Generalize splitting logic in ChromeVox

Bug: 1057904
Test: browser_tests --gtest_filter=ChromeVox*.*

Change-Id: I0f858a0a0d3fc2a34177446cb702212b6ededa84
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2095915
Commit-Queue: David Tseng <dtseng@chromium.org>
Reviewed-by: Hiroki Sato <hirokisato@chromium.org>
Cr-Commit-Position: refs/heads/master@{#748875}

chrome/browser/resources/chromeos/accessibility/chromevox/background/annotation/user_annotation_handler.js

@@ -94,6 +94,9 @@ UserAnnotationHandler = class {
    * @return {?string}
    */
   static getAnnotationForNode(node) {
+    if (!node.root) {
+      return null;
+    }
     const url = node.root.docUrl || '';
     if (!UserAnnotationHandler.instance ||
         !UserAnnotationHandler.instance.enabled || !url) {

chrome/browser/resources/chromeos/accessibility/chromevox/common/tts_background.js

@@ -260,18 +260,54 @@ TtsBackground = class extends ChromeTtsBase {
    * @private
    */
   speakSplittingText_(textString, queueMode, properties) {
-    const lines = textString.split(/\r\n|\r|\n/);
-    for (const line of lines) {
-      if (line.length > constants.OBJECT_MAX_CHARCOUNT) {
-        // Skip the paragraph when it's really a large paragraph.
-        // TODO(1057904): Try splitting based on spaces.
-        continue;
-      }
-      this.speak(line, queueMode, properties);
+    const chunks = TtsBackground.splitUntilSmall(textString, '\n\r ');
+    for (const chunk of chunks) {
+      this.speak(chunk, queueMode, properties);
       queueMode = QueueMode.QUEUE;
     }
   }
 
+  /**
+   * Splits |text| until each substring's length is smaller than or equal to
+   * constants.OBJECT_MAX_CHARCOUNT.
+   * @param {string} text
+   * @param {string} delimiters
+   * @return {!Array<string>}
+   */
+  static splitUntilSmall(text, delimiters) {
+    if (text.length == 0) {
+      return [];
+    }
+
+    if (text.length <= constants.OBJECT_MAX_CHARCOUNT) {
+      return [text];
+    }
+
+    const midIndex = text.length / 2;
+    if (!delimiters) {
+      return TtsBackground
+          .splitUntilSmall(text.substring(0, midIndex), delimiters)
+          .concat(TtsBackground.splitUntilSmall(
+              text.substring(midIndex, text.length), delimiters));
+    }
+
+    const delimiter = delimiters[0];
+    let splitIndex = text.lastIndexOf(delimiter, midIndex);
+    if (splitIndex == -1) {
+      splitIndex = text.indexOf(delimiter, midIndex);
+    }
+
+    if (splitIndex == -1) {
+      delimiters = delimiters.slice(1);
+      return TtsBackground.splitUntilSmall(text, delimiters);
+    }
+
+    return TtsBackground
+        .splitUntilSmall(text.substring(0, splitIndex), delimiters)
+        .concat(TtsBackground.splitUntilSmall(
+            text.substring(splitIndex + 1, text.length), delimiters));
+  }
+
   /**
    * Use the speech queue to handle the given speech request.
    * @param {Utterance} utterance The utterance to speak.

chrome/browser/resources/chromeos/accessibility/chromevox/common/tts_background_test.js

@@ -199,11 +199,52 @@ SYNC_TEST_F('ChromeVoxTtsBackgroundTest', 'SplitLongText', function() {
     spokenTextStrings.push(utterance.textString);
   };
 
-  const baseText = 'This is a very long text.\n';
-  const numRepeats =
-      Math.floor(constants.OBJECT_MAX_CHARCOUNT / baseText.length);
-  const text = baseText + baseText.repeat(numRepeats);
-
+  // There are three new lines, but only the first chunk exceeds the max
+  // character count.
+  const text = 'a'.repeat(constants.OBJECT_MAX_CHARCOUNT) + '\n' +
+      'a'.repeat(10) + '\n' +
+      'a'.repeat(10);
   tts.speak(text);
-  assertEquals(1 + numRepeats, spokenTextStrings.length);
+  assertEquals(2, spokenTextStrings.length);
+});
+
+SYNC_TEST_F('ChromeVoxTtsBackgroundTest', 'SplitUntilSmall', function() {
+  const split = TtsBackground.splitUntilSmall;
+
+  // A single delimiter.
+  constants.OBJECT_MAX_CHARCOUNT = 3;
+  assertEqualsJSON(['12', '345', '789'], split('12345a789', 'a'));
+
+  constants.OBJECT_MAX_CHARCOUNT = 4;
+  assertEqualsJSON(['12', '345', '789'], split('12345a789', 'a'));
+
+  constants.OBJECT_MAX_CHARCOUNT = 7;
+  assertEqualsJSON(['12345', '789'], split('12345a789', 'a'));
+
+  constants.OBJECT_MAX_CHARCOUNT = 10;
+  assertEqualsJSON(['12345a789'], split('12345a789', 'a'));
+
+  // Multiple delimiters.
+  constants.OBJECT_MAX_CHARCOUNT = 3;
+  assertEqualsJSON(['12', '34', '57', '89'], split('1234b57a89', 'ab'));
+
+  constants.OBJECT_MAX_CHARCOUNT = 4;
+  assertEqualsJSON(['1234', '57', '89'], split('1234b57a89', 'ab'));
+
+  constants.OBJECT_MAX_CHARCOUNT = 5;
+  assertEqualsJSON(['12345', '789'], split('12345b789a', 'ab'));
+  assertEqualsJSON(['12345', '789a'], split('12345b789a', 'ba'));
+
+  // No delimiters.
+  constants.OBJECT_MAX_CHARCOUNT = 3;
+  assertEqualsJSON(['12', '34', '57', '89'], split('12345789', ''));
+
+  constants.OBJECT_MAX_CHARCOUNT = 4;
+  assertEqualsJSON(['1234', '5789'], split('12345789', ''));
+
+  // Some corner cases.
+  assertEqualsJSON([], split('', ''));
+  assertEqualsJSON(['a'], split('a', ''));
+  assertEqualsJSON(['a'], split('a', 'a'));
+  assertEqualsJSON(['a'], split('a', 'b'));
 });