|
"""Sloleks is a reference morphological lexicon of Slovene.""" |
|
|
|
import xml.etree.ElementTree as ET |
|
import os |
|
|
|
import datasets |
|
|
|
NA_STR = "N/A" |
|
|
|
|
|
_CITATION = """\ |
|
@misc{sloleks3, |
|
title = {Morphological lexicon Sloleks 3.0}, |
|
author = {{\v C}ibej, Jaka and Gantar, Kaja and Dobrovoljc, Kaja and Krek, Simon and Holozan, Peter and Erjavec, Toma{\v z} and Romih, Miro and Arhar Holdt, {\v S}pela and Krsnik, Luka and Robnik-{\v S}ikonja, Marko}, |
|
url = {http://hdl.handle.net/11356/1745}, |
|
note = {Slovenian language resource repository {CLARIN}.{SI}}, |
|
copyright = {Creative Commons - Attribution-{ShareAlike} 4.0 International ({CC} {BY}-{SA} 4.0)}, |
|
year = {2022} |
|
} |
|
""" |
|
|
|
_DESCRIPTION = """\ |
|
Sloleks is a reference morphological lexicon of Slovene that was developed to be used in various NLP applications and language manuals. \ |
|
It contains Slovene lemmas, their inflected or derivative word forms and the corresponding grammatical description. \ |
|
In addition to the approx. 100,000 entries already available in Sloleks 2.0, Sloleks 3.0 contains an additional cca. \ |
|
265,000 newly generated entries from the most frequent lemmas in Gigafida 2.0 not yet included in previous versions of \ |
|
Sloleks. For verbs, adjectives, adverbs, and common nouns, the lemmas were checked manually by three annotators and \ |
|
included in Sloleks only if confirmed as legitimate by at least one annotator. \ |
|
No manual checking was performed on proper nouns. |
|
""" |
|
|
|
_HOMEPAGE = "https://viri.cjvt.si/sloleks/eng/" |
|
|
|
_LICENSE = "Creative Commons - Attribution-ShareAlike 4.0 International (CC BY-SA 4.0)" |
|
|
|
_URLS = { |
|
"sloleks3": "https://www.clarin.si/repository/xmlui/bitstream/handle/11356/1745/Sloleks.3.0.zip", |
|
} |
|
|
|
|
|
class Sloleks(datasets.GeneratorBasedBuilder): |
|
"""Sloleks is a reference morphological lexicon of Slovene.""" |
|
|
|
VERSION = datasets.Version("3.0.0") |
|
|
|
def _info(self): |
|
features = datasets.Features( |
|
{ |
|
"headword_lemma": datasets.Value("string"), |
|
"pos": datasets.Value("string"), |
|
"lex_unit": { |
|
"id": datasets.Value("string"), |
|
"form": datasets.Value("string"), |
|
"key": datasets.Value("string"), |
|
"type": datasets.Value("string") |
|
}, |
|
"word_forms": [{ |
|
"forms": datasets.Sequence(datasets.Value("string")), |
|
"accentuation": datasets.Sequence(datasets.Value("string")), |
|
"pronunciation_ipa": datasets.Sequence(datasets.Value("string")), |
|
"pronunciation_sampa": datasets.Sequence(datasets.Value("string")), |
|
"is_nonstandard": datasets.Sequence(datasets.Value("bool")), |
|
"msd": datasets.Value("string") |
|
}], |
|
"is_manually_checked": datasets.Value("bool") |
|
} |
|
) |
|
|
|
return datasets.DatasetInfo( |
|
description=_DESCRIPTION, |
|
features=features, |
|
homepage=_HOMEPAGE, |
|
license=_LICENSE, |
|
citation=_CITATION |
|
) |
|
|
|
def _split_generators(self, dl_manager): |
|
urls = _URLS["sloleks3"] |
|
data_dir = dl_manager.download_and_extract(urls) |
|
return [ |
|
datasets.SplitGenerator( |
|
name=datasets.Split.TRAIN, |
|
gen_kwargs={"data_dir": os.path.join(data_dir, "Sloleks.3.0")} |
