bb16d3a5c9
访谈与阶段 - 新增 app/agents/stage_constants.py：集中 CHAT_STAGES、章节分类/顺序、阶段到默认 memoir 类别等，与 MemoirState 默认槽位顺序对齐；减少散落在 prompts 内的重复常量。 - 新增 app/agents/chat/prompt_context.py：以 ChatPromptContext 汇总 guided 系统提示所需字段（阶段、槽位、轮次、人设、记忆证据、回复长度模式、背景声线、职业等），统一走 get_guided_conversation_prompt。 - 大幅收敛 app/agents/chat/prompts_conversation.py；调整 prompts.py、stage_prompts.py、stage_detection.py；同步 interview_agent、profile_agent、helpers 与 state_schema，使对话侧构造提示的方式一致、可测。 回忆录流水线 - memoir/prompts.py 删除已迁至 stage_constants / 独立模板的大段常量与图片占位相关逻辑；classification / extraction / fidelity / narrative agents 与 orchest（全量历史仍可用于计数，注入模型时按轮次与字符上限截断）、image_prompt_fallback_disabled。 - dependencies 增加 get_llm_provider_fast（LRU 缓存，可与默认共用密钥与 base_url）。 任务与编排 - memoir_tasks：prepare_batches 注入 llm_fast；开启独立快档模型时打结构化日志。 - chapter_cover_tasks、story_image_tasks：与图片 prompt / JSON 工具路径或策略变更对齐（import 与行为一致）。 - story_pipeline_sync 等小处同步。 其它核心 - langchain_llm、text_normalize 随上述调用链微调。 开发者体验 - .cursor/settings.json：启用 redis-development、postman 插件。 测试 - 新增 test_image_prompt_policy：覆盖「禁止回退」等图片 prompt 策略。 - 更新 test_interview_prompts、test_interview_reply_length、test_experience_regressions、test_json_and_memory_utils，匹配新常量位置、json_utils 与对话/长度行为。
108 lines
3.1 KiB
Python
108 lines
3.1 KiB
Python
"""JSON 载荷解析、证据格式化、Story 批量规划校验（纯函数）。"""
|
|
|
|
import pytest
|
|
|
|
from app.agents.chat.reply_limits import truncate_chat_segments
|
|
|
|
from app.agents.memoir.classification_agent import _normalize_llm_category
|
|
from app.agents.memoir.prompts import format_evidence_chunks_for_prompt
|
|
from app.features.memory.evidence_format import (
|
|
format_evidence_chunks_for_prompt as format_evidence_from_memory,
|
|
)
|
|
from app.agents.memoir.story_route_agent import (
|
|
StoryBatchPlan,
|
|
StoryBatchPlanUnit,
|
|
validate_story_batch_plan,
|
|
)
|
|
from app.core.json_utils import extract_json_payload
|
|
|
|
|
|
def test_extract_json_payload_strips_markdown_fence() -> None:
|
|
raw = """```json
|
|
{"a": 1}
|
|
```"""
|
|
assert '"a"' in extract_json_payload(raw)
|
|
|
|
|
|
def test_normalize_llm_category_strips_quotes() -> None:
|
|
assert _normalize_llm_category('"childhood"') == "childhood"
|
|
assert _normalize_llm_category("`beliefs`") == "beliefs"
|
|
|
|
|
|
def test_format_evidence_chunks_includes_timeline() -> None:
|
|
ev = {
|
|
"relevant_chunks": [{"content": "chunk1"}],
|
|
"relevant_facts": [
|
|
{"subject": "我", "predicate": "生于", "object_json": "1950"}
|
|
],
|
|
"timeline_hints": [
|
|
{
|
|
"id": "1",
|
|
"event_year": 1977,
|
|
"event_date": None,
|
|
"title": "恢复高考",
|
|
"description": "参加了考试",
|
|
}
|
|
],
|
|
"relevant_summaries": [],
|
|
"relevant_stories": [],
|
|
}
|
|
out = format_evidence_chunks_for_prompt(ev)
|
|
assert "chunk1" in out
|
|
assert "1950" in out or "生于" in out
|
|
assert "1977" in out or "恢复高考" in out
|
|
assert format_evidence_from_memory(ev) == out
|
|
|
|
|
|
def test_validate_story_batch_plan_ok() -> None:
|
|
ordered = ["s1", "s2"]
|
|
plan = StoryBatchPlan(
|
|
units=[
|
|
StoryBatchPlanUnit(
|
|
segment_ids=["s1", "s2"],
|
|
decision="new_story",
|
|
target_story_id=None,
|
|
new_story_title="标题",
|
|
reason=None,
|
|
)
|
|
]
|
|
)
|
|
ok, err = validate_story_batch_plan(ordered, plan, valid_story_ids=set())
|
|
assert ok is True
|
|
assert err is None
|
|
|
|
|
|
def test_truncate_chat_segments() -> None:
|
|
out = truncate_chat_segments(
|
|
["a" * 300, "b"],
|
|
max_segments=2,
|
|
max_chars_per_segment=220,
|
|
)
|
|
assert out[0] == "a" * 219 + "…"
|
|
assert len(out[0]) == 220
|
|
assert out[1] == "b"
|
|
|
|
|
|
def test_validate_story_batch_plan_duplicate_segment() -> None:
|
|
plan = StoryBatchPlan(
|
|
units=[
|
|
StoryBatchPlanUnit(
|
|
segment_ids=["s1"],
|
|
decision="new_story",
|
|
target_story_id=None,
|
|
new_story_title="A",
|
|
reason=None,
|
|
),
|
|
StoryBatchPlanUnit(
|
|
segment_ids=["s1"],
|
|
decision="new_story",
|
|
target_story_id=None,
|
|
new_story_title="B",
|
|
reason=None,
|
|
),
|
|
]
|
|
)
|
|
ok, err = validate_story_batch_plan(["s1", "s1"], plan, valid_story_ids=set())
|
|
assert ok is False
|
|
assert err == "duplicate_segment"
|