mwpersistence4git.py

git_file_persistence.py

import mwpersistence
import deltas
import mwreverts


import sys
from pathlib import Path

import git


def git_file_history(path):
	repo = git.Repo(path, search_parent_directories=True)
	git_root = repo.git.rev_parse("--show-toplevel")
	file_in_repo = str(Path(path)).replace(git_root + '/', '')

	for i, commit in enumerate(reversed(list(repo.iter_commits(paths=path)))):
		filecontents = (commit.tree / file_in_repo).data_stream.read().decode('utf-8')
		yield commit, filecontents


if __name__ == '__main__':
	path = sys.argv[1]

	revisions = list(git_file_history(path))

	state = mwpersistence.DiffState(
		deltas.SegmentMatcher(),
		revert_radius=15,
		revert_detector=mwreverts.Detector(),
	)

	for i, revision in enumerate(revisions):
		_, filecontents = revision
		tokens, _, _ = state.update(filecontents, revision=i)


	last_commit = None
	for token in tokens:
		commit, _ = revisions[token.revisions[0]]
		if last_commit != str(commit):
			print()
			print(commit.author, commit)
		print(token, end='')
		last_commit = str(commit)