StarAsh042
6cca71a249
测试此前被 .gitignore 排除，从未进入 lint 视野。纳入版本库后 `ruff check` 报 21 处（I001 / F401 / F841 等）、`ruff format` 要求改 13 个文件， 这里一次性对齐，避免 CI 的 lint job 一上线就是红的。
216 lines
8.7 KiB
Python
216 lines
8.7 KiB
Python
"""``modify_danbooru_character.py`` 的回归测试。
|
|
|
|
同时守住一条约定：两个加工脚本必须共用同一份转义实现（``danbooru_csv_tool``），
|
|
避免「改了一个忘了另一个」导致两边产出的提示词规则不一致。
|
|
"""
|
|
|
|
import csv
|
|
import os
|
|
|
|
import pytest
|
|
from ComfyUI_RollingArtist import modify_danbooru_art as art_script
|
|
from ComfyUI_RollingArtist import modify_danbooru_character as character_script
|
|
from ComfyUI_RollingArtist.danbooru_csv_tool import BACKUP_SUFFIX
|
|
from ComfyUI_RollingArtist.danbooru_csv_tool import convert as generic_convert
|
|
|
|
HEADER = ["character", "copyright", "trigger", "core_tags", "count", "solo_count", "url"]
|
|
ROWS = [
|
|
HEADER,
|
|
[
|
|
"hatsune_miku",
|
|
"vocaloid",
|
|
"hatsune miku, vocaloid",
|
|
"1girl, aqua eyes, twintails",
|
|
"103500",
|
|
"75493",
|
|
"http://x",
|
|
],
|
|
[
|
|
"admiral_(kancolle)",
|
|
"kantai_collection",
|
|
"admiral (kancolle), kantai collection",
|
|
"1girl, ponytail",
|
|
"4000",
|
|
"3000",
|
|
"http://x",
|
|
],
|
|
["low_count", "some_copyright", "low count", "1girl", "29", "10", "http://x"],
|
|
["exactly_thirty", "", "exactly thirty", "", "30", "10", "http://x"],
|
|
["", "no_character", "no character", "1girl", "99", "10", "http://x"],
|
|
["empty_trigger", "some_copyright", "", "1girl", "98", "10", "http://x"],
|
|
]
|
|
|
|
|
|
def build(path, rows=ROWS):
|
|
with open(path, "w", encoding="utf-8", newline="") as handle:
|
|
csv.writer(handle).writerows(rows)
|
|
return str(path)
|
|
|
|
|
|
# ----------------------------------------------------------------------
|
|
# 接口约定
|
|
# ----------------------------------------------------------------------
|
|
def test_output_columns_and_escape_columns():
|
|
assert character_script.OUTPUT_COLUMNS == ("character", "copyright", "trigger", "core_tags", "count")
|
|
assert character_script.ESCAPE_COLUMNS == ("trigger",)
|
|
assert character_script.DEFAULT_MIN_COUNT == 30
|
|
assert character_script.DEFAULT_INPUT.endswith("danbooru_character.csv")
|
|
assert character_script.DEFAULT_OUTPUT.endswith("danbooru_character_001.csv")
|
|
|
|
|
|
def test_both_scripts_share_one_escape_implementation():
|
|
"""两个脚本必须共用同一份转义函数，否则规则会漂移。"""
|
|
assert art_script.transform_trigger is character_script.transform_trigger
|
|
assert character_script.transform_text is art_script.transform_text
|
|
|
|
|
|
def test_escape_columns_are_identical_across_scripts():
|
|
assert art_script.ESCAPE_COLUMNS == character_script.ESCAPE_COLUMNS == ("trigger",)
|
|
|
|
|
|
# ----------------------------------------------------------------------
|
|
# 转换逻辑
|
|
# ----------------------------------------------------------------------
|
|
def test_convert_keeps_five_columns_with_header():
|
|
output, stats = character_script.convert(ROWS, 30)
|
|
assert output[0] == ["character", "copyright", "trigger", "core_tags", "count"]
|
|
assert all(len(row) == 5 for row in output)
|
|
# 6 行：low_count(29) 被阈值过滤，空 character 与空 trigger 各跳过 1 行
|
|
assert stats == {"total": 6, "kept": 3, "skipped": 2}
|
|
|
|
|
|
def test_convert_filters_by_count_and_sorts_descending():
|
|
output, _ = character_script.convert(ROWS, 30)
|
|
names = [row[0] for row in output[1:]]
|
|
counts = [int(row[4]) for row in output[1:]]
|
|
|
|
assert "low_count" not in names # count 29 被过滤
|
|
assert "exactly_thirty" in names # 边界值保留
|
|
assert counts == sorted(counts, reverse=True)
|
|
|
|
|
|
def test_convert_escapes_only_trigger():
|
|
output, _ = character_script.convert(ROWS, 0)
|
|
rows = {row[0]: row for row in output[1:]}
|
|
|
|
# trigger：逐段处理（逗号后的空格不会变成下划线），括号也被转义
|
|
assert rows["hatsune_miku"][2] == "hatsune_miku,vocaloid"
|
|
assert rows["admiral_(kancolle)"][2] == "admiral_\\(kancolle\\),kantai_collection"
|
|
|
|
# character：原样保留（含未转义括号）
|
|
assert rows["admiral_(kancolle)"][0] == "admiral_(kancolle)"
|
|
|
|
# copyright / core_tags：原样保留
|
|
assert rows["hatsune_miku"][1] == "vocaloid"
|
|
assert rows["hatsune_miku"][3] == "1girl, aqua eyes, twintails"
|
|
|
|
|
|
def test_convert_keeps_empty_copyright_but_skips_empty_character_or_trigger():
|
|
output, stats = character_script.convert(ROWS, 0)
|
|
names = [row[0] for row in output[1:]]
|
|
