multigource/multigource.py

#!/usr/bin/env python3

from argparse import ArgumentParser,Namespace
import re

STMT_CREATE='''
CREATE TABLE IF NOT EXISTS gourcelog (
    time DATETIME NOT NULL,
   	user text,
	action text,
	path text
);
'''
STMT_INSERT='''
INSERT INTO gourcelog VALUES (@time,@user,@action,@path)
'''
STMT_SELECT='''
SELECT time,user,action,path FROM gourcelog ORDER BY time ASC
'''

def main():
    p = parser()
    args = p.parse_args()
    import sys
    sys.exit(args.func(args) or 0)

def parser():
    p = ArgumentParser()
    p.set_defaults(func=lambda x: p.print_help())
    sub = p.add_subparsers()
    p_render = sub.add_parser('render')
    p_render.set_defaults(func=render)
    p_render.add_argument('--gource', default='gource')
    p_join = sub.add_parser('gitjoin')
    p_join.set_defaults(func=gitjoin)
    p_join.add_argument('source')
    p_join.add_argument('output')
    p_join.add_argument('-p','--prefix')
    return p

def gitjoin(args: Namespace):
    print(args)
    import os
    import subprocess
    import sqlite3
    GIT_CMD = ["git", "log", "--all", '--pretty=format:user:%aN%n%ct', "--reverse", "--raw", "--encoding=UTF-8", "--no-renames"]

    with sqlite3.connect(args.output) as db:
        db.execute(STMT_CREATE)
        db.commit()
        for fp in gitwalk(args.source):
            p = fp[len(args.source):]
            proc = subprocess.run(GIT_CMD, cwd=fp, stdout=subprocess.PIPE)
            if not proc.returncode == 0:
                continue
            print(fp)
            try:
                gitlog = proc.stdout.decode()
            except Exception as ex:
                print(ex)
                continue
            for l in gource_format(gitlog):
                l[3] = os.path.join(p, l[3].lstrip('/'))
                db.execute(STMT_INSERT, {
                    'time': int(l[0]),
                    'user': l[1],
                    'action': l[2],
                    'path': l[3],
                })
                # print('|'.join(l))
        db.commit()
    for it in db.execute(STMT_SELECT):
        print(it)

def gource_format(inp: str):
    import subprocess
    GOURCE_CMD = ["gource", "--log-format", "git", "--output-custom-log", "-", "-"]
    with subprocess.Popen(GOURCE_CMD, stdout=subprocess.PIPE, stdin=subprocess.PIPE) as proc:
        out, err = proc.communicate(input=inp.encode())
        if proc.returncode != 0:
            return
    for l in out.decode().splitlines():
        ls = l.split('|')
        if len(ls) != 4:
            continue
        yield ls
    # sed "s, \([ACDMRTU]\)\t, \1\t$REL/," | gource  --log-format git --output-custom-log - -

def gitwalk(root: str):
    import os
    if not os.path.isdir(root):
        return
    if root.endswith('.git'):
        yield root
    else:
        ls = os.listdir(root)
        for d in ls:
            dj = os.path.join(root,d)
            for p in gitwalk(dj):
                yield p

def tmpid(key: str):
    import hashlib
    return hashlib.sha256(key).hexdigest()

def render(args: Namespace):
    print(args)

if __name__=='__main__':
    main()