zimankit 0.2.0__py3-none-any.whl

zimankit/__init__.py

@@ -0,0 +1,60 @@
+"""zimankit - Kurdish (Central ckb / Northern kmr) NLP skills.
+
+Quick API::
+
+    from zimankit import analyze, normalize, parse, to_ipa, identify
+    analyze("کتێبەکەم")[0].segmentation   # 'کتێب+ەکە+م'
+    to_ipa("jinekê")                        # 'ʒɪnɛkeː'
+    parse("Ew baş e.").tokens[1].deprel     # 'root'
+"""
+
+from __future__ import annotations
+
+__version__ = "0.2.0"
+
+from .shared_utils.lid import identify
+from .shared_utils.text_helpers import detect_script
+
+
+def _lang(text: str, lang: str | None) -> str:
+    return lang or ("ckb" if detect_script(text) == "arabic" else "kmr")
+
+
+def normalize(text: str, lang: str | None = None) -> str:
+    if _lang(text, lang) == "ckb":
+        from .skills_ckb.orthography import normalize as f
+    else:
+        from .skills_kmr.orthography import normalize as f
+    return f(text)
+
+
+def to_ipa(word: str, lang: str | None = None) -> str:
+    if _lang(word, lang) == "ckb":
+        from .skills_ckb.phonology_ipa import to_ipa as f
+    else:
+        from .skills_kmr.phonology_ipa import to_ipa as f
+    return f(word)
+
+
+_ANALYZERS: dict = {}
+
+
+def analyze(word: str, lang: str | None = None):
+    lg = _lang(word, lang)
+    if lg not in _ANALYZERS:
+        if lg == "ckb":
+            from .skills_ckb.morphology import CkbAnalyzer as A
+        else:
+            from .skills_kmr.morphology import KmrAnalyzer as A
+        _ANALYZERS[lg] = A()
+    return _ANALYZERS[lg].analyze(word)
+
+
+def parse(sentence: str, lang: str | None = None):
+    """Dependency structure of a sentence: see zimankit.dependency."""
+    from .dependency import parse as f
+
+    return f(sentence, lang)
+
+
+__all__ = ["analyze", "normalize", "parse", "to_ipa", "identify", "detect_script", "__version__"]

zimankit/__main__.py

@@ -0,0 +1,3 @@
+from .cli import main
+
+raise SystemExit(main())

zimankit/cli.py

@@ -0,0 +1,209 @@
+"""Command-line interface: ``python -m zimankit <command> ...`` or ``zimankit ...``.
+
+Every command reads its words/text from arguments, or from stdin when none
+are given, so it composes with shell pipelines.
+"""
+
+from __future__ import annotations
+
+import argparse
+import json
+import sys
+from collections.abc import Iterable
+from dataclasses import asdict
+
+
+def _inputs(args: list[str]) -> Iterable[str]:
+    if args:
+        yield from args
+    else:
+        for line in sys.stdin:
+            line = line.rstrip("\n")
+            if line:
+                yield line
+
+
+def _lang_of(text: str, forced: str | None) -> str:
+    if forced:
+        return forced
+    from .shared_utils.text_helpers import detect_script
+
+    return "ckb" if detect_script(text) == "arabic" else "kmr"
+
+
+def cmd_normalize(a) -> None:
+    for t in _inputs(a.text):
+        lang = _lang_of(t, a.lang)
+        if lang == "ckb":
+            from .skills_ckb.orthography import normalize
+        else:
+            from .skills_kmr.orthography import normalize
+        print(normalize(t))
+
+
+def cmd_tokenize(a) -> None:
+    from .shared_utils.tokenizer import tokenize
+
+    for t in _inputs(a.text):
+        print(" ".join(tok.text for tok in tokenize(t)) if not a.json else json.dumps([asdict(tok) for tok in tokenize(t)], ensure_ascii=False))
+
+
+def cmd_lid(a) -> None:
+    from .shared_utils.lid import identify
+
+    for t in _inputs(a.text):
+        r = identify(t)
+        print(json.dumps({"lang": r.lang, "script": r.script, "confidence": r.confidence}, ensure_ascii=False) if a.json else f"{r.lang}\t{r.confidence}\t{t[:60]}")
+
+
+def cmd_ipa(a) -> None:
+    for t in _inputs(a.text):
+        lang = _lang_of(t, a.lang)
+        if lang == "ckb":
+            from .skills_ckb.phonology_ipa import G2POptions, text_to_ipa
+
+            print(text_to_ipa(t, G2POptions(mark_epenthesis=not a.plain, syllable_dots=a.syllables)))
+        else:
+            from .skills_kmr.phonology_ipa import KmrG2POptions, text_to_ipa
+
+            print(text_to_ipa(t, KmrG2POptions(narrow=a.narrow, syllable_dots=a.syllables)))
+
+
+def cmd_translit(a) -> None:
+    from .shared_utils.tokenizer import tokenize
+
+    for t in _inputs(a.text):
+        lang = _lang_of(t, a.lang)
+        if lang == "ckb":
+            from .skills_ckb.phonology_ipa import to_latin as conv
+        else:
+            from .skills_kmr.translit import to_arabic as conv
+        print(" ".join(conv(tok.text) if tok.kind == "word" else tok.text for tok in tokenize(t)))
+
+
+def _analyzer(lang: str):
+    if lang == "ckb":
+        from .skills_ckb.morphology import CkbAnalyzer
+
+        return CkbAnalyzer()
+    from .skills_kmr.morphology import KmrAnalyzer
+
+    return KmrAnalyzer()
+
+
+def cmd_analyze(a) -> None:
+    from .shared_utils.tokenizer import words
+
+    cache: dict[str, object] = {}
+    for t in _inputs(a.text):
+        lang = _lang_of(t, a.lang)
+        an = cache.setdefault(lang, _analyzer(lang))
+        for w in words(t):
+            res = an.analyze(w)[: a.top]  # type: ignore[attr-defined]
+            if a.json:
+                print(json.dumps({"token": w, "analyses": [r.to_dict() for r in res]}, ensure_ascii=False))
+            else:
+                for r in res:
+                    print(f"{w}\t{r.lemma}\t{r.pos}\t{r.segmentation}\t{r.gloss}\t{json.dumps(r.features, ensure_ascii=False)}\t{r.score}")
+
+
+def cmd_conjugate(a) -> None:
+    lang = _lang_of(a.infinitive, a.lang)
+    if lang == "ckb":
+        from .skills_ckb import verbs as V
+    else:
+        from .skills_kmr import verbs as V
+    idx = V.default_index()
+    v = idx.verbs.get(a.infinitive)
+    if v is None:
