1ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen// Copyright (c) 2011 The Chromium Authors. All rights reserved. 2c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// Use of this source code is governed by a BSD-style license that can be 3c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// found in the LICENSE file. 4c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 5c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include "net/base/net_util.h" 6c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 73345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick#include <unicode/regex.h> 8c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include <unicode/ucnv.h> 9c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include <unicode/uidna.h> 10c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include <unicode/ulocdata.h> 11c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include <unicode/uniset.h> 12c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include <unicode/uscript.h> 13c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include <unicode/uset.h> 1472a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen#include <algorithm> 15ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen#include <iterator> 1672a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen#include <map> 17c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 18c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include "build/build_config.h" 19c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 20c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#if defined(OS_WIN) 21c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include <windows.h> 22c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include <winsock2.h> 23ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen#include <iphlpapi.h> 24ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen#pragma comment(lib, "iphlpapi.lib") 25c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#elif defined(OS_POSIX) 26c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch#include <fcntl.h> 2700d26a728db2814620f390b418a7d6325ce5aca6Ben Murdoch#ifndef ANDROID 28c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch#include <ifaddrs.h> 2900d26a728db2814620f390b418a7d6325ce5aca6Ben Murdoch#endif 30c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include <netdb.h> 31c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch#include <net/if.h> 32c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch#include <netinet/in.h> 33c77745f79a9f73c4e2d409f8ebc8b031fce6ca75Kristian Monsen#if defined(__BIONIC__) && defined(ANDROID) 34c77745f79a9f73c4e2d409f8ebc8b031fce6ca75Kristian Monsen#include <netinet/in6.h> 35c77745f79a9f73c4e2d409f8ebc8b031fce6ca75Kristian Monsen#endif 36c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#endif 37c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 38c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include "base/base64.h" 39c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include "base/basictypes.h" 40c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include "base/file_path.h" 41c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include "base/file_util.h" 42c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include "base/i18n/file_util_icu.h" 43c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include "base/i18n/icu_string_conversions.h" 44c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include "base/i18n/time_formatting.h" 45c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include "base/json/string_escape.h" 46c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include "base/logging.h" 47ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen#include "base/memory/singleton.h" 48c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include "base/message_loop.h" 49731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick#include "base/metrics/histogram.h" 50c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include "base/path_service.h" 51c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include "base/stl_util-inl.h" 523345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick#include "base/string_number_conversions.h" 53c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include "base/string_piece.h" 543345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick#include "base/string_split.h" 55c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include "base/string_tokenizer.h" 56c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include "base/string_util.h" 573345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick#include "base/stringprintf.h" 5872a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen#include "base/synchronization/lock.h" 59c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include "base/sys_string_conversions.h" 60c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include "base/time.h" 61c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include "base/utf_offset_string_conversions.h" 62c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch#include "base/utf_string_conversions.h" 63c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include "googleurl/src/gurl.h" 64c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include "googleurl/src/url_canon.h" 65c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch#include "googleurl/src/url_canon_ip.h" 66c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include "googleurl/src/url_parse.h" 673345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick#include "grit/net_resources.h" 68c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch#include "net/base/dns_util.h" 69c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include "net/base/escape.h" 70c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include "net/base/net_module.h" 71c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#if defined(OS_WIN) 72c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include "net/base/winsock_init.h" 73c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#endif 74c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include "unicode/datefmt.h" 75c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 76c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottusing base::Time; 77c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 783345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merricknamespace net { 793345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick 80c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottnamespace { 81c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 82c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// what we prepend to get a file URL 83c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottstatic const FilePath::CharType kFileURLPrefix[] = 84c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott FILE_PATH_LITERAL("file:///"); 85c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 86c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// The general list of blocked ports. Will be blocked unless a specific 87c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// protocol overrides it. (Ex: ftp can use ports 20 and 21) 88c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottstatic const int kRestrictedPorts[] = { 89c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1, // tcpmux 90c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 7, // echo 91c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 9, // discard 92c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 11, // systat 93c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 13, // daytime 94c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 15, // netstat 95c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 17, // qotd 96c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 19, // chargen 97c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 20, // ftp data 98c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 21, // ftp access 99c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 22, // ssh 100c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 23, // telnet 101c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 25, // smtp 102c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 37, // time 103c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 42, // name 104c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 43, // nicname 105c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 53, // domain 106c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 77, // priv-rjs 107c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 79, // finger 108c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 87, // ttylink 109c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 95, // supdup 110c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 101, // hostriame 111c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 102, // iso-tsap 112c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 103, // gppitnp 113c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 104, // acr-nema 114c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 109, // pop2 115c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 110, // pop3 116c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 111, // sunrpc 117c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 113, // auth 118c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 115, // sftp 119c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 117, // uucp-path 120c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 119, // nntp 121c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 123, // NTP 122c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 135, // loc-srv /epmap 123c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 139, // netbios 124c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 143, // imap2 125c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 179, // BGP 126c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 389, // ldap 127c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 465, // smtp+ssl 128c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 512, // print / exec 129c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 513, // login 130c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 514, // shell 131c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 515, // printer 132c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 526, // tempo 133c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 530, // courier 134c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 531, // chat 135c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 532, // netnews 136c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 540, // uucp 137c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 556, // remotefs 138c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 563, // nntp+ssl 139c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 587, // stmp? 140c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 601, // ?? 141c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 636, // ldap+ssl 142c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 993, // ldap+ssl 143c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 995, // pop3+ssl 144c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 2049, // nfs 145c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 3659, // apple-sasl / PasswordServer 146c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 4045, // lockd 147c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 6000, // X11 148c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 6665, // Alternate IRC [Apple addition] 149c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 6666, // Alternate IRC [Apple addition] 150c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 6667, // Standard IRC [Apple addition] 151c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 6668, // Alternate IRC [Apple addition] 152c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 6669, // Alternate IRC [Apple addition] 153c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 0xFFFF, // Used to block all invalid port numbers (see 15472a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen // third_party/WebKit/Source/WebCore/platform/KURLGoogle.cpp, port()) 155c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott}; 156c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 157c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// FTP overrides the following restricted ports. 158c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottstatic const int kAllowedFtpPorts[] = { 159c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 21, // ftp data 160c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 22, // ssh 161c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott}; 162c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 163c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scotttemplate<typename STR> 164c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick ScottSTR GetSpecificHeaderT(const STR& headers, const STR& name) { 165c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // We want to grab the Value from the "Key: Value" pairs in the headers, 166c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // which should look like this (no leading spaces, \n-separated) (we format 167c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // them this way in url_request_inet.cc): 168c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // HTTP/1.1 200 OK\n 169c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // ETag: "6d0b8-947-24f35ec0"\n 170c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Content-Length: 2375\n 171c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Content-Type: text/html; charset=UTF-8\n 172c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Last-Modified: Sun, 03 Sep 2006 04:34:43 GMT\n 173c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (headers.empty()) 174c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return STR(); 175c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 176c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott STR match; 177c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott match.push_back('\n'); 178c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott match.append(name); 179c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott match.push_back(':'); 180c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 181c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott typename STR::const_iterator begin = 182c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott search(headers.begin(), headers.end(), match.begin(), match.end(), 183513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch base::CaseInsensitiveCompareASCII<typename STR::value_type>()); 184c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 185c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (begin == headers.end()) 186c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return STR(); 187c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 188c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott begin += match.length(); 189c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 190c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott typename STR::const_iterator end = find(begin, headers.end(), '\n'); 191c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 192c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott STR ret; 193c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott TrimWhitespace(STR(begin, end), TRIM_ALL, &ret); 194c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return ret; 195c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 196c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 197c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// Similar to Base64Decode. Decodes a Q-encoded string to a sequence 198c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// of bytes. If input is invalid, return false. 199c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottbool QPDecode(const std::string& input, std::string* output) { 200c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott std::string temp; 201c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott temp.reserve(input.size()); 202c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott std::string::const_iterator it = input.begin(); 203c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott while (it != input.end()) { 204c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (*it == '_') { 205c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott temp.push_back(' '); 206c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } else if (*it == '=') { 207c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (input.end() - it < 3) { 208c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return false; 209c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 210c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (IsHexDigit(static_cast<unsigned char>(*(it + 1))) && 211c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott IsHexDigit(static_cast<unsigned char>(*(it + 2)))) { 2123345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick unsigned char ch = HexDigitToInt(*(it + 1)) * 16 + 2133345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick HexDigitToInt(*(it + 2)); 214c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott temp.push_back(static_cast<char>(ch)); 215c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott ++it; 216c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott ++it; 217c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } else { 218c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return false; 219c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 220c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } else if (0x20 < *it && *it < 0x7F) { 221c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // In a Q-encoded word, only printable ASCII characters 222c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // represent themselves. Besides, space, '=', '_' and '?' are 223c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // not allowed, but they're already filtered out. 224c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott DCHECK(*it != 0x3D && *it != 0x5F && *it != 0x3F); 225c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott temp.push_back(*it); 226c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } else { 227c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return false; 228c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 229c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott ++it; 230c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 231c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott output->swap(temp); 232c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return true; 233c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 234c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 235c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottenum RFC2047EncodingType {Q_ENCODING, B_ENCODING}; 236c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottbool DecodeBQEncoding(const std::string& part, RFC2047EncodingType enc_type, 237c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott const std::string& charset, std::string* output) { 238c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott std::string decoded; 239c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (enc_type == B_ENCODING) { 240c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (!base::Base64Decode(part, &decoded)) { 241c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return false; 242c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 243c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } else { 244c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (!QPDecode(part, &decoded)) { 245c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return false; 246c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 247c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 248c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 249c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott UErrorCode err = U_ZERO_ERROR; 250c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott UConverter* converter(ucnv_open(charset.c_str(), &err)); 251c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (U_FAILURE(err)) { 252c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return false; 253c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 254c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 255c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // A single byte in a legacy encoding can be expanded to 3 bytes in UTF-8. 256c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // A 'two-byte character' in a legacy encoding can be expanded to 4 bytes 257c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // in UTF-8. Therefore, the expansion ratio is 3 at most. 258c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott int length = static_cast<int>(decoded.length()); 259c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott char* buf = WriteInto(output, length * 3); 260c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott length = ucnv_toAlgorithmic(UCNV_UTF8, converter, buf, length * 3, 261c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott decoded.data(), length, &err); 262c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott ucnv_close(converter); 263c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (U_FAILURE(err)) { 264c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return false; 265c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 266c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott output->resize(length); 267c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return true; 268c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 269c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 270c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottbool DecodeWord(const std::string& encoded_word, 271c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott const std::string& referrer_charset, 272c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch bool* is_rfc2047, 273c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott std::string* output) { 274c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch *is_rfc2047 = false; 275c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch output->clear(); 276c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (encoded_word.empty()) 277c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return true; 278c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 279c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (!IsStringASCII(encoded_word)) { 280c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Try UTF-8, referrer_charset and the native OS default charset in turn. 281c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (IsStringUTF8(encoded_word)) { 282c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott *output = encoded_word; 283c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } else { 284c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott std::wstring wide_output; 285c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (!referrer_charset.empty() && 286c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott base::CodepageToWide(encoded_word, referrer_charset.c_str(), 287c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott base::OnStringConversionError::FAIL, 288c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott &wide_output)) { 289c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott *output = WideToUTF8(wide_output); 290c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } else { 291c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott *output = WideToUTF8(base::SysNativeMBToWide(encoded_word)); 292c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 293c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 294c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 295c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return true; 296c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 297c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 298c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // RFC 2047 : one of encoding methods supported by Firefox and relatively 299c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // widely used by web servers. 300c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // =?charset?<E>?<encoded string>?= where '<E>' is either 'B' or 'Q'. 301c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // We don't care about the length restriction (72 bytes) because 302c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // many web servers generate encoded words longer than the limit. 