1c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// Copyright 2013 The Chromium Authors. All rights reserved. 2c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// Use of this source code is governed by a BSD-style license that can be 3c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// found in the LICENSE file. 4c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 5c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// Canonicalizer functions for working with and resolving relative URLs. 6c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 7c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#include "base/logging.h" 8c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#include "url/url_canon.h" 9c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#include "url/url_canon_internal.h" 10f8ee788a64d60abd8f2d742a5fdedde054ecd910Torne (Richard Coles)#include "url/url_constants.h" 11c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#include "url/url_file.h" 12c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#include "url/url_parse_internal.h" 13c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#include "url/url_util_internal.h" 14c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 150529e5d033099cbfc42635f6f6183833b09dff6eBen Murdochnamespace url { 16c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 17c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)namespace { 18c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 19c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// Firefox does a case-sensitive compare (which is probably wrong--Mozilla bug 20c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// 379034), whereas IE is case-insensetive. 21c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// 22c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// We choose to be more permissive like IE. We don't need to worry about 23c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// unescaping or anything here: neither IE or Firefox allow this. We also 24c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// don't have to worry about invalid scheme characters since we are comparing 25c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// against the canonical scheme of the base. 26c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// 27c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// The base URL should always be canonical, therefore is ASCII. 28c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)template<typename CHAR> 29c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)bool AreSchemesEqual(const char* base, 300529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch const Component& base_scheme, 31c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const CHAR* cmp, 320529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch const Component& cmp_scheme) { 33c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (base_scheme.len != cmp_scheme.len) 34c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return false; 35c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (int i = 0; i < base_scheme.len; i++) { 36c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // We assume the base is already canonical, so we don't have to 37c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // canonicalize it. 38c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (CanonicalSchemeChar(cmp[cmp_scheme.begin + i]) != 39c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) base[base_scheme.begin + i]) 40c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return false; 41c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 42c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return true; 43c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 44c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 45c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#ifdef WIN32 46c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 47c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// Here, we also allow Windows paths to be represented as "/C:/" so we can be 48c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// consistent about URL paths beginning with slashes. This function is like 49c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// DoesBeginWindowsDrivePath except that it also requires a slash at the 50c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// beginning. 51c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)template<typename CHAR> 52c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)bool DoesBeginSlashWindowsDriveSpec(const CHAR* spec, int start_offset, 53c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int spec_len) { 54c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (start_offset >= spec_len) 55c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return false; 560529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch return IsURLSlash(spec[start_offset]) && 570529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch DoesBeginWindowsDriveSpec(spec, start_offset + 1, spec_len); 58c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 59c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 60c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#endif // WIN32 61c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 62c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// See IsRelativeURL in the header file for usage. 63c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)template<typename CHAR> 64c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)bool DoIsRelativeURL(const char* base, 650529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch const Parsed& base_parsed, 66c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const CHAR* url, 67c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int url_len, 68c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool is_base_hierarchical, 69c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool* is_relative, 700529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch Component* relative_component) { 71c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) *is_relative = false; // So we can default later to not relative. 72c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 73c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Trim whitespace and construct a new range for the substring. 74c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int begin = 0; 750529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch TrimURL(url, &begin, &url_len); 76c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (begin >= url_len) { 77c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Empty URLs are relative, but do nothing. 780529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch *relative_component = Component(begin, 0); 79c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) *is_relative = true; 80c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return true; 81c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 82c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 83c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#ifdef WIN32 84c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // We special case paths like "C:\foo" so they can link directly to the 85c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // file on Windows (IE compatability). The security domain stuff should 86c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // prevent a link like this from actually being followed if its on a 87c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // web page. 88c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // 89c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // We treat "C:/foo" as an absolute URL. We can go ahead and treat "/c:/" 90c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // as relative, as this will just replace the path when the base scheme 91c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // is a file and the answer will still be correct. 92c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // 93c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // We require strict backslashes when detecting UNC since two forward 94c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // shashes should be treated a a relative URL with a hostname. 