mechanicaldiff/bin/mechanicaldiff

#!/usr/bin/env python3

import re
import sys


def trim_blank_ends(lines: list[str]) -> list[str]:
    start = 0
    end = len(lines)
    while start < end and lines[start] == "\n":
        start += 1
    while end > start and lines[end - 1] == "\n":
        end -= 1
    return lines[start:end]


def filter_change_block(
    change_lines: list[str], search: str, replace: str
) -> list[str]:
    removed_lines = []
    added_lines = []
    phase = "minus"
    for line in change_lines:
        if line.startswith("-"):
            if phase != "minus":
                raise ValueError(
                    "Non-consecutive '-' and '+' lines in change block."
                )
            removed_lines.append(line[1:])
            continue
        if line.startswith("+"):
            phase = "plus"
            added_lines.append(line[1:])
            continue
        raise ValueError("Unexpected non-change line in change block.")

    transformed_removed = [
        re.sub(search, replace, line) for line in removed_lines
    ]
    if trim_blank_ends(transformed_removed) == trim_blank_ends(added_lines):
        return change_lines

    kept_lines = []
    max_len = max(len(removed_lines), len(added_lines))
    for idx in range(max_len):
        removed = removed_lines[idx] if idx < len(removed_lines) else None
        added = added_lines[idx] if idx < len(added_lines) else None
        if removed is None:
            break
        if re.sub(search, replace, removed) == added:
            kept_lines.append(f"-{removed}")
            kept_lines.append(f"+{added}")
        else:
            kept_lines.append(f" {removed}")

    if not kept_lines:
        return []

    return kept_lines


def filter_hunk(
    hunk_lines: list[str], search: str, replace: str
) -> list[str]:
    if not hunk_lines:
        return []

    header = hunk_lines[0]
    body = hunk_lines[1:]
    output_body: list[str] = []

    i = 0
    while i < len(body):
        line = body[i]
        if line.startswith(("+", "-")):
            block = []
            while i < len(body) and body[i].startswith(("+", "-")):
                block.append(body[i])
                i += 1
            kept_block = filter_change_block(block, search, replace)
            if kept_block:
                output_body.extend(kept_block)
            continue
        output_body.append(line)
        i += 1

    if not any(line.startswith(("+", "-")) for line in output_body):
        return []

    return [header] + output_body


def main() -> None:
    if len(sys.argv) != 3:
        raise SystemExit("Usage: mechanicaldiff <search> <replace>")

    search = sys.argv[1]
    replace = sys.argv[2]

    input_data = sys.stdin.read()
    lines = input_data.splitlines(keepends=True)
    preamble_lines = []
    sections = []
    current = None

    for line in lines:
        if line.startswith("diff --git "):
            if current is not None:
                sections.append(current)
            current = {"header": [line], "hunks": []}
            continue

        if current is None:
            preamble_lines.append(line)
            continue

        if line.startswith("@@ "):
            current["hunks"].append([line])
        else:
            if current["hunks"]:
                current["hunks"][-1].append(line)
            else:
                current["header"].append(line)

    if current is not None:
        sections.append(current)

    output_lines = []
    output_lines.extend(preamble_lines)
    for section in sections:
        kept_hunks = []
        for hunk_lines in section["hunks"]:
            filtered = filter_hunk(hunk_lines, search, replace)
            if filtered:
                kept_hunks.append(filtered)

        if not kept_hunks:
            continue

        output_lines.extend(section["header"])
        for hunk_lines in kept_hunks:
            output_lines.extend(hunk_lines)

    output_data = "".join(output_lines)
    sys.stdout.write(output_data)
    if output_data != input_data:
        raise SystemExit(1)


if __name__ == "__main__":
    main()