+        known = {x.infinitive: x for x in idx.verbs.values() if not x.preverb}
+        v = V.make_verb(a.infinitive, None if a.transitive is None else a.transitive == "yes", a.present, "guess", known)
+    if v is None:
+        sys.exit(f"cannot build a paradigm for {a.infinitive!r}")
+    for f in V.conjugate(v):
+        feats = ",".join(f"{k}={x}" for k, x in f.features.items() if k != "pos")
+        print(f"{f.form}\t{feats}")
+
+
+def cmd_ergativity(a) -> None:
+    from .skills_kmr.syntax import detect_alignment
+
+    for t in _inputs(a.text):
+        print(json.dumps(detect_alignment(t).to_dict(), ensure_ascii=False))
+
+
+def cmd_parse(a) -> None:
+    from .dependency import parse
+
+    for t in _inputs(a.text):
+        p = parse(t, a.lang)
+        if a.json:
+            print(json.dumps(p.to_dict(), ensure_ascii=False))
+            continue
+        for w in p.tokens:
+            print(f"{w.id}\t{w.text}\t{w.lemma}\t{w.upos}\t{w.head}\t{w.deprel}")
+        print()
+
+
+def cmd_validate(a) -> None:
+    for t in _inputs(a.text):
+        lang = _lang_of(t, a.lang)
+        if lang == "ckb":
+            from .skills_ckb.orthography import normalize_word, validate_word
+        else:
+            from .skills_kmr.orthography import normalize_word, validate_word
+        w = normalize_word(t)
+        p = validate_word(w)
+        print(f"{t}\t{w}\t{'OK' if not p else '; '.join(p)}")
+
+
+def cmd_rules(a) -> None:
+    from .rules import export_rules
+
+    print(export_rules(a.out or f"output/{a.lang}/grammar_rules_{a.lang}.json", a.lang))
+
+
+def main(argv: list[str] | None = None) -> int:
+    from . import __version__
+
+    ap = argparse.ArgumentParser(prog="zimankit", description="Kurdish (ckb / kmr) NLP skills")
+    ap.add_argument("--version", action="version", version=f"zimankit {__version__}")
+    sub = ap.add_subparsers(dest="cmd", required=True)
+
+    def add(name, fn, help_, text=True):
+        p = sub.add_parser(name, help=help_)
+        if text:
+            p.add_argument("text", nargs="*")
+        p.add_argument("--lang", choices=["ckb", "kmr"])
+        p.set_defaults(fn=fn)
+        return p
+
+    add("normalize", cmd_normalize, "orthographic normalisation")
+    add("tokenize", cmd_tokenize, "word tokenisation").add_argument("--json", action="store_true")
+    add("lid", cmd_lid, "language identification").add_argument("--json", action="store_true")
+    p = add("ipa", cmd_ipa, "grapheme-to-phoneme (IPA)")
+    p.add_argument("--narrow", action="store_true", help="kmr: aspiration, ŋ")
+    p.add_argument("--plain", action="store_true", help="ckb: do not mark the inferred bizroke")
+    p.add_argument("--syllables", action="store_true")
+    add("translit", cmd_translit, "ckb Arabic->Latin, kmr Latin->Arabic")
+    p = add("analyze", cmd_analyze, "morphological analysis")
+    p.add_argument("--top", type=int, default=1)
+    p.add_argument("--json", action="store_true")
+    p = add("conjugate", cmd_conjugate, "full verb paradigm", text=False)
+    p.add_argument("infinitive")
+    p.add_argument("--present", help="present stem for verbs not in the reference list")
+    p.add_argument("--transitive", choices=["yes", "no"])
+    add("ergativity", cmd_ergativity, "kmr clause alignment / agreement check")
+    add("validate", cmd_validate, "orthographic validation of words")
+    add("parse", cmd_parse, "sentence structure: head and relation of each word").add_argument("--json", action="store_true")
+    p = add("rules", cmd_rules, "export grammar_rules_<lang>.json", text=False)
+    p.add_argument("--out")
+    a = ap.parse_args(argv)
+    if a.cmd == "rules" and not a.lang:
+        ap.error("rules needs --lang")
+    a.fn(a)
+    return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())

