librarian/epub.py

   1 # -*- coding: utf-8 -*-
   2 #
   3 # This file is part of Librarian, licensed under GNU Affero GPLv3 or later.
   4 # Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information.
   5 #
   6 from __future__ import with_statement
   7
   8 import os
   9 import os.path
  10 from copy import deepcopy
  11 from lxml import etree
  12 import zipfile
  13
  14 import sys
  15 sys.path.append('..') # for running from working copy
  16
  17 from librarian import XMLNamespace, RDFNS, DCNS, WLNS, NCXNS, OPFNS, NoDublinCore
  18 from librarian.dcparser import BookInfo
  19
  20
  21 def inner_xml(node):
  22     """ returns node's text and children as a string
  23
  24     >>> print inner_xml(etree.fromstring('<a>x<b>y</b>z</a>'))
  25     x<b>y</b>z
  26     """
  27
  28     nt = node.text if node.text is not None else ''
  29     return ''.join([nt] + [etree.tostring(child) for child in node])
  30
  31 def set_inner_xml(node, text):
  32     """ sets node's text and children from a string
  33
  34     >>> e = etree.fromstring('<a>b<b>x</b>x</a>')
  35     >>> set_inner_xml(e, 'x<b>y</b>z')
  36     >>> print etree.tostring(e)
  37     <a>x<b>y</b>z</a>
  38     """
  39
  40     p = etree.fromstring('<x>%s</x>' % text)
  41     node.text = p.text
  42     node[:] = p[:]
  43
  44
  45 def node_name(node):
  46     """ Find out a node's name
  47
  48     >>> print node_name(etree.fromstring('<a>X<b>Y</b>Z</a>'))
  49     XYZ
  50     """
  51
  52     tempnode = deepcopy(node)
  53
  54     for p in ('pe', 'pa', 'pt', 'pr', 'motyw'):
  55         for e in tempnode.findall('.//%s' % p):
  56             t = e.tail
  57             e.clear()
  58             e.tail = t
  59     etree.strip_tags(tempnode, '*')
  60     return tempnode.text
  61
  62
  63 def xslt(xml, sheet):
  64     if isinstance(xml, etree._Element):
  65         xml = etree.ElementTree(xml)
  66     with open(sheet) as xsltf:
  67         return xml.xslt(etree.parse(xsltf))
  68
  69
  70 _resdir = os.path.join(os.path.dirname(os.path.realpath(__file__)), 'epub')
  71 def res(fname):
  72     return os.path.join(_resdir, fname)
  73
  74
  75 def replace_characters(node):
  76     def replace_chars(text):
  77         if text is None:
  78             return None
  79         return text.replace("---", u"\u2014")\
  80                    .replace("--", u"\u2013")\
  81                    .replace(",,", u"\u201E")\
  82                    .replace('"', u"\u201D")\
  83                    .replace("'", u"\u2019")
  84     if node.tag == 'extra':
  85         node.clear()
  86     else:
  87         node.text = replace_chars(node.text)
  88         node.tail = replace_chars(node.tail)
  89         for child in node:
  90             replace_characters(child)
  91
  92
  93 def find_annotations(annotations, source, part_no):
  94     for child in source:
  95         if child.tag in ('pe', 'pa', 'pt', 'pr'):
  96             annotation = deepcopy(child)
  97             number = str(len(annotations)+1)
  98             annotation.set('number', number)
  99             annotation.set('part', str(part_no))
 100             annotation.tail = ''
 101             annotations.append(annotation)
 102             tail = child.tail
 103             child.clear()
 104             child.tail = tail
 105             child.text = number
 106         if child.tag not in ('extra', 'podtytul'):
 107             find_annotations(annotations, child, part_no)
 108
 109
 110 def replace_by_verse(tree):
 111     """ Find stanzas and create new verses in place of a '/' character """
 112
 113     stanzas = tree.findall('.//' + WLNS('strofa'))
 114     for node in stanzas:
 115         for child_node in node:
 116             if child_node.tag in ('slowo_obce', 'wyroznienie'):
 117                 foreign_verses = inner_xml(child_node).split('/\n')
 118                 if len(foreign_verses) > 1:
 119                     new_foreign = ''
 120                     for foreign_verse in foreign_verses:
 121                         if foreign_verse.startswith('<wers'):
