Changeset - 2bf5e8731154
[Not reviewed]
stable
0 3 0
Mads Kiilerich - 10 years ago 2015-12-25 12:33:13
madski@unity3d.com
git: fix parsing of git mode change + rename diffs

Assume mode lines come first, like for hg diffs.
3 files changed with 16 insertions and 3 deletions:
0 comments (0 inline, 0 general)
kallithea/lib/diffs.py
Show inline comments
 
@@ -67,197 +67,197 @@ def wrapped_diff(filenode_old, filenode_
 
        stats = (0, 0)
 
        size = 0
 

	
 
    elif cut_off_limit != -1 and (cut_off_limit is None or
 
    (filenode_old.size < cut_off_limit and filenode_new.size < cut_off_limit)):
 

	
 
        f_gitdiff = get_gitdiff(filenode_old, filenode_new,
 
                                ignore_whitespace=ignore_whitespace,
 
                                context=line_context)
 
        diff_processor = DiffProcessor(f_gitdiff, format='gitdiff')
 

	
 
        diff = diff_processor.as_html(enable_comments=enable_comments)
 
        stats = diff_processor.stat()
 
        size = len(diff or '')
 
    else:
 
        diff = wrap_to_table(_('Changeset was too big and was cut off, use '
 
                               'diff menu to display this diff'))
 
        stats = (0, 0)
 
        size = 0
 
    if not diff:
 
        submodules = filter(lambda o: isinstance(o, SubModuleNode),
 
                            [filenode_new, filenode_old])
 
        if submodules:
 
            diff = wrap_to_table(escape('Submodule %r' % submodules[0]))
 
        else:
 
            diff = wrap_to_table(_('No changes detected'))
 

	
 
    cs1 = filenode_old.changeset.raw_id
 
    cs2 = filenode_new.changeset.raw_id
 

	
 
    return size, cs1, cs2, diff, stats
 

	
 

	
 
def get_gitdiff(filenode_old, filenode_new, ignore_whitespace=True, context=3):
 
    """
 
    Returns git style diff between given ``filenode_old`` and ``filenode_new``.
 

	
 
    :param ignore_whitespace: ignore whitespaces in diff
 
    """
 
    # make sure we pass in default context
 
    context = context or 3
 
    submodules = filter(lambda o: isinstance(o, SubModuleNode),
 
                        [filenode_new, filenode_old])
 
    if submodules:
 
        return ''
 

	
 
    for filenode in (filenode_old, filenode_new):
 
        if not isinstance(filenode, FileNode):
 
            raise VCSError("Given object should be FileNode object, not %s"
 
                % filenode.__class__)
 

	
 
    repo = filenode_new.changeset.repository
 
    old_raw_id = getattr(filenode_old.changeset, 'raw_id', repo.EMPTY_CHANGESET)
 
    new_raw_id = getattr(filenode_new.changeset, 'raw_id', repo.EMPTY_CHANGESET)
 

	
 
    vcs_gitdiff = repo.get_diff(old_raw_id, new_raw_id, filenode_new.path,
 
                                ignore_whitespace, context)
 
    return vcs_gitdiff
 

	
 
NEW_FILENODE = 1
 
DEL_FILENODE = 2
 
MOD_FILENODE = 3
 
RENAMED_FILENODE = 4
 
COPIED_FILENODE = 5
 
CHMOD_FILENODE = 6
 
BIN_FILENODE = 7
 

	
 

	
 
class DiffLimitExceeded(Exception):
 
    pass
 

	
 

	
 
class LimitedDiffContainer(object):
 

	
 
    def __init__(self, diff_limit, cur_diff_size, diff):
 
        self.diff = diff
 
        self.diff_limit = diff_limit
 
        self.cur_diff_size = cur_diff_size
 

	
 
    def __iter__(self):
 
        for l in self.diff:
 
            yield l
 

	
 

	
 
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.
 
    """
 
    _chunk_re = re.compile(r'^@@ -(\d+)(?:,(\d+))? \+(\d+)(?:,(\d+))? @@(.*)')
 
    _newline_marker = re.compile(r'^\\ No newline at end of file')
 
    _git_header_re = re.compile(r"""
 
        # has already been split on this:
 
        # ^diff[ ]--git
 
            [ ]a/(?P<a_path>.+?)[ ]b/(?P<b_path>.+?)\n
 
        (?:^old[ ]mode[ ](?P<old_mode>\d+)\n
 
           ^new[ ]mode[ ](?P<new_mode>\d+)(?:\n|$))?
 
        (?:^similarity[ ]index[ ](?P<similarity_index>\d+)%\n
 
           ^rename[ ]from[ ](?P<rename_from>.+)\n
 
           ^rename[ ]to[ ](?P<rename_to>.+)(?:\n|$))?
 
