def _parse_git_diff(self, linenum):
"""Parse a Git-style diff header.
This will parse a diff header containing file mode information,
file operations, and ``diff --git`` lines, and filename information.
Args:
linenum (int):
The current line number.
Returns:
tuple:
A tuple containing the following:
1. The next line number to parse.
2. The populated :py:class:`ParsedDiffFile` instance for this
file, if any.
"""
lines = self.lines
# First check if it is a new file with no content, a file mode
# change with no content, or a deleted file with no content. If so,
# we'll skip this diff.
start_linenum = linenum
diff_git_line = lines[linenum]
linenum += 1
# Check to make sure we haven't reached the end of the diff.
if linenum >= len(lines):
return linenum, None
file_info = ParsedDiffFile(parsed_diff_change=self.parsed_diff_change)
file_info.append_data(diff_git_line)
file_info.append_data(b'\n')
file_info.binary = False
# Assume the blob / commit information is provided globally. If
# we found an index header we'll override this.
file_info.orig_file_details = self.base_commit_id
file_info.modified_file_details = self.new_commit_id
headers, linenum = self._parse_extended_headers(linenum)
# Determine the created/deleted/modified state and accompanying UNIX
# file mode.
if self._is_new_file(headers):
new_mode_header = headers[b'new file mode'][1]
file_info.append_data(new_mode_header)
file_info.orig_file_details = PRE_CREATION
file_info.new_unix_mode = self._parse_unix_mode(new_mode_header)
elif self._is_deleted_file(headers):
old_mode_header = headers[b'deleted file mode'][1]
file_info.append_data(old_mode_header)
file_info.deleted = True
file_info.old_unix_mode = self._parse_unix_mode(old_mode_header)
elif self._is_mode_change(headers):
old_mode_header = headers[b'old mode'][1]
new_mode_header = headers[b'new mode'][1]
file_info.append_data(old_mode_header)
file_info.append_data(new_mode_header)
file_info.old_unix_mode = self._parse_unix_mode(old_mode_header)
file_info.new_unix_mode = self._parse_unix_mode(new_mode_header)
# Determine whether the file has been moved or copied, and track
# that information.
if self._is_moved_file(headers):
file_info.orig_filename = headers[b'rename from'][0]
file_info.modified_filename = headers[b'rename to'][0]
file_info.moved = True
if b'similarity index' in headers:
file_info.append_data(headers[b'similarity index'][1])
file_info.append_data(headers[b'rename from'][1])
file_info.append_data(headers[b'rename to'][1])
elif self._is_copied_file(headers):
file_info.orig_filename = headers[b'copy from'][0]
file_info.modified_filename = headers[b'copy to'][0]
file_info.copied = True
if b'similarity index' in headers:
file_info.append_data(headers[b'similarity index'][1])
file_info.append_data(headers[b'copy from'][1])
file_info.append_data(headers[b'copy to'][1])
# Assume by default that the change is empty. If we find content
# later, we'll clear this.
empty_change = True
if b'index' in headers:
index_header_pair = headers[b'index']
index_range = index_header_pair[0].split()[0]
index_header = index_header_pair[1]
if b'..' in index_range:
(file_info.orig_file_details,
file_info.modified_file_details) = index_range.split(b'..')
if self.pre_creation_regexp.match(file_info.orig_file_details):
file_info.orig_file_details = PRE_CREATION
file_info.append_data(index_header)
unix_mode = self._parse_unix_mode(index_header)
if unix_mode is not None:
# This will overwrite anything set above. In theory, a Git
# diff shouldn't have multiple (conflicting) mode lines.
file_info.old_unix_mode = unix_mode
file_info.new_unix_mode = unix_mode
changes_linenum = None
# Get the changes
while linenum < len(lines):
if self._is_git_diff(linenum):
break
elif self._is_binary_patch(linenum):
file_info.binary = True
file_info.append_data(lines[linenum])
file_info.append_data(b'\n')
empty_change = False
linenum += 1
break
elif self._is_diff_fromfile_line(linenum):
orig_line = lines[linenum]
new_line = lines[linenum + 1]
orig_filename = orig_line[len(b'--- '):]
new_filename = new_line[len(b'+++ '):]
