#!/usr/bin/env python3
"""Extract the loadable .text bytes from an as34/ld34 g.out file.

g.out is not COFF. It uses the .go/linkhdr format from binutils/link.h
(LINKMAGIC 0x3abf782e), packed with #pragma pack(2).

IMPORTANT: link.h declares the header fields that hold addresses and sizes
as `long`. On the original 32-bit Unix hosts that is 4 bytes; on x86-64 it
is 8, so the written header is 156 bytes here instead of the historical 80.
as34 and ld34 are both rebuilt with the same compiler, so they agree with
each other and with this parser. A stable on-disk format would need `long`
changed to int32_t in link.h.

Layout as written by this 64-bit build (offsets in bytes):

    off  field
      0  short swapped
      2  long  magic
     10  short flags
     12  long  time
     20  long  textstart        file position of text
     28  long  textsize         size in BYTES
     36  long  datastart
     44  long  datasize
     52  long  bsssize
     60  long  relocstart
     68  long  nrelocs
     76  long  symtabstart
     84  long  nsyms
     92  long  stabstart
    100  long  nstabs
    108  long  stringstart
    116  long  stringSize
    124  long  textLoadAddress  BIT address
    132  long  dataLoadAddress
    140  long  bssLoadAddress
    148  long  entrypoint        BIT address
"""
import hashlib
import struct
import sys

LINKMAGIC = 0x3ABF782E


def main(out_path, bin_path):
    d = open(out_path, "rb").read()
    fields = struct.unpack_from("<hqh" + "q" * 18, d, 0)
    (swapped, magic, flags, _time, textstart, textsize,
     _datastart, _datasize, _bsssize, _relocstart, _nrelocs,
     _symtabstart, _nsyms, _stabstart, _nstabs,
     _stringstart, _stringsize,
     textload, _dataload, _bssload, entry) = fields

    if magic != LINKMAGIC:
        sys.exit("extract: bad magic %#010x" % magic)
    if swapped:
        sys.exit("extract: byte-swapped header not handled")

    payload = d[textstart:textstart + textsize]
    if len(payload) != textsize:
        sys.exit("extract: truncated text (%d of %d bytes)"
                 % (len(payload), textsize))

    open(bin_path, "wb").write(payload)
    print("origin=%08X (bits) size=%d bytes entry=%08X file_offset=%#x md5=%s"
          % (textload, textsize, entry, textstart,
             hashlib.md5(payload).hexdigest()))


if __name__ == "__main__":
    if len(sys.argv) != 3:
        sys.exit("usage: extract.py hello.out payload.bin")
    main(sys.argv[1], sys.argv[2])