Skip to content

Instantly share code, notes, and snippets.

@maddievision

maddievision/pyIT.py

Last active December 14, 2015 03:18
Show Gist options
  • Save maddievision/5019503 to your computer and use it in GitHub Desktop.
Save maddievision/5019503 to your computer and use it in GitHub Desktop.
Download ZIP
pyIT
Raw
pyIT.py
#!/usr/bin/env python
# -*- coding: utf-8 -*-
"""\
Python module for handling Impulse Tracker files.
(c) 2008 mike burke / mrb / mrburke@gmail.com
doesn't and won't handle old format IT instruments (cmwt < 0x200),
but I don't think these even exist in the wild.
creates an IT with the basic structure:
IT header
message
patterns
sample headers
instruments
sample data
todo:
- add some compatibility-making code: fix envelopes that have no points, etc.
- create some exceptions to replace assertion errors
- remove redundant samples and instruments (already done for patterns)
"""
import os.path
import sys
import struct
from cStringIO import StringIO
import traceback
import logging
#import psyco
#psyco.full()
import pyitcompress
class ITenvelope_node(object):
def __init__(self):
self.y_val = 0
self.tick = 0
def __len__(self):
return 3
class ITenvelope(object):
def __init__(self):
self.IsOn = False
self.LoopOn = False
self.SusloopOn = False
self.LpB = 0
self.LpE = 0
self.SLB = 0
self.SLE = 0
# xxx convert this to not have 25 nodes always, and remove numNodePoints;
# the self.Nodes list should contain the number of node points
self.numNodePoints = 0
self.Nodes = [ITenvelope_node() for i in xrange(25)] # create 25 nodes
def extraFlags(self):
return 0
def write(self, outf):
flags = 0
flags = flags | ((self.IsOn) << 0)
flags = flags | ((self.LoopOn) << 1)
flags = flags | ((self.SusloopOn) << 2)
flags = flags | self.extraFlags()
outf.write(struct.pack('<BBBB', flags, self.numNodePoints, self.LpB, self.LpE))
outf.write(struct.pack('<BB', self.SLB, self.SLE))
for node in self.Nodes:
outf.write(struct.pack('<bH', node.y_val, node.tick))
outf.write('\0')
def load(self, inf):
(flags, self.numNodePoints, self.LpB, self.LpE, self.SLB,
self.SLE) = struct.unpack('<BBBBBB', inf.read(6))
self.setFlags(flags)
self.Nodes = []
for i in xrange(25):
node = ITenvelope_node()
self.Nodes.append(node)
(node.y_val, node.tick) = struct.unpack('<bH', inf.read(3))
inf.read(1)
def setFlags(self, flags):
self.IsOn = bool(flags & 0x01)
self.LoopOn = bool(flags & 0x02)
self.SusloopOn = bool(flags & 0x04)
def __len__(self):
return 82
class ITvol_envelope(ITenvelope):
def __init__(self):
ITenvelope.__init__(self)
class ITpan_envelope(ITenvelope):
def __init__(self):
ITenvelope.__init__(self)
class ITpitch_envelope(ITenvelope):
def __init__(self):
ITenvelope.__init__(self)
self.IsFilter = False
def extraFlags(self):
if self.IsFilter:
return 0x80
else:
return 0
def setFlags(self, flags):
ITenvelope.setFlags(self, flags)
self.IsFilter = bool(flags & 0x80)
class ITinstrument(object):
def __init__(self):
self.Filename = ''
self.NNA = 0
self.DCT = 0
self.DCA = 0
self.FadeOut = 0
self.PPS = 0
self.PPC = 0x3c
self.GbV = 128
self.DfP = 128
self.RV = 0
self.RP = 0
# TrkVers and NoS are ignored (used in instrument files only)
self.InstName = ''
self.IFC = 0
self.IFR = 0
self.MCh = 0
self.MPr = 0
self.MIDIBank = 0
self.SampleTable = [[i, 0] for i in range(120)]
self.volEnv = ITvol_envelope()
self.panEnv = ITpan_envelope()
self.pitchEnv = ITpitch_envelope()
def write(self, outf):
outf.write(struct.pack('<4s12s', 'IMPI', self.Filename))
outf.write(struct.pack('<BBBB', 0, self.NNA, self.DCT, self.DCA))
outf.write(struct.pack('<HBB', self.FadeOut, self.PPS, self.PPC))
outf.write(struct.pack('<BBBB', self.GbV, self.DfP, self.RV, self.RP))
outf.write(struct.pack('<HBB', 0xadde, 0xbe, 0xef)) # unused data
outf.write(struct.pack('<26s', self.InstName[:25]+'\0'))
outf.write(struct.pack('<BBBBH', self.IFC, self.IFR, self.MCh, self.MPr, self.MIDIBank))
for smp in self.SampleTable:
outf.write(struct.pack('<BB', smp[0], smp[1]))
self.volEnv.write(outf)
self.panEnv.write(outf)
self.pitchEnv.write(outf)
outf.write('FOOB')
def load(self, inf):
"""inf must be seeked to position of instrument to be read"""
(IMPI, self.Filename) = struct.unpack('<4s12s', inf.read(16))
assert(IMPI == 'IMPI')
(zero, self.NNA, self.DCT, self.DCA, self.FadeOut, self.PPS, self.PPC,
self.GbV, self.DfP, self.RV, self.RP, discard, discard,
discard) = struct.unpack('<BBBBHBBBBBBHBB', inf.read(16))
