feat(collect): show skipped reason summaries

app/collect_skip.py

@@ -0,0 +1,70 @@
+"""Pure helpers for classifying and presenting collection skip reasons."""
+
+from __future__ import annotations
+
+ALIAS_UNMATCHED = "alias_unmatched"
+LOGIN_REQUIRED = "login_required"
+OTHER = "other"
+
+SKIP_REASON_ORDER = (ALIAS_UNMATCHED, LOGIN_REQUIRED, OTHER)
+SKIP_REASON_LABELS = {
+    ALIAS_UNMATCHED: "别名未匹配",
+    LOGIN_REQUIRED: "账号未登录",
+    OTHER: "其他原因",
+}
+
+
+def empty_skip_reason_counts():
+    return {reason: 0 for reason in SKIP_REASON_ORDER}
+
+
+def classify_skip_reason(reason="", *, alias_matched=True):
+    """Classify historical skip text without treating uncertain login checks as logout."""
+    if not alias_matched:
+        return ALIAS_UNMATCHED
+
+    text = str(reason or "").strip().lower()
+    if "别名未匹配" in text or "alias_unmatched" in text:
+        return ALIAS_UNMATCHED
+    if (
+        "账号未登录" in text
+        or "采集中途掉登录" in text
+        or "login_page" in text
+        or ("accounts.shopee." in text and "/seller/login" in text)
+    ):
+        return LOGIN_REQUIRED
+    return OTHER
+
+
+def skipped_stage_text(reason="", *, alias_matched=True):
+    reason_code = classify_skip_reason(reason, alias_matched=alias_matched)
+    return f"略过 · {SKIP_REASON_LABELS[reason_code]}"
+
+
+def normalize_skip_reason_counts(counts=None, *, skipped_total=0):
+    normalized = empty_skip_reason_counts()
+    source = counts if isinstance(counts, dict) else {}
+    for reason in SKIP_REASON_ORDER:
+        try:
+            normalized[reason] = max(0, int(source.get(reason, 0) or 0))
+        except (TypeError, ValueError):
+            normalized[reason] = 0
+
+    total = max(0, int(skipped_total or 0))
+    accounted = sum(normalized.values())
+    if accounted < total:
+        normalized[OTHER] += total - accounted
+    return normalized
+
+
+def format_skip_reason_summary(counts=None, *, skipped_total=0):
+    total = max(0, int(skipped_total or 0))
+    if total == 0:
+        return ""
+    normalized = normalize_skip_reason_counts(counts, skipped_total=total)
+    parts = [
+        f"{SKIP_REASON_LABELS[reason]}{normalized[reason]}"
+        for reason in SKIP_REASON_ORDER
+        if normalized[reason] > 0
+    ]
+    return f"略过{total}：" + "，".join(parts)