        (?:^old[ ]mode[ ](?P<old_mode>\d+)\n
 
           ^new[ ]mode[ ](?P<new_mode>\d+)(?:\n|$))?
 
        (?:^new[ ]file[ ]mode[ ](?P<new_file_mode>.+)(?:\n|$))?
 
        (?:^deleted[ ]file[ ]mode[ ](?P<deleted_file_mode>.+)(?:\n|$))?
 
        (?:^index[ ](?P<a_blob_id>[0-9A-Fa-f]+)
 
            \.\.(?P<b_blob_id>[0-9A-Fa-f]+)[ ]?(?P<b_mode>.+)?(?:\n|$))?
 
        (?:^(?P<bin_patch>GIT[ ]binary[ ]patch)(?:\n|$))?
 
        (?:^---[ ](a/(?P<a_file>.+?)|/dev/null)\t?(?:\n|$))?
 
        (?:^\+\+\+[ ](b/(?P<b_file>.+?)|/dev/null)\t?(?:\n|$))?
 
    """, re.VERBOSE | re.MULTILINE)
 
    _hg_header_re = re.compile(r"""
 
        # has already been split on this:
 
        # ^diff[ ]--git
 
            [ ]a/(?P<a_path>.+?)[ ]b/(?P<b_path>.+?)\n
 
        (?:^old[ ]mode[ ](?P<old_mode>\d+)\n
 
           ^new[ ]mode[ ](?P<new_mode>\d+)(?:\n|$))?
 
        (?:^similarity[ ]index[ ](?P<similarity_index>\d+)%(?:\n|$))?
 
        (?:^rename[ ]from[ ](?P<rename_from>.+)\n
 
           ^rename[ ]to[ ](?P<rename_to>.+)(?:\n|$))?
 
        (?:^copy[ ]from[ ](?P<copy_from>.+)\n
 
           ^copy[ ]to[ ](?P<copy_to>.+)(?:\n|$))?
 
        (?:^new[ ]file[ ]mode[ ](?P<new_file_mode>.+)(?:\n|$))?
 
        (?:^deleted[ ]file[ ]mode[ ](?P<deleted_file_mode>.+)(?:\n|$))?
 
        (?:^index[ ](?P<a_blob_id>[0-9A-Fa-f]+)
 
            \.\.(?P<b_blob_id>[0-9A-Fa-f]+)[ ]?(?P<b_mode>.+)?(?:\n|$))?
 
        (?:^(?P<bin_patch>GIT[ ]binary[ ]patch)(?:\n|$))?
 
        (?:^---[ ](a/(?P<a_file>.+?)|/dev/null)\t?(?:\n|$))?
 
        (?:^\+\+\+[ ](b/(?P<b_file>.+?)|/dev/null)\t?(?:\n|$))?
 
    """, re.VERBOSE | re.MULTILINE)
 

	
 
    # Used for inline highlighter word split, must match the substitutions in _escaper
 
    _token_re = re.compile(r'()(&amp;|&lt;|&gt;|<u>\t</u>|<u class="cr"></u>| <i></i>|\W+?)')
 

	
 
    _escape_re = re.compile(r'(&)|(<)|(>)|(\t)|(\r)|(?<=.)( \n| $)')
 

	
 

	
 
    def __init__(self, diff, vcs='hg', format='gitdiff', diff_limit=None):
 
        """
 
        :param diff:   a text in diff format
 
        :param vcs: type of version control hg or git
 
        :param format: format of diff passed, `udiff` or `gitdiff`
 
        :param diff_limit: define the size of diff that is considered "big"
 
            based on that parameter cut off will be triggered, set to None
 
            to show full diff
 
        """
 
        if not isinstance(diff, basestring):
 
            raise Exception('Diff must be a basestring got %s instead' % type(diff))
 

	
 
        self._diff = diff
 
        self._format = format
 
        self.adds = 0
 
        self.removes = 0
 
        # calculate diff size
 
        self.diff_size = len(diff)
 
        self.diff_limit = diff_limit
 
        self.cur_diff_size = 0
 
        self.parsed = False
 
        self.parsed_diff = []
 
        self.vcs = vcs
 

	
 
        if format == 'gitdiff':
 
            self.differ = self._highlight_line_difflib
 
            self._parser = self._parse_gitdiff
 
        else:
 
            self.differ = self._highlight_line_udiff
 
            self._parser = self._parse_udiff
 

	
 
    def _copy_iterator(self):
 
        """
 
        make a fresh copy of generator, we should not iterate thru
 
        an original as it's needed for repeating operations on
 
        this instance of DiffProcessor
 
        """
 
