Git diff intersection finder

diff-intersection.py

#!/usr/bin/env python3

import argparse
import sys
from pathlib import PurePath
from unidiff import PatchSet
from unidiff.constants import LINE_TYPE_CONTEXT

parser = argparse.ArgumentParser(prog="diff-intersection")

parser.add_argument("base_patch", help="The base patch to be filtered")
parser.add_argument("filter_patches", nargs="*", help="Filter the base patch down to common modifications with these patches")
parser.add_argument("--only-hunks", action='store_true', help="Don't filter lines; only filter hunks")
parser.add_argument("--ignore-files", default=[], nargs="*", help="Filter out any diffs in these file globs")

args = parser.parse_args()

with open(args.base_patch) as f:
    base_patch = PatchSet(f.read())

filter_patches = []
for path in args.filter_patches:
    with open(path) as f:
        filter_patches.append(PatchSet(f.read()))

def extract_lines(patch, output=set()):
    for file in patch:
        for hunk in file:
            current_line = None
            for line in hunk:
                if line.target_line_no is not None:
                    current_line = line.target_line_no
                if not line.is_context:
                    output.add((file.target_file or file.source_file, current_line))
    return output

filter_lines = set()
for patch in filter_patches:
    extract_lines(patch, filter_lines)

remove_files = []
for file in base_patch:
    file_ignored = False
    for ignore_file in args.ignore_files:
        if PurePath(file.target_file[2:]).match(ignore_file) or PurePath(file.source_file[2:]).match(ignore_file):
            remove_files.append(file)
            file_ignored = True
    if file_ignored:
        continue

    found_file_match = False
    remove_hunks = []
    for hunk in file:
        found_hunk_match = False
        current_line = None
        remove_lines = []
        for line in hunk:
            if line.target_line_no is not None:
                current_line = line.target_line_no
            if not line.is_context:
                matches = (file.target_file or file.source_file, current_line) in filter_lines
                if matches:
                    found_hunk_match = True
                    found_file_match = True
                elif args.only_hunks:
                    pass
                elif line.is_added:
                    line.line_type = LINE_TYPE_CONTEXT
                elif line.is_removed:
                    remove_lines.append(line)
        if not found_hunk_match:
            remove_hunks.append(hunk)
        else:
            for line in remove_lines:
                hunk.remove(line)
    if not found_file_match:
        remove_files.append(file)
    else:
        for hunk in remove_hunks:
            file.remove(hunk)

for file in remove_files:
    base_patch.remove(file)

files_changed = 0
insertions = 0
deletions = 0
for file in base_patch:
    files_changed += 1
    for hunk in file:
        for line in hunk:
            if line.is_added:
                insertions += 1
            elif line.is_removed:
                deletions += 1
sys.stderr.write(f" {files_changed} files changed, {insertions} insertions(+), {deletions} deletions(-)\n")

print(base_patch)