zimankit/config.py

@@ -0,0 +1,69 @@
+"""Configuration loading: ``config/pipeline.toml`` + ``config/sources.local.toml``."""
+
+from __future__ import annotations
+
+import os
+import tomllib
+from dataclasses import dataclass, field
+from pathlib import Path
+
+PROJECT_ROOT = Path(__file__).resolve().parent.parent
+
+
+@dataclass
+class Source:
+    id: str
+    tag: str
+    kind: str
+    lang: str
+    path: str | None = None
+    glob: list[str] = field(default_factory=list)
+    exclude: list[str] = field(default_factory=list)
+    options: dict = field(default_factory=dict)
+
+
+@dataclass
+class Config:
+    settings: dict
+    sources: list[Source]
+    deny_tokens: set[str]
+    root: Path
+
+    @property
+    def cache_dir(self) -> Path:
+        return self.root / self.settings["run"]["cache_dir"]
+
+    @property
+    def output_dir(self) -> Path:
+        return self.root / self.settings["run"]["output_dir"]
+
+    @property
+    def workers(self) -> int:
+        return int(os.environ.get("ZIMANKIT_WORKERS", self.settings["run"].get("workers", 4)))
+
+    def by_kind(self, *kinds: str) -> list[Source]:
+        return [s for s in self.sources if s.kind in kinds]
+
+
+def load_config(root: str | Path | None = None, sources_file: str | None = None) -> Config:
+    root = Path(root) if root else PROJECT_ROOT
+    with open(root / "config" / "pipeline.toml", "rb") as fh:
+        settings = tomllib.load(fh)
+    src_path = Path(sources_file) if sources_file else root / "config" / "sources.local.toml"
+    if not src_path.exists():
+        src_path = root / "config" / "sources.example.toml"
+    with open(src_path, "rb") as fh:
+        raw = tomllib.load(fh)
+    sources = []
+    known = {"id", "tag", "kind", "lang", "path", "glob", "exclude"}
+    for s in raw.get("source", []):
+        globs = s.get("glob", [])
+        sources.append(
+            Source(
+                id=s["id"], tag=s["tag"], kind=s["kind"], lang=s.get("lang", "auto"), path=s.get("path"),
+                glob=[globs] if isinstance(globs, str) else list(globs), exclude=list(s.get("exclude", [])),
+                options={k: v for k, v in s.items() if k not in known},
+            )
+        )
+    deny = {t.lower() for t in raw.get("deny_tokens", [])}
+    return Config(settings, sources, deny, root)

