Quick-and-dirty script to move lines from one source file to a new source file, preserving git authorship along the way. The new file will be made up of a number of commits, each one attributed to the original author and at the original date and time.

git-extract.py

import argparse
from datetime import (datetime, timedelta)
import os
import re
import subprocess


class Commit(object):
  def __init__(self, commit_id, fields):
    self.commit_id = commit_id
    self.author = fields['author']
    self.author_email = fields['author-mail'][1:-1]
    self.timezone = fields['author-tz']
    self.timestamp = datetime.utcfromtimestamp(int(fields['author-time'])) + timedelta(hours=int(self.timezone) / 100)
    self.line_numbers = []

  def __repr__(self):
    return str(self.__dict__)

  def __eq__(self, other):
    return self.commit_id == other.commit_id

  def __ne__(self, other):
    return self.commit_id != other.commit_id

  def __hash__(self):
    return hash(self.commit_id)

parser = argparse.ArgumentParser(
  description='Extract git-tracked source code into a new file, '\
    'preserving authors along the way.'
)

parser.add_argument(
  'file',
  type=str,
  help='The file to extract lines from, with a line range '\
    'specified in Github format, eg. /foo/file.txt#L2-L14'
)

parser.add_argument(
  'outfile',
  type=str,
  help='The new file to create'
)

def git(cmd):
  result = subprocess.run(['git', *cmd], stdout=subprocess.PIPE)
  return result.stdout.decode('utf-8')

args = parser.parse_args()
file, start, end = [part for part in re.split(r'([^#]+)#L(\d+)-L(\d+)', args.file) if part]
start = int(start)
end = int(end)

with open(file, 'r') as f:
  lines = f.read().split("\n")

blame = git(['blame', file, '--line-porcelain', '-L', f"{start},{end + 1}"])
commit_data = re.split(r'([a-zA-Z0-9]{40}) \d+ \d+ ?\d?', blame)[1:-1]
commits = {}

for i in range(0, len(commit_data), 2):
  if i + 1 >= len(commit_data):
    break

  commit_id = commit_data[i]
  fields = {}

  for field_line in commit_data[i + 1].split("\n")[0:-1]:
    field_pair = field_line.split(' ', maxsplit=1)

    if len(field_pair) == 2:
      fields[field_pair[0]] = field_pair[1]

  if not commit_id in commits:
    commits[commit_id] = Commit(commit_id, fields)

  commits[commit_id].line_numbers.append(i // 2)

commits = sorted(commits.values(), key=lambda c: c.timestamp)
cur_lines = [''] * (len(commit_data) // 2)

moving_author_name = git(['config', '--get', 'user.name']).strip()
moving_author_email = git(['config', '--get', 'user.email']).strip()
moving_author_byline = f"{moving_author_name} <{moving_author_email}>"
now = datetime.now().ctime()

for commit in commits:
  for line_number in commit.line_numbers:
    cur_lines[line_number] = lines[start + line_number - 1]

  with open(args.outfile, 'w+') as f:
    f.write("\n".join(cur_lines))

  git(['add', args.outfile])

  original_author_byline = f'{commit.author} <{commit.author_email}>'
  original_timestamp = f'{commit.timestamp} {commit.timezone}'
  original_commit_msg = git(['log', '--format=%B', '-n 1', commit.commit_id])
  new_commit_msg = f'***NOTE: These changes were originally committed in {commit.commit_id} '\
    f'and were moved from {file} by {moving_author_byline} on {now}\n\n{original_commit_msg}'

  print(f"Committing replacement for {commit.commit_id}")
  git(['-c', 'core.hooksPath=/dev/null', 'commit', '--no-verify', '--author', original_author_byline, '--date', original_timestamp, '-m', new_commit_msg])