# Some diffs may incorrectly contain filenames listed as:
#
# --- filename\t
# +++ filename\t
#
# We need to strip those single trailing tabs.
if orig_filename.endswith(b'\t'):
orig_filename = orig_filename[:-1]
if new_filename.endswith(b'\t'):
new_filename = new_filename[:-1]
# Strip the Git a/ and b/ prefixes, if set in the diff.
if orig_filename.startswith(b'a/'):
orig_filename = orig_filename[2:]
if new_filename.startswith(b'b/'):
new_filename = new_filename[2:]
if orig_filename == b'/dev/null':
file_info.orig_file_details = PRE_CREATION
file_info.orig_filename = new_filename
else:
file_info.orig_filename = orig_filename
if new_filename == b'/dev/null':
file_info.modified_filename = orig_filename
else:
file_info.modified_filename = new_filename
file_info.append_data(orig_line)
file_info.append_data(b'\n')
file_info.append_data(new_line)
file_info.append_data(b'\n')
linenum += 2
changes_linenum = linenum
else:
empty_change = False
linenum = self.parse_diff_line(linenum, file_info)
# Now that we have the UNIX file mode and changed lines, we can
# determine if this is a symlink. We need to check the new and old
# UNIX modes.
mode_for_symlink = (file_info.new_unix_mode or file_info.old_unix_mode)
if (mode_for_symlink is not None
and stat.S_ISLNK(int(mode_for_symlink, 8))):
file_info.is_symlink = True
if changes_linenum is not None:
for i in range(changes_linenum, linenum):
line = lines[i]
if line.startswith(b'-'):
file_info.old_symlink_target = line[1:].strip()
elif line.startswith(b'+'):
file_info.new_symlink_target = line[1:].strip()
if not file_info.orig_filename:
# This file didn't have any --- or +++ lines. This usually means
# the file was deleted or moved without changes. We'll need to
# fall back to parsing the diff --git line, which is more
# error-prone.
assert not file_info.modified_filename
self._parse_diff_git_line(diff_git_line, file_info, linenum)
# For an empty change, we keep the file's info only if it is a new
# 0-length file, a moved file, a copied file, or a deleted 0-length
# file.
#
# TODO: In the future, we'll want to keep empty files so we can show
# metadata changes, once that functionality is available in the
# diff viewer.
if (empty_change and file_info.orig_file_details != PRE_CREATION
and not (file_info.moved or file_info.copied
or file_info.deleted)):
# We didn't find any interesting content, so leave out this
# file's info.
#
# Note that we may want to change this in the future to preserve
# data like mode changes, but that will require filtering out
# empty changes at the diff viewer level in a sane way.
file_info.discard()
file_info = None
return linenum, file_info
def _parse_git_diff(self, linenum):
# First check if it is a new file with no content or
# a file mode change with no content or
# a deleted file with no content
# then skip
start_linenum = linenum
# Now we have a diff we are going to use so get the filenames + commits
diff_git_line = self.lines[linenum]
file_info = ParsedDiffFile()
file_info.append_data(diff_git_line)
file_info.append_data(b'\n')
file_info.binary = False
linenum += 1
# Check to make sure we haven't reached the end of the diff.
if linenum >= len(self.lines):
return linenum, None
# Assume the blob / commit information is provided globally. If
# we found an index header we'll override this.
file_info.orig_file_details = self.base_commit_id
file_info.modified_file_details = self.new_commit_id
headers, linenum = self._parse_extended_headers(linenum)
for line in self.lines[start_linenum:linenum]:
m = GitDiffParser.FILE_MODE_RE.search(line)
if m:
mode = int(m.group('mode'), 8)
if stat.S_ISLNK(mode):
file_info.is_symlink = True
break
if self._is_new_file(headers):
file_info.append_data(headers[b'new file mode'][1])
file_info.orig_file_details = PRE_CREATION
elif self._is_deleted_file(headers):
file_info.append_data(headers[b'deleted file mode'][1])
file_info.deleted = True
elif self._is_mode_change(headers):
file_info.append_data(headers[b'old mode'][1])
file_info.append_data(headers[b'new mode'][1])
if self._is_moved_file(headers):
file_info.orig_filename = headers[b'rename from'][0]
file_info.modified_filename = headers[b'rename to'][0]
file_info.moved = True
if b'similarity index' in headers:
file_info.append_data(headers[b'similarity index'][1])
file_info.append_data(headers[b'rename from'][1])
file_info.append_data(headers[b'rename to'][1])
elif self._is_copied_file(headers):
file_info.orig_filename = headers[b'copy from'][0]
file_info.modified_filename = headers[b'copy to'][0]
file_info.copied = True
if b'similarity index' in headers:
file_info.append_data(headers[b'similarity index'][1])
file_info.append_data(headers[b'copy from'][1])
file_info.append_data(headers[b'copy to'][1])
