chengma/cmshoppe
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat(collect): persist product status snapshots

Browse Source
This commit is contained in:
chengma
2026-07-18 16:34:37 +08:00
parent b4250f36f8
commit 0cf23db403
11 changed files with 464 additions and 12 deletions
Download Patch File Download Diff File Expand all files Collapse all files
app/db.py
+81 -3
View File
@@ -11,7 +11,7 @@ from dataclasses import dataclass
from datetime import datetime
from typing import Iterable, Optional
from . import appconfig
from . import appconfig, product_status
from .config import make_slug
@@ -30,6 +30,7 @@ FAILURE_STEP_LABELS = {
"open_product": "打开商品页",
"wait_ready": "页面就绪",
"read_title": "读标题",
"read_product_status": "读取商品状态",
"read_cover": "读封面",
"download_cover": "下载封面",
"db_write": "写库",
@@ -135,6 +136,9 @@ class Task:
account_name: Optional[str]
alias: str
item_id: str
product_status: Optional[str]
product_status_note: Optional[str]
product_status_at: Optional[str]
old_title: Optional[str]
old_cover_path: Optional[str]
new_title: Optional[str]
@@ -232,6 +236,9 @@ CREATE TABLE IF NOT EXISTS tasks (
account_name TEXT,
alias TEXT NOT NULL,
item_id TEXT NOT NULL,
product_status TEXT,
product_status_note TEXT,
product_status_at TEXT,
old_title TEXT,
old_cover_path TEXT,
new_title TEXT,
@@ -485,6 +492,7 @@ def init_db(path=None, conn=None) -> None:
_ensure_batch_delete_columns(database)
_ensure_task_image_task_columns(database)
_ensure_task_cover_reset_columns(database)
_ensure_task_product_status_columns(database)
_ensure_image_studio_project_suite_columns(database)
_ensure_image_studio_project_draft_columns(database)
_ensure_image_studio_job_recovery_columns(database)
@@ -516,6 +524,16 @@ def _ensure_task_cover_reset_columns(database):
database.execute("ALTER TABLE tasks ADD COLUMN cover_reset_at TEXT")
def _ensure_task_product_status_columns(database):
columns = {row["name"] for row in database.execute("PRAGMA table_info(tasks)").fetchall()}
if "product_status" not in columns:
database.execute("ALTER TABLE tasks ADD COLUMN product_status TEXT")
if "product_status_note" not in columns:
database.execute("ALTER TABLE tasks ADD COLUMN product_status_note TEXT")
if "product_status_at" not in columns:
database.execute("ALTER TABLE tasks ADD COLUMN product_status_at TEXT")
def _ensure_image_studio_project_suite_columns(database):
columns = {
row["name"]
@@ -1021,8 +1039,53 @@ def clear_image_task(task_id, path=None, conn=None) -> None:
)
def set_collected(task_id, old_title, old_cover_path, path=None, conn=None) -> None:
def set_product_status(
task_id,
product_status_value,
product_status_note=None,
product_status_at=None,
path=None,
conn=None,
) -> None:
"""Persist a status snapshot without changing the collection lifecycle."""
now = _now()
detected_at = product_status_at or now
with _connection(conn, path) as database:
with database:
database.execute(
"""
UPDATE tasks
SET product_status = ?,
product_status_note = ?,
product_status_at = ?,
updated_at = ?
WHERE id = ?
""",
(
product_status.normalize_status(product_status_value),
str(product_status_note or "")[:2000] or None,
detected_at,
now,
int(task_id),
),
)
def set_collected(
task_id,
old_title,
old_cover_path,
path=None,
conn=None,
*,
product_status_value=None,
product_status_note=None,
product_status_at=None,
) -> None:
now = _now()
has_status_snapshot = product_status_value is not None
detected_at = product_status_at or now
with _connection(conn, path) as database:
with database:
database.execute(
@@ -1035,10 +1098,25 @@ def set_collected(task_id, old_title, old_cover_path, path=None, conn=None) -> N
last_error = NULL,
collect_attempts = collect_attempts + 1,
collected_at = ?,
product_status = CASE WHEN ? THEN ? ELSE product_status END,
product_status_note = CASE WHEN ? THEN ? ELSE product_status_note END,
product_status_at = CASE WHEN ? THEN ? ELSE product_status_at END,
updated_at = ?
WHERE id = ?
