markdowntei.py
1.01 KB
'Markdown to TEI converter'
import re
from tei import Tei, Utterance
from splitter import split_sessions
class MarkdownTei:
'Converter from Markdown to TEI'
def __init__(self):
self.who = 'komentarz'
def convert(self, filename, output):
'Convert file to TEI'
self.who = 'komentarz'
data = list(filter(None, [self._utterance(par)
for par in self._read(filename)]))
sessions = split_sessions(data)
for session in sessions:
tei = Tei(session, filename)
tei.save(output)
@ staticmethod
def _read(filename):
return filter(None, [par.strip() for par in open(filename)])
def _utterance(self, par):
match = re.match(r'\*\((.*)\)\*', par)
if match:
return Utterance(who='komentarz', u=match.group(1))
match = re.match(r'\*\*(.*):\*\*', par)
if match:
self.who = match.group(1)
return None
return Utterance(who=self.who, u=par)