librarian/formats/epub/__init__.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 import os
   7 import re
   8 import urllib
   9 from copy import deepcopy
  10 from mimetypes import guess_type
  11 from tempfile import NamedTemporaryFile
  12 import zipfile
  13 from urllib2 import urlopen
  14
  15 from lxml import etree
  16 from librarian import OPFNS, NCXNS, XHTMLNS, DCNS, BuildError
  17 from librarian import core
  18 from librarian.formats import Format
  19 from librarian.formats.cover.evens import EvensCover
  20 from librarian.output import OutputFile
  21 from librarian.renderers import Register, TreeRenderer, UnknownElement
  22 from librarian.utils import Context, get_resource, extend_element
  23
  24
  25 class EpubFormat(Format):
  26     format_name = 'EPUB'
  27     format_ext = 'epub'
  28
  29     cover = EvensCover
  30     renderers = Register()
  31
  32     def __init__(self, doc, cover=None, with_fonts=True):
  33         super(EpubFormat, self).__init__(doc)
  34         self.with_fonts = with_fonts
  35         if cover is not None:
  36             self.cover = cover
  37
  38     def dc(self, tag, multiple=False):
  39         if multiple:
  40             return ', '.join(self.doc.meta.get(DCNS(tag)))
  41         else:
  42             return self.doc.meta.get_one(DCNS(tag))
  43
  44     def build(self, ctx=None):
  45
  46         def add_file(url, file_id):
  47             filename = url.rsplit('/', 1)[1]
  48             if url.startswith('file://'):
  49                 url = ctx.files_path + urllib.quote(url[7:])
  50             if url.startswith('/'):
  51                 url = 'http://milpeer.eu' + url
  52             file_content = urlopen(url).read()
  53             zip.writestr(os.path.join('OPS', filename), file_content)
  54             manifest.append(etree.fromstring(
  55                 '<item id="%s" href="%s" media-type="%s" />' % (file_id, filename, guess_type(url)[0])))
  56
  57         opf = etree.parse(get_resource('formats/epub/res/content.opf'))
  58         manifest = opf.find(OPFNS('manifest'))
  59         guide = opf.find(OPFNS('guide'))
  60         spine = opf.find(OPFNS('spine'))
  61
  62         author = ", ". join(self.doc.meta.get(DCNS('creator')) or [])
  63         title = self.doc.meta.title()
  64         opf.find('.//' + DCNS('creator')).text = author
  65         opf.find('.//' + DCNS('title')).text = title
  66
  67         output_file = NamedTemporaryFile(prefix='librarian', suffix='.epub', delete=False)
  68         zip = zipfile.ZipFile(output_file, 'w', zipfile.ZIP_DEFLATED)
  69
  70         mime = zipfile.ZipInfo()
  71         mime.filename = 'mimetype'
  72         mime.compress_type = zipfile.ZIP_STORED
  73         mime.extra = ''
  74         zip.writestr(mime, 'application/epub+zip')
  75         zip.writestr('META-INF/container.xml', '<?xml version="1.0" ?><container version="1.0" '
  76                      'xmlns="urn:oasis:names:tc:opendocument:xmlns:container">'
  77                      '<rootfiles><rootfile full-path="OPS/content.opf" '
  78                      'media-type="application/oebps-package+xml" />'
  79                      '</rootfiles></container>')
  80
  81         toc_file = etree.fromstring('<?xml version="1.0" encoding="utf-8"?><!DOCTYPE ncx PUBLIC '
  82                                     '"-//NISO//DTD ncx 2005-1//EN" "http://www.daisy.org/z3986/2005/ncx-2005-1.dtd">'
  83                                     '<ncx xmlns="http://www.daisy.org/z3986/2005/ncx/" xml:lang="pl" '
  84                                     'version="2005-1"><head></head><docTitle></docTitle><navMap>'
  85                                     '</navMap></ncx>')
  86         # nav_map = toc_file[-1]
  87
  88         if self.cover is not None:
  89             # cover_image = self.doc.meta.get(DCNS('relation.coverimage.url'))[0]
  90             cover = self.cover(self.doc)
  91             cover.set_images(ctx)
  92             cover_output = cover.build()
  93             cover_name = 'cover.%s' % cover.format_ext
  94             zip.writestr(os.path.join('OPS', cover_name), cover_output.get_string())
  95             del cover_output
  96
  97             cover_tree = etree.parse(get_resource('formats/epub/res/cover.html'))
  98             cover_tree.find('//' + XHTMLNS('img')).set('src', cover_name)
  99             zip.writestr('OPS/cover.html', etree.tostring(
 100                             cover_tree, method="html", pretty_print=True))
