1c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// Copyright 2007, Google Inc. 2c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// All rights reserved. 3c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// 4c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// Redistribution and use in source and binary forms, with or without 5c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// modification, are permitted provided that the following conditions are 6c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// met: 7c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// 8c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// * Redistributions of source code must retain the above copyright 9c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// notice, this list of conditions and the following disclaimer. 10c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// * Redistributions in binary form must reproduce the above 11c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// copyright notice, this list of conditions and the following disclaimer 12c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// in the documentation and/or other materials provided with the 13c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// distribution. 14c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// * Neither the name of Google Inc. nor the names of its 15c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// contributors may be used to endorse or promote products derived from 16c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// this software without specific prior written permission. 17c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// 18c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS 19c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT 20c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR 21c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT 22c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, 23c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT 24c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 25c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 26c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 27c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE 28c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 29c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 30c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#ifndef GOOGLEURL_SRC_GURL_H__ 31c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#define GOOGLEURL_SRC_GURL_H__ 32c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 333345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick#include <iosfwd> 34c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include <string> 35c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 36c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include "base/string16.h" 37c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include "googleurl/src/url_canon.h" 38c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include "googleurl/src/url_canon_stdstring.h" 39c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch#include "googleurl/src/url_common.h" 40c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#include "googleurl/src/url_parse.h" 41c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 42c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scottclass GURL { 43c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott public: 44c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott typedef url_canon::StdStringReplacements<std::string> Replacements; 45c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott typedef url_canon::StdStringReplacements<string16> ReplacementsW; 46c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 47c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Creates an empty, invalid URL. 48c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch GURL_API GURL(); 49c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 50c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Copy construction is relatively inexpensive, with most of the time going 51c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // to reallocating the string. It does not re-parse. 52c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch GURL_API GURL(const GURL& other); 53c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 54c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // The narrow version requires the input be UTF-8. Invalid UTF-8 input will 55c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // result in an invalid URL. 56c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // 57c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // The wide version should also take an encoding parameter so we know how to 58c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // encode the query parameters. It is probably sufficient for the narrow 59c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // version to assume the query parameter encoding should be the same as the 60c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // input encoding. 61c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch GURL_API explicit GURL(const std::string& url_string 62c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch /*, output_param_encoding*/); 63c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch GURL_API explicit GURL(const string16& url_string 64c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch /*, output_param_encoding*/); 65c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 66c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Constructor for URLs that have already been parsed and canonicalized. This 67c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // is used for conversions from KURL, for example. The caller must supply all 68c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // information associated with the URL, which must be correct and consistent. 69c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch GURL_API GURL(const char* canonical_spec, size_t canonical_spec_len, 70c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch const url_parse::Parsed& parsed, bool is_valid); 71c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 72513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch GURL_API GURL& operator=(const GURL& other); 73513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch 74c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Returns true when this object represents a valid parsed URL. When not 75c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // valid, other functions will still succeed, but you will not get canonical 76c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // data out in the format you may be expecting. Instead, we keep something 77c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // "reasonable looking" so that the user can see how it's busted if 78c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // displayed to them. 79c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott bool is_valid() const { 80c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return is_valid_; 81c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 82c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 83c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Returns true if the URL is zero-length. Note that empty URLs are also 84c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // invalid, and is_valid() will return false for them. This is provided 85c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // because some users may want to treat the empty case differently. 86c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott bool is_empty() const { 87c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return spec_.empty(); 88c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 89c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 90c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Returns the raw spec, i.e., the full text of the URL, in canonical UTF-8, 91c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // if the URL is valid. If the URL is not valid, this will assert and return 92c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // the empty string (for safety in release builds, to keep them from being 93c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // misused which might be a security problem). 94c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // 95c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // The URL will be ASCII except the reference fragment, which may be UTF-8. 96c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // It is guaranteed to be valid UTF-8. 97c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // 98c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // The exception is for empty() URLs (which are !is_valid()) but this will 99c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // return the empty string without asserting. 