[autoys] / flaccrip / flaccrip-discid

#! /bin/bash

set -e
: ${JBDIR=/mnt/jb}

###--------------------------------------------------------------------------
### CD identification algorithms.
###
### 1. CDDB
###
### CCLLLLNN [NTRACK TRACK-START... LENGTH]
###
### CC is a checksum of the track start times; LLLL is the offset of the
### leadout track, in seconds (rounded down), and NN is the total number of
### tracks.  All of these are in hexadecimal, and include the 150-frame (2
### second) pre-gap.  All of these are in hexadecimal.  Since a CD can have
### at most 99 tracks, and can contain no more than 90 minutes of audio (!),
### the other two items fit without needing reduction.
###
### The checksum is the sum of the decimal digits of the track start times,
### in seconds, reduced modulo 255.
###
### NTRACK is the number of tracks; LENGTH is the offset of the leadout in
### seconds.  These are the same as in the checksum, so repeating them is
### pointless, but it's done anyway.  The TRACK-STARTs are the track start
### offsets, in frames.
###
###
### 2. AccurateRip
###
### DA1-DA2-CDDBID
###
### CDDBID is the CDDB id as described above.  DA1 is simply the sum of the
### track starts, including the lead-out track; DA2 is the sum of the
### products TRACKNO * OFFSET for the audio tracks only, but including the
### final lead-out -- so a data track makes the last audio track look very
### long.  Another wrinkle: the OFFSET for the first track is forced to 1 if
### it's zero (to avoid the entry being lost, I presume, though I'm not sure
### why this is ever so useful).
###
###
### 3. MusicBrainz
###
### The MusicBrainz identification is a base64-encoded SHA-1 hash of the
### table of contents.  The base64 encoding uses `.', `_' and `-' in place of
### `+', `/' and `=', because the standard characters /all/ have special
### meanings in URL query strings.  (Duh.  And I'm not quite sure why we
### still need the trailing marker.)
###
### The message to be hashed is FIRST LAST LENGTH TRACK-START..., where FIRST
### and LAST are the first and last track numbers, LENGTH is the offset of
### the lead-out, in frames, and the TRACK-STARTs are the start offsets of
### the tracks, in order, also in frames.  The track numbers are two
### uppercase hex digits; the frame offsets are eight.  All of these are
### simply concatenated together.
###
### MusicBrainz only concerns itself with the audio tracks.  If there's a
### data track, then we ignore it, and the lead-out is considered to be 11400
### frames before the data track.

###--------------------------------------------------------------------------
### Command line.

format=cddb
while getopts "acCm" opt; do
  case "$opt" in
    a) format=accuraterip ;;
    c) format=cddb ;;
    C) format=cddb-tracks ;;
    m) format=musicbrainz ;;
    *) exit 1 ;;
  esac
done
shift $((OPTIND - 1))

case $# in
  0)
    ;;
  1)
    if [ -r "$1/.discid" ]; then
      exec <"$1/.discid"
    else
      exec < <($JBDIR/bin/flaccrip-toc "$1")
    fi
    ;;
  *)
    echo >&2 "Usage: $0 [-acCm] [DIRECTORY]"
    exit 1
    ;;
esac

###--------------------------------------------------------------------------
### Main work.

## Initial setup.
cddbck=0
cddbtracks=""
nt=0 nat=0
da=0 db=0
mbtracks=""

## Wander through the table of contents picking up unconsidered trifles.
while read type offset; do

  ## Bump the track numbers here.  Most things want 1-based numbering, so
  ## this is right.  Don't bump for the end marker.  Those who care
  ## (AccurateRip) will sort it out for themselves.
  case "$type" in
    T) nt=$((nt + 1)) nat=$((nat + 1));;
    D) nt=$((nt + 1)) ;;
  esac

  ## Update the CDDB state.  This is common to several formats.
  case "$type" in
    [TD])
      o=$((offset + 150))
      s=$((o/75))
      cddbtracks="${cddbtracks:+$cddbtracks }$o"
      while :; do
	case "$s" in
	  ?*) cddbck=$((cddbck + ${s:0:1})); s=${s#?} ;;
	  *) break ;;
	esac
      done
      ;;
    E)
      final=$offset
      ;;
  esac