 101
 102             if cover.uses_dc_cover:
 103                 if self.doc.meta.get_one('cover_by'):
 104                     self.doc.edoc.getroot().set('data-cover-by', self.doc.meta.get_one('cover_by'))
 105                 if self.doc.meta.get_one('cover_source'):
 106                     self.doc.edoc.getroot().set('data-cover-source', self.doc.meta.get_one('cover_source'))
 107
 108             manifest.append(etree.fromstring(
 109                 '<item id="cover" href="cover.html" media-type="application/xhtml+xml" />'))
 110             manifest.append(etree.fromstring(
 111                 '<item id="cover-image" href="%s" media-type="%s" />' % (cover_name, cover.mime_type())))
 112             spine.insert(0, etree.fromstring('<itemref idref="cover" linear="no" />'))
 113             opf.getroot()[0].append(etree.fromstring('<meta name="cover" content="cover-image"/>'))
 114             guide.append(etree.fromstring('<reference href="cover.html" type="cover" title="Okładka"/>'))
 115
 116         if not ctx:
 117             ctx = Context(format=self)
 118         else:
 119             ctx.format = self
 120         ctx.toc = TOC()
 121         ctx.toc_level = 0
 122         ctx.footnotes = Footnotes()
 123         ctx.images = []
 124         ctx.part_no = 0
 125
 126         wrap_tmpl = etree.parse(get_resource('formats/epub/res/chapter.html'))
 127         for e in self.render(self.doc.edoc.getroot(), ctx):
 128             if not len(e) and not (e.text and e.text.strip()):
 129                 continue
 130             wrap = deepcopy(wrap_tmpl)
 131             extend_element(wrap.find('//*[@id="book-text"]'), e)
 132
 133             partstr = 'part%d' % int(e.get('part_no'))
 134             manifest.append(manifest.makeelement(OPFNS('item'), attrib={
 135                                  'id': partstr,
 136                                  'href': partstr + ".html",
 137                                  'media-type': 'application/xhtml+xml',
 138                              }))
 139             spine.append(spine.makeelement(OPFNS('itemref'), attrib={
 140                         'idref': partstr,
 141                     }))
 142             zip.writestr('OPS/%s.html' % partstr, etree.tostring(wrap, method='html'))
 143
 144         for i, url in enumerate(ctx.images):
 145             add_file(url, 'image%s' % i)
 146
 147         if len(ctx.footnotes.output):
 148             ctx.toc.add("Przypisy", "footnotes.html")
 149             manifest.append(etree.Element(
 150                 OPFNS('item'), id='footnotes', href='footnotes.html',
 151                 **{'media-type': "application/xhtml+xml"}))
 152             spine.append(etree.Element('itemref', idref='footnotes'))
 153             wrap = etree.parse(get_resource('formats/epub/res/footnotes.html'))
 154             extend_element(wrap.find('//*[@id="footnotes"]'), ctx.footnotes.output)
 155
 156             # chars = chars.union(used_chars(html_tree.getroot()))
 157             zip.writestr('OPS/footnotes.html', etree.tostring(
 158                                 wrap, method="html", pretty_print=True))
 159
 160         footer_text = [
 161             'Information about the resource',
 162             'Publisher: %s' % self.dc('publisher'),
 163             'Rights: %s' % self.dc('rights'),
 164             'Intended audience: %s' % self.dc('audience', multiple=True),
 165             self.dc('description'),
 166             'Resource prepared using MIL/PEER editing platform.',
 167             'Source available at %s' % ctx.source_url,
 168         ]
 169         footer_wrap = deepcopy(wrap_tmpl)
 170         footer_body = footer_wrap.find('//*[@id="book-text"]')
 171         for line in footer_text:
 172             footer_line = etree.Element('p')
 173             footer_line.text = line
 174             footer_body.append(footer_line)
 175         manifest.append(manifest.makeelement(OPFNS('item'), attrib={
 176             'id': 'footer',
 177             'href': "footer.html",
 178             'media-type': 'application/xhtml+xml',
 179         }))
 180         spine.append(spine.makeelement(OPFNS('itemref'), attrib={
 181             'idref': 'footer',
 182         }))
 183         zip.writestr('OPS/footer.html', etree.tostring(footer_wrap, method='html'))
 184
 185         zip.writestr('OPS/content.opf', etree.tostring(opf, pretty_print=True))
 186         ctx.toc.render(toc_file[-1])
