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_PROVIDER_H_
6#define CHROME_BROWSER_AUTOCOMPLETE_AUTOCOMPLETE_PROVIDER_H_
7
8#include <string>
9
10#include "base/basictypes.h"
11#include "base/memory/ref_counted.h"
12#include "base/strings/string16.h"
13#include "chrome/browser/autocomplete/autocomplete_match.h"
14#include "chrome/common/metrics/proto/omnibox_event.pb.h"
15
16class AutocompleteInput;
17class AutocompleteProviderListener;
18class GURL;
19class Profile;
20
21typedef std::vector<metrics::OmniboxEventProto_ProviderInfo> ProvidersInfo;
22
23// The AutocompleteProviders each return different kinds of matches,
24// such as history or search matches.  These matches are given
25// "relevance" scores.  Higher scores are better matches than lower
26// scores.  The relevance scores and classes providing the respective
27// matches are as listed below.
28//
29// IMPORTANT CAVEAT: The tables below are NOT COMPLETE.  Developers
30// often forget to keep these tables in sync with the code when they
31// change scoring algorithms or add new providers.  For example,
32// neither the HistoryQuickProvider (which is a provider that appears
33// often) nor the ShortcutsProvider are listed here.  For the best
34// idea of how scoring works and what providers are affecting which
35// queries, play with chrome://omnibox/ for a while.  While the tables
36// below may have some utility, nothing compares with first-hand
37// investigation and experience.
38//
39// UNKNOWN input type:
40// --------------------------------------------------------------------|-----
41// Keyword (non-substituting or in keyword UI mode, exact match)       | 1500
42// Extension App (exact match)                                         | 1425
43// HistoryURL (good exact or inline autocomplete matches, some inexact)| 1410++
44// HistoryURL (intranet url never visited match, some inexact matches) | 1400++
45// Search Primary Provider (past query in history within 2 days)       | 1399**
46// Search Primary Provider (what you typed)                            | 1300
47// HistoryURL (what you typed, some inexact matches)                   | 1200++
48// Extension App (inexact match)                                       | 1175*~
49// Keyword (substituting, exact match)                                 | 1100
50// Search Primary Provider (past query in history older than 2 days)   | 1050--
51// HistoryURL (some inexact matches)                                   |  900++
52// BookmarkProvider (prefix match in bookmark title)                   |  900+-
53// Built-in                                                            |  860++
54// Search Primary Provider (navigational suggestion)                   |  800++
55// Search Primary Provider (suggestion)                                |  600++
56// Keyword (inexact match)                                             |  450
57// Search Secondary Provider (what you typed)                          |  250
58// Search Secondary Provider (past query in history)                   |  200--
59// Search Secondary Provider (navigational suggestion)                 |  150++
60// Search Secondary Provider (suggestion)                              |  100++
61//
62// URL input type:
63// --------------------------------------------------------------------|-----
64// Keyword (non-substituting or in keyword UI mode, exact match)       | 1500
65// Extension App (exact match)                                         | 1425
66// HistoryURL (good exact or inline autocomplete matches, some inexact)| 1410++
67// HistoryURL (intranet url never visited match, some inexact matches) | 1400++
68// HistoryURL (what you typed, some inexact matches)                   | 1200++
69// Extension App (inexact match)                                       | 1175*~
70// Keyword (substituting, exact match)                                 | 1100
71// HistoryURL (some inexact matches)                                   |  900++
72// Built-in                                                            |  860++
73// Search Primary Provider (what you typed)                            |  850
74// Search Primary Provider (navigational suggestion)                   |  800++
75// Search Primary Provider (past query in history)                     |  750--
76// Keyword (inexact match)                                             |  700
77// Search Primary Provider (suggestion)                                |  300++
78// Search Secondary Provider (what you typed)                          |  250
79// Search Secondary Provider (past query in history)                   |  200--
80// Search Secondary Provider (navigational suggestion)                 |  150++
81// Search Secondary Provider (suggestion)                              |  100++
82//
83// QUERY input type:
84// --------------------------------------------------------------------|-----
85// Search Primary or Secondary (past query in history within 2 days)   | 1599**
86// Keyword (non-substituting or in keyword UI mode, exact match)       | 1500
87// Keyword (substituting, exact match)                                 | 1450
88// Extension App (exact match)                                         | 1425
