wirevpn/scripts/patch-shim.py

#!/usr/bin/env python3
"""Minimal `patch` compatible shim (enough for the libwg-go Makefile).

Supports: patch [-pN] [-f] [-N] [-r-] [-d DIR] [-s]  (diff read from stdin)
Handles unified diffs with multiple files/hunks (git-style or plain).
"""
import sys
import os
import re


def parse_args(argv):
    strip = 1
    dir_ = "."
    args = list(argv)
    i = 0
    while i < len(args):
        a = args[i]
        if a == "-p":
            strip = int(args[i + 1]); i += 2; continue
        elif a.startswith("-p") and a[2:].isdigit():
            strip = int(a[2:]); i += 1; continue
        elif a == "-d":
            dir_ = args[i + 1]; i += 2; continue
        elif a in ("-f", "-N", "-s", "-t", "-i"):
            i += 1; continue
        elif a == "-r-":
            i += 1; continue
        elif a.startswith("-"):
            i += 1; continue
        else:
            i += 1
    return strip, dir_


def strip_path(p, strip):
    parts = p.split("/")
    if strip and len(parts) > strip:
        parts = parts[strip:]
    return "/".join(parts)


def apply_hunks(lines, hunks):
    """Apply parsed hunks (list of (old_start, old_lines, new_lines)) to lines, in order."""
    out = list(lines)
    offset = 0
    for old_start, old_count, old_content, new_content in hunks:
        idx = old_start - 1 + offset  # 0-based
        # find match allowing drift
        best = None
        for d in range(0, 200):
            for sign in (1, -1):
                cand = idx + d * sign
                if cand < 0 or cand + old_count > len(out):
                    continue
                if out[cand:cand + old_count] == old_content:
                    best = cand
                    break
            if best is not None:
                break
        if best is None:
            # try whitespace-insensitive
            def norm(l):
                return l.rstrip("\n")
            for d in range(0, 400):
                for sign in (1, -1):
                    cand = idx + d * sign
                    if cand < 0 or cand + old_count > len(out):
                        continue
                    if [norm(x) for x in out[cand:cand + old_count]] == [norm(x) for x in old_content]:
                        best = cand
                        break
                if best is not None:
                    break
        if best is None:
            raise SystemExit(f"patch: hunk at line {old_start} failed to match")
        out[best:best + old_count] = new_content
        offset += len(new_content) - old_count
    return out


def process_file(text, fname, strip, dir_):
    # text: the file's section (after "diff --git" or first "---"), up to next file
    lines = text.splitlines()
    # find hunks
    hunks = []
    old_content_all = []
    # We'll re-parse the raw section for hunk headers
    i = 0
    in_hunk = False
    old_block = []
    new_block = []
    old_start = 0
    while i < len(lines):
        ln = lines[i]
        m = re.match(r"^@@ -(\d+)(?:,(\d+))? \+(\d+)(?:,(\d+))? @@", ln)
        if m:
            if in_hunk and old_block:
                old_count = len(old_block)
                new_count = len(new_block)
                hunks.append((old_start, old_count, old_block, new_block))
            old_start = int(m.group(1))
            old_count_expect = int(m.group(2) or 1)
            new_count_expect = int(m.group(4) or 1)
            old_block = []
            new_block = []
            in_hunk = True
            i += 1
            while i < len(lines) and len(old_block) < old_count_expect and len(new_block) < new_count_expect:
                l = lines[i]
                if l.startswith("+"):
                    new_block.append(l[1:] + "\n")
                elif l.startswith("-"):
                    old_block.append(l[1:] + "\n")
                elif l.startswith(" "):
                    old_block.append(l[1:] + "\n")
                    new_block.append(l[1:] + "\n")
                elif l.startswith("\\"):
                    pass
                else:
                    break
                i += 1
            continue
        i += 1
    if in_hunk and old_block:
        old_count = len(old_block)
        new_count = len(new_block)
        hunks.append((old_start, old_count, old_block, new_block))
    target = os.path.join(dir_, strip_path(fname, strip))
    if not os.path.exists(target):
        raise SystemExit(f"patch: target {target} not found")
    with open(target, encoding="utf-8") as f:
        content = f.read()
    out = apply_hunks(content.splitlines(True), hunks)
    with open(target, "w", encoding="utf-8") as f:
        f.writelines(out)
    return target


def main():
    strip, dir_ = parse_args(sys.argv[1:])
    data = sys.stdin.read()
    # split into per-file sections
    # A new file starts at a line "diff --git " or, for plain unified diffs, at "--- "
    sections = []
    cur = []
    for ln in data.splitlines(keepends=True):
        if ln.startswith("diff --git "):
            if cur:
                sections.append("".join(cur))
            cur = [ln]
        elif ln.startswith("--- ") and not any(c.startswith("diff --git") for c in cur):
            # plain unified diff, new file
            if cur:
                sections.append("".join(cur))
            cur = [ln]
        else:
            cur.append(ln)
    if cur:
        sections.append("".join(cur))

    for sec in sections:
        slines = sec.splitlines()
        fname = None
        for l in slines:
            if l.startswith("diff --git "):
                m = re.match(r"^diff --git (\S+) b/(\S+)", l)
                if m:
                    fname = m.group(1)  # keep the a/ prefix; strip_path removes it
                break
        if fname is None:
            for l in slines:
                if l.startswith("--- "):
                    p = l[4:].strip()
                    # drop timestamp (last whitespace-separated token if it starts with a date)
                    parts = p.split("\t")
                    p = parts[0].strip()
                    if p == "/dev/null":
                        continue
                    fname = p
                    break
        if fname is None:
            continue
        # section body starts after the header; pass whole section, parser skips headers
        process_file(sec, fname, strip, dir_)


if __name__ == "__main__":
    main()