初始化提交

backend/app/modules/drive/extract.py

@@ -0,0 +1,127 @@
+"""解析模型返回为内驱力草稿（纯函数，无 IO）。
+
+好奇问题可内嵌次生目标（[内驱力生成器设计] ../../docs/domain/core-domains.md
+§6：好奇问题→转化次生目标）。缺省取**安全悲观默认**：``risk_level=L4``、
+``source=reflection``，避免模型漏字段时生成高风险探索。
+"""
+
+import json
+import re
+from dataclasses import dataclass, field
+from typing import Any
+
+
+@dataclass
+class SubgoalDraft:
+    title: str
+    description: str
+    expected_information_gain: float = 0.5
+    interruption_cost: float = 0.5
+    risk_level: str = "L4"
+
+
+@dataclass
+class CuriosityDraft:
+    question: str
+    source: str = "reflection"
+    related_entities: list[str] = field(default_factory=list)
+    expected_value: float = 0.5
+    urgency: float = 0.5
+    risk_level: str = "L4"
+    subgoals: list[SubgoalDraft] = field(default_factory=list)
+
+
+_VALID_SOURCES = {
+    "world_uncertainty",
+    "memory_gap",
+    "perception_change",
+    "goal_tension",
+    "reflection",
+    "time_rhythm",
+}
+_VALID_LEVELS = {"L0", "L1", "L2", "L3", "L4"}
+
+
+def parse_curiosities(raw: str) -> list[CuriosityDraft]:
+    drafts: list[CuriosityDraft] = []
+    for item in _extract_json_array(raw):
+        if not isinstance(item, dict):
+            continue
+        question = str(item.get("question", "")).strip()
+        if not question:
+            continue
+        drafts.append(
+            CuriosityDraft(
+                question=question,
+                source=_source(item.get("source")),
+                related_entities=_str_list(item.get("related_entities")),
+                expected_value=_clamp_float(item.get("expected_value"), 0.5),
+                urgency=_clamp_float(item.get("urgency"), 0.5),
+                risk_level=_level(item.get("risk_level")),
+                subgoals=_parse_subgoals(item.get("subgoals")),
+            )
+        )
+    return drafts
+
+
+def _parse_subgoals(value: Any) -> list[SubgoalDraft]:
+    if not isinstance(value, list):
+        return []
+    result: list[SubgoalDraft] = []
+    for item in value:
+        if not isinstance(item, dict):
+            continue
+        title = str(item.get("title", "")).strip()
+        description = str(item.get("description", "")).strip()
+        if not title or not description:
+            continue
+        result.append(
+            SubgoalDraft(
+                title=title,
+                description=description,
+                expected_information_gain=_clamp_float(item.get("expected_information_gain"), 0.5),
+                interruption_cost=_clamp_float(item.get("interruption_cost"), 0.5),
+                risk_level=_level(item.get("risk_level")),
+            )
+        )
+    return result
+
+
+def _extract_json_array(raw: str) -> list[Any]:
+    text = raw.strip()
+    matched = re.search(r"```(?:json)?\s*(\[.*?\])\s*```", text, re.DOTALL)
+    if matched:
+        text = matched.group(1)
+    else:
+        start, end = text.find("["), text.rfind("]")
+        if start != -1 and end > start:
+            text = text[start : end + 1]
+    try:
+        data = json.loads(text)
+    except json.JSONDecodeError:
+        return []
+    return data if isinstance(data, list) else []
+
+
+def _source(value: Any) -> str:
+    text = str(value or "").strip()
+    return text if text in _VALID_SOURCES else "reflection"
+
+
+def _level(value: Any) -> str:
+    text = str(value or "").strip().upper()
+    return text if text in _VALID_LEVELS else "L4"
+
+
+def _str_list(value: Any) -> list[str]:
+    if not isinstance(value, list):
+        return []
+    return [str(v) for v in value if v is not None]
+
+
+def _clamp_float(value: Any, default: float) -> float:
+    try:
+        num = float(value)
+    except (TypeError, ValueError):
+        return default
+    return max(0.0, min(1.0, num))