950529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch if (DoesBeginWindowsDriveSpec(url, begin, url_len) || 960529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch DoesBeginUNCPath(url, begin, url_len, true)) 97c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return true; 98c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#endif // WIN32 99c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 100c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // See if we've got a scheme, if not, we know this is a relative URL. 101c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // BUT: Just because we have a scheme, doesn't make it absolute. 102c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // "http:foo.html" is a relative URL with path "foo.html". If the scheme is 103c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // empty, we treat it as relative (":foo") like IE does. 1040529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch Component scheme; 105f2477e01787aa58f445919b809d89e252beef54fTorne (Richard Coles) const bool scheme_is_empty = 1060529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch !ExtractScheme(url, url_len, &scheme) || scheme.len == 0; 107f2477e01787aa58f445919b809d89e252beef54fTorne (Richard Coles) if (scheme_is_empty) { 108f2477e01787aa58f445919b809d89e252beef54fTorne (Richard Coles) if (url[begin] == '#') { 109f2477e01787aa58f445919b809d89e252beef54fTorne (Richard Coles) // |url| is a bare fragement (e.g. "#foo"). This can be resolved against 110f2477e01787aa58f445919b809d89e252beef54fTorne (Richard Coles) // any base. Fall-through. 111f2477e01787aa58f445919b809d89e252beef54fTorne (Richard Coles) } else if (!is_base_hierarchical) { 112f2477e01787aa58f445919b809d89e252beef54fTorne (Richard Coles) // Don't allow relative URLs if the base scheme doesn't support it. 113c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return false; 114f2477e01787aa58f445919b809d89e252beef54fTorne (Richard Coles) } 115c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 1160529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch *relative_component = MakeRange(begin, url_len); 117c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) *is_relative = true; 118c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return true; 119c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 120c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 121c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // If the scheme isn't valid, then it's relative. 122c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int scheme_end = scheme.end(); 123c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (int i = scheme.begin; i < scheme_end; i++) { 124c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (!CanonicalSchemeChar(url[i])) { 125a1401311d1ab56c4ed0a474bd38c108f75cb0cd9Torne (Richard Coles) if (!is_base_hierarchical) { 126a1401311d1ab56c4ed0a474bd38c108f75cb0cd9Torne (Richard Coles) // Don't allow relative URLs if the base scheme doesn't support it. 127a1401311d1ab56c4ed0a474bd38c108f75cb0cd9Torne (Richard Coles) return false; 128a1401311d1ab56c4ed0a474bd38c108f75cb0cd9Torne (Richard Coles) } 1290529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch *relative_component = MakeRange(begin, url_len); 130c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) *is_relative = true; 131c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return true; 132c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 133c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 134c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 135c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // If the scheme is not the same, then we can't count it as relative. 136c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (!AreSchemesEqual(base, base_parsed.scheme, url, scheme)) 137c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return true; 138c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 139c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // When the scheme that they both share is not hierarchical, treat the 140c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // incoming scheme as absolute (this way with the base of "data:foo", 141c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // "data:bar" will be reported as absolute. 142c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (!is_base_hierarchical) 143c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return true; 144c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 145c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int colon_offset = scheme.end(); 146c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 147c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // If it's a filesystem URL, the only valid way to make it relative is not to 148c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // supply a scheme. There's no equivalent to e.g. http:index.html. 149f8ee788a64d60abd8f2d742a5fdedde054ecd910Torne (Richard Coles) if (CompareSchemeComponent(url, scheme, kFileSystemScheme)) 150c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return true; 151c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 152c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // ExtractScheme guarantees that the colon immediately follows what it 153c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // considers to be the scheme. CountConsecutiveSlashes will handle the 154c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // case where the begin offset is the end of the input. 1550529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch int num_slashes = CountConsecutiveSlashes(url, colon_offset + 1, url_len); 156c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 157c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (num_slashes == 0 || num_slashes == 1) { 158c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // No slashes means it's a relative path like "http:foo.html". One slash 159c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // is an absolute path. "http:/home/foo.html" 160c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) *is_relative = true; 1610529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch *relative_component = MakeRange(colon_offset + 1, url_len); 162c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return true; 163c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 164c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 165c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Two or more slashes after the scheme we treat as absolute. 166c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return true; 167c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 168c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 169c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// Copies all characters in the range [begin, end) of |spec| to the output, 170c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// up until and including the last slash. There should be a slash in the 171c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// range, if not, nothing will be copied. 172c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// 173c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// The input is assumed to be canonical, so we search only for exact slashes 174c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// and not backslashes as well. We also know that it's ASCII. 175c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)void CopyToLastSlash(const char* spec, 176c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int begin, 177c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int end, 178c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) CanonOutput* output) { 179c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Find the last slash. 180c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int last_slash = -1; 181c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (int i = end - 1; i >= begin; i--) { 182c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (spec[i] == '/') { 183c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) last_slash = i; 184c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) break; 185c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 186c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 187c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (last_slash < 0) 188c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return; // No slash. 189c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 190c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Copy. 191c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (int i = begin; i <= last_slash; i++) 192c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output->push_back(spec[i]); 193c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 194c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 195c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// Copies a single component from the source to the output. This is used 196c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// when resolving relative URLs and a given component is unchanged. Since the 197c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// source should already be canonical, we don't have to do anything special, 198c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// and the input is ASCII. 