src/librarian/builders/html.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 collections import defaultdict
   5 import os
   6 from urllib.request import urlopen
   7 from lxml import etree
   8 from librarian.html import add_table_of_contents, add_table_of_themes, add_image_sizes
   9 from librarian import OutputFile
  10
  11
  12 class TreeBuilder:
  13     @property
  14     def cursor(self):
  15         return self.current_cursors[-1]
  16
  17     def enter_fragment(self, fragment):
  18         cursor = self.cursors.get(fragment, self.cursor)
  19         self.current_cursors.append(cursor)
  20
  21     def exit_fragment(self):
  22         self.current_cursors.pop()
  23
  24     def create_fragment(self, name, element):
  25         assert name not in self.cursors
  26         self.cursors[name] = element
  27
  28     def forget_fragment(self, name):
  29         del self.cursors[name]
  30
  31     def start_element(self, tag, attrib=None):
  32         self.current_cursors.append(etree.SubElement(
  33             self.cursor,
  34             tag,
  35             **(attrib or {})
  36         ))
  37
  38     def end_element(self):
  39         self.current_cursors.pop()
  40
  41     def push_text(self, text):
  42         cursor = self.cursor
  43         if len(cursor):
  44             cursor[-1].tail = (cursor[-1].tail or '') + text
  45         else:
  46             cursor.text = (cursor.text or '') + text
  47
  48
  49 class HtmlBuilder(TreeBuilder):
  50     build_method_fn = 'html_build'
  51     file_extension = "html"
  52     with_themes = True
  53     with_toc = True
  54     with_footnotes = True
  55     with_nota_red = True
  56     with_ids = True
  57     with_numbering = True
  58     no_externalities = False
  59     orphans = True
  60
  61     root_tag = 'div'
  62     root_attrib = {'id': 'book-text'}
  63
  64     def __init__(self, gallery_path=None, gallery_url=None, base_url=None):
  65         self._base_url = base_url
  66         self.gallery_path = gallery_path
  67         self.gallery_url = gallery_url
  68
  69         self.tree = text = etree.Element(self.root_tag, **self.root_attrib)
  70         self.header = etree.Element('h1')
  71
  72         self.footnotes = etree.Element('div', id='footnotes')
  73
  74         self.nota_red = etree.Element('div', id='nota_red')
  75
  76         self.cursors = {
  77             None: text,
  78             'header': self.header,
  79             'footnotes': self.footnotes,
  80             'nota_red': self.nota_red,
  81         }
  82         self.current_cursors = [text]
  83
  84     @property
  85     def base_url(self):
  86         if self._base_url is not None:
  87             return self._base_url
  88         else:
  89             return 'https://wolnelektury.pl/media/book/pictures/{}/'.format(self.document.meta.url.slug)
  90
  91     def build(self, document, element=None, **kwargs):
  92         self.document = document
  93         self.document.assign_ids()
  94         self.prepare_images()
  95
  96         if element is None:
  97             element = document.tree.getroot()
  98
  99         element.html_build(self)
 100         self.postprocess(document)
 101         return self.output()
 102
 103     def prepare_images(self):
 104         # Temporarily use the legacy method, before transitioning to external generators.
