更新

backend/rpa_engine/douyin_im/reply_payload.py

@@ -0,0 +1,330 @@
+"""自动回复内容解析与 IM 消息体构造（文本 / 网址 / 卡片）。"""
+
+from __future__ import annotations
+
+import json
+from typing import Any, Tuple
+
+from .message_content import (
+    MSG_TYPE_IMAGE,
+    MSG_TYPE_STICKER,
+    MSG_TYPE_TEXT,
+    message_preview,
+    parse_stored_content,
+    serialize_message_content,
+)
+
+
+def _normalize_reply_spec(data: dict[str, Any]) -> dict[str, Any] | None:
+    if not isinstance(data, dict):
+        return None
+    reply_type = data.get("type")
+    if reply_type in ("text", "link", "card", "image", "sticker"):
+        return data
+    return None
+
+
+def parse_reply_messages(raw: str) -> list[dict[str, Any]]:
+    """解析规则中的 reply_content，支持单条或多条回复。"""
+    raw = (raw or "").strip()
+    if not raw:
+        return [{"type": "text", "text": ""}]
+    if raw.startswith("{") or raw.startswith("["):
+        try:
+            data = json.loads(raw)
+            if isinstance(data, dict) and isinstance(data.get("messages"), list):
+                specs = [_normalize_reply_spec(item) for item in data["messages"]]
+                specs = [item for item in specs if item]
+                if specs:
+                    return specs
+            if isinstance(data, list):
+                specs = [_normalize_reply_spec(item) for item in data]
+                specs = [item for item in specs if item]
+                if specs:
+                    return specs
+            if isinstance(data, dict):
+                spec = _normalize_reply_spec(data)
+                if spec:
+                    return [spec]
+        except json.JSONDecodeError:
+            pass
+    return [{"type": "text", "text": raw}]
+
+
+def parse_reply_content(raw: str) -> dict[str, Any]:
+    """解析发送/规则中的 content。纯字符串视为文本，JSON 为结构化消息。"""
+    raw = (raw or "").strip()
+    if raw.startswith("{"):
+        try:
+            data = json.loads(raw)
+            if isinstance(data, dict):
+                spec = _normalize_reply_spec(data)
+                if spec:
+                    return spec
+        except json.JSONDecodeError:
+            pass
+    parsed = parse_stored_content(raw)
+    if parsed.get("type") != "text":
+        return parsed
+    return parse_reply_messages(raw)[0]
+
+
+def _format_reply_spec(spec: dict[str, Any]) -> str:
+    reply_type = spec.get("type", "text")
+    if reply_type in ("image", "sticker", "voice", "video"):
+        return message_preview(serialize_message_content(spec))
+    if reply_type == "text":
+        return (spec.get("text") or "").strip()
+    if reply_type == "link":
+        text = (spec.get("text") or spec.get("title") or "").strip()
+        url = (spec.get("url") or "").strip()
+        if text and url:
+            return f"{text} → {url}"
+        return text or url
+    if reply_type == "card":
+        title = (spec.get("title") or "").strip()
+        desc = (spec.get("desc") or spec.get("description") or "").strip()
+        page_url = (spec.get("url") or "").strip()
+        if title and page_url:
+            return f"[卡片] {title} → {page_url}"
+        return title or desc or page_url or "[卡片]"
+    return message_preview(serialize_message_content(spec))
+
+
+def format_reply_display(raw: str) -> str:
+    """将 reply_content 格式化为日志/列表中的可读摘要。"""
+    preview = message_preview(raw)
+    if preview:
+        return preview
+    specs = parse_reply_messages(raw)
+    parts = [_format_reply_spec(spec) for spec in specs]
+    parts = [part for part in parts if part]
+    if not parts:
+        return (raw or "").strip()
+    if len(parts) == 1:
+        return parts[0]
+    return " | ".join(parts)
+
+
+def serialize_reply_messages(specs: list[dict[str, Any]]) -> str:
+    cleaned = [spec for spec in specs if _normalize_reply_spec(spec)]
+    if not cleaned:
+        cleaned = [{"type": "text", "text": ""}]
+    if len(cleaned) == 1:
+        return serialize_reply_content(cleaned[0])
+    return json.dumps({"messages": cleaned}, ensure_ascii=False, separators=(",", ":"))
+
+
+def _extract_link_card_media_path(url: str) -> str:
+    value = (url or "").strip()
+    idx = value.find("/api/media/link-cards/")
+    return value[idx:] if idx >= 0 else ""
+
+
+def expand_card_spec(spec: dict[str, Any]) -> list[dict[str, Any]]:
+    """卡片专属发送规则：展开为「封面图片 + 标题/内容/可点击链接文字」两条消息。
+
+    抖音 Web 协议无法发原生合并卡片(type=70 带图→8004/不带图→空白)，因此卡片以
+    「图片消息(横幅) + 文本(标题/内容/链接)」组合呈现：图片提供视觉、文本提供可点击跳转。
+    图片消息仅互关用户可收（陌生人会被 8003 拦截，但文本仍可送达，不影响链接触达）。
+    """
+    title = (spec.get("title") or "").strip()
+    desc = (spec.get("desc") or spec.get("description") or "").strip()
+    target = (
+        spec.get("target_url")
+        or spec.get("url")
+        or spec.get("link_url")
+        or ""
+    ).strip()
+    if "localhost" in target or "127.0.0.1" in target:
+        target = (spec.get("target_url") or spec.get("link_url") or "").strip() or target
+
+    cover = (spec.get("image_path") or "").strip()
+    if not cover:
+        cover = _extract_link_card_media_path(spec.get("cover_url") or "")
+
+    out: list[dict[str, Any]] = []
+    if cover:
+        out.append({"type": "image", "url": cover, "text": "[图片]"})
+    lines = [x for x in (title, desc, target) if x]
+    text = "\n".join(lines) if lines else target
+    if text:
+        out.append({"type": "text", "text": text})
+    if not out:
+        out.append({"type": "text", "text": target})
+    return out
+
+
+def split_reply_payloads(raw: str) -> list[str]:
+    """将规则 reply_content 拆成可逐条发送的 payload 列表。卡片单独展开为图片+文本。"""
+    payloads: list[str] = []
+    for spec in parse_reply_messages(raw):
+        if spec.get("type") == "card":
+            payloads.extend(serialize_reply_content(s) for s in expand_card_spec(spec))
+        else:
+            payloads.append(serialize_reply_content(spec))
+    return payloads
+
+
+def serialize_reply_log(payloads: list[str]) -> str:
+    """把实际逐条发送的 payload（JSON 字符串）合并为结构化的日志内容。
+
+    单条直接返回该 payload；多条用 {"messages":[...]} 包裹，便于前端逐条渲染
+    （图片/表情正常显示为媒体，而不是被压扁成 "图片" 这样的占位文本）。
+    """
+    specs: list[dict[str, Any]] = []
+    for payload in payloads:
+        raw = (payload or "").strip()
+        if not raw:
+            continue
+        spec: dict[str, Any] | None = None
+        if raw.startswith("{"):
+            try:
+                data = json.loads(raw)
+                spec = _normalize_reply_spec(data)
+            except json.JSONDecodeError:
+                spec = None
+        if spec is None:
+            spec = {"type": "text", "text": raw}
+        specs.append(spec)
+    return serialize_reply_messages(specs)
+
+
+def serialize_reply_content(spec: dict[str, Any]) -> str:
+    return json.dumps(spec, ensure_ascii=False, separators=(",", ":"))
+
+
+def build_msg_payload(spec: dict[str, Any]) -> Tuple[dict[str, Any], int]:
+    """根据回复规格构造 IM msg_content 与 message_type。"""
+    reply_type = spec.get("type", "text")
+
+    if reply_type == "text":
+        text = (spec.get("text") or "").strip()
+        return (
+            {
+                "mention_users": [],
+                "aweType": 700,
+                "richTextInfos": [],
+                "text": text,
+            },
+            7,
+        )
+
+    if reply_type == "link":
+        display = (spec.get("text") or spec.get("title") or "").strip()
+        url = (spec.get("url") or "").strip()
+        if not display:
+            display = url
+        msg_content: dict[str, Any] = {