303c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott std::string tmp; 304c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott *is_rfc2047 = true; 305c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott int part_index = 0; 306c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott std::string charset; 307c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott StringTokenizer t(encoded_word, "?"); 308c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott RFC2047EncodingType enc_type = Q_ENCODING; 309c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott while (*is_rfc2047 && t.GetNext()) { 310c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott std::string part = t.token(); 311c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott switch (part_index) { 312c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott case 0: 313c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (part != "=") { 314c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott *is_rfc2047 = false; 315c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott break; 316c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 317c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott ++part_index; 318c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott break; 319c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott case 1: 320c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Do we need charset validity check here? 321c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott charset = part; 322c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott ++part_index; 323c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott break; 324c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott case 2: 325c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (part.size() > 1 || 326c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott part.find_first_of("bBqQ") == std::string::npos) { 327c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott *is_rfc2047 = false; 328c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott break; 329c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 330c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (part[0] == 'b' || part[0] == 'B') { 331c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott enc_type = B_ENCODING; 332c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 333c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott ++part_index; 334c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott break; 335c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott case 3: 336c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott *is_rfc2047 = DecodeBQEncoding(part, enc_type, charset, &tmp); 337c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (!*is_rfc2047) { 338c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Last minute failure. Invalid B/Q encoding. Rather than 339c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // passing it through, return now. 340c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return false; 341c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 342c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott ++part_index; 343c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott break; 344c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott case 4: 345c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (part != "=") { 346c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Another last minute failure ! 347c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Likely to be a case of two encoded-words in a row or 348c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // an encoded word followed by a non-encoded word. We can be 349c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // generous, but it does not help much in terms of compatibility, 350c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // I believe. Return immediately. 351c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott *is_rfc2047 = false; 352c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return false; 353c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 354c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott ++part_index; 355c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott break; 356c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott default: 357c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott *is_rfc2047 = false; 358c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return false; 359c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 360c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 361c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 362c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (*is_rfc2047) { 363c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (*(encoded_word.end() - 1) == '=') { 364c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott output->swap(tmp); 365c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return true; 366c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 367c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // encoded_word ending prematurelly with '?' or extra '?' 368c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott *is_rfc2047 = false; 369c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return false; 370c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 371c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 372c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // We're not handling 'especial' characters quoted with '\', but 373c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // it should be Ok because we're not an email client but a 374c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // web browser. 375c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 376513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch // What IE6/7 does: %-escaped UTF-8. 377c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott tmp = UnescapeURLComponent(encoded_word, UnescapeRule::SPACES); 378c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (IsStringUTF8(tmp)) { 379c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott output->swap(tmp); 380c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return true; 381c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // We can try either the OS default charset or 'origin charset' here, 382c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // As far as I can tell, IE does not support it. However, I've seen 383c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // web servers emit %-escaped string in a legacy encoding (usually 384c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // origin charset). 385c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // TODO(jungshik) : Test IE further and consider adding a fallback here. 386c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 387c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return false; 388c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 389c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 390c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottbool DecodeParamValue(const std::string& input, 391c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott const std::string& referrer_charset, 392c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott std::string* output) { 393c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott std::string tmp; 394c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Tokenize with whitespace characters. 395c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott StringTokenizer t(input, " \t\n\r"); 396c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott t.set_options(StringTokenizer::RETURN_DELIMS); 397c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott bool is_previous_token_rfc2047 = true; 398c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott while (t.GetNext()) { 399c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (t.token_is_delim()) { 400c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // If the previous non-delimeter token is not RFC2047-encoded, 401c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // put in a space in its place. Otheriwse, skip over it. 402c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (!is_previous_token_rfc2047) { 403c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott tmp.push_back(' '); 404c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 405c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott continue; 406c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 407c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // We don't support a single multibyte character split into 408c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // adjacent encoded words. Some broken mail clients emit headers 409c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // with that problem, but most web servers usually encode a filename 410c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // in a single encoded-word. Firefox/Thunderbird do not support 411c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // it, either. 412c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott std::string decoded; 413c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (!DecodeWord(t.token(), referrer_charset, &is_previous_token_rfc2047, 414c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott &decoded)) 415c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return false; 416c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott tmp.append(decoded); 417c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 418c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott output->swap(tmp); 419c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return true; 420c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 421c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 422c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// TODO(mpcomplete): This is a quick and dirty implementation for now. I'm 423c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// sure this doesn't properly handle all (most?) cases. 424c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scotttemplate<typename STR> 425513209b27ff55e2841eac0e4120199c23acce758Ben MurdochSTR GetHeaderParamValueT(const STR& header, const STR& param_name, 426513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch QuoteRule::Type quote_rule) { 427c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // This assumes args are formatted exactly like "bla; arg1=value; arg2=value". 428c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott typename STR::const_iterator param_begin = 429c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott search(header.begin(), header.end(), param_name.begin(), param_name.end(), 430513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch base::CaseInsensitiveCompareASCII<typename STR::value_type>()); 431c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 432c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (param_begin == header.end()) 433c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return STR(); 434c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott param_begin += param_name.length(); 435c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 436c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott STR whitespace; 437c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott whitespace.push_back(' '); 438c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott whitespace.push_back('\t'); 439c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott const typename STR::size_type equals_offset = 440c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott header.find_first_not_of(whitespace, param_begin - header.begin()); 441c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (equals_offset == STR::npos || header.at(equals_offset) != '=') 442c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return STR(); 443c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 444c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott param_begin = header.begin() + equals_offset + 1; 445c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (param_begin == header.end()) 446c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return STR(); 447c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 448c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott typename STR::const_iterator param_end; 449513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch if (*param_begin == '"' && quote_rule == QuoteRule::REMOVE_OUTER_QUOTES) { 450c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott ++param_begin; // skip past the quote. 451ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen param_end = find(param_begin, header.end(), '"'); 452ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // If the closing quote is missing, we will treat the rest of the 453ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // string as the parameter. We can't set |param_end| to the 454ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // location of the separator (';'), since the separator is 455ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // technically quoted. See: http://crbug.com/58840 456c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } else { 457c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott param_end = find(param_begin+1, header.end(), ';'); 458c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 459c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 460c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return STR(param_begin, param_end); 461c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 462c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 463c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// Does some simple normalization of scripts so we can allow certain scripts 464c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// to exist together. 465c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// TODO(brettw) bug 880223: we should allow some other languages to be 466c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// oombined such as Chinese and Latin. We will probably need a more 467c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// complicated system of language pairs to have more fine-grained control. 468c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick ScottUScriptCode NormalizeScript(UScriptCode code) { 469c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott switch (code) { 470c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott case USCRIPT_KATAKANA: 471c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott case USCRIPT_HIRAGANA: 472c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott case USCRIPT_KATAKANA_OR_HIRAGANA: 473c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott case USCRIPT_HANGUL: // This one is arguable. 474c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return USCRIPT_HAN; 475c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott default: 476c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return code; 477c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 478c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 479c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 480c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottbool IsIDNComponentInSingleScript(const char16* str, int str_len) { 481c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott UScriptCode first_script = USCRIPT_INVALID_CODE; 482c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott bool is_first = true; 483c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 484c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott int i = 0; 485c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott while (i < str_len) { 486c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott unsigned code_point; 487c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott U16_NEXT(str, i, str_len, code_point); 488c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 489c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott UErrorCode err = U_ZERO_ERROR; 490c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott UScriptCode cur_script = uscript_getScript(code_point, &err); 491c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (err != U_ZERO_ERROR) 492c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return false; // Report mixed on error. 493c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott cur_script = NormalizeScript(cur_script); 494c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 495c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // TODO(brettw) We may have to check for USCRIPT_INHERENT as well. 496c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (is_first && cur_script != USCRIPT_COMMON) { 497c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott first_script = cur_script; 498c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott is_first = false; 499c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } else { 500c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (cur_script != USCRIPT_COMMON && cur_script != first_script) 501c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return false; 502c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 503c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 504c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return true; 505c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 506c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 507c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// Check if the script of a language can be 'safely' mixed with 508c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// Latin letters in the ASCII range. 509c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottbool IsCompatibleWithASCIILetters(const std::string& lang) { 510c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // For now, just list Chinese, Japanese and Korean (positive list). 511c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // An alternative is negative-listing (languages using Greek and 512c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Cyrillic letters), but it can be more dangerous. 513c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return !lang.substr(0, 2).compare("zh") || 514c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott !lang.substr(0, 2).compare("ja") || 515c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott !lang.substr(0, 2).compare("ko"); 516c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 517c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 518c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scotttypedef std::map<std::string, icu::UnicodeSet*> LangToExemplarSetMap; 519c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 520c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottclass LangToExemplarSet { 52121d179b334e59e9a3bfcaed4c4430bef1bc5759dKristian Monsen public: 52221d179b334e59e9a3bfcaed4c4430bef1bc5759dKristian Monsen static LangToExemplarSet* GetInstance() { 52321d179b334e59e9a3bfcaed4c4430bef1bc5759dKristian Monsen return Singleton<LangToExemplarSet>::get(); 52421d179b334e59e9a3bfcaed4c4430bef1bc5759dKristian Monsen } 52521d179b334e59e9a3bfcaed4c4430bef1bc5759dKristian Monsen 526c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott private: 527c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott LangToExemplarSetMap map; 528c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott LangToExemplarSet() { } 529c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott ~LangToExemplarSet() { 530c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott STLDeleteContainerPairSecondPointers(map.begin(), map.end()); 531c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 532c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 533c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott friend class Singleton<LangToExemplarSet>; 534c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott friend struct DefaultSingletonTraits<LangToExemplarSet>; 535c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott friend bool GetExemplarSetForLang(const std::string&, icu::UnicodeSet**); 536c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott friend void SetExemplarSetForLang(const std::string&, icu::UnicodeSet*); 537c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 538c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott DISALLOW_COPY_AND_ASSIGN(LangToExemplarSet); 539c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott}; 540c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 541c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottbool GetExemplarSetForLang(const std::string& lang, 542c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott icu::UnicodeSet** lang_set) { 54321d179b334e59e9a3bfcaed4c4430bef1bc5759dKristian Monsen const LangToExemplarSetMap& map = LangToExemplarSet::GetInstance()->map; 544c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott LangToExemplarSetMap::const_iterator pos = map.find(lang); 545c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (pos != map.end()) { 546c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott *lang_set = pos->second; 547c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return true; 548c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 549c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return false; 550c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 551c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 552c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottvoid SetExemplarSetForLang(const std::string& lang, 553c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott icu::UnicodeSet* lang_set) { 55421d179b334e59e9a3bfcaed4c4430bef1bc5759dKristian Monsen LangToExemplarSetMap& map = LangToExemplarSet::GetInstance()->map; 555c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott map.insert(std::make_pair(lang, lang_set)); 556c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 557c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 55872a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsenstatic base::Lock lang_set_lock; 559c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 560c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// Returns true if all the characters in component_characters are used by 561c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// the language |lang|. 562c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottbool IsComponentCoveredByLang(const icu::UnicodeSet& component_characters, 563c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott const std::string& lang) { 564c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott static const icu::UnicodeSet kASCIILetters(0x61, 0x7a); // [a-z] 565c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott icu::UnicodeSet* lang_set; 566c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // We're called from both the UI thread and the history thread. 567c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott { 56872a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen base::AutoLock lock(lang_set_lock); 569c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (!GetExemplarSetForLang(lang, &lang_set)) { 570c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott UErrorCode status = U_ZERO_ERROR; 571c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott ULocaleData* uld = ulocdata_open(lang.c_str(), &status); 572c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // TODO(jungshik) Turn this check on when the ICU data file is 573c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // rebuilt with the minimal subset of locale data for languages 574c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // to which Chrome is not localized but which we offer in the list 575c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // of languages selectable for Accept-Languages. With the rebuilt ICU 576c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // data, ulocdata_open never should fall back to the default locale. 577c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // (issue 2078) 578c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // DCHECK(U_SUCCESS(status) && status != U_USING_DEFAULT_WARNING); 579c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (U_SUCCESS(status) && status != U_USING_DEFAULT_WARNING) { 580c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott lang_set = reinterpret_cast<icu::UnicodeSet *>( 581c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott ulocdata_getExemplarSet(uld, NULL, 0, 582c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott ULOCDATA_ES_STANDARD, &status)); 583c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // If |lang| is compatible with ASCII Latin letters, add them. 584c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (IsCompatibleWithASCIILetters(lang)) 585c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott lang_set->addAll(kASCIILetters); 586c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } else { 587c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott lang_set = new icu::UnicodeSet(1, 0); 588c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 589c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott lang_set->freeze(); 590c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott SetExemplarSetForLang(lang, lang_set); 591c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott ulocdata_close(uld); 592c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 593c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 594c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return !lang_set->isEmpty() && lang_set->containsAll(component_characters); 595c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 596c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 597c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// Returns true if the given Unicode host component is safe to display to the 598c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// user. 599c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottbool IsIDNComponentSafe(const char16* str, 600c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott int str_len, 601c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott const std::wstring& languages) { 602c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Most common cases (non-IDN) do not reach here so that we don't 603c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // need a fast return path. 604c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // TODO(jungshik) : Check if there's any character inappropriate 605c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // (although allowed) for domain names. 606c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // See http://www.unicode.org/reports/tr39/#IDN_Security_Profiles and 607c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // http://www.unicode.org/reports/tr39/data/xidmodifications.txt 608c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // For now, we borrow the list from Mozilla and tweaked it slightly. 