src/librarian/builders/html.py

   1 from lxml import etree
   2 from librarian import OutputFile
   3
   4
   5 class HtmlBuilder:
   6     file_extension = "html"
   7     identifier = "html"
   8
   9     def __init__(self, image_location='https://wolnelektury.pl/media/book/pictures/marcos-historia-kolorow/'):
  10         self.image_location = image_location
  11
  12         #self.tree = etree.Element('html')
  13         #body = etree.SubElement(self.tree, 'body')
  14         #text = etree.SubElement(body, 'div', **{'id': 'book-text'})
  15         self.tree = text = etree.Element('div', **{'id': 'book-text'})
  16         toc = etree.SubElement(text, 'div', id='toc')
  17         themes = etree.SubElement(text, 'div', id='themes')
  18         h1 = etree.SubElement(text, 'h1')
  19
  20         self.cursors = {
  21             None: text,
  22             'toc': toc,
  23             'themes': themes,
  24             'header': h1,
  25         }
  26         self.current_cursors = [None]
  27
  28     def enter_fragment(self, fragment):
  29         self.current_cursors.append(fragment)
  30
  31     def exit_fragment(self):
  32         self.current_cursors.pop()
  33
  34     def build(self, document):
  35         document.tree.getroot().html_build(self)
  36
  37         head = etree.Element('head')
  38         self.tree.insert(0, head)
  39         etree.SubElement(
  40             head,
  41             'link',
  42             href="https://static.wolnelektury.pl/css/compressed/book_text.b15153e56c0a.css",
  43             rel="stylesheet",
  44             type="text/css",
  45         )
  46
  47         return OutputFile.from_bytes(
  48             etree.tostring(
  49                 self.tree,
  50                 method='html',
  51                 encoding='utf-8',
  52                 pretty_print=True
  53             )
  54         )
  55
  56     def start_element(self, tag, attrib):
  57         self.cursors[self.current_cursors[-1]] = etree.SubElement(
  58             self.cursors[self.current_cursors[-1]],
  59             tag,
  60             **attrib
  61         )
  62         print(self.cursors)
  63
  64     def end_element(self):
  65         self.cursors[self.current_cursors[-1]] = self.cursors[self.current_cursors[-1]].getparent()
  66
  67     def push_text(self, text):
  68         cursor = self.cursors[self.current_cursors[-1]]
  69         if len(cursor):
  70             cursor.tail = (cursor[-1].tail or '') + text
  71         else:
  72             cursor.text = (cursor.text or '') + text