T-566 封面历史归档数据层

This commit is contained in:
chengma
2026-07-09 11:23:18 +08:00
parent fa27134f7c
commit 71e51d0608
6 changed files with 354 additions and 12 deletions
+65 -5
View File
@@ -292,6 +292,34 @@ def _now() -> str:
return datetime.now().isoformat(timespec="seconds")
def _cover_archive_timestamp() -> str:
return datetime.now().strftime("%Y%m%d%H%M%S")
def _archive_generated_cover_file(file_path) -> Optional[str]:
path_text = str(file_path or "").strip()
if not path_text or not os.path.isfile(path_text):
return None
directory = os.path.dirname(path_text)
filename = os.path.basename(path_text)
stem, ext = os.path.splitext(filename)
timestamp = _cover_archive_timestamp()
suffix = 1
while True:
archive_name = f"{stem}_{timestamp}{ext}" if suffix == 1 else f"{stem}_{timestamp}_{suffix}{ext}"
archive_path = os.path.join(directory, archive_name)
if not os.path.exists(archive_path):
break
suffix += 1
try:
os.rename(path_text, archive_path)
except PermissionError as exc:
raise DbError("请先关闭正在查看的封面图片再重置") from exc
except OSError as exc:
raise DbError(f"归档当前封面图片失败: {exc}") from exc
return archive_path
def _db_path(path=None) -> str:
return path or appconfig.db_path()
@@ -846,6 +874,38 @@ def update_generated_title(task_id, new_title, path=None, conn=None) -> None:
(title, now, int(task_id)),
)
def update_generated_cover(task_id, new_cover_path, path=None, conn=None) -> None:
"""Point a task at an existing local generated cover without copying the file."""
cover_path = str(new_cover_path or "").strip()
if not cover_path:
raise DbError("新封面路径不能为空")
abs_cover_path = os.path.abspath(cover_path)
if not os.path.isfile(abs_cover_path):
raise DbError("新封面图片不存在")
with _connection(conn, path) as database:
task = get_task(task_id, conn=database)
if task is None:
raise DbError(f"任务不存在: {task_id}")
if task.status == "running":
raise DbError("任务正在运行,不能修改新封面")
now = _now()
with database:
database.execute(
"""
UPDATE tasks
SET new_cover_path = ?,
stage = 'generated',
status = 'pending',
last_error = NULL,
updated_at = ?
WHERE id = ?
""",
(abs_cover_path, now, int(task_id)),
)
def set_applied(task_id, committed, error=None, path=None, conn=None, step=None) -> None:
now = _now()
success = bool(committed) and error is None
@@ -901,10 +961,9 @@ def reset_generated(
if before is None:
raise DbError(f"任务不存在: {task_id}")
new_cover_path = before.new_cover_path
deleted_file = None
if delete_file and reset_cover and new_cover_path and os.path.isfile(str(new_cover_path)):
os.remove(str(new_cover_path))
deleted_file = new_cover_path
archived_file = None
if reset_cover:
archived_file = _archive_generated_cover_file(new_cover_path)
now = _now()
new_title = None if reset_title else before.new_title
new_cover = None if reset_cover else before.new_cover_path
@@ -932,7 +991,8 @@ def reset_generated(
"before": before,
"after": after,
"new_cover_path": new_cover_path,
"deleted_file": deleted_file,
"archived_file": archived_file,
"deleted_file": None,
"reset_title": bool(reset_title),
"reset_cover": bool(reset_cover),
}
+38
View File
@@ -1,4 +1,5 @@
import os
import re
from .config import make_slug
@@ -22,6 +23,27 @@ def task_image_path(image_root, task, account=None, suffix="old", ext=".jpg"):
)
def list_task_cover_candidates(image_root, task, account=None):
"""Return generated cover candidates for one task from the image directory."""