609c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // (e.g. Characters like U+00A0, U+3000, U+3002 are omitted because 610c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // they're gonna be canonicalized to U+0020 and full stop before 611c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // reaching here.) 612c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // The original list is available at 613c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // http://kb.mozillazine.org/Network.IDN.blacklist_chars and 614c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // at http://mxr.mozilla.org/seamonkey/source/modules/libpref/src/init/all.js#703 615c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 616c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott UErrorCode status = U_ZERO_ERROR; 617c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#ifdef U_WCHAR_IS_UTF16 618c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott icu::UnicodeSet dangerous_characters(icu::UnicodeString( 619c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott L"[[\\ \u00bc\u00bd\u01c3\u0337\u0338" 620c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott L"\u05c3\u05f4\u06d4\u0702\u115f\u1160][\u2000-\u200b]" 621c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott L"[\u2024\u2027\u2028\u2029\u2039\u203a\u2044\u205f]" 622c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott L"[\u2154-\u2156][\u2159-\u215b][\u215f\u2215\u23ae" 623c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott L"\u29f6\u29f8\u2afb\u2afd][\u2ff0-\u2ffb][\u3014" 624c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott L"\u3015\u3033\u3164\u321d\u321e\u33ae\u33af\u33c6\u33df\ufe14" 625c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott L"\ufe15\ufe3f\ufe5d\ufe5e\ufeff\uff0e\uff06\uff61\uffa0\ufff9]" 626c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott L"[\ufffa-\ufffd]]"), status); 6273345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick DCHECK(U_SUCCESS(status)); 6283345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick icu::RegexMatcher dangerous_patterns(icu::UnicodeString( 6293345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick // Lone katakana no, so, or n 6303345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick L"[^\\p{Katakana}][\u30ce\u30f3\u30bd][^\\p{Katakana}]" 6313345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick // Repeating Japanese accent characters 6323345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick L"|[\u3099\u309a\u309b\u309c][\u3099\u309a\u309b\u309c]"), 6333345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick 0, status); 634c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#else 635c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott icu::UnicodeSet dangerous_characters(icu::UnicodeString( 636c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott "[[\\u0020\\u00bc\\u00bd\\u01c3\\u0337\\u0338" 637c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott "\\u05c3\\u05f4\\u06d4\\u0702\\u115f\\u1160][\\u2000-\\u200b]" 638c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott "[\\u2024\\u2027\\u2028\\u2029\\u2039\\u203a\\u2044\\u205f]" 639c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott "[\\u2154-\\u2156][\\u2159-\\u215b][\\u215f\\u2215\\u23ae" 640c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott "\\u29f6\\u29f8\\u2afb\\u2afd][\\u2ff0-\\u2ffb][\\u3014" 641c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott "\\u3015\\u3033\\u3164\\u321d\\u321e\\u33ae\\u33af\\u33c6\\u33df\\ufe14" 642c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott "\\ufe15\\ufe3f\\ufe5d\\ufe5e\\ufeff\\uff0e\\uff06\\uff61\\uffa0\\ufff9]" 643c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott "[\\ufffa-\\ufffd]]", -1, US_INV), status); 6443345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick DCHECK(U_SUCCESS(status)); 6453345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick icu::RegexMatcher dangerous_patterns(icu::UnicodeString( 6463345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick // Lone katakana no, so, or n 6473345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick "[^\\p{Katakana}][\\u30ce\\u30f3\u30bd][^\\p{Katakana}]" 6483345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick // Repeating Japanese accent characters 6493345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick "|[\\u3099\\u309a\\u309b\\u309c][\\u3099\\u309a\\u309b\\u309c]"), 6503345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick 0, status); 651c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#endif 652c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott DCHECK(U_SUCCESS(status)); 653c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott icu::UnicodeSet component_characters; 6543345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick icu::UnicodeString component_string(str, str_len); 6553345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick component_characters.addAll(component_string); 656c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (dangerous_characters.containsSome(component_characters)) 657c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return false; 658c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 6593345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick DCHECK(U_SUCCESS(status)); 6603345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick dangerous_patterns.reset(component_string); 6613345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick if (dangerous_patterns.find()) 6623345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick return false; 6633345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick 664c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // If the language list is empty, the result is completely determined 665c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // by whether a component is a single script or not. This will block 666c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // even "safe" script mixing cases like <Chinese, Latin-ASCII> that are 667c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // allowed with |languages| (while it blocks Chinese + Latin letters with 668c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // an accent as should be the case), but we want to err on the safe side 669c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // when |languages| is empty. 670c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (languages.empty()) 671c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return IsIDNComponentInSingleScript(str, str_len); 672c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 673c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // |common_characters| is made up of ASCII numbers, hyphen, plus and 674c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // underscore that are used across scripts and allowed in domain names. 675c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // (sync'd with characters allowed in url_canon_host with square 676c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // brackets excluded.) See kHostCharLookup[] array in url_canon_host.cc. 677c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott icu::UnicodeSet common_characters(UNICODE_STRING_SIMPLE("[[0-9]\\-_+\\ ]"), 678c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott status); 679c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott DCHECK(U_SUCCESS(status)); 680c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Subtract common characters because they're always allowed so that 681c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // we just have to check if a language-specific set contains 682c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // the remainder. 683c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott component_characters.removeAll(common_characters); 684c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 685c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott std::string languages_list(WideToASCII(languages)); 686c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott StringTokenizer t(languages_list, ","); 687c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott while (t.GetNext()) { 688c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (IsComponentCoveredByLang(component_characters, t.token())) 689c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return true; 690c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 691c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return false; 692c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 693c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 694c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// Converts one component of a host (between dots) to IDN if safe. The result 695c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// will be APPENDED to the given output string and will be the same as the input 696c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// if it is not IDN or the IDN is unsafe to display. Returns whether any 697c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// conversion was performed. 698c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottbool IDNToUnicodeOneComponent(const char16* comp, 699c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott size_t comp_len, 700c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott const std::wstring& languages, 701c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott string16* out) { 702c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott DCHECK(out); 703c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (comp_len == 0) 704c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return false; 705c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 706c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Only transform if the input can be an IDN component. 707c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott static const char16 kIdnPrefix[] = {'x', 'n', '-', '-'}; 708c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if ((comp_len > arraysize(kIdnPrefix)) && 709c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott !memcmp(comp, kIdnPrefix, arraysize(kIdnPrefix) * sizeof(char16))) { 710c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Repeatedly expand the output string until it's big enough. It looks like 711c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // ICU will return the required size of the buffer, but that's not 712c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // documented, so we'll just grow by 2x. This should be rare and is not on a 713c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // critical path. 714c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott size_t original_length = out->length(); 715c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott for (int extra_space = 64; ; extra_space *= 2) { 716c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott UErrorCode status = U_ZERO_ERROR; 717c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott out->resize(out->length() + extra_space); 718c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott int output_chars = uidna_IDNToUnicode(comp, 719c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott static_cast<int32_t>(comp_len), &(*out)[original_length], extra_space, 720c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott UIDNA_DEFAULT, NULL, &status); 721c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (status == U_ZERO_ERROR) { 722c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Converted successfully. 723c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott out->resize(original_length + output_chars); 724c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (IsIDNComponentSafe(out->data() + original_length, output_chars, 725c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott languages)) 726c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return true; 727c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 728c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 729c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (status != U_BUFFER_OVERFLOW_ERROR) 730c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott break; 731c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 732c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Failed, revert back to original string. 733c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott out->resize(original_length); 734c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 735c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 736c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // We get here with no IDN or on error, in which case we just append the 737c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // literal input. 738c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott out->append(comp, comp_len); 739c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return false; 740c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 741c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 742ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsenstruct SubtractFromOffset { 743ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen explicit SubtractFromOffset(size_t amount) 744ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen : amount(amount) {} 745ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen void operator()(size_t& offset) { 746ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (offset != std::wstring::npos) { 747ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (offset >= amount) 748ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen offset -= amount; 749ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen else 750ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen offset = std::wstring::npos; 751ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } 752ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } 753ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 754ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen size_t amount; 755ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen}; 756ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 757ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsenstruct AddToOffset { 758ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen explicit AddToOffset(size_t amount) 759ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen : amount(amount) {} 760ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen void operator()(size_t& offset) { 761ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (offset != std::wstring::npos) 762ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen offset += amount; 763ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } 764ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 765ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen size_t amount; 766ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen}; 767ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 768ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsenstd::vector<size_t> OffsetsIntoSection( 769ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::vector<size_t>* offsets_for_adjustment, 770ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen size_t section_begin) { 771ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::vector<size_t> offsets_into_section; 772ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (offsets_for_adjustment) { 773ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::transform(offsets_for_adjustment->begin(), 774ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen offsets_for_adjustment->end(), 775ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::back_inserter(offsets_into_section), 776ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen ClampComponentOffset(section_begin)); 777ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::for_each(offsets_into_section.begin(), offsets_into_section.end(), 778ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen SubtractFromOffset(section_begin)); 779ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } 780ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen return offsets_into_section; 781ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen} 782ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 783ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsenvoid ApplySectionAdjustments(const std::vector<size_t>& offsets_into_section, 784ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::vector<size_t>* offsets_for_adjustment, 785ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen size_t old_section_len, 786ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen size_t new_section_len, 787ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen size_t section_begin) { 788ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (offsets_for_adjustment) { 789ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen DCHECK_EQ(offsets_for_adjustment->size(), offsets_into_section.size()); 790ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::vector<size_t>::const_iterator host_offsets_iter = 791ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen offsets_into_section.begin(); 792ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen for (std::vector<size_t>::iterator offsets_iter = 793ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen offsets_for_adjustment->begin(); 794ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen offsets_iter != offsets_for_adjustment->end(); 795ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen ++offsets_iter, ++host_offsets_iter) { 796ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen size_t offset = *offsets_iter; 797ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (offset == std::wstring::npos || offset < section_begin) { 798ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // The offset is before the host section so leave it as is. 799ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen continue; 800ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } 801ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (offset >= section_begin + old_section_len) { 802ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // The offset is after the host section so adjust by host length delta. 803ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen offset += new_section_len - old_section_len; 804ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } else if (*host_offsets_iter != std::wstring::npos) { 805ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // The offset is within the host and valid so adjust by the host 806ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // reformatting offsets results. 807ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen offset = section_begin + *host_offsets_iter; 808ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } else { 809ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // The offset is invalid. 810ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen offset = std::wstring::npos; 811ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } 812ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen *offsets_iter = offset; 813ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } 814ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } 815ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen} 816ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 817c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch// If |component| is valid, its begin is incremented by |delta|. 818c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdochvoid AdjustComponent(int delta, url_parse::Component* component) { 819c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (!component->is_valid()) 820c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return; 821c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 822c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch DCHECK(delta >= 0 || component->begin >= -delta); 823c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch component->begin += delta; 824c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch} 825c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 826c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch// Adjusts all the components of |parsed| by |delta|, except for the scheme. 827c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdochvoid AdjustComponents(int delta, url_parse::Parsed* parsed) { 828c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch AdjustComponent(delta, &(parsed->username)); 829c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch AdjustComponent(delta, &(parsed->password)); 830c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch AdjustComponent(delta, &(parsed->host)); 831c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch AdjustComponent(delta, &(parsed->port)); 832c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch AdjustComponent(delta, &(parsed->path)); 833c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch AdjustComponent(delta, &(parsed->query)); 834c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch AdjustComponent(delta, &(parsed->ref)); 835c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch} 836c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 8373345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrickstd::wstring FormatUrlInternal(const GURL& url, 8383345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick const std::wstring& languages, 8393345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick FormatUrlTypes format_types, 8403345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick UnescapeRule::Type unescape_rules, 8413345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick url_parse::Parsed* new_parsed, 8423345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick size_t* prefix_end, 843ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::vector<size_t>* offsets_for_adjustment); 8443345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick 8453345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick// Helper for FormatUrl()/FormatUrlInternal(). 846c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottstd::wstring FormatViewSourceUrl(const GURL& url, 847c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott const std::wstring& languages, 848ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen FormatUrlTypes format_types, 849c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott UnescapeRule::Type unescape_rules, 850c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott url_parse::Parsed* new_parsed, 851c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott size_t* prefix_end, 852ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::vector<size_t>* offsets_for_adjustment) { 853c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott DCHECK(new_parsed); 854ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen DCHECK(offsets_for_adjustment); 855c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott const wchar_t* const kWideViewSource = L"view-source:"; 856c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott const size_t kViewSourceLengthPlus1 = 12; 857ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::vector<size_t> saved_offsets(*offsets_for_adjustment); 858c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 859c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott GURL real_url(url.possibly_invalid_spec().substr(kViewSourceLengthPlus1)); 860ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // Clamp the offsets to the source area. 861ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::for_each(offsets_for_adjustment->begin(), 862ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen offsets_for_adjustment->end(), 863ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen SubtractFromOffset(kViewSourceLengthPlus1)); 8643345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick std::wstring result = FormatUrlInternal(real_url, languages, format_types, 865ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen unescape_rules, new_parsed, prefix_end, offsets_for_adjustment); 866c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott result.insert(0, kWideViewSource); 867c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 868c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Adjust position values. 869c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (new_parsed->scheme.is_nonempty()) { 870c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Assume "view-source:real-scheme" as a scheme. 871c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott new_parsed->scheme.len += kViewSourceLengthPlus1; 872c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } else { 873c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott new_parsed->scheme.begin = 0; 874c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott new_parsed->scheme.len = kViewSourceLengthPlus1 - 1; 875c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 876c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch AdjustComponents(kViewSourceLengthPlus1, new_parsed); 877c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (prefix_end) 878c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott *prefix_end += kViewSourceLengthPlus1; 879ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::for_each(offsets_for_adjustment->begin(), 880ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen offsets_for_adjustment->end(), 881ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen AddToOffset(kViewSourceLengthPlus1)); 882ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // Restore all offsets which were not affected by FormatUrlInternal. 883ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen DCHECK_EQ(saved_offsets.size(), offsets_for_adjustment->size()); 884ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen for (size_t i = 0; i < saved_offsets.size(); ++i) { 885ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (saved_offsets[i] < kViewSourceLengthPlus1) 886ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen (*offsets_for_adjustment)[i] = saved_offsets[i]; 887c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 888c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return result; 889c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 890c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 891c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// Appends the substring |in_component| inside of the URL |spec| to |output|, 892c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// and the resulting range will be filled into |out_component|. |unescape_rules| 893ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen// defines how to clean the URL for human readability. |offsets_for_adjustment| 894ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen// is an array of offsets into |output| each of which will be adjusted based on 895ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen// how it maps to the component being converted; if it is less than 896ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen// output->length(), it will be untouched, and if it is greater than 897ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen// output->length() + in_component.len it will be adjusted by the difference in 898ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen// lengths between the input and output components. Otherwise it points into 899ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen// the component being converted, and is adjusted to point to the same logical 900ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen// place in |output|. |offsets_for_adjustment| may not be NULL. 9013345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrickvoid AppendFormattedComponent(const std::string& spec, 9023345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick const url_parse::Component& in_component, 9033345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick UnescapeRule::Type unescape_rules, 9043345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick std::wstring* output, 9053345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick url_parse::Component* out_component, 906ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::vector<size_t>* offsets_for_adjustment) { 9073345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick DCHECK(output); 908ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen DCHECK(offsets_for_adjustment); 9093345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick if (in_component.is_nonempty()) { 910ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen size_t component_begin = output->length(); 911ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen out_component->begin = static_cast<int>(component_begin); 912ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 913ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // Compose a list of offsets within the component area. 914ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::vector<size_t> offsets_into_component = 915ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen OffsetsIntoSection(offsets_for_adjustment, component_begin); 916ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 9173345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick if (unescape_rules == UnescapeRule::NONE) { 918ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen output->append(UTF8ToWideAndAdjustOffsets( 9193345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick spec.substr(in_component.begin, in_component.len), 920ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen &offsets_into_component)); 9213345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick } else { 922ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen output->append(UTF16ToWideHack( 923ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen UnescapeAndDecodeUTF8URLComponentWithOffsets( 924ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen spec.substr(in_component.begin, in_component.len), unescape_rules, 925ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen &offsets_into_component))); 9263345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick } 927ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen size_t new_component_len = output->length() - component_begin; 928ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen out_component->len = static_cast<int>(new_component_len); 929ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 930ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // Apply offset adjustments. 931ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen size_t old_component_len = static_cast<size_t>(in_component.len); 932ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen ApplySectionAdjustments(offsets_into_component, offsets_for_adjustment, 933ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen old_component_len, new_component_len, component_begin); 9343345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick } else { 9353345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick out_component->reset(); 9363345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick } 9373345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick} 9383345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick 9393345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick// TODO(viettrungluu): This is really the old-fashioned version, made internal. 9403345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick// I need to really convert |FormatUrl()|. 9413345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrickstd::wstring FormatUrlInternal(const GURL& url, 9423345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick const std::wstring& languages, 9433345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick FormatUrlTypes format_types, 9443345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick UnescapeRule::Type unescape_rules, 9453345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick url_parse::Parsed* new_parsed, 9463345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick size_t* prefix_end, 947ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::vector<size_t>* offsets_for_adjustment) { 9483345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick url_parse::Parsed parsed_temp; 9493345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick if (!new_parsed) 9503345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick new_parsed = &parsed_temp; 9513345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick else 9523345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick *new_parsed = url_parse::Parsed(); 953ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 954ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::vector<size_t> offsets_temp; 955ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (!offsets_for_adjustment) 956ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen offsets_for_adjustment = &offsets_temp; 9573345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick 9583345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick std::wstring url_string; 9593345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick 9603345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick // Check for empty URLs or 0 available text width. 9613345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick if (url.is_empty()) { 9623345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick if (prefix_end) 9633345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick *prefix_end = 0; 964ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::for_each(offsets_for_adjustment->begin(), 965ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen offsets_for_adjustment->end(), 966ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen LimitOffset<std::wstring>(0)); 9673345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick return url_string; 9683345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick } 9693345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick 9703345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick // Special handling for view-source:. Don't use chrome::kViewSourceScheme 9713345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick // because this library shouldn't depend on chrome. 9723345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick const char* const kViewSource = "view-source"; 9733345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick // Reject "view-source:view-source:..." to avoid deep recursion. 9743345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick const char* const kViewSourceTwice = "view-source:view-source:"; 9753345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick if (url.SchemeIs(kViewSource) && 9763345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick !StartsWithASCII(url.possibly_invalid_spec(), kViewSourceTwice, false)) { 9773345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick return FormatViewSourceUrl(url, languages, format_types, 978ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen unescape_rules, new_parsed, prefix_end, offsets_for_adjustment); 9793345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick } 9803345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick 9813345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick // We handle both valid and invalid URLs (this will give us the spec 9823345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick // regardless of validity). 9833345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick const std::string& spec = url.possibly_invalid_spec(); 9843345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick const url_parse::Parsed& parsed = url.parsed_for_possibly_invalid_spec(); 985ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen size_t spec_length = spec.length(); 986ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::for_each(offsets_for_adjustment->begin(), 987ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen offsets_for_adjustment->end(), 988ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen LimitOffset<std::wstring>(spec_length)); 9893345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick 9903345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick // Copy everything before the username (the scheme and the separators.) 9913345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick // These are ASCII. 99272a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen url_string.insert(url_string.end(), spec.begin(), 9933345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick spec.begin() + parsed.CountCharactersBefore(url_parse::Parsed::USERNAME, 99472a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen true)); 9953345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick 9963345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick const wchar_t kHTTP[] = L"http://"; 9973345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick const char kFTP[] = "ftp."; 9983345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick // URLFixerUpper::FixupURL() treats "ftp.foo.com" as ftp://ftp.foo.com. This 9993345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick // means that if we trim "http://" off a URL whose host starts with "ftp." and 10003345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick // the user inputs this into any field subject to fixup (which is basically 10013345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick // all input fields), the meaning would be changed. (In fact, often the 10023345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick // formatted URL is directly pre-filled into an input field.) For this reason 10033345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick // we avoid stripping "http://" in this case. 10043345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick bool omit_http = 10053345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick (format_types & kFormatUrlOmitHTTP) && (url_string == kHTTP) && 10063345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick (url.host().compare(0, arraysize(kFTP) - 1, kFTP) != 0); 10073345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick 10083345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick new_parsed->scheme = parsed.scheme; 10093345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick 10103345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick if ((format_types & kFormatUrlOmitUsernamePassword) != 0) { 10113345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick // Remove the username and password fields. We don't want to display those 10123345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick // to the user since they can be used for attacks, 10133345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick // e.g. "http://google.com:search@evil.ru/" 10143345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick new_parsed->username.reset(); 10153345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick new_parsed->password.reset(); 1016ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // Update the offsets based on removed username and/or password. 1017ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (!offsets_for_adjustment->empty() && 10183345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick (parsed.username.is_nonempty() || parsed.password.is_nonempty())) { 1019ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen AdjustOffset::Adjustments adjustments; 10203345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick if (parsed.username.is_nonempty() && parsed.password.is_nonempty()) { 10213345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick // The seeming off-by-one and off-by-two in these first two lines are to 10223345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick // account for the ':' after the username and '@' after the password. 1023ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen adjustments.push_back(AdjustOffset::Adjustment( 1024ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen static_cast<size_t>(parsed.username.begin), 1025ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen static_cast<size_t>(parsed.username.len + parsed.password.len + 1026ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 2), 0)); 10273345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick } else { 10283345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick const url_parse::Component* nonempty_component = 10293345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick parsed.username.is_nonempty() ? &parsed.username : &parsed.password; 1030ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // The seeming off-by-one in below is to account for the '@' after the 1031ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // username/password. 1032ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen adjustments.push_back(AdjustOffset::Adjustment( 1033ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen static_cast<size_t>(nonempty_component->begin), 1034ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen static_cast<size_t>(nonempty_component->len + 1), 0)); 10353345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick } 1036ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 1037ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // Make offset adjustment. 1038ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::for_each(offsets_for_adjustment->begin(), 1039ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen offsets_for_adjustment->end(), 1040ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen AdjustOffset(adjustments)); 10413345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick } 10423345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick } else { 10433345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick AppendFormattedComponent(spec, parsed.username, unescape_rules, &url_string, 1044ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen &new_parsed->username, offsets_for_adjustment); 10453345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick if (parsed.password.is_valid()) 10463345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick url_string.push_back(':'); 10473345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick AppendFormattedComponent(spec, parsed.password, unescape_rules, &url_string, 1048ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen &new_parsed->password, offsets_for_adjustment); 10493345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick if (parsed.username.is_valid() || parsed.password.is_valid()) 10503345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick url_string.push_back('@'); 10513345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick } 10523345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick if (prefix_end) 10533345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick *prefix_end = static_cast<size_t>(url_string.length()); 10543345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick 1055ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen AppendFormattedHostWithOffsets(url, languages, &url_string, new_parsed, 1056ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen offsets_for_adjustment); 10573345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick 10583345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick // Port. 10593345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick if (parsed.port.is_nonempty()) { 10603345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick url_string.push_back(':'); 10613345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick new_parsed->port.begin = url_string.length(); 106272a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen url_string.insert(url_string.end(), 106372a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen spec.begin() + parsed.port.begin, 106472a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen spec.begin() + parsed.port.end()); 10653345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick new_parsed->port.len = url_string.length() - new_parsed->port.begin; 10663345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick } else { 10673345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick new_parsed->port.reset(); 10683345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick } 10693345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick 10703345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick // Path and query both get the same general unescape & convert treatment. 10713345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick if (!(format_types & kFormatUrlOmitTrailingSlashOnBareHostname) || 10723345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick !CanStripTrailingSlash(url)) { 10733345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick AppendFormattedComponent(spec, parsed.path, unescape_rules, &url_string, 1074ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen &new_parsed->path, offsets_for_adjustment); 10753345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick } 10763345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick if (parsed.query.is_valid()) 10773345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick url_string.push_back('?'); 10783345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick AppendFormattedComponent(spec, parsed.query, unescape_rules, &url_string, 1079ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen &new_parsed->query, offsets_for_adjustment); 10803345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick 10813345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick // Reference is stored in valid, unescaped UTF-8, so we can just convert. 10823345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick if (parsed.ref.is_valid()) { 10833345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick url_string.push_back('#'); 1084ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen size_t ref_begin = url_string.length(); 1085ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen new_parsed->ref.begin = static_cast<int>(ref_begin); 1086ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 1087ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // Compose a list of offsets within the section. 1088ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::vector<size_t> offsets_into_ref = 1089ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen OffsetsIntoSection(offsets_for_adjustment, ref_begin); 1090ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 10913345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick if (parsed.ref.len > 0) { 1092ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen url_string.append(UTF8ToWideAndAdjustOffsets(spec.substr(parsed.ref.begin, 1093ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen parsed.ref.len), 1094ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen &offsets_into_ref)); 10953345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick } 1096ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen size_t old_ref_len = static_cast<size_t>(parsed.ref.len); 1097ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen size_t new_ref_len = url_string.length() - new_parsed->ref.begin; 1098ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen new_parsed->ref.len = static_cast<int>(new_ref_len); 1099ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 1100ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // Apply offset adjustments. 1101ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen ApplySectionAdjustments(offsets_into_ref, offsets_for_adjustment, 1102ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen old_ref_len, new_ref_len, ref_begin); 11033345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick } 11043345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick 11053345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick // If we need to strip out http do it after the fact. This way we don't need 11063345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick // to worry about how offset_for_adjustment is interpreted. 11073345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick const size_t kHTTPSize = arraysize(kHTTP) - 1; 11083345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick if (omit_http && !url_string.compare(0, kHTTPSize, kHTTP)) { 11093345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick url_string = url_string.substr(kHTTPSize); 1110ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen AdjustOffset::Adjustments adjustments; 1111ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen adjustments.push_back(AdjustOffset::Adjustment(0, kHTTPSize, 0)); 1112ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::for_each(offsets_for_adjustment->begin(), 1113ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen offsets_for_adjustment->end(), 1114ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen AdjustOffset(adjustments)); 11153345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick if (prefix_end) 11163345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick *prefix_end -= kHTTPSize; 11173345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick 11183345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick // Adjust new_parsed. 11193345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick DCHECK(new_parsed->scheme.is_valid()); 11203345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick int delta = -(new_parsed->scheme.len + 3); // +3 for ://. 