100c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // 101c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Used invalid_spec() below to get the unusable spec of an invalid URL. This 102c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // separation is designed to prevent errors that may cause security problems 103c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // that could result from the mistaken use of an invalid URL. 104c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch GURL_API const std::string& spec() const; 105c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 106c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Returns the potentially invalid spec for a the URL. This spec MUST NOT be 107c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // modified or sent over the network. It is designed to be displayed in error 108c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // messages to the user, as the apperance of the spec may explain the error. 109c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // If the spec is valid, the valid spec will be returned. 110c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // 111c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // The returned string is guaranteed to be valid UTF-8. 112c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott const std::string& possibly_invalid_spec() const { 113c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return spec_; 114c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 115c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 116c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Getter for the raw parsed structure. This allows callers to locate parts 117c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // of the URL within the spec themselves. Most callers should consider using 118c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // the individual component getters below. 119c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // 120c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // The returned parsed structure will reference into the raw spec, which may 121c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // or may not be valid. If you are using this to index into the spec, BE 122c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // SURE YOU ARE USING possibly_invalid_spec() to get the spec, and that you 123c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // don't do anything "important" with invalid specs. 124c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott const url_parse::Parsed& parsed_for_possibly_invalid_spec() const { 125c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return parsed_; 126c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 127c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 128c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Defiant equality operator! 129c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott bool operator==(const GURL& other) const { 130c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return spec_ == other.spec_; 131c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 132c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott bool operator!=(const GURL& other) const { 133c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return spec_ != other.spec_; 134c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 135c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 136c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Allows GURL to used as a key in STL (for example, a std::set or std::map). 137c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott bool operator<(const GURL& other) const { 138c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return spec_ < other.spec_; 139c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 140c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 141c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Resolves a URL that's possibly relative to this object's URL, and returns 142c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // it. Absolute URLs are also handled according to the rules of URLs on web 143c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // pages. 144c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // 145c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // It may be impossible to resolve the URLs properly. If the input is not 146c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // "standard" (SchemeIsStandard() == false) and the input looks relative, we 147c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // can't resolve it. In these cases, the result will be an empty, invalid 148c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // GURL. 149c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // 150c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // The result may also be a nonempty, invalid URL if the input has some kind 151c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // of encoding error. In these cases, we will try to construct a "good" URL 152c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // that may have meaning to the user, but it will be marked invalid. 153c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // 154c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // It is an error to resolve a URL relative to an invalid URL. The result 155c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // will be the empty URL. 156c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch GURL_API GURL Resolve(const std::string& relative) const; 157c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch GURL_API GURL Resolve(const string16& relative) const; 158c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 159c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Like Resolve() above but takes a character set encoder which will be used 160c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // for any query text specified in the input. The charset converter parameter 161c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // may be NULL, in which case it will be treated as UTF-8. 162c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // 163c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // TODO(brettw): These should be replaced with versions that take something 164c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // more friendly than a raw CharsetConverter (maybe like an ICU character set 165c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // name). 166c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch GURL_API GURL ResolveWithCharsetConverter( 167c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott const std::string& relative, 168c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott url_canon::CharsetConverter* charset_converter) const; 169c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch GURL_API GURL ResolveWithCharsetConverter( 170c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott const string16& relative, 171c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott url_canon::CharsetConverter* charset_converter) const; 172c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 173c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Creates a new GURL by replacing the current URL's components with the 174c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // supplied versions. See the Replacements class in url_canon.h for more. 175c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // 176c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // These are not particularly quick, so avoid doing mutations when possible. 177c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Prefer the 8-bit version when possible. 178c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // 179c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // It is an error to replace components of an invalid URL. The result will 180c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // be the empty URL. 181c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // 182c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Note that we use the more general url_canon::Replacements type to give 183c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // callers extra flexibility rather than our override. 