canonical_path = task_image_path(image_root, task, account=account, suffix="new", ext=".jpg")
directory = os.path.dirname(canonical_path)
prefix = os.path.splitext(os.path.basename(canonical_path))[0]
if not os.path.isdir(directory):
return []
candidates = []
for filename in os.listdir(directory):
full_path = os.path.join(directory, filename)
if not os.path.isfile(full_path):
continue
stem, ext = os.path.splitext(filename)
if ext.lower() != ".jpg":
continue
if stem == prefix or stem.startswith(prefix + "_"):
candidates.append(os.path.abspath(full_path))
return sorted(candidates, key=lambda path: _cover_candidate_sort_key(path, prefix))
def _account_slug(account, task):
slug = _get(account, "slug")
if slug:
@@ -32,6 +54,22 @@ def _account_slug(account, task):
return "unknown_account"
def _cover_candidate_sort_key(path, prefix):
filename = os.path.basename(path)
stem, _ext = os.path.splitext(filename)
if stem == prefix:
return (0, 0, 0, filename)
marker = prefix + "_"
if stem.startswith(marker):
suffix = stem[len(marker):]
match = re.fullmatch(r"(\d{14})(?:_(\d+))?", suffix)
if match:
timestamp = int(match.group(1))
collision = int(match.group(2) or 1)
return (1, -timestamp, -collision, filename)
return (2, 0, 0, filename)
def _get(obj, name, default=None):
if obj is None:
return default
+17 -2
View File
@@ -3,7 +3,7 @@ id: T-566
title: 封面历史归档数据层:重置改名保留旧图 + update_generated_cover + 目录枚举
phase: 7
deps: [T-564]
status: TODO
status: DONE
created: 2026-07-09
---
@@ -84,4 +84,19 @@ created: 2026-07-09
## 执行记录
(做完在这里写:改了什么文件、跑了什么验证命令及结果、遇到的阻塞、关键决策。)
- 2026-07-09 完成 T-566。
- 修改 `app/db.py`:
- `reset_generated(..., reset_cover=True)` 在清空 DB 指针前先把当前 `new_cover_path` 指向的本地文件改名归档为 `{stem}_YYYYMMDDHHMMSS{ext}`;同秒撞名追加 `_2/_3`;旧 `delete_file=True` 不再物理删除,归档取代删除,返回结果新增 `archived_file`,保留 `deleted_file=None` 兼容旧调用。
- 文件不存在时幂等跳过归档;`PermissionError` 抛中文 `DbError` 并保持 DB 指针不变。
- 新增 `update_generated_cover(task_id, new_cover_path)`,只切换当前生效封面指针,写入后退回 `generated/pending`,不增加 attempts,不清 `committed/applied_at/apply_attempts/image_task_id/image_task_key`。
- 修改 `app/image_paths.py`:新增 `list_task_cover_candidates(image_root, task, account=None)`,只读扫描任务目录下 `_new*.jpg` 候选,返回绝对路径,排序为规范名优先、归档时间倒序、同秒后缀数字倒序、异常文件名最后。
- 修改测试:
- `tests/test_db.py` 覆盖归档撞名、文件锁失败 DB 不变、缺失文件幂等、`update_generated_cover` 状态语义、运行中拒绝切换、批次软删除不删图片目录。
- `tests/test_image_paths.py` 覆盖候选图过滤和稳定排序。
- `tests/test_gui.py` 同步重置后本地文件“归档保留而非原名保留”的新语义。
- 验证通过:
- `py -3.10 -m unittest tests.test_db tests.test_image_paths`
- `python -m ruff check app tests main.py`
- `py -3.10 -m compileall app main.py`
- `py -3.10 -m unittest discover -s tests`(285 tests;PySide6 字体目录警告不影响结果)
- `git diff --check`
+183 -2
View File
@@ -1,6 +1,7 @@
import os
import sys
import unittest
from unittest import mock
sys.path.insert(0, os.path.dirname(__file__))
@@ -302,8 +303,10 @@ class DbTests(TempDirMixin, unittest.TestCase):
reset_generated = db.reset_generated(task.id, path=db_path)
after_generated = reset_generated["after"]
self.assertEqual(new_cover, reset_generated["new_cover_path"])
self.assertIsNotNone(reset_generated["archived_file"])
self.assertIsNone(reset_generated["deleted_file"])
self.assertTrue(os.path.exists(new_cover))
self.assertFalse(os.path.exists(new_cover))