11213345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick new_parsed->scheme.reset(); 11223345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick AdjustComponents(delta, new_parsed); 11233345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick } 11243345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick 11253345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick return url_string; 11263345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick} 11273345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick 1128ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsenchar* do_strdup(const char* src) { 1129ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen#if defined(OS_WIN) 1130ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen return _strdup(src); 1131ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen#else 1132ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen return strdup(src); 1133ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen#endif 1134ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen} 1135ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 11363345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick} // namespace 11373345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick 11383345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrickconst FormatUrlType kFormatUrlOmitNothing = 0; 11393345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrickconst FormatUrlType kFormatUrlOmitUsernamePassword = 1 << 0; 11403345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrickconst FormatUrlType kFormatUrlOmitHTTP = 1 << 1; 11413345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrickconst FormatUrlType kFormatUrlOmitTrailingSlashOnBareHostname = 1 << 2; 11423345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrickconst FormatUrlType kFormatUrlOmitAll = kFormatUrlOmitUsernamePassword | 11433345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick kFormatUrlOmitHTTP | kFormatUrlOmitTrailingSlashOnBareHostname; 11443345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick 11453345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick// TODO(viettrungluu): We don't want non-POD globals; change this. 114621d179b334e59e9a3bfcaed4c4430bef1bc5759dKristian Monsenstd::multiset<int> explicitly_allowed_ports; 1147c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1148c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick ScottGURL FilePathToFileURL(const FilePath& path) { 1149c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Produce a URL like "file:///C:/foo" for a regular file, or 1150c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // "file://///server/path" for UNC. The URL canonicalizer will fix up the 1151c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // latter case to be the canonical UNC form: "file://server/path" 1152c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott FilePath::StringType url_string(kFileURLPrefix); 1153c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott url_string.append(path.value()); 1154c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1155c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Now do replacement of some characters. Since we assume the input is a 1156c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // literal filename, anything the URL parser might consider special should 1157c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // be escaped here. 1158c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1159c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // must be the first substitution since others will introduce percents as the 1160c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // escape character 1161c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott ReplaceSubstringsAfterOffset(&url_string, 0, 1162c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott FILE_PATH_LITERAL("%"), FILE_PATH_LITERAL("%25")); 1163c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1164c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // semicolon is supposed to be some kind of separator according to RFC 2396 1165c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott ReplaceSubstringsAfterOffset(&url_string, 0, 1166c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott FILE_PATH_LITERAL(";"), FILE_PATH_LITERAL("%3B")); 1167c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1168c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott ReplaceSubstringsAfterOffset(&url_string, 0, 1169c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott FILE_PATH_LITERAL("#"), FILE_PATH_LITERAL("%23")); 1170c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1171c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#if defined(OS_POSIX) 1172c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott ReplaceSubstringsAfterOffset(&url_string, 0, 1173c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott FILE_PATH_LITERAL("\\"), FILE_PATH_LITERAL("%5C")); 1174c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#endif 1175c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1176c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return GURL(url_string); 1177c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 1178c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1179c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottstd::wstring GetSpecificHeader(const std::wstring& headers, 1180c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott const std::wstring& name) { 1181c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return GetSpecificHeaderT(headers, name); 1182c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 1183c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1184c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottstd::string GetSpecificHeader(const std::string& headers, 1185c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott const std::string& name) { 1186c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return GetSpecificHeaderT(headers, name); 1187c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 1188c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1189513209b27ff55e2841eac0e4120199c23acce758Ben Murdochbool DecodeCharset(const std::string& input, 1190513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch std::string* decoded_charset, 1191513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch std::string* value) { 1192513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch StringTokenizer t(input, "'"); 1193513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch t.set_options(StringTokenizer::RETURN_DELIMS); 1194513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch std::string temp_charset; 1195513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch std::string temp_value; 1196513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch int numDelimsSeen = 0; 1197513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch while (t.GetNext()) { 1198513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch if (t.token_is_delim()) { 1199513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch ++numDelimsSeen; 1200513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch continue; 1201513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch } else { 1202513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch switch (numDelimsSeen) { 1203513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch case 0: 1204513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch temp_charset = t.token(); 1205513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch break; 1206513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch case 1: 1207513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch // Language is ignored. 1208513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch break; 1209513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch case 2: 1210513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch temp_value = t.token(); 1211513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch break; 1212513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch default: 1213513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch return false; 1214513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch } 1215513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch } 1216513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch } 1217513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch if (numDelimsSeen != 2) 1218513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch return false; 1219513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch if (temp_charset.empty() || temp_value.empty()) 1220513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch return false; 1221513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch decoded_charset->swap(temp_charset); 1222513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch value->swap(temp_value); 1223513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch return true; 1224513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch} 1225513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch 1226c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottstd::string GetFileNameFromCD(const std::string& header, 1227c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott const std::string& referrer_charset) { 1228513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch std::string decoded; 1229513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch std::string param_value = GetHeaderParamValue(header, "filename*", 1230513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch QuoteRule::KEEP_OUTER_QUOTES); 1231513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch if (!param_value.empty()) { 1232513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch if (param_value.find('"') == std::string::npos) { 1233513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch std::string charset; 1234513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch std::string value; 1235513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch if (DecodeCharset(param_value, &charset, &value)) { 1236513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch // RFC 5987 value should be ASCII-only. 1237513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch if (!IsStringASCII(value)) 1238513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch return std::string(); 123921d179b334e59e9a3bfcaed4c4430bef1bc5759dKristian Monsen std::string tmp = UnescapeURLComponent( 124021d179b334e59e9a3bfcaed4c4430bef1bc5759dKristian Monsen value, 124121d179b334e59e9a3bfcaed4c4430bef1bc5759dKristian Monsen UnescapeRule::SPACES | UnescapeRule::URL_SPECIAL_CHARS); 1242513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch if (base::ConvertToUtf8AndNormalize(tmp, charset, &decoded)) 1243513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch return decoded; 1244513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch } 1245513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch } 1246513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch } 1247513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch param_value = GetHeaderParamValue(header, "filename", 1248513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch QuoteRule::REMOVE_OUTER_QUOTES); 1249c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (param_value.empty()) { 1250c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Some servers use 'name' parameter. 1251513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch param_value = GetHeaderParamValue(header, "name", 1252513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch QuoteRule::REMOVE_OUTER_QUOTES); 1253c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 1254c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (param_value.empty()) 1255c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return std::string(); 1256c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (DecodeParamValue(param_value, referrer_charset, &decoded)) 1257c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return decoded; 1258c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return std::string(); 1259c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 1260c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1261c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottstd::wstring GetHeaderParamValue(const std::wstring& field, 1262513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch const std::wstring& param_name, 1263513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch QuoteRule::Type quote_rule) { 1264513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch return GetHeaderParamValueT(field, param_name, quote_rule); 1265c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 1266c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1267c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottstd::string GetHeaderParamValue(const std::string& field, 1268513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch const std::string& param_name, 1269513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch QuoteRule::Type quote_rule) { 1270513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch return GetHeaderParamValueT(field, param_name, quote_rule); 1271c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 1272c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1273c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// TODO(brettw) bug 734373: check the scripts for each host component and 1274c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// don't un-IDN-ize if there is more than one. Alternatively, only IDN for 1275c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// scripts that the user has installed. For now, just put the entire 1276c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// path through IDN. Maybe this feature can be implemented in ICU itself? 1277c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// 1278c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// We may want to skip this step in the case of file URLs to allow unicode 1279c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// UNC hostnames regardless of encodings. 1280ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsenstd::wstring IDNToUnicodeWithOffsets( 1281ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen const char* host, 1282ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen size_t host_len, 1283ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen const std::wstring& languages, 1284ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::vector<size_t>* offsets_for_adjustment) { 1285c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Convert the ASCII input to a wide string for ICU. 1286c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott string16 input16; 1287c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott input16.reserve(host_len); 128872a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen input16.insert(input16.end(), host, host + host_len); 1289c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1290c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Do each component of the host separately, since we enforce script matching 1291c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // on a per-component basis. 1292ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen AdjustOffset::Adjustments adjustments; 1293ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen string16 out16; 1294c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott for (size_t component_start = 0, component_end; 1295c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott component_start < input16.length(); 1296c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott component_start = component_end + 1) { 1297c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Find the end of the component. 1298c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott component_end = input16.find('.', component_start); 1299c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (component_end == string16::npos) 1300c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott component_end = input16.length(); // For getting the last component. 1301c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott size_t component_length = component_end - component_start; 1302ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen size_t new_component_start = out16.length(); 1303c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott bool converted_idn = false; 1304c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (component_end > component_start) { 1305c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Add the substring that we just found. 1306c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott converted_idn = IDNToUnicodeOneComponent(input16.data() + component_start, 1307c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott component_length, languages, &out16); 1308c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 1309ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen size_t new_component_length = out16.length() - new_component_start; 1310c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1311ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (converted_idn && offsets_for_adjustment) { 1312ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen adjustments.push_back(AdjustOffset::Adjustment( 1313ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen component_start, component_length, new_component_length)); 1314c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 1315c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1316c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Need to add the dot we just found (if we found one). 1317c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (component_end < input16.length()) 1318c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott out16.push_back('.'); 1319c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 1320c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1321ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // Make offset adjustment. 1322ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (offsets_for_adjustment && !adjustments.empty()) { 1323ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::for_each(offsets_for_adjustment->begin(), 1324ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen offsets_for_adjustment->end(), 1325ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen AdjustOffset(adjustments)); 1326ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } 1327c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1328ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen return UTF16ToWideAndAdjustOffsets(out16, offsets_for_adjustment); 1329ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen} 1330ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 1331ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsenstd::wstring IDNToUnicode(const char* host, 1332ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen size_t host_len, 1333ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen const std::wstring& languages, 1334ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen size_t* offset_for_adjustment) { 1335ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::vector<size_t> offsets; 1336ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (offset_for_adjustment) 1337ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen offsets.push_back(*offset_for_adjustment); 1338ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::wstring result = 1339ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen IDNToUnicodeWithOffsets(host, host_len, languages, &offsets); 1340ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (offset_for_adjustment) 1341ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen *offset_for_adjustment = offsets[0]; 1342ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen return result; 1343c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 1344c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1345c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottstd::string CanonicalizeHost(const std::string& host, 1346c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott url_canon::CanonHostInfo* host_info) { 1347c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Try to canonicalize the host. 1348c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott const url_parse::Component raw_host_component( 1349c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 0, static_cast<int>(host.length())); 1350c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott std::string canon_host; 1351c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott url_canon::StdStringCanonOutput canon_host_output(&canon_host); 1352c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott url_canon::CanonicalizeHostVerbose(host.c_str(), raw_host_component, 1353c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott &canon_host_output, host_info); 1354c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1355c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (host_info->out_host.is_nonempty() && 1356c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott host_info->family != url_canon::CanonHostInfo::BROKEN) { 1357c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Success! Assert that there's no extra garbage. 1358c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott canon_host_output.Complete(); 1359c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott DCHECK_EQ(host_info->out_host.len, static_cast<int>(canon_host.length())); 1360c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } else { 1361c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Empty host, or canonicalization failed. We'll return empty. 1362c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott canon_host.clear(); 1363c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 1364c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1365c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return canon_host; 1366c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 1367c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1368c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottstd::string CanonicalizeHost(const std::wstring& host, 1369c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott url_canon::CanonHostInfo* host_info) { 1370c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott std::string converted_host; 1371c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott WideToUTF8(host.c_str(), host.length(), &converted_host); 1372c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return CanonicalizeHost(converted_host, host_info); 1373c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 1374c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1375c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottstd::string GetDirectoryListingHeader(const string16& title) { 1376c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott static const base::StringPiece header( 1377c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott NetModule::GetResource(IDR_DIR_HEADER_HTML)); 1378c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // This can be null in unit tests. 1379c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott DLOG_IF(WARNING, header.empty()) << 1380c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott "Missing resource: directory listing header"; 1381c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1382c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott std::string result; 1383c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (!header.empty()) 1384c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott result.assign(header.data(), header.size()); 1385c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1386c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott result.append("<script>start("); 1387c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott base::JsonDoubleQuote(title, true, &result); 1388c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott result.append(");</script>\n"); 1389c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1390c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return result; 1391c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 1392c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1393c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottinline bool IsHostCharAlpha(char c) { 1394c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // We can just check lowercase because uppercase characters have already been 1395c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // normalized. 1396c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return (c >= 'a') && (c <= 'z'); 1397c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 1398c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1399c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottinline bool IsHostCharDigit(char c) { 1400c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return (c >= '0') && (c <= '9'); 1401c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 1402c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1403c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdochbool IsCanonicalizedHostCompliant(const std::string& host, 1404c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch const std::string& desired_tld) { 1405c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (host.empty()) 1406c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return false; 1407c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1408c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott bool in_component = false; 1409c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott bool most_recent_component_started_alpha = false; 1410c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott bool last_char_was_hyphen_or_underscore = false; 1411c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1412c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott for (std::string::const_iterator i(host.begin()); i != host.end(); ++i) { 1413c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott const char c = *i; 1414c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (!in_component) { 1415c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott most_recent_component_started_alpha = IsHostCharAlpha(c); 1416c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (!most_recent_component_started_alpha && !IsHostCharDigit(c)) 1417c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return false; 1418c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott in_component = true; 1419c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } else { 1420c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (c == '.') { 1421c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (last_char_was_hyphen_or_underscore) 1422c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return false; 1423c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott in_component = false; 1424c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } else if (IsHostCharAlpha(c) || IsHostCharDigit(c)) { 1425c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott last_char_was_hyphen_or_underscore = false; 1426c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } else if ((c == '-') || (c == '_')) { 1427c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott last_char_was_hyphen_or_underscore = true; 1428c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } else { 1429c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return false; 1430c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 1431c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 1432c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 1433c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1434c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return most_recent_component_started_alpha || 1435c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch (!desired_tld.empty() && IsHostCharAlpha(desired_tld[0])); 1436c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 1437c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1438c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottstd::string GetDirectoryListingEntry(const string16& name, 1439c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott const std::string& raw_bytes, 1440c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott bool is_dir, 1441c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott int64 size, 1442c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott Time modified) { 1443c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott std::string result; 1444c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott result.append("<script>addRow("); 1445c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott base::JsonDoubleQuote(name, true, &result); 1446c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott result.append(","); 1447c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (raw_bytes.empty()) { 1448c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott base::JsonDoubleQuote(EscapePath(UTF16ToUTF8(name)), 1449c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott true, &result); 1450c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } else { 1451c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott base::JsonDoubleQuote(EscapePath(raw_bytes), true, &result); 1452c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 1453c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (is_dir) { 1454c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott result.append(",1,"); 1455c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } else { 1456c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott result.append(",0,"); 1457c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 1458c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1459c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott base::JsonDoubleQuote( 14603345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick FormatBytes(size, GetByteDisplayUnits(size), true), 14613345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick true, 1462c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott &result); 1463c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1464c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott result.append(","); 1465c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1466c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott string16 modified_str; 1467c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // |modified| can be NULL in FTP listings. 