199c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)void CopyOneComponent(const char* source, 2000529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch const Component& source_component, 201c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) CanonOutput* output, 2020529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch Component* output_component) { 203c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (source_component.len < 0) { 204c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // This component is not present. 2050529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch *output_component = Component(); 206c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return; 207c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 208c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 209c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output_component->begin = output->length(); 210c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int source_end = source_component.end(); 211c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (int i = source_component.begin; i < source_end; i++) 212c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output->push_back(source[i]); 213c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output_component->len = output->length() - output_component->begin; 214c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 215c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 216c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#ifdef WIN32 217c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 218c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// Called on Windows when the base URL is a file URL, this will copy the "C:" 219c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// to the output, if there is a drive letter and if that drive letter is not 220c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// being overridden by the relative URL. Otherwise, do nothing. 221c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// 222c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// It will return the index of the beginning of the next character in the 223c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// base to be processed: if there is a "C:", the slash after it, or if 224c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// there is no drive letter, the slash at the beginning of the path, or 225c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// the end of the base. This can be used as the starting offset for further 226c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// path processing. 227c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)template<typename CHAR> 228c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)int CopyBaseDriveSpecIfNecessary(const char* base_url, 229c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int base_path_begin, 230c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int base_path_end, 231c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const CHAR* relative_url, 232c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int path_start, 233c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int relative_url_len, 234c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) CanonOutput* output) { 235c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (base_path_begin >= base_path_end) 236c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return base_path_begin; // No path. 237c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 238c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // If the relative begins with a drive spec, don't do anything. The existing 239c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // drive spec in the base will be replaced. 2400529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch if (DoesBeginWindowsDriveSpec(relative_url, path_start, relative_url_len)) { 241c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return base_path_begin; // Relative URL path is "C:/foo" 242c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 243c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 244c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // The path should begin with a slash (as all canonical paths do). We check 245c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // if it is followed by a drive letter and copy it. 246c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (DoesBeginSlashWindowsDriveSpec(base_url, 247c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) base_path_begin, 248c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) base_path_end)) { 249c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Copy the two-character drive spec to the output. It will now look like 250c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // "file:///C:" so the rest of it can be treated like a standard path. 251c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output->push_back('/'); 252c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output->push_back(base_url[base_path_begin + 1]); 253c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output->push_back(base_url[base_path_begin + 2]); 254c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return base_path_begin + 3; 255c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 256c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 257c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return base_path_begin; 258c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 259c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 260c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#endif // WIN32 261c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 262c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// A subroutine of DoResolveRelativeURL, this resolves the URL knowning that 263c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// the input is a relative path or less (qyuery or ref). 264c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)template<typename CHAR> 265c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)bool DoResolveRelativePath(const char* base_url, 2660529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch const Parsed& base_parsed, 267c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool base_is_file, 268c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const CHAR* relative_url, 2690529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch const Component& relative_component, 270c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) CharsetConverter* query_converter, 271c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) CanonOutput* output, 2720529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch Parsed* out_parsed) { 273c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool success = true; 274c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 275c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // We know the authority section didn't change, copy it to the output. We 276c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // also know we have a path so can copy up to there. 2770529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch Component path, query, ref; 2780529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch ParsePathInternal(relative_url, relative_component, &path, &query, &ref); 279c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Canonical URLs always have a path, so we can use that offset. 280c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output->Append(base_url, base_parsed.path.begin); 281c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 282c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (path.len > 0) { 283c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // The path is replaced or modified. 