        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 '&amp;'
 
            if groups[1]:
 
                return '&lt;'
 
            if groups[2]:
 
                return '&gt;'
kallithea/tests/fixtures/git_diff_rename_file.diff
Show inline comments
 
diff --git a/work-horus.xls b/file.xls
 
similarity index 100%
 
rename from work-horus.xls
 
rename to file.xls
 
diff --git a/files/var/www/favicon.ico b/files/var/www/favicon.ico/DEFAULT
 
old mode 100644
 
new mode 100755
 
similarity index 100%
 
rename from files/var/www/favicon.ico
 
rename to files/var/www/favicon.ico/DEFAULT
kallithea/tests/models/test_diff_parsers.py
Show inline comments
 
@@ -40,193 +40,200 @@ DIFF_FIXTURES = {
 
          'ops': {RENAMED_FILENODE: 'file renamed from README to README.rst'}}),
 
    ],
 
    'hg_diff_del_single_binary_file.diff': [
 
        ('US Warszawa.jpg', 'D',
 
         {'added': 0,
 
          'deleted': 0,
 
          'binary': True,
 
          'ops': {DEL_FILENODE: 'deleted file',
 
                  BIN_FILENODE: 'binary diff not shown'}}),
 
    ],
 
    'hg_diff_chmod_and_mod_single_binary_file.diff': [
 
        ('gravatar.png', 'M',
 
         {'added': 0,
 
          'deleted': 0,
 
          'binary': True,
 
          'ops': {CHMOD_FILENODE: 'modified file chmod 100644 => 100755',
 
                  BIN_FILENODE: 'binary diff not shown'}}),
 
    ],
 
    'hg_diff_chmod.diff': [
 
        ('file', 'M',
 
         {'added': 0,
 
          'deleted': 0,
 
          'binary': True,
 
          'ops': {CHMOD_FILENODE: 'modified file chmod 100755 => 100644'}}),
 
    ],
 
    'hg_diff_rename_file.diff': [
 
        ('file_renamed', 'R',
 
         {'added': 0,
 
          'deleted': 0,
 
          'binary': True,
 
          'ops': {RENAMED_FILENODE: 'file renamed from file to file_renamed'}}),
 
    ],
 
    'hg_diff_rename_and_chmod_file.diff': [
 
        ('README', 'R',
 
         {'added': 0,
 
          'deleted': 0,
 
          'binary': True,
 
          'ops': {CHMOD_FILENODE: 'modified file chmod 100644 => 100755',
 
                  RENAMED_FILENODE: 'file renamed from README.rst to README'}}),
 
    ],
 
    'hg_diff_binary_and_normal.diff': [
 
        ('img/baseline-10px.png', 'A',
 
         {'added': 0,
 
          'deleted': 0,
 
          'binary': True,
 
          'ops': {NEW_FILENODE: 'new file 100644',
 
                  BIN_FILENODE: 'binary diff not shown'}}),
 
        ('img/baseline-20px.png', 'D',
 
         {'added': 0,
 
          'deleted': 0,
 
          'binary': True,
 
          'ops': {DEL_FILENODE: 'deleted file',
 
                  BIN_FILENODE: 'binary diff not shown'}}),
 
        ('index.html', 'M',
 
         {'added': 3,
 
          'deleted': 2,
 
          'binary': False,
 
          'ops': {MOD_FILENODE: 'modified file'}}),
 
        ('js/global.js', 'D',
 
         {'added': 0,
 
          'deleted': 75,
 
          'binary': False,
 
          'ops': {DEL_FILENODE: 'deleted file'}}),
 
        ('js/jquery/hashgrid.js', 'A',
 
         {'added': 340,
 
          'deleted': 0,
 
          'binary': False,
 
          'ops': {NEW_FILENODE: 'new file 100755'}}),
 
        ('less/docs.less', 'M',
 
         {'added': 34,
 
          'deleted': 0,
 
          'binary': False,
 
          'ops': {MOD_FILENODE: 'modified file'}}),
 
        ('less/scaffolding.less', 'M',
 
         {'added': 1,
 
          'deleted': 3,
 
          'binary': False,
 
          'ops': {MOD_FILENODE: 'modified file'}}),
 
        ('readme.markdown', 'M',
 
         {'added': 1,
 
          'deleted': 10,
 
          'binary': False,
 
          'ops': {MOD_FILENODE: 'modified file'}}),
 
    ],
 
    'git_diff_chmod.diff': [
 
        ('work-horus.xls', 'M',
 
         {'added': 0,
 
          'deleted': 0,
 
          'binary': True,
 
          'ops': {CHMOD_FILENODE: 'modified file chmod 100644 => 100755'}})
 
    ],
 
    'git_diff_rename_file.diff': [
 
        ('file.xls', 'R',
 
         {'added': 0,
 
          'deleted': 0,
 
          'binary': True,
 
          'ops': {RENAMED_FILENODE: 'file renamed from work-horus.xls to file.xls'}})
 
          'ops': {RENAMED_FILENODE: 'file renamed from work-horus.xls to file.xls'}}),
 
        ('files/var/www/favicon.ico/DEFAULT',
 
         'R',
 
         {'added': 0,
 
          'binary': True,
 
          'deleted': 0,
 
          'ops': {4: 'file renamed from files/var/www/favicon.ico to files/var/www/favicon.ico/DEFAULT',
 
                  6: 'modified file chmod 100644 => 100755'}})
 
    ],
 
    'git_diff_mod_single_binary_file.diff': [
 
        ('US Warszawa.jpg', 'M',
 
         {'added': 0,
 
          'deleted': 0,
 
          'binary': True,
 
          'ops': {MOD_FILENODE: 'modified file',
 
                  BIN_FILENODE: 'binary diff not shown'}})
 
    ],
 
    'git_diff_binary_and_normal.diff': [
 
        ('img/baseline-10px.png', 'A',
 
         {'added': 0,
 
          'deleted': 0,
 
          'binary': True,
 
          'ops': {NEW_FILENODE: 'new file 100644',
 
                  BIN_FILENODE: 'binary diff not shown'}}),
 
        ('img/baseline-20px.png', 'D',
 
         {'added': 0,
 
          'deleted': 0,
 
          'binary': True,
 
          'ops': {DEL_FILENODE: 'deleted file',
 
                  BIN_FILENODE: 'binary diff not shown'}}),
 
        ('index.html', 'M',
 
         {'added': 3,
 
          'deleted': 2,
 
          'binary': False,
 
          'ops': {MOD_FILENODE: 'modified file'}}),
 
        ('js/global.js', 'D',
 
         {'added': 0,
 
          'deleted': 75,
 
          'binary': False,
 
          'ops': {DEL_FILENODE: 'deleted file'}}),
 
        ('js/jquery/hashgrid.js', 'A',
 
         {'added': 340,
 
          'deleted': 0,
 
          'binary': False,
 
          'ops': {NEW_FILENODE: 'new file 100755'}}),
 
        ('less/docs.less', 'M',
 
         {'added': 34,
 
          'deleted': 0,
 
          'binary': False,
 
          'ops': {MOD_FILENODE: 'modified file'}}),
 
        ('less/scaffolding.less', 'M',
 
         {'added': 1,
 
          'deleted': 3,
 
          'binary': False,
 
          'ops': {MOD_FILENODE: 'modified file'}}),
 
        ('readme.markdown', 'M',
 
         {'added': 1,
 
          'deleted': 10,
 
          'binary': False,
 
          'ops': {MOD_FILENODE: 'modified file'}}),
 
    ],
 
    'diff_with_diff_data.diff': [
 
        ('vcs/backends/base.py', 'M',
 
         {'added': 18,
 
          'deleted': 2,
 
          'binary': False,
 
          'ops': {MOD_FILENODE: 'modified file'}}),
 
        ('vcs/backends/git/repository.py', 'M',
 
         {'added': 46,
 
          'deleted': 15,
 
          'binary': False,
 
          'ops': {MOD_FILENODE: 'modified file'}}),
 
        ('vcs/backends/hg.py', 'M',
 
         {'added': 22,
 
          'deleted': 3,
 
          'binary': False,
 
          'ops': {MOD_FILENODE: 'modified file'}}),
 
        ('vcs/tests/test_git.py', 'M',
 
         {'added': 5,
 
          'deleted': 5,
 
          'binary': False,
 
          'ops': {MOD_FILENODE: 'modified file'}}),
 
        ('vcs/tests/test_repository.py', 'M',
 
         {'added': 174,
 
          'deleted': 2,
 
          'binary': False,
 
          'ops': {MOD_FILENODE: 'modified file'}}),
 
    ],
 
    'git_diff_modify_binary_file.diff': [
 
        ('file.name', 'M',
 
         {'added': 0,
 
          'deleted': 0,
 
          'binary': True,
 
          'ops': {MOD_FILENODE: 'modified file',
 
                  BIN_FILENODE: 'binary diff not shown'}})
 
    ],
 
    'hg_diff_copy_file.diff': [
 
        ('file2', 'M',
 
         {'added': 0,
 
          'deleted': 0,
 
          'binary': True,
 
          'ops': {COPIED_FILENODE: 'file copied from file1 to file2'}}),
 
    ],
 
    'hg_diff_copy_and_modify_file.diff': [
0 comments (0 inline, 0 general)