history_quick_provider.cc revision 731df977c0511bca2206b5f333555b1205ff1f43
1bda42a81ee5f9b20d2bebedcf0bbef1e30e5b293Kristian Monsen// Copyright (c) 2010 The Chromium Authors. All rights reserved.
2bda42a81ee5f9b20d2bebedcf0bbef1e30e5b293Kristian Monsen// Use of this source code is governed by a BSD-style license that can be
3bda42a81ee5f9b20d2bebedcf0bbef1e30e5b293Kristian Monsen// found in the LICENSE file.
4bda42a81ee5f9b20d2bebedcf0bbef1e30e5b293Kristian Monsen
5bda42a81ee5f9b20d2bebedcf0bbef1e30e5b293Kristian Monsen#include "chrome/browser/autocomplete/history_quick_provider.h"
6bda42a81ee5f9b20d2bebedcf0bbef1e30e5b293Kristian Monsen
7731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick#include "base/basictypes.h"
8731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick#include "base/i18n/word_iterator.h"
9731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick#include "base/string_util.h"
10731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick#include "base/logging.h"
11731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick#include "base/utf_string_conversions.h"
12731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick#include "chrome/browser/history/history.h"
13731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick#include "chrome/browser/prefs/pref_service.h"
14731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick#include "chrome/browser/profile.h"
15731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick#include "chrome/browser/history/in_memory_url_index.h"
16731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick#include "chrome/browser/net/url_fixer_upper.h"
17731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick#include "chrome/browser/plugin_service.h"
18731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick#include "chrome/common/notification_source.h"
19731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick#include "chrome/common/notification_type.h"
20731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick#include "chrome/common/pref_names.h"
21731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick#include "chrome/common/url_constants.h"
22731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick#include "googleurl/src/url_util.h"
23731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick#include "net/base/escape.h"
24731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick#include "net/base/net_util.h"
25731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick
26731df977c0511bca2206b5f333555b1205ff1f43Iain Merrickusing history::InMemoryURLIndex;
27731df977c0511bca2206b5f333555b1205ff1f43Iain Merrickusing history::ScoredHistoryMatch;
28731df977c0511bca2206b5f333555b1205ff1f43Iain Merrickusing history::ScoredHistoryMatches;
29731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick
30731df977c0511bca2206b5f333555b1205ff1f43Iain MerrickHistoryQuickProvider::HistoryQuickProvider(ACProviderListener* listener,
31731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick                                           Profile* profile)
32731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick    : HistoryProvider(listener, profile, "HistoryQuickProvider"),
33731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick      trim_http_(false),
34731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick      languages_(profile_->GetPrefs()->GetString(prefs::kAcceptLanguages)) {}
35731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick
36731df977c0511bca2206b5f333555b1205ff1f43Iain MerrickHistoryQuickProvider::~HistoryQuickProvider() {}
37731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick
38bda42a81ee5f9b20d2bebedcf0bbef1e30e5b293Kristian Monsenvoid HistoryQuickProvider::Start(const AutocompleteInput& input,
39bda42a81ee5f9b20d2bebedcf0bbef1e30e5b293Kristian Monsen                                 bool minimal_changes) {
40731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  matches_.clear();
41731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick
42731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  if ((input.type() == AutocompleteInput::INVALID) ||
43731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick      (input.type() == AutocompleteInput::FORCED_QUERY))
44731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick    return;
45731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick
46731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  autocomplete_input_ = input;
47731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  trim_http_ = !HasHTTPScheme(input.text());
48731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick
49731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  // Do some fixup on the user input before matching against it, so we provide
50731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  // good results for local file paths, input with spaces, etc.
51731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  // NOTE: This purposefully doesn't take input.desired_tld() into account; if
52731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  // it did, then holding "ctrl" would change all the results from the
53731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  // HistoryQuickProvider provider, not just the What You Typed Result.
54731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  const std::wstring fixed_text(FixupUserInput(input));
55731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  if (fixed_text.empty()) {
56731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick    // Conceivably fixup could result in an empty string (although I don't
57731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick    // have cases where this happens offhand).  We can't do anything with
58731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick    // empty input, so just bail; otherwise we'd crash later.
59731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick    return;
60731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  }
61731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  autocomplete_input_.set_text(fixed_text);
62731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick
63731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  // TODO(pkasting): We should just block here until this loads.  Any time
64731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  // someone unloads the history backend, we'll get inconsistent inline
65731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  // autocomplete behavior here.
66731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  if (GetIndex()) {
67731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick    DoAutocomplete();
68731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick    UpdateStarredStateOfMatches();
69731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  }
70731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick}
71731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick
72731df977c0511bca2206b5f333555b1205ff1f43Iain Merrickvoid HistoryQuickProvider::DoAutocomplete() {
73731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  // Get the matching URLs from the DB.