# seems some mods (saved by a bad schismtracker, maybe?)
# don't have zero = 0x0
#assert(zero == 0x0)
self.InstName = inf.read(26).replace('\0', ' ')[:25]
(self.IFC, self.IFR, self.MCh, self.MPr,
self.MIDIBank) = struct.unpack('<BBBBH', inf.read(6))
self.SampleTable = []
for i in xrange(120):
self.SampleTable.append(list(struct.unpack('<BB', inf.read(2))))
self.volEnv = ITvol_envelope()
self.panEnv = ITpan_envelope()
self.pitchEnv = ITpitch_envelope()
self.volEnv.load(inf)
self.panEnv.load(inf)
self.pitchEnv.load(inf)
inf.read(4) # dummy read
def __len__(self):
return 554
class ITsample(object):
def __init__(self):
self.Filename = ''
self.GvL = 64
self.IsSample = False
self.Is16bit = False
self.IsStereo = False
self.IsCompressed = False
self.IsLooped = False
self.IsSusLooped = False
self.IsPingPongLoop = False
self.IsPingPongSusLoop = False
self.Vol = 64
self.SampleName = ''
self.Cvt = 0x01
self.DfP = 0x00
# length is determined by sample data
# note, lengths and loop indices are in SAMPLES, not BYTES
self.LoopBegin = 0
self.LoopEnd = 0
self.C5Speed = 8363
self.SusLoopBegin = 0
self.SusLoopEnd = 0
self.ViS = 0
self.ViD = 0
self.ViT = 0
self.ViR = 0
self.SampleData = ''
self.CompressedSampleData = None
self._original_sample_data = self.SampleData
def sampleDataLen(self):
"""
Return the length of the sample data in SAMPLES.
"""
divider = 1
if self.Is16bit:
divider = divider * 2
if self.IsStereo:
divider = divider * 2
return len(self.SampleData) / divider
def rawSampleData(self):
"""
Return the raw sample data.
If you are saving the sample data, this is the correct function to call
as it will return the COMPRESSED data if possible.
If you are modifying the sample data, DO NOT call this function; use
SampleData directly. It's ok, I promise.
"""
self._check_compression_status()
if self.IsCompressed and self.CompressedSampleData is not None:
return self.CompressedSampleData
else:
return self.SampleData
def _check_compression_status(self):
"""
Check if the data should be stored compressed.
If modifications were made to the sample data, and it was compressed,
we have to save it uncompressed, because re-compression is not
implemented.
However, if it wasn't modified, we try to re-save the original
compressed data.
"""
if self.IsCompressed and self.modified():
self.IsCompressed = False
def write(self, outf, sample_offset):
log = logging.getLogger('pyIT.ITsample.save')
if not self.IsSample:
self.SampleData = ''
self._check_compression_status()
flags = 0
flags = flags | ((self.IsSample) << 0)
flags = flags | ((self.Is16bit) << 1)
flags = flags | ((self.IsStereo) << 2)
flags = flags | ((self.IsCompressed) << 3)
flags = flags | ((self.IsLooped) << 4)
flags = flags | ((self.IsSusLooped) << 5)
flags = flags | ((self.IsPingPongLoop) << 6)
flags = flags | ((self.IsPingPongSusLoop) << 7)
#log.debug(" Cvt (convert) = 0x%02x" % (self.Cvt,))
#log.debug(" Flg (flags) = 0x%02x" % (flags,))
#self.Cvt = 0x01
outf.write(struct.pack('<4s12s', 'IMPS', self.Filename))
outf.write(struct.pack('<BBBB', 0, self.GvL, flags, self.Vol))
outf.write(struct.pack('<26s', self.SampleName[:25]+'\0'))
outf.write(struct.pack('<BB', self.Cvt, self.DfP))
outf.write(struct.pack('<I', self.sampleDataLen()))
outf.write(struct.pack('<III', self.LoopBegin, self.LoopEnd, self.C5Speed))
outf.write(struct.pack('<II', self.SusLoopBegin, self.SusLoopEnd))
outf.write(struct.pack('<I', sample_offset))
outf.write(struct.pack('<BBBB', self.ViS, self.ViD, self.ViT, self.ViR))
def load(self, inf):
log = logging.getLogger('pyIT.ITsample.load')
(IMPS, self.Filename) = struct.unpack('<4s12s', inf.read(16))
assert(IMPS == 'IMPS')
(zero, self.GvL, flags, self.Vol) = struct.unpack('<BBBB', inf.read(4))