zimankit/data/ckb_function_words.tsv

@@ -0,0 +1,119 @@
+# Central Kurdish closed-class words (authored reference data).
+# word	upos	features
+من	PRON	person=1sg|case=dir
+تۆ	PRON	person=2sg|case=dir
+ئەو	PRON	person=3sg|deixis=distal
+ئێمە	PRON	person=1pl
+ئێوە	PRON	person=2pl
+ئەوان	PRON	person=3pl
+خۆ	PRON	reflexive=yes
+خۆم	PRON	reflexive=yes|person=1sg
+خۆت	PRON	reflexive=yes|person=2sg
+خۆی	PRON	reflexive=yes|person=3sg
+خۆمان	PRON	reflexive=yes|person=1pl
+خۆتان	PRON	reflexive=yes|person=2pl
+خۆیان	PRON	reflexive=yes|person=3pl
+ئەم	DET	deixis=proximal
+ئەمە	PRON	deixis=proximal
+ئەوە	PRON	deixis=distal
+ئەمانە	PRON	deixis=proximal|number=pl
+ئەوانە	PRON	deixis=distal|number=pl
+کێ	PRON	prontype=int
+چی	PRON	prontype=int
+کام	DET	prontype=int
+چۆن	ADV	prontype=int
+کەی	ADV	prontype=int
+بۆچی	ADV	prontype=int
+لەکوێ	ADV	prontype=int
+کوا	ADV	prontype=int
+چەند	DET	prontype=int
+هەموو	DET	prontype=tot
+هەر	DET	prontype=tot
+هیچ	DET	prontype=neg
+هەندێک	DET	prontype=ind
+لە	ADP	adptype=prep
+بە	ADP	adptype=prep
+بۆ	ADP	adptype=prep
+لەگەڵ	ADP	adptype=prep
+تا	ADP	adptype=prep
+بێ	ADP	adptype=prep
+بەبێ	ADP	adptype=prep
+وەک	ADP	adptype=prep
+لەسەر	ADP	adptype=prep
+لەژێر	ADP	adptype=prep
+لەناو	ADP	adptype=prep
+لەبەر	ADP	adptype=prep
+لەپێش	ADP	adptype=prep
+لەدوای	ADP	adptype=prep
+بەرەو	ADP	adptype=prep
+دوای	ADP	adptype=prep
+پێش	ADP	adptype=prep
+نێوان	ADP	adptype=prep
+دژی	ADP	adptype=prep
+و	CCONJ
+یان	CCONJ
+یا	CCONJ
+بەڵام	CCONJ
+کە	SCONJ
+ئەگەر	SCONJ
+چونکە	SCONJ
+هەتا	SCONJ
+تاکو	SCONJ
+هەروەها	ADV
+بەڵێ	INTJ
+نەخێر	INTJ
+ئا	INTJ
+نا	PART	polarity=neg
+نە	PART	polarity=neg
+زۆر	ADV	degree=yes
+کەم	ADJ
+ئێستا	ADV
+ئەمڕۆ	ADV
+دوێنێ	ADV
+سبەی	ADV
+لێرە	ADV
+لەوێ	ADV
+یەک	NUM	numtype=card
+دوو	NUM	numtype=card
+سێ	NUM	numtype=card
+چوار	NUM	numtype=card
+پێنج	NUM	numtype=card
+شەش	NUM	numtype=card
+حەوت	NUM	numtype=card
+هەشت	NUM	numtype=card
+نۆ	NUM	numtype=card
+دە	NUM	numtype=card
+سەد	NUM	numtype=card
+هەزار	NUM	numtype=card
+یەکەم	ADJ	numtype=ord
+دووەم	ADJ	numtype=ord
+سێیەم	ADJ	numtype=ord
+هەیە	VERB	existential=yes|polarity=pos
+نییە	VERB	existential=yes|polarity=neg
+ی	PART	ezafe=yes
+دا	ADP	adptype=post
+ەوە	ADP	adptype=post
+بۆیە	ADV	
+بەڵکو	CCONJ	
+لەلایەن	ADP	adptype=prep
+بەرامبەر	ADP	adptype=prep
+بەسەر	ADP	adptype=prep
+چەندین	DET	prontype=ind
+زیاتر	ADV	degree=cmp
+کەمتر	ADV	degree=cmp
+پێویستە	AUX	mood=necessitative
+دەبێت	AUX	mood=necessitative
+ئەوەیە	PRON	deixis=distal|copula=yes
+تەنها	ADV	
+تەنیا	ADV	
+هێشتا	ADV	
+پاشان	ADV	
+بەتایبەتی	ADV	
+کاک	NOUN	title=yes
+هەمیشە	ADV	
+هەردوو	DET	prontype=tot
+ئایا	PART	prontype=int
+ژێر	ADP	adptype=prep
+دژ	ADP	adptype=prep
+ڕەنگە	ADV	mood=potential
+زۆربەی	DET	prontype=ind