 122                             new_foreign += foreign_verse
 123                         else:
 124                             new_foreign += ''.join(('<wers_normalny>', foreign_verse, '</wers_normalny>'))
 125                     set_inner_xml(child_node, new_foreign)
 126         verses = inner_xml(node).split('/\n')
 127         if len(verses) > 1:
 128             modified_inner_xml = ''
 129             for verse in verses:
 130                 if verse.startswith('<wers') or verse.startswith('<extra'):
 131                     modified_inner_xml += verse
 132                 else:
 133                     modified_inner_xml += ''.join(('<wers_normalny>', verse, '</wers_normalny>'))
 134             set_inner_xml(node, modified_inner_xml)
 135
 136
 137 def add_to_manifest(manifest, partno):
 138     """ Adds a node to the manifest section in content.opf file """
 139
 140     partstr = 'part%d' % partno
 141     e = manifest.makeelement(OPFNS('item'), attrib={
 142                                  'id': partstr,
 143                                  'href': partstr + '.html',
 144                                  'media-type': 'application/xhtml+xml',
 145                              })
 146     manifest.append(e)
 147
 148
 149 def add_to_spine(spine, partno):
 150     """ Adds a node to the spine section in content.opf file """
 151
 152     e = spine.makeelement(OPFNS('itemref'), attrib={'idref': 'part%d' % partno});
 153     spine.append(e)
 154
 155
 156 class TOC(object):
 157     def __init__(self, name=None, part_number=None):
 158         self.children = []
 159         self.name = name
 160         self.part_number = part_number
 161         self.sub_number = None
 162
 163     def add(self, name, part_number, level=0, is_part=True):
 164         if level > 0 and self.children:
 165             return self.children[-1].add(name, part_number, level-1, is_part)
 166         else:
 167             t = TOC(name)
 168             t.part_number = part_number
 169             self.children.append(t)
 170             if not is_part:
 171                 t.sub_number = len(self.children) + 1
 172                 return t.sub_number
 173
 174     def append(self, toc):
 175         self.children.append(toc)
 176
 177     def extend(self, toc):
 178         self.children.extend(toc.children)
 179
 180     def depth(self):
 181         if self.children:
 182             return max((c.depth() for c in self.children)) + 1
 183         else:
 184             return 0
 185
 186     def write_to_xml(self, nav_map, counter):
 187         for child in self.children:
 188             nav_point = nav_map.makeelement(NCXNS('navPoint'))
 189             nav_point.set('id', 'NavPoint-%d' % counter)
 190             nav_point.set('playOrder', str(counter))
 191
 192             nav_label = nav_map.makeelement(NCXNS('navLabel'))
 193             text = nav_map.makeelement(NCXNS('text'))
 194             text.text = child.name
 195             nav_label.append(text)
 196             nav_point.append(nav_label)
 197
 198             content = nav_map.makeelement(NCXNS('content'))
 199             src = 'part%d.html' % child.part_number
 200             if child.sub_number is not None:
 201                 src += '#sub%d' % child.sub_number
 202             content.set('src', src)
 203             nav_point.append(content)
 204             nav_map.append(nav_point)
 205             counter = child.write_to_xml(nav_point, counter + 1)
 206         return counter
 207
 208
 209 def chop(main_text):
 210     """ divide main content of the XML file into chunks """
 211
 212     # prepare a container for each chunk
 213     part_xml = etree.Element('utwor')
 214     etree.SubElement(part_xml, 'master')
 215     main_xml_part = part_xml[0] # master
 216
 217     last_node_part = False
 218     for one_part in main_text:
 219         name = one_part.tag
 220         if name == 'naglowek_czesc':
 221             yield part_xml
 222             last_node_part = True
 223             main_xml_part[:] = [deepcopy(one_part)]