# seems some mods (saved by a bad schismtracker, maybe?)
# don't have zero = 0x0
#assert(zero == 0x0)
self.IsSample = bool(flags & 0x01)
self.Is16bit = bool(flags & 0x02)
self.IsStereo = bool(flags & 0x04)
self.IsCompressed = bool(flags & 0x08)
self.IsLooped = bool(flags & 0x10)
self.IsSusLooped = bool(flags & 0x20)
self.IsPingPongLoop = bool(flags & 0x40)
self.IsPingPongSusLoop = bool(flags & 0x80)
self.SampleName = inf.read(26).replace('\0', ' ')[:25]
log.debug("=> Loading sample %s" % (self.SampleName,))
(self.Cvt, self.DfP) = struct.unpack('<BB', inf.read(2))
log.debug(" Cvt (convert) = 0x%02x" % (self.Cvt,))
self.IT215Compression = self.IsCompressed and bool(self.Cvt & 0x04)
(length, self.LoopBegin, self.LoopEnd, self.C5Speed) = struct.unpack('<IIII', inf.read(16))
(self.SusLoopBegin, self.SusLoopEnd, offs_sampledata, self.ViS,
self.ViD, self.ViT, self.ViR) = struct.unpack('<IIIBBBB', inf.read(16))
# load sample, if there is one
if self.IsSample and length > 0:
# first, find length in bytes (not samples!)
mult = 1
if self.Is16bit:
mult = mult * 2
if self.IsStereo:
mult = mult * 2
log.debug(" length in samples is %d" % (length,))
if self.IsCompressed:
log.debug(" compressed!")
# real sample decompression
decompressedbuf = StringIO()
if self.Is16bit:
decompressor = pyitcompress.it_decompress16
log.debug(" 16-bit compressed sample at %d" % (offs_sampledata,))
else:
decompressor = pyitcompress.it_decompress8
log.debug(" 8-bit compressed sample at %d" % (offs_sampledata,))
inf.seek(offs_sampledata)
try:
# Load compressed sample
if self.IT215Compression:
log.debug(" IT 2.15 sample compression")
compressed_len = decompressor(decompressedbuf, length, inf, self.IT215Compression)
self.SampleData = decompressedbuf.getvalue()
log.debug(" compressed length: %d; decompressed length: %d" % (compressed_len, len(self.SampleData)))
# Load actual compressed sample data in case we want
# to re-save it later
inf.seek(offs_sampledata)
self.CompressedSampleData = inf.read(compressed_len)
# Retain reference to original sample data; we can use
# this with modified() to determine if the sample was
# modified.
#
# This is used later for re-saving compressed data.
self._original_sample_data = self.SampleData
except:
print
traceback.print_exc()
else:
# Load uncompressed sample
length = length * mult
log.debug(" length in bytes is %s" % (length,))
inf.seek(offs_sampledata)
self.SampleData = inf.read(length)
self.CompressedSampleData = None
self._original_sample_data = self.SampleData
def modified(self):
return (self.SampleData is not self._original_sample_data)
def __len__(self):
return 80
class ITnote(object):
def __init__(self):
self.Note = None
self.Instrument = None
self.Volume = None
self.Effect = None
self.EffectArg = None
def __eq__(self, other):
return (self.Note == other.Note and
self.Instrument == other.Instrument and
self.Volume == other.Volume and
self.Effect == other.Effect and
self.EffectArg == other.EffectArg)
def __ne__(self, other):
return not (self == other)
def note_num_as_str(self, note_num):
# C C# D D# E F F# G G# A A# B
if self.Note is None:
return '...'
if self.Note == 254:
return '^^^'
if self.Note == 255:
return '==='
note_list = [
'C', 'C#', 'D', 'D#', 'E', 'F', 'F#', 'G', 'G#', 'A', 'A#', 'B']
pitch = note_list[note_num % 12]
octave = note_num / 12
return ('%-2s%d' % (pitch, octave)).replace(' ', '-')
def __str__(self):
if self.Instrument is None:
instrument = ".."
else:
instrument = "%02d" % self.Instrument
if self.Volume is None:
volume = ".."
else:
volume = "%02d" % self.Volume
if self.Effect is None:
effect = ".."
else:
effect = "%02d" % self.Effect
if self.EffectArg is None:
effectarg = ".."
else:
effectarg = "%02x" % self.EffectArg
return "%s %s %s %s%s" % (self.note_num_as_str(self.Note),
instrument,
volume,
effect,
effectarg
)
class ITpattern(object):
def __init__(self):
# Fill pattern with a bunch of empty ITnote instances.
# self.Rows[4][2] would return the note on the third channel in
# the fifth row.
self.Rows = [[ITnote() for i in xrange(64)] for j in xrange(64)]
def __len__(self):
return len(self.pack()) + 8
def __eq__(self, other):
return self.Rows == other.Rows
def __ne__(self, other):
return not (self == other)
def isEmpty(self):
""" 'empty' here uses the IT definition of a 64-row pattern with no note data. """
return self == ITpattern()
def write(self, outf):
ptndata = self.pack()
outf.write(struct.pack('<HH4s', len(ptndata), len(self.Rows), '\0'*4))
outf.write(ptndata)
def unpack(self, rows, ptndata):
"""
Unpack the raw pattern data stored in self.ptnData.
