src/librarian/builders/html.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 try:
   5     from urllib.request import urlopen
   6 except ImportError:
   7     from urllib2 import urlopen
   8 from lxml import etree
   9 from librarian.html import add_anchors, add_table_of_contents, add_table_of_themes
  10 from librarian import OutputFile
  11
  12
  13 class HtmlBuilder:
  14     file_extension = "html"
  15     with_anchors = True
  16     with_themes = True
  17     with_toc = True
  18     with_footnotes = True
  19     with_nota_red = True
  20     no_externalities = False
  21     orphans = True
  22
  23     def __init__(self, base_url=None):
  24         self._base_url = base_url
  25
  26         self.tree = text = etree.Element('div', **{'id': 'book-text'})
  27         self.header = etree.Element('h1')
  28
  29         self.footnotes = etree.Element('div', id='footnotes')
  30         self.footnote_counter = 0
  31
  32         self.nota_red = etree.Element('div', id='nota_red')
  33
  34         self.cursors = {
  35             None: text,
  36             'header': self.header,
  37             'footnotes': self.footnotes,
  38             'nota_red': self.nota_red,
  39         }
  40         self.current_cursors = [text]
  41
  42     @property
  43     def base_url(self):
  44         if self._base_url is not None:
  45             return self._base_url
  46         else:
  47             return 'https://wolnelektury.pl/media/book/pictures/{}/'.format(self.document.meta.url.slug)
  48
  49     @property
  50     def cursor(self):
  51         return self.current_cursors[-1]
  52
  53     def enter_fragment(self, fragment):
  54         self.current_cursors.append(self.cursors[fragment])
  55
  56     def exit_fragment(self):
  57         self.current_cursors.pop()
  58
  59     def create_fragment(self, name, element):
  60         assert name not in self.cursors
  61         self.cursors[name] = element
  62
  63     def forget_fragment(self, name):
  64         del self.cursors[name]
  65
  66     def preprocess(self, document):
  67         document._compat_assign_ordered_ids()
  68         document._compat_assign_section_ids()
  69
  70     def build(self, document, **kwargs):
  71         self.document = document
  72
  73         self.preprocess(document)
  74         document.tree.getroot().html_build(self)
  75         self.postprocess(document)
  76         return self.output()
  77
  78     def output(self):
  79         return OutputFile.from_bytes(
  80             etree.tostring(
  81                 self.tree,
  82                 method='html',
  83                 encoding='utf-8',
  84                 pretty_print=True
  85             )
  86         )
  87
  88     def postprocess(self, document):
  89         _ = document.tree.getroot().master.gettext
  90
  91         if document.meta.translators:
  92             self.enter_fragment('header')
  93             self.start_element('span', {'class': 'translator'})
  94             self.push_text(_("translated by") + " ")
  95             self.push_text(
  96                 ", ".join(
  97                     translator.readable()
  98                     for translator in document.meta.translators
  99                 )
 100             )
 101             self.exit_fragment()
 102
 103         if self.with_anchors:
 104             add_anchors(self.tree)
 105         if self.with_nota_red and len(self.nota_red):
 106             self.tree.append(self.nota_red)
 107         if self.with_themes:
 108             add_table_of_themes(self.tree)
 109         if self.with_toc:
 110             add_table_of_contents(self.tree)
 111
 112         if len(self.header):
 113             self.tree.insert(0, self.header)
 114
 115         if self.footnote_counter:
 116             fnheader = etree.Element("h3")
 117             fnheader.text = _("Footnotes")
 118             self.footnotes.insert(0, fnheader)
 119             self.tree.append(self.footnotes)
 120
 121     def start_element(self, tag, attrib=None):
 122         self.current_cursors.append(etree.SubElement(
 123             self.cursor,
 124             tag,
 125             **(attrib or {})
 126         ))
 127
 128     def end_element(self):
 129         self.current_cursors.pop()
 130
 131     def push_text(self, text):
 132         cursor = self.cursor
 133         if len(cursor):
 134             cursor[-1].tail = (cursor[-1].tail or '') + text
 135         else:
 136             cursor.text = (cursor.text or '') + text
 137
 138
 139 class StandaloneHtmlBuilder(HtmlBuilder):
 140     css_url = "https://static.wolnelektury.pl/css/compressed/book_text.css"
 141
 142     def postprocess(self, document):
 143         super(StandaloneHtmlBuilder, self).postprocess(document)
 144
 145         tree = etree.Element('html')
 146         body = etree.SubElement(tree, 'body')
 147         body.append(self.tree)
 148         self.tree = tree
 149
 150         head = etree.Element('head')
 151         tree.insert(0, head)
 152
 153
 154         etree.SubElement(head, 'meta', charset='utf-8')
 155         etree.SubElement(head, 'title').text = document.meta.title
 156
 157         etree.SubElement(
 158             head,
 159             'meta',
 160             name="viewport",
 161             content="width=device-width, initial-scale=1, maximum-scale=1"
 162         )
 163
 164         if self.no_externalities:
 165             etree.SubElement(
 166                 head, 'style',
 167             ).text = urlopen(self.css_url).read().decode('utf-8')
 168         else:
 169             etree.SubElement(
 170                 head,
 171                 'link',
 172                 href=self.css_url,
 173                 rel="stylesheet",
 174                 type="text/css",
 175             )
 176
 177             etree.SubElement(
 178                 body, 'script',
 179                 src="https://ajax.googleapis.com/ajax/libs/jquery/1/jquery.min.js"
 180             )
 181
 182             etree.SubElement(
 183                 body,
 184                 "script",
 185                 src="http://malsup.github.io/min/jquery.cycle2.min.js"
 186             )
 187
 188
 189 class SnippetHtmlBuilder(HtmlBuilder):
 190     with_anchors = False
 191     with_themes = False
 192     with_toc = False
 193     with_footnotes = False
 194     with_nota_red = False
 195     with_refs = False
 196
 197
 198 class DaisyHtmlBuilder(StandaloneHtmlBuilder):
 199     file_extension = 'xhtml'
 200     with_anchors = False
 201     with_themes = False
 202     with_toc = False
 203     with_footnotes = False
 204     with_nota_red = False
 205     with_deep_identifiers = False
 206     no_externalities = True
 207
 208     def output(self):
 209         tree = etree.ElementTree(self.tree)
 210         tree.docinfo.public_id = '-//W3C//DTD XHTML 1.0 Transitional//EN'
 211         tree.docinfo.system_url = 'http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd'
 212         return OutputFile.from_bytes(
 213             etree.tostring(
 214                 tree,
 215                 encoding='utf-8',
 216                 pretty_print=True,
 217                 xml_declaration=True
 218             )
 219         )
 220