#
# reprocess.py
#
# (C) Jack Whitham 2009
# URL site update by David Hodgkinson (2024)
# Use with Python 2.4 or higher. Run without parameters for instructions.
# This program requires you the SoX tools, mpg123 and wget
#
# $Id: reprocess.py,v 1.2 2009/10/25 01:52:57 jack Exp jack $
#

import os, pickle, sys, shutil
HEADER = 0x30
REF_RATE = 1e5
SAMPLE_RATE = 44100
BYTES_PER_SAMPLE = 4
DUMP = True
DATA = "subdat"
BASE_URL = "http://www.hoth.uk/magrathea"
#BASE_URL = "http://www.jwhitham.org.uk/magrathea"
#BASE_URL = "http://localhost/magrathea"
BS = 65536

def Samples(x):
    return int(SAMPLE_RATE * x)

def Main():
    if ( 4 <= len(sys.argv) <= 5 ):
        return Normal_Mode(sys.argv)

    if ( len(sys.argv) > 1 ):
        if ( sys.argv[ 1 ] == 'diff' ):
            return Diff_Mode()
        elif ( sys.argv[ 1 ] == 'patch' ):
            return Patch_Mode()

    print "Usage 1 - patch:"
    print "   %s patch" % sys.argv[ 0 ]
    print "   Automatically patches episodes copied off your store-bought CDs"
    print "   to return them to the original broadcast version."
    print "   See %s/adiff.html for instructions." % BASE_URL
    print ""
    print "Usage 2 - align an audio file based on waypoint data"
    print "   %s <input audio file> <input matching file>" % sys.argv[ 0 ]
    print "         <output audio file>"
    print "   Used to generate files for a listening test, comparing"
    print "   off-air recordings to the CD recordings."
    return 1

def Grab_File(dir_name, file_name):
    print 'Downloading: %s' % file_name
    out = "copy_%s" % file_name
    rc = os.system("wget --no-verbose -c -O %s %s/%s/%s" % (
                out, BASE_URL, dir_name, file_name))
    if ((rc != 0)
    or (not os.path.exists(out))):
        print ""
        print "Failed to download file %s." % file_name
        sys.exit(1)

    return out

def Sort_Substitution(a, b):
    ((a_episode_nr, _), (a_begin, _, _)) = a
    ((b_episode_nr, _), (b_begin, _, _)) = b
    a = (a_episode_nr, a_begin)
    b = (b_episode_nr, b_begin)
    return cmp(a, b)
    
def Patch_Mode():
    subdat_file = Grab_File(".", DATA)
    try:
        data = pickle.load(file(subdat_file, "rb"))
        subdata = data[ 'subdata' ].items()
    except:
        print ""
        print "Error reading substitution data file."
        print "You probably need to download a new copy of this program."
        print "Please visit %s." % BASE_URL
        sys.exit(1)

    subdata.sort(cmp=Sort_Substitution)
    subdata.reverse()
    episodes = set()
    for ((episode_nr, dir_name), (begin, end, size)) in subdata:
        episodes.add(episode_nr)

    episode_name = dict()
    episodes = list(episodes)
    episodes.sort()
    for episode_nr in episodes:
        name = "%02d.wav" % episode_nr
        assert os.path.exists(name)
        episode_name[ episode_nr ] = name
       
    clip = SAMPLE_RATE * 2 # clip 2 seconds from the beginning and the end
    uid = 0
    for ((episode_nr, dir_name), (begin, end, size)) in subdata:
        print 'Processing episode %d' % episode_nr

        mp3 = Grab_File(dir_name, "%s-original.mp3" % dir_name)
        wav = "decoded_%s.wav" % dir_name

        print 'Decoding: %s' % mp3
        rc = os.system("mpg123 -q -w %s %s" % (wav, mp3))
        assert rc == 0

        begin += clip
        end -= clip
        size -= clip * 2

        name = "temp_%u.wav" % uid
        uid += 1

        print 'Patching: %s' % dir_name

        old_name = episode_name[ episode_nr ]
        cd_in = file(old_name, "rb")
        cd_out = file(name, "wb")
        patch2 = Gen_Header(cd_out)
        offair_in = file(wav, "rb")
        offair_in.seek(HEADER + (clip * BYTES_PER_SAMPLE)) 

        Copy(cd_in, cd_out, begin * BYTES_PER_SAMPLE)
        Copy(offair_in, cd_out, size * BYTES_PER_SAMPLE)
        cd_in.seek((end - begin) * BYTES_PER_SAMPLE, 1)
        Copy(cd_in, cd_out)
        Patch_Header(cd_out, patch2)
        cd_out.close()
        cd_in.close()
        offair_in.close()
        if (old_name.startswith("temp")):
            os.unlink(old_name)

        episode_name[ episode_nr ] = name

    for (episode_nr, name) in episode_name.iteritems():
        print 'Completed episode %d' % episode_nr
        os.rename(name, "%02d_original.wav" % episode_nr)

    print 'All done.'

