librarian/formats/html/__init__.py

   1 # -*- coding: utf-8 -*-
   2 #
   3 # This file is part of Librarian, licensed under GNU Affero GPLv3 or later.
   4 # Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information.
   5 #
   6 import re
   7 from lxml import etree
   8 from librarian.formats import Format
   9 from librarian.output import OutputFile
  10 from librarian.renderers import Register, TreeRenderer
  11 from librarian.utils import Context, get_resource
  12 from librarian import core
  13
  14
  15 class HtmlFormat(Format):
  16     format_name = 'HTML'
  17     format_ext = 'html'
  18
  19     renderers = Register()
  20
  21     def __init__(self, doc, standalone=False):
  22         super(HtmlFormat, self).__init__(doc)
  23         self.standalone = standalone
  24
  25     def build(self):
  26         if self.standalone:
  27             tmpl = get_resource("formats/html/res/html_standalone.html")
  28         else:
  29             tmpl = get_resource("formats/html/res/html.html")
  30         t = etree.parse(tmpl)
  31
  32         ctx = Context(format=self)
  33         ctx.toc = TOC()
  34         ctx.toc_level = 0
  35         ctx.footnotes = Footnotes()
  36
  37         if self.standalone:
  38             t.find('head/title').text = u"%s (%s)" % (self.doc.meta.title(), self.doc.meta.author())
  39
  40         t.find('.//div[@id="content"]').extend(
  41             self.render(self.doc.edoc.getroot(), ctx))
  42         t.find('.//div[@id="toc"]').append(ctx.toc.render())
  43         t.find('.//div[@id="footnotes"]').extend(ctx.footnotes.output)
  44
  45         return OutputFile.from_string(etree.tostring(
  46             t, encoding='utf-8', method="html"))
  47
  48     def render(self, element, ctx):
  49         return self.renderers.get_for(element).render(element, ctx)
  50
  51
  52 # Helpers
  53
  54 class NaturalText(TreeRenderer):
  55     def render_text(self, text, ctx):
  56         root, inner = self.text_container()
  57         chunks = re.split('(?<=\s\w) ', text)
  58         inner.text = chunks[0]
  59         for chunk in chunks[1:]:
  60             x = etree.Entity("nbsp")
  61             x.tail = chunk
  62             inner.append(x)
  63         return root
  64
  65
  66 class LiteralText(TreeRenderer):
  67     pass
  68
  69
  70 class Footnotes(object):
  71     def __init__(self):
  72         self.counter = 0
  73         self.output = etree.Element("_")
  74
  75     def append(self, item):
  76         self.counter += 1
  77         e = etree.Element("a",
  78             href="#footnote-anchor-%d" % self.counter,
  79             id="footnote-%d" % self.counter,
  80             style="float:left;margin-right:1em")
  81         e.text = "[%d]" % self.counter
  82         e.tail = " "
  83         self.output.append(e)
  84         self.output.extend(item)
  85         anchor = etree.Element("a",
  86             id="footnote-anchor-%d" % self.counter,
  87             href="#footnote-%d" % self.counter)
  88         anchor.text = "[%d]" % self.counter
  89         return anchor
  90
  91
  92 class TOC(object):
  93     def __init__(self):
  94         self.items = []
  95         self.counter = 0
  96
  97     def add(self, title, level=0):
  98         self.counter += 1
  99         self.items.append((level, title, self.counter))
 100         return self.counter
 101
 102     def render(self):
 103         out = etree.Element("ul", id="toc")
 104         curr_level = 0
 105         cursor = out
 106         for level, title, counter in self.items:
 107             while level > curr_level:
 108                 ins = etree.Element("ul")
 109                 cursor.append(ins)
 110                 cursor = ins
 111                 curr_level += 1
 112             while level < curr_level:
 113                 cursor = cursor.getparent()
 114                 curr_level -= 1
 115             ins = etree.Element("li")
 116             ins.append(etree.Element("a", href="#sect%d" % counter))
 117             ins[0].text = title
 118             cursor.append(ins)
 119         return out
 120
 121
 122 # Renderers
 123
 124 HtmlFormat.renderers.register(core.Aside, None, NaturalText('aside'))
 125
 126 class AsideFootnote(NaturalText):
 127     def render(self, element, ctx):
 128         output = super(AsideFootnote, self).render(element, ctx)
 129         anchor = ctx.footnotes.append(output)
 130         root, inner = self.container()
 131         inner.append(anchor)
 132         return root
 133 HtmlFormat.renderers.register(core.Aside, 'footnote', AsideFootnote())
 134
 135
 136 HtmlFormat.renderers.register(core.Header, None, NaturalText('h1'))
 137
 138
 139 HtmlFormat.renderers.register(core.Div, None, NaturalText('div'))
 140 HtmlFormat.renderers.register(core.Div, 'item', NaturalText('li'))
 141 HtmlFormat.renderers.register(core.Div, 'list', NaturalText('ul'))
 142 HtmlFormat.renderers.register(core.Div, 'p', NaturalText('p'))
 143
 144
 145 class Section(NaturalText):
 146     def subcontext(self, element, ctx):
 147         return Context(ctx, toc_level=ctx.toc_level + 1)
 148
 149     def render(self, element, ctx):
 150         counter = ctx.toc.add(element.meta.title(), ctx.toc_level)
 151         root = super(Section, self).render(element, ctx)
 152         root[0].set("id", "sect%d" % counter)
 153         return root
 154 HtmlFormat.renderers.register(core.Section, None, Section('section'))
 155
 156
 157 HtmlFormat.renderers.register(core.Span, None, NaturalText('span'))
 158 HtmlFormat.renderers.register(core.Span, 'cite', NaturalText('cite'))
 159 HtmlFormat.renderers.register(core.Span, 'cite.code', LiteralText('code'))
 160 HtmlFormat.renderers.register(core.Span, 'emph', NaturalText('em'))
 161
 162 class SpanUri(LiteralText):
 163     def render(self, element, ctx):
 164         root = super(SpanUri, self).render(element, ctx)
 165         root[0].attrib['href'] = element.text
 166         return root
 167 HtmlFormat.renderers.register(core.Span, 'uri', SpanUri('a'))