# Assume by default that the change is empty. If we find content
# later, we'll clear this.
empty_change = True
if b'index' in headers:
index_range = headers[b'index'][0].split()[0]
if b'..' in index_range:
(file_info.orig_file_details,
file_info.modified_file_details) = index_range.split(b'..')
if self.pre_creation_regexp.match(file_info.orig_file_details):
file_info.orig_file_details = PRE_CREATION
file_info.append_data(headers[b'index'][1])
# Get the changes
while linenum < len(self.lines):
if self._is_git_diff(linenum):
break
elif self._is_binary_patch(linenum):
file_info.binary = True
file_info.append_data(self.lines[linenum])
file_info.append_data(b'\n')
empty_change = False
linenum += 1
break
elif self._is_diff_fromfile_line(linenum):
orig_line = self.lines[linenum]
new_line = self.lines[linenum + 1]
orig_filename = orig_line[len(b'--- '):]
new_filename = new_line[len(b'+++ '):]
# Some diffs may incorrectly contain filenames listed as:
#
# --- filename\t
# +++ filename\t
#
# We need to strip those single trailing tabs.
if orig_filename.endswith(b'\t'):
orig_filename = orig_filename[:-1]
if new_filename.endswith(b'\t'):
new_filename = new_filename[:-1]
# Strip the Git a/ and b/ prefixes, if set in the diff.
if orig_filename.startswith(b'a/'):
orig_filename = orig_filename[2:]
if new_filename.startswith(b'b/'):
new_filename = new_filename[2:]
if orig_filename == b'/dev/null':
file_info.orig_file_details = PRE_CREATION
file_info.orig_filename = new_filename
else:
file_info.orig_filename = orig_filename
if new_filename == b'/dev/null':
file_info.modified_filename = orig_filename
else:
file_info.modified_filename = new_filename
file_info.append_data(orig_line)
file_info.append_data(b'\n')
file_info.append_data(new_line)
file_info.append_data(b'\n')
linenum += 2
else:
empty_change = False
linenum = self.parse_diff_line(linenum, file_info)
if not file_info.orig_filename:
# This file didn't have any --- or +++ lines. This usually means
# the file was deleted or moved without changes. We'll need to
# fall back to parsing the diff --git line, which is more
# error-prone.
assert not file_info.modified_filename
self._parse_diff_git_line(diff_git_line, file_info, linenum)
# For an empty change, we keep the file's info only if it is a new
# 0-length file, a moved file, a copied file, or a deleted 0-length
# file.
if (empty_change and file_info.orig_file_details != PRE_CREATION
and not (file_info.moved or file_info.copied
or file_info.deleted)):
# We didn't find any interesting content, so leave out this
# file's info.
#
# Note that we may want to change this in the future to preserve
# data like mode changes, but that will require filtering out
# empty changes at the diff viewer level in a sane way.
file_info = None
return linenum, file_info
def _parse_git_diff(self, linenum):
# First check if it is a new file with no content or
# a file mode change with no content or
# a deleted file with no content
# then skip
start_linenum = linenum
# Now we have a diff we are going to use so get the filenames + commits
diff_git_line = self.lines[linenum]
file_info = ParsedDiffFile()
file_info.append_data(diff_git_line)
file_info.append_data(b'\n')
file_info.binary = False
linenum += 1
# Check to make sure we haven't reached the end of the diff.
if linenum >= len(self.lines):