def Copy(cd_in, cd_out, size=-1):
    while size != 0:
        if size > 0:
            to_copy = min(BS, size)
        else:
            to_copy = BS
        data = cd_in.read(to_copy)
        cd_out.write(data)
        if size > 0:
            size -= len(data)
        if (( to_copy == 0 )
        or ( len(data) != to_copy )):
            break
    
def Diff_Mode():
    def X(m, s, plus=0):
        return int(((m * 60.0) + s + plus) * SAMPLE_RATE * 2)

    e = Make_Patch("magrathea", "Missing Magrathea Scene", 3,
            86325083, 86587831, 10, 0)

    e = Make_Patch("mice", "Dinner with Frankie and Benjy Mouse", 4,
            2*52454879, 2*64349117, 6, 0)
    e = Make_Patch("micecredits", "Credits", 4,
            2*74214127, 2*76542003, 4, e)

    e = Make_Patch("allitnils", "Lintilla and the Allitnils", 12,
            X(4, 35.692), X(10, 20), 5, 0)
    e = Make_Patch("passengercabin", "Escape from the Passenger Cabin", 12,
            X(10, 58.148), X(10, 58.148, 27), 5, e)
    e = Make_Patch("autopilot", '"Return to your seats!"', 12,
            X(12, 53.585), X(12, 53.585, 30), 5, e)
    e = Make_Patch("autobiography", "Marvin's Autobiography", 12,
            X(14, 4.511), X(14, 4.511, 10), 5, e)
    e = Make_Patch("flashback", '"Reality is on the blink again"', 12,
            X(15, 44.619), X(15, 44.619, 20), 5, e)
    e = Make_Patch("shack", "The Man in the Shack", 12,
            X(19, 39), X(20, 10), 10, e)
    e = Make_Patch("ending", "The End", 12,
            X(24, 55.301), X(25, 12.155, 11), 10, e)


def Make_Patch(dir_name, english_name, 
            episode_nr, begin, end, margin_size, offset):

    print dir_name

    relation_list = Get_Relation_List("tracks/%02d_lineup.txt" % episode_nr)
    gradient = Calc_Gradient(relation_list)
    cd_locations = (begin / 2, end / 2)
    offair_locations = [
            Translate(relation_list, x, False) for x in cd_locations ]

    cd_locations = Expand(cd_locations, margin_size)
    offair_locations = Expand(offair_locations, margin_size * gradient)
    (begin, end) = offair_locations

    try:
        shutil.rmtree(dir_name)
    except:
        pass
    os.mkdir(dir_name)
    html = file("%s/info.html" % dir_name, "wt")

    html.write('<table align=left border bgcolor="#c0c0c0">')
    html.write('<tr><td><table>')
    html.write("<tr><th colspan=7 align=center>%s"
            " (episode %u)</th></tr>" % (english_name, episode_nr))
    html.write("<tr><td colspan=7>&nbsp;&nbsp;</td></tr>\n\n")

    offair_begin = offair_locations[ 0 ] / gradient
    cd_begin = cd_locations[ 0 ] * 1.0
    correction = cd_begin - offair_begin

    size = []

    for (colour, symbol, version, f, locations, c) in [ 
                ("red", "&lt;&lt;", "original", 
                    1.0 / gradient, offair_locations, correction + offset),
                ("blue", "&gt;&gt;", "CD", 
                    1.0, cd_locations, 0) ]:

        (begin, end) = locations 
        begin = ( begin * f ) + c 
        end = ( end * f ) + c
        size.append(end - begin)
        href = "%s/%s-%s.mp3" % (dir_name, dir_name, version.lower())
        html.write("<tr>"
            '<td>&nbsp;&nbsp;&nbsp;&nbsp;</td>'
            '<td><a href="%s"><font color=%s>%s</text> %s version</td></a>'
            '<td>&nbsp;&nbsp;&nbsp;&nbsp;</td>'
            '<td>Location: <a href="%s">%s - %s</a></td>'
            '<td>&nbsp;&nbsp;&nbsp;&nbsp;</td>'
            '<td>Length: %s</td>' 
            '<td>&nbsp;&nbsp;&nbsp;&nbsp;</td></tr>' % (
                href, colour, symbol, version,
                href, Get_Stamp(begin).strip(), Get_Stamp(end).strip(),
                Get_Stamp(end - begin).strip()))

    html.write("<tr><td colspan=7>&nbsp;&nbsp;</td></tr>")
    html.write("</table></td></tr></table>\n\n")
    html.close()
    offset += size[ 0 ] # original
    offset -= size[ 1 ] # CD

    try:
        data = pickle.load(file(DATA, "rb"))
    except:
        data = dict()
        data[ 'subdata' ] = dict()

