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