1 # This file is part of Librarian, licensed under GNU Affero GPLv3 or later.
2 # Copyright © Fundacja Wolne Lektury. See NOTICE for more information.
6 from ebooklib import epub
8 from librarian.dcparser import Person
9 from librarian import get_resource
12 def _register_function(f):
13 """ Register extension function with lxml """
14 ns = etree.FunctionNamespace('http://wolnelektury.pl/functions')
18 def reg_substitute_entities():
19 entity_substitutions = [
27 def substitute_entities(context, text):
28 """XPath extension function converting all entites in passed text."""
29 if isinstance(text, list):
31 for entity, substitutution in entity_substitutions:
32 text = text.replace(entity, substitutution)
35 _register_function(substitute_entities)
39 def strip(context, text):
40 """Remove unneeded whitespace from beginning and end"""
41 if isinstance(text, list):
43 return re.sub(r'\s+', ' ', text).strip()
44 _register_function(strip)
47 def reg_starts_white():
48 def starts_white(context, text):
49 if isinstance(text, list):
53 return text[0].isspace()
54 _register_function(starts_white)
58 def ends_white(context, text):
59 if isinstance(text, list):
63 return text[-1].isspace()
64 _register_function(ends_white)
68 def wrap_words(context, text, wrapping):
70 XPath extension function automatically wrapping words
73 if isinstance(text, list):
78 words = re.split(r'\s', text)
83 line_length += len(word) + 1
84 if line_length > wrapping:
85 # Max line length was exceeded. We create new line
87 line_length = len(word)
88 lines[-1].append(word)
89 return '\n'.join(' '.join(line) for line in lines)
90 _register_function(wrap_words)
93 def reg_person_name():
94 def person_name(context, text):
95 """ Converts "Name, Forename" to "Forename Name" """
96 if isinstance(text, list):
98 return Person.from_text(text).readable()
99 _register_function(person_name)
102 def reg_texcommand():
103 def texcommand(context, text):
104 """Remove non-letters"""
105 if isinstance(text, list):
107 return re.sub(r'[^a-zA-Z]', '', text).strip()
108 _register_function(texcommand)
111 def lang_code_3to2(text):
112 """Convert 3-letter language code to 2-letter code"""
115 with open(get_resource('res/ISO-639-2_8859-1.txt'), 'rb') as f:
116 for line in f.read().decode('latin1').split('\n'):
117 codes = line.strip().split('|')
126 def mathml_latex(context, trees):
127 from librarian.embeds.mathml import MathML
128 text = MathML(trees[0]).to_latex().data
129 # Remove invisible multiplications, they produce unwanted spaces.
130 text = text.replace('\u2062', '')
134 def reg_mathml_latex():
135 _register_function(mathml_latex)
138 def reg_mathml_epub(output):
139 from librarian.embeds.mathml import MathML
141 def mathml(context, trees):
142 data = MathML(trees[0]).to_latex().to_png().data
143 name = "math%d.png" % mathml.count
147 uid='math%d' % mathml.count,
149 media_type='image/png',
156 _register_function(mathml)