284c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int true_path_begin = output->length(); 285c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 286c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // For file: URLs on Windows, we don't want to treat the drive letter and 287c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // colon as part of the path for relative file resolution when the 288c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // incoming URL does not provide a drive spec. We save the true path 289c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // beginning so we can fix it up after we are done. 290c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int base_path_begin = base_parsed.path.begin; 291c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#ifdef WIN32 292c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (base_is_file) { 293c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) base_path_begin = CopyBaseDriveSpecIfNecessary( 294c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) base_url, base_parsed.path.begin, base_parsed.path.end(), 295c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) relative_url, relative_component.begin, relative_component.end(), 296c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output); 297c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Now the output looks like either "file://" or "file:///C:" 298c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // and we can start appending the rest of the path. |base_path_begin| 299c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // points to the character in the base that comes next. 300c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 301c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#endif // WIN32 302c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 3030529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch if (IsURLSlash(relative_url[path.begin])) { 304c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Easy case: the path is an absolute path on the server, so we can 305c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // just replace everything from the path on with the new versions. 306c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Since the input should be canonical hierarchical URL, we should 307c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // always have a path. 308c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) success &= CanonicalizePath(relative_url, path, 309c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output, &out_parsed->path); 310c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } else { 311c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Relative path, replace the query, and reference. We take the 312c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // original path with the file part stripped, and append the new path. 313c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // The canonicalizer will take care of resolving ".." and "." 314c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int path_begin = output->length(); 315c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) CopyToLastSlash(base_url, base_path_begin, base_parsed.path.end(), 316c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output); 317c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) success &= CanonicalizePartialPath(relative_url, path, path_begin, 318c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output); 3190529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch out_parsed->path = MakeRange(path_begin, output->length()); 320c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 321c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Copy the rest of the stuff after the path from the relative path. 322c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 323c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 324c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Finish with the query and reference part (these can't fail). 325c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) CanonicalizeQuery(relative_url, query, query_converter, 326c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output, &out_parsed->query); 327c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) CanonicalizeRef(relative_url, ref, output, &out_parsed->ref); 328c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 329c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Fix the path beginning to add back the "C:" we may have written above. 3300529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch out_parsed->path = MakeRange(true_path_begin, out_parsed->path.end()); 331c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return success; 332c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 333c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 334c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // If we get here, the path is unchanged: copy to output. 335c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) CopyOneComponent(base_url, base_parsed.path, output, &out_parsed->path); 336c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 337c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (query.is_valid()) { 338c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Just the query specified, replace the query and reference (ignore 339c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // failures for refs) 340c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) CanonicalizeQuery(relative_url, query, query_converter, 341c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output, &out_parsed->query); 342c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) CanonicalizeRef(relative_url, ref, output, &out_parsed->ref); 343c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return success; 344c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 345c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 346c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // If we get here, the query is unchanged: copy to output. Note that the 347c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // range of the query parameter doesn't include the question mark, so we 348c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // have to add it manually if there is a component. 349c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (base_parsed.query.is_valid()) 350c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output->push_back('?'); 351c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) CopyOneComponent(base_url, base_parsed.query, output, &out_parsed->query); 352c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 353c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (ref.is_valid()) { 354c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Just the reference specified: replace it (ignoring failures). 355c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) CanonicalizeRef(relative_url, ref, output, &out_parsed->ref); 356c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return success; 357c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 358c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 359c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // We should always have something to do in this function, the caller checks 360c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // that some component is being replaced. 361c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) DCHECK(false) << "Not reached"; 362c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return success; 363c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 364c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 365c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// Resolves a relative URL that contains a host. Typically, these will 366c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// be of the form "//www.google.com/foo/bar?baz#ref" and the only thing which 367c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// should be kept from the original URL is the scheme. 368c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)template<typename CHAR> 369c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)bool DoResolveRelativeHost(const char* base_url, 3700529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch const Parsed& base_parsed, 371c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const CHAR* relative_url, 3720529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch const Component& relative_component, 373c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) CharsetConverter* query_converter, 374c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) CanonOutput* output, 3750529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch Parsed* out_parsed) { 376c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Parse the relative URL, just like we would for anything following a 377c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // scheme. 3780529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch Parsed relative_parsed; // Everything but the scheme is valid. 