feat(ingestion): import bypos-collector JSONL into goods

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-06-24 02:33:45 +00:00
parent a2ef7319e9
commit f04da0a135
5 changed files with 518 additions and 0 deletions
@@ -0,0 +1,116 @@
+"""Adapter for the bypos-collector output (central product library zc.bypos.net).
+
+The ``bypos-collector`` tool (see ``tools/bypos-collector``) queries the same
+central product library that the 云店 POS uses when adding a product by barcode,
+and writes one JSON object per line (JSONL) with these fields::
+
+    barcode name spec unit area manufacturer license
+    in_price sell_price status retmsg fetched_at source
+
+This module turns one such record into the internal product shape consumed by
+:func:`opengoods.etl.load.load_bypos_record`. It is a pure function (no DB, no
+network) so it is easy to unit-test.
+
+The central library is a Chinese retail catalogue: it provides 品名/规格/单位/
+产地/厂商/建议价 but **not** ingredients or nutrition, so no ``food`` block is
+produced.
+"""
+
+from __future__ import annotations
+
+from decimal import Decimal, InvalidOperation
+
+from opengoods.etl.transform import _clamp, is_valid_gtin, parse_quantity
+from opengoods.units import UnitError, normalize
+
+SOURCE_NAME = "bypos中心库"
+SOURCE_HOMEPAGE = "https://zc.bypos.net"
+# Vendor catalogue data; not an open-licensed dataset. Display facts only.
+SOURCE_LICENSE = "proprietary"
+SOURCE_TRUST = 0.6
+
+
+def _to_price(text: str | None) -> Decimal | None:
+    """Parse a price string to a positive Decimal, or None for empty/zero."""
+    if not text:
+        return None
+    try:
+        amount = Decimal(str(text).strip())
+    except (InvalidOperation, ValueError):
+        return None
+    if amount <= 0:
+        return None
+    return amount
+
+
+def _net_content(spec: str | None) -> tuple[Decimal | None, str | None, Decimal | None]:
+    """Best-effort parse a spec like '500mL'/'5kg' to (value, unit, canonical).
+
+    Packaging-style specs ('20支', '1X24', '盒') have no mass/volume unit and
+    yield ``(None, None, None)`` — the raw spec is kept in attributes instead.
+    """
+    parsed = parse_quantity(spec or "")
+    if not parsed:
+        return None, None, None
+    value, unit = parsed
+    try:
+        norm = normalize(value, unit)
+    except UnitError:
+        return None, None, None
+    return norm.value, norm.unit, norm.canonical_value
+
+
+def transform_bypos(rec: dict) -> dict | None:
+    """Transform one bypos-collector JSONL record into an internal product dict.
+
+    Returns ``None`` for non-hit rows or rows without a usable name.
+    """
+    if rec.get("status") != "hit":
+        return None
+    name = (rec.get("name") or "").strip()
+    if not name:
+        return None
+
+    barcode = str(rec.get("barcode") or "").strip()
+    gtin = barcode if barcode and is_valid_gtin(barcode) else None
+
+    net_value, net_unit, net_canonical = _net_content(rec.get("spec"))
+
+    spec = (rec.get("spec") or "").strip()
+    pack_unit = (rec.get("unit") or "").strip()
+    area = (rec.get("area") or "").strip()
+    manufacturer = (rec.get("manufacturer") or "").strip() or None
+    license_no = (rec.get("license") or "").strip()
+    in_price = _to_price(rec.get("in_price"))
+    sell_price = _to_price(rec.get("sell_price"))
+
+    attributes: dict[str, object] = {}
+    if spec:
+        attributes["spec"] = spec
+    if pack_unit:
+        attributes["pack_unit"] = pack_unit
+    if area:
+        attributes["origin_area"] = area
+    if license_no:
+        attributes["production_license"] = license_no
+    if in_price is not None:
+        attributes["suggested_in_price"] = float(in_price)
+    if sell_price is not None:
+        attributes["suggested_retail_price"] = float(sell_price)
+
+    # Domestic GS1-China barcodes (69x) are China-made; area is a province/city,
+    # kept separately in attributes.origin_area.
+    country = "中国" if gtin and gtin.startswith("69") else None
+
+    return {
+        "gtin": gtin,
+        "name": name,
+        "manufacturer": manufacturer,
+        "net_content_value": net_value,
+        "net_content_unit": _clamp(net_unit, 16),
+        "net_content_canonical": net_canonical,
+        "country_of_origin": country,
+        "attributes": attributes,
+        "msrp": sell_price,
+        "fetched_at": (rec.get("fetched_at") or "").strip() or None,
+    }