1 // Copyright (c) 2012 The Chromium Authors. All rights reserved. 2 // Use of this source code is governed by a BSD-style license that can be 3 // found in the LICENSE file. 4 5 #ifndef CHROME_BROWSER_AUTOCOMPLETE_AUTOCOMPLETE_INPUT_H_ 6 #define CHROME_BROWSER_AUTOCOMPLETE_AUTOCOMPLETE_INPUT_H_ 7 8 #include <string> 9 10 #include "base/basictypes.h" 11 #include "base/gtest_prod_util.h" 12 #include "base/strings/string16.h" 13 #include "url/gurl.h" 14 #include "url/url_parse.h" 15 16 // The user input for an autocomplete query. Allows copying. 17 class AutocompleteInput { 18 public: 19 // Note that the type below may be misleading. For example, "http:/" alone 20 // cannot be opened as a URL, so it is marked as a QUERY; yet the user 21 // probably intends to type more and have it eventually become a URL, so we 22 // need to make sure we still run it through inline autocomplete. 23 enum Type { 24 INVALID, // Empty input 25 UNKNOWN, // Valid input whose type cannot be determined 26 URL, // Input autodetected as a URL 27 QUERY, // Input autodetected as a query 28 FORCED_QUERY, // Input forced to be a query by an initial '?' 29 }; 30 31 // Enumeration of the possible match query types. Callers who only need some 32 // of the matches for a particular input can get answers more quickly by 33 // specifying that upfront. 34 enum MatchesRequested { 35 // Only the best match in the whole result set matters. Providers should at 36 // most return synchronously-available matches, and if possible do even less 37 // work, so that it's safe to ask for these repeatedly in the course of one 38 // higher-level "synchronous" query. 39 BEST_MATCH, 40 41 // Only synchronous matches should be returned. 42 SYNCHRONOUS_MATCHES, 43 44 // All matches should be fetched. 45 ALL_MATCHES, 46 }; 47 48 // The type of page currently displayed. 49 // Note: when adding an element to this enum, please add it at the end 50 // and update omnibox_event.proto::PageClassification and 51 // omnibox_edit_model.cc::ClassifyPage() too. 52 enum PageClassification { 53 // An invalid URL; shouldn't happen. 54 INVALID_SPEC = 0, 55 56 // chrome://newtab/. This can be either the built-in version or a 57 // replacement new tab page from an extension. Note that when Instant 58 // Extended is enabled, the new tab page will be reported as either 59 // INSTANT_NTP_WITH_OMNIBOX_AS_STARTING_FOCUS or 60 // INSTANT_NTP_WITH_FAKEBOX_AS_STARTING_FOCUS below, 61 // unless an extension is replacing the new tab page, in which case 62 // it will still be reported as NTP. 63 NTP = 1, 64 65 // about:blank. 66 BLANK = 2, 67 68 // The user's home page. Note that if the home page is set to any 69 // of the new tab page versions or to about:blank, then we'll 70 // classify the page into those categories, not HOME_PAGE. 71 HOME_PAGE = 3, 72 73 // The catch-all entry of everything not included somewhere else 74 // on this list. 75 OTHER = 4, 76 77 // The user is on a search result page that's doing search term 78 // replacement, meaning the search terms should've appeared in the omnibox 79 // before the user started editing it, not the URL of the page. 80 SEARCH_RESULT_PAGE_DOING_SEARCH_TERM_REPLACEMENT = 6, 81 82 // The new tab page in which this omnibox interaction first started 83 // with the user having focus in the omnibox. 84 INSTANT_NTP_WITH_OMNIBOX_AS_STARTING_FOCUS = 7, 85 86 // The new tab page in which this omnibox interaction first started 87 // with the user having focus in the fakebox. 88 INSTANT_NTP_WITH_FAKEBOX_AS_STARTING_FOCUS = 8, 89 90 // The user is on a search result page that's not doing search term 91 // replacement, meaning the URL of the page should've appeared in the 92 // omnibox before the user started editing it, not the search terms. 93 SEARCH_RESULT_PAGE_NO_SEARCH_TERM_REPLACEMENT = 9 94 }; 95 96 AutocompleteInput(); 97 // |text| and |cursor_position| represent the input query and location of 98 // the cursor with the query respectively. |cursor_position| may be set to 99 // base::string16::npos if the input |text| doesn't come directly from the 100 // user's typing. 101 // 102 // |desired_tld| is the user's desired TLD, if one is not already present in 103 // the text to autocomplete. When this is non-empty, it also implies that 104 // "www." should be prepended to the domain where possible. The |desired_tld| 105 // should not contain a leading '.' (use "com" instead of ".com"). 106 // 107 // If |current_url| is set to a valid search result page URL, providers can 108 // use it to perform query refinement. For example, if it is set to an image 109 // search result page, the search provider may generate an image search URL. 110 // Query refinement is only used by mobile ports, so only these set 111 // |current_url| to a non-empty string. 112 // 113 // |current_page_classification| represents the type of page the user is 114 // viewing and manner in which the user is accessing the omnibox; it's 115 // more than simply the URL. It includes, for example, whether the page 116 // is a search result page doing search term replacement or not. 117 // 118 // |prevent_inline_autocomplete| is true if the generated result set should 119 // not require inline autocomplete for the default match. This is difficult 120 // to explain in the abstract; the practical use case is that after the user 121 // deletes text in the edit, the HistoryURLProvider should make sure not to 122 // promote a match requiring inline autocomplete too highly. 123 // 124 // |prefer_keyword| should be true when the keyword UI is onscreen; this will 125 // bias the autocomplete result set toward the keyword provider when the input 126 // string is a bare keyword. 127 // 128 // |allow_exact_keyword_match| should be false when triggering keyword mode on 129 // the input string would be surprising or wrong, e.g. when highlighting text 130 // in a page and telling the browser to search for it or navigate to it. This 131 // parameter only applies to substituting keywords. 132 // 133 // If |matches_requested| is BEST_MATCH or SYNCHRONOUS_MATCHES the controller 134 // asks the providers to only return matches which are synchronously 135 // available, which should mean that all providers will be done immediately. 136 AutocompleteInput(const base::string16& text, 137 size_t cursor_position, 138 const base::string16& desired_tld, 139 const GURL& current_url, 140 PageClassification current_page_classification, 141 bool prevent_inline_autocomplete, 142 bool prefer_keyword, 143 bool allow_exact_keyword_match, 144 MatchesRequested matches_requested); 145 ~AutocompleteInput(); 146 147 // If type is |FORCED_QUERY| and |text| starts with '?', it is removed. 148 // Returns number of leading characters removed. 149 static size_t RemoveForcedQueryStringIfNecessary(Type type, 150 base::string16* text); 151 152 // Converts |type| to a string representation. Used in logging. 153 static std::string TypeToString(Type type); 154 155 // Parses |text| and returns the type of input this will be interpreted as. 156 // The components of the input are stored in the output parameter |parts|, if 157 // it is non-NULL. The scheme is stored in |scheme| if it is non-NULL. The 158 // canonicalized URL is stored in |canonicalized_url|; however, this URL is 159 // not guaranteed to be valid, especially if the parsed type is, e.g., QUERY. 160 static Type Parse(const base::string16& text, 161 const base::string16& desired_tld, 162 url_parse::Parsed* parts, 163 base::string16* scheme, 164 GURL* canonicalized_url); 165 166 // Parses |text| and fill |scheme| and |host| by the positions of them. 167 // The results are almost as same as the result of Parse(), but if the scheme 168 // is view-source, this function returns the positions of scheme and host 169 // in the URL qualified by "view-source:" prefix. 