kevin/life-echo
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

chore/ 精简展示AI活动的日志

Browse Source
This commit is contained in:
Kevin
2026-04-03 13:49:24 +08:00
parent 43d1689e9c
commit 4cfa3843a7
11 changed files with 194 additions and 24 deletions
Download Patch File Download Diff File Expand all files Collapse all files

21
api/app/agents/chat/helpers.py
View File

@@ -1,5 +1,6 @@
"""聊天 Agent 共享工具：历史获取、格式化、存储"""
import hashlib
from dataclasses import dataclass
from datetime import datetime
from typing import Any, List
@@ -68,12 +69,28 @@ async def get_history_messages(conversation_id: str) -> List[Any]:
return _lc_messages_from_rows(_human_ai_rows(history))
def format_history_string(messages: List[Any]) -> str:
def _sha12_utf8(text: str) -> str:
return hashlib.sha256((text or "").encode("utf-8")).hexdigest()[:12]
def format_history_string(
messages: List[Any], *, omit_system_body: bool = False
) -> str:
"""将 LangChain 消息列表格式化为调试日志用多段文本（含 System，不静默跳过）。"""
history_parts: list[str] = []
for msg in messages:
if isinstance(msg, SystemMessage):
history_parts.append(f"System: {msg.content}")
if omit_system_body:
c = (
(msg.content or "")
if isinstance(msg.content, str)
else str(msg.content)
)
history_parts.append(
f"System: <omitted total_len={len(c)} sha12={_sha12_utf8(c)}>"
)
else:
history_parts.append(f"System: {msg.content}")
elif isinstance(msg, HumanMessage):
history_parts.append(f"Human: {msg.content}")
elif isinstance(msg, AIMessage):

10
api/app/agents/chat/interview_agent.py
View File

@@ -176,7 +176,10 @@ class InterviewAgent:
log_agent_payload(
logger,
"InterviewAgent.generate_response.prompt",
format_history_string(messages),
format_history_string(
messages,
omit_system_body=settings.agent_log_omit_system_message_body,
),
)
chat_llm = self.llm.bind(max_tokens=reply_plan.max_tokens)
prompt_chars = _message_contents_char_count(messages)
@@ -276,7 +279,10 @@ class InterviewAgent:
log_agent_payload(
logger,
"InterviewAgent.opening.prompt",
format_history_string(messages),
format_history_string(
messages,
omit_system_body=settings.agent_log_omit_system_message_body,
),
)
opening_llm = self.llm.bind(max_tokens=settings.chat_opening_max_tokens)
prompt_chars = _message_contents_char_count(messages)

32
api/app/agents/chat/orchestrator.py
View File

@@ -58,38 +58,46 @@ async def _fetch_interview_memory_evidence(
from app.features.memory.service import MemoryService
if not settings.chat_memory_retrieval_enabled:
logger.debug(
"event=chat_memory_retrieval_skip reason=disabled user_id={}", user_id
)
return ""
msg = (user_message or "").strip()
if not msg:
logger.debug(
"event=chat_memory_retrieval_skip reason=empty user_id={}", user_id
)
return ""
if (
settings.chat_memory_retrieval_require_substantive
and not should_run_chat_stage_memory_heavy_work(msg)
):
logger.debug(
"event=chat_memory_retrieval_skip reason=not_substantive user_id={}",
user_id,
)
return ""
try:
emb = get_embedding_provider()
ms = MemoryService(db, embedding_provider=emb)
bundle = await ms.retrieve(user_id, msg, top_k=settings.chat_memory_top_k)
bd = bundle.model_dump()
vector_ok = emb.is_available()
logger.info(
"memory_evidence_retrieved user_id={} chunks={} facts={} summaries={} timeline={} stories={} vector_ok={}",
user_id,
len(bd.get("relevant_chunks") or []),
len(bd.get("relevant_facts") or []),
len(bd.get("relevant_summaries") or []),
len(bd.get("timeline_hints") or []),
len(bd.get("relevant_stories") or []),
vector_ok,
)
text = format_evidence_chunks_for_prompt(bd)
t = (text or "").strip()
if not t:
logger.debug(
"event=memory_evidence_for_prompt user_id={} formatted_chars=0",
user_id,
)
return ""
max_c = settings.chat_memory_evidence_max_chars
if len(t) > max_c:
return t[: max_c - 3] + "..."
t = t[: max_c - 3] + "..."
logger.info(
"event=memory_evidence_for_prompt user_id={} formatted_chars={}",
user_id,
len(t),
)
return t
except Exception as e:
try:

12
api/app/agents/chat/profile_agent.py
View File

@@ -188,7 +188,10 @@ class ProfileAgent:
log_agent_payload(
logger,
"ProfileAgent.followup.prompt",
format_history_string(messages),
format_history_string(
messages,
omit_system_body=settings.agent_log_omit_system_message_body,
),
)
prompt_chars = _message_contents_char_count(messages)
logger.info(
@@ -246,7 +249,12 @@ class ProfileAgent:
else:
messages.append(HumanMessage(content="（请说出资料收集开场白。）"))
log_agent_payload(
logger, "ProfileAgent.greeting.prompt", format_history_string(messages)
logger,
"ProfileAgent.greeting.prompt",
format_history_string(
messages,
omit_system_body=settings.agent_log_omit_system_message_body,
),
)
prompt_chars = _message_contents_char_count(messages)
logger.info(

23
api/app/core/agent_logging.py
View File

@@ -6,6 +6,10 @@ Agent / LLM 诊断日志：耗时、输入输出规模、截断预览。
便于生产环境在不把全局日志调到 DEBUG 的情况下排查 Agent 性能与路径。
敏感内容：DEBUG 下会记录用户相关文本截断预览，生产环境请勿长期开启 DEBUG。
配置（节选）：``AGENT_LOG_OMIT_SYSTEM_MESSAGE_BODY``（默认 true）省略聊天 System 正文，仅打 len+sha12；
``AGENT_LOG_JSON_PROMPT_PREFIX_CHARS`` + ``AGENT_LOG_JSON_PROMPT_PREFIX_ONLY_IF_LEN_GT`` 在 DEBUG 下跳过
超长单段 prompt 的前缀再预览。
"""
from __future__ import annotations
@@ -96,10 +100,23 @@ def log_agent_payload(
"""在 DEBUG 下记录文本长度与截断预览。"""
if not agent_verbose_enabled():
return
preview = truncate_for_log(text, max_chars=max_chars)
raw = text or ""
total_len = len(raw)
preview_source = raw
extra_note = ""
if (
label.endswith(".prompt")
and settings.agent_log_json_prompt_prefix_chars > 0
and total_len > settings.agent_log_json_prompt_prefix_only_if_len_gt
):
skip = settings.agent_log_json_prompt_prefix_chars
preview_source = raw[skip:]
extra_note = f" skipped_prefix_chars={skip}"
preview = truncate_for_log(preview_source, max_chars=max_chars)
logger.debug(
"agent_payload {} total_len={} preview={}",
"agent_payload {} total_len={}{} preview={}",
label,
len(text or ""),
total_len,
extra_note,
preview,
)

20
api/app/core/config.py
View File

@@ -188,6 +188,14 @@ class Settings(BaseSettings):
log_agent_verbose: bool = False
# AGENT_LOG_MAX_CHARS：DEBUG 下记录 prompt/响应预览时的最大字符数
agent_log_max_chars: int = Field(default=4096, ge=256, le=100_000)
# AGENT_LOG_OMIT_SYSTEM_MESSAGE_BODY：DEBUG 下访谈/资料聊天日志省略 System 正文（仅 len+sha12）
agent_log_omit_system_message_body: bool = True
# AGENT_LOG_JSON_PROMPT_PREFIX_CHARS：DEBUG 下 *.prompt 总长超过下项时再跳过前 N 字符后预览（0=不跳过）
agent_log_json_prompt_prefix_chars: int = Field(default=0, ge=0, le=500_000)
# AGENT_LOG_JSON_PROMPT_PREFIX_ONLY_IF_LEN_GT：触发“跳过前缀”的最小 prompt 长度
agent_log_json_prompt_prefix_only_if_len_gt: int = Field(
default=4000, ge=0, le=2_000_000
)
# 第三方 stdlib logging（空=自动：LOG_LEVEL 为 DEBUG/TRACE 时 Celery→INFO、httpx/httpcore→WARNING）
celery_log_level: str = ""
httpx_log_level: str = ""
@@ -201,6 +209,18 @@ class Settings(BaseSettings):
return False
return str(v).strip().lower() in ("1", "true", "yes", "on")
@field_validator("agent_log_omit_system_message_body", mode="before")
@classmethod
def _coerce_agent_log_omit_system_message_body(cls, v: object) -> bool:
if isinstance(v, bool):
return v
if v is None:
return True
s = str(v).strip().lower()
if s in ("0", "false", "no", "off"):
return False
return True
# ── Misc ─────────────────────────────────────────────────
enable_test_subscription: int = 0
enable_test_plan: str = "" # "1" / "true" / "yes" 为 True

64
api/app/features/memory/service.py
View File

@@ -70,16 +70,20 @@ class MemoryService:
await self._db.flush()
from app.core.config import settings
vectors_written = 0
# Embedding: 若有 provider 则写入
if self._embedding and chunk_records:
texts = [c for _, c in chunk_records]
embeddings = await self._embedding.embed_texts(texts)
for (chunk_id, _), emb in zip(chunk_records, embeddings):
if emb:
vectors_written += 1
await update_chunk_embedding(self._db, chunk_id, emb)
enrichment_ok: bool | None = None
try:
from app.core.config import settings
from app.core.dependencies import get_llm_provider_fast
from app.features.memory.enrichment import enrich_memory_after_ingest_async
@@ -88,12 +92,28 @@ class MemoryService:
await enrich_memory_after_ingest_async(
self._db, user_id, source.id, llm
)
enrichment_ok = True
except Exception as e:
if settings.memory_enrichment_enabled:
enrichment_ok = False
logger.warning(
"memory enrichment 跳过: {} exc_type={}", e, type(e).__name__
)
await self._db.commit()
emb_ok = self._embedding.is_available() if self._embedding else False
logger.info(
"event=memory_ingest_done user_id={} conversation_id={} source_id={} "
"chunks={} vectors_written={} embedding_available={} enrichment_enabled={} enrichment_ok={}",
user_id,
conversation_id,
source.id,
len(chunk_records),
vectors_written,
emb_ok,
settings.memory_enrichment_enabled,
enrichment_ok,
)
return source.id
async def retrieve(
@@ -104,7 +124,23 @@ class MemoryService:
retriever = HybridRetriever(self._db, embedding_provider=self._embedding)
raw = await retriever.retrieve(user_id=user_id, query=query, top_k=top_k)
return EvidenceBundle.model_validate(raw)
bundle = EvidenceBundle.model_validate(raw)
bd = bundle.model_dump()
vec_ok = self._embedding.is_available() if self._embedding else False
logger.info(
"event=memory_retrieve_done user_id={} query_len={} top_k={} "
"chunks={} facts={} summaries={} timeline={} stories={} vector_ok={}",
user_id,
len((query or "").strip()),
top_k,
len(bd.get("relevant_chunks") or []),
len(bd.get("relevant_facts") or []),
len(bd.get("relevant_summaries") or []),
len(bd.get("timeline_hints") or []),
len(bd.get("relevant_stories") or []),
vec_ok,
)
return bundle
async def exclude_chunk(
self, user_id: str, chunk_id: str, *, reason: str = ""
@@ -215,29 +251,51 @@ def ingest_transcript_sync(
session.flush()
chunk_records.append((chunk.id, content))
from app.core.config import settings
vectors_written = 0
embedding_available = False
try:
embedding_provider = get_embedding_provider()
if embedding_provider is not None:
embedding_available = embedding_provider.is_available()
if chunk_records and embedding_provider is not None:
texts = [content for _, content in chunk_records]
embeddings = embedding_provider.embed_texts_sync(texts)
for (chunk_id, _), emb in zip(chunk_records, embeddings):
if emb:
vectors_written += 1
update_chunk_embedding_sync(session, chunk_id, emb)
except Exception as e:
logger.warning(
"memory embedding 跳过(sync): {} exc_type={}", e, type(e).__name__
)
enrichment_ok: bool | None = None
try:
from app.core.config import settings
from app.features.memory.enrichment import enrich_memory_after_ingest_sync
if settings.memory_enrichment_enabled:
enrich_memory_after_ingest_sync(session, user_id, source.id, llm=None)
enrichment_ok = True
except Exception as e:
if settings.memory_enrichment_enabled:
enrichment_ok = False
logger.warning(
"memory enrichment 跳过(sync): {} exc_type={}", e, type(e).__name__
)
session.commit()
logger.info(
"event=memory_ingest_done user_id={} conversation_id={} source_id={} "
"chunks={} vectors_written={} embedding_available={} enrichment_enabled={} enrichment_ok={} sync=1",
user_id,
conversation_id,
source.id,
len(chunk_records),
vectors_written,
embedding_available,
settings.memory_enrichment_enabled,
enrichment_ok,
)
return source.id