mkm3u: Overhaul video filename parsing again.

[epls] / mkm3u

#! /usr/bin/python3
### -*- mode: python; coding: utf-8 -*-

from contextlib import contextmanager
import optparse as OP
import os as OS
import re as RX
import sys as SYS

class ExpectedError (Exception): pass

@contextmanager
def location(loc):
  global LOC
  old, LOC = LOC, loc
  yield loc
  LOC = old

def filter(value, func = None, dflt = None):
  if value is None: return dflt
  elif func is None: return value
  else: return func(value)

def check(cond, msg):
  if not cond: raise ExpectedError(msg)

def lookup(dict, key, msg):
  try: return dict[key]
  except KeyError: raise ExpectedError(msg)

def forget(dict, key):
  try: del dict[key]
  except KeyError: pass

def getint(s):
  if not s.isdigit(): raise ExpectedError("bad integer `%s'" % s)
  return int(s)

class Words (object):
  def __init__(me, s):
    me._s = s
    me._i, me._n = 0, len(s)
  def _wordstart(me):
    s, i, n = me._s, me._i, me._n
    while i < n:
      if not s[i].isspace(): return i
      i += 1
    return -1
  def nextword(me):
    s, n = me._s, me._n
    begin = i = me._wordstart()
    if begin < 0: return None
    while i < n and not s[i].isspace(): i += 1
    me._i = i
    return s[begin:i]
  def rest(me):
    s, n = me._s, me._n
    begin = me._wordstart()
    if begin < 0: return None
    else: return s[begin:].rstrip()

URL_SAFE_P = 256*[False]
for ch in \
    b"ABCDEFGHIJKLMNOPQRSTUVWXYZ" \
    b"abcdefghijklmnopqrstuvwxyz" \
    b"0123456789" b"!$%-.,/":
  URL_SAFE_P[ch] = True
def urlencode(s):
  return "".join((URL_SAFE_P[ch] and chr(ch) or "%%%02x" % ch
                  for ch in s.encode("UTF-8")))

PROG = OS.path.basename(SYS.argv[0])

class BaseLocation (object):
  def report(me, exc):
    SYS.stderr.write("%s: %s%s\n" % (PROG, me._loc(), exc))

class DummyLocation (BaseLocation):
  def _loc(me): return ""

class FileLocation (BaseLocation):
  def __init__(me, fn, lno = 1): me._fn, me._lno = fn, lno
  def _loc(me): return "%s:%d: " % (me._fn, me._lno)
  def stepline(me): me._lno += 1

LOC = DummyLocation()

class Source (object):
  PREFIX = ""
  TITLEP = CHAPTERP = False
  def __init__(me, fn):
    me.fn = fn
    me.neps = None
    me.used_titles = dict()
    me.used_chapters = set()
    me.nuses = 0
  def url(me, title = None, start_chapter = None, end_chapter = None):
    if title == "-":
      if me.TITLEP: raise ExpectedError("missing title number")
      if start_chapter is not None or end_chapter is not None:
        raise ExpectedError("can't specify chapter without title")
      suffix = ""
    elif not me.TITLEP:
      raise ExpectedError("can't specify title with `%s'" % me.fn)
    elif start_chapter is None:
      if end_chapter is not None:
        raise ExpectedError("can't specify end chapter without start chapter")
      suffix = "#%d" % title
    elif not me.CHAPTERP:
      raise ExpectedError("can't specify chapter with `%s'" % me.fn)
    elif end_chapter is None:
      suffix = "#%d:%d" % (title, start_chapter)
    else:
      suffix = "#%d:%d-%d:%d" % (title, start_chapter, title, end_chapter - 1)
    if end_chapter is not None:
      keys = [(title, ch) for ch in range(start_chapter, end_chapter)]
      set = me.used_chapters
    else:
      keys, set = [title], me.used_titles
    for k in keys:
      if k in set:
        if title == "-":
          raise ExpectedError("`%s' already used" % me.fn)
        elif end_chapter is None:
          raise ExpectedError("`%s' title %d already used" % (me.fn, title))
        else:
          raise ExpectedError("`%s' title %d chapter %d already used" %
                              (me.fn, title, k[1]))
    if end_chapter is not None:
      for ch in range(start_chapter, end_chapter):
        me.used_chapters.add((title, ch))
    return me.PREFIX + ROOT + urlencode(me.fn) + suffix

class VideoDisc (Source):
  PREFIX = "dvd://"
  TITLEP = CHAPTERP = True

  def __init__(me, fn, *args, **kw):
    super().__init__(fn, *args, **kw)
    me.neps = 0

class VideoSeason (object):
  def __init__(me, i, title):
    me.i = i
    me.title = title
    me.episodes = {}
  def set_episode_disc(me, i, disc):
    if i in me.episodes:
      raise ExpectedError("season %d episode %d already taken" % (me.i, i))
    me.episodes[i] = disc; disc.neps += 1

def match_group(m, *groups, dflt = None, mustp = False):
  for g in groups:
    try: s = m.group(g)
    except IndexError: continue
    if s is not None: return s
  if mustp: raise ValueError("no match found")
  else: return dflt

class VideoDir (object):