 224         elif not last_node_part and name in ("naglowek_rozdzial", "naglowek_akt", "srodtytul"):
 225             yield part_xml
 226             main_xml_part[:] = [deepcopy(one_part)]
 227         else:
 228             main_xml_part.append(deepcopy(one_part))
 229             last_node_part = False
 230     yield part_xml
 231
 232
 233 def transform_chunk(chunk_xml, chunk_no, annotations):
 234     """ transforms one chunk, returns a HTML string and a TOC object """
 235
 236     toc = TOC()
 237     for element in chunk_xml[0]:
 238         if element.tag in ("naglowek_czesc", "naglowek_rozdzial", "naglowek_akt", "srodtytul"):
 239             toc.add(node_name(element), chunk_no)
 240         elif element.tag in ('naglowek_podrozdzial', 'naglowek_scena'):
 241             subnumber = toc.add(node_name(element), chunk_no, level=1, is_part=False)
 242             element.set('sub', str(subnumber))
 243     find_annotations(annotations, chunk_xml, chunk_no)
 244     replace_by_verse(chunk_xml)
 245     output_html = etree.tostring(xslt(chunk_xml, res('xsltScheme.xsl')), pretty_print=True)
 246     return output_html, toc
 247
 248
 249 def transform(provider, slug, output_file=None, output_dir=None):
 250     """ produces an epub
 251
 252     provider is a DocProvider
 253     either output_file (a file-like object) or output_dir (path to file/dir) should be specified
 254     if output_dir is specified, file will be written to <output_dir>/<author>/<slug>.epub
 255     """
 256
 257     def transform_file(input_xml, chunk_counter=1, first=True):
 258         """ processes one input file and proceeds to its children """
 259
 260         children = [child.text for child in input_xml.findall('.//'+DCNS('relation.hasPart'))]
 261
 262         # every input file will have a TOC entry,
 263         # pointing to starting chunk
 264         toc = TOC(node_name(input_xml.find('.//'+DCNS('title'))), chunk_counter)
 265         if first:
 266             # write book title page
 267             zip.writestr('OPS/title.html',
 268                  etree.tostring(xslt(input_xml, res('xsltTitle.xsl')), pretty_print=True))
 269         elif children:
 270             # write title page for every parent
 271             zip.writestr('OPS/part%d.html' % chunk_counter,
 272                 etree.tostring(xslt(input_xml, res('xsltChunkTitle.xsl')), pretty_print=True))
 273             add_to_manifest(manifest, chunk_counter)
 274             add_to_spine(spine, chunk_counter)
 275             chunk_counter += 1
 276
 277         if len(input_xml.getroot()) > 1:
 278             # rdf before style master
 279             main_text = input_xml.getroot()[1]
 280         else:
 281             # rdf in style master
 282             main_text = input_xml.getroot()[0]
 283             if main_text.tag == RDFNS('RDF'):
 284                 main_text = None
 285
 286         if main_text is not None:
 287             replace_characters(main_text)
 288
 289             for chunk_no, chunk_xml in enumerate(chop(main_text), chunk_counter):
 290                 chunk_html, chunk_toc = transform_chunk(chunk_xml, chunk_counter, annotations)
 291                 toc.extend(chunk_toc)
 292                 zip.writestr('OPS/part%d.html' % chunk_counter, chunk_html)
 293                 add_to_manifest(manifest, chunk_counter)
 294                 add_to_spine(spine, chunk_counter)
 295                 chunk_counter += 1
 296
 297         if children:
 298             for child in children:
 299                 child_xml = etree.parse(provider.by_uri(child))
 300                 child_toc, chunk_counter = transform_file(child_xml, chunk_counter, first=False)
 301                 toc.append(child_toc)
 302
 303         return toc, chunk_counter
 304
 305     # read metadata from the first file
 306     input_xml = etree.parse(provider[slug])
 307     metadata = input_xml.find('.//'+RDFNS('Description'))
 308     if metadata is None:
 309         raise NoDublinCore('Document has no DublinCore - which is required.')
