c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Always escape '#' since it would mark the ref. 1091c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"as#df", L"as#df", NULL, "?as%23df"}, 1092c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Escape some questionable 8-bit characters, but never unescape. 1093c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"\x02hello\x7f bye", L"\x02hello\x7f bye", NULL, "?%02hello%7F%20bye"}, 1094c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"%40%41123", L"%40%41123", NULL, "?%40%41123"}, 1095c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Chinese input/output 1096c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"q=\xe4\xbd\xa0\xe5\xa5\xbd", L"q=\x4f60\x597d", NULL, "?q=%E4%BD%A0%E5%A5%BD"}, 1097c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"q=\xe4\xbd\xa0\xe5\xa5\xbd", L"q=\x4f60\x597d", "gb2312", "?q=%C4%E3%BA%C3"}, 1098c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"q=\xe4\xbd\xa0\xe5\xa5\xbd", L"q=\x4f60\x597d", "big5", "?q=%A7A%A6n"}, 1099c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Unencodable character in the destination character set should be 1100c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // escaped. The escape sequence unescapes to be the entity name: 1101c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // "?q=你" 1102c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"q=Chinese\xef\xbc\xa7", L"q=Chinese\xff27", "iso-8859-1", "?q=Chinese%26%2365319%3B"}, 1103c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Invalid UTF-8/16 input should be replaced with invalid characters. 1104c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"q=\xed\xed", L"q=\xd800\xd800", NULL, "?q=%EF%BF%BD%EF%BF%BD"}, 1105c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Don't allow < or > because sometimes they are used for XSS if the 1106c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // URL is echoed in content. Firefox does this, IE doesn't. 1107c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"q=<asdf>", L"q=<asdf>", NULL, "?q=%3Casdf%3E"}, 1108c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Escape double quotemarks in the query. 1109c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"q=\"asdf\"", L"q=\"asdf\"", NULL, "?q=%22asdf%22"}, 1110c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) }; 1111c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1112c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (size_t i = 0; i < ARRAYSIZE(query_cases); i++) { 1113c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component out_comp; 1114c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1115c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) UConvScoper conv(query_cases[i].encoding); 1116c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) ASSERT_TRUE(!query_cases[i].encoding || conv.converter()); 1117c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::ICUCharsetConverter converter(conv.converter()); 1118c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1119c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Map NULL to a NULL converter pointer. 1120c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::ICUCharsetConverter* conv_pointer = &converter; 1121c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (!query_cases[i].encoding) 1122c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) conv_pointer = NULL; 1123c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1124c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (query_cases[i].input8) { 1125c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int len = static_cast<int>(strlen(query_cases[i].input8)); 1126c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component in_comp(0, len); 1127c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string out_str; 1128c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1129c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output(&out_str); 1130c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::CanonicalizeQuery(query_cases[i].input8, in_comp, 1131c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) conv_pointer, &output, &out_comp); 1132c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output.Complete(); 1133c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1134c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(query_cases[i].expected, out_str); 1135c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 1136c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1137c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (query_cases[i].input16) { 11387d4cd473f85ac64c3747c96c277f9e506a0d2246Torne (Richard Coles) base::string16 input16(WStringToUTF16(query_cases[i].input16)); 1139c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int len = static_cast<int>(input16.length()); 1140c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component in_comp(0, len); 1141c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string out_str; 1142c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1143c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output(&out_str); 1144c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::CanonicalizeQuery(input16.c_str(), in_comp, 1145c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) conv_pointer, &output, &out_comp); 1146c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output.Complete(); 1147c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1148c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(query_cases[i].expected, out_str); 1149c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 1150c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 1151c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1152c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Extra test for input with embedded NULL; 1153c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string out_str; 1154c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output(&out_str); 1155c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component out_comp; 1156c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::CanonicalizeQuery("a \x00z\x01", url_parse::Component(0, 5), NULL, 1157c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) &output, &out_comp); 1158c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output.Complete(); 1159c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ("?a%20%00z%01", out_str); 1160c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 1161c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1162c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)TEST(URLCanonTest, Ref) { 1163c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Refs are trivial, it just checks the encoding. 1164c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) DualComponentCase ref_cases[] = { 1165c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Regular one, we shouldn't escape spaces, et al. 1166c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"hello, world", L"hello, world", "#hello, world", url_parse::Component(1, 12), true}, 1167c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // UTF-8/wide input should be preserved 1168c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"\xc2\xa9", L"\xa9", "#\xc2\xa9", url_parse::Component(1, 2), true}, 1169c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Test a characer that takes > 16 bits (U+10300 = old italic letter A) 1170c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"\xF0\x90\x8C\x80ss", L"\xd800\xdf00ss", "#\xF0\x90\x8C\x80ss", url_parse::Component(1, 6), true}, 1171c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Escaping should be preserved unchanged, even invalid ones 1172c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"%41%a", L"%41%a", "#%41%a", url_parse::Component(1, 5), true}, 1173c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Invalid UTF-8/16 input should be flagged and the input made valid 1174c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"\xc2", NULL, "#\xef\xbf\xbd", url_parse::Component(1, 3), true}, 1175c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {NULL, L"\xd800\x597d", "#\xef\xbf\xbd\xe5\xa5\xbd", url_parse::Component(1, 6), true}, 1176c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Test a Unicode invalid character. 1177c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"a\xef\xb7\x90", L"a\xfdd0", "#a\xef\xbf\xbd", url_parse::Component(1, 4), true}, 1178c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Refs can have # signs and we should preserve them. 1179c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"asdf#qwer", L"asdf#qwer", "#asdf#qwer", url_parse::Component(1, 9), true}, 1180c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"#asdf", L"#asdf", "##asdf", url_parse::Component(1, 5), true}, 1181c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) }; 1182c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1183c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (size_t i = 0; i < arraysize(ref_cases); i++) { 1184c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // 8-bit input 1185c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (ref_cases[i].input8) { 1186c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int len = static_cast<int>(strlen(ref_cases[i].input8)); 1187c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component in_comp(0, len); 1188c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component out_comp; 1189c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1190c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string out_str; 1191c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output(&out_str); 1192c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::CanonicalizeRef(ref_cases[i].input8, in_comp, 1193c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) &output, &out_comp); 1194c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output.Complete(); 1195c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1196c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(ref_cases[i].expected_component.begin, out_comp.begin); 1197c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(ref_cases[i].expected_component.len, out_comp.len); 1198c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(ref_cases[i].expected, out_str); 1199c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 1200c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1201c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // 16-bit input 1202c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (ref_cases[i].input16) { 12037d4cd473f85ac64c3747c96c277f9e506a0d2246Torne (Richard Coles) base::string16 input16(WStringToUTF16(ref_cases[i].input16)); 1204c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int len = static_cast<int>(input16.length()); 1205c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component in_comp(0, len); 1206c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component out_comp; 1207c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1208c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string out_str; 1209c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output(&out_str); 1210c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::CanonicalizeRef(input16.c_str(), in_comp, &output, &out_comp); 1211c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output.Complete(); 1212c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1213c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(ref_cases[i].expected_component.begin, out_comp.begin); 1214c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(ref_cases[i].expected_component.len, out_comp.len); 1215c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(ref_cases[i].expected, out_str); 1216c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 1217c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 1218c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1219c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Try one with an embedded NULL. It should be stripped. 