Changeset - 42d18c125aa8
[Not reviewed]
default
0 1 0
Mads Kiilerich - 11 years ago 2014-12-15 13:47:36
madski@unity3d.com
diff: refactor header parsing, give good error messages when it fails
1 file changed with 9 insertions and 6 deletions:
0 comments (0 inline, 0 general)
kallithea/lib/diffs.py
Show inline comments
 
@@ -305,106 +305,109 @@ class DiffProcessor(object):
 
            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)
 
            do(next_)
 

	
 
    def _get_header(self, diff_chunk):
 
        """
 
        parses the diff header, and returns parts, and leftover diff
 
        parts consists of 14 elements::
 

	
 
            a_path, b_path, similarity_index, rename_from, rename_to,
 
            old_mode, new_mode, new_file_mode, deleted_file_mode,
 
            a_blob_id, b_blob_id, b_mode, a_file, b_file
 

	
 
        :param diff_chunk:
 
        """
 

	
 
        match = None
 
        if self.vcs == 'git':
 
            match = self._git_header_re.match(diff_chunk)
 
            diff = diff_chunk[match.end():]
 
            return match.groupdict(), imap(self._escaper, diff.splitlines(1))
 
        elif self.vcs == 'hg':
 
            match = self._hg_header_re.match(diff_chunk)
 
            diff = diff_chunk[match.end():]
 
            return match.groupdict(), imap(self._escaper, diff.splitlines(1))
 
        else:
 
        if match is None:
 
            raise Exception('VCS type %s is not supported' % self.vcs)
 
        groups = match.groupdict()
 
        rest = diff_chunk[match.end():]
 
        if rest and not rest.startswith('@') and not rest.startswith('literal '):
 
            raise Exception('cannot parse diff header: %r followed by %r' % (diff_chunk[:match.end()], rest[:1000]))
 
        difflines = imap(self._escaper, rest.splitlines(True))
 
        return groups, difflines
 

	
 
    def _clean_line(self, line, command):
 
        if command in ['+', '-', ' ']:
 
            #only modify the line if it's actually a diff thing
 
            line = line[1:]
 
        return line
 

	
 
    def _parse_gitdiff(self, inline_diff=True):
 
        _files = []
 
        diff_container = lambda arg: arg
 

	
 
        ##split the diff in chunks of separate --git a/file b/file chunks
 
        for raw_diff in ('\n' + self._diff).split('\ndiff --git')[1:]:
 
            head, diff = self._get_header(raw_diff)
 

	
 
            op = None
 
            stats = {
 
                'added': 0,
 
                'deleted': 0,
 
                'binary': False,
 
                'ops': {},
 
            }
 

	
 
            if head['deleted_file_mode']:
 
                op = 'D'
 
                stats['binary'] = True
 
                stats['ops'][DEL_FILENODE] = 'deleted file'
 

	
 
            elif head['new_file_mode']:
 
                op = 'A'
 
                stats['binary'] = True
 
                stats['ops'][NEW_FILENODE] = 'new file %s' % head['new_file_mode']
 
            else:  # modify operation, can be cp, rename, chmod
 
                # CHMOD
 
                if head['new_mode'] and head['old_mode']:
 
                    op = 'M'
 
                    stats['binary'] = True
 
                    stats['ops'][CHMOD_FILENODE] = ('modified file chmod %s => %s'
 
                                        % (head['old_mode'], head['new_mode']))
 
                # RENAME
 
                if (head['rename_from'] and head['rename_to']
 
                      and head['rename_from'] != head['rename_to']):
 
                    op = 'R'
 
                    stats['binary'] = True
 
                    stats['ops'][RENAMED_FILENODE] = ('file renamed from %s to %s'
 
                                    % (head['rename_from'], head['rename_to']))
 
                # COPY
 
                if head.get('copy_from') and head.get('copy_to'):
 
@@ -455,97 +458,97 @@ class DiffProcessor(object):
 
            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
 
        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)
 
                except StopIteration:
 
                    pass
 

	
 
        return diff_container(_files)
 

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

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

	
 
        lineiter = iter(diff)
 
        stats = [0, 0]
 

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

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

	
 
                match = self._chunk_re.match(line)
 

	
 
                if not match:
 
                    break
 

	
 
                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,
 
                        })
 

	
 
                line = lineiter.next()
 

	
 
                while old_line < old_end or new_line < new_end:
 
                    command = ' '
 
                    if line:
 
                        command = line[0]
 

	
 
                    affects_old = affects_new = False
 

	
 
                    # ignore those if we don't expect them
0 comments (0 inline, 0 general)