1468c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (!modified.is_null()) { 146921d179b334e59e9a3bfcaed4c4430bef1bc5759dKristian Monsen modified_str = base::TimeFormatShortDateAndTime(modified); 1470c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 1471c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott base::JsonDoubleQuote(modified_str, true, &result); 1472c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1473c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott result.append(");</script>\n"); 1474c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1475c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return result; 1476c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 1477c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 14783345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrickstring16 StripWWW(const string16& text) { 14793345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick const string16 www(ASCIIToUTF16("www.")); 1480ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen return StartsWith(text, www, true) ? text.substr(www.length()) : text; 1481c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 1482c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 148372a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsenstring16 GetSuggestedFilename(const GURL& url, 1484c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott const std::string& content_disposition, 1485c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott const std::string& referrer_charset, 148672a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen const string16& default_name) { 148772a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen // TODO: this function to be updated to match the httpbis recommendations. 148872a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen // Talk to abarth for the latest news. 148972a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen 1490c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // We don't translate this fallback string, "download". If localization is 1491c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // needed, the caller should provide localized fallback default_name. 149272a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen static const char* kFinalFallbackName = "download"; 1493c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1494c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // about: and data: URLs don't have file names, but esp. data: URLs may 1495c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // contain parts that look like ones (i.e., contain a slash). 1496c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Therefore we don't attempt to divine a file name out of them. 1497c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (url.SchemeIs("about") || url.SchemeIs("data")) { 149872a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen return default_name.empty() ? ASCIIToUTF16(kFinalFallbackName) 149972a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen : default_name; 1500c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 1501c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 150272a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen std::string filename = GetFileNameFromCD(content_disposition, 150372a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen referrer_charset); 1504c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1505c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (!filename.empty()) { 150672a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen // Replace any path information the server may have sent, by changing 150772a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen // path separators with underscores. 150872a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen ReplaceSubstringsAfterOffset(&filename, 0, "/", "_"); 150972a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen ReplaceSubstringsAfterOffset(&filename, 0, "\\", "_"); 1510c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1511c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Next, remove "." from the beginning and end of the file name to avoid 1512c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // tricks with hidden files, "..", and "." 151372a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen TrimString(filename, ".", &filename); 1514c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 1515c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (filename.empty()) { 1516c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (url.is_valid()) { 1517c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott const std::string unescaped_url_filename = UnescapeURLComponent( 1518c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott url.ExtractFileName(), 1519c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott UnescapeRule::SPACES | UnescapeRule::URL_SPECIAL_CHARS); 1520c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 1521c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // The URL's path should be escaped UTF-8, but may not be. 1522c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch std::string decoded_filename = unescaped_url_filename; 1523c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (!IsStringASCII(decoded_filename)) { 1524c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch bool ignore; 1525c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // TODO(jshin): this is probably not robust enough. To be sure, we 1526c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // need encoding detection. 1527c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch DecodeWord(unescaped_url_filename, referrer_charset, &ignore, 1528c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch &decoded_filename); 1529c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch } 1530c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 1531c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch filename = decoded_filename; 1532c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 1533c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 1534c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1535c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch#if defined(OS_WIN) 1536c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch { // Handle CreateFile() stripping trailing dots and spaces on filenames 1537c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // http://support.microsoft.com/kb/115827 153872a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen std::string::size_type pos = filename.find_last_not_of(" ."); 1539c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (pos == std::string::npos) 1540c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch filename.resize(0); 1541c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch else 1542c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch filename.resize(++pos); 1543c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch } 1544c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch#endif 1545c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Trim '.' once more. 154672a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen TrimString(filename, ".", &filename); 1547c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1548c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // If there's no filename or it gets trimed to be empty, use 1549c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // the URL hostname or default_name 1550c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (filename.empty()) { 1551c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (!default_name.empty()) { 155272a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen return default_name; 1553c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } else if (url.is_valid()) { 1554c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Some schemes (e.g. file) do not have a hostname. Even though it's 1555c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // not likely to reach here, let's hardcode the last fallback name. 1556c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // TODO(jungshik) : Decode a 'punycoded' IDN hostname. (bug 1264451) 155772a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen filename = url.host().empty() ? kFinalFallbackName : url.host(); 1558c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } else { 1559c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott NOTREACHED(); 1560c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 1561c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 1562c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 156372a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen#if defined(OS_WIN) 156472a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen string16 path = UTF8ToUTF16(filename); 156572a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen file_util::ReplaceIllegalCharactersInPath(&path, '-'); 156672a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen return path; 156772a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen#else 156872a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen std::string path = filename; 156972a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen file_util::ReplaceIllegalCharactersInPath(&path, '-'); 157072a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen return UTF8ToUTF16(path); 157172a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen#endif 1572c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 1573c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1574c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottbool IsPortAllowedByDefault(int port) { 1575c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott int array_size = arraysize(kRestrictedPorts); 1576c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott for (int i = 0; i < array_size; i++) { 1577c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (kRestrictedPorts[i] == port) { 1578c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return false; 1579c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 1580c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 1581c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return true; 1582c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 1583c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1584c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottbool IsPortAllowedByFtp(int port) { 1585c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott int array_size = arraysize(kAllowedFtpPorts); 1586c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott for (int i = 0; i < array_size; i++) { 1587c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (kAllowedFtpPorts[i] == port) { 1588c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return true; 1589c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 1590c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 1591c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Port not explicitly allowed by FTP, so return the default restrictions. 1592c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return IsPortAllowedByDefault(port); 1593c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 1594c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1595c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottbool IsPortAllowedByOverride(int port) { 1596c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (explicitly_allowed_ports.empty()) 1597c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return false; 1598c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 159921d179b334e59e9a3bfcaed4c4430bef1bc5759dKristian Monsen return explicitly_allowed_ports.count(port) > 0; 1600c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 1601c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1602c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottint SetNonBlocking(int fd) { 1603c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#if defined(OS_WIN) 1604c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott unsigned long no_block = 1; 1605c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return ioctlsocket(fd, FIONBIO, &no_block); 1606c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#elif defined(OS_POSIX) 1607c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott int flags = fcntl(fd, F_GETFL, 0); 1608c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (-1 == flags) 1609c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return flags; 1610c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return fcntl(fd, F_SETFL, flags | O_NONBLOCK); 1611c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#endif 1612c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 1613c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1614c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottbool ParseHostAndPort(std::string::const_iterator host_and_port_begin, 1615c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott std::string::const_iterator host_and_port_end, 1616c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott std::string* host, 1617c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott int* port) { 1618c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (host_and_port_begin >= host_and_port_end) 1619c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return false; 1620c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1621c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // When using url_parse, we use char*. 1622c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott const char* auth_begin = &(*host_and_port_begin); 1623c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott int auth_len = host_and_port_end - host_and_port_begin; 1624c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1625c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott url_parse::Component auth_component(0, auth_len); 1626c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott url_parse::Component username_component; 1627c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott url_parse::Component password_component; 1628c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott url_parse::Component hostname_component; 1629c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott url_parse::Component port_component; 1630c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1631c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott url_parse::ParseAuthority(auth_begin, auth_component, &username_component, 1632c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott &password_component, &hostname_component, &port_component); 1633c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1634c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // There shouldn't be a username/password. 1635c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (username_component.is_valid() || password_component.is_valid()) 1636c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return false; 1637c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1638c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (!hostname_component.is_nonempty()) 1639c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return false; // Failed parsing. 1640c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1641c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott int parsed_port_number = -1; 1642c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (port_component.is_nonempty()) { 1643c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott parsed_port_number = url_parse::ParsePort(auth_begin, port_component); 1644c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1645c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // If parsing failed, port_number will be either PORT_INVALID or 1646c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // PORT_UNSPECIFIED, both of which are negative. 1647c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (parsed_port_number < 0) 1648c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return false; // Failed parsing the port number. 1649c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 1650c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1651c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (port_component.len == 0) 1652c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return false; // Reject inputs like "foo:" 1653c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1654c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Pass results back to caller. 1655c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott host->assign(auth_begin + hostname_component.begin, hostname_component.len); 1656c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott *port = parsed_port_number; 1657c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1658c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return true; // Success. 1659c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 1660c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1661c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottbool ParseHostAndPort(const std::string& host_and_port, 1662c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott std::string* host, 1663c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott int* port) { 1664c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return ParseHostAndPort( 1665c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott host_and_port.begin(), host_and_port.end(), host, port); 1666c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 1667c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1668c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottstd::string GetHostAndPort(const GURL& url) { 1669c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // For IPv6 literals, GURL::host() already includes the brackets so it is 1670c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // safe to just append a colon. 16713345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick return base::StringPrintf("%s:%d", url.host().c_str(), 16723345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick url.EffectiveIntPort()); 1673c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 1674c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1675c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottstd::string GetHostAndOptionalPort(const GURL& url) { 1676c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // For IPv6 literals, GURL::host() already includes the brackets 1677c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // so it is safe to just append a colon. 1678c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (url.has_port()) 16793345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick return base::StringPrintf("%s:%s", url.host().c_str(), url.port().c_str()); 1680c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return url.host(); 1681c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 1682c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1683c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottstd::string NetAddressToString(const struct addrinfo* net_address) { 168472a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen return NetAddressToString(net_address->ai_addr, net_address->ai_addrlen); 168572a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen} 168672a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen 168772a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsenstd::string NetAddressToString(const struct sockaddr* net_address, 168872a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen socklen_t address_len) { 1689c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#if defined(OS_WIN) 1690c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott EnsureWinsockInit(); 1691c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#endif 1692c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1693c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // This buffer is large enough to fit the biggest IPv6 string. 1694c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott char buffer[INET6_ADDRSTRLEN]; 1695c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 169672a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen int result = getnameinfo(net_address, address_len, buffer, sizeof(buffer), 169772a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen NULL, 0, NI_NUMERICHOST); 1698c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1699c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (result != 0) { 170072a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen DVLOG(1) << "getnameinfo() failed with " << result << ": " 170172a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen << gai_strerror(result); 1702c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott buffer[0] = '\0'; 1703c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 1704c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return std::string(buffer); 1705c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 1706c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1707c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdochstd::string NetAddressToStringWithPort(const struct addrinfo* net_address) { 170872a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen return NetAddressToStringWithPort( 170972a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen net_address->ai_addr, net_address->ai_addrlen); 171072a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen} 171172a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsenstd::string NetAddressToStringWithPort(const struct sockaddr* net_address, 171272a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen socklen_t address_len) { 171372a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen std::string ip_address_string = NetAddressToString(net_address, address_len); 1714c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (ip_address_string.empty()) 1715c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return std::string(); // Failed. 1716c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 171772a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen int port = GetPortFromSockaddr(net_address, address_len); 1718c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 1719c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (ip_address_string.find(':') != std::string::npos) { 1720c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // Surround with square brackets to avoid ambiguity. 17213345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick return base::StringPrintf("[%s]:%d", ip_address_string.c_str(), port); 1722c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch } 1723c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 17243345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick return base::StringPrintf("%s:%d", ip_address_string.c_str(), port); 1725c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch} 1726c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 1727c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottstd::string GetHostName() { 1728c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#if defined(OS_WIN) 1729c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott EnsureWinsockInit(); 1730c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#endif 1731c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1732c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Host names are limited to 255 bytes. 1733c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott char buffer[256]; 1734c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott int result = gethostname(buffer, sizeof(buffer)); 1735c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (result != 0) { 1736731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick DVLOG(1) << "gethostname() failed with " << result; 1737c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott buffer[0] = '\0'; 1738c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 1739c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return std::string(buffer); 1740c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 1741c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1742c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottvoid GetIdentityFromURL(const GURL& url, 17433345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick string16* username, 17443345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick string16* password) { 1745c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch UnescapeRule::Type flags = 1746c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch UnescapeRule::SPACES | UnescapeRule::URL_SPECIAL_CHARS; 17473345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick *username = UnescapeAndDecodeUTF8URLComponent(url.username(), flags, NULL); 17483345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick *password = UnescapeAndDecodeUTF8URLComponent(url.password(), flags, NULL); 1749c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 1750c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1751c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdochstd::string GetHostOrSpecFromURL(const GURL& url) { 1752ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen return url.has_host() ? TrimEndingDot(url.host()) : url.spec(); 1753c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch} 1754c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 1755ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsenvoid AppendFormattedHostWithOffsets( 1756ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen const GURL& url, 1757ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen const std::wstring& languages, 1758ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::wstring* output, 1759ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen url_parse::Parsed* new_parsed, 1760ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::vector<size_t>* offsets_for_adjustment) { 1761c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott DCHECK(output); 1762c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott const url_parse::Component& host = 1763c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott url.parsed_for_possibly_invalid_spec().host; 1764c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1765c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (host.is_nonempty()) { 1766c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Handle possible IDN in the host name. 1767ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen size_t host_begin = output->length(); 1768c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (new_parsed) 1769ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen new_parsed->host.begin = static_cast<int>(host_begin); 1770ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen size_t old_host_len = static_cast<size_t>(host.len); 1771ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 1772ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // Compose a list of offsets within the host area. 1773ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::vector<size_t> offsets_into_host = 1774ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen OffsetsIntoSection(offsets_for_adjustment, host_begin); 1775c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1776c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott const std::string& spec = url.possibly_invalid_spec(); 1777c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott DCHECK(host.begin >= 0 && 1778c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott ((spec.length() == 0 && host.begin == 0) || 1779c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott host.begin < static_cast<int>(spec.length()))); 1780ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen output->append(IDNToUnicodeWithOffsets(&spec[host.begin], old_host_len, 1781ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen languages, &offsets_into_host)); 1782c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1783ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen size_t new_host_len = output->length() - host_begin; 1784c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (new_parsed) 1785ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen new_parsed->host.len = static_cast<int>(new_host_len); 1786ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 1787ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // Apply offset adjustments. 1788ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen ApplySectionAdjustments(offsets_into_host, offsets_for_adjustment, 1789ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen old_host_len, new_host_len, host_begin); 1790c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } else if (new_parsed) { 1791c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott new_parsed->host.reset(); 1792c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 1793c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 1794c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1795ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsenvoid AppendFormattedHost(const GURL& url, 1796ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen const std::wstring& languages, 1797ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::wstring* output, 1798ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen url_parse::Parsed* new_parsed, 1799ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen size_t* offset_for_adjustment) { 1800ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::vector<size_t> offsets; 1801ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (offset_for_adjustment) 1802ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen offsets.push_back(*offset_for_adjustment); 1803ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen AppendFormattedHostWithOffsets(url, languages, output, new_parsed, &offsets); 1804ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (offset_for_adjustment) 1805ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen *offset_for_adjustment = offsets[0]; 1806ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen} 1807ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 18083345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick// TODO(viettrungluu): convert the wstring |FormatUrlInternal()|. 1809ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsenstring16 FormatUrlWithOffsets(const GURL& url, 1810ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen const std::string& languages, 1811ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen FormatUrlTypes format_types, 1812ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen UnescapeRule::Type unescape_rules, 1813ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen url_parse::Parsed* new_parsed, 1814ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen size_t* prefix_end, 1815ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::vector<size_t>* offsets_for_adjustment) { 1816ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen return WideToUTF16Hack( 1817ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen FormatUrlInternal(url, ASCIIToWide(languages), format_types, 1818ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen unescape_rules, new_parsed, prefix_end, 1819ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen offsets_for_adjustment)); 1820ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen} 1821ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 18223345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrickstring16 FormatUrl(const GURL& url, 18233345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick const std::string& languages, 18243345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick FormatUrlTypes format_types, 18253345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick UnescapeRule::Type unescape_rules, 18263345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick url_parse::Parsed* new_parsed, 18273345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick size_t* prefix_end, 18283345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick size_t* offset_for_adjustment) { 1829ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::vector<size_t> offsets; 1830ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (offset_for_adjustment) 1831ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen offsets.push_back(*offset_for_adjustment); 1832ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen string16 result = WideToUTF16Hack( 18333345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick FormatUrlInternal(url, ASCIIToWide(languages), format_types, 1834ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen unescape_rules, new_parsed, prefix_end, &offsets)); 1835ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (offset_for_adjustment) 1836ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen *offset_for_adjustment = offsets[0]; 1837ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen return result; 1838c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 1839c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1840c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdochbool CanStripTrailingSlash(const GURL& url) { 1841c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // Omit the path only for standard, non-file URLs with nothing but "/" after 1842c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // the hostname. 1843c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return url.IsStandard() && !url.SchemeIsFile() && !url.has_query() && 1844c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch !url.has_ref() && url.path() == "/"; 1845c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch} 1846c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 1847c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick ScottGURL SimplifyUrlForRequest(const GURL& url) { 1848c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott DCHECK(url.is_valid()); 1849c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott GURL::Replacements replacements; 1850c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott replacements.ClearUsername(); 1851c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott replacements.ClearPassword(); 1852c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott replacements.ClearRef(); 1853c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return url.ReplaceComponents(replacements); 1854c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 1855c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1856c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// Specifies a comma separated list of port numbers that should be accepted 1857c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// despite bans. If the string is invalid no allowed ports are stored. 18583345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrickvoid SetExplicitlyAllowedPorts(const std::string& allowed_ports) { 1859c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (allowed_ports.empty()) 1860c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return; 1861c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 186221d179b334e59e9a3bfcaed4c4430bef1bc5759dKristian Monsen std::multiset<int> ports; 1863c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott size_t last = 0; 1864c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott size_t size = allowed_ports.size(); 1865c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // The comma delimiter. 18663345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick const std::string::value_type kComma = ','; 1867c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 1868c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Overflow is still possible for evil user inputs. 1869c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott for (size_t i = 0; i <= size; ++i) { 1870c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // The string should be composed of only digits and commas. 1871c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (i != size && !IsAsciiDigit(allowed_ports[i]) && 1872c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott (allowed_ports[i] != kComma)) 1873c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return; 1874c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (i == size || allowed_ports[i] == kComma) { 1875513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch if (i > last) { 18763345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick int port; 1877513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch base::StringToInt(allowed_ports.begin() + last, 1878513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch allowed_ports.begin() + i, 1879513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch &port); 18803345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick ports.insert(port); 18813345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick } 1882c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott last = i + 1; 1883c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 1884c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 1885c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott explicitly_allowed_ports = ports; 1886c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} 1887c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 188821d179b334e59e9a3bfcaed4c4430bef1bc5759dKristian MonsenScopedPortException::ScopedPortException(int port) : port_(port) { 188921d179b334e59e9a3bfcaed4c4430bef1bc5759dKristian Monsen explicitly_allowed_ports.insert(port); 189021d179b334e59e9a3bfcaed4c4430bef1bc5759dKristian Monsen} 189121d179b334e59e9a3bfcaed4c4430bef1bc5759dKristian Monsen 189221d179b334e59e9a3bfcaed4c4430bef1bc5759dKristian MonsenScopedPortException::~ScopedPortException() { 189321d179b334e59e9a3bfcaed4c4430bef1bc5759dKristian Monsen std::multiset<int>::iterator it = explicitly_allowed_ports.find(port_); 189421d179b334e59e9a3bfcaed4c4430bef1bc5759dKristian Monsen if (it != explicitly_allowed_ports.end()) 189521d179b334e59e9a3bfcaed4c4430bef1bc5759dKristian Monsen explicitly_allowed_ports.erase(it); 189621d179b334e59e9a3bfcaed4c4430bef1bc5759dKristian Monsen else 189721d179b334e59e9a3bfcaed4c4430bef1bc5759dKristian Monsen NOTREACHED(); 189821d179b334e59e9a3bfcaed4c4430bef1bc5759dKristian Monsen} 189921d179b334e59e9a3bfcaed4c4430bef1bc5759dKristian Monsen 1900c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdochenum IPv6SupportStatus { 1901c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch IPV6_CANNOT_CREATE_SOCKETS, 1902c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch IPV6_CAN_CREATE_SOCKETS, 1903c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch IPV6_GETIFADDRS_FAILED, 1904c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch IPV6_GLOBAL_ADDRESS_MISSING, 1905c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch IPV6_GLOBAL_ADDRESS_PRESENT, 1906c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch IPV6_INTERFACE_ARRAY_TOO_SHORT, 1907c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch IPV6_SUPPORT_MAX // Bounding values for enumeration. 1908c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch}; 1909c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 1910c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdochstatic void IPv6SupportResults(IPv6SupportStatus result) { 1911c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch static bool run_once = false; 1912c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (!run_once) { 1913c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch run_once = true; 1914c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch UMA_HISTOGRAM_ENUMERATION("Net.IPv6Status", result, IPV6_SUPPORT_MAX); 1915c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch } else { 1916c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch UMA_HISTOGRAM_ENUMERATION("Net.IPv6Status_retest", result, 1917c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch IPV6_SUPPORT_MAX); 1918c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch } 1919c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch} 1920c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 1921c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch// TODO(jar): The following is a simple estimate of IPv6 support. We may need 1922c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch// to do a test resolution, and a test connection, to REALLY verify support. 1923c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch// static 1924c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdochbool IPv6Supported() { 192500d26a728db2814620f390b418a7d6325ce5aca6Ben Murdoch#ifdef ANDROID 192600d26a728db2814620f390b418a7d6325ce5aca6Ben Murdoch // Android does not have the ifaddrs.h header 192700d26a728db2814620f390b418a7d6325ce5aca6Ben Murdoch return false; 192800d26a728db2814620f390b418a7d6325ce5aca6Ben Murdoch#elif defined(OS_POSIX) 1929c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch int test_socket = socket(AF_INET6, SOCK_STREAM, 0); 1930c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (test_socket == -1) { 1931c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch IPv6SupportResults(IPV6_CANNOT_CREATE_SOCKETS); 1932c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return false; 1933c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch } 1934c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch close(test_socket); 1935c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 1936c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // Check to see if any interface has a IPv6 address. 1937c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch struct ifaddrs* interface_addr = NULL; 1938c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch int rv = getifaddrs(&interface_addr); 1939c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (rv != 0) { 1940c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch IPv6SupportResults(IPV6_GETIFADDRS_FAILED); 1941c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return true; // Don't yet block IPv6. 1942c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch } 1943c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 1944c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch bool found_ipv6 = false; 1945c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch for (struct ifaddrs* interface = interface_addr; 1946c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch interface != NULL; 1947c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch interface = interface->ifa_next) { 1948c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (!(IFF_UP & interface->ifa_flags)) 1949c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch continue; 1950c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (IFF_LOOPBACK & interface->ifa_flags) 1951c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch continue; 1952c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch struct sockaddr* addr = interface->ifa_addr; 1953c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (!addr) 1954c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch continue; 1955c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (addr->sa_family != AF_INET6) 1956c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch continue; 1957c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // Safe cast since this is AF_INET6. 1958c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch struct sockaddr_in6* addr_in6 = 1959c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch reinterpret_cast<struct sockaddr_in6*>(addr); 1960c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch struct in6_addr* sin6_addr = &addr_in6->sin6_addr; 1961c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (IN6_IS_ADDR_LOOPBACK(sin6_addr) || IN6_IS_ADDR_LINKLOCAL(sin6_addr)) 1962c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch continue; 1963c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch found_ipv6 = true; 1964c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch break; 1965c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch } 1966c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch freeifaddrs(interface_addr); 1967c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (!found_ipv6) { 1968c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch IPv6SupportResults(IPV6_GLOBAL_ADDRESS_MISSING); 1969c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return false; 1970c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch } 1971c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 1972c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch IPv6SupportResults(IPV6_GLOBAL_ADDRESS_PRESENT); 1973c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return true; 1974c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch#elif defined(OS_WIN) 1975c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch EnsureWinsockInit(); 1976c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch SOCKET test_socket = socket(AF_INET6, SOCK_STREAM, 0); 1977c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (test_socket == INVALID_SOCKET) { 1978c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch IPv6SupportResults(IPV6_CANNOT_CREATE_SOCKETS); 1979c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return false; 1980c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch } 1981c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch closesocket(test_socket); 1982c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 1983c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // Check to see if any interface has a IPv6 address. 1984ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // The GetAdaptersAddresses MSDN page recommends using a size of 15000 to 1985ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // avoid reallocation. 1986ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen ULONG adapters_size = 15000; 1987ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen scoped_ptr_malloc<IP_ADAPTER_ADDRESSES> adapters; 1988ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen ULONG error; 1989ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen int num_tries = 0; 1990ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen do { 1991ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen adapters.reset( 1992ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen reinterpret_cast<PIP_ADAPTER_ADDRESSES>(malloc(adapters_size))); 1993ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // Return only unicast addresses. 1994ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen error = GetAdaptersAddresses(AF_UNSPEC, 1995ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen GAA_FLAG_SKIP_ANYCAST | 1996ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen GAA_FLAG_SKIP_MULTICAST | 1997ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen GAA_FLAG_SKIP_DNS_SERVER | 1998ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen GAA_FLAG_SKIP_FRIENDLY_NAME, 1999ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen NULL, adapters.get(), &adapters_size); 2000ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen num_tries++; 2001ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } while (error == ERROR_BUFFER_OVERFLOW && num_tries <= 3); 2002ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (error == ERROR_NO_DATA) { 2003ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen IPv6SupportResults(IPV6_GLOBAL_ADDRESS_MISSING); 2004ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen return false; 2005ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } 2006ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (error != ERROR_SUCCESS) { 2007ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen IPv6SupportResults(IPV6_GETIFADDRS_FAILED); 2008c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return true; // Don't yet block IPv6. 2009c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch } 2010ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 2011ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen PIP_ADAPTER_ADDRESSES adapter; 2012ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen for (adapter = adapters.get(); adapter; adapter = adapter->Next) { 2013ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (adapter->OperStatus != IfOperStatusUp) 2014c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch continue; 2015ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (adapter->IfType == IF_TYPE_SOFTWARE_LOOPBACK) 2016c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch continue; 2017ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen PIP_ADAPTER_UNICAST_ADDRESS unicast_address; 2018ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen for (unicast_address = adapter->FirstUnicastAddress; 2019ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen unicast_address; 2020ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen unicast_address = unicast_address->Next) { 2021ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (unicast_address->Address.lpSockaddr->sa_family != AF_INET6) 2022ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen continue; 2023ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // Safe cast since this is AF_INET6. 2024ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen struct sockaddr_in6* addr_in6 = reinterpret_cast<struct sockaddr_in6*>( 2025ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen unicast_address->Address.lpSockaddr); 2026ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen struct in6_addr* sin6_addr = &addr_in6->sin6_addr; 2027ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (IN6_IS_ADDR_LOOPBACK(sin6_addr) || IN6_IS_ADDR_LINKLOCAL(sin6_addr)) 2028ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen continue; 2029ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen IPv6SupportResults(IPV6_GLOBAL_ADDRESS_PRESENT); 2030ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen return true; 2031ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } 2032c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch } 2033c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 2034c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch IPv6SupportResults(IPV6_GLOBAL_ADDRESS_MISSING); 2035c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return false; 2036c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch#else 2037c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch NOTIMPLEMENTED(); 2038c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return true; 2039c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch#endif // defined(various platforms) 2040c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch} 2041c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 20423345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrickbool HaveOnlyLoopbackAddresses() { 2043dd0e069b1c2e5079f99024c4d54c7d06ef81d11bIain Merrick#if defined(ANDROID) 2044dd0e069b1c2e5079f99024c4d54c7d06ef81d11bIain Merrick // Android has no <ifaddrs.h> 2045dd0e069b1c2e5079f99024c4d54c7d06ef81d11bIain Merrick return false; 2046dd0e069b1c2e5079f99024c4d54c7d06ef81d11bIain Merrick#elif defined(OS_POSIX) 20473345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick struct ifaddrs* interface_addr = NULL; 20483345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick int rv = getifaddrs(&interface_addr); 20493345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick if (rv != 0) { 2050731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick DVLOG(1) << "getifaddrs() failed with errno = " << errno; 20513345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick return false; 20523345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick } 20533345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick 20543345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick bool result = true; 20553345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick for (struct ifaddrs* interface = interface_addr; 20563345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick interface != NULL; 20573345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick interface = interface->ifa_next) { 20583345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick if (!