1220c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char null_input[5] = "ab\x00z"; 1221c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component null_input_component(0, 4); 1222c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component out_comp; 1223c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1224c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string out_str; 1225c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output(&out_str); 1226c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::CanonicalizeRef(null_input, null_input_component, 1227c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) &output, &out_comp); 1228c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output.Complete(); 1229c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1230c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(1, out_comp.begin); 1231c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(3, out_comp.len); 1232c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ("#abz", out_str); 1233c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 1234c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1235c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)TEST(URLCanonTest, CanonicalizeStandardURL) { 1236c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // The individual component canonicalize tests should have caught the cases 1237c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // for each of those components. Here, we just need to test that the various 1238c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // parts are included or excluded properly, and have the correct separators. 1239c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) struct URLCase { 1240c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* input; 1241c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* expected; 1242c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool expected_success; 1243c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } cases[] = { 1244c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://www.google.com/foo?bar=baz#", "http://www.google.com/foo?bar=baz#", true}, 1245c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://[www.google.com]/", "http://[www.google.com]/", false}, 1246c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"ht\ttp:@www.google.com:80/;p?#", "ht%09tp://www.google.com:80/;p?#", false}, 1247c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http:////////user:@google.com:99?foo", "http://user@google.com:99/?foo", true}, 1248c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"www.google.com", ":www.google.com/", true}, 1249c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://192.0x00A80001", "http://192.168.0.1/", true}, 1250c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://www/foo%2Ehtml", "http://www/foo.html", true}, 1251c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://user:pass@/", "http://user:pass@/", false}, 1252c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://%25DOMAIN:foobar@foodomain.com/", "http://%25DOMAIN:foobar@foodomain.com/", true}, 1253c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1254c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Backslashes should get converted to forward slashes. 1255c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http:\\\\www.google.com\\foo", "http://www.google.com/foo", true}, 1256c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1257c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Busted refs shouldn't make the whole thing fail. 1258c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://www.google.com/asdf#\xc2", "http://www.google.com/asdf#\xef\xbf\xbd", true}, 1259c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1260c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Basic port tests. 1261c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://foo:80/", "http://foo/", true}, 1262c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://foo:81/", "http://foo:81/", true}, 1263c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"httpa://foo:80/", "httpa://foo:80/", true}, 1264c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://foo:-80/", "http://foo:-80/", false}, 1265c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1266c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"https://foo:443/", "https://foo/", true}, 1267c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"https://foo:80/", "https://foo:80/", true}, 1268c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"ftp://foo:21/", "ftp://foo/", true}, 1269c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"ftp://foo:80/", "ftp://foo:80/", true}, 1270c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"gopher://foo:70/", "gopher://foo/", true}, 1271c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"gopher://foo:443/", "gopher://foo:443/", true}, 1272c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"ws://foo:80/", "ws://foo/", true}, 1273c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"ws://foo:81/", "ws://foo:81/", true}, 1274c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"ws://foo:443/", "ws://foo:443/", true}, 1275c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"ws://foo:815/", "ws://foo:815/", true}, 1276c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"wss://foo:80/", "wss://foo:80/", true}, 1277c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"wss://foo:81/", "wss://foo:81/", true}, 1278c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"wss://foo:443/", "wss://foo/", true}, 1279c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"wss://foo:815/", "wss://foo:815/", true}, 1280c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) }; 1281c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1282c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (size_t i = 0; i < ARRAYSIZE(cases); i++) { 1283c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int url_len = static_cast<int>(strlen(cases[i].input)); 1284c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Parsed parsed; 1285c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::ParseStandardURL(cases[i].input, url_len, &parsed); 1286c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1287c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Parsed out_parsed; 1288c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string out_str; 1289c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output(&out_str); 1290c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool success = url_canon::CanonicalizeStandardURL( 1291c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) cases[i].input, url_len, parsed, NULL, &output, &out_parsed); 1292c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output.Complete(); 1293c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1294c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(cases[i].expected_success, success); 1295c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(cases[i].expected, out_str); 1296c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 1297c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 1298c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1299c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// The codepath here is the same as for regular canonicalization, so we just 1300c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// need to test that things are replaced or not correctly. 1301c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)TEST(URLCanonTest, ReplaceStandardURL) { 1302c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) ReplaceCase replace_cases[] = { 1303c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Common case of truncating the path. 1304c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://www.google.com/foo?bar=baz#ref", NULL, NULL, NULL, NULL, NULL, "/", kDeleteComp, kDeleteComp, "http://www.google.com/"}, 1305c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Replace everything 1306c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://a:b@google.com:22/foo;bar?baz@cat", "https", "me", "pw", "host.com", "99", "/path", "query", "ref", "https://me:pw@host.com:99/path?query#ref"}, 1307c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Replace nothing 1308c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://a:b@google.com:22/foo?baz@cat", NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, "http://a:b@google.com:22/foo?baz@cat"}, 1309c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Replace scheme with filesystem. The result is garbage, but you asked 1310c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // for it. 1311c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://a:b@google.com:22/foo?baz@cat", "filesystem", NULL, NULL, NULL, NULL, NULL, NULL, NULL, "filesystem://a:b@google.com:22/foo?baz@cat"}, 1312c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) }; 1313c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1314c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (size_t i = 0; i < arraysize(replace_cases); i++) { 1315c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const ReplaceCase& cur = replace_cases[i]; 1316c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int base_len = static_cast<int>(strlen(cur.base)); 1317c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Parsed parsed; 1318c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::ParseStandardURL(cur.base, base_len, &parsed); 1319c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1320c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::Replacements<char> r; 1321c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) typedef url_canon::Replacements<char> R; // Clean up syntax. 1322c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1323c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Note that for the scheme we pass in a different clear function since 1324c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // there is no function to clear the scheme. 1325c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetScheme, &R::ClearRef, &r, cur.scheme); 1326c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetUsername, &R::ClearUsername, &r, cur.username); 1327c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetPassword, &R::ClearPassword, &r, cur.password); 1328c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetHost, &R::ClearHost, &r, cur.host); 1329c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetPort, &R::ClearPort, &r, cur.port); 1330c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetPath, &R::ClearPath, &r, cur.path); 1331c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetQuery, &R::ClearQuery, &r, cur.query); 1332c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetRef, &R::ClearRef, &r, cur.ref); 1333c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1334c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string out_str; 1335c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output(&out_str); 1336c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Parsed out_parsed; 1337c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::ReplaceStandardURL(replace_cases[i].base, parsed, 1338c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) r, NULL, &output, &out_parsed); 1339c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output.Complete(); 1340c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1341c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(replace_cases[i].expected, out_str); 1342c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 1343c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1344c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // The path pointer should be ignored if the address is invalid. 