89// Search Primary Provider (past query in history within 2 days)       | 1399**
90// Search Primary Provider (what you typed)                            | 1300
91// Extension App (inexact match)                                       | 1175*~
92// Search Primary Provider (past query in history older than 2 days)   | 1050--
93// HistoryURL (inexact match)                                          |  900++
94// BookmarkProvider (prefix match in bookmark title)                   |  900+-
95// Search Primary Provider (navigational suggestion)                   |  800++
96// Search Primary Provider (suggestion)                                |  600++
97// Keyword (inexact match)                                             |  450
98// Search Secondary Provider (what you typed)                          |  250
99// Search Secondary Provider (past query in history)                   |  200--
100// Search Secondary Provider (navigational suggestion)                 |  150++
101// Search Secondary Provider (suggestion)                              |  100++
102//
103// FORCED_QUERY input type:
104// --------------------------------------------------------------------|-----
105// Extension App (exact match on title only, not url)                  | 1425
106// Search Primary Provider (past query in history within 2 days)       | 1399**
107// Search Primary Provider (what you typed)                            | 1300
108// Extension App (inexact match on title only, not url)                | 1175*~
109// Search Primary Provider (past query in history older than 2 days)   | 1050--
110// Search Primary Provider (navigational suggestion)                   |  800++
111// Search Primary Provider (suggestion)                                |  600++
112//
113// (A search keyword is a keyword with a replacement string; a bookmark keyword
114// is a keyword with no replacement string, that is, a shortcut for a URL.)
115//
116// There are two possible providers for search suggestions. If the user has
117// typed a keyword, then the primary provider is the keyword provider and the
118// secondary provider is the default provider. If the user has not typed a
119// keyword, then the primary provider corresponds to the default provider.
120//
121// Search providers may supply relevance values along with their results to be
122// used in place of client-side calculated values.
123//
124// The value column gives the ranking returned from the various providers.
125// ++: a series of matches with relevance from n up to (n + max_matches).
126// --: relevance score falls off over time (discounted 50 points @ 15 minutes,
127//     450 points @ two weeks)
128// **: relevance score falls off over two days (discounted 99 points after two
129//     days).
130// *~: Partial matches get a score on a sliding scale from about 575-1125 based
131//     on how many times the URL for the Extension App has been typed and how
132//     many of the letters match.
133// +-: A base score that the provider will adjust upward or downward based on
134//     provider-specific metrics.
135//
136// A single result provider for the autocomplete system.  Given user input, the
137// provider decides what (if any) matches to return, their relevance, and their
138// classifications.
139class AutocompleteProvider
140    : public base::RefCountedThreadSafe<AutocompleteProvider> {
141 public:
142  // Different AutocompleteProvider implementations.
143  enum Type {
144    TYPE_BOOKMARK         = 1 << 0,
145    TYPE_BUILTIN          = 1 << 1,
146    TYPE_CONTACT          = 1 << 2,
147    TYPE_EXTENSION_APP    = 1 << 3,
148    TYPE_HISTORY_QUICK    = 1 << 4,
149    TYPE_HISTORY_URL      = 1 << 5,
150    TYPE_KEYWORD          = 1 << 6,
151    TYPE_SEARCH           = 1 << 7,
152    TYPE_SHORTCUTS        = 1 << 8,
153    TYPE_ZERO_SUGGEST     = 1 << 9,
154  };
155
156  AutocompleteProvider(AutocompleteProviderListener* listener,
157                       Profile* profile,
158                       Type type);
159
160  // Returns a string describing a particular AutocompleteProvider type.
161  static const char* TypeToString(Type type);
162
163  // Called to start an autocomplete query.  The provider is responsible for
164  // tracking its matches for this query and whether it is done processing the
165  // query.  When new matches are available or the provider finishes, it
166  // calls the controller's OnProviderUpdate() method.  The controller can then
167  // get the new matches using the provider's accessors.
168  // Exception: Matches available immediately after starting the query (that
169  // is, synchronously) do not cause any notifications to be sent.  The
170  // controller is expected to check for these without prompting (since
171  // otherwise, starting each provider running would result in a flurry of
172  // notifications).
173  //
174  // Once Stop() has been called, no more notifications should be sent.
175  //
176  // |minimal_changes| is an optimization that lets the provider do less work
177  // when the |input|'s text hasn't changed.  See the body of
178  // OmniboxPopupModel::StartAutocomplete().
