Retry failed XMP parse with combined standard+extended chunks

#240
mattiasw · Jul 6, 2023 · b4c1dbc · b4c1dbc
1 parent 88f602a
commit b4c1dbc
Show file tree

Hide file tree

Showing 4 changed files with 53 additions and 6 deletions.
diff --git a/dist/exif-reader.js b/dist/exif-reader.js
diff --git a/dist/exif-reader.js.map b/dist/exif-reader.js.map
diff --git a/src/xmp-tags.js b/src/xmp-tags.js
@@ -11,10 +11,31 @@ export default {
 };
 
 function read(dataView, chunks) {
+    const tags = {};
+
     if (typeof dataView === 'string') {
-        return readTags({}, dataView);
+        readTags(tags, dataView);
+        return tags;
+    }
+
+    const [standardXmp, extendedXmp] = extractCompleteChunks(dataView, chunks);
+
+    const hasStandardTags = readTags(tags, standardXmp);
+
+    if (extendedXmp) {
+        const hasExtendedTags = readTags(tags, extendedXmp);
+
+        if (!hasStandardTags && !hasExtendedTags) {
+            // Some writers are not spec-compliant in that they split an XMP
+            // metadata tree over both the standard XMP block and the extended
+            // XMP block. If we failed parsing both of the XMPs in the regular
+            // way, we try to combine them to see if that works better.
+            delete tags._raw;
+            readTags(tags, combineChunks(dataView, chunks));
+        }
     }
-    return extractCompleteChunks(dataView, chunks).reduce(readTags, {});
+
+    return tags;
 }
 
 // The first chunk is always the regular XMP document. Then there is something
@@ -54,9 +75,10 @@ function readTags(tags, chunkDataView) {
         tags._raw = (tags._raw || '') + raw;
         const rdf = getRDF(doc);
 
-        return objectAssign(tags, parseXMPObject(convertToObject(rdf, true)));
+        objectAssign(tags, parseXMPObject(convertToObject(rdf, true)));
+        return true;
     } catch (error) {
-        return tags;
+        return false;
     }
 }
 

diff --git a/test/unit/xmp-tags-spec.js b/test/unit/xmp-tags-spec.js
@@ -878,6 +878,31 @@ describe('xmp-tags', function () {
         });
     });
 
+    // This is non-spec but there are files in the wild using this format.
+    it('should be able to handle multiple chunks where they are all part of a single XMP metadata tree', function () {
+        const xmlString = getXmlString(`
+            <rdf:Description xmlns:xmp="http://ns.example.com/xmp" xmp:MyXMPTag="4711">
+            </rdf:Description>
+        `);
+        const xmlString0 = xmlString.substr(0, 40);
+        const xmlString1 = xmlString.substr(40);
+        const dataView = getDataView(xmlString0 + xmlString1);
+
+        const tags = XmpTags.read(dataView, [
+            {dataOffset: 0, length: xmlString0.length},
+            {dataOffset: xmlString0.length, length: xmlString1.length},
+        ]);
+
+        expect(tags).to.deep.equal({
+            _raw: xmlString0 + xmlString1,
+            MyXMPTag: {
+                value: '4711',
+                attributes: {},
+                description: '4711'
+            }
+        });
+    });
+
     it('should handle when input is a regular string', () => {
         const xmlString = getXmlString(`
             <rdf:Description xmlns:xmp="http://ns.example.com/xmp" xmp:MyXMPTag0="4711">