"""
log = logging.getLogger("pyIT.ITpattern.unpack")
log.info("load pattern: rows = %d, len = %d" %(rows, len(ptndata),))
ptn_reader = StringIO(ptndata)
masks = [0] * 64 # prepare mask variables
last_note = [ITnote() for i in xrange(64)] # last note storage
# Reset row data
self.Rows = [[ITnote() for i in xrange(64)] for j in xrange(rows)]
row_num = 0
while True:
chan_data = ptn_reader.read(1)
if chan_data == '': # end of data
break
chan_data = struct.unpack('<B', chan_data)[0]
if chan_data == 0: # end of row
row_num = row_num + 1
continue
chan_num = (chan_data-1) & 63 # get channel number for this data
if chan_data & 128: # new value for this channel's mask variable
masks[chan_num] = struct.unpack('<B', ptn_reader.read(1))[0]
mask = masks[chan_num]
if mask & 1:
self.Rows[row_num][chan_num].Note = struct.unpack('<B', ptn_reader.read(1))[0]
last_note[chan_num].Note = self.Rows[row_num][chan_num].Note
if mask & 2:
self.Rows[row_num][chan_num].Instrument = struct.unpack('<B', ptn_reader.read(1))[0]
last_note[chan_num].Instrument = self.Rows[row_num][chan_num].Instrument
if mask & 4:
self.Rows[row_num][chan_num].Volume = struct.unpack('<B', ptn_reader.read(1))[0]
last_note[chan_num].Volume = self.Rows[row_num][chan_num].Volume
if mask & 8:
(self.Rows[row_num][chan_num].Effect,
self.Rows[row_num][chan_num].EffectArg) = struct.unpack('<BB', ptn_reader.read(2))
last_note[chan_num].Effect = self.Rows[row_num][chan_num].Effect
last_note[chan_num].EffectArg = self.Rows[row_num][chan_num].EffectArg
if mask & 16:
self.Rows[row_num][chan_num].Note = last_note[chan_num].Note
if mask & 32:
self.Rows[row_num][chan_num].Instrument = last_note[chan_num].Instrument
if mask & 64:
self.Rows[row_num][chan_num].Volume = last_note[chan_num].Volume
if mask & 128:
self.Rows[row_num][chan_num].Effect = last_note[chan_num].Effect
self.Rows[row_num][chan_num].EffectArg = last_note[chan_num].EffectArg
#row_num = 0
#for row in self.Rows:
# pretty_row = ' | '.join([str(row[x]) for x in xrange(4)])
# log.debug("Row %02d: %s" % (row_num, pretty_row))
# row_num = row_num + 1
def pack(self):
"""
Pack pattern data back and return it as a string of raw data.
"""
log = logging.getLogger("pyIT.ITpattern.unpack")
ptn_writer = StringIO()
masks = [0] * 64 # prepare mask variables
last_note = [ITnote() for i in xrange(64)] # last note storage
empty_note = ITnote()
for row_data in self.Rows:
for chan_num in xrange(64):
# Anything in channel?
note = row_data[chan_num]
if note == empty_note:
continue
# Find out what mask variable should be, and pack note data
# in a temporary StringIO.
#
# This needs to be stored in a temporary place, as chan_data
# and mask won't be known until after we've looked at the
# entire note.
mask = 0
packed_note = StringIO()
if note.Note is not None:
if note.Note == last_note[chan_num].Note:
mask |= 16
else:
packed_note.write(struct.pack('<B', note.Note))
last_note[chan_num].Note = note.Note
mask |= 1
if note.Instrument is not None:
if note.Instrument == last_note[chan_num].Instrument:
mask |= 32
else:
packed_note.write(struct.pack('<B', note.Instrument))
last_note[chan_num].Instrument = note.Instrument
mask |= 2
if note.Volume is not None:
if note.Volume == last_note[chan_num].Volume:
mask |= 64
else:
packed_note.write(struct.pack('<B', note.Volume))
last_note[chan_num].Volume = note.Volume
mask |= 4
if note.Effect is not None or note.EffectArg is not None:
if (note.Effect == last_note[chan_num].Effect and
note.EffectArg == last_note[chan_num].EffectArg):
mask |= 128
else:
mask |= 8
write_effect = note.Effect
write_effectarg = note.EffectArg
if write_effect is None:
write_effect = 0
if write_effectarg is None:
write_effectarg = 0
last_note[chan_num].Effect = write_effect
last_note[chan_num].EffectArg = write_effectarg
packed_note.write(struct.pack('<BB',
write_effect,
write_effectarg))
# Check if we will reuse last mask
if mask == masks[chan_num]:
ptn_writer.write(struct.pack('<B', (chan_num + 1)))
else:
ptn_writer.write(struct.pack('<BB',
(chan_num + 1) | 128,
mask))
masks[chan_num] = mask
ptn_writer.write(packed_note.getvalue())
# Write end-row marker.
ptn_writer.write("\x00")
return ptn_writer.getvalue()
def load(self, inf):
"""Load IT pattern data from inf. inf should already be seeked to
the offset of the pattern to be loaded."""
