When extracting body of a document in HTML source format (i.e. without pandoc conversion before display), use BeautifilSoup, which is more tolerant, instead of xml.dom.minidom, which crashes too often.
import chardet
import re
def to_unicode(input):
if type(input) == str:
res = None
encodings = ['utf8', 'latin1']
doc_enc = chardet.detect(input)['encoding']
if doc_enc:
encodings = [doc_enc,] + encodings
for encoding in encodings:
try:
res = unicode(input, encoding)
break;
except UnicodeDecodeError:
pass
if not res:
raise Exception('UnicodeDecodeError: could not decode')
return res
return input
# strip carriage returns
def strip_cr(input):
return re.sub('\r\n|\r|\n', '\n', input)