Parse binary patch diffs

src/diff-parser.js

@@ -32,6 +32,9 @@
     var oldLine2 = null; // Used for combined diff
     var newLine = null;
 
+    var possibleOldName;
+    var possibleNewName;
+
     /* Diff Header */
     var oldFileNameHeader = '--- ';
     var newFileNameHeader = '+++ ';
@@ -50,10 +53,23 @@
      * if it has name (to avoid binary files errors)
      */
     function saveFile() {
-      if (currentFile && currentFile.newName) {
+      if (currentFile) {
-        files.push(currentFile);
+        if (!currentFile.oldName) {
-        currentFile = null;
+          currentFile.oldName = possibleOldName;
+        }
+
+        if (!currentFile.newName) {
+          currentFile.newName = possibleNewName;
+        }
+
+        if (currentFile.newName) {
+          files.push(currentFile);
+          currentFile = null;
+        }
       }
+
+      possibleOldName = undefined;
+      possibleNewName = undefined;
     }
 
     /* Create file structure */
@@ -196,6 +212,7 @@
     var index = /^index ([0-9a-z]+)\.\.([0-9a-z]+)\s*(\d{6})?/;
 
     var binaryFiles = /^Binary files (.*) and (.*) differ/;
+    var binaryDiff = /^GIT binary patch/;
 
     /* Combined Diff */
     var combinedIndex = /^index ([0-9a-z]+),([0-9a-z]+)\.\.([0-9a-z]+)/;
@@ -211,12 +228,23 @@
         return;
       }
 
+      // Used to store regex capture groups
+      var values;
+
       var prevLine = diffLines[lineIndex - 1];
       var nxtLine = diffLines[lineIndex + 1];
       var afterNxtLine = diffLines[lineIndex + 2];
 
       if (utils.startsWith(line, 'diff')) {
         startFile();
+
+        // diff --git a/blocked_delta_results.png b/blocked_delta_results.png
+        var gitDiffStart = /^diff --git "?(.+)"? "?(.+)"?/;
+        if ((values = gitDiffStart.exec(line))) {
+          possibleOldName = _getFilename(null, values[1], config.dstPrefix);
+          possibleNewName = _getFilename(null, values[2], config.srcPrefix);
+        }
+
         currentFile.isGitDiff = true;
         return;
       }
@@ -234,8 +262,6 @@
         startFile();
       }
 
-      var values;
-
       /*
        * We need to make sure that we have the three lines of the header.
        * This avoids cases like the ones described in:
@@ -328,8 +354,12 @@
         currentFile.isRename = true;
       } else if ((values = binaryFiles.exec(line))) {
         currentFile.isBinary = true;
-        currentFile.oldName = _getFilename(null, values[1], [config.srcPrefix]);
+        currentFile.oldName = _getFilename(null, values[1], config.srcPrefix);
-        currentFile.newName = _getFilename(null, values[2], [config.dstPrefix]);
+        currentFile.newName = _getFilename(null, values[2], config.dstPrefix);
+        startBlock('Binary file');
+      } else if ((values = binaryDiff.exec(line))) {
+        currentFile.isBinary = true;
+        startBlock(line);
       } else if ((values = similarityIndex.exec(line))) {
         currentFile.unchangedPercentage = values[1];
       } else if ((values = dissimilarityIndex.exec(line))) {
@@ -369,26 +399,19 @@
   }
 
   function getSrcFilename(line, cfg) {
-    var prefixes = ['a/', 'i/', 'w/', 'c/', 'o/'];
+    return _getFilename('---', line, cfg.srcPrefix);
-
-    if (cfg.srcPrefix) {
-      prefixes.push(cfg.srcPrefix);
-    }
-
-    return _getFilename('---', line, prefixes);
   }
 
   function getDstFilename(line, cfg) {
-    var prefixes = ['b/', 'i/', 'w/', 'c/', 'o/'];
+    return _getFilename('\\+\\+\\+', line, cfg.dstPrefix);
-
-    if (cfg.dstPrefix) {
-      prefixes.push(cfg.dstPrefix);
-    }
-
-    return _getFilename('\\+\\+\\+', line, prefixes);
   }
 
-  function _getFilename(linePrefix, line, prefixes) {
+  function _getFilename(linePrefix, line, extraPrefix) {
+    var prefixes = ['a/', 'b/', 'i/', 'w/', 'c/', 'o/'];
+    if (extraPrefix) {
+      prefixes.push(extraPrefix);
+    }
+
     var FilenameRegExp;
     if (linePrefix) {
       FilenameRegExp = new RegExp('^' + linePrefix + ' "?(.+?)"?$');

test/diff-parser-tests.js

@@ -627,10 +627,23 @@ describe('DiffParser', function() {
         'diff --git "\ttardis.png" "\ttardis.png"\n' +
         'new file mode 100644\n' +
         'index 0000000..d503a29\n' +
-        'Binary files /dev/null and "\ttardis.png" differ\n';
+        'Binary files /dev/null and "\ttardis.png" differ\n' +
+        'diff --git a/src/test-bar.js b/src/test-baz.js\n' +
+        'similarity index 98%\n' +
+        'rename from src/test-bar.js\n' +
+        'rename to src/test-baz.js\n' +
+        'index e01513b..f14a870 100644\n' +
+        '--- a/src/test-bar.js\n' +
+        '+++ b/src/test-baz.js\n' +
+        '@@ -1,4 +1,32 @@\n' +
+        ' function foo() {\n' +
+        '-var bar = "Whoops!";\n' +
+        '+var baz = "Whoops!";\n' +
+        ' }\n' +
+        ' ';
 
