src/librarian/builders/daisy.py

   1 # This file is part of Librarian, licensed under GNU Affero GPLv3 or later.
   2 # Copyright © Fundacja Wolne Lektury. See NOTICE for more information.
   3 #
   4 from copy import deepcopy
   5 import subprocess
   6 import tempfile
   7 import zipfile
   8 from lxml import etree
   9 import mutagen
  10 from librarian import OutputFile, get_resource
  11 from librarian.html import raw_printable_text
  12 from .html import DaisyHtmlBuilder
  13
  14
  15 def get_duration(path):
  16     return float(
  17         subprocess.run(
  18             [
  19                 "ffprobe",
  20                 "-i",
  21                 path,
  22                 "-show_entries",
  23                 "format=duration",
  24                 "-v",
  25                 "quiet",
  26                 "-of",
  27                 "csv=p=0",
  28             ],
  29             capture_output=True,
  30             text=True,
  31             check=True,
  32         ).stdout
  33     )
  34
  35
  36 def format_hms(t):
  37     seconds = t % 60
  38     t -= seconds
  39     t /= 60
  40     minutes = t % 60
  41     t -= minutes
  42     hours = t / 60
  43     return "%02d:%02d:%02.3f" % (hours, minutes, seconds)
  44
  45
  46 def populate(element, context):
  47     if element.text:
  48         element.text = element.text.format(**context)
  49     if element.tail:
  50         element.tail = element.tail.format(**context)
  51     for k, v in element.attrib.items():
  52         element.attrib[k] = v.format(**context)
  53     for child in element:
  54         populate(child, context)
  55
  56
  57 class DaisyBuilder:
  58     file_extension = 'daisy.zip'
  59
  60     def build(self, document, mp3, split_on=None, **kwargs):
  61         if not mp3:
  62             raise ValueError("Need MP3 files")
  63
  64         outfile = tempfile.NamedTemporaryFile(delete=False)
  65         zipf = zipfile.ZipFile(outfile, 'w')
  66
  67         directory = document.meta.url.slug + '/'
  68
  69         if split_on:
  70             documents = []
  71             headers = []
  72             present = True
  73             n = 0
  74             while present:
  75                 present = False
  76                 n += 1
  77                 newdoc = deepcopy(document)
  78                 newdoc.tree.getroot().document = newdoc
  79
  80                 master = newdoc.tree.getroot()[-1]
  81                 i = 0
  82                 for item in list(master):
  83                     if item.tag == split_on:
  84                         # TODO: clear
  85                         i += 1
  86                         if i == n:
  87                             headers.append(raw_printable_text(item))
  88                     if i != n and not (n == 1 and not i):
  89                         master.remove(item)
  90                     else:
  91                         present = True
  92                 if present:
  93                     documents.append(newdoc)
  94         else:
  95             documents = [document]
  96             headers = [document.meta.title]
  97
  98         assert len(documents) == len(mp3)
  99
 100         narrator = mutagen.File(mp3[0]).get('TPE1')
 101         narrator = narrator.text[0] if narrator else ''
 102
 103         durations = []
 104         for i, part in enumerate(documents):
 105             print('part', i)
 106             html = DaisyHtmlBuilder().build(part)
 107             zipf.write(
 108                 html.get_filename(),
 109                 directory + 'book%d.html' % i,
 110             )
 111
 112             durations.append(get_duration(mp3[i]))
 113             zipf.write(
 114                 mp3[i],
 115                 directory + "book%d.mp3" % i,
 116             )
 117
 118             config_string = "task_language=pol|is_text_type=unparsed|is_text_unparsed_id_regex=sec\\d+$|is_text_unparsed_id_sort=numeric|os_task_file_format=tab"
 119             task = Task(config_string=config_string)
 120
 121             with tempfile.TemporaryDirectory() as temp:
 122                 syncfile = temp + "/sync"
 123                 task.audio_file_path_absolute = mp3[i]
 124                 task.text_file_path_absolute = html.get_filename()
 125                 task.sync_map_file_path_absolute = syncfile
 126
 127                 ExecuteTask(task).execute()
 128                 task.output_sync_map_file()
 129
 130                 sync = []
 131                 with open(syncfile) as f:
 132                     for line in f:
 133                         start, end, sec = line.strip().split('\t')
 134                         start = float(start)