(ptnlen, rows, discard) = struct.unpack('<HH4s', inf.read(8))
ptndata = inf.read(ptnlen)
self.unpack(rows, ptndata)
class ITfile(object):
Orderlist_offs = 192 # length of IT header before any dynamic data (order list)
pyIT_Cwt_v = 0x4101 # This value will be written into Cwt_v ("created with
# tracker version") upon write().
def __init__(self):
self.SongName = ''
self.PHilight_minor = 4
self.PHilight_major = 16
# OrdNum, InsNum, SmpNum, PatNum are used only when loading files;
# the actual numbers will be stored as len(lists)
self.Cwt_v = ITfile.pyIT_Cwt_v
self.Cmwt = 0x0214
self.Flags = 0x000d
self.Special = 0x0006
self.GV = 128 # global vol
self.MV = 48 # mixing vol
self.IS = 6 # initial speed
self.IT = 125 # initial tempo
self.Sep = 128 # stereo separation
self.PWD = 0x00
# msglen is also collected by actual message length
self.Message = ''
self.ChannelPans = 64*[32]
self.ChannelVols = 64*[64]
self.Orders = []
self.Instruments = []
self.Samples = []
self.Patterns = []
def open(self, infilename):
log = logging.getLogger("pyIT.ITfile.open")
inf = file(infilename, "rb")
buf = inf.read(30)
(IMPM, self.SongName) = struct.unpack('<4s26s', buf)
assert(IMPM == 'IMPM')
self.SongName = self.SongName.split('\0')[0]
buf = inf.read(34)
(self.PHilight_minor, self.PHilight_major, n_ords, n_insts, n_samps,
n_ptns, self.Cwt_v, self.Cmwt, self.Flags, self.Special, self.GV, self.MV,
self.IS, self.IT, self.Sep, self.PWD, msglen, offs_msg, reserved) = struct.unpack(
'<BBHHHHHHHHBBBBBBHII', buf)
offs_ords = ITfile.Orderlist_offs
offs_instoffs = offs_ords + n_ords
offs_sampoffs = offs_instoffs + n_insts*4
offs_ptnoffs = offs_sampoffs + n_samps*4
assert(inf.tell() == 0x40)
self.ChannelPans = []
for i in xrange(64):
self.ChannelPans.append(struct.unpack('<B', inf.read(1))[0])
self.ChannelVols = []
for i in xrange(64):
self.ChannelVols.append(struct.unpack('<B', inf.read(1))[0])
assert(inf.tell() == offs_ords)
self.Orders = []
for i in xrange(n_ords):
self.Orders.append(struct.unpack('<B', inf.read(1))[0])
assert(inf.tell() == offs_instoffs)
offs_insts = []
for i in xrange(n_insts):
offs_insts.append(struct.unpack('<I', inf.read(4))[0])
assert(inf.tell() == offs_sampoffs)
offs_samps = []
for i in xrange(n_samps):
offs_samps.append(struct.unpack('<I', inf.read(4))[0])
assert(inf.tell() == offs_ptnoffs)
offs_ptns = []
for i in xrange(n_ptns):
offs_ptns.append(struct.unpack('<I', inf.read(4))[0])
# load song message
if (self.Special & 0x0001) and (msglen > 0):
inf.seek(offs_msg)
self.Message = inf.read(msglen).replace('\0', ' ').replace('\r', '\n')[:-1]
else:
self.Message = ''
# load patterns
self.Patterns = []
for offs_ptn in offs_ptns:
ptn = ITpattern()
if offs_ptn != 0:
inf.seek(offs_ptn)
ptn.load(inf)
self.Patterns.append(ptn)
# load instruments
self.Instruments = []
for offs_inst in offs_insts:
inf.seek(offs_inst)
inst = ITinstrument()
try:
inst.load(inf)
except:
# the instrument failed to load, but we'll pretend it didn't
pass
self.Instruments.append(inst)
self.Samples = []
for offs_samp in offs_samps:
inf.seek(offs_samp)
samp = ITsample()
try:
samp.load(inf)
except:
# the sample failed to load, but we'll pretend it didn't
# we might need to do some cleanup...
pass
self.Samples.append(samp)
inf.close()
def write(self, outfilename):
log = logging.getLogger("pyIT.ITfile.write")
outf = file(outfilename, "wb")
# This is a comment. I like comments.
if (len(self.Message) > 0):
self.Special = self.Special | 0x0001
message = self.Message.replace('\n', '\r') + '\0'
else:
self.Special = self.Special & (~0x0001)
message = ''
