Merge pull request #584 from peakon/feature/decodeSharedStringEscapes

guyonroche · web-flow · commit ec4cd23ea704 · 2018-06-26T13:32:41.000+01:00
Decode the _x&lt;4 hex chars&gt;_ escape notation in shared strings
diff --git a/lib/xlsx/xform/strings/text-xform.js b/lib/xlsx/xform/strings/text-xform.js
@@ -28,7 +28,9 @@ utils.inherits(TextXform, BaseXform, {
   },
   
   get model() {
-    return this._text.join('');
+    return this._text.join('').replace(/_x([0-9A-F]{4})_/g, function($0, $1) {
+      return String.fromCharCode(parseInt($1, 16));
+    });
   },
 
   parseOpen: function(node) {
diff --git a/spec/integration/data/shared_string_with_escape.xlsx b/spec/integration/data/shared_string_with_escape.xlsx
diff --git a/spec/integration/workbook-xlsx-reader.spec.js b/spec/integration/workbook-xlsx-reader.spec.js
@@ -127,4 +127,20 @@ describe('WorkbookReader', function() {
       });
     });
   });
+
+  describe('with a spreadsheet that contains a shared string with an escaped underscore', function() {
+    before(function() {
+      var testContext = this;
+      var workbook = new Excel.Workbook();
+      return workbook.xlsx.read(fs.createReadStream('./spec/integration/data/shared_string_with_escape.xlsx'))
+        .then(function() {
+          testContext.worksheet = workbook.getWorksheet();
+        });
+    });
+
+    it('should decode the underscore', function() {
+      const cell = this.worksheet.getCell('A1');
+      expect(cell.value).to.equal('_x000D_');
+    });
+  });
 });