  _R_ISO_PRE = list(map(lambda pat: RX.compile("^" + pat + r"\.iso$", RX.X),
    [r""" S? (?P<si> \d+) [A-Z]? \. \ (?P<stitle> .*) —
            (?: D \d+ \. \ )?
          (?P<epex> .*) """,
     r""" S (?P<si> \d+) (?: D \d+)? \. \ (?P<epex> .*) """,
     r""" S (?P<si> \d+) \. \ (?P<epex> E \d+ .*) """,
     r""" S (?P<si> \d+) \. \ (?P<epex> E \d+ .*) """,
     r""" S (?P<si> \d+) (?P<epex> E \d+) \. \ .* """,
     r""" \d+ \. \ (?P<epex> [ES] \d+ .*) """,
     r""" (?P<epnum> \d+ ) \. \ .* """]))

  _R_ISO_EP = RX.compile(r""" ^
        (?: S (?P<si> \d+) \ )?
        E (?P<ei> \d+) (?: – (?P<ej> \d+))? $
  """, RX.X)

  def __init__(me, dir):
    me.dir = dir
    fns = OS.listdir(OS.path.join(ROOT, dir))
    fns.sort()
    season = None
    seasons = {}
    for fn in fns:
      path = OS.path.join(dir, fn)
      if not fn.endswith(".iso"): continue
      #print(";; `%s'" % path, file = SYS.stderr)
      for r in me._R_ISO_PRE:
        m = r.match(fn)
        if m: break
      else:
        #print(";;\tignored (regex mismatch)", file = SYS.stderr)
        continue

      si = filter(match_group(m, "si"), int)
      stitle = match_group(m, "stitle")

      check(si is not None or stitle is None,
            "explicit season title without number in `%s'" % fn)
      if si is not None:
        if season is None or si != season.i:
          check(season is None or si == season.i + 1,
                "season %d /= %d" %
                  (si, season is None and -1 or season.i + 1))
          check(si not in seasons, "season %d already seen" % si)
          seasons[si] = season = VideoSeason(si, stitle)
        else:
          check(stitle == season.title,
                "season title `%s' /= `%s'" % (stitle, season.title))

      disc = VideoDisc(path)
      ts = season
      any, bad = False, False
      epnum = match_group(m, "epnum")
      if epnum is not None: eplist = ["E" + epnum]
      else: eplist = match_group(m, "epex", mustp = True).split(", ")
      for eprange in eplist:
        mm = me._R_ISO_EP.match(eprange)
        if mm is None:
          #print(";;\t`%s'?" % eprange, file = SYS.stderr)
          bad = True; continue
        if not any: any = True
        i = filter(mm.group("si"), int)
        if i is not None:
          try: ts = seasons[i]
          except KeyError: ts = seasons[i] = VideoSeason(i, None)
        if ts is None:
          ts = season = seasons[1] = VideoSeason(1, None)
        start = filter(mm.group("ei"), int)
        end = filter(mm.group("ej"), int, start)
        for k in range(start, end + 1):
          ts.set_episode_disc(k, disc)
          #print(";;\tepisode %d.%d" % (ts.i, k), file = SYS.stderr)
      if not any:
        #print(";;\tignored", file = SYS.stderr)
        pass
      elif bad:
        raise ExpectedError("bad ep list in `%s'", fn)
    me.seasons = seasons

class AudioDisc (Source):
  PREFIX = "file://"
  TITLEP = CHAPTERP = False

class AudioEpisode (Source):
  PREFIX = "file://"
  TITLEP = CHAPTERP = False
  def __init__(me, fn, i, *args, **kw):
    super().__init__(fn, *args, **kw)
    me.i = i

class AudioDir (object):

  _R_FLAC = RX.compile(r""" ^
          E (\d+)
          (?: \. \ (.*))?
          \. flac $
  """, RX.X)

  def __init__(me, dir):
    me.dir = dir
    fns = OS.listdir(OS.path.join(ROOT, dir))
    fns.sort()
    episodes = {}
    last_i = 0
    for fn in fns:
      path = OS.path.join(dir, fn)
      if not fn.endswith(".flac"): continue
      m = me._R_FLAC.match(fn)
      if not m: continue
      i = filter(m.group(1), int)
      etitle = m.group(2)
      check(i == last_i + 1, "episode %d /= %d" % (i, last_i + 1))
      episodes[i] = AudioEpisode(path, i)
      last_i = i
    me.episodes = episodes

class Chapter (object):
  def __init__(me, episode, title, i):
    me.title, me.i = title, i
    me.url = episode.source.url(episode.tno, i, i + 1)

class Episode (object):
  def __init__(me, season, i, neps, title, src, tno = None,
               startch = None, endch = None):
    me.season = season
    me.i, me.neps, me.title = i, neps, title
    me.chapters = []
    me.source, me.tno = src, tno
    me.url = src.url(tno, startch, endch)
  def add_chapter(me, title, j):
    ch = Chapter(me, title, j)
    me.chapters.append(ch)
    return ch
  def label(me):
    return me.season._eplabel(me.i, me.neps, me.title)

class BaseSeason (object):
  def __init__(me, series, implicitp = False):
    me.series = series
    me.episodes = []
    me.implicitp = implicitp
    me.ep_i, episodes = 1, []
  def add_episode(me, j, neps, title, src, tno, startch, endch):
    ep = Episode(me, j, neps, title, src, tno, startch, endch)
    me.episodes.append(ep)
    src.nuses += neps; me.ep_i += neps
    return ep
  def _epnames(me, i, neps):
    playlist = me.series.playlist
    if neps == 1: return playlist.epname, "%d" % i
    elif neps == 2: return playlist.epnames, "%d, %d" % (i, i + 1)
    else: return playlist.epnames, "%d–%d" % (i, i + neps - 1)

class Season (BaseSeason):
  def __init__(me, series, title, i, *args, **kw):
    super().__init__(series, *args, **kw)
    me.title, me.i = title, i
  def _eplabel(me, i, neps, title):
    epname, epn = me._epnames(i, neps)
    if title is None:
      if me.implicitp: label = "%s %s" % (epname, epn)
      elif me.title is None: label = "%s %d.%s" % (epname, me.i, epn)
      else: label = "%s—%s %s" % (me.title, epname, epn)
    else:
      if me.implicitp: label = "%s. %s" % (epn, title)
      elif me.title is None: label = "%d.%s. %s" % (me.i, epn, title)
      else: label = "%s—%s. %s" % (me.title, epn, title)
    return label

class MovieSeason (BaseSeason):
  def __init__(me, series, title, *args, **kw):
    super().__init__(series, *args, **kw)
    me.title = title
    me.i = None
  def add_episode(me, j, neps, title, src, tno, startch, endch):
    if me.title is None and title is None:
      raise ExpectedError("movie or movie season must have a title")
    return super().add_episode(j, neps, title, src, tno, startch, endch)
  def _eplabel(me, i, neps, title):
    if me.title is None:
      label = title
    elif title is None:
      epname, epn = me._epnames(i, neps)
      label = "%s—%s %s" % (me.title, epname, epn)
    else:
      label = "%s—%s" % (me.title, title)
    return label

class Series (object):
  def __init__(me, playlist, name, title = None, wantedp = True):
    me.playlist = playlist
    me.name, me.title = name, title
    me.cur_season = None
    me.wantedp = wantedp
  def _add_season(me, season):
    me.cur_season = season
  def add_season(me, title, i, implicitp = False):
    me._add_season(Season(me, title, i, implicitp))
  def add_movies(me, title = None):
    me._add_season(MovieSeason(me, title))
  def ensure_season(me):
    if me.cur_season is None: me.add_season(None, 1, implicitp = True)
    return me.cur_season
  def end_season(me):
    me.cur_season = None

class Playlist (object):
  def __init__(me):
    me.seasons = []
    me.episodes = []
    me.epname, me.epnames = "Episode", "Episodes"
    me.nseries = 0
  def add_episode(me, episode):
    me.episodes.append(episode)
  def done_season(me):
    if me.episodes:
      me.seasons.append(me.episodes)
      me.episodes = []
  def write(me, f):
    f.write("#EXTM3U\n")
    for season in me.seasons:
      f.write("\n")
      for ep in season:
        label = ep.label()
        if me.nseries > 1: label = ep.season.series.title + " " + label
        if not ep.chapters:
          f.write("#EXTINF:0,,%s\n%s\n" % (label, ep.url))
        else:
          for ch in ep.chapters:
            f.write("#EXTINF:0,,%s: %s\n%s\n" %
                    (label, ch.title, ch.url))

MODE_UNSET = 0
MODE_SINGLE = 1
MODE_MULTI = 2

class EpisodeListParser (object):

  def __init__(me, series_wanted = None, chapters_wanted_p = False):
    me._pl = Playlist()
    me._cur_episode = me._cur_chapter = None
    me._series = {}; me._vdirs = {}; me._audirs = {}; me._isos = {}
    me._series_wanted = series_wanted
    me._chaptersp = chapters_wanted_p
    if series_wanted is None: me._mode = MODE_UNSET
    else: me._mode = MODE_MULTI

  def _bad_keyval(me, cmd, k, v):
    raise ExpectedError("invalid `!%s' option `%s'" %
                          (cmd, v if k is None else k))

  def _keyvals(me, opts):
    if opts is not None:
      for kv in opts.split(","):
        try: sep = kv.index("=")
        except ValueError: yield None, kv
        else: yield kv[:sep], kv[sep + 1:]

  def _set_mode(me, mode):
    if me._mode == MODE_UNSET:
      me._mode = mode
    elif me._mode != mode:
      raise ExpectedError("inconsistent single-/multi-series usage")

  def _get_series(me, name):
    if name is None:
      me._set_mode(MODE_SINGLE)
      try: series = me._series[None]
      except KeyError:
        series = me._series[None] = Series(me._pl, None)
        me._pl.nseries += 1
    else:
      me._set_mode(MODE_MULTI)
      series = lookup(me._series, name, "unknown series `%s'" % name)
    return series

  def _opts_series(me, cmd, opts):
    name = None
    for k, v in me._keyvals(opts):
      if k is None: name = v
      else: me._bad_keyval(cmd, k, v)
    return me._get_series(name)

  def _auto_epsrc(me, series):
    dir = lookup(me._vdirs, series.name, "no active video directory")
    season = series.ensure_season()
    check(season.i is not None, "must use explicit iso for movie seasons")
    vseason = lookup(dir.seasons, season.i,
                     "season %d not found in video dir `%s'" %
                       (season.i, dir.dir))
    src = lookup(vseason.episodes, season.ep_i,
                 "episode %d.%d not found in video dir `%s'" %
                   (season.i, season.ep_i, dir.dir))
    return src

  def _process_cmd(me, ww):

    cmd = ww.nextword(); check(cmd is not None, "missing command")
    try: sep = cmd.index(":")
    except ValueError: opts = None
    else: cmd, opts = cmd[:sep], cmd[sep + 1:]

    if cmd == "series":
      name = None
      for k, v in me._keyvals(opts):
        if k is None: name = v
        else: me._bad_keyval(cmd, k, v)
      check(name is not None, "missing series name")
      check(name not in me._series, "series `%s' already defined" % name)
      title = ww.rest(); check(title is not None, "missing title")
      me._set_mode(MODE_MULTI)
      me._series[name] = series = Series(me._pl, name, title,
                                         me._series_wanted is None or
                                           name in me._series_wanted)
      if series.wantedp: me._pl.nseries += 1

    elif cmd == "season":
      series = me._opts_series(cmd, opts)
      w = ww.nextword();
      check(w is not None, "missing season number")
      if w == "-":
        if not series.wantedp: return
        series.add_movies(ww.rest())
      else:
        title = ww.rest(); i = getint(w)
        if not series.wantedp: return
        series.add_season(ww.rest(), getint(w), implicitp = False)
      me._cur_episode = me._cur_chapter = None
      me._pl.done_season()

    elif cmd == "epname":
      for k, v in me._keyvals(opts): me._bad_keyval("epname", k, v)
      name = ww.rest(); check(name is not None, "missing episode name")
      try: sep = name.index(",")
      except ValueError: names = name + "s"
      else: name, names = name[:sep], name[sep + 1:]
      me._pl.epname, me._pl.epnames = name, names

    elif cmd == "epno":
      series = me._opts_series(cmd, opts)
      w = ww.rest(); check(w is not None, "missing episode number")
      epi = getint(w)
      if not series.wantedp: return
      series.ensure_season().ep_i = epi

    elif cmd == "iso":
      series = me._opts_series(cmd, opts)
      fn = ww.rest(); check(fn is not None, "missing filename")
      if not series.wantedp: return
      if fn == "-": forget(me._isos, series.name)
      else:
        check(OS.path.exists(OS.path.join(ROOT, fn)),
              "iso file `%s' not found" % fn)
        me._isos[series.name] = VideoDisc(fn)

    elif cmd == "vdir":
      series = me._opts_series(cmd, opts)
      dir = ww.rest(); check(dir is not None, "missing directory")
      if not series.wantedp: return
      if dir == "-": forget(me._vdirs, series.name)
      else: me._vdirs[series.name] = VideoDir(dir)