# We set "Compatible with" to IT 2.15 when saving IT 2.15 samples,
# so that modplug-based loaders knows what the hell is up.
#
# Let's scan all our samples to see.
self.Cwt_v = ITfile.pyIT_Cwt_v
self.Cmwt = 0x0214
for sample in self.Samples:
sample._check_compression_status()
if sample.IsCompressed and sample.IT215Compression:
log.debug("Song contains at least one IT 2.15 sample; setting cmwt == 0x0215")
self.Cmwt = 0x0215
break
instoffs_offs = ITfile.Orderlist_offs + len(self.Orders)
sampoffs_offs = instoffs_offs + len(self.Instruments)*4
ptnoffs_offs = sampoffs_offs + len(self.Samples)*4
msg_offs = ptnoffs_offs + len(self.Patterns)*4
ptn_offs = msg_offs + len(message)
# pack patterns so we can predict total pattern data length, and
# next offset
(pattern_list, unique_ITpatterns) = self.pack_ptns()
ptn_offsets = {}
offs = ptn_offs
for x in pattern_list:
if x is not False and x not in ptn_offsets:
# unknown pattern
# store new pattern offset
ptn_offsets[x] = offs
offs = offs + len(unique_ITpatterns[x])
#samp_offs = ptn_offs + sum([len(x) for x in self.Patterns])
samp_offs = offs
inst_offs = samp_offs + sum([len(x) for x in self.Samples])
sampledata_offs = inst_offs + sum([len(x) for x in self.Instruments])
# write header
songname = self.SongName[:25].ljust(26, '\x00')
outf.write(struct.pack('<4s26sBB', 'IMPM', songname, self.PHilight_minor, self.PHilight_major))
outf.write(struct.pack('<HHHHHHHH', len(self.Orders), len(self.Instruments),
len(self.Samples), len(self.Patterns),
self.Cwt_v, self.Cmwt, self.Flags, self.Special))
outf.write(struct.pack('<BBBBBBHII', self.GV, self.MV, self.IS, self.IT,
self.Sep, self.PWD, len(message), msg_offs, 0))
for x in self.ChannelPans:
# x >= 128 == muted
if (x > 64 and x < 128):
x = 100 # surround
elif x < 0:
x = 0
outf.write(struct.pack('<B', x))
for x in self.ChannelVols:
if (x > 64):
x = 64
elif x < 0:
x = 0
outf.write(struct.pack('<B', x))
assert(outf.tell() == ITfile.Orderlist_offs)
for x in self.Orders:
if (x > 199):
if (x < 254):
x = 199
elif (x > 255):
x = 255
elif x < 0:
x = 0
outf.write(struct.pack('<B', x))
assert(outf.tell() == instoffs_offs)
offs = inst_offs
for x in self.Instruments:
outf.write(struct.pack('<I', offs))
offs = offs + len(x)
assert(outf.tell() == sampoffs_offs)
offs = samp_offs
for x in self.Samples:
outf.write(struct.pack('<I', offs))
offs = offs + len(x)
assert(outf.tell() == ptnoffs_offs)
# save patterns (packed)
for x in pattern_list:
if x is False:
log.debug("write empty pattern offs")
ptnoffs = 0
else:
log.debug("write real pattern offs")
ptnoffs = ptn_offsets[x]
outf.write(struct.pack('<I', ptnoffs))
assert(outf.tell() == msg_offs)
if message:
outf.write(message)
assert(outf.tell() == ptn_offs)
for ptn in unique_ITpatterns:
log.debug("write pattern")
ptn.write(outf)
assert(outf.tell() == samp_offs)
# next_smpoffs is the actual offset of the sample data for each sample.
# It's stored in the header, so writing the header needs to know it.
next_smpoffs = sampledata_offs
for samp in self.Samples:
samp.write(outf, next_smpoffs)
next_smpoffs = next_smpoffs + len(samp.rawSampleData())
eof = next_smpoffs
assert(outf.tell() == inst_offs)
for inst in self.Instruments:
inst.write(outf)
assert(outf.tell() == sampledata_offs)
for samp in self.Samples:
outf.write(samp.rawSampleData())
assert(outf.tell() == eof)
outf.close()
def pack_ptns(self):
"""Returns a tuple(pattern_list, unique_ITpatterns)"""
ptnlist = []
ptns = []
for ptn in self.Patterns:
if ptn.isEmpty():
# empty pattern is empty
ptnlist.append(False)
elif ptn in ptns:
# already in pattern set, create a reference only
ptnlist.append(ptns.index(ptn))
else:
# doesn't exist in pattern set, add it and create a reference to it
ptns.append(ptn)
ptnlist.append(ptns.index(ptn))
return (ptnlist, ptns)
def process():
#logging.basicConfig(level=logging.DEBUG, format="%(asctime)s - %(name)s - %(levelname)s - %(message)s")
logging.basicConfig(level=logging.DEBUG, format="%(name)-24s %(levelname)-7s %(message)s")
# pyitcompress is like, really noisy on the DEBUG channel,
# and sample compression slows down a shitload if you let it print
# all that shit to your screen.
#logging.getLogger("pyitcompress").setLevel(level=logging.WARNING)
itf = ITfile()
assert(len(sys.argv) == 2)
itf.open(sys.argv[1])
#logging.info("Cwt_v is 0x%04x" % (itf.Cwt_v,))
## set all samples to "uncompressed" (should prevent re-saving of
## compressed samples in favour of uncompressed versions)
for samp in itf.Samples:
samp.IsCompressed = False
## modify all samples very slightly (should prevent re-saving of compressed
## samples in favour of uncompressed versions)
#for samp in itf.Samples:
#samp.SampleData = samp.SampleData + "\0\0\0\0"
#for samp in itf.Samples:
# print samp.SampleName.decode('cp437')
itf.write('new.it')
# Create a mostly-empty .IT file
#itf = ITfile()
#itf.Orders.append(0)
#itf.Instruments.append(ITinstrument())
#itf.Instruments[0].Filename = 'fallow'
#itf.Instruments[0].InstName = 'aaaaaa!!'