  ## Update other bits of information.
  case "$type" in
    T)
      da=$((da + offset))
      db=$((db + nat*(offset > 0 ? offset : 1)))
      mbtracks="$mbtracks$(printf "%08X" $((offset + 150)))"
      ;;
    D)
      mbfinal=$((offset - 11250))
      ;;
    E)
      da=$((da + offset))
      db=$((db + (nat + 1)*(offset > 0 ? offset : 1)))
      case "${mbfinal+t}" in
	t) ;;
	*) mbfinal=$((offset + 150)) ;;
      esac
      ;;
  esac
done

## Sort out the CDDB id.
cddbid=$(printf "%02x%04x%02x" $((cddbck%255)) $((final/75)) $nt)

###--------------------------------------------------------------------------
### Produce the answer.

case "$format" in
  cddb)
    echo "$cddbid"
    ;;
  cddb-tracks)
    echo "$cddbid $nt $cddbtracks $((final/75 + 2))"
    ;;
  accuraterip)
    printf "%03d-%08x-%08x-%s\n" $nat $da $db $cddbid
    ;;
  musicbrainz)
    mb=$(printf "%02X%02X%08X%s" 1 $nat $mbfinal $mbtracks)
    for ((i = nat; i < 99; i++)); do
      mb="${mb}00000000"
    done
    printf "%s" $mb |
      openssl dgst -sha1 -binary |
      openssl base64 | tr '+/=' '._-'
    ;;
esac
Commit	Line	Data
583b7e4a MW	1	#! /bin/bash
	2
	3	set -e
	4	: ${JBDIR=/mnt/jb}
	5
	6	###--------------------------------------------------------------------------
	7	### CD identification algorithms.
	8	###
	9	### 1. CDDB
	10	###
	11	### CCLLLLNN [NTRACK TRACK-START... LENGTH]
	12	###
	13	### CC is a checksum of the track start times; LLLL is the offset of the
	14	### leadout track, in seconds (rounded down), and NN is the total number of
	15	### tracks. All of these are in hexadecimal, and include the 150-frame (2
	16	### second) pre-gap. All of these are in hexadecimal. Since a CD can have
	17	### at most 99 tracks, and can contain no more than 90 minutes of audio (!),
	18	### the other two items fit without needing reduction.
	19	###
	20	### The checksum is the sum of the decimal digits of the track start times,
	21	### in seconds, reduced modulo 255.
	22	###
	23	### NTRACK is the number of tracks; LENGTH is the offset of the leadout in
	24	### seconds. These are the same as in the checksum, so repeating them is
	25	### pointless, but it's done anyway. The TRACK-STARTs are the track start
	26	### offsets, in frames.
	27	###
	28	###
	29	### 2. AccurateRip
	30	###
	31	### DA1-DA2-CDDBID
	32	###
	33	### CDDBID is the CDDB id as described above. DA1 is simply the sum of the
	34	### track starts, including the lead-out track; DA2 is the sum of the
	35	### products TRACKNO * OFFSET for the audio tracks only, but including the
	36	### final lead-out -- so a data track makes the last audio track look very
	37	### long. Another wrinkle: the OFFSET for the first track is forced to 1 if
	38	### it's zero (to avoid the entry being lost, I presume, though I'm not sure
	39	### why this is ever so useful).
	40	###
	41	###
	42	### 3. MusicBrainz
	43	###
	44	### The MusicBrainz identification is a base64-encoded SHA-1 hash of the
	45	### table of contents. The base64 encoding uses `.', `_' and `-' in place of
	46	### `+', `/' and `=', because the standard characters /all/ have special
	47	### meanings in URL query strings. (Duh. And I'm not quite sure why we
	48	### still need the trailing marker.)
	49	###
