talk / backend /tests /test_laugh_split.py
user
Deploy StyleVoiceChat backend
aa9892b
Raw History Blame Contribute Delete
3.93 kB
"""laugh_split unit tests β€” μ›ƒμŒ 토큰 μ •κ·œμ‹ + span 감지 νšŒκ·€ λ°©μ§€."""
from __future__ import annotations
import sys
from pathlib import Path
# allow `from services import laugh_split` from this test file
sys.path.insert(0, str(Path(__file__).resolve().parents[1]))
from services.laugh_split import ( # noqa: E402
classify_laugh_length,
find_laugh_spans,
is_pure_laughter,
)
# ----- is_pure_laughter --------------------------------------------------
def test_pure_laughter_basic() -> None:
assert is_pure_laughter("ν‘Έν•˜ν•˜")
assert is_pure_laughter("ν‘Έν•˜ν•˜ν•˜ν•˜!")
assert is_pure_laughter("γ…‹γ…‹γ…‹γ…‹")
assert is_pure_laughter("ν•˜ν•˜ ν•˜ν•˜ν•˜")
assert is_pure_laughter("μœΌν•˜ν•˜ν•˜")
assert is_pure_laughter("크크크")
assert is_pure_laughter("히히히")
assert is_pure_laughter("κΉ”κΉ”")
# μ–‘ 끝 곡백/ꡬ두점 ν—ˆμš©
assert is_pure_laughter(" ν‘Έν•˜ν•˜! ")
assert is_pure_laughter("ν‘Έν•˜ν•˜.")
def test_pure_laughter_negative() -> None:
assert not is_pure_laughter("였빠 μ§„μ§œ 웃긴닀!")
assert not is_pure_laughter("μ•„ 정말?")
assert not is_pure_laughter("μœΌν•˜ν•˜ν•˜ λ„ˆλ¬΄ 웃겨") # μ›ƒμŒ λΆ€λΆ„ < 70%
assert not is_pure_laughter("ν•˜") # λ„ˆλ¬΄ 짧음
assert not is_pure_laughter("")
assert not is_pure_laughter(" ")
# ----- find_laugh_spans --------------------------------------------------
def test_spans_basic() -> None:
spans = find_laugh_spans("ν‘Έν•˜ν•˜")
assert len(spans) == 1
s, e = spans[0]
assert (s, e) == (0, 3)
def test_spans_in_sentence() -> None:
text = "이거 μ§„μ§œ λŒ€λ°•μ΄μ•Ό. λ°”λ‚˜λ‚˜κ°€ μ›ƒμœΌλ©΄ λ°”λ‚˜λ‚˜ν‚₯. ν‘Έν•˜ν•˜ 이거 μ™„μ „ 졜고 μ•„λ‹ˆμ•Ό?"
spans = find_laugh_spans(text)
assert len(spans) >= 1
# "ν‘Έν•˜ν•˜" μœ„μΉ˜λ₯Ό μ •ν™•νžˆ μž‘λŠ”κ°€?
found = any(text[s:e].strip() == "ν‘Έν•˜ν•˜" for s, e in spans)
assert found, f"ν‘Έν•˜ν•˜ span 미감지: spans={spans!r}, text={text!r}"
def test_spans_skip_single_char() -> None:
# 1κΈ€μž (λ‹¨μˆœ "ν•˜") λŠ” μ§„μ§œ μ›ƒμŒμ΄ μ•„λ‹ˆλ―€λ‘œ μ œμ™Έ
spans = find_laugh_spans("λ‚˜λŠ” μ•Ό 그래 정도 μ•Œμ§€ ν•˜ κ·Έλž˜λ„")
# "ν•˜" 단독 κΈ€μžλŠ” λ§€μΉ­ μ•ˆ λ˜μ–΄μ•Ό 함
one_char_spans = [(s, e) for s, e in spans if e - s == 1]
assert not one_char_spans, f"1κΈ€μž λ§€μΉ­ λ°œμƒ: {one_char_spans}"
def test_spans_skip_single_char_with_trailing_space() -> None:
# "히 " 처럼 단일 κΈ€μž + 곡백 쑰합도 strip ν›„ 1κΈ€μžλΌ μ œμ™ΈλΌμ•Ό 함.
# (regex group 3 의 \s κ°€ 맀치λ₯Ό 2자둜 λ§Œλ“€μ–΄ νšŒκ·€ν•˜λ˜ 버그.)
for sample in ["히 ", " 히 ", "말이야. 히 κ·ΈλŸ¬λ‹ˆκΉŒ", "응. 히. μ•Œκ² μ–΄"]:
spans = find_laugh_spans(sample)
single_core = [(s, e) for s, e in spans
if sample[s:e].strip() == "히"]
assert not single_core, f"단일 '히' span λ°œμƒ: {sample!r} β†’ {spans}"
def test_spans_merge_adjacent() -> None:
# μΈμ ‘ν•œ μ—¬λŸ¬ λ§€μΉ˜κ°€ ν•˜λ‚˜μ˜ span 으둜 λ³‘ν•©λ˜λŠ”κ°€?
text = "ν‘Έν•˜ν•˜ γ…‹γ…‹"
spans = find_laugh_spans(text)
# 곡백 1개 갭은 병합 β†’ 단일 span
assert len(spans) <= 2 # 병합 λ˜λŠ” 뢄리 λ‘˜ λ‹€ OK; 0κ°œλŠ” μ•ˆ 됨
assert spans, "μ›ƒμŒ span 0개"
# ----- classify_laugh_length ---------------------------------------------
def test_classify_short() -> None:
assert classify_laugh_length("ν‘Έν•˜") == "short"
assert classify_laugh_length("ν•˜ν•˜") == "short"
assert classify_laugh_length("γ…‹γ…‹γ…‹") == "short"
def test_classify_medium() -> None:
assert classify_laugh_length("ν‘Έν•˜ν•˜ν•˜") == "medium"
assert classify_laugh_length("μœΌν•˜ν•˜ν•˜!") == "medium"
def test_classify_long() -> None:
assert classify_laugh_length("ν‘Έν•˜ν•˜ν•˜ν•˜ λ„ˆλ¬΄ 웃겨") == "long"
assert classify_laugh_length("μœΌν•˜ν•˜ν•˜ν•˜ν•˜ν•˜ν•˜") == "long" # 8 chars