|
) |
|
] |
|
|
|
def _generate_examples(self, data_dir): |
|
all_files = [os.path.join(data_dir, fname) |
|
for fname in os.listdir(data_dir) |
|
if os.path.isfile(os.path.join(data_dir, fname)) and fname.startswith("sloleks_3.0")] |
|
all_files = sorted(all_files) |
|
|
|
_idx_ex = 0 |
|
for file_path in all_files: |
|
curr_doc = ET.parse(file_path) |
|
root = curr_doc.getroot() |
|
|
|
for entry in root.findall("entry"): |
|
head_tag = entry.find("head") |
|
|
|
is_manually_checked = head_tag.find("status").text.strip() == "MANUAL" |
|
assert head_tag.find("status").text.strip() in {"MANUAL", "AUTOMATIC"} |
|
|
|
headword_lemma = head_tag.find("headword").find("lemma").text.strip() |
|
|
|
""" |
|
<lexicalUnit sloleksId="LE_..." sloleksKey="..." type="..."> |
|
<lexeme>...</lexeme> |
|
</lexicalUnit> |
|
""" |
|
lex_unit_tag = head_tag.find("lexicalUnit") |
|
assert "sloleksId" in lex_unit_tag.attrib |
|
lu_id = lex_unit_tag.attrib["sloleksId"] |
|
lu_key = lex_unit_tag.attrib["sloleksKey"] |
|
lu_type = lex_unit_tag.attrib["type"] |
|
lu_text = lex_unit_tag.find("lexeme").text.strip() |
|
|
|
""" |
|
<grammar> |
|
<category>verb</category> |
|
<subcategory type="pronunciation">Slovene G2P</subcategory> |
|
<grammarFeature name="type">main</grammarFeature> |
|
<grammarFeature name="aspect">biaspectual</grammarFeature> |
|
</grammar> |
|
""" |
|
grammar_tag = head_tag.find("grammar") |
|
|
|
|
|
category_tag = grammar_tag.find("category") |
|
category_str = NA_STR |
|
if category_tag is not None: |
|
category_str = category_tag.text.strip() |
|
|
|
|
|
related_tag = head_tag.find("relatedEntryList") |
|
related_entries = [] |
|
if related_tag is not None: |
|
for _related in related_tag: |
|
related_entries.append(_related.attrib["origin"].strip()) |
|
|
|
body_tag = entry.find("body") |
|
word_forms = [] |
|
for _form in body_tag.iterfind(".//wordForm"): |
|
msd_str = _form.find("msd").text.strip() |
|
|
|
orthography, accentuation, pronunciation_ipa, pronunciation_sampa = [], [], [], [] |
|
is_nonstandard = [] |
|
for _tag in _form.findall(".//orthography"): |
|
orthography.append(_tag.find("form").text.strip()) |
|
is_nonstandard.append(_tag.attrib.get("norm", "standard") == "non-standard") |
|
|
|
for _tag in _form.findall(".//accentuation"): |
|
accentuation.append(_tag.find("form").text.strip()) |
|
|
|
for _tag in _form.findall(".//pronunciation"): |
|
for _pronunciation_form in _tag.findall("form"): |
|
if _pronunciation_form.attrib["script"] == "IPA": |
|
pronunciation_ipa.append(_pronunciation_form.text.strip()) |
|
else: |
|
pronunciation_sampa.append(_pronunciation_form.text.strip()) |
|
|
|
word_forms.append({ |
|
"forms": orthography, |
|
"accentuation": accentuation, |
|
"pronunciation_ipa": pronunciation_ipa, |
|
"pronunciation_sampa": pronunciation_sampa, |
|
"is_nonstandard": is_nonstandard, |
|
"msd": msd_str |
|
}) |
|
|
|
yield _idx_ex, { |
|
"headword_lemma": headword_lemma, |
|
"pos": category_str, |
|
"lex_unit": { |
|
"id": lu_id, |
|
"form": lu_text, |
|
"key": lu_key, |
|
"type": lu_type |
|
}, |
|
"word_forms": word_forms, |
|
"is_manually_checked": is_manually_checked |
|
} |
|
_idx_ex += 1 |
|
|