self.assertTrue(os.path.exists(reset_generated["archived_file"]))
self.assertEqual("generated", after_generated.stage)
self.assertEqual("success", after_generated.status)
self.assertIsNone(after_generated.new_title)
@@ -378,8 +381,10 @@ class DbTests(TempDirMixin, unittest.TestCase):
self.assertEqual("success", after_cover.status)
self.assertEqual("手动标题B", after_cover.new_title)
self.assertIsNone(after_cover.new_cover_path)
self.assertEqual(second_cover, cover_reset["deleted_file"])
self.assertIsNone(cover_reset["deleted_file"])
self.assertIsNotNone(cover_reset["archived_file"])
self.assertFalse(os.path.exists(second_cover))
self.assertTrue(os.path.exists(cover_reset["archived_file"]))
with self.assertRaises(db.DbError):
db.reset_generated(first.id, reset_title=False, reset_cover=False, path=db_path)
@@ -389,6 +394,175 @@ class DbTests(TempDirMixin, unittest.TestCase):
self.assert_removed(temp_dir)
def test_reset_generated_archives_cover_with_collision_name(self):
with self.make_temp_dir() as temp_dir:
db_path = os.path.join(temp_dir, "cmshopee.db")
db.init_db(db_path)
batch_id = db.create_batch(["input.xlsx"], path=db_path)
db.insert_tasks(
batch_id,
[
{
"source_file_abs": os.path.join(temp_dir, "input.xlsx"),
"source_sheet": "Sheet1",
"source_row": 2,
"account_name": "shop",
"alias": "alias",
"item_id": "51100639510",
}
],
path=db_path,
)
task = db.list_tasks(batch_id=batch_id, path=db_path)[0]
cover = os.path.join(temp_dir, "42_51100639510_new.jpg")
collision = os.path.join(temp_dir, "42_51100639510_new_20260709094800.jpg")
expected_archive = os.path.join(temp_dir, "42_51100639510_new_20260709094800_2.jpg")
with open(cover, "wb") as fh:
fh.write(b"current")
with open(collision, "wb") as fh:
fh.write(b"existing")
db.set_collected(task.id, "旧标题", "old.jpg", path=db_path)
db.set_generated(task.id, "新标题", cover, path=db_path)
with mock.patch("app.db._cover_archive_timestamp", return_value="20260709094800"):
result = db.reset_generated(
task.id,
reset_title=False,
reset_cover=True,
path=db_path,
)
self.assertEqual(expected_archive, result["archived_file"])
self.assertFalse(os.path.exists(cover))
self.assertTrue(os.path.exists(collision))
self.assertTrue(os.path.exists(expected_archive))
with open(collision, "rb") as fh:
self.assertEqual(b"existing", fh.read())
with open(expected_archive, "rb") as fh:
self.assertEqual(b"current", fh.read())
after = db.get_task(task.id, path=db_path)
self.assertEqual("新标题", after.new_title)
self.assertIsNone(after.new_cover_path)
self.assert_removed(temp_dir)
def test_reset_generated_permission_error_keeps_db_pointer(self):
with self.make_temp_dir() as temp_dir:
db_path = os.path.join(temp_dir, "cmshopee.db")
db.init_db(db_path)
batch_id = db.create_batch(["input.xlsx"], path=db_path)
db.insert_tasks(
batch_id,
[
{
"source_file_abs": os.path.join(temp_dir, "input.xlsx"),
"source_sheet": "Sheet1",
"source_row": 2,
"account_name": "shop",
"alias": "alias",
"item_id": "51100639510",
}
],
path=db_path,
)
task = db.list_tasks(batch_id=batch_id, path=db_path)[0]
cover = os.path.join(temp_dir, "new.jpg")
with open(cover, "wb") as fh:
fh.write(b"jpeg")
db.set_generated(task.id, "新标题", cover, path=db_path)
with mock.patch("app.db.os.rename", side_effect=PermissionError):
