def handle_possession(ud_tag: UdTag) -> UmFeat:
psor_parts = [p for p in ud_tag if "[psor]" in p]
if not psor_parts:
return EMPTY_FEAT
if "None" in str(psor_parts):
return UmFeat("PSSD")
try:
assert len(psor_parts) <= 2
except AssertionError:
print(psor_parts)
raise
if "Number[psor]=Plur" in psor_parts:
number = "P"
elif "Number[psor]=Sing" in psor_parts:
number = "S"
elif "Number[psor]=Dual" in psor_parts:
number = "D"
else:
assert "Number" not in str(psor_parts)
number = ""
if "Person[psor]=1" in psor_parts:
person = "1"
elif "Person[psor]=2" in psor_parts:
person = "2"
elif "Person[psor]=3" in psor_parts:
person = "3"
else:
assert "Person" not in str(psor_parts)
person = ""
return UmFeat(f"PSS{person}{number}")
def ud2um(ud_tag: UdTag) -> UmTag:
um_tag: List[UmFeat] = []
possession = handle_possession(ud_tag)
um_tag.append(possession)
arguments = handle_arguments(ud_tag)
um_tag.extend(arguments)
for part in ud_tag:
if "," not in part:
tag = process_tag(part)
um_tag.append(tag)
else:
key, vals = part.split("=")
vals = vals.split(",")
all_parts = []
for val in vals:
tag = process_tag(UdFeat(f"{key}={val}"))
all_parts.append(tag)
all_parts = [p for p in all_parts if p != "_"]
um_tag.append(
UmFeat(f"{{{'/'.join(all_parts)}}}"
) if all_parts else EMPTY_FEAT)
um_tag = [f for f in um_tag if str(f) != "_"] or [EMPTY_FEAT]
# print(um_tag)
return UmTag(";".join(um_tag))
import re
from collections import defaultdict
from typing import List, Set
from languages import languages
from utils import CoNLLRow, UdFeat, UdTag, UmFeat, UmTag, ud2um_mapping
EMPTY_FEAT = UmFeat("_")
def handle_arguments(ud: UdTag) -> List[UmFeat]:
def handle_argument(parts):
parts = str(parts)
if "[psed]" in parts or "[gram]" in parts:
return "_"
if "[erg]" in parts:
kind = "ER"
elif "[dat]" in parts:
kind = "DA"
elif "[abs]" in parts:
kind = "AB"
else:
print(parts)
raise AssertionError
if "=Plur" in parts:
number = "P"
elif "=Sing" in parts:
number = "S"
elif "=Dual" in parts: