Changeset - 6b723a49a9a1
[Not reviewed]
default
0 3 0
Mads Kiilerich - 9 years ago 2016-08-12 03:13:59
madski@unity3d.com
diff: only highlight of difference between del and add line for one-liners

Comparing a single del line with the first of several add lines would often be
quite arbitrary and misleading.

Note: the non-gitdiff code paths in diffs.py are incomplete and unused and
unusable.
3 files changed with 35 insertions and 18 deletions:
0 comments (0 inline, 0 general)
kallithea/lib/diffs.py
Show inline comments
 
@@ -238,105 +238,103 @@ class DiffProcessor(object):
 
        """
 
        self.__udiff, iterator_copy = tee(self.__udiff)
 
        return iterator_copy
 

	
 
    def _escaper(self, string):
 
        """
 
        Escaper for diff escapes special chars and checks the diff limit
 

	
 
        :param string:
 
        """
 

	
 
        self.cur_diff_size += len(string)
 

	
 
        # escaper gets iterated on each .next() call and it checks if each
 
        # parsed line doesn't exceed the diff limit
 
        if self.diff_limit is not None and self.cur_diff_size > self.diff_limit:
 
            raise DiffLimitExceeded('Diff Limit Exceeded')
 

	
 
        def substitute(m):
 
            groups = m.groups()
 
            if groups[0]:
 
                return '&'
 
            if groups[1]:
 
                return '<'
 
            if groups[2]:
 
                return '>'
 
            if groups[3]:
 
                return '<u>\t</u>'
 
            if groups[4]:
 
                return '<u class="cr"></u>'
 
            if groups[5]:
 
                return ' <i></i>'
 
            assert False
 

	
 
        return self._escape_re.sub(substitute, safe_unicode(string))
 

	
 
    def _line_counter(self, l):
 
        """
 
        Checks each line and bumps total adds/removes for this diff
 

	
 
        :param l:
 
        """
 
        if l.startswith('+') and not l.startswith('+++'):
 
            self.adds += 1
 
        elif l.startswith('-') and not l.startswith('---'):
 
            self.removes += 1
 
        return safe_unicode(l)
 

	
 
    def _highlight_line_difflib(self, line, next_):
 
    def _highlight_line_difflib(self, old, new):
 
        """
 
        Highlight inline changes in both lines.
 
        """
 

	
 
        if line['action'] == 'del':
 
            old, new = line, next_
 
        else:
 
            old, new = next_, line
 
        assert old['action'] == 'del'
 
        assert new['action'] == 'add'
 

	
 
        oldwords = self._token_re.split(old['line'])
 
        newwords = self._token_re.split(new['line'])
 
        sequence = difflib.SequenceMatcher(None, oldwords, newwords)
 

	
 
        oldfragments, newfragments = [], []
 
        for tag, i1, i2, j1, j2 in sequence.get_opcodes():
 
            oldfrag = ''.join(oldwords[i1:i2])
 
            newfrag = ''.join(newwords[j1:j2])
 
            if tag != 'equal':
 
                if oldfrag:
 
                    oldfrag = '<del>%s</del>' % oldfrag
 
                if newfrag:
 
                    newfrag = '<ins>%s</ins>' % newfrag
 
            oldfragments.append(oldfrag)
 
            newfragments.append(newfrag)
 

	
 
        old['line'] = "".join(oldfragments)
 
        new['line'] = "".join(newfragments)
 

	
 
    def _highlight_line_udiff(self, line, next_):
 
        """
 
        Highlight inline changes in both lines.
 
        """
 
        start = 0
 
        limit = min(len(line['line']), len(next_['line']))
 
        while start < limit and line['line'][start] == next_['line'][start]:
 
            start += 1
 
        end = -1
 
        limit -= start
 
        while -end <= limit and line['line'][end] == next_['line'][end]:
 
            end -= 1
 
        end += 1
 
        if start or end:
 
            def do(l):
 
                last = end + len(l['line'])
 
                if l['action'] == 'add':
 
                    tag = 'ins'
 
                else:
 
                    tag = 'del'
 
                l['line'] = '%s<%s>%s</%s>%s' % (
 
                    l['line'][:start],
 
                    tag,
 
                    l['line'][start:last],
 
                    tag,
 
                    l['line'][last:]
 
                )
 
            do(line)
 
@@ -439,109 +437,124 @@ class DiffProcessor(object):
 

	
 