with self.assertRaisesRegex(db.DbError, "请先关闭正在查看的封面图片再重置"):
db.reset_generated(task.id, reset_title=False, reset_cover=True, path=db_path)
after = db.get_task(task.id, path=db_path)
self.assertEqual(cover, after.new_cover_path)
self.assertEqual("新标题", after.new_title)
self.assertTrue(os.path.exists(cover))
self.assert_removed(temp_dir)
def test_update_generated_cover_switches_pointer_and_keeps_apply_history(self):
with self.make_temp_dir() as temp_dir:
db_path = os.path.join(temp_dir, "cmshopee.db")
db.init_db(db_path)
batch_id = db.create_batch(["input.xlsx"], path=db_path)
db.insert_tasks(
batch_id,
[
{
"source_file_abs": os.path.join(temp_dir, "input.xlsx"),
"source_sheet": "Sheet1",
"source_row": 2,
"account_name": "shop",
"alias": "alias",
"item_id": "51100639510",
}
],
path=db_path,
)
task = db.list_tasks(batch_id=batch_id, path=db_path)[0]
current_cover = os.path.join(temp_dir, "current.jpg")
selected_cover = os.path.join(temp_dir, "selected.jpg")
for cover in (current_cover, selected_cover):
with open(cover, "wb") as fh:
fh.write(b"jpeg")
db.set_collected(task.id, "旧标题", "old.jpg", path=db_path)
db.set_generated(task.id, "新标题", current_cover, path=db_path)
key = db.ensure_image_task_key(task.id, path=db_path)
db.set_image_task_submitted(task.id, "cmhub-task-1", key, path=db_path)
db.set_applied(task.id, True, path=db_path)
before = db.get_task(task.id, path=db_path)
db.update_generated_cover(task.id, selected_cover, path=db_path)
after = db.get_task(task.id, path=db_path)
self.assertEqual(os.path.abspath(selected_cover), after.new_cover_path)
self.assertEqual("generated", after.stage)
self.assertEqual("pending", after.status)
self.assertIsNone(after.last_error)
self.assertEqual(before.generate_attempts, after.generate_attempts)
self.assertEqual(before.apply_attempts, after.apply_attempts)
self.assertEqual(before.committed, after.committed)
self.assertEqual(before.applied_at, after.applied_at)
self.assertEqual(before.image_task_id, after.image_task_id)
self.assertEqual(before.image_task_key, after.image_task_key)
with self.assertRaisesRegex(db.DbError, "新封面图片不存在"):
db.update_generated_cover(task.id, os.path.join(temp_dir, "missing.jpg"), path=db_path)
self.assert_removed(temp_dir)
def test_update_generated_cover_rejects_running_task(self):
with self.make_temp_dir() as temp_dir:
db_path = os.path.join(temp_dir, "cmshopee.db")
db.init_db(db_path)
batch_id = db.create_batch(["input.xlsx"], path=db_path)
db.insert_tasks(
batch_id,
[
{
"source_file_abs": os.path.join(temp_dir, "input.xlsx"),
"source_sheet": "Sheet1",
"source_row": 2,
"account_name": "shop",
"alias": "alias",
"item_id": "51100639510",
}
],
path=db_path,
)
task = db.list_tasks(batch_id=batch_id, path=db_path)[0]
cover = os.path.join(temp_dir, "selected.jpg")
with open(cover, "wb") as fh:
fh.write(b"jpeg")
db.mark_running(task.id, "generate", path=db_path)
with self.assertRaisesRegex(db.DbError, "任务正在运行"):
db.update_generated_cover(task.id, cover, path=db_path)
self.assert_removed(temp_dir)
def test_delete_batch_soft_hides_batch_and_tasks(self):
with self.make_temp_dir() as temp_dir:
db_path = os.path.join(temp_dir, "cmshopee.db")
@@ -419,6 +593,11 @@ class DbTests(TempDirMixin, unittest.TestCase):
