Joey
195 lines
8.3 KiB
Python
195 lines
8.3 KiB
Python
"""出图方案规划器：DeepSeek 根据采集的商品信息生成套图方案。
|
||
|
||
方案每项 = 一类图（标题 + 说明 + 生图提示 + 张数 + 可选 SKU 绑定），
|
||
生成时按方案逐张出图；参考图可按 variant_name 精确绑定到对应 SKU 图。
|
||
"""
|
||
from __future__ import annotations
|
||
|
||
import json
|
||
import logging
|
||
import re
|
||
|
||
import httpx
|
||
|
||
from config import get_settings
|
||
|
||
log = logging.getLogger("suite.planner")
|
||
|
||
# 规划器可选用的图类型（与 prompt.py 的 builder 对应）
|
||
ALLOWED_KINDS = [
|
||
"white_bg", "key_features", "selling_pt", "material",
|
||
"lifestyle", "multi_scene", "ecommerce_detail",
|
||
"size_chart", "sku_collection", "custom",
|
||
]
|
||
|
||
SYSTEM_PROMPT = """你是一名资深电商视觉策划。根据商品信息规划一套电商套图的出图方案。
|
||
|
||
## 输出硬性约束（违反即失败）
|
||
1. 输出必须是**单行紧凑 JSON**：无换行、无缩进、无空格填充、无注释、无 markdown 围栏。
|
||
2. 顶层只有 summary 和 items 两个字段；每个 item 严格只有 kind/title/detail/prompt_hint/count/variant_name 六个字段，不得增删。
|
||
3. 文本长度上限（中文字符/英文单词数）：summary ≤ 25 字；title ≤ 8 字；detail ≤ 20 字；prompt_hint ≤ 15 个英文词。超限必须删减，不得省略号截断。
|
||
4. count 默认 1，仅当该类图确需多个变体时才 >1，最大 3。总张数 8-15。
|
||
5. variant_name 只能从「SKU规格」列表原样照抄；没有绑定就输出 null。
|
||
|
||
## 规划规则
|
||
1. SKU 主图：每个带图 SKU 出 1 张独立主图（kind=white_bg），variant_name 填对应规格名；单 SKU 出 1 张（variant_name=null）。
|
||
2. 场景图（kind=lifestyle）：按核心使用场景出 2-4 张，每张聚焦一个场景。
|
||
3. 细节图（kind=material 或 custom）：按关键细节/材质/结构出 2-3 张，每张聚焦一个卖点。
|
||
4. 尺寸标注图（kind=size_chart）：参数含长宽高/尺寸时出 1 张。
|
||
5. SKU 合集图（kind=sku_collection）：SKU >1 时出 1 张。
|
||
6. kind 枚举：white_bg / key_features / selling_pt / material / lifestyle / multi_scene / ecommerce_detail / size_chart / sku_collection / custom。
|
||
7. title 用中文短语（如「主图·粉色」「浴室壁挂」）；detail 中文说明这张图展示什么；prompt_hint 用英文描述构图要点。
|
||
|
||
## 输出示例（紧凑单行）
|
||
{"summary":"三色收纳盒全套图","items":[{"kind":"white_bg","title":"主图·粉色","detail":"粉色SKU白底主视觉","prompt_hint":"front view on white background","count":1,"variant_name":"粉色"}]}"""
|
||
|
||
|
||
def _normalize_items(raw_items: list, sku_variants: list[str]) -> list[dict]:
|
||
"""清洗模型输出：kind 白名单、count 钳制、variant 必须真实存在。"""
|
||
items: list[dict] = []
|
||
for it in raw_items:
|
||
if not isinstance(it, dict):
|
||
continue
|
||
kind = str(it.get("kind") or "custom")
|
||
if kind not in ALLOWED_KINDS:
|
||
kind = "custom"
|
||
title = str(it.get("title") or "").strip()[:20]
|
||
if not title:
|
||
continue
|
||
try:
|
||
count = max(0, min(3, int(it.get("count", 1))))
|
||
except (TypeError, ValueError):
|
||
count = 1
|
||
variant = str(it.get("variant_name") or "").strip() or None
|
||
if variant and variant not in sku_variants:
|
||
variant = None # 幻觉规格：丢弃绑定，回退主图
|
||
items.append({
|
||
"kind": kind,
|
||
"title": title,
|
||
"detail": str(it.get("detail") or "").strip()[:80],
|
||
"prompt_hint": str(it.get("prompt_hint") or "").strip()[:300],
|
||
"count": count,
|
||
"variant_name": variant,
|
||
})
|
||
return items
|
||
|
||
|
||
def _repair_truncated(s: str) -> dict | None:
|
||
"""截断修复：从最后一个完整的 '}' 处截断，剥尾逗号后按括号配平补全闭合。
|
||
|
||
适用于「items 数组中途被 max_tokens 截断」的场景——截断点在完整对象边界，
|
||
此前的字符串必然已闭合，简单计数配平即可。
|
||
"""
|
||
for cut in (m.end() for m in reversed(list(re.finditer(r'\}', s)))):
|
||
cand = s[:cut].rstrip().rstrip(',')
|
||
opens: list[str] = []
|
||
for ch in cand:
|
||
if ch in '{[':
|
||
opens.append(ch)
|
||
elif ch == '}' and opens and opens[-1] == '{':
|
||
opens.pop()
|
||
elif ch == ']' and opens and opens[-1] == '[':
|
||
opens.pop()
|
||
suffix = ''.join('}' if o == '{' else ']' for o in reversed(opens))
|
||
try:
|
||
data = json.loads(cand + suffix)
|
||
if isinstance(data, dict):
|
||
return data
|
||
except json.JSONDecodeError:
|
||
continue
|
||
return None
|
||
|
||
|
||
def _extract_json(text: str) -> dict:
|
||
"""从模型输出提取 JSON：剥离思考块/markdown 围栏，截断时尝试修复。"""
|
||
s = (text or '').strip()
|
||
# 剥离思考块（思考型模型会把推理过程放进 <think>）
|
||
s = re.sub(r'<think>.*?</think>', '', s, flags=re.S).strip()
|
||
# 剥离 markdown 代码围栏
|
||
m = re.search(r'```(?:json)?\s*(.*?)```', s, flags=re.S)
|
||
if m:
|
||
s = m.group(1).strip()
|
||
try:
|
||
return json.loads(s)
|
||
except json.JSONDecodeError:
|
||
pass
|
||
start = s.find('{')
|
||
if start >= 0:
|
||
repaired = _repair_truncated(s[start:])
|
||
if repaired is not None:
|
||
log.warning("规划器输出疑似被截断，已自动截断修复（可能丢失末尾部分方案项）")
|
||
return repaired
|
||
raise ValueError("模型输出无法解析为 JSON")
|
||
|
||
|
||
async def generate_plan(
|
||
product_info: dict,
|
||
sku_variants: list[str],
|
||
image_stats: dict,
|
||
platform: str,
|
||
) -> dict:
|
||
"""调用 DeepSeek 生成方案。返回 {summary, items}。"""
|
||
s = get_settings()
|
||
if not s.deepseek_api_key:
|
||
raise RuntimeError("未配置 DEEPSEEK_API_KEY（.env）")
|
||
|
||
user_content = json.dumps({
|
||
"商品信息": product_info, # {title, desc, params:[{key,value}], sellingPoints, price}
|
||
"SKU规格": sku_variants, # 带图的 SKU 规格名（variant_name 只能从中选）
|
||
"图片统计": image_stats, # {main: n, sku: n, detail: n}
|
||
"目标平台": platform, # ozon/wb/cn（决定图内文案语言）
|
||
}, ensure_ascii=False)
|
||
|
||
async with httpx.AsyncClient(timeout=90, verify=False) as client:
|
||
resp = await client.post(
|
||
f"{s.deepseek_base_url.rstrip('/')}/chat/completions",
|
||
headers={"Authorization": f"Bearer {s.deepseek_api_key}", "Content-Type": "application/json"},
|
||
json={
|
||
"model": s.deepseek_model,
|
||
"messages": [
|
||
{"role": "system", "content": SYSTEM_PROMPT},
|
||
{"role": "user", "content": user_content},
|
||
],
|
||
"response_format": {"type": "json_object"},
|
||
"temperature": 0.3,
|
||
"max_tokens": 8000,
|
||
},
|
||
)
|
||
resp.raise_for_status()
|
||
body = resp.json()
|
||
message = body["choices"][0]["message"]
|
||
finish_reason = body["choices"][0].get("finish_reason", "")
|
||
usage = body.get("usage") or {}
|
||
log.info(
|
||
"规划器 token 用量: prompt=%s completion=%s finish=%s",
|
||
usage.get("prompt_tokens", "?"), usage.get("completion_tokens", "?"), finish_reason,
|
||
)
|
||
content = message.get("content") or ""
|
||
# 思考型输出：content 为空时从 reasoning_content 里捞
|
||
if not content.strip() and message.get("reasoning_content"):
|
||
content = message["reasoning_content"]
|
||
|
||
try:
|
||
data = _extract_json(content)
|
||
except ValueError as exc:
|
||
log.error(
|
||
"规划器输出解析失败 finish_reason=%s content[:200]=%s",
|
||
finish_reason, content[:200],
|
||
)
|
||
raise RuntimeError("规划器输出解析失败") from exc
|
||
|
||
items = _normalize_items(data.get("items") or [], sku_variants)
|
||
if not items:
|
||
raise RuntimeError("规划器未返回有效方案项")
|
||
# 总量保护：超过 18 张时按比例截断
|
||
total = sum(i["count"] for i in items)
|
||
while total > 18 and items:
|
||
last = items[-1]
|
||
if last["count"] > 1:
|
||
last["count"] -= 1
|
||
else:
|
||
items.pop()
|
||
total = sum(i["count"] for i in items)
|
||
|
||
return {"summary": str(data.get("summary") or "").strip()[:100], "items": items}
|