#
#itf.Samples.append(ITsample())
#itf.Samples[0].Filename = 'HUUU'
#itf.Samples[0].SampleName = 'Bubuuu!!!'
#
#itf.Message = 'ahahaha!'
#itf.write('bloo.it')
if __name__ == '__main__':
process()
Raw
pyitcompress.py
# -*- coding: utf-8 -*-
"""
* Schism Tracker - a cross-platform Impulse Tracker clone
* copyright (c) 2003-2005 Storlek <storlek@rigelseven.com>
* copyright (c) 2005-2008 Mrs. Brisby <mrs.brisby@nimh.org>
* copyright (c) 2009 Storlek & Mrs. Brisby
* URL: http://schismtracker.org/
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
"""
#define NEED_BYTESWAP
#include "headers.h"
#include "fmt.h"
# ------------------------------------------------------------------------------------------------------------
# IT decompression code from itsex.c (Cubic Player) and load_it.cpp (Modplug)
# (I suppose this could be considered a merge between the two.)
import sys
import logging
class ReadBitsState:
def __init__(self):
self.bitbuf = 0
self.bitnum = 0
def MIN(a, b):
if a < b:
return a
else:
return b
def it_readbits(n, state, stream):
value = 0
i = n
#log = logging.getLogger("pyitcompress.it_readbits")
#log.debug("bitnum=%d, bitbuf=%d, n=%d" % (state.bitnum, state.bitbuf, n))
# this could be better
while i:
i -= 1
if not state.bitnum:
state.bitbuf = ord(stream.read(1))
state.bitnum = 8
value >>= 1
#logging.debug("state.bitbuf = 0x%x, state.bitbuf<<31 = 0x%x" % (state.bitbuf, state.bitbuf<<31))
value |= (state.bitbuf << 31) & 0xffffffff
state.bitbuf >>= 1
state.bitnum -= 1
return value >> (32 - n)
##
## Converting an unsigned value to a signed one.
##
#
# byte_signer_pack = struct.Struct("@B")
# byte_signer_unpack = struct.Struct("@b")
#
# >>> def signbyte(b): return b - 256 if b > 127 else b
# ...
# >>> def signbyte_s(b): return byte_signer_unpack.unpack(byte_signer_pack.pack(b))[0]
# ...
# >>> def signbyte_s2(b): return struct.unpack("@b", struct.pack("@B", b))[0]
# ...
# >>> timeit.repeat(lambda: signbyte(234))
# [0.3214220080452037, 0.3246569789375826, 0.3103202065507844]
# >>> timeit.repeat(lambda: signbyte_s(234))
# [0.665931344571618, 0.6369421357301803, 0.6587316597669997]
# >>> timeit.repeat(lambda: signbyte_s2(234))
# [0.8120824689951291, 0.7978116412757572, 0.7951949434834091]
#
##
## Word (16-bit) conversions have similar characteristics.
##
def signbyte(b):
return b - 256 if b > 127 else b
def unsignbyte(b):
#logging.getLogger("pyitcompress.unsignbyte").debug("converting %d" % (b,))
return b & 0xff
def signword(w):
#logging.getLogger("pyitcompress.signword").debug("converting %d" % (b,))
return w - 65536 if w > 32767 else w
def unsignword(w):
#logging.getLogger("pyitcompress.unsignword").debug("converting %d" % (b,))
return w & 0xffff
def it_decompress8(dest, len, srcbuf, it215):
"""
dest: (file-like object) output buffer for decompressed data
len: number of samples
srcbuf: (file-like object) input
it215: (bool) use it215 algorithm
RETURN: actual size (in bytes) of COMPRESSED data
"""
#const uint8_t *filebuf # source buffer containing compressed sample data
#const uint8_t *srcbuf # current position in source buffer
#int8_t *destpos # position in destination buffer which will be returned
#uint16_t blklen # length of compressed data block in samples
#uint16_t blkpos # position in block
#uint8_t width # actual "bit width"
#uint16_t value # value read from file to be processed
#int8_t d1, d2 # integrator buffers (d2 for it2.15)
#int8_t v # sample value
state = ReadBitsState() # state for it_readbits
log = logging.getLogger("pyitcompress.it_decompress8")
startpos = srcbuf.tell()
#log.debug("startpos = %d" % (startpos,))
# now unpack data till the dest buffer is full
while (len):
# read a new block of compressed data and reset variables
# block layout: word size, <size> bytes data
# removed: error handling when data is truncated
if not srcbuf.read(2):
return
state.bitbuf = state.bitnum = 0
blklen = MIN(0x8000, len)
blkpos = 0
#log.debug("new block, len = %d", blklen)
width = 9 # start with width of 9 bits
d1 = d2 = 0 # reset integrator buffers
# now uncompress the data block
while (blkpos < blklen):
#log.debug("while2: blkpos = %d, blklen = %d" % (blkpos, blklen))
value = it_readbits(width, state, srcbuf)
if (width < 7):
# method 1 (1-6 bits)
#log.debug("method 1: width=0x%x, value=0x%x" % (width, value))
