1 // Copyright 2014 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.
5 #ifndef COMPONENTS_OMNIBOX_BROWSER_AUTOCOMPLETE_MATCH_H_
6 #define COMPONENTS_OMNIBOX_BROWSER_AUTOCOMPLETE_MATCH_H_
12 #include "base/memory/scoped_ptr.h"
13 #include "components/omnibox/browser/autocomplete_input.h"
14 #include "components/omnibox/browser/autocomplete_match_type.h"
15 #include "components/search_engines/template_url.h"
16 #include "ui/base/page_transition_types.h"
19 class AutocompleteProvider
;
20 class SuggestionAnswer
;
22 class TemplateURLService
;
28 const char kACMatchPropertyInputText
[] = "input text";
29 const char kACMatchPropertyContentsPrefix
[] = "match contents prefix";
30 const char kACMatchPropertyContentsStartIndex
[] = "match contents start index";
32 // AutocompleteMatch ----------------------------------------------------------
34 // A single result line with classified spans. The autocomplete popup displays
35 // the 'contents' and the 'description' (the description is optional) in the
36 // autocomplete dropdown, and fills in 'fill_into_edit' into the textbox when
37 // that line is selected. fill_into_edit may be the same as 'description' for
38 // things like URLs, but may be different for searches or other providers. For
39 // example, a search result may say "Search for asdf" as the description, but
40 // "asdf" should appear in the box.
41 struct AutocompleteMatch
{
42 // Autocomplete matches contain strings that are classified according to a
43 // separate vector of styles. This vector associates flags with particular
44 // string segments, and must be in sorted order. All text must be associated
45 // with some kind of classification. Even if a match has no distinct
46 // segments, its vector should contain an entry at offset 0 with no flags.
48 // Example: The user typed "goog"
49 // http://www.google.com/ Google
54 // This structure holds the classification information for each span.
55 struct ACMatchClassification
{
56 // The values in here are not mutually exclusive -- use them like a
57 // bitfield. This also means we use "int" instead of this enum type when
58 // passing the values around, so the compiler doesn't complain.
61 URL
= 1 << 0, // A URL
62 MATCH
= 1 << 1, // A match for the user's search term
63 DIM
= 1 << 2, // "Helper text"
66 ACMatchClassification(size_t offset
, int style
)
71 // Offset within the string that this classification starts
77 typedef std::vector
<ACMatchClassification
> ACMatchClassifications
;
79 // Type used by providers to attach additional, optional information to
80 // an AutocompleteMatch.
81 typedef std::map
<std::string
, std::string
> AdditionalInfo
;
83 // The type of this match.
84 typedef AutocompleteMatchType::Type Type
;
86 // Null-terminated array of characters that are not valid within |contents|
87 // and |description| strings.
88 static const base::char16 kInvalidChars
[];
91 AutocompleteMatch(AutocompleteProvider
* provider
,
95 AutocompleteMatch(const AutocompleteMatch
& match
);
98 // Converts |type| to a string representation. Used in logging and debugging.
99 AutocompleteMatch
& operator=(const AutocompleteMatch
& match
);
101 // Converts |type| to a resource identifier for the appropriate icon for this
102 // type to show in the completion popup.
103 static int TypeToIcon(Type type
);
105 // Comparison function for determining when one match is better than another.
106 static bool MoreRelevant(const AutocompleteMatch
& elem1
,
107 const AutocompleteMatch
& elem2
);
109 // Comparison function for removing matches with duplicate destinations.
110 // Destinations are compared using |stripped_destination_url|. Pairs of
111 // matches with empty destinations are treated as differing, since empty
112 // destinations are expected for non-navigable matches.
113 static bool DestinationsEqual(const AutocompleteMatch
& elem1
,
114 const AutocompleteMatch
& elem2
);
116 // Helper functions for classes creating matches:
117 // Fills in the classifications for |text|, using |style| as the base style
118 // and marking the first instance of |find_text| as a match. (This match
119 // will also not be dimmed, if |style| has DIM set.)
120 static void ClassifyMatchInString(const base::string16
& find_text
,
121 const base::string16
& text
,
123 ACMatchClassifications
* classifications
);
125 // Similar to ClassifyMatchInString(), but for cases where the range to mark
126 // as matching is already known (avoids calling find()). This can be helpful
127 // when find() would be misleading (e.g. you want to mark the second match in
128 // a string instead of the first).
