Fix filename parsing with special characters and prefixes

* Add support for diffs with prefixes through configuration
* Fix parsing of filenames with special chars

src/diff-parser.js

@@ -23,7 +23,7 @@
 
   DiffParser.prototype.LINE_TYPE = LINE_TYPE;
 
-  DiffParser.prototype.generateDiffJson = function(diffInput) {
+  DiffParser.prototype.generateDiffJson = function(diffInput, config) {
     var files = [];
     var currentFile = null;
     var currentBlock = null;
@@ -133,11 +133,11 @@
     var deletedFileMode = /^deleted file mode (\d{6})/;
     var newFileMode = /^new file mode (\d{6})/;
 
-    var copyFrom = /^copy from (.+)/;
-    var copyTo = /^copy to (.+)/;
+    var copyFrom = /^copy from "?(.+?)"?/;
+    var copyTo = /^copy to "?(.+?)"?/;
 
-    var renameFrom = /^rename from (.+)/;
-    var renameTo = /^rename to (.+)/;
+    var renameFrom = /^rename from "?(.+?)"?/;
+    var renameTo = /^rename to "?(.+?)"?/;
 
     var similarityIndex = /^similarity index (\d+)%/;
     var dissimilarityIndex = /^dissimilarity index (\d+)%/;
@@ -160,10 +160,10 @@
       var values = [];
       if (utils.startsWith(line, 'diff')) {
         startFile();
-      } else if (currentFile && !currentFile.oldName && (values = /^--- [aiwco]\/(.+)$/.exec(line))) {
+      } else if (currentFile && !currentFile.oldName && (values = getSrcFilename(line, config))) {
         currentFile.oldName = values[1];
         currentFile.language = getExtension(currentFile.oldName, currentFile.language);
-      } else if (currentFile && !currentFile.newName && (values = /^\+\+\+ [biwco]?\/(.+)$/.exec(line))) {
+      } else if (currentFile && !currentFile.newName && (values = getDstFilename(line, config))) {
         currentFile.newName = values[1];
         currentFile.language = getExtension(currentFile.newName, currentFile.language);
       } else if (currentFile && utils.startsWith(line, '@@')) {
@@ -226,6 +226,27 @@
     return language;
   }
 
+  function getSrcFilename(line, cfg) {
+    var prefixes = ["a\\/", "i\\/", "w\\/", "c\\/", "o\\/"];
+
+    if (cfg.srcPrefix) prefixes.push(cfg.srcPrefix);
+
+    return _getFilename('---', line, prefixes);
+  }
+
+  function getDstFilename(line, cfg) {
+    var prefixes = ["b\\/", "i\\/", "w\\/", "c\\/", "o\\/"];
+
+    if (cfg.dstPrefix) prefixes.push(cfg.dstPrefix);
+
+    return _getFilename('\\+\\+\\+', line, prefixes);
+  }
+
+  function _getFilename(linePrefix, line, prefixes) {
+    var prefixesStr = prefixes.join("|");
+    return new RegExp('^' + linePrefix + ' "?(?:' + prefixesStr + ')(.+?)"?$').exec(line);
+  }
+
   module.exports.DiffParser = new DiffParser();
 
 })();

src/diff2html.js

@@ -26,8 +26,9 @@
   /*
    * Generates json object from string diff input
    */
-  Diff2Html.prototype.getJsonFromDiff = function(diffInput) {
-    return diffParser.generateDiffJson(diffInput);
+  Diff2Html.prototype.getJsonFromDiff = function(diffInput, config) {
+    var configOrEmpty = config || {};
+    return diffParser.generateDiffJson(diffInput, configOrEmpty);
   };
 
   /*
@@ -38,7 +39,7 @@
 
     var diffJson = diffInput;
     if (!configOrEmpty.inputFormat || configOrEmpty.inputFormat === 'diff') {
-      diffJson = diffParser.generateDiffJson(diffInput);
+      diffJson = diffParser.generateDiffJson(diffInput, configOrEmpty);
     }
 
     var fileList = '';

test/diff-parser-tests.js

@@ -64,5 +64,47 @@ describe('DiffParser', function() {
       assert.equal(1, file1.blocks.length);
     }
 
+    it('should parse diff with special characters', function() {
+      var diff =
+        'diff --git "a/bla with \ttab.scala" "b/bla with \ttab.scala"\n' +
+        'index 4c679d7..e9bd385 100644\n' +
+        '--- "a/bla with \ttab.scala"\n' +
+        '+++ "b/bla with \ttab.scala"\n' +
+        '@@ -1 +1,2 @@\n' +
+        '-cenas\n' +
+        '+cenas com ananas\n' +
+        '+bananas';
+
+      var result = Diff2Html.getJsonFromDiff(diff);
+      var file1 = result[0];
+      assert.equal(1, result.length);
+      assert.equal(2, file1.addedLines);
+      assert.equal(1, file1.deletedLines);
+      assert.equal('bla with \ttab.scala', file1.oldName);
+      assert.equal('bla with \ttab.scala', file1.newName);
+      assert.equal(1, file1.blocks.length);
+    });
+
+    it('should parse diff with prefix', function() {
+      var diff =
+        'diff --git "\tbla with \ttab.scala" "\tbla with \ttab.scala"\n' +
+        'index 4c679d7..e9bd385 100644\n' +
+        '--- "\tbla with \ttab.scala"\n' +
+        '+++ "\tbla with \ttab.scala"\n' +
+        '@@ -1 +1,2 @@\n' +
+        '-cenas\n' +
+        '+cenas com ananas\n' +
+        '+bananas';
+
+      var result = Diff2Html.getJsonFromDiff(diff, {"srcPrefix": "\t", "dstPrefix": "\t"});
+      var file1 = result[0];
+      assert.equal(1, result.length);
+      assert.equal(2, file1.addedLines);
+      assert.equal(1, file1.deletedLines);
+      assert.equal('bla with \ttab.scala', file1.oldName);
+      assert.equal('bla with \ttab.scala', file1.newName);
+      assert.equal(1, file1.blocks.length);
+    });
+
   });
 });