179  virtual void Start(const AutocompleteInput& input, bool minimal_changes) = 0;
180
181  // Called when a provider must not make any more callbacks for the current
182  // query. This will be called regardless of whether the provider is already
183  // done.  If the provider caches any results, it should clear the cache based
184  // on the value of |clear_cached_results|.
185  virtual void Stop(bool clear_cached_results);
186
187  // Returns the enum equivalent to the name of this provider.
188  // TODO(derat): Make metrics use AutocompleteProvider::Type directly, or at
189  // least move this method to the metrics directory.
190  metrics::OmniboxEventProto_ProviderType AsOmniboxEventProviderType() const;
191
192  // Called to delete a match and the backing data that produced it.  This
193  // match should not appear again in this or future queries.  This can only be
194  // called for matches the provider marks as deletable.  This should only be
195  // called when no query is running.
196  // NOTE: Remember to call OnProviderUpdate() if matches_ is updated.
197  virtual void DeleteMatch(const AutocompleteMatch& match);
198
199  // Called when an omnibox event log entry is generated.  This gives
200  // a provider the opportunity to add diagnostic information to the
201  // logs.  A provider is expected to append a single entry of whatever
202  // information it wants to |provider_info|.
203  virtual void AddProviderInfo(ProvidersInfo* provider_info) const;
204
205  // Called when a new omnibox session starts or the current session ends.
206  // This gives the opportunity to reset the internal state, if any, associated
207  // with the previous session.
208  virtual void ResetSession();
209
210  // A convenience function to call net::FormatUrl() with the current set of
211  // "Accept Languages" when check_accept_lang is true.  Otherwise, it's called
212  // with an empty list.
213  base::string16 StringForURLDisplay(const GURL& url,
214                                     bool check_accept_lang,
215                                     bool trim_http) const;
216
217  // Returns the set of matches for the current query.
218  const ACMatches& matches() const { return matches_; }
219
220  // Returns whether the provider is done processing the query.
221  bool done() const { return done_; }
222
223  // Returns this provider's type.
224  Type type() const { return type_; }
225
226  // Returns a string describing this provider's type.
227  const char* GetName() const;
228
229#ifdef UNIT_TEST
230  void set_listener(AutocompleteProviderListener* listener) {
231    listener_ = listener;
232  }
233#endif
234  // A suggested upper bound for how many matches a provider should return.
235  // TODO(pkasting): http://b/1111299 , http://b/933133 This should go away once
236  // we have good relevance heuristics; the controller should handle all
237  // culling.
238  static const size_t kMaxMatches;
239
240 protected:
241  friend class base::RefCountedThreadSafe<AutocompleteProvider>;
242
243  virtual ~AutocompleteProvider();
244
245  // Updates the starred state of each of the matches in matches_ from the
246  // profile's bookmark bar model.
247  void UpdateStarredStateOfMatches();
248
249  // Fixes up user URL input to make it more possible to match against.  Among
250  // many other things, this takes care of the following:
251  // * Prepending file:// to file URLs
252  // * Converting drive letters in file URLs to uppercase
253  // * Converting case-insensitive parts of URLs (like the scheme and domain)
254  //   to lowercase
255  // * Convert spaces to %20s
256  // Note that we don't do this in AutocompleteInput's constructor, because if
257  // e.g. we convert a Unicode hostname to punycode, other providers will show
258  // output that surprises the user ("Search Google for xn--6ca.com").
259  // Returns false if the fixup attempt resulted in an empty string (which
260  // providers generally can't do anything with).
261  static bool FixupUserInput(AutocompleteInput* input);
262
263  // Trims "http:" and up to two subsequent slashes from |url|.  Returns the
264  // number of characters that were trimmed.
265  // NOTE: For a view-source: URL, this will trim from after "view-source:" and
266  // return 0.
267  static size_t TrimHttpPrefix(base::string16* url);
268
269  // The profile associated with the AutocompleteProvider.  Reference is not
270  // owned by us.
271  Profile* profile_;
272
273  AutocompleteProviderListener* listener_;
274  ACMatches matches_;
275  bool done_;
276
277  Type type_;
278
279 private:
280  DISALLOW_COPY_AND_ASSIGN(AutocompleteProvider);
281};
282
283typedef std::vector<AutocompleteProvider*> ACProviders;
284
285#endif  // CHROME_BROWSER_AUTOCOMPLETE_AUTOCOMPLETE_PROVIDER_H_
286