 310     book_info = BookInfo.from_element(input_xml)
 311     metadata = etree.ElementTree(metadata)
 312
 313     # if output to dir, create the file
 314     if output_dir is not None:
 315         author = unicode(book_info.author)
 316         author_dir = os.path.join(output_dir, author)
 317         try:
 318             os.makedirs(author_dir)
 319         except OSError:
 320             pass
 321         output_file = open(os.path.join(author_dir, '%s.epub' % slug), 'w')
 322
 323
 324     zip = zipfile.ZipFile(output_file, 'w', zipfile.ZIP_DEFLATED)
 325
 326     # write static elements
 327     mime = zipfile.ZipInfo()
 328     mime.filename = 'mimetype'
 329     mime.compress_type = zipfile.ZIP_STORED
 330     mime.extra = ''
 331     zip.writestr(mime, 'application/epub+zip')
 332     zip.writestr('META-INF/container.xml', '<?xml version="1.0" ?><container version="1.0" ' \
 333                        'xmlns="urn:oasis:names:tc:opendocument:xmlns:container">' \
 334                        '<rootfiles><rootfile full-path="OPS/content.opf" ' \
 335                        'media-type="application/oebps-package+xml" />' \
 336                        '</rootfiles></container>')
 337     for fname in 'style.css', 'logo_wolnelektury.png':
 338         zip.write(res(fname), os.path.join('OPS', fname))
 339
 340     opf = xslt(metadata, res('xsltContent.xsl'))
 341     manifest = opf.find('.//' + OPFNS('manifest'))
 342     spine = opf.find('.//' + OPFNS('spine'))
 343
 344     annotations = etree.Element('annotations')
 345
 346     toc_file = etree.fromstring('<?xml version="1.0" encoding="utf-8"?><!DOCTYPE ncx PUBLIC ' \
 347                                '"-//NISO//DTD ncx 2005-1//EN" "http://www.daisy.org/z3986/2005/ncx-2005-1.dtd">' \
 348                                '<ncx xmlns="http://www.daisy.org/z3986/2005/ncx/" xml:lang="pl" ' \
 349                                'version="2005-1"><head></head><docTitle></docTitle><navMap>' \
 350                                '<navPoint id="NavPoint-1" playOrder="1"><navLabel>' \
 351                                '<text>Strona tytułowa</text></navLabel><content src="title.html" />' \
 352                                '</navPoint><navPoint id="NavPoint-2" playOrder="2"><navLabel>' \
 353                                '<text>Początek utworu</text></navLabel><content src="part1.html" />' \
 354                                '</navPoint></navMap></ncx>')
 355     nav_map = toc_file[-1]
 356
 357     toc, chunk_counter = transform_file(input_xml)
 358     toc_counter = toc.write_to_xml(nav_map, 3) # we already have 2 navpoints
 359
 360     # Last modifications in container files and EPUB creation
 361     if len(annotations) > 0:
 362         nav_map.append(etree.fromstring(
 363             '<navPoint id="NavPoint-%(i)d" playOrder="%(i)d" ><navLabel><text>Przypisy</text>'\
 364             '</navLabel><content src="annotations.html" /></navPoint>' % {'i': toc_counter}))
 365         manifest.append(etree.fromstring(
 366             '<item id="annotations" href="annotations.html" media-type="application/xhtml+xml" />'))
 367         spine.append(etree.fromstring(
 368             '<itemref idref="annotations" />'))
 369         replace_by_verse(annotations)
 370         zip.writestr('OPS/annotations.html', etree.tostring(
 371                             xslt(annotations, res("xsltAnnotations.xsl")), pretty_print=True))
 372
 373     zip.writestr('OPS/content.opf', etree.tostring(opf, pretty_print=True))
 374     contents = []
 375     title = node_name(etree.ETXPath('.//'+DCNS('title'))(input_xml)[0])
 376     attributes = "dtb:uid", "dtb:depth", "dtb:totalPageCount", "dtb:maxPageNumber"
 377     for st in attributes:
 378         meta = toc_file.makeelement(NCXNS('meta'))
 379         meta.set('name', st)
 380         meta.set('content', '0')
 381         toc_file[0].append(meta)
 382     toc_file[0][0].set('content', ''.join((title, 'WolneLektury.pl')))
 383     toc_file[0][1].set('content', str(toc.depth()))
 384     set_inner_xml(toc_file[1], ''.join(('<text>', title, '</text>')))
 385     zip.writestr('OPS/toc.ncx', etree.tostring(toc_file, pretty_print=True))
 386     zip.close()
 387
 388
 389 if __name__ == '__main__':
 390     from librarian import DirDocProvider
 391
 392     if len(sys.argv) < 2:
 393         print >> sys.stderr, 'Usage: python epub.py <input file>'
 394         sys.exit(1)
 395
 396     main_input = sys.argv[1]
 397     basepath, ext = os.path.splitext(main_input)
 398     path, slug = os.path.realpath(basepath).rsplit('/', 1)
 399     provider = DirDocProvider(path)
 400     transform(provider, slug, output_dir=path)
 401