diff -r 2b510a3a0f7d -r d2d28468b410 script/utils/export_twitter_alchemy.py --- a/script/utils/export_twitter_alchemy.py Sun Sep 23 17:14:07 2018 +0200 +++ b/script/utils/export_twitter_alchemy.py Mon Oct 22 17:04:30 2018 +0200 @@ -96,9 +96,34 @@ else: return None +def parse_polemics_3(tw, extended_mode): + """ + parse polemics in text and return a list of polemic code. None if not polemic found + """ + polemics = {} + for m in re.finditer("(\+\+|\?\?|\*\*|\=\=)",tw.text): + pol_link = { + '++' : u'OK', + '??' : u'KO', + '**' : u'REF', + '==' : u'Q'}[m.group(1)] + polemics[pol_link] = pol_link + + if extended_mode: + for entity in tw.entity_list: + if entity.type == "entity_url": + polemics["REF"] = "REF" + + if len(polemics) > 0: + return polemics.keys() + else: + return None + + protocol_version_map = { "1" : parse_polemics_1, - "2" : parse_polemics_2 + "2" : parse_polemics_2, + "3" : parse_polemics_3 } def get_options():