TRX/docs/update_functions

#!/usr/bin/python3
import sys
import re
import typing as T
from dataclasses import dataclass
from pathlib import Path


DOCS_DIR = Path(__file__).parent
PROGRESS_TXT_FILE = DOCS_DIR / "progress.txt"


@dataclass
class IdaFunction:
    name: str
    offset: int
    size: int
    flags: str


class BaseProgressLine:
    def to_progress_line() -> str:
        raise NotImplementedError("not implemented")


@dataclass
class MyFunction(BaseProgressLine):
    name: str
    offset: int
    size: int
    flags: str

    def to_progress_line(self) -> str:
        return f"{self.name:31s} 0x{self.offset:08X}  0x{self.size:08X}  {self.flags}"


@dataclass
class MyComment(BaseProgressLine):
    text: str

    def to_progress_line(self) -> str:
        return self.text


def collect_ida_functions() -> T.Iterable[IdaFunction]:
    for line in sys.stdin:
        line = line.strip()
        if not line.startswith("#") and line:
            (
                name,
                _section,
                offset_str,
                size_str,
                _locals,
                _arguments,
                flags,
            ) = re.split(r"\s+", line, maxsplit=6)
            flags = re.sub(r"\s+", "", flags)
            if "L" not in flags:
                yield IdaFunction(
                    name=name,
                    offset=int(offset_str, 16),
                    size=int(size_str, 16),
                    flags=flags,
                )


def collect_my_functions() -> T.Iterable[BaseProgressLine]:
    for line in PROGRESS_TXT_FILE.open("r", encoding="utf-8"):
        line = line.strip()
        if line.startswith("#") or not line:
            yield MyComment(text=line)
            continue

        name, offset_str, size_str, flags = re.split(r"\s+", line)
        yield MyFunction(
            name=name,
            offset=int(offset_str, 16),
            size=int(size_str, 16),
            flags=flags,
        )


def update_functions(
    ida_functions: T.Iterable[IdaFunction],
    my_functions: T.Iterable[BaseProgressLine],
) -> int:
    results = list(my_functions)

    # update names
    ida_func_map = {func.offset: func for func in ida_functions}
    for my_function in results:
        if isinstance(my_function, MyFunction):
            ida_func = ida_func_map.get(my_function.offset)
            if ida_func and ida_func.name != my_function.name:
                my_function.name = ida_func.name

    # insert missing functions
    my_func_map = {
        func.offset: i
        for i, func in enumerate(results)
        if isinstance(func, MyFunction)
    }
    last_i = -1
    for ida_function in reversed(ida_functions):
        if ida_function.offset not in my_func_map:
            results.insert(
                last_i,
                MyFunction(
                    name=ida_function.name,
                    offset=ida_function.offset,
                    size=ida_function.size,
                    flags="-",
                ),
            )
        else:
            last_i = my_func_map[ida_function.offset]

    return results


def main() -> None:
    ida_functions = list(collect_ida_functions())
    my_functions = list(collect_my_functions())
    updated_functions = list(update_functions(ida_functions, my_functions))

    new_text = (
        "\n".join(
            function.to_progress_line() for function in updated_functions
        )
        + "\n"
    )
    if new_text != PROGRESS_TXT_FILE.read_text():
        PROGRESS_TXT_FILE.write_text(new_text)
        print("changed")


if __name__ == "__main__":
    main()