feat: Parse XPath (#6470)

Required by #5247 Simple XPath parser for tXml utils. Not used for now, but needed for MPD Patch support.
shaka-project · Apr 24, 2024 · 0883c32 · 0883c32
1 parent c9b61fe
commit 0883c32
Show file tree

Hide file tree

Showing 3 changed files with 60 additions and 0 deletions.
diff --git a/externs/shaka/player.js b/externs/shaka/player.js
@@ -881,6 +881,8 @@ shaka.extern.InitDataTransform;
  *   The child nodes or string body of the element
  * @property {?shaka.extern.xml.Node} parent
  *   The parent of the current element
+ *
+ * @exportDoc
  */
 shaka.extern.xml.Node;
 

diff --git a/lib/util/tXml.js b/lib/util/tXml.js
@@ -735,6 +735,34 @@ shaka.util.TXml = class {
   }
 
 
+  /**
+   * Parse xPath strings for segments and id targets.
+   * @param {string} exprString
+   * @return {!Array<!shaka.util.TXml.PathNode>}
+   */
+  static parseXpath(exprString) {
+    const returnPaths = [];
+    // Split string by paths but ignore '/' in quotes
+    const paths = exprString
+        .split(/\/+(?=(?:[^'"]*['"][^'"]*['"])*[^'"]*$)/);
+    for (const path of paths) {
+      const nodeName = path.match(/\b([A-Z])\w+/);
+
+      // We only want the id attribute in which case
+      // /'(.*?)'/ will suffice to get it.
+      const idAttr = path.match(/(@id='(.*?)')/);
+      if (nodeName) {
+        returnPaths.push({
+          name: nodeName[0],
+          id: idAttr ?
+            idAttr[0].match(/'(.*?)'/)[0].replaceAll('\'', '') : null,
+        });
+      }
+    }
+    return returnPaths;
+  }
+
+
   /**
    * Converts a tXml node to DOM element.
    * @param {shaka.extern.xml.Node} node
@@ -775,3 +803,12 @@ shaka.util.TXml = class {
 };
 
 shaka.util.TXml.knownNameSpaces_ = new Map([]);
+
+
+/**
+ * @typedef {{
+ *   name: string,
+ *   id: ?string
+ * }}
+ */
+shaka.util.TXml.PathNode;
diff --git a/test/util/tXml_unit.js b/test/util/tXml_unit.js
@@ -416,4 +416,25 @@ describe('tXml', () => {
     expect(TXml.parseFloat(HUGE_NUMBER_STRING)).toBe(Infinity);
     expect(TXml.parseFloat('-' + HUGE_NUMBER_STRING)).toBe(-Infinity);
   });
+
+  it('parseXpath', () => {
+    expect(TXml.parseXpath('/MPD')).toEqual([{name: 'MPD', id: null}]);
+    expect(TXml.parseXpath('/MPD/@type'))
+        .toEqual([{name: 'MPD', id: null}]);
+
+    const timelinePath = '/' + [
+      'MPD',
+      'Period[@id=\'6469\']',
+      'AdaptationSet[@id=\'7\']',
+      'SegmentTemplate',
+      'SegmentTimeline',
+    ].join('/');
+    expect(TXml.parseXpath(timelinePath)).toEqual([
+      {name: 'MPD', id: null},
+      {name: 'Period', id: '6469'},
+      {name: 'AdaptationSet', id: '7'},
+      {name: 'SegmentTemplate', id: null},
+      {name: 'SegmentTimeline', id: null},
+    ]);
+  });
 });