184c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch GURL_API GURL ReplaceComponents( 185c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott const url_canon::Replacements<char>& replacements) const; 186c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch GURL_API GURL ReplaceComponents( 187c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott const url_canon::Replacements<char16>& replacements) const; 188c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 189c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // A helper function that is equivalent to replacing the path with a slash 190c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // and clearing out everything after that. We sometimes need to know just the 191c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // scheme and the authority. If this URL is not a standard URL (it doesn't 192c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // have the regular authority and path sections), then the result will be 193c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // an empty, invalid GURL. Note that this *does* work for file: URLs, which 194c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // some callers may want to filter out before calling this. 195c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // 196c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // It is an error to get an empty path on an invalid URL. The result 197c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // will be the empty URL. 198c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch GURL_API GURL GetWithEmptyPath() const; 199c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 200c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // A helper function to return a GURL containing just the scheme, host, 201c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // and port from a URL. Equivalent to clearing any username and password, 202c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // replacing the path with a slash, and clearing everything after that. If 203c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // this URL is not a standard URL, then the result will be an empty, 204c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // invalid GURL. If the URL has neither username nor password, this 205c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // degenerates to GetWithEmptyPath(). 206c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // 207c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // It is an error to get the origin of an invalid URL. The result 208c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // will be the empty URL. 209c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch GURL_API GURL GetOrigin() const; 210c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 211c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Returns true if the scheme for the current URL is a known "standard" 212c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // scheme. Standard schemes have an authority and a path section. This 213c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // includes file:, which some callers may want to filter out explicitly by 214c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // calling SchemeIsFile. 215c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch GURL_API bool IsStandard() const; 216c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 217c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Returns true if the given parameter (should be lower-case ASCII to match 218c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // the canonicalized scheme) is the scheme for this URL. This call is more 219c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // efficient than getting the scheme and comparing it because no copies or 220c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // object constructions are done. 221c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch GURL_API bool SchemeIs(const char* lower_ascii_scheme) const; 222c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 223c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // We often need to know if this is a file URL. File URLs are "standard", but 224c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // are often treated separately by some programs. 225c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott bool SchemeIsFile() const { 226c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return SchemeIs("file"); 227c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 228c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 229c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // If the scheme indicates a secure connection 230c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott bool SchemeIsSecure() const { 231c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return SchemeIs("https"); 232c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 233c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 234c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Returns true if the hostname is an IP address. Note: this function isn't 235c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // as cheap as a simple getter because it re-parses the hostname to verify. 236c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // This currently identifies only IPv4 addresses (bug 822685). 237c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch GURL_API bool HostIsIPAddress() const; 238c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 239c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Getters for various components of the URL. The returned string will be 240c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // empty if the component is empty or is not present. 241c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott std::string scheme() const { // Not including the colon. See also SchemeIs. 242c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return ComponentString(parsed_.scheme); 243c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 244c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott std::string username() const { 245c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return ComponentString(parsed_.username); 246c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 247c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott std::string password() const { 248c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return ComponentString(parsed_.password); 249c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 250c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Note that this may be a hostname, an IPv4 address, or an IPv6 literal 251c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // surrounded by square brackets, like "[2001:db8::1]". To exclude these 252c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // brackets, use HostNoBrackets() below. 253c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott std::string host() const { 254c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return ComponentString(parsed_.host); 255c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 256c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott std::string port() const { // Returns -1 if "default" 257c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return ComponentString(parsed_.port); 258c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 259c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott std::string path() const { // Including first slash following host 260c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return ComponentString(parsed_.path); 261c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 262c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott std::string query() const { // Stuff following '?' 263c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return ComponentString(parsed_.query); 264c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 265c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott std::string ref() const { // Stuff following '#' 266c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return ComponentString(parsed_.ref); 267c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 268c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 269c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Existance querying. These functions will return true if the corresponding 270c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // URL component exists in this URL. Note that existance is different than 271c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // being nonempty. http://www.google.com/? has a query that just happens to 272c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // be empty, and has_query() will return true. 273c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott bool has_scheme() const { 274c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return parsed_.scheme.len >= 0; 275c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 276c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott bool has_username() const { 277c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return parsed_.username.len >= 0; 278c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 279c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott bool has_password() const { 280c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return parsed_.password.len >= 0; 281c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 282c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott bool has_host() const { 283c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Note that hosts are special, absense of host means length 0. 