	50	### The message to be hashed is FIRST LAST LENGTH TRACK-START..., where FIRST
	51	### and LAST are the first and last track numbers, LENGTH is the offset of
	52	### the lead-out, in frames, and the TRACK-STARTs are the start offsets of
	53	### the tracks, in order, also in frames. The track numbers are two
	54	### uppercase hex digits; the frame offsets are eight. All of these are
	55	### simply concatenated together.
	56	###
	57	### MusicBrainz only concerns itself with the audio tracks. If there's a
	58	### data track, then we ignore it, and the lead-out is considered to be 11400
	59	### frames before the data track.
	60
	61	###--------------------------------------------------------------------------
	62	### Command line.
	63
	64	format=cddb
65	while getopts "acCm" opt; do
66	case "$opt" in
67	a) format=accuraterip ;;
68	c) format=cddb ;;
69	C) format=cddb-tracks ;;
70	m) format=musicbrainz ;;
71	*) exit 1 ;;
72	esac
73	done
74	shift $((OPTIND - 1))
75
76	case $# in
77	0)
78	;;
79	1)
80	if [ -r "$1/.discid" ]; then
81	exec <"$1/.discid"
82	else
83	exec < <($JBDIR/bin/flaccrip-toc "$1")
84	fi
85	;;
86	*)
87	echo >&2 "Usage: $0 [-acCm] [DIRECTORY]"
88	exit 1
89	;;
90	esac
91
92	###--------------------------------------------------------------------------
93	### Main work.
94
95	## Initial setup.
96	cddbck=0
97	cddbtracks=""
98	nt=0 nat=0
99	da=0 db=0
100	mbtracks=""
101
102	## Wander through the table of contents picking up unconsidered trifles.
103	while read type offset; do
104
105	## Bump the track numbers here. Most things want 1-based numbering, so
106	## this is right. Don't bump for the end marker. Those who care
107	## (AccurateRip) will sort it out for themselves.
108	case "$type" in
109	T) nt=$((nt + 1)) nat=$((nat + 1));;
110	D) nt=$((nt + 1)) ;;
111	esac
112
113	## Update the CDDB state. This is common to several formats.
114	case "$type" in
115	[TD])
116	o=$((offset + 150))
117	s=$((o/75))
118	cddbtracks="${cddbtracks:+$cddbtracks }$o"
119	while :; do
120	case "$s" in
121	?*) cddbck=$((cddbck + ${s:0:1})); s=${s#?} ;;
122	*) break ;;
123	esac
124	done
125	;;
126	E)
127	final=$offset
128	;;
129	esac
130
131	## Update other bits of information.
132	case "$type" in
133	T)
134	da=$((da + offset))
135	db=$((db + nat*(offset > 0 ? offset : 1)))
136	mbtracks="$mbtracks$(printf "%08X" $((offset + 150)))"
137	;;
138	D)
139	mbfinal=$((offset - 11250))
140	;;
141	E)
142	da=$((da + offset))
143	db=$((db + (nat + 1)*(offset > 0 ? offset : 1)))
144	case "${mbfinal+t}" in
145	t) ;;
146	*) mbfinal=$((offset + 150)) ;;
147	esac
148	;;
149	esac
150	done
151
152	## Sort out the CDDB id.
153	cddbid=$(printf "%02x%04x%02x" $((cddbck%255)) $((final/75)) $nt)
154
155	###--------------------------------------------------------------------------
156	### Produce the answer.
157
158	case "$format" in
159	cddb)
160	echo "$cddbid"
161	;;
162	cddb-tracks)
163	echo "$cddbid $nt $cddbtracks $((final/75 + 2))"
164	;;
165	accuraterip)
166	printf "%03d-%08x-%08x-%s\n" $nat $da $db $cddbid
167	;;
168	musicbrainz)
169	mb=$(printf "%02X%02X%08X%s" 1 $nat $mbfinal $mbtracks)
170	for ((i = nat; i < 99; i++)); do
171	mb="${mb}00000000"
172	done
173	printf "%s" $mb \|
174	openssl dgst -sha1 -binary \|
175	openssl base64 \| tr '+/=' '._-'
176	;;
177	esac