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         cursor = self.cursors.get(fragment, self.cursor)
  55         self.current_cursors.append(cursor)
  56
  57     def exit_fragment(self):
  58         self.current_cursors.pop()
  59
  60     def create_fragment(self, name, element):
  61         assert name not in self.cursors
  62         self.cursors[name] = element
  63
  64     def forget_fragment(self, name):
  65         del self.cursors[name]
  66
  67     def preprocess(self, document):
  68         document._compat_assign_ordered_ids()
  69         document._compat_assign_section_ids()
  70
  71     def build(self, document, **kwargs):
  72         self.document = document
  73
  74         self.preprocess(document)
  75         document.tree.getroot().html_build(self)
  76         self.postprocess(document)
  77         return self.output()
  78
  79     def output(self):
  80         return OutputFile.from_bytes(
  81             etree.tostring(
  82                 self.tree,
  83                 method='html',
  84                 encoding='utf-8',
  85                 pretty_print=True
  86             )
  87         )
  88
  89     def postprocess(self, document):
  90         _ = document.tree.getroot().master.gettext
  91
  92         if document.meta.translators:
  93             self.enter_fragment('header')
  94             self.start_element('span', {'class': 'translator'})
  95             self.push_text(_("translated by") + " ")
  96             self.push_text(
  97                 ", ".join(
  98                     translator.readable()
  99                     for translator in document.meta.translators
 100                 )
 101             )
 102             self.exit_fragment()
 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 len(self.header):
 114             self.tree.insert(0, self.header)
 115
 116         if self.footnote_counter:
 117             fnheader = etree.Element("h3")
 118             fnheader.text = _("Footnotes")
 119             self.footnotes.insert(0, fnheader)
 120             self.tree.append(self.footnotes)
 121
 122     def start_element(self, tag, attrib=None):
 123         self.current_cursors.append(etree.SubElement(
 124             self.cursor,
 125             tag,
 126             **(attrib or {})
 127         ))
 128
 129     def end_element(self):
 130         self.current_cursors.pop()
 131
 132     def push_text(self, text):
 133         cursor = self.cursor
 134         if len(cursor):
 135             cursor[-1].tail = (cursor[-1].tail or '') + text
 136         else:
 137             cursor.text = (cursor.text or '') + text
 138
 139
 140 class StandaloneHtmlBuilder(HtmlBuilder):
 141     css_url = "https://static.wolnelektury.pl/css/compressed/book_text.css"
 142
 143     def postprocess(self, document):
 144         super(StandaloneHtmlBuilder, self).postprocess(document)
 145
 146         tree = etree.Element('html')
 147         body = etree.SubElement(tree, 'body')
 148         body.append(self.tree)
 149         self.tree = tree
 150
 151         head = etree.Element('head')
 152         tree.insert(0, head)
 153
 154
 155         etree.SubElement(head, 'meta', charset='utf-8')
 156         etree.SubElement(head, 'title').text = document.meta.title
 157
 158         etree.SubElement(
 159             head,
 160             'meta',
 161             name="viewport",
 162             content="width=device-width, initial-scale=1, maximum-scale=1"
 163         )
 164
 165         if self.no_externalities:
 166             etree.SubElement(
 167                 head, 'style',
 168             ).text = urlopen(self.css_url).read().decode('utf-8')
 169         else:
 170             etree.SubElement(
 171                 head,
 172                 'link',
 173                 href=self.css_url,
 174                 rel="stylesheet",
 175                 type="text/css",
 176             )
 177
 178             etree.SubElement(
 179                 body, 'script',
 180                 src="https://ajax.googleapis.com/ajax/libs/jquery/1/jquery.min.js"
 181             )
 182
 183             etree.SubElement(
 184                 body,
 185                 "script",
 186                 src="http://malsup.github.io/min/jquery.cycle2.min.js"
 187             )
 188
 189
 190 class SnippetHtmlBuilder(HtmlBuilder):
 191     with_anchors = False
 192     with_themes = False
 193     with_toc = False
 194     with_footnotes = False
 195     with_nota_red = False
 196     with_refs = False
 197
 198
 199 class DaisyHtmlBuilder(StandaloneHtmlBuilder):
 200     file_extension = 'xhtml'
 201     with_anchors = False
 202     with_themes = False
 203     with_toc = False
 204     with_footnotes = False
 205     with_nota_red = False
 206     with_deep_identifiers = False
 207     no_externalities = True
 208
 209     def output(self):
 210         tree = etree.ElementTree(self.tree)
 211         tree.docinfo.public_id = '-//W3C//DTD XHTML 1.0 Transitional//EN'
 212         tree.docinfo.system_url = 'http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd'
 213         return OutputFile.from_bytes(
 214             etree.tostring(
 215                 tree,
 216                 encoding='utf-8',
 217                 pretty_print=True,
 218                 xml_declaration=True
 219             )
 220         )
 221