return linenum, None
# Assume the blob / commit information is provided globally. If
# we found an index header we'll override this.
file_info.origInfo = self.base_commit_id
file_info.newInfo = self.new_commit_id
headers, linenum = self._parse_extended_headers(linenum)
for line in self.lines[start_linenum:linenum]:
m = GitDiffParser.FILE_MODE_RE.search(line)
if m:
mode = int(m.group('mode'), 8)
if stat.S_ISLNK(mode):
file_info.is_symlink = True
break
if self._is_new_file(headers):
file_info.append_data(headers[b'new file mode'][1])
file_info.origInfo = PRE_CREATION
elif self._is_deleted_file(headers):
file_info.append_data(headers[b'deleted file mode'][1])
file_info.deleted = True
elif self._is_mode_change(headers):
file_info.append_data(headers[b'old mode'][1])
file_info.append_data(headers[b'new mode'][1])
if self._is_moved_file(headers):
file_info.origFile = headers[b'rename from'][0]
file_info.newFile = headers[b'rename to'][0]
file_info.moved = True
if b'similarity index' in headers:
file_info.append_data(headers[b'similarity index'][1])
file_info.append_data(headers[b'rename from'][1])
file_info.append_data(headers[b'rename to'][1])
elif self._is_copied_file(headers):
file_info.origFile = headers[b'copy from'][0]
file_info.newFile = headers[b'copy to'][0]
file_info.copied = True
if b'similarity index' in headers:
file_info.append_data(headers[b'similarity index'][1])
file_info.append_data(headers[b'copy from'][1])
file_info.append_data(headers[b'copy to'][1])
# Assume by default that the change is empty. If we find content
# later, we'll clear this.
empty_change = True
if b'index' in headers:
index_range = headers[b'index'][0].split()[0]
if '..' in index_range:
file_info.origInfo, file_info.newInfo = index_range.split("..")
if self.pre_creation_regexp.match(file_info.origInfo):
file_info.origInfo = PRE_CREATION
file_info.append_data(headers[b'index'][1])
# Get the changes
while linenum < len(self.lines):
if self._is_git_diff(linenum):
break
elif self._is_binary_patch(linenum):
file_info.binary = True
file_info.append_data(self.lines[linenum])
file_info.append_data(b'\n')
empty_change = False
linenum += 1
break
elif self._is_diff_fromfile_line(linenum):
orig_line = self.lines[linenum]
new_line = self.lines[linenum + 1]
orig_filename = orig_line[len(b'--- '):]
new_filename = new_line[len(b'+++ '):]
# Some diffs may incorrectly contain filenames listed as:
#
# --- filename\t
# +++ filename\t
#
# We need to strip those single trailing tabs.
if orig_filename.endswith(b'\t'):
orig_filename = orig_filename[:-1]
if new_filename.endswith(b'\t'):
new_filename = new_filename[:-1]
# Strip the Git a/ and b/ prefixes, if set in the diff.
if orig_filename.startswith(b'a/'):
orig_filename = orig_filename[2:]
if new_filename.startswith(b'b/'):
new_filename = new_filename[2:]
if orig_filename == b'/dev/null':
file_info.origInfo = PRE_CREATION
file_info.origFile = new_filename
else:
file_info.origFile = orig_filename
if new_filename == b'/dev/null':
file_info.newFile = orig_filename
else:
file_info.newFile = new_filename
file_info.append_data(orig_line)
file_info.append_data(b'\n')
file_info.append_data(new_line)
file_info.append_data(b'\n')
linenum += 2
else:
empty_change = False
linenum = self.parse_diff_line(linenum, file_info)
if not file_info.origFile:
# This file didn't have any --- or +++ lines. This usually means
# the file was deleted or moved without changes. We'll need to
# fall back to parsing the diff --git line, which is more
# error-prone.
assert not file_info.newFile
self._parse_diff_git_line(diff_git_line, file_info, linenum)
if isinstance(file_info.origFile, six.binary_type):
file_info.origFile = file_info.origFile.decode('utf-8')
if isinstance(file_info.newFile, six.binary_type):
file_info.newFile = file_info.newFile.decode('utf-8')
# For an empty change, we keep the file's info only if it is a new
# 0-length file, a moved file, a copied file, or a deleted 0-length
# file.
if (empty_change and
file_info.origInfo != PRE_CREATION and
not (file_info.moved or file_info.copied or file_info.deleted)):
# We didn't find any interesting content, so leave out this
# file's info.
#
# Note that we may want to change this in the future to preserve
# data like mode changes, but that will require filtering out
# empty changes at the diff viewer level in a sane way.
file_info = None
return linenum, file_info