OLD | NEW |
---|---|
(Empty) | |
1 // Copyright 2016 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 #include "chrome/browser/android/contextualsearch/search_action.h" | |
6 | |
7 #include <set> | |
8 | |
9 #include "base/android/jni_string.h" | |
10 #include "base/bind.h" | |
11 #include "base/memory/weak_ptr.h" | |
12 #include "base/strings/utf_string_conversions.h" | |
13 #include "chrome/browser/android/contextualsearch/contextual_search_context.h" | |
14 #include "content/public/browser/render_frame_host.h" | |
15 #include "content/public/browser/web_contents.h" | |
16 #include "jni/SearchAction_jni.h" | |
17 #include "url/gurl.h" | |
18 | |
19 using base::android::ConvertUTF8ToJavaString; | |
20 using base::android::JavaParamRef; | |
21 using base::android::ScopedJavaGlobalRef; | |
22 using content::WebContents; | |
23 | |
24 namespace { | |
25 const int kSurroundingTextSize = 1536; | |
Theresa
2016/09/01 16:49:37
nit: where did 1536 come from?
Donn Denman
2016/09/02 16:40:43
Done: added the comment "Context length is set to
| |
26 } | |
27 | |
28 SearchAction::SearchAction(JNIEnv* env, jobject obj) { | |
29 java_object_.Reset(env, obj); | |
30 } | |
31 | |
32 SearchAction::~SearchAction() { | |
33 JNIEnv* env = base::android::AttachCurrentThread(); | |
34 Java_SearchAction_clearNativePointer(env, java_object_.obj()); | |
35 } | |
36 | |
37 void SearchAction::Destroy(JNIEnv* env, const JavaParamRef<jobject>& obj) { | |
38 delete this; | |
39 } | |
40 | |
41 void SearchAction::RequestSurroundingText( | |
42 JNIEnv* env, | |
43 const JavaParamRef<jobject>& obj, | |
44 const JavaParamRef<jobject>& j_web_contents) { | |
45 WebContents* web_contents = | |
46 WebContents::FromJavaWebContents(j_web_contents.obj()); | |
47 DCHECK(web_contents); | |
48 | |
49 GURL url(web_contents->GetURL()); | |
50 std::string encoding(web_contents->GetEncoding()); | |
51 search_context_.reset(new ContextualSearchContext(true, url, encoding)); | |
52 | |
53 content::RenderFrameHost* focused_frame = web_contents->GetFocusedFrame(); | |
54 if (!focused_frame) { | |
55 OnSurroundingTextResponse(base::string16(), 0, 0); | |
56 } else { | |
57 focused_frame->RequestTextSurroundingSelection( | |
58 base::Bind(&SearchAction::OnSurroundingTextResponse, AsWeakPtr()), | |
59 kSurroundingTextSize); | |
60 } | |
61 } | |
62 | |
63 void SearchAction::OnSurroundingTextResponse( | |
64 const base::string16& surrounding_text, | |
65 int focus_start, | |
66 int focus_end) { | |
67 // Record the context, which can be accessed later on demand. | |
68 search_context_->surrounding_text = surrounding_text; | |
69 search_context_->start_offset = focus_start; | |
70 search_context_->end_offset = focus_end; | |
71 | |
72 // Notify Java. | |
73 JNIEnv* env = base::android::AttachCurrentThread(); | |
74 Java_SearchAction_onSurroundingTextResponse(env, java_object_.obj()); | |
75 } | |
76 | |
77 std::string SearchAction::FindFocusedWord() { | |
78 DCHECK(search_context_); | |
79 // Find the focused word. | |
80 std::pair<int, int> focus = FindFocusedWord(search_context_->surrounding_text, | |
81 search_context_->start_offset, | |
82 search_context_->end_offset); | |
83 return base::UTF16ToUTF8(search_context_->surrounding_text.substr( | |
84 focus.first, focus.second - focus.first)); | |
85 } | |
86 | |
87 std::string SearchAction::GetSampleText(int sample_length) { | |
88 DCHECK(search_context_); | |
89 // Trim surroundings to the requested sample size. | |
90 std::pair<base::string16, int> sample = SampleSurroundings( | |
91 search_context_->surrounding_text, search_context_->start_offset, | |
92 search_context_->end_offset, sample_length); | |
93 return base::UTF16ToUTF8(sample.first); | |
94 } | |
95 | |
96 // static | |
97 std::pair<int, int> SearchAction::FindFocusedWord( | |
pedro (no code reviews)
2016/08/31 22:38:22
Nit: Why not merge this method with the one with s
Donn Denman
2016/09/02 16:40:43
Done. It was tricky to figure out how to test non
| |
98 const base::string16& surrounding_text, | |
99 int focus_start, | |
100 int focus_end) { | |