129 static void ClassifyLocationInString(size_t match_location
,
131 size_t overall_length
,
133 ACMatchClassifications
* classifications
);
135 // Returns a new vector of classifications containing the merged contents of
136 // |classifications1| and |classifications2|.
137 static ACMatchClassifications
MergeClassifications(
138 const ACMatchClassifications
& classifications1
,
139 const ACMatchClassifications
& classifications2
);
141 // Converts classifications to and from a serialized string representation
142 // (using comma-separated integers to sequentially list positions and styles).
143 static std::string
ClassificationsToString(
144 const ACMatchClassifications
& classifications
);
145 static ACMatchClassifications
ClassificationsFromString(
146 const std::string
& serialized_classifications
);
148 // Adds a classification to the end of |classifications| iff its style is
149 // different from the last existing classification. |offset| must be larger
150 // than the offset of the last classification in |classifications|.
151 static void AddLastClassificationIfNecessary(
152 ACMatchClassifications
* classifications
,
156 // Returns true if at least one style in |classifications| is of type MATCH.
157 static bool HasMatchStyle(const ACMatchClassifications
& classifications
);
159 // Removes invalid characters from |text|. Should be called on strings coming
160 // from external sources (such as extensions) before assigning to |contents|
162 static base::string16
SanitizeString(const base::string16
& text
);
164 // Convenience function to check if |type| is a search (as opposed to a URL or
166 static bool IsSearchType(Type type
);
168 // Convenience function to check if |type| is a special search suggest type -
169 // like entity, personalized, profile or postfix.
170 static bool IsSpecializedSearchType(Type type
);
172 // A static version GetTemplateURL() that takes the match's keyword and
173 // match's hostname as parameters. In short, returns the TemplateURL
174 // associated with |keyword| if it exists; otherwise returns the TemplateURL
175 // associated with |host| if it exists.
176 static TemplateURL
* GetTemplateURLWithKeyword(
177 TemplateURLService
* template_url_service
,
178 const base::string16
& keyword
,
179 const std::string
& host
);
181 // Returns |url| altered by stripping off "www.", converting https protocol
182 // to http, and stripping excess query parameters. These conversions are
183 // merely to allow comparisons to remove likely duplicates; these URLs are
184 // not used as actual destination URLs. If |template_url_service| is not
185 // NULL, it is used to get a template URL corresponding to this match. If
186 // the match's keyword is known, it can be passed in. Otherwise, it can be
187 // left empty and the template URL (if any) is determined from the
188 // destination's hostname. The template URL is used to strip off query args
189 // other than the search terms themselves that would otherwise prevent doing
190 // proper deduping. |input| is used to decide if the scheme is allowed to
191 // be altered during stripping. If this URL, minus the scheme and separator,
192 // starts with any the terms in input.terms_prefixed_by_http_or_https(), we
193 // avoid converting an HTTPS scheme to HTTP. This means URLs that differ
194 // only by these schemes won't be marked as dupes, since the distinction
195 // seems to matter to the user. |languages| is used to format punycoded
196 // domain names to UTF-8 for the aforementioned duplicate detection.
197 static GURL
GURLToStrippedGURL(const GURL
& url
,
198 const AutocompleteInput
& input
,
199 const std::string
& languages
,
200 TemplateURLService
* template_url_service
,
201 const base::string16
& keyword
);
203 // Computes the stripped destination URL (via GURLToStrippedGURL()) and
204 // stores the result in |stripped_destination_url|. |input| and |languages|
205 // are used for the same purpose as in GURLToStrippedGURL().
206 void ComputeStrippedDestinationURL(
207 const AutocompleteInput
& input
,
208 const std::string
& languages
,
209 TemplateURLService
* template_url_service
);
211 // Sets |allowed_to_be_default_match| to true if this match is effectively
212 // the URL-what-you-typed match (i.e., would be dupped against the UWYT
213 // match when AutocompleteResult merges matches). |languages| is used
214 // for the same purpose as in GURLToStrippedGURL().
215 void EnsureUWYTIsAllowedToBeDefault(
216 const AutocompleteInput
& input
,
217 const std::string
& languages
,
218 TemplateURLService
* template_url_service
);
220 // Gets data relevant to whether there should be any special keyword-related
221 // UI shown for this match. If this match represents a selected keyword, i.e.