1345c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) { 1346c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char src[] = "http://www.google.com/here_is_the_path"; 1347c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int src_len = static_cast<int>(strlen(src)); 1348c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1349c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Parsed parsed; 1350c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::ParseStandardURL(src, src_len, &parsed); 1351c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1352c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Replace the path to 0 length string. By using 1 as the string address, 1353c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // the test should get an access violation if it tries to dereference it. 1354c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::Replacements<char> r; 1355c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) r.SetPath(reinterpret_cast<char*>(0x00000001), url_parse::Component(0, 0)); 1356c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string out_str1; 1357c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output1(&out_str1); 1358c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Parsed new_parsed; 1359c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::ReplaceStandardURL(src, parsed, r, NULL, &output1, &new_parsed); 1360c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output1.Complete(); 1361c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_STREQ("http://www.google.com/", out_str1.c_str()); 1362c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1363c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Same with an "invalid" path. 1364c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) r.SetPath(reinterpret_cast<char*>(0x00000001), url_parse::Component()); 1365c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string out_str2; 1366c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output2(&out_str2); 1367c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::ReplaceStandardURL(src, parsed, r, NULL, &output2, &new_parsed); 1368c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output2.Complete(); 1369c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_STREQ("http://www.google.com/", out_str2.c_str()); 1370c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 1371c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 1372c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1373c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)TEST(URLCanonTest, ReplaceFileURL) { 1374c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) ReplaceCase replace_cases[] = { 1375c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Replace everything 1376c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file:///C:/gaba?query#ref", NULL, NULL, NULL, "filer", NULL, "/foo", "b", "c", "file://filer/foo?b#c"}, 1377c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Replace nothing 1378c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file:///C:/gaba?query#ref", NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, "file:///C:/gaba?query#ref"}, 1379c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Clear non-path components (common) 1380c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file:///C:/gaba?query#ref", NULL, NULL, NULL, NULL, NULL, NULL, kDeleteComp, kDeleteComp, "file:///C:/gaba"}, 1381c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Replace path with something that doesn't begin with a slash and make 1382c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // sure it gets added properly. 1383c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file:///C:/gaba", NULL, NULL, NULL, NULL, NULL, "interesting/", NULL, NULL, "file:///interesting/"}, 1384c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file:///home/gaba?query#ref", NULL, NULL, NULL, "filer", NULL, "/foo", "b", "c", "file://filer/foo?b#c"}, 1385c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file:///home/gaba?query#ref", NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, "file:///home/gaba?query#ref"}, 1386c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file:///home/gaba?query#ref", NULL, NULL, NULL, NULL, NULL, NULL, kDeleteComp, kDeleteComp, "file:///home/gaba"}, 1387c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file:///home/gaba", NULL, NULL, NULL, NULL, NULL, "interesting/", NULL, NULL, "file:///interesting/"}, 1388c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Replace scheme -- shouldn't do anything. 1389c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file:///C:/gaba?query#ref", "http", NULL, NULL, NULL, NULL, NULL, NULL, NULL, "file:///C:/gaba?query#ref"}, 1390c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) }; 1391c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1392c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (size_t i = 0; i < arraysize(replace_cases); i++) { 1393c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const ReplaceCase& cur = replace_cases[i]; 1394c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int base_len = static_cast<int>(strlen(cur.base)); 1395c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Parsed parsed; 1396c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::ParseFileURL(cur.base, base_len, &parsed); 1397c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1398c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::Replacements<char> r; 1399c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) typedef url_canon::Replacements<char> R; // Clean up syntax. 1400c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetScheme, &R::ClearRef, &r, cur.scheme); 1401c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetUsername, &R::ClearUsername, &r, cur.username); 1402c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetPassword, &R::ClearPassword, &r, cur.password); 1403c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetHost, &R::ClearHost, &r, cur.host); 1404c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetPort, &R::ClearPort, &r, cur.port); 1405c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetPath, &R::ClearPath, &r, cur.path); 1406c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetQuery, &R::ClearQuery, &r, cur.query); 1407c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetRef, &R::ClearRef, &r, cur.ref); 1408c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1409c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string out_str; 1410c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output(&out_str); 1411c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Parsed out_parsed; 1412c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::ReplaceFileURL(cur.base, parsed, 1413c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) r, NULL, &output, &out_parsed); 1414c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output.Complete(); 1415c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1416c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(replace_cases[i].expected, out_str); 1417c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 1418c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 1419c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1420c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)TEST(URLCanonTest, ReplaceFileSystemURL) { 1421c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) ReplaceCase replace_cases[] = { 1422c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Replace everything in the outer URL. 1423c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"filesystem:file:///temporary/gaba?query#ref", NULL, NULL, NULL, NULL, NULL, "/foo", "b", "c", "filesystem:file:///temporary/foo?b#c"}, 1424c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Replace nothing 1425c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"filesystem:file:///temporary/gaba?query#ref", NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, "filesystem:file:///temporary/gaba?query#ref"}, 1426c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Clear non-path components (common) 1427c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"filesystem:file:///temporary/gaba?query#ref", NULL, NULL, NULL, NULL, NULL, NULL, kDeleteComp, kDeleteComp, "filesystem:file:///temporary/gaba"}, 1428c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Replace path with something that doesn't begin with a slash and make 1429c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // sure it gets added properly. 1430c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"filesystem:file:///temporary/gaba?query#ref", NULL, NULL, NULL, NULL, NULL, "interesting/", NULL, NULL, "filesystem:file:///temporary/interesting/?query#ref"}, 1431c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Replace scheme -- shouldn't do anything. 1432c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"filesystem:http://u:p@bar.com/t/gaba?query#ref", "http", NULL, NULL, NULL, NULL, NULL, NULL, NULL, "filesystem:http://u:p@bar.com/t/gaba?query#ref"}, 1433c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Replace username -- shouldn't do anything. 1434c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"filesystem:http://u:p@bar.com/t/gaba?query#ref", NULL, "u2", NULL, NULL, NULL, NULL, NULL, NULL, "filesystem:http://u:p@bar.com/t/gaba?query#ref"}, 1435c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Replace password -- shouldn't do anything. 1436c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"filesystem:http://u:p@bar.com/t/gaba?query#ref", NULL, NULL, "pw2", NULL, NULL, NULL, NULL, NULL, "filesystem:http://u:p@bar.com/t/gaba?query#ref"}, 1437c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Replace host -- shouldn't do anything. 1438c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"filesystem:http://u:p@bar.com/t/gaba?query#ref", NULL, NULL, NULL, "foo.com", NULL, NULL, NULL, NULL, "filesystem:http://u:p@bar.com/t/gaba?query#ref"}, 1439c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Replace port -- shouldn't do anything. 1440c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"filesystem:http://u:p@bar.com:40/t/gaba?query#ref", NULL, NULL, NULL, NULL, "41", NULL, NULL, NULL, "filesystem:http://u:p@bar.com:40/t/gaba?query#ref"}, 1441c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) }; 1442c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1443c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (size_t i = 0; i < arraysize(replace_cases); i++) { 1444c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const ReplaceCase& cur = replace_cases[i]; 1445c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int base_len = static_cast<int>(strlen(cur.base)); 1446c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Parsed parsed; 1447c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::ParseFileSystemURL(cur.base, base_len, &parsed); 1448c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1449c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::Replacements<char> r; 1450c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) typedef url_canon::Replacements<char> R; // Clean up syntax. 1451c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetScheme, &R::ClearRef, &r, cur.scheme); 1452c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetUsername, &R::ClearUsername, &r, cur.username); 1453c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetPassword, &R::ClearPassword, &r, cur.password); 1454c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetHost, &R::ClearHost, &r, cur.host); 1455c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetPort, &R::ClearPort, &r, cur.port); 1456c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetPath, &R::ClearPath, &r, cur.path); 1457c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetQuery, &R::ClearQuery, &r, cur.query); 1458c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetRef, &R::ClearRef, &r, cur.ref); 1459c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1460c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string out_str; 1461c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output(&out_str); 1462c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Parsed out_parsed; 1463c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::ReplaceFileSystemURL(cur.base, parsed, r, NULL, 1464c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) &output, &out_parsed); 1465c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output.Complete(); 1466c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1467c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(replace_cases[i].expected, out_str); 1468c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 1469c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 1470c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1471c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)TEST(URLCanonTest, ReplacePathURL) { 1472c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) ReplaceCase replace_cases[] = { 1473c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Replace everything 1474c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"data:foo", "javascript", NULL, NULL, NULL, NULL, "alert('foo?');", NULL, NULL, "javascript:alert('foo?');"}, 1475c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Replace nothing 1476c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"data:foo", NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, "data:foo"}, 1477c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Replace one or the other 1478c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"data:foo", "javascript", NULL, NULL, NULL, NULL, NULL, NULL, NULL, "javascript:foo"}, 1479c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"data:foo", NULL, NULL, NULL, NULL, NULL, "bar", NULL, NULL, "data:bar"}, 1480c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"data:foo", NULL, NULL, NULL, NULL, NULL, kDeleteComp, NULL, NULL, "data:"}, 1481c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) }; 1482c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1483c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (size_t i = 0; i < arraysize(replace_cases); i++) { 1484c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const ReplaceCase& cur = replace_cases[i]; 1485c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int base_len = static_cast<int>(strlen(cur.base)); 1486c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Parsed parsed; 1487c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::ParsePathURL(cur.base, base_len, &parsed); 1488c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1489c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::Replacements<char> r; 1490c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) typedef url_canon::Replacements<char> R; // Clean up syntax. 