X-Git-Url: https://git.mdrn.pl/librarian.git/blobdiff_plain/167d19bdfe76a77754d2e14a80ac0c25c725c379..2c73c162844627d26991915fecc7e39f02bc34b8:/librarian/functions.py?ds=inline diff --git a/librarian/functions.py b/librarian/functions.py index 8427ba8..b88a7fb 100644 --- a/librarian/functions.py +++ b/librarian/functions.py @@ -6,13 +6,16 @@ from lxml import etree import re +from librarian.dcparser import Person +from librarian import get_resource + def _register_function(f): """ Register extension function with lxml """ ns = etree.FunctionNamespace('http://wolnelektury.pl/functions') ns[f.__name__] = f -def reg_substitute_entities(): +def reg_substitute_entities(): ENTITY_SUBSTITUTIONS = [ (u'---', u'—'), (u'--', u'–'), @@ -68,9 +71,9 @@ def reg_wrap_words(): text = ''.join(text) if not wrapping: return text - + words = re.split(r'\s', text) - + line_length = 0 lines = [[]] for word in words: @@ -89,7 +92,32 @@ def reg_person_name(): """ Converts "Name, Forename" to "Forename Name" """ if isinstance(text, list): text = ''.join(text) - return ' '.join([t.strip() for t in text.split(',', 1)[::-1]]) + return Person.from_text(text).readable() _register_function(person_name) +def reg_texcommand(): + def texcommand(context, text): + """Remove non-letters""" + if isinstance(text, list): + text = ''.join(text) + return re.sub(r'[^a-zA-Z]', '', text).strip() + _register_function(texcommand) + +def reg_lang_code_3to2(): + def lang_code_3to2(context, text): + """Convert 3-letter language code to 2-letter code""" + result = '' + text = ''.join(text) + with open(get_resource('res/ISO-639-2_8859-1.txt'), 'rb') as f: + for line in f: + list = line.strip().split('|') + if list[0] == text: + result=list[2] + if result == '': + return text + else: + return result + _register_function(lang_code_3to2) + +