            # a real non-binary diff
 
            if head['a_file'] or head['b_file']:
 
                try:
 
                    chunks, _stats = self._parse_lines(diff)
 
                    stats['binary'] = False
 
                    stats['added'] = _stats[0]
 
                    stats['deleted'] = _stats[1]
 
                    # explicit mark that it's a modified file
 
                    if op == 'M':
 
                        stats['ops'][MOD_FILENODE] = 'modified file'
 

	
 
                except DiffLimitExceeded:
 
                    diff_container = lambda _diff: \
 
                        LimitedDiffContainer(self.diff_limit,
 
                                            self.cur_diff_size, _diff)
 
                    break
 
            else:  # Git binary patch (or empty diff)
 
                # Git binary patch
 
                if head['bin_patch']:
 
                    stats['ops'][BIN_FILENODE] = 'binary diff not shown'
 
                chunks = []
 

	
 
            if op == 'D' and chunks:
 
                # a way of seeing deleted content could perhaps be nice - but
 
                # not with the current UI
 
                chunks = []
 

	
 
            chunks.insert(0, [{
 
                'old_lineno': '',
 
                'new_lineno': '',
 
                'action':     'context',
 
                'line':       msg,
 
                } for _op, msg in stats['ops'].iteritems()
 
                  if _op not in [MOD_FILENODE]])
 

	
 
            _files.append({
 
                'filename':         head['b_path'],
 
                'old_revision':     head['a_blob_id'],
 
                'new_revision':     head['b_blob_id'],
 
                'chunks':           chunks,
 
                'operation':        op,
 
                'stats':            stats,
 
            })
 

	
 
        if not inline_diff:
 
            return diff_container(_files)
 

	
 
        # highlight inline changes
 
        # highlight inline changes when one del is followed by one add
 
        for diff_data in _files:
 
            for chunk in diff_data['chunks']:
 
                lineiter = iter(chunk)
 
                try:
 
                    while 1:
 
                        line = lineiter.next()
 
                        if line['action'] not in ['unmod', 'context']:
 
                            nextline = lineiter.next()
 
                            if nextline['action'] in ['unmod', 'context'] or \
 
                               nextline['action'] == line['action']:
 
                                continue
 
                            self.differ(line, nextline)
 
                    peekline = lineiter.next()
 
                    while True:
 
                        # find a first del line
 
                        while peekline['action'] != 'del':
 
                            peekline = lineiter.next()
 
                        delline = peekline
 
                        peekline = lineiter.next()
 
                        # if not followed by add, eat all following del lines
 
                        if peekline['action'] != 'add':
 
                            while peekline['action'] == 'del':
 
                                peekline = lineiter.next()
 
                            continue
 
                        # found an add - make sure it is the only one
 
                        addline = peekline
 
                        try:
 
                            peekline = lineiter.next()
 
                        except StopIteration:
 
                            # add was last line - ok
 
                            self.differ(delline, addline)
 
                            raise
 
                        if peekline['action'] != 'add':
 
                            # there was only one add line - ok
 
                            self.differ(delline, addline)
 
                except StopIteration:
 
                    pass
 

	
 
        return diff_container(_files)
 

	
 
    def _parse_udiff(self, inline_diff=True):
 
        raise NotImplementedError()
 

	
 
    def _parse_lines(self, diff):
 
        """
 
        Parse the diff and return data for the template.
 
        """
 

	
 
        stats = [0, 0]
 
        (old_line, old_end, new_line, new_end) = (None, None, None, None)
 

	
 
        try:
 
            chunks = []
 
            line = diff.next()
 

	
 
            while True:
 
                lines = []
 
                chunks.append(lines)
 

	
 
                match = self._chunk_re.match(line)
 

	
 
                if not match:
 
                    raise Exception('error parsing diff @@ line %r' % line)
 

	
 
                gr = match.groups()
 
                (old_line, old_end,
 
                 new_line, new_end) = [int(x or 1) for x in gr[:-1]]
 
                old_line -= 1
 
                new_line -= 1
 

	
 
                context = len(gr) == 5
 
                old_end += old_line
 
                new_end += new_line
 

	
 
                if context:
 
                    # skip context only if it's first line
 
                    if int(gr[0]) > 1:
 
                        lines.append({
 
                            'old_lineno': '...',
 
                            'new_lineno': '...',
 
                            'action':     'context',
 
                            'line':       line,
 
                        })
kallithea/tests/fixtures/markuptest.diff
Show inline comments
 
diff --git a/f b/f
 
--- a/f	
 
+++ b/f	
 
@@ -51,5 +51,12 @@
 
@@ -51,6 +51,13 @@
 
 	begin();
 