 187         zip.writestr('OPS/toc.ncx', etree.tostring(toc_file, pretty_print=True))
 188         zip.close()
 189         return OutputFile.from_filename(output_file.name)
 190
 191     def render(self, element, ctx):
 192         return self.renderers.get_for(element).render(element, ctx)
 193
 194
 195 # Helpers
 196
 197 class EpubRenderer(TreeRenderer):
 198     """ Renders insides as XML in a <_/> container. """
 199     def container(self, ctx):
 200         root, inner = super(EpubRenderer, self).container()
 201         root.set("part_no", str(ctx.part_no))
 202         return root, inner
 203
 204     def render(self, element, ctx):
 205         subctx = self.subcontext(element, ctx)
 206         wrapper, inside = self.container(ctx)
 207         if element.text:
 208             extend_element(inside, self.render_text(element.text, ctx))
 209         for child in element:
 210             try:
 211                 child_renderer = ctx.format.renderers.get_for(child)
 212             except UnknownElement:
 213                 continue
 214             else:
 215                 if getattr(child_renderer, 'epub_separate', False):
 216                     yield wrapper
 217                     ctx.part_no += 1
 218                     for child_part in child_renderer.render(child, subctx):
 219                         yield child_part
 220                     wrapper, inside = self.container(ctx)
 221                 else:
 222                     child_parts = list(child_renderer.render(child, subctx))
 223                     extend_element(inside, child_parts[0])
 224                     if len(child_parts) > 1:
 225                         yield wrapper
 226                         for child_part in child_parts[1:-1]:
 227                             yield child_part
 228                         wrapper, inside = self.container(ctx)
 229                         extend_element(inside, child_parts[-1])
 230             finally:
 231                 if child.tail:
 232                     extend_element(inside, self.render_text(child.tail, ctx))
 233         yield wrapper
 234
 235
 236 class NaturalText(EpubRenderer):
 237     def render_text(self, text, ctx):
 238         root, inner = self.text_container()
 239         chunks = re.split('(?<=\s\w) ', text)
 240         inner.text = chunks[0]
 241         for chunk in chunks[1:]:
 242             x = etree.Entity("nbsp")
 243             x.tail = chunk
 244             inner.append(x)
 245         return root
 246
 247
 248 class Silent(EpubRenderer):
 249     def render_text(self, text, ctx):
 250         root, inner = self.text_container()
 251         return root
 252
 253
 254 class Footnotes(object):
 255     def __init__(self):
 256         self.counter = 0
 257         self.output = etree.Element("_")
 258
 259     def append(self, items):
 260         self.counter += 1
 261         e = etree.Element(
 262             "a", href="part%d.html#footnote-anchor-%d" % (int(items[0].get('part_no')), self.counter),
 263             id="footnote-%d" % self.counter,
 264             style="float:left;margin-right:1em")
 265         e.text = "[%d]" % self.counter
 266         e.tail = " "
 267         self.output.append(e)
 268         for item in items:
 269             extend_element(self.output, item)
 270         anchor = etree.Element(
 271             "a", href="footnotes.html#footnote-%d" % self.counter, id="footnote-anchor-%d" % self.counter)
 272         anchor.text = "[%d]" % self.counter
 273         return anchor
 274
 275
 276 class TOC(object):
 277     def __init__(self, title=None, href="", root=None):
 278         if root is None:
 279             self.counter = 0
 280             self.root = self
 281         else:
 282             self.root = root
 283         self.children = []
 284         self.title = title
 285         self.href = href.format(counter=self.root.counter)
 286         self.number = self.root.counter
 287         self.root.counter += 1
 288
 289     def add(self, title, href):
 290         subtoc = type(self)(title, href, root=self.root)
 291         self.children.append(subtoc)
 292         return subtoc
 293
 294     def render(self, nav_map):
 295         for child in self.children:
 296             nav_point = etree.Element(NCXNS('navPoint'))
 297             nav_point.set('id', 'NavPoint-%d' % child.number)
 298             nav_point.set('playOrder', str(child.number))
 299
 300             nav_label = etree.Element(NCXNS('navLabel'))