222 // the UI should be "in keyword mode", |keyword| will be set to the keyword
223 // and |is_keyword_hint| will be set to false. If this match has a non-NULL
224 // |associated_keyword|, i.e. we should show a "Press [tab] to search ___"
225 // hint and allow the user to toggle into keyword mode, |keyword| will be set
226 // to the associated keyword and |is_keyword_hint| will be set to true. Note
227 // that only one of these states can be in effect at once. In all other
228 // cases, |keyword| will be cleared, even when our member variable |keyword|
229 // is non-empty -- such as with non-substituting keywords or matches that
230 // represent searches using the default search engine. See also
231 // GetSubstitutingExplicitlyInvokedKeyword().
232 void GetKeywordUIState(TemplateURLService
* template_url_service
,
233 base::string16
* keyword
,
234 bool* is_keyword_hint
) const;
236 // Returns |keyword|, but only if it represents a substituting keyword that
237 // the user has explicitly invoked. If for example this match represents a
238 // search with the default search engine (and the user didn't explicitly
239 // invoke its keyword), this returns the empty string. The result is that
240 // this function returns a non-empty string in the same cases as when the UI
241 // should show up as being "in keyword mode".
242 base::string16
GetSubstitutingExplicitlyInvokedKeyword(
243 TemplateURLService
* template_url_service
) const;
245 // Returns the TemplateURL associated with this match. This may be NULL if
246 // the match has no keyword OR if the keyword no longer corresponds to a valid
247 // TemplateURL. See comments on |keyword| below.
248 // If |allow_fallback_to_destination_host| is true and the keyword does
249 // not map to a valid TemplateURL, we'll then check for a TemplateURL that
250 // corresponds to the destination_url's hostname.
251 TemplateURL
* GetTemplateURL(TemplateURLService
* template_url_service
,
252 bool allow_fallback_to_destination_host
) const;
254 // Adds optional information to the |additional_info| dictionary.
255 void RecordAdditionalInfo(const std::string
& property
,
256 const std::string
& value
);
257 void RecordAdditionalInfo(const std::string
& property
, int value
);
258 void RecordAdditionalInfo(const std::string
& property
,
259 const base::Time
& value
);
261 // Returns the value recorded for |property| in the |additional_info|
262 // dictionary. Returns the empty string if no such value exists.
263 std::string
GetAdditionalInfo(const std::string
& property
) const;
265 // Returns whether this match is a "verbatim" match: a URL navigation directly
266 // to the user's input, a search for the user's input with the default search
267 // engine, or a "keyword mode" search for the query portion of the user's
268 // input. Note that rare or unusual types that could be considered verbatim,
269 // such as keyword engine matches or extension-provided matches, aren't
270 // detected by this IsVerbatimType, as the user will not be able to infer
271 // what will happen when he or she presses enter in those cases if the match
273 bool IsVerbatimType() const;
275 // Returns whether this match or any duplicate of this match can be deleted.
276 // This is used to decide whether we should call DeleteMatch().
277 bool SupportsDeletion() const;
279 // Swaps the contents and description fields, and their associated
280 // classifications, if this is a match for which we should emphasize the
281 // title (stored in the description field) over the URL (in the contents
282 // field). Intended to only be used at the UI level before displaying, lest
283 // other omnibox systems get confused about which is which. See the code
284 // that sets |swap_contents_and_description| for conditions under which
286 void PossiblySwapContentsAndDescriptionForDisplay();
288 // The provider of this match, used to remember which provider the user had
289 // selected when the input changes. This may be NULL, in which case there is
290 // no provider (or memory of the user's selection).
291 AutocompleteProvider
* provider
;
293 // The relevance of this match. See table in autocomplete.h for scores
294 // returned by various providers. This is used to rank matches among all
295 // responding providers, so different providers must be carefully tuned to
296 // supply matches with appropriate relevance.
298 // TODO(pkasting): http://b/1111299 This should be calculated algorithmically,
299 // rather than being a fairly fixed value defined by the table above.
302 // How many times this result was typed in / selected from the omnibox.
303 // Only set for some providers and result_types. If it is not set,
304 // its value is -1. At the time of writing this comment, it is only
305 // set for matches from HistoryURL and HistoryQuickProvider.
308 // True if the user should be able to delete this match.
311 // This string is loaded into the location bar when the item is selected
312 // by pressing the arrow keys. This may be different than a URL, for example,
313 // for search suggestions, this would just be the search terms.
314 base::string16 fill_into_edit
;
316 // The inline autocompletion to display after the user's typing in the
317 // omnibox, if this match becomes the default match. It may be empty.
