def _parse_hunk_line(self, line):
"""
Given a hunk line in `git diff` output, return the line number
at the start of the hunk. A hunk is a segment of code that
contains changes.
The format of the hunk line is:
@@ -k,l +n,m @@ TEXT
where `k,l` represent the start line and length before the changes
and `n,m` represent the start line and length after the changes.
`git diff` will sometimes put a code excerpt from within the hunk
in the `TEXT` section of the line.
"""
# Split the line at the @@ terminators (start and end of the line)
components = line.split('@@')
# The first component should be an empty string, because
# the line starts with '@@'. The second component should
# be the hunk information, and any additional components
# are excerpts from the code.
if len(components) >= 2:
hunk_info = components[1]
groups = self.HUNK_LINE_RE.findall(hunk_info)
if len(groups) == 1:
try:
return int(groups[0])
except ValueError:
msg = "Could not parse '{}' as a line number".format(
groups[0])
raise GitDiffError(msg)
else:
msg = "Could not find start of hunk in line '{}'".format(line)
raise GitDiffError(msg)
else:
msg = "Could not parse hunk in line '{}'".format(line)
raise GitDiffError(msg)
def _parse_source_line(self, line):
"""
Given a source line in `git diff` output, return the path
to the source file.
"""
if "--git" in line:
regex = self.SRC_FILE_RE
elif "--cc" in line:
regex = self.MERGE_CONFLICT_RE
else:
msg = f"Do not recognize format of source in line '{line}'"
raise GitDiffError(msg)
# Parse for the source file path
groups = regex.findall(line)
if len(groups) == 1:
return groups[0]
msg = f"Could not parse source path in line '{line}'"
raise GitDiffError(msg)
def _parse_source_sections(self, diff_str):
"""
Given the output of `git diff`, return a dictionary
with keys that are source file paths.
Each value is a list of lines from the `git diff` output
related to the source file.
Raises a `GitDiffError` if `diff_str` is in an invalid format.
"""
# Create a dict to map source files to lines in the diff output
source_dict = dict()
# Keep track of the current source file
src_path = None
# Signal that we've found a hunk (after starting a source file)
found_hunk = False
# Parse the diff string into sections by source file
for line in diff_str.split('\n'):
# If the line starts with "diff --git"
# or "diff --cc" (in the case of a merge conflict)
# then it is the start of a new source file
if line.startswith('diff --git') or line.startswith('diff --cc'):
# Retrieve the name of the source file
src_path = self._parse_source_line(line)
# Create an entry for the source file, if we don't
# already have one.
if src_path not in source_dict:
source_dict[src_path] = []
# Signal that we're waiting for a hunk for this source file
found_hunk = False
# Every other line is stored in the dictionary for this source file
# once we find a hunk section
else:
# Only add lines if we're in a hunk section
# (ignore index and files changed lines)
if found_hunk or line.startswith('@@'):
# Remember that we found a hunk
found_hunk = True
if src_path is not None:
source_dict[src_path].append(line)
else:
# We tolerate other information before we have
# a source file defined, unless it's a hunk line
if line.startswith("@@"):
msg = "Hunk has no source file: '{0}'".format(line)
raise GitDiffError(msg)
return source_dict