 135                         end = float(end)
 136                         sync.append([start, end, sec])
 137
 138             hms = format_hms(durations[i])
 139             elapsed_hms = format_hms(sum(durations[:i]))
 140
 141             context = {
 142                 "VERSION": "1.10",
 143
 144                 "HHMMSSmmm": hms,
 145                 "HHMMSS": hms.split('.')[0],
 146                 "Sd": "%.1f" % durations[i],
 147                 "ELAPSED": elapsed_hms,
 148
 149                 "TITLE": document.meta.title,
 150                 "PUBLISHER": document.meta.publisher[0],
 151                 "YEAR": document.meta.created_at[:4],
 152                 "MONTH": document.meta.created_at[5:7],
 153                 "AUTHOR": document.meta.author.readable(),
 154
 155                 "NARRATOR": narrator,
 156             }
 157
 158             with open(get_resource('res/daisy/content.smil')) as f:
 159                 tree = etree.parse(f)
 160             populate(tree.getroot(), context)
 161
 162             seq = tree.find('//seq')
 163             for si, item in enumerate(sync):
 164                 par = etree.SubElement(seq, 'par', id="par%06d" % (si + 1), endsync="last")
 165                 etree.SubElement(
 166                     par,
 167                     "text",
 168                     src="book%d.html#%s" % (i, item[2]))
 169
 170                 audio = etree.SubElement(
 171                     par,
 172                     "audio",
 173                     src="book%d.mp3" % i,
 174                     **{
 175                         "clip-begin": "npt=%.3fs" % item[0],
 176                         "clip-end": "npt=%.3fs" % item[1],
 177                     }
 178                 )
 179
 180             zipf.writestr(
 181                 directory + 'content%d.smil' % i,
 182                 etree.tostring(
 183                     tree,
 184                     xml_declaration=True,
 185                     pretty_print=True,
 186                 ),
 187             )
 188
 189         for fname in ('smil10.dtd', 'xhtml1-transitional.dtd', 'xhtml-lat1.ent', 'xhtml-special.ent', 'xhtml-symbol.ent'):
 190             zipf.write(
 191                 get_resource('res/daisy/' + fname),
 192                 directory + fname)
 193
 194         duration = sum(durations)
 195         hms = format_hms(duration)
 196         context = {
 197             "VERSION": "1.10",
 198             "HHMMSSmmm": hms,
 199             "HHMMSS": hms.split('.')[0],
 200             "Sd": "%.1f" % duration,
 201             "TITLE": document.meta.title,
 202             "PUBLISHER": document.meta.publisher[0],
 203             "YEAR": document.meta.created_at[:4],
 204             "MONTH": document.meta.created_at[5:7],
 205             "AUTHOR": document.meta.author.readable(),
 206             "NARRATOR": narrator,
 207         }
 208
 209         tree = etree.parse(get_resource('res/daisy/er_book_info.xml'))
 210         cont = tree.getroot()[0]
 211         for i, dur in enumerate(durations):
 212             etree.SubElement(cont, 'smil', nr=str(i+1), Name="content%i.smil" % i, dur="%.1f" % dur)
 213         zipf.writestr(
 214             directory + 'er_book_info.xml',
 215             etree.tostring(tree, xml_declaration=True))
 216
 217         tree = etree.parse(get_resource('res/daisy/master.smil'))
 218         populate(tree.getroot(), context)
 219         cont = tree.getroot()[-1]
 220         for i, header in enumerate(headers):
 221             etree.SubElement(cont, 'ref', title=header, src="content%d.smil#seq000001" % i, id="smil_%04d" % i)
 222         zipf.writestr(
 223             directory + 'master.smil',
 224             etree.tostring(tree, xml_declaration=True))
 225
 226         tree = etree.parse(get_resource('res/daisy/ncc.html'))
 227         populate(tree.getroot(), context)
 228         cont = tree.getroot()[-1]
 229         for i, header in enumerate(headers):
 230             if not i:
 231                 h1 = etree.SubElement(
 232                     cont, 'h1', id='content', **{"class": "title"})
 233                 etree.SubElement(
 234                     h1, "a", href='content%d.smil#par000001' % i).text = document.meta.title
 235             else:
 236                 h2 = etree.SubElement(
 237                     cont, 'h2', id='content', **{"class": "chapter"})
 238                 etree.SubElement(
 239                     h2, "a", href='content%d.smil#par000001' % i).text = header
 240
 241         zipf.writestr(
 242             directory + 'ncc.html',
 243             etree.tostring(tree, xml_declaration=True))
 244
 245         zipf.close()
 246         return OutputFile.from_filename(outfile.name)