318 base::string16 inline_autocompletion
;
320 // If false, the omnibox should prevent this match from being the
321 // default match. Providers should set this to true only if the
322 // user's input, plus any inline autocompletion on this match, would
323 // lead the user to expect a navigation to this match's destination.
324 // For example, with input "foo", a search for "bar" or navigation
325 // to "bar.com" should not set this flag; a navigation to "foo.com"
326 // should only set this flag if ".com" will be inline autocompleted;
327 // and a navigation to "foo/" (an intranet host) or search for "foo"
328 // should set this flag.
329 bool allowed_to_be_default_match
;
331 // The URL to actually load when the autocomplete item is selected. This URL
332 // should be canonical so we can compare URLs with strcmp to avoid dupes.
333 // It may be empty if there is no possible navigation.
334 GURL destination_url
;
336 // The destination URL with "www." stripped off for better dupe finding.
337 GURL stripped_destination_url
;
339 // The main text displayed in the address bar dropdown.
340 base::string16 contents
;
341 ACMatchClassifications contents_class
;
343 // Additional helper text for each entry, such as a title or description.
344 base::string16 description
;
345 ACMatchClassifications description_class
;
347 // If true, UI-level code should swap the contents and description fields
348 // before displaying.
349 bool swap_contents_and_description
;
351 // TODO(jdonnelly): Remove the first two properties once the downstream
352 // clients are using the SuggestionAnswer.
353 // A rich-format version of the display for the dropdown.
354 base::string16 answer_contents
;
355 base::string16 answer_type
;
356 scoped_ptr
<SuggestionAnswer
> answer
;
358 // The transition type to use when the user opens this match. By default
359 // this is TYPED. Providers whose matches do not look like URLs should set
361 ui::PageTransition transition
;
363 // Type of this match.
366 // Set with a keyword provider match if this match can show a keyword hint.
367 // For example, if this is a SearchProvider match for "www.amazon.com",
368 // |associated_keyword| could be a KeywordProvider match for "amazon.com".
370 // When this is set, the popup will show a ">" symbol at the right edge of the
371 // line for this match, and tab/shift-tab will toggle in and out of keyword
372 // mode without disturbing the rest of the popup. See also
373 // OmniboxPopupModel::SetSelectedLineState().
374 scoped_ptr
<AutocompleteMatch
> associated_keyword
;
376 // The keyword of the TemplateURL the match originated from. This is nonempty
377 // for both explicit "keyword mode" matches as well as matches for the default
378 // search provider (so, any match for which we're doing substitution); it
379 // doesn't imply (alone) that the UI is going to show a keyword hint or
380 // keyword mode. For that, see GetKeywordUIState() or
381 // GetSubstitutingExplicitlyInvokedKeyword().
383 // CAUTION: The TemplateURL associated with this keyword may be deleted or
384 // modified while the AutocompleteMatch is alive. This means anyone who
385 // accesses it must perform any necessary sanity checks before blindly using
387 base::string16 keyword
;
389 // True if this match is from a previous result.
392 // Optional search terms args. If present,
393 // AutocompleteController::UpdateAssistedQueryStats() will incorporate this
394 // data with additional data it calculates and pass the completed struct to
395 // TemplateURLRef::ReplaceSearchTerms() to reset the match's |destination_url|
396 // after the complete set of matches in the AutocompleteResult has been chosen
397 // and sorted. Most providers will leave this as NULL, which will cause the
398 // AutocompleteController to do no additional transformations.
399 scoped_ptr
<TemplateURLRef::SearchTermsArgs
> search_terms_args
;
401 // Information dictionary into which each provider can optionally record a
402 // property and associated value and which is presented in chrome://omnibox.
403 AdditionalInfo additional_info
;
405 // A list of matches culled during de-duplication process, retained to
406 // ensure if a match is deleted, the duplicates are deleted as well.
407 std::vector
<AutocompleteMatch
> duplicate_matches
;
410 // Does a data integrity check on this match.
411 void Validate() const;
413 // Checks one text/classifications pair for valid values.
414 void ValidateClassifications(
415 const base::string16
& text
,
416 const ACMatchClassifications
& classifications
) const;
420 typedef AutocompleteMatch::ACMatchClassification ACMatchClassification
;
421 typedef std::vector
<ACMatchClassification
> ACMatchClassifications
;
422 typedef std::vector
<AutocompleteMatch
> ACMatches
;
424 #endif // COMPONENTS_OMNIBOX_BROWSER_AUTOCOMPLETE_MATCH_H_