Class: DTAS::Mlib

Inherits:
Object
  • Object
show all
Includes:
Process
Defined in:
lib/dtas/mlib.rb

Overview

For the DTAS Music Library, based on what MPD uses.

Defined Under Namespace

Classes: Job

Constant Summary collapse

DM_DIR = 
-1
DM_IGN = 
-2
TAGS =

same capitalization as in mpd

Hash[*(
    %w(Artist ArtistSort
Album AlbumSort
AlbumArtist AlbumArtistSort
Title Track Name
Genre Date Composer Performer Comment Disc
MUSICBRAINZ_ARTISTID MUSICBRAINZ_ALBUMID
MUSICBRAINZ_ALBUMARTISTID
MUSICBRAINZ_TRACKID
MUSICBRAINZ_RELEASETRACKID).map! { |x| [ x.downcase, x ]

Constants included from Process

Process::PIDS

Instance Attribute Summary collapse

Instance Method Summary collapse

Methods included from Process

#dtas_spawn, #env_expand, #env_expand_ary, #env_expand_i, #qx, reaper

Methods included from SpawnFix

#spawn

Methods included from XS

#xs

Constructor Details

#initialize(db) ⇒ Mlib

Returns a new instance of Mlib.



34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
 
# File 'lib/dtas/mlib.rb', line 34

def initialize(db)
  if String === db
    db = "sqlite://#{db}" unless db.include?('://')
    require 'sequel/no_core_ext'
    db = Sequel.connect(db, single_threaded: true)
  end
  if db.class.to_s.downcase.include?('sqlite')
    db.transaction_mode = :immediate
    db.synchronous = :off
    db.case_sensitive_like = false
  else
    warn 'non-SQLite databases may not work in the future'
  end
  @db = db
  @pwd = nil
  @follow_outside_symlinks = true
  @follow_inside_symlinks = true
  @root_node = nil
  @tags = TAGS.dup
  @tag_map = nil
  @suffixes = nil
  @work = nil
end

Instance Attribute Details

Returns the value of attribute follow_inside_symlinks.



13
14
15
 
# File 'lib/dtas/mlib.rb', line 13

def follow_inside_symlinks
  @follow_inside_symlinks
end

#follow_outside_symlinks ⇒ Object

:nodoc:



12
13
14
 
# File 'lib/dtas/mlib.rb', line 12

def follow_outside_symlinks
  @follow_outside_symlinks
end

#tags ⇒ Object

Returns the value of attribute tags.



14
15
16
 
# File 'lib/dtas/mlib.rb', line 14

def tags
  @tags
end

Instance Method Details

#cd(path) ⇒ Object 



275
276
277
278
279
280
281
282
283
 
# File 'lib/dtas/mlib.rb', line 275

def cd(path)
  prev_wd = @pwd
  Dir.chdir(path)
  cur = @pwd = Dir.pwd.b
  yield
ensure
  Dir.chdir(prev_wd) if cur && prev_wd
  @pwd = prev_wd
end

#count_distinct(tag) ⇒ Object 



368
369
370
371
 
# File 'lib/dtas/mlib.rb', line 368

def count_distinct(tag)
  s = 'SELECT COUNT(DISTINCT(val_id)) FROM comments WHERE tag_id = ?'
  @db.fetch(s, @tag_map[tag]).single_value
end

#count_songs ⇒ Object 



373
374
375
 
# File 'lib/dtas/mlib.rb', line 373

def count_songs
  @db.fetch('SELECT COUNT(*) FROM nodes WHERE tlen >= 0').single_value
end

#db_playtime ⇒ Object 



377
378
379
 
# File 'lib/dtas/mlib.rb', line 377

def db_playtime
  @db.fetch('SELECT SUM(tlen) FROM nodes WHERE tlen >= 0').single_value
end

#dir_vivify(parts, ctime) ⇒ Object 



236
237
238
239
240
241
242
243
244
245
 
# File 'lib/dtas/mlib.rb', line 236

def dir_vivify(parts, ctime)
  @db.transaction do
    dir = root_node
    last = parts.pop
    parts.each do |name|
      dir = node_ensure(dir[:id], name, DM_DIR)
    end
    node_ensure(dir[:id], last, DM_DIR, ctime)
  end
end

#dump(path, cache, cb) ⇒ Object

returns an array on error



341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
 
# File 'lib/dtas/mlib.rb', line 341

def dump(path, cache, cb)
  dir = path
  base = nil
  retried = false
  begin
    found = cd(dir) { find_dump_part(root_node, base) }
  rescue Errno::ENOTDIR
    raise if retried || found
    dir, base = File.split(path)
    retried = true
    retry
  end
  return found if Array === found # error

  # success
  load_tags
  @tag_rmap = @tag_map.invert
  if found[:tlen] == DM_DIR
    emit_recurse(found, cache, cb)
  else
    parent = @db[:nodes][id: found[:parent_id]]
    parent or abort "missing parent for #{found.inspect}"
    parent[:dirname] ||= path_of(parent, cache)
    emit_1(found, parent, cache, cb)
  end
end

#emit_1(node, parent, cb) ⇒ Object 



418
419
420
421
422
423
424
425
426
 
# File 'lib/dtas/mlib.rb', line 418

def emit_1(node, parent, cb)
  comments = Hash.new { |h,k| h[k] = [] }
  @db['SELECT c.tag_id, v.val FROM comments c ' \
      'LEFT JOIN vals v ON v.id = c.val_id ' \
      "WHERE c.node_id = #{node[:id]} ORDER BY c.tag_id"].map do |c|
    comments[@tag_rmap[c[:tag_id]]] << c[:val]
  end
  cb.call(parent, node, comments)
end

#emit_recurse(node, cache, cb) ⇒ Object 



405
406
407
408
409
410
411
412
413
414
415
416
 
# File 'lib/dtas/mlib.rb', line 405

def emit_recurse(node, cache, cb)
  node[:dirname] ||= path_of(node, cache)
  @db[:nodes].where(parent_id: node[:id]).order(:name).each do |nd|
    next if nd[:id] == node[:id] # root_node
    case nd[:tlen]
    when DM_DIR then emit_recurse(nd, cache, cb)
    when DM_IGN then next
    else
      emit_1(nd, node, cb)
    end
  end
end

#find(type, what, offset = 0, limit = nil) ⇒ Object

based on the MPD command of the same name, unstable API



456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
 
# File 'lib/dtas/mlib.rb', line 456

def find(type, what, offset = 0, limit = nil)
  load_tags
  type = type.downcase
  q = []
  case type
  when 'any'
    # TODO: add path name matches
    q << 'SELECT DISTINCT(n.id),n.* FROM nodes n ' \
         'LEFT JOIN comments c ON c.node_id = n.id ' \
         'LEFT JOIN vals v ON v.id = c.val_id ' \
         'WHERE v.val = ?'
  when *(@tags.keys)
    tag_id = @tag_map[type]
    q << 'SELECT DISTINCT(n.id),n.* FROM nodes n ' \
         'LEFT JOIN comments c ON c.node_id = n.id ' \
         'LEFT JOIN vals v ON v.id = c.val_id ' \
         'LEFT JOIN tags t ON t.id = c.tag_id ' \
         "WHERE v.val = ? AND t.id = #{tag_id}"
  else
    raise ArgumentError, "invalid type=#{type.inspect}"
  end
  q << 'ORDER by n.parent_id,n.name'
  offset_limit(q, offset, limit)
  @db[q.join(' '), what].each { |node| yield node }
end

#find_dump_part(cur, base) ⇒ Object 



320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
 
# File 'lib/dtas/mlib.rb', line 320

def find_dump_part(cur, base)
  parts = @pwd.split(%r{/+}n)
  parts.shift # no first part
  parts << base if base
  parts.each do |name|
    if cur = node_lookup(cur[:id], name)
      case cur[:tlen]
      when DM_DIR then next # keep going
      when DM_IGN then return [ :ignored, cur ]
      else # regular audio
        return cur if name.object_id == parts[-1].object_id
        return [ :notdir, cur ]
      end
    else
      return [ :missing, name ]
    end
  end
  cur
end

#ignore(job) ⇒ Object 



79
80
81
82
83
 
# File 'lib/dtas/mlib.rb', line 79

def ignore(job)
  @db.transaction do
    node_ensure(job.parent_id, job.path, DM_IGN, job.ctime)
  end
end

