#!/usr/bin/env python3
"""Find likely x64 RIP-relative references to known string RVAs in a PE file."""

from __future__ import annotations

import argparse
import dataclasses
import struct
from pathlib import Path


@dataclasses.dataclass(frozen=True)
class Section:
    name: str
    virtual_address: int
    virtual_size: int
    raw_offset: int
    raw_size: int
    characteristics: int

    @property
    def is_code(self) -> bool:
        image_scn_cnt_code = 0x00000020
        image_scn_mem_execute = 0x20000000
        return bool(self.characteristics & image_scn_cnt_code) and bool(self.characteristics & image_scn_mem_execute)

    def contains_rva(self, rva: int) -> bool:
        size = max(self.virtual_size, self.raw_size)
        return self.virtual_address <= rva < self.virtual_address + size

    def rva_to_offset(self, rva: int) -> int:
        return self.raw_offset + (rva - self.virtual_address)


def parse_pe(data: bytes) -> tuple[int, list[Section]]:
    if data[:2] != b"MZ":
        raise ValueError("not an MZ executable")

    pe_offset = struct.unpack_from("<I", data, 0x3C)[0]
    if data[pe_offset : pe_offset + 4] != b"PE\0\0":
        raise ValueError("not a PE executable")

    coff_offset = pe_offset + 4
    section_count = struct.unpack_from("<H", data, coff_offset + 2)[0]
    optional_size = struct.unpack_from("<H", data, coff_offset + 16)[0]
    optional_offset = coff_offset + 20
    magic = struct.unpack_from("<H", data, optional_offset)[0]
    if magic != 0x20B:
        raise ValueError("expected PE32+ executable")

    image_base = struct.unpack_from("<Q", data, optional_offset + 24)[0]
    section_offset = optional_offset + optional_size

    sections: list[Section] = []
    for index in range(section_count):
        offset = section_offset + index * 40
        name = data[offset : offset + 8].split(b"\0", 1)[0].decode("ascii", "replace")
        virtual_size, virtual_address, raw_size, raw_offset = struct.unpack_from("<IIII", data, offset + 8)
        characteristics = struct.unpack_from("<I", data, offset + 36)[0]
        sections.append(Section(name, virtual_address, virtual_size, raw_offset, raw_size, characteristics))

    return image_base, sections


def section_for_rva(sections: list[Section], rva: int) -> Section | None:
    for section in sections:
        if section.contains_rva(rva):
            return section
    return None


def iter_rip_xrefs(data: bytes, section: Section, image_base: int, target_vas: set[int]) -> dict[int, list[int]]:
    start = section.raw_offset
    end = min(len(data), section.raw_offset + section.raw_size)
    hits: dict[int, list[int]] = {target_va: [] for target_va in target_vas}

    # On x64, most string references are encoded as a signed 32-bit displacement
    # relative to the address immediately after the displacement.
    for offset in range(start, max(start, end - 4)):
        disp = struct.unpack_from("<i", data, offset)[0]
        instr_end_va = image_base + section.virtual_address + (offset - section.raw_offset) + 4
        target_va = instr_end_va + disp
        if target_va in target_vas:
            hits[target_va].append(section.virtual_address + (offset - section.raw_offset))

    return hits


def main() -> int:
    parser = argparse.ArgumentParser()
    parser.add_argument("pe", type=Path)
    parser.add_argument("targets", nargs="+", help="name=rva_hex pairs")
    args = parser.parse_args()

    data = args.pe.read_bytes()
    image_base, sections = parse_pe(data)
    code_sections = [section for section in sections if section.is_code]

    targets: list[tuple[str, int, int]] = []
    target_vas: set[int] = set()
    for item in args.targets:
        name, _, rva_text = item.partition("=")
        if not name or not rva_text:
            raise ValueError(f"target must be name=rva_hex: {item}")
        target_rva = int(rva_text, 16)
        target_va = image_base + target_rva
        targets.append((name, target_rva, target_va))
        target_vas.add(target_va)

    xrefs: dict[int, list[int]] = {target_va: [] for target_va in target_vas}
    for section in code_sections:
        section_hits = iter_rip_xrefs(data, section, image_base, target_vas)
        for target_va, hits in section_hits.items():
            xrefs[target_va].extend(hits)

    print(f"image_base=0x{image_base:x}")
    for name, target_rva, target_va in targets:
        target_section = section_for_rva(sections, target_rva)
        section_name = target_section.name if target_section else "<none>"
        print(f"{name}: target_rva=0x{target_rva:x} target_section={section_name}")

        all_hits = xrefs[target_va]
        if not all_hits:
            print("  no code xrefs found")
            continue

        for hit in all_hits[:32]:
            print(f"  code_xref_disp_rva=0x{hit:x} probable_instr_rva=0x{max(0, hit - 7):x}..0x{hit:x}")
        if len(all_hits) > 32:
            print(f"  ... {len(all_hits) - 32} more")

    return 0


if __name__ == "__main__":
    raise SystemExit(main())