       var result = DiffParser.generateDiffJson(diff, {'srcPrefix': '\t', 'dstPrefix': '\t'});
-      assert.equal(2, result.length);
+      assert.equal(3, result.length);
 
       var file1 = result[0];
       assert.equal(2, file1.addedLines);
@@ -644,6 +657,76 @@ describe('DiffParser', function() {
       var file2 = result[1];
       assert.equal('/dev/null', file2.oldName);
       assert.equal('tardis.png', file2.newName);
+
+      var file3 = result[2];
+      assert.equal(1, file3.addedLines);
+      assert.equal(1, file3.deletedLines);
+      assert.equal('src/test-bar.js', file3.oldName);
+      assert.equal('src/test-baz.js', file3.newName);
+      assert.equal(1, file3.blocks.length);
+      assert.equal(5, file3.blocks[0].lines.length);
+      var linesContent = file3.blocks[0].lines.map(function(line) {
+        return line.content;
+      });
+      assert.deepEqual(linesContent,
+        [' function foo() {', '-var bar = "Whoops!";', '+var baz = "Whoops!";', ' }', ' ']);
+    });
+
+    it('should parse binary with content', function() {
+      var diff =
+        'diff --git a/favicon.png b/favicon.png\n' +
+        'deleted file mode 100644\n' +
+        'index 2a9d516a5647205d7be510dd0dff93a3663eff6f..0000000000000000000000000000000000000000\n' +
+        'GIT binary patch\n' +
+        'literal 0\n' +
+        'HcmV?d00001\n' +
+        '\n' +
+        'literal 471\n' +
+        'zcmeAS@N?(olHy`uVBq!ia0vp^0wB!61|;P_|4#%`EX7WqAsj$Z!;#Vf<Z~8yL>4nJ\n' +
+        'za0`Jj<E6WGe}IBwC9V-A&PAz-C7Jno3L%-fsSJk3`UaNzMkcGzh!g=;$beJ?=ckpF\n' +
+        'zCl;kLIHu$$r7E~(7NwTw7iAYKI0u`(*t4mJfq_xq)5S5wqIc=!hrWj$cv|<b{x!c(\n' +
+        'z;3r#y;31Y&=1q>qPVOAS4ANVKzqmCp=Cty@U^(7zk!jHsvT~YI{F^=Ex6g|gox78w\n' +
+        'z+Sn2Du3GS9U7qU`1*NYYlJi3u-!<?H-eky}wyIIL;8VU@wCDrb0``&v(jQ*DWSR4K\n' +
+        'zPq(3;isEyho{emNa=%%!jDPE`l3u;5d=q=<+v8kO-=C`*G#t-*AiE-D>-_B#8k9H0\n' +
+        'zGl{FnZs<2$wz5^=Q2h-1XI^s{LQL1#T4epqNPC%Orl(tD_@!*EY++~^Lt2<2&!&%=\n' +
+        'z`m>(TYj6uS7jDdt=eH>iOyQg(QMR<-Fw8)Dk^ZG)XQTuzEgl{`GpS?Cfq9818R9~=\n' +
+        'z{&h9@9n8F^?|qusoPy{k#%tVHzu7H$t26CR`BJZk*Ixf&u36WuS=?6m2^ho-p00i_\n' +
+        'I>zopr0Nz-&lmGw#\n' +
+        'diff --git a/src/test-bar.js b/src/test-baz.js\n' +
+        'similarity index 98%\n' +
+        'rename from src/test-bar.js\n' +
+        'rename to src/test-baz.js\n' +
+        'index e01513b..f14a870 100644\n' +
+        '--- a/src/test-bar.js\n' +
+        '+++ b/src/test-baz.js\n' +
+        '@@ -1,4 +1,32 @@\n' +
+        ' function foo() {\n' +
+        '-var bar = "Whoops!";\n' +
+        '+var baz = "Whoops!";\n' +
+        ' }\n' +
+        ' ';
+
+      var result = DiffParser.generateDiffJson(diff);
+      assert.equal(2, result.length);
+
+      var file1 = result[0];
+      assert.equal('favicon.png', file1.oldName);
+      assert.equal('favicon.png', file1.newName);
+      assert.equal(1, file1.blocks.length);
+      assert.equal(0, file1.blocks[0].lines.length);
+
+      var file2 = result[1];
+      assert.equal(1, file2.addedLines);
+      assert.equal(1, file2.deletedLines);
+      assert.equal('src/test-bar.js', file2.oldName);
+      assert.equal('src/test-baz.js', file2.newName);
+      assert.equal(1, file2.blocks.length);
+      assert.equal(5, file2.blocks[0].lines.length);
+      var linesContent = file2.blocks[0].lines.map(function(line) {
+        return line.content;
+      });
+      assert.deepEqual(linesContent,
+        [' function foo() {', '-var bar = "Whoops!";', '+var baz = "Whoops!";', ' }', ' ']);
     });
   });
 });