machfs/bin/MakeHFS

#!/usr/bin/env python3

import argparse
import os
import os.path as path
import re
from datetime import datetime
from machfs import Volume, Folder, File
import macresources

########################################################################

def hfsdat(x):
    if x.lower() == 'now':
        x = datetime.now().isoformat()

    if len(x) == 8 and all(c in '0123456789ABCDEF' for c in x.upper()):
        try:
            return int(x, base=16)
        except ValueError:
            pass

    epoch = '19040101000000' # ISO8601 with the non-numerics stripped

    # strip non-numerics and pad out using the epoch (cheeky)
    stripped = ''.join(c for c in x if c in '0123456789')
    stripped = stripped[:len(epoch)] + epoch[len(stripped):]

    tformat = '%Y%m%d%H%M%S'

    delta = datetime.strptime(stripped, tformat) - datetime.strptime(epoch, tformat)
    delta = int(delta.total_seconds())

    if not 0 <= delta <= 0xFFFFFFFF:
        print('Warning: moving %r into the legacy MacOS date range (1904-2040)' % x)

    delta = min(delta, 0xFFFFFFFF)
    delta = max(delta, 0)

    return delta

def imgsize(x):
    x = x.upper()
    x = x.replace('B', '').replace('I', '')
    if x.endswith('K'):
        factor = 1024
    elif x.endswith('M'):
        factor = 1024*1024
    elif x.endswith('G'):
        factor = 1024*1024*1024
    else:
        factor = 1
        x += 'b'
    return int(x[:-1]) * factor

def hfspathtpl(s):
    return tuple(c for c in s.split(':') if c)

args = argparse.ArgumentParser()

args.add_argument('dest', metavar='OUTPUT', nargs=1, help='Destination file')
args.add_argument('-n', '--name', default='untitled', action='store', help='volume name (default: untitled)')
args.add_argument('-i', '--dir', action='store', help='folder to copy into the image')
args.add_argument('-a', '--app', default=None, type=hfspathtpl, help='Path:To:Startup:App')
args.add_argument('-s', '--size', default='800k', type=imgsize, action='store', help='volume size (default: size of OUTPUT)')
args.add_argument('-d', '--date', default='1994', type=hfsdat, action='store', help='creation & mod date (ISO-8601 or "now")')
args.add_argument('--mpw-dates', action='store_true', help='''
    preserve the modification order of files by setting on-disk dates
    that differ by 1-minute increments, so that MPW Make can decide
    which files to rebuild
''')

args = args.parse_args()

########################################################################

vol = Volume()
vol.name = args.name
vol.crdate = vol.mddate = vol.bkdate = args.date

########################################################################

def includefilter(n):
    if n.startswith('.'): return False
    if n.endswith('.rdump'): return True
    if n.endswith('.idump'): return True
    return True

def swapsep(n):
    return n.replace(':', path.sep)

def mkbasename(n):
    base, ext = path.splitext(n)
    if ext in ('.rdump', '.idump'):
        return base
    else:
        return n

if args.dir is not None:
    tmptree = {args.dir: vol}

    for dirpath, dirnames, filenames in os.walk(args.dir):
        dirnames[:] = [swapsep(x) for x in dirnames if includefilter(x)]
        filenames[:] = [swapsep(x) for x in filenames if includefilter(x)]

        for dn in dirnames:
            newdir = Folder()
            newdir.crdate = newdir.mddate = newdir.bkdate = args.date
            tmptree[dirpath][dn] = newdir
            tmptree[path.join(dirpath, dn)] = newdir

        for fn in filenames:
            basename = mkbasename(fn)
            fullbase = path.join(dirpath, basename)
            fullpath = path.join(dirpath, fn)

            try:
                thefile = tmptree[fullbase]
            except KeyError:
                thefile = File()
                thefile.real_t = 0 # for the MPW hack
                thefile.crdate = thefile.mddate = thefile.bkdate = args.date
                thefile.contributors = []
                tmptree[fullbase] = thefile

            if fn.endswith('.idump'):
                with open(fullpath, 'rb') as f:
                    thefile.type = f.read(4)
                    thefile.creator = f.read(4)
            elif fn.endswith('rdump'):
                rez = open(fullpath, 'rb').read()
                resources = macresources.parse_rez_code(rez)
                resfork = macresources.make_file(resources, align=4)
                thefile.rsrc = resfork
            else:
                thefile.data = open(fullpath, 'rb').read()

            thefile.contributors.append(fullpath)
            if args.mpw_dates:
                thefile.real_t = max(thefile.real_t, path.getmtime(fullpath))

            tmptree[dirpath][basename] = thefile

    for pathtpl, obj in vol.iter_paths():
        try:
            if obj.type == b'TEXT':
                obj.data = obj.data.decode('utf8').replace('\r\n', '\r').replace('\n', '\r').encode('mac_roman')
        except AttributeError:
            pass

########################################################################

if args.mpw_dates:
    all_real_times = set()
    for pathtpl, obj in vol.iter_paths():
        try:
            all_real_times.add(obj.real_t)
        except AttributeError:
            pass
    ts2idx = {ts: idx for (idx, ts) in enumerate(sorted(set(all_real_times)))}

    for pathtpl, obj in vol.iter_paths():
        try:
            real_t = obj.real_t
        except AttributeError:
            pass
        else:
            fake_t = obj.crdate + 60 * ts2idx[real_t]
            obj.crdate = obj.mddate = obj.bkdate = fake_t

########################################################################

image = vol.write(args.size, startapp=args.app)
with open(args.dest[0], 'wb') as f:
    f.write(image)