fix(TTML): Fix trim surrounding spaces with xml:space="default" (#6395)

Fixes #4974
shaka-project · Apr 4, 2024 · bcedec3 · bcedec3
1 parent 286126e
commit bcedec3
Show file tree

Hide file tree

Showing 2 changed files with 14 additions and 16 deletions.
diff --git a/lib/text/ttml_text_parser.js b/lib/text/ttml_text_parser.js
@@ -99,7 +99,7 @@ shaka.text.TtmlTextParser = class {
           shaka.util.Error.Code.INVALID_XML,
           'Invalid xml:space value: ' + spaceStyle);
     }
-    const whitespaceTrim = spaceStyle == 'default';
+    const collapseMultipleSpaces = spaceStyle == 'default';
 
     const rateInfo = new TtmlTextParser.RateInfo_(
         frameRate, subFrameRate, frameRateMultiplier, tickRate);
@@ -146,7 +146,7 @@ shaka.text.TtmlTextParser = class {
 
     const cue = TtmlTextParser.parseCue_(
         body, time, rateInfo, metadataElements, styles,
-        regionElements, cueRegions, whitespaceTrim,
+        regionElements, cueRegions, collapseMultipleSpaces,
         cellResolutionInfo, /* parentCueElement= */ null,
         /* isContent= */ false, uri);
     if (cue) {
@@ -172,7 +172,7 @@ shaka.text.TtmlTextParser = class {
    * @param {!Array.<!shaka.extern.xml.Node>} styles
    * @param {!Array.<!shaka.extern.xml.Node>} regionElements
    * @param {!Array.<!shaka.text.CueRegion>} cueRegions
-   * @param {boolean} whitespaceTrim
+   * @param {boolean} collapseMultipleSpaces
    * @param {?{columns: number, rows: number}} cellResolution
    * @param {?shaka.extern.xml.Node} parentCueElement
    * @param {boolean} isContent
@@ -182,8 +182,8 @@ shaka.text.TtmlTextParser = class {
    */
   static parseCue_(
       cueNode, timeContext, rateInfo, metadataElements, styles, regionElements,
-      cueRegions, whitespaceTrim, cellResolution, parentCueElement, isContent,
-      uri) {
+      cueRegions, collapseMultipleSpaces, cellResolution, parentCueElement,
+      isContent, uri) {
     const TXml = shaka.util.TXml;
     const StringUtils = shaka.util.StringUtils;
     /** @type {shaka.extern.xml.Node} */
@@ -246,9 +246,9 @@ shaka.text.TtmlTextParser = class {
     const parentIsContent = isContent;
 
     const spaceStyle = cueElement.attributes['xml:space'] ||
-        (whitespaceTrim ? 'default' : 'preserve');
+        (collapseMultipleSpaces ? 'default' : 'preserve');
 
-    const localWhitespaceTrim = spaceStyle == 'default';
+    const localCollapseMultipleSpaces = spaceStyle == 'default';
 
     // Parse any nested cues first.
     const isLeafNode = cueElement.children.every(TXml.isText);
@@ -265,7 +265,7 @@ shaka.text.TtmlTextParser = class {
             styles,
             regionElements,
             cueRegions,
-            localWhitespaceTrim,
+            localCollapseMultipleSpaces,
             cellResolution,
             cueElement,
             isContent,
@@ -300,7 +300,7 @@ shaka.text.TtmlTextParser = class {
         // as some information could be held by its attributes.
         // <p /> won't, as it would not be displayed.
         return null;
-      } else if (localWhitespaceTrim) {
+      } else if (localCollapseMultipleSpaces) {
         // Disregards empty anonymous spans when (local) trim is true.
         return null;
       }
@@ -361,9 +361,7 @@ shaka.text.TtmlTextParser = class {
       // If the childNodes are all text, this is a leaf node.  Get the payload.
       payload = StringUtils.htmlUnescape(
           shaka.util.TXml.getTextContents(cueElement) || '');
-      if (localWhitespaceTrim) {
-        // Trim leading and trailing whitespace.
-        payload = payload.trim();
+      if (localCollapseMultipleSpaces) {
         // Collapse multiple spaces into one.
         payload = payload.replace(/\s+/g, ' ');
       }

diff --git a/test/text/ttml_text_parser_unit.js b/test/text/ttml_text_parser_unit.js
@@ -47,7 +47,7 @@ describe('TtmlTextParser', () => {
             startTime: 62.03,
             endTime: 62.05,
             nestedCues: [{
-              payload: 'A B C',
+              payload: ' A B C ',
               startTime: 62.03,
               endTime: 62.05,
             }],
@@ -90,7 +90,7 @@ describe('TtmlTextParser', () => {
             startTime: 62.03,
             endTime: 62.05,
             nestedCues: [{
-              payload: 'A B C',
+              payload: ' A B C ',
               startTime: 62.03,
               endTime: 62.05,
             }],
@@ -2055,7 +2055,7 @@ describe('TtmlTextParser', () => {
               {
                 startTime: 0,
                 endTime: 60,
-                payload: 'A',
+                payload: ' A',
                 fontSize: '16px',
               },
               {
@@ -2067,7 +2067,7 @@ describe('TtmlTextParser', () => {
               {
                 startTime: 0,
                 endTime: 60,
-                payload: 'B',
+                payload: 'B ',
                 fontSize: '16px',
               },
             ],