Git Blame Results for Linter Violations

linterblame.py

#!/usr/bin/env python

import argparse
from collections import defaultdict
import subprocess


def lint(linter, dirs):
    files = defaultdict(lambda: defaultdict(list))
    for dir in dirs:
        try:
            subprocess.check_output([linter, dir])
        except subprocess.CalledProcessError as ex:
            for line in ex.output.split('\n'):
                if not line.strip():
                    continue
                loc, desc = line.split(' ', 1)
                filename, line_number = loc.rsplit(':', 1)
                files[filename][int(line_number)].append(desc)
    return files


def blame(files):
    blames = defaultdict(list)
    for filename, line_mapping in files.items():
        commit_details = defaultdict(dict)
        output = subprocess.check_output(['git', 'blame', '-e', '-w', '--incremental', filename])
        lines = output.strip().split('\n')
        line_blame = {}
        while lines:
            line = lines.pop(0)
            sha, source_line_number, result_line_number, num_lines = line.split(' ')
            # iterate until we have the filename line
            while lines:
                line = lines.pop(0)
                key, value = line.split(' ', 1)
                if key == 'filename':
                    break
                commit_details[sha][key] = value
            line_blame[xrange(int(result_line_number), int(result_line_number) + int(num_lines))] = commit_details[sha]
        
        for line, descriptions in line_mapping.items():
            for line_range, details in line_blame.items():
                if line not in line_range:
                    continue
                for desc in descriptions:
                    blames[(details['author'], details.get('author-email', None))].append(
                        '{filename}:{line}\t{desc}'.format(filename=filename, line=line, desc=desc)
                    )
    return blames


if __name__ == '__main__':
    parser = argparse.ArgumentParser('linterblame.py')
    parser.add_argument('--linter', help='lint command to run')
    parser.add_argument('dir', nargs='+', help='where to run linterblame')
    args = parser.parse_args()
    files = lint(args.linter, args.dir)
    blames = blame(files)
    for author, violations in blames.items():
        name, email = author
        print
        print name
        if email:
            print email
        for violation in violations:
            print violation