|
|
|
@ -15,6 +15,10 @@ logger = logging.getLogger(__name__)
|
|
|
|
|
n_kres_files = -1 # for logging
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
def enriched_lemma(token):
|
|
|
|
|
return (token["lemma"] if token["msd"][0] == "G" else token["lemma"] + "_")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
def _helper_tid_to_token(tid, tokens):
|
|
|
|
|
for t in tokens:
|
|
|
|
|
if t["tid"] == tid:
|
|
|
|
@ -29,7 +33,7 @@ def _db_preprocess(e):
|
|
|
|
|
else:
|
|
|
|
|
hw_tids = list(set([x["from"] for x in e["srl_links"]]))
|
|
|
|
|
hw_tokens = [_helper_tid_to_token(tid, e["tokens"]) for tid in hw_tids]
|
|
|
|
|
headwords = [(t["lemma"] if t["msd"][0] == "G" else t["lemma"] + "_") for t in hw_tokens]
|
|
|
|
|
headwords = [enriched_lemma(t) for t in hw_tokens]
|
|
|
|
|
e["headwords"] = headwords
|
|
|
|
|
|
|
|
|
|
functors = list(set([x["afun"] for x in e["srl_links"]]))
|
|
|
|
|