 	
 
+	int foo;
 
+	int bar; 
 
+	int baz;	
 
+	int space; 
 
+	int tab;	
 
+	
 
  
 
-	#define MAX_STEPS (48)
 
+	
 
+	#define MAX_STEPS (64)
 
 
 
-	#define MIN_STEPS (48)
 
+	#define MIN_STEPS (42)
kallithea/tests/models/test_diff_parsers.py
Show inline comments
 
@@ -253,63 +253,65 @@ DIFF_FIXTURES = {
 
                  CHMOD_FILENODE: 'modified file chmod 100644 => 100755'}}),
 
    ],
 
    'hg_diff_copy_chmod_and_edit_file.diff': [
 
        ('file5', 'M',
 
         {'added': 2,
 
          'deleted': 1,
 
          'binary': False,
 
          'ops': {COPIED_FILENODE: 'file copied from file4 to file5',
 
                  CHMOD_FILENODE: 'modified file chmod 100755 => 100644',
 
                  MOD_FILENODE: 'modified file'}}),
 
    ],
 
    'hg_diff_rename_space_cr.diff': [
 
        ('oh yes', 'R',
 
         {'added': 3,
 
          'deleted': 2,
 
          'binary': False,
 
          'ops': {RENAMED_FILENODE: 'file renamed from oh no to oh yes'}}),
 
    ],
 
}
 

	
 

	
 
class TestDiffLib(TestController):
 

	
 
    @parametrize('diff_fixture', DIFF_FIXTURES)
 
    def test_diff(self, diff_fixture):
 
        diff = fixture.load_resource(diff_fixture, strip=False)
 
        vcs = 'hg'
 
        if diff_fixture.startswith('git_'):
 
            vcs = 'git'
 
        diff_proc = DiffProcessor(diff, vcs=vcs)
 
        diff_proc_d = diff_proc.prepare()
 
        data = [(x['filename'], x['operation'], x['stats']) for x in diff_proc_d]
 
        expected_data = DIFF_FIXTURES[diff_fixture]
 
        assert expected_data == data
 

	
 
    def test_diff_markup(self):
 
        diff = fixture.load_resource('markuptest.diff', strip=False)
 
        diff_proc = DiffProcessor(diff)
 
        diff_proc_d = diff_proc.prepare()
 
        chunks = diff_proc_d[0]['chunks']
 
        assert not chunks[0]
 
        #from pprint import pprint; pprint(chunks[1])
 
        l = ['\n']
 
        for d in chunks[1]:
 
            l.append('%(action)-7s %(new_lineno)3s %(old_lineno)3s %(line)r\n' % d)
 
        s = ''.join(l)
 
        print s
 
        assert s == r'''
 
context ... ... u'@@ -51,5 +51,12 @@\n'
 
context ... ... u'@@ -51,6 +51,13 @@\n'
 
unmod    51  51 u'<u>\t</u>begin();\n'
 
unmod    52  52 u'<u>\t</u>\n'
 
add      53     u'<u>\t</u>int foo;<u class="cr"></u>\n'
 
add      54     u'<u>\t</u>int bar; <u class="cr"></u>\n'
 
add      55     u'<u>\t</u>int baz;<u>\t</u><u class="cr"></u>\n'
 
add      56     u'<u>\t</u>int space; <i></i>'
 
add      57     u'<u>\t</u>int tab;<u>\t</u>\n'
 
add      58     u'<u>\t</u>\n'
 
unmod    59  53 u' <i></i>'
 
del          54 u'<u>\t</u><del>#define MAX_STEPS (48)</del>\n'
 
add      60     u'<u>\t</u><ins><u class="cr"></u></ins>\n'
 
del          54 u'<u>\t</u>#define MAX_STEPS (48)\n'
 
add      60     u'<u>\t</u><u class="cr"></u>\n'
 
add      61     u'<u>\t</u>#define MAX_STEPS (64)<u class="cr"></u>\n'
 
unmod    62  55 u'\n'
 
del          56 u'<u>\t</u>#define MIN_STEPS (<del>48</del>)\n'
 
add      63     u'<u>\t</u>#define MIN_STEPS (<ins>42</ins>)\n'
 
'''
0 comments (0 inline, 0 general)