    elif cmd == "adir":
      series = me._opts_series(cmd, opts)
      dir = ww.rest(); check(dir is not None, "missing directory")
      if not series.wantedp: return
      if dir == "-": forget(me._audirs, series.name)
      else: me._audirs[series.name] = AudioDir(dir)

    elif cmd == "displaced":
      series = me._opts_series(cmd, opts)
      w = ww.rest(); check(w is not None, "missing count"); n = getint(w)
      src = me._auto_epsrc(series)
      src.nuses += n
    else:
      raise ExpectedError("unknown command `%s'" % cmd)

  def _process_episode(me, ww):

    opts = ww.nextword(); check(opts is not None, "missing title/options")
    ti = None; sname = None; neps = 1; epi = None; loch = hich = None
    for k, v in me._keyvals(opts):
      if k is None:
        if v.isdigit(): ti = int(v)
        elif v == "-": ti = "-"
        else: sname = v
      elif k == "s": sname = v
      elif k == "n": neps = getint(v)
      elif k == "ep": epi = getint(v)
      elif k == "ch":
        try: sep = v.index("-")
        except ValueError: loch, hich = getint(v), None
        else: loch, hich = getint(v[:sep]), getint(v[sep + 1:]) + 1
      else: raise ExpectedError("unknown episode option `%s'" % k)
    check(ti is not None, "missing title number")
    series = me._get_series(sname)
    me._cur_chapter = None

    title = ww.rest()
    if not series.wantedp: return
    season = series.ensure_season()
    if epi is None: epi = season.ep_i

    if ti == "-":
      check(season.implicitp, "audio source, but explicit season")
      dir = lookup(me._audirs, series.name,
                   "no title, and no audio directory")
      src = lookup(dir.episodes, season.ep_i,
                   "episode %d not found in audio dir `%s'" % (epi, dir.dir))

    else:
      try: src = me._isos[series.name]
      except KeyError: src = me._auto_epsrc(series)

    episode = season.add_episode(epi, neps, title, src, ti, loch, hich)
    me._pl.add_episode(episode)
    me._cur_episode = episode

  def _process_chapter(me, ww):
    check(me._cur_episode is not None, "no current episode")
    check(me._cur_episode.source.CHAPTERP,
          "episode source doesn't allow chapters")
    if me._chaptersp:
      if me._cur_chapter is None: i = 1
      else: i = me._cur_chapter.i + 1
      me._cur_chapter = me._cur_episode.add_chapter(ww.rest(), i)

  def parse_file(me, fn):
    with location(FileLocation(fn, 0)) as floc:
      with open(fn, "r") as f:
        for line in f:
          floc.stepline()
          sline = line.lstrip()
          if sline == "" or sline.startswith(";"): continue

          if line.startswith("!"): me._process_cmd(Words(line[1:]))
          elif not line[0].isspace(): me._process_episode(Words(line))
          else: me._process_chapter(Words(line))
    me._pl.done_season()

  def done(me):
    discs = set()
    for name, vdir in me._vdirs.items():
      if not me._series[name].wantedp: continue
      for s in vdir.seasons.values():
        for d in s.episodes.values():
          discs.add(d)
    for adir in me._audirs.values():
      for d in adir.episodes.values():
        discs.add(d)
    for d in sorted(discs, key = lambda d: d.fn):
      if d.neps is not None and d.neps != d.nuses:
        raise ExpectedError("disc `%s' has %d episodes, used %d times" %
                            (d.fn, d.neps, d.nuses))
    return me._pl

ROOT = "/mnt/dvd/archive/"

op = OP.OptionParser \
  (usage = "%prog [-c] [-s SERIES] EPLS",
   description = "Generate M3U playlists from an episode list.")
op.add_option("-c", "--chapters",
              dest = "chaptersp", action = "store_true", default = False,
              help = "Output individual chapter names")
op.add_option("-s", "--series",
              dest = "series", type = "str", default = None,
              help = "Output only the listed SERIES (comma-separated)")
opts, argv = op.parse_args()
if len(argv) != 1: op.print_usage(file = SYS.stderr); SYS.exit(2)
if opts.series is None:
  series_wanted = None
else:
  series_wanted = set()
  for name in opts.series.split(","): series_wanted.add(name)
try:
  ep = EpisodeListParser(series_wanted, opts.chaptersp)
  ep.parse_file(argv[0])
  pl = ep.done()
  pl.write(SYS.stdout)
except (ExpectedError, IOError, OSError) as e:
  LOC.report(e)
  SYS.exit(2)