1491c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetScheme, &R::ClearRef, &r, cur.scheme); 1492c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetUsername, &R::ClearUsername, &r, cur.username); 1493c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetPassword, &R::ClearPassword, &r, cur.password); 1494c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetHost, &R::ClearHost, &r, cur.host); 1495c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetPort, &R::ClearPort, &r, cur.port); 1496c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetPath, &R::ClearPath, &r, cur.path); 1497c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetQuery, &R::ClearQuery, &r, cur.query); 1498c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetRef, &R::ClearRef, &r, cur.ref); 1499c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1500c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string out_str; 1501c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output(&out_str); 1502c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Parsed out_parsed; 1503c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::ReplacePathURL(cur.base, parsed, 1504c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) r, &output, &out_parsed); 1505c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output.Complete(); 1506c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1507c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(replace_cases[i].expected, out_str); 1508c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 1509c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 1510c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1511c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)TEST(URLCanonTest, ReplaceMailtoURL) { 1512c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) ReplaceCase replace_cases[] = { 1513c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Replace everything 1514c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"mailto:jon@foo.com?body=sup", "mailto", NULL, NULL, NULL, NULL, "addr1", "to=tony", NULL, "mailto:addr1?to=tony"}, 1515c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Replace nothing 1516c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"mailto:jon@foo.com?body=sup", NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, "mailto:jon@foo.com?body=sup"}, 1517c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Replace the path 1518c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"mailto:jon@foo.com?body=sup", NULL, NULL, NULL, NULL, NULL, "jason", NULL, NULL, "mailto:jason?body=sup"}, 1519c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Replace the query 1520c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"mailto:jon@foo.com?body=sup", NULL, NULL, NULL, NULL, NULL, NULL, "custom=1", NULL, "mailto:jon@foo.com?custom=1"}, 1521c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Replace the path and query 1522c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"mailto:jon@foo.com?body=sup", NULL, NULL, NULL, NULL, NULL, "jason", "custom=1", NULL, "mailto:jason?custom=1"}, 1523c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Set the query to empty (should leave trailing question mark) 1524c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"mailto:jon@foo.com?body=sup", NULL, NULL, NULL, NULL, NULL, NULL, "", NULL, "mailto:jon@foo.com?"}, 1525c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Clear the query 1526c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"mailto:jon@foo.com?body=sup", NULL, NULL, NULL, NULL, NULL, NULL, "|", NULL, "mailto:jon@foo.com"}, 1527c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Clear the path 1528c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"mailto:jon@foo.com?body=sup", NULL, NULL, NULL, NULL, NULL, "|", NULL, NULL, "mailto:?body=sup"}, 1529c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Clear the path + query 1530c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"mailto:", NULL, NULL, NULL, NULL, NULL, "|", "|", NULL, "mailto:"}, 1531c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Setting the ref should have no effect 1532c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"mailto:addr1", NULL, NULL, NULL, NULL, NULL, NULL, NULL, "BLAH", "mailto:addr1"}, 1533c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) }; 1534c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1535c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (size_t i = 0; i < arraysize(replace_cases); i++) { 1536c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const ReplaceCase& cur = replace_cases[i]; 1537c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int base_len = static_cast<int>(strlen(cur.base)); 1538c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Parsed parsed; 1539c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::ParseMailtoURL(cur.base, base_len, &parsed); 1540c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1541c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::Replacements<char> r; 1542c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) typedef url_canon::Replacements<char> R; 1543c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetScheme, &R::ClearRef, &r, cur.scheme); 1544c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetUsername, &R::ClearUsername, &r, cur.username); 1545c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetPassword, &R::ClearPassword, &r, cur.password); 1546c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetHost, &R::ClearHost, &r, cur.host); 1547c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetPort, &R::ClearPort, &r, cur.port); 1548c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetPath, &R::ClearPath, &r, cur.path); 1549c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetQuery, &R::ClearQuery, &r, cur.query); 1550c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) SetupReplComp(&R::SetRef, &R::ClearRef, &r, cur.ref); 1551c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1552c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string out_str; 1553c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output(&out_str); 1554c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Parsed out_parsed; 1555c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::ReplaceMailtoURL(cur.base, parsed, 1556c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) r, &output, &out_parsed); 1557c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output.Complete(); 1558c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1559c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(replace_cases[i].expected, out_str); 1560c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 1561c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 1562c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1563c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)TEST(URLCanonTest, CanonicalizeFileURL) { 1564c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) struct URLCase { 1565c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* input; 1566c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* expected; 1567c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool expected_success; 1568c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component expected_host; 1569c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component expected_path; 1570c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } cases[] = { 1571c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#ifdef _WIN32 1572c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Windows-style paths 1573c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file:c:\\foo\\bar.html", "file:///C:/foo/bar.html", true, url_parse::Component(), url_parse::Component(7, 16)}, 1574c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {" File:c|////foo\\bar.html", "file:///C:////foo/bar.html", true, url_parse::Component(), url_parse::Component(7, 19)}, 1575c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file:", "file:///", true, url_parse::Component(), url_parse::Component(7, 1)}, 1576c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file:UNChost/path", "file://unchost/path", true, url_parse::Component(7, 7), url_parse::Component(14, 5)}, 1577c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // CanonicalizeFileURL supports absolute Windows style paths for IE 1578c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // compatability. Note that the caller must decide that this is a file 1579c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // URL itself so it can call the file canonicalizer. This is usually 1580c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // done automatically as part of relative URL resolving. 1581c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"c:\\foo\\bar", "file:///C:/foo/bar", true, url_parse::Component(), url_parse::Component(7, 11)}, 1582c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"C|/foo/bar", "file:///C:/foo/bar", true, url_parse::Component(), url_parse::Component(7, 11)}, 1583c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"/C|\\foo\\bar", "file:///C:/foo/bar", true, url_parse::Component(), url_parse::Component(7, 11)}, 1584c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"//C|/foo/bar", "file:///C:/foo/bar", true, url_parse::Component(), url_parse::Component(7, 11)}, 1585c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"//server/file", "file://server/file", true, url_parse::Component(7, 6), url_parse::Component(13, 5)}, 1586c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"\\\\server\\file", "file://server/file", true, url_parse::Component(7, 6), url_parse::Component(13, 5)}, 1587c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"/\\server/file", "file://server/file", true, url_parse::Component(7, 6), url_parse::Component(13, 5)}, 1588c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // We should preserve the number of slashes after the colon for IE 1589c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // compatability, except when there is none, in which case we should 1590c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // add one. 1591c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file:c:foo/bar.html", "file:///C:/foo/bar.html", true, url_parse::Component(), url_parse::Component(7, 16)}, 1592c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file:/\\/\\C:\\\\//foo\\bar.html", "file:///C:////foo/bar.html", true, url_parse::Component(), url_parse::Component(7, 19)}, 1593c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Three slashes should be non-UNC, even if there is no drive spec (IE 1594c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // does this, which makes the resulting request invalid). 1595c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file:///foo/bar.txt", "file:///foo/bar.txt", true, url_parse::Component(), url_parse::Component(7, 12)}, 1596c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // TODO(brettw) we should probably fail for invalid host names, which 1597c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // would change the expected result on this test. We also currently allow 1598c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // colon even though it's probably invalid, because its currently the 1599c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // "natural" result of the way the canonicalizer is written. There doesn't 1600c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // seem to be a strong argument for why allowing it here would be bad, so 1601c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // we just tolerate it and the load will fail later. 1602c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"FILE:/\\/\\7:\\\\//foo\\bar.html", "file://7:////foo/bar.html", false, url_parse::Component(7, 2), url_parse::Component(9, 16)}, 1603c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file:filer/home\\me", "file://filer/home/me", true, url_parse::Component(7, 5), url_parse::Component(12, 8)}, 1604c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Make sure relative paths can't go above the "C:" 1605c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file:///C:/foo/../../../bar.html", "file:///C:/bar.html", true, url_parse::Component(), url_parse::Component(7, 12)}, 1606c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Busted refs shouldn't make the whole thing fail. 1607c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file:///C:/asdf#\xc2", "file:///C:/asdf#\xef\xbf\xbd", true, url_parse::Component(), url_parse::Component(7, 8)}, 1608c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#else 1609c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Unix-style paths 1610c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file:///home/me", "file:///home/me", true, url_parse::Component(), url_parse::Component(7, 8)}, 1611c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Windowsy ones should get still treated as Unix-style. 1612c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file:c:\\foo\\bar.html", "file:///c:/foo/bar.html", true, url_parse::Component(), url_parse::Component(7, 16)}, 1613c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file:c|//foo\\bar.html", "file:///c%7C//foo/bar.html", true, url_parse::Component(), url_parse::Component(7, 19)}, 1614c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // file: tests from WebKit (LayoutTests/fast/loader/url-parse-1.html) 1615c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"//", "file:///", true, url_parse::Component(), url_parse::Component(7, 1)}, 1616c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"///", "file:///", true, url_parse::Component(), url_parse::Component(7, 1)}, 1617c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"///test", "file:///test", true, url_parse::Component(), url_parse::Component(7, 5)}, 1618c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file://test", "file://test/", true, url_parse::Component(7, 4), url_parse::Component(11, 1)}, 1619c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file://localhost", "file://localhost/", true, url_parse::Component(7, 9), url_parse::Component(16, 1)}, 1620c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file://localhost/", "file://localhost/", true, url_parse::Component(7, 9), url_parse::Component(16, 1)}, 1621c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file://localhost/test", "file://localhost/test", true, url_parse::Component(7, 9), url_parse::Component(16, 5)}, 1622c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#endif // _WIN32 1623c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) }; 1624c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1625c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (size_t i = 0; i < ARRAYSIZE(cases); i++) { 1626c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int url_len = static_cast<int>(strlen(cases[i].input)); 1627c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Parsed parsed; 1628c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::ParseFileURL(cases[i].input, url_len, &parsed); 1629c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1630c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Parsed out_parsed; 1631c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string out_str; 1632c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output(&out_str); 1633c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool success = url_canon::CanonicalizeFileURL(cases[i].input, url_len, 1634c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) parsed, NULL, &output, 1635c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) &out_parsed); 1636c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output.Complete(); 1637c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1638c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(cases[i].expected_success, success); 1639c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(cases[i].expected, out_str); 1640c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1641c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Make sure the spec was properly identified, the file canonicalizer has 1642c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // different code for writing the spec. 1643c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(0, out_parsed.scheme.begin); 1644c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(4, out_parsed.scheme.len); 1645c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1646c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(cases[i].expected_host.begin, out_parsed.host.begin); 1647c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(cases[i].expected_host.len, out_parsed.host.len); 1648c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1649c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(cases[i].expected_path.begin, out_parsed.path.begin); 1650c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(cases[i].expected_path.len, out_parsed.path.len); 1651c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 1652c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 1653c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1654c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)TEST(URLCanonTest, CanonicalizeFileSystemURL) { 1655c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) struct URLCase { 1656c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* input; 1657c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* expected; 1658c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool expected_success; 1659c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } cases[] = { 1660c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"Filesystem:htTp://www.Foo.com:80/tempoRary", "filesystem:http://www.foo.com/tempoRary/", true}, 1661c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"filesystem:httpS://www.foo.com/temporary/", "filesystem:https://www.foo.com/temporary/", true}, 1662c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"filesystem:http://www.foo.com//", "filesystem:http://www.foo.com//", false}, 1663c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"filesystem:http://www.foo.com/persistent/bob?query#ref", "filesystem:http://www.foo.com/persistent/bob?query#ref", true}, 1664c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"filesystem:fIle://\\temporary/", "filesystem:file:///temporary/", true}, 1665c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"filesystem:fiLe:///temporary", "filesystem:file:///temporary/", true}, 1666c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"filesystem:File:///temporary/Bob?qUery#reF", "filesystem:file:///temporary/Bob?qUery#reF", true}, 1667c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) }; 1668c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1669c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (size_t i = 0; i < ARRAYSIZE(cases); i++) { 1670c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int url_len = static_cast<int>(strlen(cases[i].input)); 1671c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Parsed parsed; 1672c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::ParseFileSystemURL(cases[i].input, url_len, &parsed); 1673c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1674c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Parsed out_parsed; 1675c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string out_str; 1676c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output(&out_str); 1677c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool success = url_canon::CanonicalizeFileSystemURL(cases[i].input, url_len, 1678c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) parsed, NULL, &output, 1679c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) &out_parsed); 1680c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output.Complete(); 1681c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1682c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(cases[i].expected_success, success); 1683c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(cases[i].expected, out_str); 1684c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1685c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Make sure the spec was properly identified, the filesystem canonicalizer 1686c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // has different code for writing the spec. 1687c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(0, out_parsed.scheme.begin); 1688c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(10, out_parsed.scheme.len); 1689c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (success) 1690c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_GT(out_parsed.path.len, 0); 1691c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 1692c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 1693c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1694c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)TEST(URLCanonTest, CanonicalizePathURL) { 1695c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Path URLs should get canonicalized schemes but nothing else. 1696c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) struct PathCase { 1697c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* input; 1698c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* expected; 1699c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } path_cases[] = { 1700c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"javascript:", "javascript:"}, 1701c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"JavaScript:Foo", "javascript:Foo"}, 1702c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {":\":This /is interesting;?#", ":\":This /is interesting;?#"}, 1703c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) }; 1704c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1705c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (size_t i = 0; i < ARRAYSIZE(path_cases); i++) { 1706c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int url_len = static_cast<int>(strlen(path_cases[i].input)); 1707c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Parsed parsed; 1708c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::ParsePathURL(path_cases[i].input, url_len, &parsed); 1709c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1710c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Parsed out_parsed; 1711c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string out_str; 1712c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output(&out_str); 1713c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool success = url_canon::CanonicalizePathURL(path_cases[i].input, url_len, 1714c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) parsed, &output, 1715c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) &out_parsed); 1716c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output.Complete(); 1717c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1718c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_TRUE(success); 1719c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(path_cases[i].expected, out_str); 1720c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1721c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(0, out_parsed.host.begin); 1722c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(-1, out_parsed.host.len); 1723c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1724c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // When we end with a colon at the end, there should be no path. 1725c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (path_cases[i].input[url_len - 1] == ':') { 1726c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(0, out_parsed.path.begin); 1727c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(-1, out_parsed.path.len); 1728c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 1729c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 1730c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 1731c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1732c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)TEST(URLCanonTest, CanonicalizeMailtoURL) { 1733c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) struct URLCase { 1734c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* input; 1735c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* expected; 1736c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool expected_success; 1737c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component expected_path; 1738c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component expected_query; 1739c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } cases[] = { 1740c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"mailto:addr1", "mailto:addr1", true, url_parse::Component(7, 5), url_parse::Component()}, 1741c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"mailto:addr1@foo.com", "mailto:addr1@foo.com", true, url_parse::Component(7, 13), url_parse::Component()}, 1742c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Trailing whitespace is stripped. 