[TTS] Gather surrounding text on Tap before any UX.

chrome/browser/android/contextualsearch/search_action.cc

Index: chrome/browser/android/contextualsearch/search_action.cc
diff --git a/chrome/browser/android/contextualsearch/search_action.cc b/chrome/browser/android/contextualsearch/search_action.cc
new file mode 100644
index 0000000000000000000000000000000000000000..67db3a7f6f58c5d794e875b36965b9b29bcf5125
--- /dev/null
+++ b/chrome/browser/android/contextualsearch/search_action.cc
@@ -0,0 +1,205 @@
+// Copyright 2016 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#include "chrome/browser/android/contextualsearch/search_action.h"
+
+#include <set>
+
+#include "base/android/jni_string.h"
+#include "base/bind.h"
+#include "base/memory/weak_ptr.h"
+#include "base/strings/utf_string_conversions.h"
+#include "chrome/browser/android/contextualsearch/contextual_search_context.h"
+#include "chrome/browser/profiles/profile_manager.h"
+#include "content/public/browser/render_frame_host.h"
+#include "content/public/browser/web_contents.h"
+#include "jni/SearchAction_jni.h"
+#include "url/gurl.h"
+
+using base::android::JavaParamRef;
+
+namespace {
+
+const int kSurroundingTextSize = 1536;
+const int kSurroundingTextSampleSize = 200;
+}
+
+SearchAction::SearchAction(JNIEnv* env, jobject obj) {
+  java_object_.Reset(env, obj);
+}
+
+SearchAction::~SearchAction() {
+  LOG(ERROR) << "ctxs ===== native SearchAction::~SearchAction";
+  JNIEnv* env = base::android::AttachCurrentThread();
+  Java_SearchAction_clearNativePointer(env, java_object_.obj());
+}
+
+void SearchAction::Destroy(JNIEnv* env, const JavaParamRef<jobject>& obj) {
+  LOG(ERROR) << "ctxs ===== native SearchAction::Destroy";
+  delete this;
+}
+
+void SearchAction::RequestSurroundingText(
+    JNIEnv* env,
+    const JavaParamRef<jobject>& obj,
+    const JavaParamRef<jobject>& j_web_contents) {
+  content::WebContents* web_contents =
+      content::WebContents::FromJavaWebContents(j_web_contents.obj());
+  DCHECK(web_contents);
+
+  GURL url(web_contents->GetURL());
+  std::string encoding(web_contents->GetEncoding());
+  search_context_.reset(new ContextualSearchContext(true, url, encoding));
+
+  content::RenderFrameHost* focused_frame = web_contents->GetFocusedFrame();
+  if (!focused_frame) {
+    OnSurroundingTextResponse(base::string16(), 0, 0);
+  } else {
+    focused_frame->RequestTextSurroundingSelection(
+        base::Bind(&SearchAction::OnSurroundingTextResponse, AsWeakPtr()),
+        kSurroundingTextSize);
+  }
+}
+
+void SearchAction::OnSurroundingTextResponse(
+    const base::string16& surrounding_text,
+    int focus_start,
+    int focus_end) {
+  search_context_->surrounding_text = surrounding_text;
+
+  // Find the focused word.
+  std::pair<int, int> focus =
+      FindFocusedWord(surrounding_text, focus_start, focus_end);
+  std::string selected_text = base::UTF16ToUTF8(
+      surrounding_text.substr(focus.first, focus.second - focus.first));
+
+  // Trim surroundings to a sample small enough to be suitable to send to Java.
+  std::pair<base::string16, int> sample = SampleSurroundings(
+      surrounding_text, focus_start, focus_end, kSurroundingTextSampleSize);
+  JNIEnv* env = base::android::AttachCurrentThread();
+
+  LOG(ERROR) << "ctxs --- surroundings";
+  LOG(ERROR) << "ctxs --- '" << sample.first << "'";
+
+  base::android::ScopedJavaLocalRef<jstring> j_surrounding_text_sample =
+      base::android::ConvertUTF16ToJavaString(env, sample.first);
+
+  LOG(ERROR) << "ctxs --- WORD [" << selected_text << "]  focus " << focus_start
+             << ", " << focus_end << "  focused word [" << focus.first << ", "
+             << focus.second << ") substring '" << sample.first << "' offset "
+             << sample.second;
+
+  // Notify Java.
+  Java_SearchAction_onSurroundingTextResponse(
+      env, java_object_.obj(), j_surrounding_text_sample.obj(), sample.second,
+      focus_start, focus_end, focus.first, focus.second);
+}
+
+// static
+std::pair<int, int> SearchAction::FindFocusedWord(
+    const base::string16& surrounding_text,
+    int focus_start,
+    int focus_end) {
+  int surrounding_text_length = surrounding_text.length();
+
+  // First, we need to find the focused word boundaries.
+  int focused_word_start = focus_start;
+  int focused_word_end = focus_end;
+
+  if (surrounding_text_length > 0 &&
+      IsValidCharacter(surrounding_text[focused_word_start])) {
+    // Find focused word start (inclusive)
+    for (int i = focus_start; i >= 0; i--) {
+      focused_word_start = i;
+
+      wchar_t character;
+      if (i > 0) {
+        character = surrounding_text[i - 1];
+
+        if (!IsValidCharacter(character))
+          break;
+      }
+    }
+
+    // Find focused word end (non inclusive)
+    for (int i = focus_end; i < surrounding_text_length; i++) {
+      wchar_t character = surrounding_text[i];
+
+      if (IsValidCharacter(character)) {
+        focused_word_end = i + 1;
+      } else {
+        focused_word_end = i;
+        break;
+      }
+    }
+  }
+
+  return std::pair<int, int>(focused_word_start, focused_word_end);
+}
+
+// static
+std::pair<base::string16, int> SearchAction::SampleSurroundings(
+    const base::string16& surrounding_text,
+    int focus_start,
+    int focus_end,
+    int surrounding_text_sample_limit) {
+  int surrounding_text_length = surrounding_text.length();
+
+  //---------------------------------------------------------------------------
+  // Cut the surrounding size so it can be sent to the Java side.
+
+  // Start equally distributing the size around the focal point.
+  int focal_point_size = focus_end - focus_start;
+  int sample_margin = (surrounding_text_sample_limit - focal_point_size) / 2;
+  int sample_start = focus_start - sample_margin;
+  int sample_end = focus_end + sample_margin;
+
+  // If the the start is out of bounds, compensate the end side.
+  if (sample_start < 0) {
+    sample_end -= sample_start;
+    sample_start = 0;
+  }
+
+  // If the the end is out of bounds, compensate the start side.
+  if (sample_end > surrounding_text_length) {
+    int diff = sample_end - surrounding_text_length;
+    sample_end = surrounding_text_length;
+    sample_start -= diff;
+  }
+
+  // Trim the start and end to make sure they are within bounds.
+  sample_start = std::max(0, sample_start);
+  sample_end = std::min(surrounding_text_length, sample_end);
+  base::string16 sample_string(
+      surrounding_text.substr(sample_start, sample_end - sample_start));
+  std::pair<base::string16, int> sample(sample_string, sample_start);
+  return sample;
+}
+
+// static
+bool SearchAction::IsValidCharacter(int char_code) {
+  // See http://www.unicode.org/Public/UCD/latest/ucd/NamesList.txt
+  // See http://jrgraphix.net/research/unicode_blocks.php
+
+  // TODO(pedrosimonetti): should not include CJK characters!

[Theresa, 2016/08/16 15:41:49]

Let's be sure to get some manual test coverage on CJK to make sure we're not
breaking anything there.

[Donn Denman, 2016/08/17 04:35:22]

Acknowledged.

[Donn Denman, 2016/08/17 04:35:22]

Pedro, why shouldn't CJK characters be included here?  Maybe you meant some
other CJK items like punctuation? 

[pedro (no code reviews), 2016/08/22 20:54:17]

If we include all CJK charactes, as we are doing, then it will select all
characters until it finds a non-character, like space or punctuation. This is a
problem because our resolution engine will "expand" words when resolving, but
not "shrink". This means resolution will be broken for CJK languages. Instead,
the code will select chunks of characters that not necessarily translate to
actual words.

Example: "San" might be resolved to "San Francisco", but "San Francisco plus
some extra characters" won't be resolved to "San Francisco".

[Donn Denman, 2016/08/23 23:21:46]

Acknowledged.

+
+  if ((char_code >= 11264 && char_code <= 55295) ||

[Theresa, 2016/08/16 15:41:49]

Can we comment on whta these are as well?

[Donn Denman, 2016/08/17 04:35:22]

Done.

+      (char_code >= 192 && char_code <= 8191) ||
+      (char_code >= 97 && char_code <= 122) ||  // Lowercase letters
+      (char_code >= 65 && char_code <= 90) ||   // Uppercase letters
+      (char_code >= 48 && char_code <= 57)) {   // Numbers
+    return true;
+  }
+
+  return false;
+}
+
+bool RegisterSearchAction(JNIEnv* env) {
+  return RegisterNativesImpl(env);
+}
+
+jlong Init(JNIEnv* env, const JavaParamRef<jobject>& obj) {
+  SearchAction* content = new SearchAction(env, obj);
+  return reinterpret_cast<intptr_t>(content);
+}