""",
(old_title, old_cover_path, now, now, int(task_id)),
(
old_title,
old_cover_path,
now,
int(has_status_snapshot),
product_status.normalize_status(product_status_value),
int(has_status_snapshot),
str(product_status_note or "")[:2000] or None,
int(has_status_snapshot),
detected_at,
now,
int(task_id),
),
)
app/editor.py
+31 -1
View File
@@ -5,7 +5,7 @@ import os
import time
from urllib.parse import urlparse
from . import appconfig, image_paths
from . import appconfig, image_paths, product_status
from .cdp import (
CDP,
close_tab,
@@ -112,6 +112,17 @@ JS_RECTS = (
"return JSON.stringify(a);})()"
)
JS_PRODUCT_STATUS_ALERTS = (
"(function(){"
"var nodes=[].slice.call(document.querySelectorAll('.eds-alert.eds-alert--warning'));"
"return JSON.stringify(nodes.map(function(node){"
"var title=node.querySelector('.eds-alert-title');"
"var desc=node.querySelector('.eds-alert-desc');"
"return {title:(title&&(title.innerText||title.textContent)||'').trim(),"
"description:(desc&&(desc.innerText||desc.textContent)||'').trim()};"
"}));})()"
)
JS_UPLOAD_STATE = (
"(function(){"
@@ -1153,6 +1164,8 @@ def collect(account, task, on_step=None) -> dict:
cdp = open_product(account, item_id, on_step=on_step, bring_to_front=False)
result = None
try:
_notify_collect_step(on_step, "read_product_status")
status_snapshot = read_product_status(cdp)
_notify_collect_step(on_step, "read_title")
old_title = read_title(cdp)
_notify_collect_step(on_step, "read_cover")
@@ -1171,6 +1184,7 @@ def collect(account, task, on_step=None) -> dict:
"old_title": old_title,
"old_cover_src": old_cover_src,
"old_cover_path": old_cover_path,
**status_snapshot,
}
finally:
close_target_confirmed = _close_collected_product(cdp)
@@ -1178,6 +1192,22 @@ def collect(account, task, on_step=None) -> dict:
return result
def read_product_status(cdp) -> dict:
"""Read the current product's warning state without retaining page HTML."""
try:
alerts = _json_value(cdp, JS_PRODUCT_STATUS_ALERTS, default=None)
if not isinstance(alerts, list):
raise ValueError("商品状态提示读取结果无效")
snapshot = product_status.classify_alerts(alerts)
snapshot["product_status_error"] = None
return snapshot
except Exception as exc:
snapshot = product_status.classify_alerts(None)
snapshot["product_status_error"] = str(exc) or exc.__class__.__name__
return snapshot
def _notify_collect_step(callback, step):
if callback is None:
app/gui/workers.py
+10
View File
@@ -2152,8 +2152,18 @@ class CollectWorker(BaseWorker):
task.id,
result.get("old_title", ""),
result.get("old_cover_path", ""),
product_status_value=result.get("product_status"),
product_status_note=result.get("product_status_note"),
path=self.db_path,
)
if result.get("product_status_error"):
self._write_diagnostic_log(
"商品状态检测失败，已按状态未知保存",
level="WARNING",
step="read_product_status",
task=task,
payload={"error": result.get("product_status_error")},
)
collected += 1
elapsed_ms = self._elapsed_ms(started)
self.row_updated.emit(
app/product_status.py
+109
View File
@@ -0,0 +1,109 @@
"""Shared product-status rules for the collection, generation, and apply flows."""
from __future__ import annotations
import re
STATUS_NORMAL = "normal"
STATUS_UNLISTED = "unlisted"
STATUS_REVIEWING = "reviewing"
STATUS_UNKNOWN = "unknown"
VALID_PRODUCT_STATUSES = frozenset(
{
STATUS_NORMAL,
STATUS_UNLISTED,
STATUS_REVIEWING,
STATUS_UNKNOWN,
}
)
PRODUCT_STATUS_LABELS = {
STATUS_NORMAL: "正常",
STATUS_UNLISTED: "未上架",
STATUS_REVIEWING: "审核中",
STATUS_UNKNOWN: "状态未知",
}
_WHITESPACE_RE = re.compile(r"\s+")
_NOTE_LIMIT = 2000
def normalize_status(value) -> str:
"""Return a stable status code; legacy and invalid values are unknown."""
normalized = str(value or "").strip().lower()
if normalized in VALID_PRODUCT_STATUSES:
return normalized
return STATUS_UNKNOWN
def status_label(value) -> str:
return PRODUCT_STATUS_LABELS[normalize_status(value)]
def is_normal(value) -> bool:
return normalize_status(value) == STATUS_NORMAL
def is_known_abnormal(value) -> bool:
return normalize_status(value) in {STATUS_UNLISTED, STATUS_REVIEWING}
def normalize_text(value) -> str:
return _WHITESPACE_RE.sub(" ", str(value or "").strip())
def classify_alerts(alerts) -> dict:
"""Classify normalized EDS warning-alert data without retaining page HTML."""
if not isinstance(alerts, list):
return _snapshot(STATUS_UNKNOWN, "")
if not alerts:
return _snapshot(STATUS_NORMAL, "")
first_note = ""
valid_alert_seen = False
for alert in alerts:
if not isinstance(alert, dict):
continue
valid_alert_seen = True
title = normalize_text(alert.get("title"))
description = normalize_text(alert.get("description"))
note = _alert_note(title, description)
if not first_note:
first_note = note
if title in {"審核中", "审核中"}:
return _snapshot(STATUS_REVIEWING, note)
if "您的商品未上架" in title:
return _snapshot(STATUS_UNLISTED, note)
if not valid_alert_seen:
return _snapshot(STATUS_UNKNOWN, "")
return _snapshot(STATUS_UNKNOWN, first_note)
def partition_tasks(tasks) -> dict:
"""Return tasks grouped by normalized status for callers' run planners."""
grouped = {status: [] for status in VALID_PRODUCT_STATUSES}
for task in tasks:
grouped[normalize_status(getattr(task, "product_status", None))].append(task)
return grouped
def _alert_note(title: str, description: str) -> str:
parts = []
if title:
parts.append(f"标题：{title}")
if description:
parts.append(f"说明：{description}")
return "；".join(parts)[:_NOTE_LIMIT]
def _snapshot(status: str, note: str) -> dict:
return {
"product_status": normalize_status(status),
"product_status_note": str(note or "")[:_NOTE_LIMIT] or None,
}