1743c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"MaIlTo:addr1 \t ", "mailto:addr1", true, url_parse::Component(7, 5), url_parse::Component()}, 1744c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"MaIlTo:addr1?to=jon", "mailto:addr1?to=jon", true, url_parse::Component(7, 5), url_parse::Component(13,6)}, 1745c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"mailto:addr1,addr2", "mailto:addr1,addr2", true, url_parse::Component(7, 11), url_parse::Component()}, 1746c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"mailto:addr1, addr2", "mailto:addr1, addr2", true, url_parse::Component(7, 12), url_parse::Component()}, 1747c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"mailto:addr1%2caddr2", "mailto:addr1%2caddr2", true, url_parse::Component(7, 13), url_parse::Component()}, 1748c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"mailto:\xF0\x90\x8C\x80", "mailto:%F0%90%8C%80", true, url_parse::Component(7, 12), url_parse::Component()}, 1749c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Null character should be escaped to %00 1750c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"mailto:addr1\0addr2?foo", "mailto:addr1%00addr2?foo", true, url_parse::Component(7, 13), url_parse::Component(21, 3)}, 1751c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Invalid -- UTF-8 encoded surrogate value. 1752c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"mailto:\xed\xa0\x80", "mailto:%EF%BF%BD", false, url_parse::Component(7, 9), url_parse::Component()}, 1753c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"mailto:addr1?", "mailto:addr1?", true, url_parse::Component(7, 5), url_parse::Component(13, 0)}, 1754c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) }; 1755c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1756c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Define outside of loop to catch bugs where components aren't reset 1757c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Parsed parsed; 1758c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Parsed out_parsed; 1759c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1760c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (size_t i = 0; i < ARRAYSIZE(cases); i++) { 1761c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int url_len = static_cast<int>(strlen(cases[i].input)); 1762c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (i == 8) { 1763c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // The 9th test case purposely has a '\0' in it -- don't count it 1764c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // as the string terminator. 1765c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_len = 22; 1766c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 1767c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::ParseMailtoURL(cases[i].input, url_len, &parsed); 1768c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1769c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string out_str; 1770c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output(&out_str); 1771c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool success = url_canon::CanonicalizeMailtoURL(cases[i].input, url_len, 1772c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) parsed, &output, 1773c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) &out_parsed); 1774c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output.Complete(); 1775c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1776c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(cases[i].expected_success, success); 1777c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(cases[i].expected, out_str); 1778c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1779c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Make sure the spec was properly identified 1780c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(0, out_parsed.scheme.begin); 1781c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(6, out_parsed.scheme.len); 1782c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1783c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(cases[i].expected_path.begin, out_parsed.path.begin); 1784c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(cases[i].expected_path.len, out_parsed.path.len); 1785c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1786c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(cases[i].expected_query.begin, out_parsed.query.begin); 1787c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(cases[i].expected_query.len, out_parsed.query.len); 1788c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 1789c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 1790c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1791c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#ifndef WIN32 1792c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1793c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)TEST(URLCanonTest, _itoa_s) { 1794c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // We fill the buffer with 0xff to ensure that it's getting properly 1795c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // null-terminated. We also allocate one byte more than what we tell 1796c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // _itoa_s about, and ensure that the extra byte is untouched. 1797c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) char buf[6]; 1798c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) memset(buf, 0xff, sizeof(buf)); 1799c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(0, url_canon::_itoa_s(12, buf, sizeof(buf) - 1, 10)); 1800c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_STREQ("12", buf); 1801c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ('\xFF', buf[3]); 1802c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1803c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Test the edge cases - exactly the buffer size and one over 1804c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) memset(buf, 0xff, sizeof(buf)); 1805c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(0, url_canon::_itoa_s(1234, buf, sizeof(buf) - 1, 10)); 1806c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_STREQ("1234", buf); 1807c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ('\xFF', buf[5]); 1808c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1809c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) memset(buf, 0xff, sizeof(buf)); 1810c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(EINVAL, url_canon::_itoa_s(12345, buf, sizeof(buf) - 1, 10)); 1811c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ('\xFF', buf[5]); // should never write to this location 1812c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1813c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Test the template overload (note that this will see the full buffer) 1814c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) memset(buf, 0xff, sizeof(buf)); 1815c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(0, url_canon::_itoa_s(12, buf, 10)); 1816c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_STREQ("12", buf); 1817c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ('\xFF', buf[3]); 1818c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1819c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) memset(buf, 0xff, sizeof(buf)); 1820c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(0, url_canon::_itoa_s(12345, buf, 10)); 1821c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_STREQ("12345", buf); 1822c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1823c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(EINVAL, url_canon::_itoa_s(123456, buf, 10)); 1824c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1825c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Test that radix 16 is supported. 1826c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) memset(buf, 0xff, sizeof(buf)); 1827c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(0, url_canon::_itoa_s(1234, buf, sizeof(buf) - 1, 16)); 1828c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_STREQ("4d2", buf); 1829c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ('\xFF', buf[5]); 1830c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 1831c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1832c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)TEST(URLCanonTest, _itow_s) { 1833c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // We fill the buffer with 0xff to ensure that it's getting properly 1834c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // null-terminated. We also allocate one byte more than what we tell 1835c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // _itoa_s about, and ensure that the extra byte is untouched. 18367d4cd473f85ac64c3747c96c277f9e506a0d2246Torne (Richard Coles) base::char16 buf[6]; 1837c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char fill_mem = 0xff; 18387d4cd473f85ac64c3747c96c277f9e506a0d2246Torne (Richard Coles) const base::char16 fill_char = 0xffff; 1839c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) memset(buf, fill_mem, sizeof(buf)); 1840c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(0, url_canon::_itow_s(12, buf, sizeof(buf) / 2 - 1, 10)); 18417d4cd473f85ac64c3747c96c277f9e506a0d2246Torne (Richard Coles) EXPECT_EQ(WStringToUTF16(L"12"), base::string16(buf)); 1842c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(fill_char, buf[3]); 1843c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1844c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Test the edge cases - exactly the buffer size and one over 1845c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(0, url_canon::_itow_s(1234, buf, sizeof(buf) / 2 - 1, 10)); 18467d4cd473f85ac64c3747c96c277f9e506a0d2246Torne (Richard Coles) EXPECT_EQ(WStringToUTF16(L"1234"), base::string16(buf)); 1847c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(fill_char, buf[5]); 1848c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1849c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) memset(buf, fill_mem, sizeof(buf)); 1850c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(EINVAL, url_canon::_itow_s(12345, buf, sizeof(buf) / 2 - 1, 10)); 1851c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(fill_char, buf[5]); // should never write to this location 1852c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1853c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Test the template overload (note that this will see the full buffer) 1854c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) memset(buf, fill_mem, sizeof(buf)); 1855c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(0, url_canon::_itow_s(12, buf, 10)); 18567d4cd473f85ac64c3747c96c277f9e506a0d2246Torne (Richard Coles) EXPECT_EQ(WStringToUTF16(L"12"), base::string16(buf)); 1857c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(fill_char, buf[3]); 1858c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1859c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) memset(buf, fill_mem, sizeof(buf)); 1860c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(0, url_canon::_itow_s(12345, buf, 10)); 18617d4cd473f85ac64c3747c96c277f9e506a0d2246Torne (Richard Coles) EXPECT_EQ(WStringToUTF16(L"12345"), base::string16(buf)); 1862c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1863c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(EINVAL, url_canon::_itow_s(123456, buf, 10)); 1864c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 1865c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1866c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#endif // !WIN32 1867c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1868c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// Returns true if the given two structures are the same. 1869c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)static bool ParsedIsEqual(const url_parse::Parsed& a, 1870c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const url_parse::Parsed& b) { 1871c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return a.scheme.begin == b.scheme.begin && a.scheme.len == b.scheme.len && 1872c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) a.username.begin == b.username.begin && a.username.len == b.username.len && 1873c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) a.password.begin == b.password.begin && a.password.len == b.password.len && 1874c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) a.host.begin == b.host.begin && a.host.len == b.host.len && 1875c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) a.port.begin == b.port.begin && a.port.len == b.port.len && 1876c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) a.path.begin == b.path.begin && a.path.len == b.path.len && 1877c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) a.query.begin == b.query.begin && a.query.len == b.query.len && 1878c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) a.ref.begin == b.ref.begin && a.ref.len == b.ref.len; 1879c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 1880c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1881c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)TEST(URLCanonTest, ResolveRelativeURL) { 1882c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) struct RelativeCase { 1883c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* base; // Input base URL: MUST BE CANONICAL 1884c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool is_base_hier; // Is the base URL hierarchical 1885c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool is_base_file; // Tells us if the base is a file URL. 1886c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* test; // Input URL to test against. 