    (begin, end) = offair_locations 
    size = int(( end - begin ) / gradient)
    data[ 'subdata' ][ (episode_nr, dir_name) ] = tuple(
            list(cd_locations) + [ size ])
    pickle.dump(data, file(DATA, "wb"))

    Extract_Clip("tracks/%02d_reference.wav" % episode_nr,
            "%s/%s-cd.wav" % (dir_name, dir_name), 
            cd_locations)
    Extract_Clip("tracks/%02d_offair.wav" % episode_nr,
            "%s/%s-original.wav" % (dir_name, dir_name), 
            offair_locations, gradient)
    for x in ("cd", "original"):
        rc = os.system("lame -h -V 3 %s/%s-%s.wav %s/%s-%s.mp3" % (
                dir_name, dir_name, x, dir_name, dir_name, x))
        assert rc == 0
        os.unlink("%s/%s-%s.wav" % (dir_name, dir_name, x))

    return offset

def Expand(locations, margin_seconds):
    (begin, end) = locations
    size = end - begin
    assert size > 0
    begin -= int(margin_seconds * SAMPLE_RATE)
    end += int(margin_seconds * SAMPLE_RATE)
    return (begin, end)
    
def Extract_Clip(input_audio, output_audio, locations, new_rate=1.0):
    offair_in = file(input_audio, "rb")
    offair_out = file(output_audio, "wb")

    (begin, end) = locations
    size = end - begin
    assert size > 0

    patch2 = Gen_Header(offair_out)
    offair_in.seek(begin * BYTES_PER_SAMPLE, 0)
    data = offair_in.read(size * BYTES_PER_SAMPLE)
    if new_rate != 1.0:
        data = Resample(data, new_rate)
    offair_out.write(data)
    Patch_Header(offair_out, patch2)

def Normal_Mode(argv):
    assert ( 4 <= len(argv) <= 5 )
    input_audio = argv[ 1 ]
    input_matching = argv[ 2 ]
    output_audio = argv[ 3 ]
    dump = None
    if len(argv) > 4:
        dump = file(argv[ 4 ], "rb")

    return Do_Work(input_audio=input_audio, 
            input_matching=input_matching, 
            output_audio=output_audio, 
            dump=dump)

def Calc_Gradient(relation_list):
    relation_list = relation_list[:]
    relation_list.reverse()
    vectorX = vectorY = 0
    for i in xrange(len(relation_list) - 1):
        (x1, y1) = relation_list[ i ]
        (x2, y2) = relation_list[ i + 1 ]
        assert x2 > x1
        assert y2 > y1
        sizeX = x2 - x1
        sizeY = y2 - y1
        if (( sizeX > ( SAMPLE_RATE * 10 ))
        and ( sizeY > ( SAMPLE_RATE * 10 ))):
            gradient = float(sizeY) / float(sizeX)
            if ( 0.96 < gradient < 1.04 ):
                vectorX += sizeX
                vectorY += sizeY

    return float(vectorY) / float(vectorX)
        
def Translate(relation_list, sample, reverse_direction):
    if reverse_direction:
        relation_list = [ (y, x) for (x, y) in relation_list ] 
    else:
        relation_list = relation_list[:]

    relation_list.reverse()
    for i in xrange(len(relation_list) - 1):
        (x1, y1) = relation_list[ i ]
        (x2, y2) = relation_list[ i + 1 ]
        assert x2 > x1
        assert y2 > y1
        if (x1 <= sample < x2):
            interp = float(sample - x1) / float(x2 - x1)
            return (int(float(y2 - y1) * interp) + y1) & ~3
    
    return None # Not translatable
    
def Get_Relation_List(input_matching):
    cd0 = cd1 = offair0 = offair1 = 0
    relation_list = []

    for line in file(input_matching):
        (cd, offair) = line.split()
        cd0 = int(cd) / 2
        offair0 = int(offair) / 2

        if ( cd0 < 0 ):
            offair0 = offair1 - ( cd1 * ( offair1 - offair0 )) / ( cd1 - cd0 )
            cd0 = 0

        relation_list.append((cd0, offair0))
        cd1 = cd0
        offair1 = offair0

    return relation_list

def Do_Work(input_audio, input_matching, output_audio, dump):

    if ( os.path.exists(output_audio) ):
        print 'Output file already exists - refusing to overwrite'
        return 1

    relation_list = Get_Relation_List(input_matching)
    offair_in = file(input_audio, "rb")
    offair_out = file(output_audio, "wb")
    offair_in.read(HEADER)
    patch2 = Gen_Header(offair_out)

    
    (cd0, offair0) = relation_list.pop()
    if ( offair0 > 0 ):
        offair_in.read(offair0 * BYTES_PER_SAMPLE)
    if ( cd0 > 0 ):
        offair_out.write('\0' * (cd0 * BYTES_PER_SAMPLE))
    if ( dump != None ):
        dump.read(HEADER)
        dump.read(cd0 * BYTES_PER_SAMPLE)

    error = 0
    serial = 1
    previous_rate = SAMPLE_RATE

    while ( len(relation_list) != 0 ):    
        tstamp = Get_Stamp(cd0).strip()
        cd1 = cd0
        offair1 = offair0
        (cd0, offair0) = relation_list.pop()
        print '%7.2e  %7.2e ' % (offair0 - offair1, cd0 - cd1),
      
        reference = None
        if dump != None:
            reference = dump.read((cd0 - cd1) * BYTES_PER_SAMPLE)

        data = offair_in.read((offair0 - offair1) * BYTES_PER_SAMPLE)
        new_rate = 1.0 / (float(cd0 - cd1) / 
                float(offair0 - offair1))
        edit = 100.0 * (( new_rate ) - 1.0 )
        print '%10.2f' % ( new_rate * REF_RATE ),
        if ((dump != None) and not (-3.0 < edit < 3.0)):
            temp1 = file("temp1.raw", "wb")
            temp1.write(data)
            temp1.close()
            temp3 = file("temp3.raw", "wb")
            temp3.write(reference)
            temp3.close()
            cmd = ("sox -r %1.0f -c 2 -s -2 -t raw temp3.raw "
                "%03d_reference_%s.wav" % (SAMPLE_RATE, serial, tstamp))
            rc = os.system(cmd)
            assert rc == 0
            cmd = ("sox -r %1.0f -c 2 -s -2 -t raw temp1.raw -r %1.0f "
                "%03d_offair_%s.wav" % (previous_rate, SAMPLE_RATE,
                                            serial, tstamp))
            rc = os.system(cmd)
            assert rc == 0
            serial += 1
        else:
            previous_rate = SAMPLE_RATE / (float(cd0 - cd1) / 
                float(offair0 - offair1))

        data = Resample(data, new_rate)
        cd0 = (len(data) / BYTES_PER_SAMPLE) + cd1
        offair_out.write(data)

        tstamp = Get_Stamp(cd0)
        print '%s  %4.1f%%' % (tstamp, edit)

    temp1 = file("temp1.raw", "wb")
    temp1.write(offair_in.read())
    temp1.close()
    rc = os.system(cmd)
    assert rc == 0
    temp2 = file("temp2.raw", "rb")
    data = temp2.read()
    temp2.close()
    offair_out.write(data)
    print len(data), 'extra bytes'

    Patch_Header(offair_out, patch2)

    offair_out.close()
    return 0

def Resample(data, new_rate):
    temp1 = file("temp1.raw", "wb")
    temp1.write(data)
    temp1.close()
    cmd = ("sox -r %1.0f -c 2 -s -2 -t raw temp1.raw "
        "-r %1.0f temp2.raw" % (REF_RATE * new_rate, REF_RATE))
    rc = os.system(cmd)
    assert rc == 0
    temp2 = file("temp2.raw", "rb")
    data = temp2.read()
    temp2.close()
    return data

def Get_Stamp(cd0):
    if type(cd0) == float:
        cd0 = int(cd0)
    assert type(cd0) == int
    ms = ( cd0 * 1000 ) / SAMPLE_RATE
    s = ms / 1000
    m = s / 60
    ms %= 1000
    s %= 60
    return '%5d:%02d.%03d' % (m, s, ms)

def Gen_Header(offair_out):
    offair_out.write("RIFF----WAVEfmt "
        "\x10\x00\x00\x00\x01\x00\x02\x00")
    for x in (SAMPLE_RATE, SAMPLE_RATE * 4):
        for y in xrange(0, 32, 8):
            offair_out.write(chr(( int(x) >> y ) & 0xff))
    offair_out.write("\x04\x00\x10\x00data----")
    patch2 = offair_out.tell() - 4
    return patch2

def Patch_Header(offair_out, patch2):
    for (p, x) in [ (4, offair_out.tell() - 8),
                (patch2, offair_out.tell() - 44) ]:
        offair_out.seek(p, 0)
        for y in xrange(0, 32, 8):
            offair_out.write(chr(( int(x) >> y ) & 0xff))
    

if ( __name__ == "__main__" ):
    sys.exit(Main())