freedoom/scripts/fix-map-names

#!/usr/bin/env python3
# SPDX-License-Identifier: BSD-3-Clause
#
# fix-map-names - Fix map names in WAD files
#
# Fix the map name, which is the name of the first lump. If the "-t" option is
# passed for test mode then incorrect map names are displayed, but not fixed.
#
# This script can be invoked with make target "fix-map-names" (no "-t" option)
# or make target "test-map-names" ("-t" option). Make target "test"
# ("-t" option) will run this and any other test.

# Imports

import argparse
import os
import re
import struct
import sys

# Globals

args = {}  # Command line arguments.
error_count = 0
fixes_needed = 0
freedoom_1_re = re.compile(r"^C(\d)M(\d)$")  # FD #1 maps
freedoom_dm_re = re.compile(r"^DM(\d\d)$")  # FD DM maps
header_shown = False
ignored_wads = set(["dummy.wad", "test_levels.wad"])
last_error = None
map_name_re = re.compile(r"^((E\dM\d)|(MAP\d\d))$")
output_line = "%-17s %-9s %-7s %s"

# Functions

# Handle error 'msg'. Pass None to reset 'last_error'.
def error(msg):
    global error_count
    global last_error

    last_error = msg
    if msg:
        error_count += 1


# Given WAD path 'wad' return the expected map name as a function of the
# filename.
def get_expected_map_name(wad):
    # Strip of the directory, upper case, remove ".wad".
    name = os.path.basename(wad).upper()
    if name.endswith(".WAD"):
        name = name[:-4]

    # Convert from Freedoom name to Doom names.
    name = freedoom_1_re.sub(r"E\1M\2", name)
    name = freedoom_dm_re.sub(r"MAP\1", name)

    if map_name_re.match(name):
        return name
    else:
        return None


# Parse the command line arguments and store the result in 'args'.
def parse_args():
    global args

    parser = argparse.ArgumentParser(
        description="Fix map names in WAD files.",
        formatter_class=argparse.ArgumentDefaultsHelpFormatter,
    )

    # The following is sorted by long argument.

    parser.add_argument(
        "-f",
        "--force",
        action="store_true",
        help="Force. Fix map name regardless of the existing map name.",
    )
    parser.add_argument(
        "-q", "--quiet", action="store_true", help="Quiet (minimum output)."
    )
    parser.add_argument(
        "-r",
        "--recursive",
        action="store_true",
        help="Recurse into directories.",
    )
    parser.add_argument(
        "-t",
        "--test",
        action="store_true",
        help="Test mode. Don't make any changes.",
    )
    parser.add_argument(
        "paths",
        metavar="PATH",
        nargs="+",
        help="WAD paths, files and directories.",
    )

    args = parser.parse_args()

    return args


# Process path 'path' which is at depth 'depth'. If 'depth' is 0 then this is
# a top level path passed in on the command line.
def process_path(path, depth):
    if os.path.isdir(path):
        # Directory. If not recursive then only consider this directory if it
        # was specified explicitly.
        if args.recursive or not depth:
            path_list = os.listdir(path)
            path_list.sort()
            for base in path_list:
                process_path(path + "/" + base, depth + 1)
    else:
        # File. Only process WAD files that were specified explicitly
        # (depth 0), or that have the expected suffix.
        if (not depth) or path.lower().endswith(".wad"):
            process_wad(path)


# Process the paths passed in on the command line.
def process_paths():
    for path in args.paths:
        process_path(path, 0)


# Process WAD path 'wad'.
def process_wad(wad):
    global header_shown
    global last_error
    global fixes_needed

    if os.path.basename(wad).lower() in ignored_wads:
        # A known WAD that should not be processed.
        return

    try:
        # Reset everything.
        error(None)
        lump_name = None
        fix_needed = False

        expected_name = get_expected_map_name(wad)
        if not expected_name:
            raise Exception("Unable to get the expected name")
        with open(wad, "rb" if args.test else "r+b") as fhand:
            magic = fhand.read(4)
            if not isinstance(magic, str):
                # magic is bytes in Python 3.
                magic = magic.decode("UTF-8")
            if not magic == "PWAD":
                raise Exception("Not a PWAD. magic=" + magic)
            # Directory at offset 0x8 in the header.
            fhand.seek(0x08)
            directory_offset, = struct.unpack("<I", fhand.read(4))
            fhand.seek(directory_offset)
            # The first lump in the directory, which should be the 0 byte map
            # name one.
            lump_data_offset, lump_size, lump_name = struct.unpack(
                "<II8s", fhand.read(16)
            )
            if not isinstance(lump_name, str):
                # lump_name is bytes in Python 3.
                lump_name = lump_name.decode("UTF-8")
            # Get rid of the null suffix.
            lump_name = lump_name.partition("\0")[0]
            if lump_size:
                # The first lump should be 0 bytes.
                error(
                    "First lump size non-zero with "
                    + str(lump_size)
                    + " bytes"
                )
            elif not (args.force or map_name_re.match(lump_name)):
                # A sanity check to make sure we read the right part.
                error("Actual name unexpected")
            elif expected_name != lump_name:
                # The name is not what we thought it should be.
                fix_needed = True
                fixes_needed += 1
            if fix_needed and not args.test:
                # Seek to the lump name and the overwrite the lump name with
                # the expected name.
                fhand.seek(directory_offset + 8)
                fhand.write(struct.pack("8s", expected_name.encode("UTF-8")))
    except IOError as err:
        # Probably the WAD file couldn't be open for read (test) or read and
        # write (default).
        error(
            "Unable to open for read"
            + ("" if args.test else " and write")
            + ": "
            + str(err)
        )
    except struct.error as err:
        # This is probably the reason since seek silently succeeds even when
        # the location is not possible, but then unpack fails due to the short
        # read.
        error("File too small: " + str(err))
    except Exception as err:
        # This was probably explicitly thrown by this script.
        error(str(err))

    if (last_error or fix_needed) and not args.quiet:
        # Map None to "".
        expected_name, lump_name, last_error = [
            x if x else "" for x in (expected_name, lump_name, last_error)
        ]
        if not header_shown:
            print(output_line % ("WAD", "Expected", "Actual", "Error"))
            print(output_line % ("---", "--------", "------", "-----"))
            header_shown = True
        print(output_line % (wad, expected_name, lump_name, last_error))


# Summarize what happened, and then exit with the appropriate exit code.
def summarize():
    if not args.quiet:
        if fixes_needed:
            print(
                "\n%s %d WADs with the incorrect map name."
                % ("Found" if args.test else "Fixed", fixes_needed)
            )
        else:
            print("\nAll WADs had the correct map name.")
        if error_count:
            print("There were %d errors." % error_count)
    sys.exit(1 if (error_count or (args.test and fixes_needed)) else 0)


# Main

parse_args()
process_paths()
summarize()