 301             text = etree.Element(NCXNS('text'))
 302             text.text = child.title
 303             nav_label.append(text)
 304             nav_point.append(nav_label)
 305
 306             content = etree.Element(NCXNS('content'))
 307             content.set('src', child.href)
 308             nav_point.append(content)
 309             nav_map.append(nav_point)
 310             child.render(nav_point)
 311
 312
 313 # Renderers
 314
 315 class AsideR(NaturalText):
 316     def render(self, element, ctx):
 317         outputs = list(super(AsideR, self).render(element, ctx))
 318         anchor = ctx.footnotes.append(outputs)
 319         wrapper, inside = self.text_container()  # etree.Element('_', part_no=str(ctx.part_no))
 320         inside.append(anchor)
 321         yield wrapper
 322 EpubFormat.renderers.register(core.Aside, None, AsideR('div'))
 323
 324 EpubFormat.renderers.register(core.Aside, 'comment', Silent())
 325
 326
 327 class DivR(NaturalText):
 328     def container(self, ctx):
 329         root, inner = super(DivR, self).container(ctx)
 330         if getattr(ctx, 'inline', False):
 331             inner.tag = 'span'
 332             inner.set('style', 'display: block;')
 333         return root, inner
 334 EpubFormat.renderers.register(core.Div, None, DivR('div'))
 335 EpubFormat.renderers.register(core.Div, 'p', NaturalText('p'))
 336
 337 EpubFormat.renderers.register(core.Div, 'list', NaturalText('ul'))
 338 EpubFormat.renderers.register(core.Div, 'list.enum', NaturalText('ol'))
 339 EpubFormat.renderers.register(core.Div, 'item', NaturalText('li'))
 340 EpubFormat.renderers.register(core.Span, 'item', NaturalText('li'))
 341
 342
 343 class DivImageR(EpubRenderer):
 344     def render(self, element, ctx):
 345         src = element.attrib.get('src', '')
 346         ctx.images.append(src)
 347         if '/' not in src:
 348             raise BuildError('Bad image URL')
 349         src = src.rsplit('/', 1)[1]
 350         return super(DivImageR, self).render(element, Context(ctx, src=src))
 351
 352     def container(self, ctx):
 353         root, inner = super(DivImageR, self).container(ctx)
 354         src = getattr(ctx, 'src', '')
 355         inner.set('src', src)
 356         # inner.set('style', 'display: block; width: 60%; margin: 3em auto')
 357         return root, inner
 358 EpubFormat.renderers.register(core.Div, 'img', DivImageR('img'))
 359
 360
 361 class DivVideoR(Silent):
 362     def render(self, element, ctx):
 363         src = 'https://www.youtube.com/watch?v=%s' % element.attrib.get('videoid', '')
 364         return super(DivVideoR, self).render(element, Context(ctx, src=src))
 365
 366     def container(self, ctx):
 367         root, inner = super(DivVideoR, self).container(ctx)
 368         src = getattr(ctx, 'src', '')
 369         link = etree.Element('a', {'href': src})
 370         link.text = src
 371         inner.append(link)
 372         return root, inner
 373 EpubFormat.renderers.register(core.Div, 'video', DivVideoR('p'))
 374
 375
 376 class HeaderR(NaturalText):
 377     def subcontext(self, element, ctx):
 378         return Context(ctx, inline=True)
 379 EpubFormat.renderers.register(core.Header, None, HeaderR('h1'))
 380
 381
 382 class SectionR(NaturalText):
 383     epub_separate = True
 384
 385     def render(self, element, ctx):
 386         # Add 'poczatek'?
 387         if element.getparent() is not None:
 388             tocitem = ctx.toc.add(element.meta.title(), 'part%d.html' % ctx.part_no)
 389             ctx = Context(ctx, toc=tocitem)
 390         return super(SectionR, self).render(element, ctx)
 391 EpubFormat.renderers.register(core.Section, None, SectionR())
 392
 393
 394 class SpanR(NaturalText):
 395     pass
 396 EpubFormat.renderers.register(core.Span, None, SpanR('span'))
 397 EpubFormat.renderers.register(core.Span, 'cite', SpanR('i'))
 398 EpubFormat.renderers.register(core.Span, 'emp', SpanR('b'))
 399 EpubFormat.renderers.register(core.Span, 'emph', SpanR('i'))
 400
 401
 402 class SpanLink(EpubRenderer):
 403     def render(self, element, ctx):
 404         parts = super(SpanLink, self).render(element, ctx)
 405         for part in parts:
 406             src = element.attrib.get('href', '')
 407             if src.startswith('file://'):
 408                 src = ctx.files_path + src[7:]
 409             part[0].attrib['href'] = src
 410             yield part
 411 EpubFormat.renderers.register(core.Span, 'link', SpanLink('a'))