# check for "100..."
if (value == 1 << (width - 1)):
# yes!
value = it_readbits(3, state, srcbuf) + 1 # read new width
width = value if (value < width) else value + 1 # and expand it
continue # ... next value
elif (width < 9):
# method 2 (7-8 bits)
#log.debug("method 2: width=0x%x, value=0x%x" % (width, value))
border = (0xFF >> (9 - width)) - 4 # lower border for width chg
if (value > border and value <= (border + 8)):
value -= border # convert width to 1-8
width = value if (value < width) else value + 1 # and expand it
continue # ... next value
elif (width == 9):
# method 3 (9 bits)
# bit 8 set?
#log.debug("method 3: width=0x%x, value=0x%x" % (width, value))
if (value & 0x100):
width = (value + 1) & 0xff # new width...
continue # ... and next value
else:
# illegal width, abort
log.error("Illegal width")
return
# now expand value to signed byte
if (width < 8):
shift = 8 - width
v = signbyte((value << shift) & 0xff)
v >>= shift
v = (v & 0xff)
else:
v = value & 0xff
# integrate upon the sample values
d1 = (d1 + v) & 0xff
d2 = (d2 + d1) & 0xff
# .. and store it into the buffer
dest.write(chr(d2) if it215 else chr(d1))
blkpos += 1
# now subtract block length from total length and go on
len -= blklen
compressed_len = srcbuf.tell() - startpos
#log.debug("size of compressed data: %d bytes" % (compressed_len))
return compressed_len
def it_decompress16(dest, len, srcbuf, it215):
"""
dest: (file-like object) output buffer for decompressed data
len: number of samples
srcbuf: (file-like object) input
it215: (bool) use it215 algorithm
"""
#const uint8_t *filebuf # source buffer containing compressed sample data
#const uint8_t *srcbuf # current position in source buffer
#int16_t *destpos # position in destination buffer which will be returned
#uint16_t blklen # length of compressed data block in samples
#uint16_t blkpos # position in block
#uint8_t width # actual "bit width"
#uint32_t value # value read from file to be processed
#int16_t d1, d2 # integrator buffers (d2 for it2.15)
#int16_t v # sample value
state = ReadBitsState() # state for it_readbits
log = logging.getLogger("pyitcompress.it_decompress16")
startpos = srcbuf.tell()
#log.debug("startpos = %d" % (startpos,))
# now unpack data till the dest buffer is full
while (len):
# read a new block of compressed data and reset variables
# block layout: word size, <size> bytes data
# removed: error handling when data is truncated
if not srcbuf.read(2):
return
state.bitbuf = state.bitnum = 0
blklen = MIN(0x4000, len)
blkpos = 0
width = 17 # start with width of 17 bits
d1 = d2 = 0 # reset integrator buffers
# now uncompress the data block
while (blkpos < blklen):
value = it_readbits(width, state, srcbuf)
if (width < 7):
# method 1 (1-6 bits)
# check for "100..."
if (value == 1 << (width - 1)):
# yes!
value = it_readbits(4, state, srcbuf) + 1 # read new width
width = value if (value < width) else value + 1 # and expand it
continue # ... next value
elif (width < 17):
# method 2 (7-17 bits)
border = (0xFFFF >> (17 - width)) - 8 # lower border for width chg
if (value > border and value <= (border + 16)):
value -= border # convert width to 1-8
width = value if (value < width) else value + 1 # and expand it
continue # ... next value
elif (width == 17):
# method 3 (17 bits)
# bit 8 set?
if (value & 0x10000):
width = (value + 1) & 0xff # new width...
continue # ... and next value
else:
# illegal width, abort
log.error("Illegal width")
return
# now expand value to signed byte
if (width < 16):
shift = 16 - width
v = signword((value << shift) & 0xffff)
v >>= shift
else:
v = (value & 0xffff)
# integrate upon the sample values
d1 = (d1 + v) & 0xffff
d2 = (d2 + d1) & 0xffff
# .. and store it into the buffer
outval = d2 if it215 else d1
dest.write(chr(outval & 0xff))
dest.write(chr(unsignbyte(outval >> 8)))
blkpos += 1
# now subtract block length from total length and go on
len -= blklen
compressed_len = srcbuf.tell() - startpos
#log.debug("size of compressed data: %d bytes" % (compressed_len))
return compressed_len
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment