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     def simple_element(self, tag, text='', attrib=None):
  49         self.start_element(tag, attrib)
  50         self.push_text(text)
  51         self.end_element()
  52
  53
  54 class HtmlBuilder(TreeBuilder):
  55     build_method_fn = 'html_build'
  56     file_extension = "html"
  57     with_themes = True
  58     with_toc = True
  59     with_footnotes = True
  60     with_nota_red = True
  61     with_ids = True
  62     with_numbering = True
  63     no_externalities = False
  64     orphans = True
  65
  66     root_tag = 'div'
  67     root_attrib = {'id': 'book-text'}
  68
  69     def __init__(self, gallery_path=None, gallery_url=None, base_url=None):
  70         self._base_url = base_url
  71         self.gallery_path = gallery_path
  72         self.gallery_url = gallery_url
  73
  74         self.tree = text = etree.Element(self.root_tag, **self.root_attrib)
  75         self.header = etree.Element('h1')
  76
  77         self.footnotes = etree.Element('div', id='footnotes')
  78
  79         self.nota_red = etree.Element('div', id='nota_red')
  80
  81         self.cursors = {
  82             None: text,
  83             'header': self.header,
  84             'footnotes': self.footnotes,
  85             'nota_red': self.nota_red,
  86         }
  87         self.current_cursors = [text]
  88
  89     @property
  90     def base_url(self):
  91         if self._base_url is not None:
  92             return self._base_url
  93         else:
  94             return 'https://wolnelektury.pl/media/book/pictures/{}/'.format(self.document.meta.url.slug)
  95
  96     def build(self, document, element=None, **kwargs):
  97         self.document = document
  98         self.document.assign_ids()
  99         self.prepare_images()
 100
 101         if element is None:
 102             element = document.tree.getroot()
 103
 104         element.html_build(self)
 105         self.postprocess(document)
 106         return self.output()
 107
 108     def prepare_images(self):
 109         # Temporarily use the legacy method, before transitioning to external generators.
 110         if self.gallery_path is None:
 111             return
 112         try:
 113             os.makedirs(self.gallery_path)
 114         except OSError:
 115             pass
 116         add_image_sizes(self.document.tree, self.gallery_path, self.gallery_url, self.base_url)
 117
 118     def output(self):
 119         if not len(self.tree):
 120             return None
 121         return OutputFile.from_bytes(
 122             etree.tostring(
 123                 self.tree,
 124                 method='html',
 125                 encoding='utf-8',
 126                 pretty_print=True
 127             )
 128         )
 129
 130     def postprocess(self, document):
 131         _ = document.tree.getroot().gettext
 132
 133         if document.meta.translators:
 134             self.enter_fragment('header')
 135             self.start_element('span', {'class': 'translator'})
 136             self.push_text(_("translated by") + " ")
 137             self.push_text(
 138                 ", ".join(
 139                     translator.readable()
 140                     for translator in document.meta.translators
 141                 )
 142             )
 143             self.exit_fragment()
 144
 145         if len(self.header):
 146             self.tree.insert(0, self.header)
 147
 148         if self.with_nota_red and len(self.nota_red):
 149             self.tree.append(self.nota_red)
 150         if self.with_themes:
 151             add_table_of_themes(self.tree)
 152         if self.with_toc:
 153             add_table_of_contents(self.tree)
 154
 155         if len(self.footnotes):
 156             fnheader = etree.Element("h3")
 157             fnheader.text = _("Footnotes")
 158             self.footnotes.insert(0, fnheader)
 159             self.tree.append(self.footnotes)
 160
 161     def add_visible_number(self, element):
 162         assert '_id' in element.attrib, etree.tostring(element)
 163         self.start_element('a', {
 164             'href': f'#{element.attrib["_id"]}',
 165             'class': 'wl-num',
 166         })
 167         self.push_text(element.attrib['_visible_numbering'])
 168         self.end_element()
 169
 170
 171 class StandaloneHtmlBuilder(HtmlBuilder):
 172     css_url = "https://static.wolnelektury.pl/css/compressed/book_text.css"
 173
 174     def postprocess(self, document):
 175         super(StandaloneHtmlBuilder, self).postprocess(document)
 176
 177         tree = etree.Element('html')
 178         body = etree.SubElement(tree, 'body')
 179         body.append(self.tree)
 180         self.tree = tree
 181
 182         head = etree.Element('head')
 183         tree.insert(0, head)
 184
 185         etree.SubElement(head, 'meta', charset='utf-8')
 186         etree.SubElement(head, 'title').text = document.meta.title
 187
 188         etree.SubElement(
 189             head,
 190             'meta',
 191             name="viewport",
 192             content="width=device-width, initial-scale=1, maximum-scale=1"
 193         )
 194
 195         if self.no_externalities:
 196             etree.SubElement(
 197                 head, 'style',
 198             ).text = urlopen(self.css_url).read().decode('utf-8')
 199         else:
 200             etree.SubElement(
 201                 head,
 202                 'link',
 203                 href=self.css_url,
 204                 rel="stylesheet",
 205                 type="text/css",
 206             )
 207
 208             etree.SubElement(
 209                 body, 'script',
 210                 src="https://ajax.googleapis.com/ajax/libs/jquery/1/jquery.min.js"
 211             )
 212
 213             etree.SubElement(
 214                 body,
 215                 "script",
 216                 src="http://malsup.github.io/min/jquery.cycle2.min.js"
 217             )
 218
 219
 220 class SnippetHtmlBuilder(HtmlBuilder):
 221     with_themes = False
 222     with_toc = False
 223     with_footnotes = False
 224     with_nota_red = False
 225     with_ids = False
 226     with_numbering = False
 227
 228
 229 class AbstraktHtmlBuilder(HtmlBuilder):
 230     with_themes = False
 231     with_toc = False
 232     with_footnotes = False
 233     with_nota_red = False
 234     with_ids = False
 235     with_numbering = False
 236
 237     root_tag = 'blockquote'
 238     root_attrib = {}
 239
 240     def build(self, document, element=None, **kwargs):
 241         if element is None:
 242             element = document.tree.find('//abstrakt')
 243         if element is None:
 244             return OutputFile.from_bytes(b'')
 245         element.attrib['_force'] = '1'
 246         return super().build(document, element, **kwargs)
 247
 248
 249 class DaisyHtmlBuilder(StandaloneHtmlBuilder):
 250     file_extension = 'xhtml'
 251     with_themes = False
 252     with_toc = False
 253     with_footnotes = False
 254     with_nota_red = False
 255     with_deep_identifiers = False
 256     no_externalities = True
 257     with_numbering = False
 258
 259     def output(self):
 260         tree = etree.ElementTree(self.tree)
 261         tree.docinfo.public_id = '-//W3C//DTD XHTML 1.0 Transitional//EN'
 262         tree.docinfo.system_url = 'http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd'
 263         return OutputFile.from_bytes(
 264             etree.tostring(
 265                 tree,
 266                 encoding='utf-8',
 267                 pretty_print=True,
 268                 xml_declaration=True
 269             )
 270         )
 271