(IFF_UP & interface->ifa_flags)) 20593345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick continue; 20603345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick if (IFF_LOOPBACK & interface->ifa_flags) 20613345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick continue; 20623345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick const struct sockaddr* addr = interface->ifa_addr; 20633345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick if (!addr) 20643345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick continue; 20653345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick if (addr->sa_family == AF_INET6) { 20663345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick // Safe cast since this is AF_INET6. 20673345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick const struct sockaddr_in6* addr_in6 = 20683345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick reinterpret_cast<const struct sockaddr_in6*>(addr); 20693345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick const struct in6_addr* sin6_addr = &addr_in6->sin6_addr; 20703345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick if (IN6_IS_ADDR_LOOPBACK(sin6_addr) || IN6_IS_ADDR_LINKLOCAL(sin6_addr)) 20713345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick continue; 20723345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick } 20733345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick if (addr->sa_family != AF_INET6 && addr->sa_family != AF_INET) 20743345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick continue; 20753345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick 20763345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick result = false; 20773345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick break; 20783345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick } 20793345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick freeifaddrs(interface_addr); 20803345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick return result; 2081ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen#elif defined(OS_WIN) 2082ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // TODO(wtc): implement with the GetAdaptersAddresses function. 2083ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen NOTIMPLEMENTED(); 2084ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen return false; 20853345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick#else 20863345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick NOTIMPLEMENTED(); 20873345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick return false; 20883345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick#endif // defined(various platforms) 20893345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick} 20903345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick 2091c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdochbool ParseIPLiteralToNumber(const std::string& ip_literal, 2092c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch IPAddressNumber* ip_number) { 2093c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // |ip_literal| could be either a IPv4 or an IPv6 literal. If it contains 2094c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // a colon however, it must be an IPv6 address. 2095c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (ip_literal.find(':') != std::string::npos) { 2096c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // GURL expects IPv6 hostnames to be surrounded with brackets. 2097c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch std::string host_brackets = "[" + ip_literal + "]"; 2098c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch url_parse::Component host_comp(0, host_brackets.size()); 2099c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 2100c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // Try parsing the hostname as an IPv6 literal. 2101c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch ip_number->resize(16); // 128 bits. 2102c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return url_canon::IPv6AddressToNumber(host_brackets.data(), 2103c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch host_comp, 2104c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch &(*ip_number)[0]); 2105c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch } 2106c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 2107c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // Otherwise the string is an IPv4 address. 2108c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch ip_number->resize(4); // 32 bits. 2109c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch url_parse::Component host_comp(0, ip_literal.size()); 2110c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch int num_components; 2111c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch url_canon::CanonHostInfo::Family family = url_canon::IPv4AddressToNumber( 2112c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch ip_literal.data(), host_comp, &(*ip_number)[0], &num_components); 2113c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return family == url_canon::CanonHostInfo::IPV4; 2114c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch} 2115c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 2116c407dc5cd9bdc5668497f21b26b09d988ab439deBen MurdochIPAddressNumber ConvertIPv4NumberToIPv6Number( 2117c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch const IPAddressNumber& ipv4_number) { 2118c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch DCHECK(ipv4_number.size() == 4); 2119c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 2120c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // IPv4-mapped addresses are formed by: 2121c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // <80 bits of zeros> + <16 bits of ones> + <32-bit IPv4 address>. 2122c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch IPAddressNumber ipv6_number; 2123c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch ipv6_number.reserve(16); 2124c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch ipv6_number.insert(ipv6_number.end(), 10, 0); 2125c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch ipv6_number.push_back(0xFF); 2126c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch ipv6_number.push_back(0xFF); 2127c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch ipv6_number.insert(ipv6_number.end(), ipv4_number.begin(), ipv4_number.end()); 2128c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return ipv6_number; 2129c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch} 2130c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 2131c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdochbool ParseCIDRBlock(const std::string& cidr_literal, 2132c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch IPAddressNumber* ip_number, 2133c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch size_t* prefix_length_in_bits) { 2134c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // We expect CIDR notation to match one of these two templates: 2135c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // <IPv4-literal> "/" <number of bits> 2136c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // <IPv6-literal> "/" <number of bits> 2137c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 2138c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch std::vector<std::string> parts; 2139731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick base::SplitString(cidr_literal, '/', &parts); 2140c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (parts.size() != 2) 2141c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return false; 2142c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 2143c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // Parse the IP address. 2144c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (!ParseIPLiteralToNumber(parts[0], ip_number)) 2145c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return false; 2146c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 2147c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // Parse the prefix length. 2148c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch int number_of_bits = -1; 21493345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick if (!base::StringToInt(parts[1], &number_of_bits)) 2150c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return false; 2151c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 2152c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // Make sure the prefix length is in a valid range. 2153c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (number_of_bits < 0 || 2154c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch number_of_bits > static_cast<int>(ip_number->size() * 8)) 2155c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return false; 2156c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 2157c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch *prefix_length_in_bits = static_cast<size_t>(number_of_bits); 2158c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return true; 2159c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch} 2160c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 2161c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdochbool IPNumberMatchesPrefix(const IPAddressNumber& ip_number, 2162c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch const IPAddressNumber& ip_prefix, 2163c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch size_t prefix_length_in_bits) { 2164c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // Both the input IP address and the prefix IP address should be 2165c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // either IPv4 or IPv6. 2166c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch DCHECK(ip_number.size() == 4 || ip_number.size() == 16); 2167c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch DCHECK(ip_prefix.size() == 4 || ip_prefix.size() == 16); 2168c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 2169c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch DCHECK_LE(prefix_length_in_bits, ip_prefix.size() * 8); 2170c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 2171c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // In case we have an IPv6 / IPv4 mismatch, convert the IPv4 addresses to 2172c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // IPv6 addresses in order to do the comparison. 2173c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (ip_number.size() != ip_prefix.size()) { 2174c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (ip_number.size() == 4) { 2175c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return IPNumberMatchesPrefix(ConvertIPv4NumberToIPv6Number(ip_number), 2176c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch ip_prefix, prefix_length_in_bits); 2177c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch } 2178c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return IPNumberMatchesPrefix(ip_number, 2179c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch ConvertIPv4NumberToIPv6Number(ip_prefix), 2180c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 96 + prefix_length_in_bits); 2181c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch } 2182c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 2183c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // Otherwise we are comparing two IPv4 addresses, or two IPv6 addresses. 2184c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // Compare all the bytes that fall entirely within the prefix. 2185c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch int num_entire_bytes_in_prefix = prefix_length_in_bits / 8; 2186c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch for (int i = 0; i < num_entire_bytes_in_prefix; ++i) { 2187c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (ip_number[i] != ip_prefix[i]) 2188c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return false; 2189c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch } 2190c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 2191c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // In case the prefix was not a multiple of 8, there will be 1 byte 2192c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // which is only partially masked. 2193c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch int remaining_bits = prefix_length_in_bits % 8; 2194c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (remaining_bits != 0) { 2195c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch unsigned char mask = 0xFF << (8 - remaining_bits); 2196c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch int i = num_entire_bytes_in_prefix; 2197c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if ((ip_number[i] & mask) != (ip_prefix[i] & mask)) 2198c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return false; 2199c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch } 2200c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 2201c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return true; 2202c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch} 2203c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 2204ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsenstruct addrinfo* CreateCopyOfAddrinfo(const struct addrinfo* info, 2205ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen bool recursive) { 2206ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen DCHECK(info); 2207ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen struct addrinfo* copy = new addrinfo; 2208ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 2209ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // Copy all the fields (some of these are pointers, we will fix that next). 2210ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen memcpy(copy, info, sizeof(addrinfo)); 2211ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 2212ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // ai_canonname is a NULL-terminated string. 2213ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (info->ai_canonname) { 2214ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen copy->ai_canonname = do_strdup(info->ai_canonname); 2215ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } 2216ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 2217ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // ai_addr is a buffer of length ai_addrlen. 2218ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (info->ai_addr) { 2219ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen copy->ai_addr = reinterpret_cast<sockaddr *>(new char[info->ai_addrlen]); 2220ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen memcpy(copy->ai_addr, info->ai_addr, info->ai_addrlen); 2221ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } 2222ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 2223ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // Recursive copy. 2224ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (recursive && info->ai_next) 2225ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen copy->ai_next = CreateCopyOfAddrinfo(info->ai_next, recursive); 2226ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen else 2227ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen copy->ai_next = NULL; 2228ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 2229ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen return copy; 2230ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen} 2231ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 2232ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsenvoid FreeCopyOfAddrinfo(struct addrinfo* info) { 2233ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen DCHECK(info); 2234ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (info->ai_canonname) 2235ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen free(info->ai_canonname); // Allocated by strdup. 2236ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 2237ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (info->ai_addr) 2238ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen delete [] reinterpret_cast<char*>(info->ai_addr); 2239ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 2240ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen struct addrinfo* next = info->ai_next; 2241ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 2242ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen delete info; 2243ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 2244ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // Recursive free. 2245ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (next) 2246ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen FreeCopyOfAddrinfo(next); 2247ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen} 2248ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 2249c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch// Returns the port field of the sockaddr in |info|. 225072a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsenuint16* GetPortFieldFromAddrinfo(struct addrinfo* info) { 225172a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen const struct addrinfo* const_info = info; 225272a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen const uint16* port_field = GetPortFieldFromAddrinfo(const_info); 225372a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen return const_cast<uint16*>(port_field); 225472a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen} 225572a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen 225672a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsenconst uint16* GetPortFieldFromAddrinfo(const struct addrinfo* info) { 2257c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch DCHECK(info); 225872a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen const struct sockaddr* address = info->ai_addr; 225972a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen DCHECK(address); 226072a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen DCHECK_EQ(info->ai_family, address->sa_family); 226172a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen return GetPortFieldFromSockaddr(address, info->ai_addrlen); 226272a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen} 226372a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen 226472a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsenint GetPortFromAddrinfo(const struct addrinfo* info) { 226572a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen const uint16* port_field = GetPortFieldFromAddrinfo(info); 226672a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen if (!port_field) 226772a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen return -1; 226872a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen return ntohs(*port_field); 226972a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen} 227072a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen 227172a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsenconst uint16* GetPortFieldFromSockaddr(const struct sockaddr* address, 227272a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen socklen_t address_len) { 227372a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen if (address->sa_family == AF_INET) { 227472a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen DCHECK_LE(sizeof(sockaddr_in), static_cast<size_t>(address_len)); 227572a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen const struct sockaddr_in* sockaddr = 227672a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen reinterpret_cast<const struct sockaddr_in*>(address); 2277c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return &sockaddr->sin_port; 227872a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen } else if (address->sa_family == AF_INET6) { 227972a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen DCHECK_LE(sizeof(sockaddr_in6), static_cast<size_t>(address_len)); 228072a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen const struct sockaddr_in6* sockaddr = 228172a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen reinterpret_cast<const struct sockaddr_in6*>(address); 2282c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return &sockaddr->sin6_port; 2283c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch } else { 2284c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch NOTREACHED(); 2285c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return NULL; 2286c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch } 2287c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch} 2288c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 228972a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsenint GetPortFromSockaddr(const struct sockaddr* address, socklen_t address_len) { 229072a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen const uint16* port_field = GetPortFieldFromSockaddr(address, address_len); 2291c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (!port_field) 2292c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return -1; 2293c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return ntohs(*port_field); 2294c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch} 2295c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 2296ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsenbool IsLocalhost(const std::string& host) { 2297ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (host == "localhost" || 2298ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen host == "localhost.localdomain" || 2299ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen host == "localhost6" || 2300ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen host == "localhost6.localdomain6") 2301ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen return true; 2302ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 2303ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen IPAddressNumber ip_number; 2304ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (ParseIPLiteralToNumber(host, &ip_number)) { 2305ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen size_t size = ip_number.size(); 2306ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen switch (size) { 2307ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen case kIPv4AddressSize: { 2308ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen IPAddressNumber localhost_prefix; 2309ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen localhost_prefix.push_back(127); 2310ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen for (int i = 0; i < 3; ++i) { 2311ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen localhost_prefix.push_back(0); 2312ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } 2313ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen return IPNumberMatchesPrefix(ip_number, localhost_prefix, 8); 2314ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } 2315ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 2316ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen case kIPv6AddressSize: { 2317ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen struct in6_addr sin6_addr; 2318ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen memcpy(&sin6_addr, &ip_number[0], kIPv6AddressSize); 2319ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen return !!IN6_IS_ADDR_LOOPBACK(&sin6_addr); 2320ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } 2321ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 2322ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen default: 2323ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen NOTREACHED(); 2324ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } 2325ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } 2326ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 2327ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen return false; 2328ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen} 2329ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 2330ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian MonsenNetworkInterface::NetworkInterface() { 2331ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen} 2332ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 2333ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian MonsenNetworkInterface::NetworkInterface(const std::string& name, 2334ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen const IPAddressNumber& address) 2335ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen : name(name), address(address) { 2336ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen} 2337ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 2338ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian MonsenNetworkInterface::~NetworkInterface() { 2339ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen} 2340ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 2341ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian MonsenClampComponentOffset::ClampComponentOffset(size_t component_start) 2342ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen : component_start(component_start) {} 2343ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 2344ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsensize_t ClampComponentOffset::operator()(size_t offset) { 2345ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen return (offset >= component_start) ? 2346ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen offset : std::wstring::npos; 2347ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen} 2348ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 2349c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott} // namespace net 2350