zimankit/data/ckb_verbs_core.tsv

@@ -0,0 +1,104 @@
+# Central Kurdish core verbs (authored reference data).
+# infinitive	past_stem	present_stem	transitivity	preverb	notes
+# transitivity: T = transitive (past tense is ergative), I = intransitive
+# preverb: a separable prefix written before the aspect marker (هەڵ دەگرێت)
+کردن	کرد	کە	T
+بوون	بوو	ب	I
+هاتن	هات	ێ	I		irregular present دێم, negative نایەم, imperative وەرە
+چوون	چوو	چ	I		imperative بڕۆ is suppletive (from ڕۆیشتن)
+گرتن	گرت	گر	T		passive stem گیر
+دان	دا	دە	T
+خواردن	خوارد	خۆ	T
+بردن	برد	بە	T
+هێنان	هێنا	هێن	T
+زانین	زانی	زان	T
+ویستن	ویست	وێ	T		present uses a clitic subject: دەمەوێت
+بینین	بینی	بین	T
+دیتن	دیت	بین	T
+گوتن	گوت	ڵێ	T
+وتن	وت	ڵێ	T
+نووسین	نووسی	نووس	T
+خوێندن	خوێند	خوێن	T
+کوشتن	کوشت	کوژ	T
+مردن	مرد	مر	I
+ڕۆیشتن	ڕۆیشت	ڕۆ	I
+نیشتن	نیشت	نیش	I
+هەستان	هەستا	هەست	I
+ناردن	نارد	نێر	T
+کڕین	کڕی	کڕ	T
+فرۆشتن	فرۆشت	فرۆش	T
+شکاندن	شکاند	شکێن	T
+شکان	شکا	شکێ	I
+سووتان	سووتا	سووتێ	I
+سووتاندن	سووتاند	سووتێن	T
+کەوتن	کەوت	کەو	I
+خستن	خست	خە	T
+ژیان	ژیا	ژی	I
+ترسان	ترسا	ترس	I
+پرسین	پرسی	پرس	T
+گەڕان	گەڕا	گەڕێ	I
+گەیشتن	گەیشت	گە	I
+گەیاندن	گەیاند	گەیەن	T
+شوشتن	شوشت	شۆ	T
+بەستن	بەست	بەست	T
+ناسین	ناسی	ناس	T
+خەوتن	خەوت	خەو	I
+گریان	گریا	گری	I
+کێشان	کێشا	کێش	T
+ڕشتن	ڕشت	ڕێژ	T
+هێشتن	هێشت	هێڵ	T
+مان	ما	مێن	I
+بیستن	بیست	بیس	T
+توانین	توانی	توان	T
+کەندن	کەند	کەن	T
+چاندن	چاند	چێن	T
+فڕین	فڕی	فڕ	I
+بارین	باری	بار	I
+گۆڕین	گۆڕی	گۆڕ	T
+گۆڕان	گۆڕا	گۆڕێ	I
+بڕین	بڕی	بڕ	T
+دۆزین	دۆزی	دۆز	T
+نووستن	نووست	نوو	I
+پێچان	پێچا	پێچ	T
+بژاردن	بژارد	بژێر	T
+پاراستن	پاراست	پارێز	T
+ژماردن	ژمارد	ژمێر	T
+سپاردن	سپارد	سپێر	T
+کوتان	کوتا	کوت	T
+لەرزین	لەرزی	لەرز	I
+بزووتن	بزووت	بزوو	I
+ڕووخان	ڕووخا	ڕووخێ	I
+کشان	کشا	کشێ	I
+تەقین	تەقی	تەق	I
+تەقاندن	تەقاند	تەقێن	T
+مژین	مژی	مژ	T
+گەزین	گەزی	گەز	T
+خنکان	خنکا	خنکێ	I
+خنکاندن	خنکاند	خنکێن	T
+ڕوانین	ڕوانی	ڕوان	I
+ئێشان	ئێشا	ئێش	I
+کەنین	کەنی	کەن	I
+خوێندنەوە	خوێند	خوێن	T		suffix -ەوە ("again, back")
+کردنەوە	کرد	کە	T
+هاتنەوە	هات	ێ	I
+چوونەوە	چوو	چ	I
+گەڕانەوە	گەڕا	گەڕێ	I
+دۆزینەوە	دۆزی	دۆز	T
+مانەوە	ما	مێن	I
+هەڵگرتن	گرت	گر	T	هەڵ
+هەڵبژاردن	بژارد	بژێر	T	هەڵ
+وەرگرتن	گرت	گر	T	وەر
+دەرکردن	کرد	کە	T	دەر
+تێگەیشتن	گەیشت	گە	I	تێ
+داخستن	خست	خە	T	دا
+دانیشتن	نیشت	نیش	I	دا
+ڕاگەیاندن	گەیاند	گەیەن	T	ڕا
+ڕاکردن	کرد	کە	I	ڕا
+لێدان	دا	دە	T	لێ
+پێکەنین	کەنی	کەن	I	پێ
+دەرکەوتن	کەوت	کەو	I	دەر
+دەرچوون	چوو	چ	I	دەر
+هەڵسان	هەڵسا	هەڵس	I
+پێکهاتن	هات	ێ	I	پێک
+بەکارهێنان	هێنا	هێن	T	بەکار
+دەستپێکردن	کرد	کە	T	دەستپێ