3790529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch ParseAfterScheme(relative_url, relative_component.end(), 3800529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch relative_component.begin, &relative_parsed); 381c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 382c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Now we can just use the replacement function to replace all the necessary 383c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // parts of the old URL with the new one. 384c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) Replacements<CHAR> replacements; 385c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) replacements.SetUsername(relative_url, relative_parsed.username); 386c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) replacements.SetPassword(relative_url, relative_parsed.password); 387c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) replacements.SetHost(relative_url, relative_parsed.host); 388c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) replacements.SetPort(relative_url, relative_parsed.port); 389c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) replacements.SetPath(relative_url, relative_parsed.path); 390c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) replacements.SetQuery(relative_url, relative_parsed.query); 391c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) replacements.SetRef(relative_url, relative_parsed.ref); 392c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 393c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return ReplaceStandardURL(base_url, base_parsed, replacements, 394c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) query_converter, output, out_parsed); 395c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 396c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 397c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// Resolves a relative URL that happens to be an absolute file path. Examples 398c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// include: "//hostname/path", "/c:/foo", and "//hostname/c:/foo". 399c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)template<typename CHAR> 400c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)bool DoResolveAbsoluteFile(const CHAR* relative_url, 4010529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch const Component& relative_component, 402c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) CharsetConverter* query_converter, 403c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) CanonOutput* output, 4040529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch Parsed* out_parsed) { 405c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Parse the file URL. The file URl parsing function uses the same logic 406c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // as we do for determining if the file is absolute, in which case it will 407c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // not bother to look for a scheme. 4080529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch Parsed relative_parsed; 4090529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch ParseFileURL(&relative_url[relative_component.begin], relative_component.len, 4100529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch &relative_parsed); 411c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 412c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return CanonicalizeFileURL(&relative_url[relative_component.begin], 413c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) relative_component.len, relative_parsed, 414c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) query_converter, output, out_parsed); 415c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 416c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 417c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)// TODO(brettw) treat two slashes as root like Mozilla for FTP? 418c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)template<typename CHAR> 419c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)bool DoResolveRelativeURL(const char* base_url, 4200529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch const Parsed& base_parsed, 421c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool base_is_file, 422c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const CHAR* relative_url, 4230529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch const Component& relative_component, 424c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) CharsetConverter* query_converter, 425c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) CanonOutput* output, 4260529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch Parsed* out_parsed) { 427c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Starting point for our output parsed. We'll fix what we change. 428c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) *out_parsed = base_parsed; 429c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 430c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Sanity check: the input should have a host or we'll break badly below. 431c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // We can only resolve relative URLs with base URLs that have hosts and 432c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // paths (even the default path of "/" is OK). 433c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // 434c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // We allow hosts with no length so we can handle file URLs, for example. 435c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (base_parsed.path.len <= 0) { 436c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // On error, return the input (resolving a relative URL on a non-relative 437c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // base = the base). 438c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int base_len = base_parsed.Length(); 439c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) for (int i = 0; i < base_len; i++) 440c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output->push_back(base_url[i]); 441c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return false; 442c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 443c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 444c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (relative_component.len <= 0) { 445c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Empty relative URL, leave unchanged, only removing the ref component. 446c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int base_len = base_parsed.Length(); 447c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) base_len -= base_parsed.ref.len + 1; 448c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) out_parsed->ref.reset(); 449c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) output->Append(base_url, base_len); 450c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return true; 451c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 452c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 4530529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch int num_slashes = CountConsecutiveSlashes( 454c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) relative_url, relative_component.begin, relative_component.end()); 455c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 456c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#ifdef WIN32 457c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // On Windows, two slashes for a file path (regardless of which direction 458c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // they are) means that it's UNC. Two backslashes on any base scheme mean 459c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // that it's an absolute UNC path (we use the base_is_file flag to control 460c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // how strict the UNC finder is). 