74731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  string16 term_string(WideToUTF16(autocomplete_input_.text()));
75731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  term_string = UnescapeURLComponent(term_string,
76731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick      UnescapeRule::SPACES | UnescapeRule::URL_SPECIAL_CHARS);
77731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  history::InMemoryURLIndex::String16Vector terms(
78731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick      HistoryQuickProvider::WordVectorFromString16(term_string));
79731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  ScoredHistoryMatches matches = GetIndex()->HistoryItemsForTerms(terms);
80731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick
81731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  size_t match_num = matches.size() - 1;
82731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  for (ScoredHistoryMatches::const_iterator match_iter = matches.begin();
83731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick       match_iter != matches.end(); ++match_iter, --match_num) {
84731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick    const ScoredHistoryMatch& history_match(*match_iter);
85731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick    AutocompleteMatch ac_match =
86731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick        QuickMatchToACMatch(history_match, NORMAL, match_num);
87731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick    matches_.push_back(ac_match);
88731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  }
89731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick}
90731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick
91731df977c0511bca2206b5f333555b1205ff1f43Iain MerrickAutocompleteMatch HistoryQuickProvider::QuickMatchToACMatch(
92731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick    const ScoredHistoryMatch& history_match,
93731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick    MatchType match_type,
94731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick    size_t match_number) {
95731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  const history::URLRow& info = history_match.url_info;
96731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  int score = CalculateRelevance(history_match.raw_score,
97731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick                                 autocomplete_input_.type(),
98731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick                                 match_type, match_number);
99731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  AutocompleteMatch match(this, score, !!info.visit_count(),
100731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick                          AutocompleteMatch::HISTORY_URL);
101731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  match.destination_url = info.url();
102731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  DCHECK(match.destination_url.is_valid());
103731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  size_t inline_autocomplete_offset =
104731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick      history_match.input_location + autocomplete_input_.text().length();
105731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  const net::FormatUrlTypes format_types = net::kFormatUrlOmitAll &
106731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick      ~((trim_http_ && !history_match.match_in_scheme) ?
107731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick          0 : net::kFormatUrlOmitHTTP);
108731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  std::string languages =
109731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick      match_type == WHAT_YOU_TYPED ? std::string() : languages_;
110731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  match.fill_into_edit =
111731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick      AutocompleteInput::FormattedStringWithEquivalentMeaning(info.url(),
112731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick          UTF16ToWide(net::FormatUrl(info.url(), languages, format_types,
113731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick                                     UnescapeRule::SPACES, NULL, NULL,
114731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick                                     &inline_autocomplete_offset)));
115731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  if (!autocomplete_input_.prevent_inline_autocomplete())
116731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick    match.inline_autocomplete_offset = inline_autocomplete_offset;
117731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  DCHECK((match.inline_autocomplete_offset == std::wstring::npos) ||
118731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick         (match.inline_autocomplete_offset <= match.fill_into_edit.length()));
119731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick
120731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  size_t match_start = history_match.input_location;
121731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  match.contents =
122731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick      UTF16ToWide(net::FormatUrl(info.url(), languages, format_types,
123731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick                                 UnescapeRule::SPACES, NULL, NULL,
124731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick                                 &match_start));
125731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  if ((match_start != std::wstring::npos) &&
126731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick      (inline_autocomplete_offset != std::wstring::npos) &&
127731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick      (inline_autocomplete_offset != match_start)) {
128731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick    DCHECK(inline_autocomplete_offset > match_start);
129731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick    AutocompleteMatch::ClassifyLocationInString(match_start,
130731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick        inline_autocomplete_offset - match_start, match.contents.length(),
131731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick        ACMatchClassification::URL, &match.contents_class);
132731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  } else {
133731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick    AutocompleteMatch::ClassifyLocationInString(std::wstring::npos, 0,
134731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick        match.contents.length(), ACMatchClassification::URL,
135731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick        &match.contents_class);
136731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  }
137731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  match.description = UTF16ToWide(info.title());
138731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  AutocompleteMatch::ClassifyMatchInString(autocomplete_input_.text(),
139731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick                                           UTF16ToWide(info.title()),
140731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick                                           ACMatchClassification::NONE,
141731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick                                           &match.description_class);
142731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick
143731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  return match;
144bda42a81ee5f9b20d2bebedcf0bbef1e30e5b293Kristian Monsen}
145731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick
146731df977c0511bca2206b5f333555b1205ff1f43Iain Merrickhistory::InMemoryURLIndex* HistoryQuickProvider::GetIndex() {
147731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  if (index_for_testing_.get())
148731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick    return index_for_testing_.get();
149731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick
150731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  HistoryService* const history_service =
151731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick      profile_->GetHistoryService(Profile::EXPLICIT_ACCESS);
152731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  if (!history_service)
153731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick    return NULL;
154731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick
155731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  return history_service->InMemoryIndex();
156731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick}
157731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick
158731df977c0511bca2206b5f333555b1205ff1f43Iain Merrickvoid HistoryQuickProvider::SetIndexForTesting(
159731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick    history::InMemoryURLIndex* index) {
160731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  DCHECK(index);
161731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  index_for_testing_.reset(index);
162731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick}
163731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick
164731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick// Utility Functions
165731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick
166731df977c0511bca2206b5f333555b1205ff1f43Iain Merrickhistory::InMemoryURLIndex::String16Vector
167731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick    HistoryQuickProvider::WordVectorFromString16(const string16& uni_string) {
168731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  history::InMemoryURLIndex::String16Vector words;
169731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  WordIterator iter(&uni_string, WordIterator::BREAK_WORD);
170731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  if (iter.Init()) {
171731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick    while (iter.Advance()) {
172731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick      if (iter.IsWord())
173731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick        words.push_back(iter.GetWord());
174731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick    }
175731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  }
176731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  return words;
177731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick}
178731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick
179731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick// static
180731df977c0511bca2206b5f333555b1205ff1f43Iain Merrickint HistoryQuickProvider::CalculateRelevance(int raw_score,
181731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick                                             AutocompleteInput::Type input_type,
182731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick                                             MatchType match_type,
183731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick                                             size_t match_number) {
184731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  switch (match_type) {
185731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick    case INLINE_AUTOCOMPLETE:
186731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick      return 1400;
187731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick
188731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick    case WHAT_YOU_TYPED:
189731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick      return 1200;
190731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick
191731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick    default:
192731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick      return 900 + static_cast<int>(match_number);
193731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick  }
194731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick}
195731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick
196