AI核对部分退款：按退款数量分摊优惠金额

{
  "policy": "优惠按各商品行原价金额比例分摊；分位余数按商品ID升序补给，退款按行内件数平均分摊。",
  "lines": [
    {
      "id": "A",
      "qty": 2,
      "unit_price": "10.00"
    },
    {
      "id": "B",
      "qty": 1,
      "unit_price": "10.00"
    }
  ],
  "discount": "1.00",
  "return": {
    "line": "A",
    "qty": 1
  }
}

python3 -m venv .venv
source .venv/bin/activate
python -m pip install -r requirements.txt
python run.py

{
  "model": "gemini-2.5-flash-lite",
  "messages": [
    {
      "role": "system",
      "content": "严格只输出单个JSON对象，不要Markdown代码围栏，不要解释。只处理给定原创虚构材料，只输出JSON。提取退款约定：allocation_basis为line_original_amount，rounding为largest_remainder_cent，refund_units正整数。优惠按原价金额比例分摊到商品行，再按行内件数分摊，分位余数给排序更前的行/件。 输出结构：{\"allocation_basis\": \"string\", \"rounding\": \"string\", \"refund_units\": \"number\"}"
    },
    {
      "role": "user",
      "content": "任务约束（只输出JSON对象，不用代码围栏）：\n只处理给定原创虚构材料，只输出JSON。提取退款约定：allocation_basis为line_original_amount，rounding为largest_remainder_cent，refund_units正整数。优惠按原价金额比例分摊到商品行，再按行内件数分摊，分位余数给排序更前的行/件。 输出结构：{\"allocation_basis\": \"string\", \"rounding\": \"string\", \"refund_units\": \"number\"}\n待处理合成材料：\n{\"policy\": \"优惠按各商品行原价金额比例分摊；分位余数按商品ID升序补给，退款按行内件数平均分摊。\", \"lines\": [{\"id\": \"A\", \"qty\": 2, \"unit_price\": \"10.00\"}, {\"id\": \"B\", \"qty\": 1, \"unit_price\": \"10.00\"}], \"discount\": \"1.00\", \"return\": {\"line\": \"A\", \"qty\": 1}}"
    }
  ],
  "temperature": 0,
  "max_tokens": 1024,
  "provider": {
    "sort": "success_rate"
  },
  "response_format": {
    "type": "json_object"
  }
}

lines = data['lines']
gross = {r['id']: D(r['unit_price']) * r['qty'] for r in lines}
total = sum(gross.values())
discount_cents = int(D(data['discount']) * 100)
raw = {k: D(discount_cents) * v / total for k, v in gross.items()}
allocated = {k: int(v) for k, v in raw.items()}
remainder = discount_cents - sum(allocated.values())
for key in sorted(raw, key=lambda k: (-(raw[k] - int(raw[k])), k))[:remainder]:
    allocated[key] += 1
rows = []
refund = 0
for line in lines:
    base, extra = divmod(allocated[line['id']], line['qty'])
    for ordinal in range(1, line['qty'] + 1):
        off = base + (ordinal <= extra)
        net = int(D(line['unit_price']) * 100) - off
        returned = line['id'] == data['return']['line'] and ordinal <= output['refund_units']
        if returned:
            refund += net
        rows.append({'line': line['id'], 'unit': ordinal, 'original_cents': int(D(line['unit_price']) * 100), 'discount_cents': off, 'net_cents': net, 'returned': returned})
paid = int(total * 100) - discount_cents
remaining = sum((r['net_cents'] for r in rows if not r['returned']))
check('优惠总额分位守恒', sum((r['discount_cents'] for r in rows)) == discount_cents)
check('行分摊A67分B33分', allocated == {'A': 67, 'B': 33})
check('退款加剩余货款仍等于实付', refund + remaining == paid)
check('部分退货金额9.66元', refund == 966)
cout('unit-refund-ledger.csv', rows)
result = {'line_discount_cents': allocated, 'refund': money(D(refund) / 100), 'remaining_paid': money(D(remaining) / 100), 'original_paid': money(D(paid) / 100), 'rounding_note': '先分摊优惠到商品行，再把行优惠分摊到各件；相同余数按ID/件序升序。'}

{
  "line_discount_cents": {
    "A": 67,
    "B": 33
  },
  "refund": "9.66",
  "remaining_paid": "19.34",
  "original_paid": "29.00",
  "rounding_note": "先分摊优惠到商品行，再把行优惠分摊到各件；相同余数按ID/件序升序。"
}