zimankit/data/kmr_function_words.tsv

@@ -0,0 +1,97 @@
+# Northern Kurdish closed-class words (authored reference data).
+# word	upos	features
+ez	PRON	person=1sg|case=dir
+tu	PRON	person=2sg|case=dir
+ew	PRON	person=3sg|case=dir|deixis=distal
+em	PRON	person=1pl|case=dir
+hûn	PRON	person=2pl|case=dir
+ev	PRON	person=3sg|case=dir|deixis=proximal
+min	PRON	person=1sg|case=obl
+te	PRON	person=2sg|case=obl
+wî	PRON	person=3sg|case=obl|gender=m|deixis=distal
+wê	PRON	person=3sg|case=obl|gender=f|deixis=distal
+vî	PRON	person=3sg|case=obl|gender=m|deixis=proximal
+vê	PRON	person=3sg|case=obl|gender=f|deixis=proximal
+me	PRON	person=1pl|case=obl
+we	PRON	person=2pl|case=obl
+wan	PRON	person=3pl|case=obl|deixis=distal
+van	PRON	person=3pl|case=obl|deixis=proximal
+xwe	PRON	reflexive=yes
+yê	PRON	prontype=ezafe|gender=m|number=sg
+ya	PRON	prontype=ezafe|gender=f|number=sg
+yên	PRON	prontype=ezafe|number=pl
+kî	PRON	prontype=int
+çi	PRON	prontype=int
+kîjan	DET	prontype=int
+çend	DET	prontype=int
+kengî	ADV	prontype=int
+çawa	ADV	prontype=int
+çima	ADV	prontype=int
+her	DET	prontype=tot
+hemû	DET	prontype=tot
+tu	DET	prontype=neg
+hin	DET	prontype=ind
+di	ADP	adptype=prep
+ji	ADP	adptype=prep
+bi	ADP	adptype=prep
+li	ADP	adptype=prep
+bo	ADP	adptype=prep
+jibo	ADP	adptype=prep
+ber	ADP	adptype=prep
+bê	ADP	adptype=prep
+heta	ADP	adptype=prep
+ta	ADP	adptype=prep
+ligel	ADP	adptype=prep
+digel	ADP	adptype=prep
+nav	ADP	adptype=prep
+navbera	ADP	adptype=prep
+ser	ADP	adptype=prep
+bin	ADP	adptype=prep
+wek	ADP	adptype=prep
+de	ADP	adptype=post
+da	ADP	adptype=post
+re	ADP	adptype=post
+ra	ADP	adptype=post
+ve	ADP	adptype=post
+û	CCONJ
+an	CCONJ
+yan	CCONJ
+lê	CCONJ
+belê	CCONJ
+ku	SCONJ
+heke	SCONJ
+eger	SCONJ
+wekî	SCONJ
+jî	PART	additive=yes
+ne	PART	polarity=neg
+na	PART	polarity=neg
+dê	AUX	tense=fut
+ê	AUX	tense=fut
+ma	PART	prontype=int
+e	AUX	copula=yes|person=3sg
+ye	AUX	copula=yes|person=3sg
+im	AUX	copula=yes|person=1sg
+in	AUX	copula=yes|person=pl
+heye	VERB	existential=yes|polarity=pos
+tune	VERB	existential=yes|polarity=neg
+tuneye	VERB	existential=yes|polarity=neg
+gelek	ADV	degree=yes
+pir	ADV	degree=yes
+niha	ADV
+îro	ADV
+duh	ADV
+sibê	ADV
+vir	ADV
+wir	ADV
+yek	NUM	numtype=card
+du	NUM	numtype=card
+sê	NUM	numtype=card
+çar	NUM	numtype=card
+pênc	NUM	numtype=card
+şeş	NUM	numtype=card
+heft	NUM	numtype=card
+heşt	NUM	numtype=card
+neh	NUM	numtype=card
+deh	NUM	numtype=card
+sed	NUM	numtype=card
+hezar	NUM	numtype=card