+            "mention_users": [],
+            "aweType": 700,
+            "richTextInfos": [],
+            "text": display,
+        }
+        if url and display:
+            msg_content["richTextInfos"] = [
+                {
+                    "start": 0,
+                    "end": len(display),
+                    "type": 2,
+                    "link": url,
+                    "text": display,
+                }
+            ]
+        return msg_content, 7
+
+    if reply_type == "card":
+        # 卡片在 split_reply_payloads 阶段已展开为「图片 + 文本」两条，正常不会走到这里。
+        # 兜底：万一收到未展开的卡片 spec，退化为「标题/内容/链接」文本，确保可送达。
+        title = (spec.get("title") or "").strip()
+        desc = (spec.get("desc") or spec.get("description") or "").strip()
+        target = (
+            spec.get("target_url")
+            or spec.get("url")
+            or spec.get("link_url")
+            or ""
+        ).strip()
+        lines = [x for x in (title, desc, target) if x]
+        text = "\n".join(lines) if lines else target
+        return (
+            {"mention_users": [], "aweType": 700, "richTextInfos": [], "text": text},
+            MSG_TYPE_TEXT,
+        )
+
+    if reply_type == "image":
+        uri = (spec.get("uri") or "").strip().lstrip("/")
+        url = (spec.get("url") or "").strip()
+        width = spec.get("width")
+        height = spec.get("height")
+        md5 = (spec.get("md5") or "").strip()
+        url_list = spec.get("url_list")
+        if not isinstance(url_list, list):
+            url_list = [url] if url.startswith("http") else []
+
+        clean_urls = [str(u).strip() for u in url_list if str(u).strip().startswith("http")]
+
+        resource_url: dict[str, Any] = {}
+        if uri:
+            resource_url["uri"] = uri
+        if clean_urls:
+            resource_url["url_list"] = clean_urls
+        if md5:
+            resource_url["md5"] = md5
+        if width:
+            try:
+                resource_url["width"] = int(width)
+            except (TypeError, ValueError):
+                pass
+        if height:
+            try:
+                resource_url["height"] = int(height)
+            except (TypeError, ValueError):
+                pass
+
+        msg_content: dict[str, Any] = {
+            "aweType": 2702,
+            "from_gallery": 1,
+            "create_type": 0,
+        }
+        if resource_url:
+            msg_content["resource_url"] = resource_url
+        if uri:
+            msg_content["local_path"] = uri
+        if md5:
+            msg_content["md5"] = md5
+        if width:
+            try:
+                msg_content["cover_width"] = int(width)
+            except (TypeError, ValueError):
+                pass
+        if height:
+            try:
+                msg_content["cover_height"] = int(height)
+            except (TypeError, ValueError):
+                pass
+        return msg_content, MSG_TYPE_IMAGE
+
+    if reply_type == "sticker":
+        url = (spec.get("url") or "").strip()
+        sticker_id = spec.get("sticker_id") or spec.get("id")
+        msg_content = {
+            "display_name": (spec.get("name") or spec.get("text") or "[表情包]").strip(),
+        }
+        if sticker_id:
+            msg_content["id"] = sticker_id
+            msg_content["sticker_id"] = sticker_id
+        if url:
+            msg_content["static_url"] = {"url_list": [url]}
+            msg_content["animate_url"] = {"url_list": [url]}
+        return msg_content, MSG_TYPE_STICKER
+
+    text = format_reply_display(serialize_reply_content(spec))
+    return (
+        {
+            "mention_users": [],
+            "aweType": 700,
+            "richTextInfos": [],
+            "text": text,
+        },
+        7,
+    )