tasks = db.list_tasks(batch_id=batch_id, path=db_path)
old_cover = os.path.join(temp_dir, "old.jpg")
new_cover = os.path.join(temp_dir, "new.jpg")
batch_image_dir = os.path.join(temp_dir, "images", batch_id, "alias")
os.makedirs(batch_image_dir, exist_ok=True)
generated_cover = os.path.join(batch_image_dir, "1_51100639510_new.jpg")
with open(generated_cover, "wb") as fh:
fh.write(b"jpeg")
db.set_collected(tasks[0].id, "旧标题", old_cover, path=db_path)
db.set_generated(tasks[0].id, "新标题", new_cover, path=db_path)
db.set_applied(tasks[0].id, True, path=db_path)
@@ -440,6 +619,8 @@ class DbTests(TempDirMixin, unittest.TestCase):
self.assertEqual("导错文件", deleted_batch.deleted_reason)
deleted_tasks = db.list_tasks(batch_id=batch_id, path=db_path, include_deleted=True)
self.assertEqual(2, len(deleted_tasks))
self.assertTrue(os.path.exists(batch_image_dir))
self.assertTrue(os.path.exists(generated_cover))
self.assert_removed(temp_dir)
+8 -1
View File
@@ -2151,7 +2151,14 @@ class GuiTests(TempDirMixin, unittest.TestCase):
self.assertEqual("success", updated.status)
self.assertIsNone(updated.new_title)
self.assertIsNone(updated.new_cover_path)
self.assertTrue(os.path.exists(new_cover))
self.assertFalse(os.path.exists(new_cover))
archived_covers = [
filename
for filename in os.listdir(temp_dir)
if filename.startswith("new_") and filename.endswith(".jpg")
]
self.assertEqual(1, len(archived_covers))
self.assertTrue(os.path.exists(os.path.join(temp_dir, archived_covers[0])))
run_log = db.list_run_logs(limit=1, run_type="reset", path=cfg["db_path"])[0]
self.assertEqual("done", run_log.status)
self.assertEqual("reset_generated", run_log.options["action"])
+41
View File
@@ -1,5 +1,6 @@
import os
import sys
import tempfile
import unittest
from types import SimpleNamespace
@@ -33,6 +34,46 @@ class ImagePathTests(unittest.TestCase):
path,
)
def test_list_task_cover_candidates_filters_and_sorts_generated_covers(self):
with tempfile.TemporaryDirectory() as temp_dir:
task = SimpleNamespace(id=42, batch_id="batch-a", item_id="51100639510", alias="alias-a")
canonical = image_paths.task_image_path(temp_dir, task, suffix="new")
directory = os.path.dirname(canonical)
prefix = os.path.splitext(os.path.basename(canonical))[0]
os.makedirs(directory, exist_ok=True)
filenames = [
f"{prefix}_20260709094800.jpg",
f"{prefix}_20260709094800_2.jpg",
f"{prefix}_20260709094900.jpg",
f"{prefix}.jpg",
f"{prefix}_bad.jpg",
f"{prefix.replace('_new', '_old')}.jpg",
f"{prefix}.png",
"99_51100639510_new.jpg",
]
for filename in filenames:
with open(os.path.join(directory, filename), "wb") as fh:
fh.write(b"jpeg")
paths = image_paths.list_task_cover_candidates(temp_dir, task)
self.assertEqual(
[
canonical,
os.path.join(directory, f"{prefix}_20260709094900.jpg"),
os.path.join(directory, f"{prefix}_20260709094800_2.jpg"),
os.path.join(directory, f"{prefix}_20260709094800.jpg"),
os.path.join(directory, f"{prefix}_bad.jpg"),
],
paths,
)
def test_list_task_cover_candidates_returns_empty_for_missing_directory(self):
with tempfile.TemporaryDirectory() as temp_dir:
task = SimpleNamespace(id=42, batch_id="batch-a", item_id="51100639510", alias="alias-a")
self.assertEqual([], image_paths.list_task_cover_candidates(temp_dir, task))
if __name__ == "__main__":
unittest.main()