1887c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool succeed_relative; // Whether we expect IsRelativeURL to succeed 1888c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool is_rel; // Whether we expect |test| to be relative or not. 1889c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool succeed_resolve; // Whether we expect ResolveRelativeURL to succeed. 1890c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* resolved; // What we expect in the result when resolving. 1891c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } rel_cases[] = { 1892c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Basic absolute input. 1893c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a", true, false, "http://another/", true, false, false, NULL}, 1894c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a", true, false, "http:////another/", true, false, false, NULL}, 1895c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Empty relative URLs should only remove the ref part of the URL, 1896c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // leaving the rest unchanged. 1897c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://foo/bar", true, false, "", true, true, true, "http://foo/bar"}, 1898c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://foo/bar#ref", true, false, "", true, true, true, "http://foo/bar"}, 1899c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://foo/bar#", true, false, "", true, true, true, "http://foo/bar"}, 1900c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Spaces at the ends of the relative path should be ignored. 1901c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://foo/bar", true, false, " another ", true, true, true, "http://foo/another"}, 1902c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://foo/bar", true, false, " . ", true, true, true, "http://foo/"}, 1903c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://foo/bar", true, false, " \t ", true, true, true, "http://foo/bar"}, 1904c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Matching schemes without two slashes are treated as relative. 1905c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a", true, false, "http:path", true, true, true, "http://host/path"}, 1906c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a/", true, false, "http:path", true, true, true, "http://host/a/path"}, 1907c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a", true, false, "http:/path", true, true, true, "http://host/path"}, 1908c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a", true, false, "HTTP:/path", true, true, true, "http://host/path"}, 1909c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Nonmatching schemes are absolute. 1910c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a", true, false, "https:host2", true, false, false, NULL}, 1911c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a", true, false, "htto:/host2", true, false, false, NULL}, 1912c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Absolute path input 1913c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a", true, false, "/b/c/d", true, true, true, "http://host/b/c/d"}, 1914c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a", true, false, "\\b\\c\\d", true, true, true, "http://host/b/c/d"}, 1915c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a", true, false, "/b/../c", true, true, true, "http://host/c"}, 1916c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a?b#c", true, false, "/b/../c", true, true, true, "http://host/c"}, 1917c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a", true, false, "\\b/../c?x#y", true, true, true, "http://host/c?x#y"}, 1918c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a?b#c", true, false, "/b/../c?x#y", true, true, true, "http://host/c?x#y"}, 1919c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Relative path input 1920c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a", true, false, "b", true, true, true, "http://host/b"}, 1921c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a", true, false, "bc/de", true, true, true, "http://host/bc/de"}, 1922c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a/", true, false, "bc/de?query#ref", true, true, true, "http://host/a/bc/de?query#ref"}, 1923c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a/", true, false, ".", true, true, true, "http://host/a/"}, 1924c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a/", true, false, "..", true, true, true, "http://host/"}, 1925c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a/", true, false, "./..", true, true, true, "http://host/"}, 1926c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a/", true, false, "../.", true, true, true, "http://host/"}, 1927c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a/", true, false, "././.", true, true, true, "http://host/a/"}, 1928c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a?query#ref", true, false, "../../../foo", true, true, true, "http://host/foo"}, 1929c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Query input 1930c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a", true, false, "?foo=bar", true, true, true, "http://host/a?foo=bar"}, 1931c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a?x=y#z", true, false, "?", true, true, true, "http://host/a?"}, 1932c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a?x=y#z", true, false, "?foo=bar#com", true, true, true, "http://host/a?foo=bar#com"}, 1933c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Ref input 1934c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a", true, false, "#ref", true, true, true, "http://host/a#ref"}, 1935c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a#b", true, false, "#", true, true, true, "http://host/a#"}, 1936c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a?foo=bar#hello", true, false, "#bye", true, true, true, "http://host/a?foo=bar#bye"}, 1937c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Non-hierarchical base: no relative handling. Relative input should 1938c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // error, and if a scheme is present, it should be treated as absolute. 1939c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"data:foobar", false, false, "baz.html", false, false, false, NULL}, 1940c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"data:foobar", false, false, "data:baz", true, false, false, NULL}, 1941c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"data:foobar", false, false, "data:/base", true, false, false, NULL}, 1942c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Non-hierarchical base: absolute input should succeed. 1943c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"data:foobar", false, false, "http://host/", true, false, false, NULL}, 1944c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"data:foobar", false, false, "http:host", true, false, false, NULL}, 1945c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Invalid schemes should be treated as relative. 1946c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://foo/bar", true, false, "./asd:fgh", true, true, true, "http://foo/asd:fgh"}, 1947c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://foo/bar", true, false, ":foo", true, true, true, "http://foo/:foo"}, 1948c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://foo/bar", true, false, " hello world", true, true, true, "http://foo/hello%20world"}, 1949c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"data:asdf", false, false, ":foo", false, false, false, NULL}, 1950c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // We should treat semicolons like any other character in URL resolving 1951c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a", true, false, ";foo", true, true, true, "http://host/;foo"}, 1952c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a;", true, false, ";foo", true, true, true, "http://host/;foo"}, 1953c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a", true, false, ";/../bar", true, true, true, "http://host/bar"}, 1954c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Relative URLs can also be written as "//foo/bar" which is relative to 1955c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // the scheme. In this case, it would take the old scheme, so for http 1956c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // the example would resolve to "http://foo/bar". 1957c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a", true, false, "//another", true, true, true, "http://another/"}, 1958c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a", true, false, "//another/path?query#ref", true, true, true, "http://another/path?query#ref"}, 1959c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a", true, false, "///another/path", true, true, true, "http://another/path"}, 1960c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a", true, false, "//Another\\path", true, true, true, "http://another/path"}, 1961c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a", true, false, "//", true, true, false, "http:"}, 1962c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // IE will also allow one or the other to be a backslash to get the same 1963c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // behavior. 1964c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a", true, false, "\\/another/path", true, true, true, "http://another/path"}, 1965c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a", true, false, "/\\Another\\path", true, true, true, "http://another/path"}, 1966c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#ifdef WIN32 1967c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Resolving against Windows file base URLs. 1968c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file:///C:/foo", true, true, "http://host/", true, false, false, NULL}, 1969c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file:///C:/foo", true, true, "bar", true, true, true, "file:///C:/bar"}, 1970c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file:///C:/foo", true, true, "../../../bar.html", true, true, true, "file:///C:/bar.html"}, 1971c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file:///C:/foo", true, true, "/../bar.html", true, true, true, "file:///C:/bar.html"}, 1972c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // But two backslashes on Windows should be UNC so should be treated 1973c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // as absolute. 1974c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a", true, false, "\\\\another\\path", true, false, false, NULL}, 1975c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // IE doesn't support drive specs starting with two slashes. It fails 1976c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // immediately and doesn't even try to load. We fix it up to either 1977c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // an absolute path or UNC depending on what it looks like. 1978c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file:///C:/something", true, true, "//c:/foo", true, true, true, "file:///C:/foo"}, 1979c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file:///C:/something", true, true, "//localhost/c:/foo", true, true, true, "file:///C:/foo"}, 1980c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Windows drive specs should be allowed and treated as absolute. 