Changeset - 52e756b40a2b
[Not reviewed]
default
0 1 0
Mads Kiilerich - 8 years ago 2017-10-03 00:14:40
mads@kiilerich.com
diffs: do check of cut_off_limit on the raw diff before any parsing and allocation

This redefines the exact meaning of cut_off_limit but it will still be
approximately the same thing. It makes it more a limitation of what amount of
work should be done, more than how much html should be outputted.

It could make sense to push the limit further back to vcs to also prevent
computing or allocating memory for huge diffs.
1 file changed with 12 insertions and 26 deletions:
0 comments (0 inline, 0 general)
kallithea/lib/diffs.py
Show inline comments
 
@@ -134,16 +134,12 @@ MOD_FILENODE = 3
 
RENAMED_FILENODE = 4
 
COPIED_FILENODE = 5
 
CHMOD_FILENODE = 6
 
BIN_FILENODE = 7
 

	
 

	
 
class DiffLimitExceeded(Exception):
 
    pass
 

	
 

	
 
class DiffProcessor(object):
 
    """
 
    Give it a unified or git diff and it returns a list of the files that were
 
    mentioned in the diff together with a dict of meta information that
 
    can be used to render it in a HTML template.
 
    """
 
@@ -201,30 +197,21 @@ class DiffProcessor(object):
 
        if not isinstance(diff, basestring):
 
            raise Exception('Diff must be a basestring got %s instead' % type(diff))
 

	
 
        self._diff = diff
 
        self.adds = 0
 
        self.removes = 0
 
        # calculate diff size
 
        self.diff_limit = diff_limit
 
        self.cur_diff_size = 0
 
        self.limited_diff = False
 
        self.vcs = vcs
 
        self.parsed = self._parse_gitdiff(inline_diff=inline_diff)
 

	
 
    def _escaper(self, string):
 
        """
 
        Do HTML escaping/markup and check the diff limit
 
        Do HTML escaping/markup
 
        """
 
        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 '<'
 
@@ -301,12 +288,16 @@ class DiffProcessor(object):
 
        _files = [] # list of dicts with meta info and chunks
 

	
 
        starts = [m.start() for m in self._diff_git_re.finditer(self._diff)]
 
        starts.append(len(self._diff))
 

	
 
        for start, end in zip(starts, starts[1:]):
 
            if self.diff_limit and end > self.diff_limit:
 
                self.limited_diff = True
 
                continue
 

	
 
            head, diff_lines = self._get_header(buffer(self._diff, start, end - start))
 

	
 
            op = None
 
            stats = {
 
                'added': 0,
 
                'deleted': 0,
 
@@ -360,24 +351,19 @@ class DiffProcessor(object):
 
                    op = 'M'
 
                    stats['binary'] = True
 
                    stats['ops'][MOD_FILENODE] = 'modified file'
 

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

	
 
                except DiffLimitExceeded:
 
                    self.limited_diff = True
 
                    break
 
                chunks, added, deleted = self._parse_lines(diff_lines)
 
                stats['binary'] = False
 
                stats['added'] = added
 
                stats['deleted'] = deleted
 
                # explicit mark that it's a modified file
 
                if op == 'M':
 
                    stats['ops'][MOD_FILENODE] = 'modified file'
 
            else:  # Git binary patch (or empty diff)
 
                # Git binary patch
 
                if head['bin_patch']:
 
                    stats['ops'][BIN_FILENODE] = 'binary diff not shown'
 
                chunks = []
 

	
0 comments (0 inline, 0 general)