101 int surrounding_text_length = surrounding_text.length(); | |
102 | |
103 // First, we need to find the focused word boundaries. | |
104 int focused_word_start = focus_start; | |
105 int focused_word_end = focus_end; | |
106 | |
107 if (surrounding_text_length > 0 && | |
108 IsValidCharacter(surrounding_text[focused_word_start])) { | |
109 // Find focused word start (inclusive) | |
110 for (int i = focus_start; i >= 0; i--) { | |
111 focused_word_start = i; | |
112 | |
113 wchar_t character; | |
114 if (i > 0) { | |
115 character = surrounding_text[i - 1]; | |
116 | |
117 if (!IsValidCharacter(character)) | |
118 break; | |
119 } | |
120 } | |
121 | |
122 // Find focused word end (non inclusive) | |
123 for (int i = focus_end; i < surrounding_text_length; i++) { | |
124 wchar_t character = surrounding_text[i]; | |
125 | |
126 if (IsValidCharacter(character)) { | |
127 focused_word_end = i + 1; | |
128 } else { | |
129 focused_word_end = i; | |
130 break; | |
131 } | |
132 } | |
133 } | |
134 | |
135 return std::pair<int, int>(focused_word_start, focused_word_end); | |
136 } | |
137 | |
138 // static | |
139 std::pair<base::string16, int> SearchAction::SampleSurroundings( | |
140 const base::string16& surrounding_text, | |
141 int focus_start, | |
142 int focus_end, | |
143 int surrounding_text_sample_limit) { | |
144 int surrounding_text_length = surrounding_text.length(); | |
145 | |
146 //--------------------------------------------------------------------------- | |
147 // Cut the surrounding size so it can be sent to the Java side. | |
148 | |
149 // Start equally distributing the size around the focal point. | |
150 int focal_point_size = focus_end - focus_start; | |
151 int sample_margin = (surrounding_text_sample_limit - focal_point_size) / 2; | |
152 int sample_start = focus_start - sample_margin; | |
153 int sample_end = focus_end + sample_margin; | |
154 | |
155 // If the the start is out of bounds, compensate the end side. | |
156 if (sample_start < 0) { | |
157 sample_end -= sample_start; | |
158 sample_start = 0; | |
159 } | |
160 | |
161 // If the the end is out of bounds, compensate the start side. | |
162 if (sample_end > surrounding_text_length) { | |
163 int diff = sample_end - surrounding_text_length; | |
164 sample_end = surrounding_text_length; | |
165 sample_start -= diff; | |
166 } | |
167 | |
168 // Trim the start and end to make sure they are within bounds. | |
169 sample_start = std::max(0, sample_start); | |
170 sample_end = std::min(surrounding_text_length, sample_end); | |
171 base::string16 sample_string( | |
172 surrounding_text.substr(sample_start, sample_end - sample_start)); | |
173 std::pair<base::string16, int> sample(sample_string, sample_start); | |
174 return sample; | |
175 } | |
176 | |
177 // static | |
178 bool SearchAction::IsValidCharacter(int char_code) { | |
179 // See http://www.unicode.org/Public/UCD/latest/ucd/NamesList.txt | |
180 // See http://jrgraphix.net/research/unicode_blocks.php | |
181 | |
182 // TODO(pedrosimonetti): should not include CJK characters! | |
pedro (no code reviews)
2016/08/31 22:38:22
Take over this TODO? This needs to be addressed or
Donn Denman
2016/09/02 16:40:43
Done.
Note: With the current implementation sur
| |
183 // TODO(donnd): consider using regular expressions. | |
184 | |
185 if ((char_code >= 11264 && char_code <= 55295) || // Asian language symbols | |
186 (char_code >= 192 && char_code <= 8191) || // Letters in many languages | |
187 (char_code >= 97 && char_code <= 122) || // Lowercase letters | |
188 (char_code >= 65 && char_code <= 90) || // Uppercase letters | |
189 (char_code >= 48 && char_code <= 57)) { // Numbers | |
190 return true; | |
191 } | |
192 | |
193 return false; | |
194 } | |
195 | |
196 bool RegisterSearchAction(JNIEnv* env) { | |
197 return RegisterNativesImpl(env); | |
198 } | |
199 | |
200 jlong Init(JNIEnv* env, const JavaParamRef<jobject>& obj) { | |
201 SearchAction* content = new SearchAction(env, obj); | |
202 return reinterpret_cast<intptr_t>(content); | |
203 } | |
OLD | NEW |