1981c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file:///C:/foo", true, true, "c:", true, false, false, NULL}, 1982c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file:///C:/foo", true, true, "c:/foo", true, false, false, NULL}, 1983c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a", true, false, "c:\\foo", true, false, false, NULL}, 1984c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Relative paths with drive letters should be allowed when the base is 1985c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // also a file. 1986c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file:///C:/foo", true, true, "/z:/bar", true, true, true, "file:///Z:/bar"}, 1987c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Treat absolute paths as being off of the drive. 1988c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file:///C:/foo", true, true, "/bar", true, true, true, "file:///C:/bar"}, 1989c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file://localhost/C:/foo", true, true, "/bar", true, true, true, "file://localhost/C:/bar"}, 1990c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file:///C:/foo/com/", true, true, "/bar", true, true, true, "file:///C:/bar"}, 1991c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // On Windows, two slashes without a drive letter when the base is a file 1992c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // means that the path is UNC. 1993c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file:///C:/something", true, true, "//somehost/path", true, true, true, "file://somehost/path"}, 1994c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"file:///C:/something", true, true, "/\\//somehost/path", true, true, true, "file://somehost/path"}, 1995c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#else 1996c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // On Unix we fall back to relative behavior since there's nothing else 1997c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // reasonable to do. 1998c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a", true, false, "\\\\Another\\path", true, true, true, "http://another/path"}, 1999c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#endif 2000c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Even on Windows, we don't allow relative drive specs when the base 2001c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // is not file. 2002c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a", true, false, "/c:\\foo", true, true, true, "http://host/c:/foo"}, 2003c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/a", true, false, "//c:\\foo", true, true, true, "http://c/foo"}, 2004c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Filesystem URL tests; filesystem URLs are only valid and relative if 2005c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // they have no scheme, e.g. "./index.html". There's no valid equivalent 2006c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // to http:index.html. 2007c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"filesystem:http://host/t/path", true, false, "filesystem:http://host/t/path2", true, false, false, NULL}, 2008c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"filesystem:http://host/t/path", true, false, "filesystem:https://host/t/path2", true, false, false, NULL}, 2009c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"filesystem:http://host/t/path", true, false, "http://host/t/path2", true, false, false, NULL}, 2010c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"http://host/t/path", true, false, "filesystem:http://host/t/path2", true, false, false, NULL}, 2011c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"filesystem:http://host/t/path", true, false, "./path2", true, true, true, "filesystem:http://host/t/path2"}, 2012c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"filesystem:http://host/t/path/", true, false, "path2", true, true, true, "filesystem:http://host/t/path/path2"}, 2013c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"filesystem:http://host/t/path", true, false, "filesystem:http:path2", true, false, false, NULL}, 2014c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Absolute URLs are still not relative to a non-standard base URL. 2015c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"about:blank", false, false, "http://X/A", true, false, true, ""}, 2016c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) {"about:blank", false, false, "content://content.Provider/", true, false, true, ""}, 2017c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) }; 2018c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 2019c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (size_t i = 0; i < ARRAYSIZE(rel_cases); i++) { 2020c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const RelativeCase& cur_case = rel_cases[i]; 2021c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 2022c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Parsed parsed; 2023c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int base_len = static_cast<int>(strlen(cur_case.base)); 2024c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (cur_case.is_base_file) 2025c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::ParseFileURL(cur_case.base, base_len, &parsed); 2026c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) else if (cur_case.is_base_hier) 2027c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::ParseStandardURL(cur_case.base, base_len, &parsed); 2028c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) else 2029c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::ParsePathURL(cur_case.base, base_len, &parsed); 2030c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 2031c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // First see if it is relative. 2032c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int test_len = static_cast<int>(strlen(cur_case.test)); 2033c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool is_relative; 2034c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component relative_component; 2035c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool succeed_is_rel = url_canon::IsRelativeURL( 2036c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) cur_case.base, parsed, cur_case.test, test_len, cur_case.is_base_hier, 2037c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) &is_relative, &relative_component); 2038c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 2039c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(cur_case.succeed_relative, succeed_is_rel) << 2040c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) "succeed is rel failure on " << cur_case.test; 2041c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(cur_case.is_rel, is_relative) << 2042c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) "is rel failure on " << cur_case.test; 2043c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Now resolve it. 2044c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (succeed_is_rel && is_relative && cur_case.is_rel) { 2045c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string resolved; 2046c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput output(&resolved); 2047c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Parsed resolved_parsed; 2048c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 2049c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool succeed_resolve = url_canon::ResolveRelativeURL( 2050c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) cur_case.base, parsed, cur_case.is_base_file, 2051c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) cur_case.test, relative_component, NULL, &output, &resolved_parsed); 2052c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output.Complete(); 2053c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 2054c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(cur_case.succeed_resolve, succeed_resolve); 2055c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_EQ(cur_case.resolved, resolved) << " on " << cur_case.test; 2056c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 2057c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Verify that the output parsed structure is the same as parsing a 2058c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // the URL freshly. 2059c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Parsed ref_parsed; 2060c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int resolved_len = static_cast<int>(resolved.size()); 2061c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (cur_case.is_base_file) 2062c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::ParseFileURL(resolved.c_str(), resolved_len, &ref_parsed); 2063c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) else if (cur_case.is_base_hier) 2064c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::ParseStandardURL(resolved.c_str(), resolved_len, &ref_parsed); 2065c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) else 2066c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::ParsePathURL(resolved.c_str(), resolved_len, &ref_parsed); 2067c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_TRUE(ParsedIsEqual(ref_parsed, resolved_parsed)); 2068c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 2069c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 2070c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 2071c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 2072c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// It used to be when we did a replacement with a long buffer of UTF-16 2073c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// characters, we would get invalid data in the URL. This is because the buffer 2074c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// it used to hold the UTF-8 data was resized, while some pointers were still 2075c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// kept to the old buffer that was removed. 2076c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)TEST(URLCanonTest, ReplacementOverflow) { 2077c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char src[] = "file:///C:/foo/bar"; 2078c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int src_len = static_cast<int>(strlen(src)); 2079c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Parsed parsed; 2080c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::ParseFileURL(src, src_len, &parsed); 2081c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 2082c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Override two components, the path with something short, and the query with 2083c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // sonething long enough to trigger the bug. 20847d4cd473f85ac64c3747c96c277f9e506a0d2246Torne (Richard Coles) url_canon::Replacements<base::char16> repl; 20857d4cd473f85ac64c3747c96c277f9e506a0d2246Torne (Richard Coles) base::string16 new_query; 2086c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (int i = 0; i < 4800; i++) 2087c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) new_query.push_back('a'); 2088c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 20897d4cd473f85ac64c3747c96c277f9e506a0d2246Torne (Richard Coles) base::string16 new_path(WStringToUTF16(L"/foo")); 2090c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) repl.SetPath(new_path.c_str(), url_parse::Component(0, 4)); 2091c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) repl.SetQuery(new_query.c_str(), 2092c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Component(0, static_cast<int>(new_query.length()))); 2093c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 2094c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Call ReplaceComponents on the string. It doesn't matter if we call it for 2095c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // standard URLs, file URLs, etc, since they will go to the same replacement 2096c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // function that was buggy. 2097c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_parse::Parsed repl_parsed; 2098c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string repl_str; 2099c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::StdStringCanonOutput repl_output(&repl_str); 2100c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) url_canon::ReplaceFileURL(src, parsed, repl, NULL, &repl_output, &repl_parsed); 2101c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) repl_output.Complete(); 2102c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 2103c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Generate the expected string and check. 2104c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) std::string expected("file:///foo?"); 2105c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (size_t i = 0; i < new_query.length(); i++) 2106c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) expected.push_back('a'); 2107c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) EXPECT_TRUE(expected == repl_str); 2108c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 2109