#init_suffixes ⇒ Object 



58
59
60
61
62
 
# File 'lib/dtas/mlib.rb', line 58

def init_suffixes
  `sox --help 2>/dev/null` =~ /\nAUDIO FILE FORMATS:\s*([^\n]+)/s
  re = $1.split(/\s+/).map { |x| Regexp.quote(x) }.join('|')
  @suffixes = Regexp.new("\\.(?:#{re})\\z", Regexp::IGNORECASE)
end

#load_tags ⇒ Object 



170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
 
# File 'lib/dtas/mlib.rb', line 170

def load_tags
  return @tag_map if @tag_map
  tag_map = {}
  tags = @db[:tags]
  @tags.each do |lc, mc|
    unless q = tags[tag: mc]
      q = { tag: mc }
      q[:id] = tags.insert(q)
    end
    tag_map[lc] = q[:id]
  end

  # Xiph tags use "tracknumber" and "discnumber"
  %w(track disc).each do |x|
    tag_id = tag_map[x] and tag_map["#{x}number"] = tag_id
  end
  @tag_map = tag_map.freeze
end

#migrate ⇒ Object 



160
161
162
163
164
165
166
167
168
 
# File 'lib/dtas/mlib.rb', line 160

def migrate
  require 'sequel'
  Sequel.extension(:migration, :core_extensions) # ugh...
  @db.transaction do
    Sequel::Migrator.apply(@db, "#{File.dirname(__FILE__)}/mlib/migrations")
    root_node # ensure this exists
    load_tags
  end
end

#node_ensure(parent_id, name, tlen, ctime = nil) ⇒ Object 



261
262
263
264
265
266
267
268
269
270
271
272
273
 
# File 'lib/dtas/mlib.rb', line 261

def node_ensure(parent_id, name, tlen, ctime = nil)
  q = { name: name, parent_id: parent_id }
  if node = @db[:nodes][q]
    node_update_maybe(node, tlen, ctime)
  else
    # brand new node
    node = q.dup
    node[:tlen] = tlen
    node[:ctime] = ctime
    node[:id] = @db[:nodes].insert(node)
  end
  node
end

#node_lookup(parent_id, name) ⇒ Object 



257
258
259
 
# File 'lib/dtas/mlib.rb', line 257

def node_lookup(parent_id, name)
  @db[:nodes][name: name, parent_id: parent_id]
end

#node_update_maybe(node, tlen, ctime) ⇒ Object 



247
248
249
250
251
252
253
254
255
 
# File 'lib/dtas/mlib.rb', line 247

def node_update_maybe(node, tlen, ctime)
  q = {}
  q[:ctime] = ctime if ctime && ctime != node[:ctime]
  q[:tlen] = tlen if tlen != node[:tlen]
  return if q.empty?
  node_id = node.delete(:id)
  @db[:nodes].where(id: node_id).update(node.merge(q))
  node[:id] = node_id
end

#offset_limit(q, offset, limit) ⇒ Object 



445
446
447
448
449
450
451
452
453
 
# File 'lib/dtas/mlib.rb', line 445

def offset_limit(q, offset, limit)
  offset = offset.to_s
  limit = limit.to_s
  if limit =~ %r{\A\d+\z}
    q << "LIMIT #{limit}"
    q << "OFFSET #{offset}" if offset =~ %r{\A\d+\z} && offset != '0'
  end
  q
end

#path_of(node, cache) ⇒ Object 



387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
 
# File 'lib/dtas/mlib.rb', line 387

def path_of(node, cache)
  base = node[:name]
  return '/' if base == '' # root_node
  parent_id = node[:parent_id]
  base += '/' unless node[:tlen] >= 0
  ppath = cache[parent_id] and return "#{ppath}/#{base}"
  parts = []
  begin
    node = @db[:nodes][id: node[:parent_id]]
    break if node[:id] == node[:parent_id]
    parts.unshift node[:name]
  end while true
  parts.unshift('')
  cache[parent_id] = parts.join('/')
  parts << base
  parts.join('/').freeze
end

#remove_entry(node) ⇒ Object 



428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
 
# File 'lib/dtas/mlib.rb', line 428

def remove_entry(node)
  root_id = root_node[:id]
  node_id = node[:id]
  q = { parent_id: node_id }
  nodes = @db[:nodes]
  nodes.where(q).each do |nd|
    next if nd[:id] == root_id
    case nd[:tlen]
    when DM_DIR, DM_IGN
      remove_entry(nd)
    end
  end
  nodes.where(q).delete
  @db[:comments].where(node_id: node_id).delete
  nodes.where(id: node_id).delete
end

#root_node ⇒ Object 



212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
 
# File 'lib/dtas/mlib.rb', line 212

def root_node
  q = @root_node and return q
  # root node always has parent_id: 1
  q = {
    parent_id: 1, # self
    name: '',
  }
  node = @db[:nodes][q]
  if node
    node[:dirname] = ''
    @root_node = node
    return node
  end
  begin
    q[:tlen] = DM_DIR
    q[:id] = @db[:nodes].insert(q)
    q
  rescue Sequel::DatabaseError
    # we may conflict on insert if we didn't use a transaction
    raise if @db.in_transaction?
    @root_node = @db[:paths][q] or raise
  end
end

#scan_any(path, parent_id) ⇒ Object 



189
190
191
192
193
194
195
196
197
198
 
# File 'lib/dtas/mlib.rb', line 189

def scan_any(path, parent_id)
  st = File.lstat(path) rescue return
  if st.directory?
    scan_dir(path, st, parent_id)
  elsif st.file?
    scan_file(path, st, parent_id)
  # elsif st.symlink? TODO
    # scan_link(path, st, parent_id)
  end
end

#scan_dir(path, st, parent_id = nil) ⇒ Object 



285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
 
# File 'lib/dtas/mlib.rb', line 285

def scan_dir(path, st, parent_id = nil)
  cd(path) do
    # TODO: use parent_id if given
    dir = dir_vivify(@pwd.split(%r{/+}n), st.ctime.to_i)
    dir_id = dir[:id]

    @db[:nodes].where(parent_id: dir_id).each do |node|
      File.exist?(node[:name]) or remove_entry(node)
    end

    Dir.foreach('.', encoding: Encoding::BINARY) do |x|
      case x
      when '.', '..', %r{\n}n
        # files with newlines in them are rare and last I checked (in 2008),
        # mpd could not support them, either.  So lets not bother for now.
        next
      else
        scan_any(x, dir_id)
      end
    end
  end
end

#scan_file(path, st, parent_id) ⇒ Object 



200
201
202
203
204
205
206
207
208
209
210
 
# File 'lib/dtas/mlib.rb', line 200

def scan_file(path, st, parent_id)
  return if @suffixes !~ path || st.size == 0

  # no-op if no change
  if node = @db[:nodes][name: path, parent_id: parent_id]
    return if st.ctime.to_i == node[:ctime] || node[:tlen] == DM_IGN
  end

  job = Job.new(@pwd, st.ctime.to_i, parent_id, path)
  send_harder(@work, Marshal.dump(job))
end

#search(type, what, offset = 0, limit = nil) ⇒ Object

based on the MPD command of the same name



483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
 
# File 'lib/dtas/mlib.rb', line 483

def search(type, what, offset = 0, limit = nil)
  load_tags
  type = type.downcase
  q = []
  what = @db.literal(%Q(%#{what}%))
  case type
  when 'any'
    # TODO: add path name matches
    q << 'SELECT DISTINCT(n.id),n.* FROM nodes n ' \
         'LEFT JOIN comments c ON c.node_id = n.id ' \
         'LEFT JOIN vals v ON v.id = c.val_id ' \
         "WHERE v.val LIKE #{what}"
  when *(@tags.keys)
    tag_id = @tag_map[type]
    q << 'SELECT DISTINCT(n.id),n.* FROM nodes n ' \
         'LEFT JOIN comments c ON c.node_id = n.id ' \
         'LEFT JOIN vals v ON v.id = c.val_id ' \
         'LEFT JOIN tags t ON t.id = c.tag_id ' \
         "WHERE t.id = #{tag_id} AND v.val LIKE #{what}"
  else
    raise ArgumentError, "invalid type=#{type.inspect}"
  end
  q << 'ORDER by n.parent_id,n.name'
  offset_limit(q, offset, limit)
  @db[q.join(' ')].each { |node| yield node }
end

#send_harder(sock, msg) ⇒ Object 



308
309
310
311
312
313
314
315
316
317
318
 
# File 'lib/dtas/mlib.rb', line 308

def send_harder(sock, msg)
  sock.sendmsg(msg)
rescue Errno::EMSGSIZE
  sock.setsockopt(:SOL_SOCKET, :SO_SNDBUF, msg.bytesize + 1024)
  # if it still fails, oh well...
  begin
    sock.sendmsg(msg)
  rescue => e
    warn "#{msg.bytesize} too big, dropped #{e.class}"
  end
end

#stats ⇒ Object 



381
382
383
384
385
 
# File 'lib/dtas/mlib.rb', line 381

def stats
  rv = { songs: count_songs, db_playtime: db_playtime }
  %w(artist album).each { |k| rv[:"#{k}s"] = count_distinct(k) }
  rv
end

#update(path, opts = nil) ⇒ Object 



136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
 
# File 'lib/dtas/mlib.rb', line 136

def update(path, opts = nil)
  # n.b. "jobs" is for CPU concurrency.  Audio media is typically stored
  # on high-latency media or slow network file systems; so we use a high
  # number of jobs by default to compensate for the seek-heavy workload
  # this generates
  opts ||= {}
  jobs = opts[:jobs] || 8

  init_suffixes
  st = File.stat(path) # we always follow the first dir even if it's a symlink
  st.directory? or
    raise ArgumentError, "path: #{path.inspect} is not a directory"
  @work and raise 'update already running'
  todo, @work = UNIXSocket.pair(:SOCK_SEQPACKET)
  @db.disconnect
  jobs.times { |i| fork { worker(todo) } }
  todo.close
  scan_dir(path, st)
  @work.close
  Process.waitall
ensure
  @work = nil
end

#worker(todo) ⇒ Object 



64
65
66
67
68
69
70
71
72
73
74
75
76
77
 
# File 'lib/dtas/mlib.rb', line 64

def worker(todo)
  @work.close
  @db.tables # reconnect before chdir
  @pwd = Dir.pwd.b
  begin
    buf = todo.recv(16384) # 4x bigger than PATH_MAX ought to be enough
    exit if buf.empty?
    job = Marshal.load(buf)
    buf.clear
    worker_work(job)
  rescue => e
    warn "#{e.message} (#{e.class}) #{e.backtrace.join("\n")}\n"
  end while true
end

#worker_work(job) ⇒ Object 



85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
 
# File 'lib/dtas/mlib.rb', line 85

def worker_work(job)
  tlen = nil
  wd = job.wd
  if wd != @pwd
    Dir.chdir(wd)
    @pwd = wd
  end
  tmp = {}
  path = job.path
  tlen = qx(%W(soxi -D #{path}), no_raise: true)
  return ignore(job) unless String === tlen
  tlen = tlen.to_f
  return ignore(job) if tlen < 0
  tlen = tlen.round
  buf = qx(%W(soxi -a #{path}), no_raise: true)
  return ignore(job) unless String === buf

  # no, we don't support comments with newlines in them
  buf = buf.split("\n")
  while line = buf.shift
    tag, value = line.split('=', 2)
    tag && value or next
    tag.downcase!
    tag_id = @tag_map[tag] or next
    value.strip!

    # FIXME: this fallback needs testing
    [ Encoding::UTF_8, Encoding::ISO_8859_1 ].each do |enc|
      value.force_encoding(enc)
      if value.valid_encoding?
        value.encode!(Encoding::UTF_8) if enc != Encoding::UTF_8
        tmp[tag_id] = value
        break
      end
    end
  end
  @db.transaction do
    node_id = node_ensure(job.parent_id, path, tlen, job.ctime)[:id]
    vals = @db[:vals]
    comments = @db[:comments]
    q = { node_id: node_id }
    comments.where(q).delete
    tmp.each do |tid, val|
      v = vals[val: val]
      q[:val_id] = v ? v[:id] : vals.insert(val: val)
      q[:tag_id] = tid
      comments.insert(q)
    end
  end
end