170 static void ParseForEmphasizeComponents(const base::string16& text, 171 url_parse::Component* scheme, 172 url_parse::Component* host); 173 174 // Code that wants to format URLs with a format flag including 175 // net::kFormatUrlOmitTrailingSlashOnBareHostname risk changing the meaning if 176 // the result is then parsed as AutocompleteInput. Such code can call this 177 // function with the URL and its formatted string, and it will return a 178 // formatted string with the same meaning as the original URL (i.e. it will 179 // re-append a slash if necessary). 180 static base::string16 FormattedStringWithEquivalentMeaning( 181 const GURL& url, 182 const base::string16& formatted_url); 183 184 // Returns the number of non-empty components in |parts| besides the host. 185 static int NumNonHostComponents(const url_parse::Parsed& parts); 186 187 // Returns whether |text| begins "http:" or "view-source:http:". 188 static bool HasHTTPScheme(const base::string16& text); 189 190 // User-provided text to be completed. 191 const base::string16& text() const { return text_; } 192 193 // Returns 0-based cursor position within |text_| or base::string16::npos if 194 // not used. 195 size_t cursor_position() const { return cursor_position_; } 196 197 // Use of this setter is risky, since no other internal state is updated 198 // besides |text_|, |cursor_position_| and |parts_|. Only callers who know 199 // that they're not changing the type/scheme/etc. should use this. 200 void UpdateText(const base::string16& text, 201 size_t cursor_position, 202 const url_parse::Parsed& parts); 203 204 // The current URL, or an invalid GURL if query refinement is not desired. 205 const GURL& current_url() const { return current_url_; } 206 207 // The type of page that is currently behind displayed and how it is 208 // displayed (e.g., with search term replacement or without). 209 AutocompleteInput::PageClassification current_page_classification() const { 210 return current_page_classification_; 211 } 212 213 // The type of input supplied. 214 Type type() const { return type_; } 215 216 // Returns parsed URL components. 217 const url_parse::Parsed& parts() const { return parts_; } 218 219 // The scheme parsed from the provided text; only meaningful when type_ is 220 // URL. 221 const base::string16& scheme() const { return scheme_; } 222 223 // The input as an URL to navigate to, if possible. 224 const GURL& canonicalized_url() const { return canonicalized_url_; } 225 226 // Returns whether inline autocompletion should be prevented. 227 bool prevent_inline_autocomplete() const { 228 return prevent_inline_autocomplete_; 229 } 230 231 // Returns whether, given an input string consisting solely of a substituting 232 // keyword, we should score it like a non-substituting keyword. 233 bool prefer_keyword() const { return prefer_keyword_; } 234 235 // Returns whether this input is allowed to be treated as an exact 236 // keyword match. If not, the default result is guaranteed not to be a 237 // keyword search, even if the input is "<keyword> <search string>". 238 bool allow_exact_keyword_match() const { return allow_exact_keyword_match_; } 239 240 // See description of enum for details. 241 MatchesRequested matches_requested() const { return matches_requested_; } 242 243 // Resets all internal variables to the null-constructed state. 244 void Clear(); 245 246 private: 247 FRIEND_TEST_ALL_PREFIXES(AutocompleteProviderTest, GetDestinationURL); 248 249 // NOTE: Whenever adding a new field here, please make sure to update Clear() 250 // method. 251 base::string16 text_; 252 size_t cursor_position_; 253 GURL current_url_; 254 AutocompleteInput::PageClassification current_page_classification_; 255 Type type_; 256 url_parse::Parsed parts_; 257 base::string16 scheme_; 258 GURL canonicalized_url_; 259 bool prevent_inline_autocomplete_; 260 bool prefer_keyword_; 261 bool allow_exact_keyword_match_; 262 MatchesRequested matches_requested_; 263 }; 264 265 #endif // CHROME_BROWSER_AUTOCOMPLETE_AUTOCOMPLETE_INPUT_H_ 266
