feat(collect): persist product status snapshots

app/product_status.py

@@ -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,
+    }