 105         if self.gallery_path is None:
 106             return
 107         try:
 108             os.makedirs(self.gallery_path)
 109         except OSError:
 110             pass
 111         add_image_sizes(self.document.tree, self.gallery_path, self.gallery_url, self.base_url)
 112
 113     def output(self):
 114         if not len(self.tree):
 115             return None
 116         return OutputFile.from_bytes(
 117             etree.tostring(
 118                 self.tree,
 119                 method='html',
 120                 encoding='utf-8',
 121                 pretty_print=True
 122             )
 123         )
 124
 125     def postprocess(self, document):
 126         _ = document.tree.getroot().gettext
 127
 128         if document.meta.translators:
 129             self.enter_fragment('header')
 130             self.start_element('span', {'class': 'translator'})
 131             self.push_text(_("translated by") + " ")
 132             self.push_text(
 133                 ", ".join(
 134                     translator.readable()
 135                     for translator in document.meta.translators
 136                 )
 137             )
 138             self.exit_fragment()
 139
 140         if len(self.header):
 141             self.tree.insert(0, self.header)
 142
 143         if self.with_nota_red and len(self.nota_red):
 144             self.tree.append(self.nota_red)
 145         if self.with_themes:
 146             add_table_of_themes(self.tree)
 147         if self.with_toc:
 148             add_table_of_contents(self.tree)
 149
 150         if len(self.footnotes):
 151             fnheader = etree.Element("h3")
 152             fnheader.text = _("Footnotes")
 153             self.footnotes.insert(0, fnheader)
 154             self.tree.append(self.footnotes)
 155
 156     def add_visible_number(self, element):
 157         assert '_id' in element.attrib, etree.tostring(element)
 158         self.start_element('a', {
 159             'href': f'#{element.attrib["_id"]}',
 160             'class': 'wl-num',
 161         })
 162         self.push_text(element.attrib['_visible_numbering'])
 163         self.end_element()
 164
 165
 166 class StandaloneHtmlBuilder(HtmlBuilder):
 167     css_url = "https://static.wolnelektury.pl/css/compressed/book_text.css"
 168
 169     def postprocess(self, document):
 170         super(StandaloneHtmlBuilder, self).postprocess(document)
 171
 172         tree = etree.Element('html')
 173         body = etree.SubElement(tree, 'body')
 174         body.append(self.tree)
 175         self.tree = tree
 176
 177         head = etree.Element('head')
 178         tree.insert(0, head)
 179
 180         etree.SubElement(head, 'meta', charset='utf-8')
 181         etree.SubElement(head, 'title').text = document.meta.title
 182
 183         etree.SubElement(
 184             head,
 185             'meta',
 186             name="viewport",
 187             content="width=device-width, initial-scale=1, maximum-scale=1"
 188         )
 189
 190         if self.no_externalities:
 191             etree.SubElement(
 192                 head, 'style',
 193             ).text = urlopen(self.css_url).read().decode('utf-8')
 194         else:
 195             etree.SubElement(
 196                 head,
 197                 'link',
 198                 href=self.css_url,
 199                 rel="stylesheet",
 200                 type="text/css",
 201             )
 202
 203             etree.SubElement(
 204                 body, 'script',
 205                 src="https://ajax.googleapis.com/ajax/libs/jquery/1/jquery.min.js"
 206             )
 207
 208             etree.SubElement(
 209                 body,
 210                 "script",
 211                 src="http://malsup.github.io/min/jquery.cycle2.min.js"
 212             )
 213
 214
 215 class SnippetHtmlBuilder(HtmlBuilder):
 216     with_themes = False
 217     with_toc = False
 218     with_footnotes = False
 219     with_nota_red = False
 220     with_ids = False
 221     with_numbering = False
 222
 223
 224 class AbstraktHtmlBuilder(HtmlBuilder):
 225     with_themes = False
 226     with_toc = False
 227     with_footnotes = False
 228     with_nota_red = False
 229     with_ids = False
 230     with_numbering = False
 231
 232     root_tag = 'blockquote'
 233     root_attrib = {}
 234
 235     def build(self, document, element=None, **kwargs):
 236         if element is None:
 237             element = document.tree.find('//abstrakt')
 238         if element is None:
 239             return OutputFile.from_bytes(b'')
 240         element.attrib['_force'] = '1'
 241         return super().build(document, element, **kwargs)
 242
 243
 244 class DaisyHtmlBuilder(StandaloneHtmlBuilder):
 245     file_extension = 'xhtml'
 246     with_themes = False
 247     with_toc = False
 248     with_footnotes = False
 249     with_nota_red = False
 250     with_deep_identifiers = False
 251     no_externalities = True
 252     with_numbering = False
 253
 254     def output(self):
 255         tree = etree.ElementTree(self.tree)
 256         tree.docinfo.public_id = '-//W3C//DTD XHTML 1.0 Transitional//EN'
 257         tree.docinfo.system_url = 'http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd'
 258         return OutputFile.from_bytes(
 259             etree.tostring(
 260                 tree,
 261                 encoding='utf-8',
 262                 pretty_print=True,
 263                 xml_declaration=True
 264             )
 265         )
 266