284c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return parsed_.host.len > 0; 285c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 286c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott bool has_port() const { 287c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return parsed_.port.len >= 0; 288c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 289c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott bool has_path() const { 290c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Note that http://www.google.com/" has a path, the path is "/". This can 291c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // return false only for invalid or nonstandard URLs. 292c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return parsed_.path.len >= 0; 293c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 294c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott bool has_query() const { 295c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return parsed_.query.len >= 0; 296c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 297c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott bool has_ref() const { 298c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return parsed_.ref.len >= 0; 299c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 300c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 301c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Returns a parsed version of the port. Can also be any of the special 302c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // values defined in Parsed for ExtractPort. 303c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch GURL_API int IntPort() const; 304c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 305c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Returns the port number of the url, or the default port number. 306c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // If the scheme has no concept of port (or unknown default) returns 307c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // PORT_UNSPECIFIED. 308c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch GURL_API int EffectiveIntPort() const; 309c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 310c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Extracts the filename portion of the path and returns it. The filename 311c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // is everything after the last slash in the path. This may be empty. 312c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch GURL_API std::string ExtractFileName() const; 313c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 314c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Returns the path that should be sent to the server. This is the path, 315c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // parameter, and query portions of the URL. It is guaranteed to be ASCII. 316c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch GURL_API std::string PathForRequest() const; 317c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 318c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Returns the host, excluding the square brackets surrounding IPv6 address 319c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // literals. This can be useful for passing to getaddrinfo(). 320c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch GURL_API std::string HostNoBrackets() const; 321c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 322c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Returns true if this URL's host matches or is in the same domain as 323c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // the given input string. For example if this URL was "www.google.com", 324c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // this would match "com", "google.com", and "www.google.com 325c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // (input domain should be lower-case ASCII to match the canonicalized 326c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // scheme). This call is more efficient than getting the host and check 327c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // whether host has the specific domain or not because no copies or 328c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // object constructions are done. 329c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // 330c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // If function DomainIs has parameter domain_len, which means the parameter 331c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // lower_ascii_domain does not gurantee to terminate with NULL character. 332c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch GURL_API bool DomainIs(const char* lower_ascii_domain, int domain_len) const; 333c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 334c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // If function DomainIs only has parameter lower_ascii_domain, which means 335c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // domain string should be terminate with NULL character. 336c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott bool DomainIs(const char* lower_ascii_domain) const { 337c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return DomainIs(lower_ascii_domain, 338c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott static_cast<int>(strlen(lower_ascii_domain))); 339c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 340c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 341c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Swaps the contents of this GURL object with the argument without doing 342c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // any memory allocations. 343c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch GURL_API void Swap(GURL* other); 344c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 345c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Returns a reference to a singleton empty GURL. This object is for callers 346c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // who return references but don't have anything to return in some cases. 347c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // This function may be called from any thread. 348c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch GURL_API static const GURL& EmptyGURL(); 349c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 350c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott private: 351c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Returns the substring of the input identified by the given component. 352c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott std::string ComponentString(const url_parse::Component& comp) const { 353c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott if (comp.len <= 0) 354c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return std::string(); 355c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott return std::string(spec_, comp.begin, comp.len); 356c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott } 357c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 358c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // The actual text of the URL, in canonical ASCII form. 359c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott std::string spec_; 360c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 361c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Set when the given URL is valid. Otherwise, we may still have a spec and 362c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // components, but they may not identify valid resources (for example, an 363c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // invalid port number, invalid characters in the scheme, etc.). 364c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott bool is_valid_; 365c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 366c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // Identified components of the canonical spec. 367c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott url_parse::Parsed parsed_; 368c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 369c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott // TODO bug 684583: Add encoding for query params. 370c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott}; 371c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 372c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott// Stream operator so GURL can be used in assertion statements. 37321d179b334e59e9a3bfcaed4c4430bef1bc5759dKristian MonsenGURL_API std::ostream& operator<<(std::ostream& out, const GURL& url); 374c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott 375c7f5f8508d98d5952d42ed7648c2a8f30a4da156Patrick Scott#endif // GOOGLEURL_SRC_GURL_H__ 376