assert "exactly_thirty" in names # copyright 为空不影响保留
|
|
assert "empty_trigger" not in names # trigger 为空 -> 跳过
|
|
assert "" not in names # character 为空 -> 跳过
|
|
assert stats["skipped"] == 2
|
|
|
|
|
|
def test_convert_skips_short_and_bad_rows():
|
|
rows = [
|
|
HEADER,
|
|
["ok", "c", "ok trigger", "tags", "50", "1", "http://x"],
|
|
["short", "c", "trigger"],
|
|
["bad_count", "c", "bad count trigger", "tags", "abc", "1", "http://x"],
|
|
[],
|
|
]
|
|
output, stats = generic_convert(
|
|
rows, 30, character_script.OUTPUT_COLUMNS, character_script.ESCAPE_COLUMNS
|
|
)
|
|
assert [row[0] for row in output[1:]] == ["ok"]
|
|
assert stats == {"total": 3, "kept": 1, "skipped": 2}
|
|
|
|
|
|
def test_convert_rejects_missing_column():
|
|
with pytest.raises(ValueError, match="core_tags"):
|
|
character_script.column_indexes(["character", "copyright", "trigger", "count"])
|
|
|
|
|
|
def test_convert_rejects_empty_file():
|
|
with pytest.raises(ValueError, match="空"):
|
|
character_script.convert([], 30)
|
|
|
|
|
|
# ----------------------------------------------------------------------
|
|
# 命令行
|
|
# ----------------------------------------------------------------------
|
|
def test_main_end_to_end(tmp_path, capsys):
|
|
source = build(tmp_path / "character.csv")
|
|
output = tmp_path / "character_001.csv"
|
|
|
|
assert character_script.main(["-i", source, "-o", str(output), "-n", "30"]) == 0
|
|
with open(output, encoding="utf-8", newline="") as handle:
|
|
body = list(csv.reader(handle))
|
|
assert body[0] == ["character", "copyright", "trigger", "core_tags", "count"]
|
|
assert len(body) == 4
|
|
text = capsys.readouterr().out
|
|
assert "保留 3 行" in text
|
|
assert "转义列：trigger" in text
|
|
|
|
|
|
def test_main_backs_up_existing_output(tmp_path):
|
|
source = build(tmp_path / "character.csv")
|
|
output = tmp_path / "character_001.csv"
|
|
output.write_text("placeholder\n", encoding="utf-8")
|
|
|
|
assert character_script.main(["-i", source, "-o", str(output)]) == 0
|
|
backup = tmp_path / f"character_001{BACKUP_SUFFIX}"
|
|
assert backup.read_text(encoding="utf-8") == "placeholder\n"
|
|
|
|
|
|
def test_main_no_backup_flag(tmp_path):
|
|
source = build(tmp_path / "character.csv")
|
|
output = tmp_path / "character_001.csv"
|
|
output.write_text("placeholder\n", encoding="utf-8")
|
|
|
|
assert character_script.main(["-i", source, "-o", str(output), "--no-backup"]) == 0
|
|
assert not (tmp_path / f"character_001{BACKUP_SUFFIX}").exists()
|
|
|
|
|
|
def test_main_refuses_same_path_and_missing_input(tmp_path, capsys):
|
|
same = build(tmp_path / "same.csv")
|
|
assert character_script.main(["-i", same, "-o", same]) == 1
|
|
assert "同一个文件" in capsys.readouterr().out
|
|
assert character_script.main(["-i", str(tmp_path / "nope.csv")]) == 1
|
|
assert "输入文件不存在" in capsys.readouterr().out
|
|
|
|
|
|
def test_main_reports_missing_column(tmp_path, capsys):
|
|
source = build(tmp_path / "bad.csv", [["character", "count"], ["a", "50"]])
|
|
assert character_script.main(["-i", source, "-o", str(tmp_path / "o.csv")]) == 1
|
|
assert "trigger" in capsys.readouterr().out
|
|
|
|
|
|
# ----------------------------------------------------------------------
|
|
# 仓库里已生成的处理结果
|
|
# ----------------------------------------------------------------------
|
|
def test_processed_character_csv_shape():
|
|
"""处理后的角色 CSV（若已生成）必须符合：五列、count>=30、trigger 已转义。"""
|
|
path = character_script.DEFAULT_OUTPUT
|
|
if not os.path.isfile(path):
|
|
pytest.skip("尚未生成 danbooru_character_001.csv")
|
|
|
|
with open(path, encoding="utf-8-sig", newline="") as handle:
|
|
rows = list(csv.reader(handle))
|
|
header, body = rows[0], rows[1:]
|
|
assert header == ["character", "copyright", "trigger", "core_tags", "count"]
|
|
assert body, "处理结果为空"
|
|
assert all(len(row) == 5 for row in body)
|
|
assert min(int(row[4]) for row in body) >= 30
|
|
assert not any(" " in row[2] for row in body), "trigger 列不应包含空格"
|
|
assert not any("(" in row[2].replace("\\(", "") for row in body), "trigger 列不应包含未转义括号"
|
|
# 脏下划线守卫：,_ / _, / __ / 首尾下划线
|
|
assert not any(",_" in row[2] or "_," in row[2] for row in body), "trigger 列不应出现逗号紧邻的下划线"
|
|
assert not any("__" in row[2] for row in body), "trigger 列不应出现连续下划线"
|
|
assert not any(row[2].startswith("_") or row[2].endswith("_") for row in body), (
|
|
"trigger 列不应以短横线/空格转来的下划线开头或结尾"
|
|
)
|