461c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // 462c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // We also allow Windows absolute drive specs on any scheme (for example 463c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // "c:\foo") like IE does. There must be no preceeding slashes in this 464c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // case (we reject anything like "/c:/foo") because that should be treated 465c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // as a path. For file URLs, we allow any number of slashes since that would 466c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // be setting the path. 467c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // 468c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // This assumes the absolute path resolver handles absolute URLs like this 4690529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch // properly. DoCanonicalize does this. 470c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int after_slashes = relative_component.begin + num_slashes; 4710529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch if (DoesBeginUNCPath(relative_url, relative_component.begin, 4720529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch relative_component.end(), !base_is_file) || 473c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) ((num_slashes == 0 || base_is_file) && 4740529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch DoesBeginWindowsDriveSpec( 4750529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch relative_url, after_slashes, relative_component.end()))) { 476c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return DoResolveAbsoluteFile(relative_url, relative_component, 477c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) query_converter, output, out_parsed); 478c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 479c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#else 480c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Other platforms need explicit handling for file: URLs with multiple 481c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // slashes because the generic scheme parsing always extracts a host, but a 482d0247b1b59f9c528cb6df88b4f2b9afaf80d181eTorne (Richard Coles) // file: URL only has a host if it has exactly 2 slashes. Even if it does 483d0247b1b59f9c528cb6df88b4f2b9afaf80d181eTorne (Richard Coles) // have a host, we want to use the special host detection logic for file 484d0247b1b59f9c528cb6df88b4f2b9afaf80d181eTorne (Richard Coles) // URLs provided by DoResolveAbsoluteFile(), as opposed to the generic host 485d0247b1b59f9c528cb6df88b4f2b9afaf80d181eTorne (Richard Coles) // detection logic, for consistency with parsing file URLs from scratch. 486d0247b1b59f9c528cb6df88b4f2b9afaf80d181eTorne (Richard Coles) // This also handles the special case where the URL is only slashes, 487d0247b1b59f9c528cb6df88b4f2b9afaf80d181eTorne (Richard Coles) // since that doesn't have a host part either. 488c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (base_is_file && 489d0247b1b59f9c528cb6df88b4f2b9afaf80d181eTorne (Richard Coles) (num_slashes >= 2 || num_slashes == relative_component.len)) { 490c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return DoResolveAbsoluteFile(relative_url, relative_component, 491c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) query_converter, output, out_parsed); 492c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 493c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)#endif 494c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 495c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // Any other double-slashes mean that this is relative to the scheme. 496c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) if (num_slashes >= 2) { 497c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return DoResolveRelativeHost(base_url, base_parsed, 498c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) relative_url, relative_component, 499c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) query_converter, output, out_parsed); 500c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) } 501c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 502c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) // When we get here, we know that the relative URL is on the same host. 503c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return DoResolveRelativePath(base_url, base_parsed, base_is_file, 504c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) relative_url, relative_component, 505c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) query_converter, output, out_parsed); 506c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 507c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 508c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} // namespace 509c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 510c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)bool IsRelativeURL(const char* base, 5110529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch const Parsed& base_parsed, 512c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* fragment, 513c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int fragment_len, 514c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool is_base_hierarchical, 515c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool* is_relative, 5160529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch Component* relative_component) { 517c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return DoIsRelativeURL<char>( 518c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) base, base_parsed, fragment, fragment_len, is_base_hierarchical, 519c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) is_relative, relative_component); 520c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 521c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 522c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)bool IsRelativeURL(const char* base, 5230529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch const Parsed& base_parsed, 5247d4cd473f85ac64c3747c96c277f9e506a0d2246Torne (Richard Coles) const base::char16* fragment, 525c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) int fragment_len, 526c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool is_base_hierarchical, 527c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool* is_relative, 5280529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch Component* relative_component) { 5297d4cd473f85ac64c3747c96c277f9e506a0d2246Torne (Richard Coles) return DoIsRelativeURL<base::char16>( 530c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) base, base_parsed, fragment, fragment_len, is_base_hierarchical, 531c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) is_relative, relative_component); 532c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 533c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 534c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)bool ResolveRelativeURL(const char* base_url, 5350529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch const Parsed& base_parsed, 536c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool base_is_file, 537c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) const char* relative_url, 5380529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch const Component& relative_component, 539c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) CharsetConverter* query_converter, 540c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) CanonOutput* output, 5410529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch Parsed* out_parsed) { 542c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) return DoResolveRelativeURL<char>( 543c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) base_url, base_parsed, base_is_file, relative_url, 544c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) relative_component, query_converter, output, out_parsed); 545c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 546c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 547c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)bool ResolveRelativeURL(const char* base_url, 5480529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch const Parsed& base_parsed, 549c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) bool base_is_file, 5507d4cd473f85ac64c3747c96c277f9e506a0d2246Torne (Richard Coles) const base::char16* relative_url, 5510529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch const Component& relative_component, 552c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) CharsetConverter* query_converter, 553c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) CanonOutput* output, 5540529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch Parsed* out_parsed) { 5557d4cd473f85ac64c3747c96c277f9e506a0d2246Torne (Richard Coles) return DoResolveRelativeURL<base::char16>( 556c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) base_url, base_parsed, base_is_file, relative_url, 557c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) relative_component, query_converter, output, out_parsed); 558c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles)} 559c2e0dbddbe15c98d52c4786dac06cb8952a8ae6dTorne (Richard Coles) 5600529e5d033099cbfc42635f6f6183833b09dff6eBen Murdoch} // namespace url 561