2fded6fbd9
- Drop interview_reply_length and utterance_substance; always run stage LLM and memory retrieval when enabled; trim Settings fields and .env.example. - Replace guided/opening prompts with compact fact blocks plus unified behavior guidance; slim background_voice and persona to tone hints. - InterviewAgent uses fixed chat_interview max_tokens/chars/segments. Also includes stacked work: profile followup/extract path, evaluation rubric and judge schema updates, transcript SPLIT handling in execution service, user export markdown split tests, and golden case fixture.
282 lines
14 KiB
Plaintext
282 lines
14 KiB
Plaintext
# =============================================================================
|
||
# Life Echo API — 模板（example）
|
||
#
|
||
# 本地：复制为 .env.development（勿提交密钥），再运行 api/development.sh 会在首次自动生成 .env（从
|
||
# .env.development 复制）；Settings 只读 .env（见 app/core/config.py）。
|
||
# 服务端：仓库维护 .env.staging / .env.production；workflow 按目标环境上传并复制为运行时 .env，compose 的 env_file 统一指向 .env。
|
||
# 不要把真实密钥提交到仓库。
|
||
# =============================================================================
|
||
|
||
# =============================================================================
|
||
# Docker Compose（宿主机独立 Caddy 反代到本 API）
|
||
# =============================================================================
|
||
# 映射到宿主机的端口，默认 8000；与同机其它项目冲突时改为未占用端口，并在独立 Caddy 的 Caddyfile 中 reverse_proxy 到 127.0.0.1:该端口。
|
||
# LIFE_ECHO_API_HOST_PORT=8000
|
||
# 若 Caddy 跑在独立容器且非 host 网络，不要用 127.0.0.1，应把 Caddy 加入与本 compose 相同的 Docker 网络，并对 http://life-echo-api-prod:8000 做 reverse_proxy。
|
||
|
||
# =============================================================================
|
||
# Logging（loguru sink 最低级别：TRACE / DEBUG / INFO / WARNING / ERROR / CRITICAL）
|
||
# =============================================================================
|
||
LOG_LEVEL=INFO
|
||
# Agent 单行 INFO 摘要（耗时、路由、段落规模）；与 LOG_LEVEL 独立，便于生产短暂排查
|
||
# LOG_AGENT_VERBOSE=0
|
||
# DEBUG 下 prompt/响应预览最大字符数
|
||
# AGENT_LOG_MAX_CHARS=4096
|
||
# DEBUG 下访谈/资料：省略 SystemMessage 正文（仅 total_len+sha12）；0/false=打出全文
|
||
# AGENT_LOG_OMIT_SYSTEM_MESSAGE_BODY=1
|
||
# DEBUG 下超长单段 *.prompt：总长超过下一项时，先跳过前 N 字符再预览（0=不跳过）
|
||
# AGENT_LOG_JSON_PROMPT_PREFIX_CHARS=0
|
||
# AGENT_LOG_JSON_PROMPT_PREFIX_ONLY_IF_LEN_GT=4000
|
||
# 第三方 stdlib logging（空=自动：LOG_LEVEL 为 DEBUG/TRACE 时 Celery→INFO、httpx/httpcore→WARNING，减少刷屏）
|
||
# CELERY_LOG_LEVEL=
|
||
# HTTPX_LOG_LEVEL=
|
||
|
||
# =============================================================================
|
||
# LLM / DeepSeek
|
||
# =============================================================================
|
||
DEEPSEEK_API_KEY=your_deepseek_api_key
|
||
DEEPSEEK_BASE_URL=https://api.deepseek.com
|
||
DEEPSEEK_MODEL=deepseek-chat
|
||
|
||
# =============================================================================
|
||
# Memory 向量（智谱 BigModel 国内 embedding-3；与 DeepSeek/OpenAI 用途分离）
|
||
# 文档：https://docs.bigmodel.cn/cn/guide/models/embedding/embedding-3
|
||
# 本期固定 1024 维；库表经迁移与 MEMORY_EMBEDDING_DIMENSION 一致。
|
||
# =============================================================================
|
||
ZHIPU_API_KEY=your_zhipu_api_key
|
||
# 默认国内通用端点（与 ZhipuAiClient 一致）
|
||
# EMBEDDING_BASE_URL=https://open.bigmodel.cn/api/paas/v4
|
||
EMBEDDING_MODEL=embedding-3
|
||
|
||
# Chat 访谈：每轮根据用户内容判定主人生阶段（关则仅用关键词，省一次 LLM）
|
||
# CHAT_STAGE_DETECTION_ENABLED=true
|
||
# CHAT_STAGE_DETECTION_MAX_TOKENS=128
|
||
# 访谈性格（InterviewAgent）：default | warm_listener | curious_guide
|
||
# CHAT_INTERVIEW_PERSONA=default
|
||
# 访谈主回复：统一 max_tokens / 单段字数（代码截断），不再分 brief/expanded 档
|
||
# CHAT_INTERVIEW_MAX_TOKENS=380
|
||
# CHAT_INTERVIEW_MAX_CHARS_PER_SEGMENT=260
|
||
# CHAT_INTERVIEW_MAX_SEGMENTS=2
|
||
# 访谈：是否按本轮用户话检索记忆并注入提示词（关则不调 retrieve）
|
||
# CHAT_MEMORY_RETRIEVAL_ENABLED=true
|
||
# CHAT_MEMORY_TOP_K=8
|
||
# CHAT_MEMORY_EVIDENCE_MAX_CHARS=4096
|
||
|
||
# Memoir：批处理/抽取更新 slot 时是否允许改写 MemoirState.current_stage（默认 false，访谈 switch_stage 仍可推进）
|
||
# True 时仅当 proposed 与 existing 在同一 chat_bucket 才对齐 current_stage
|
||
# MEMOIR_EXTRACTION_UPDATES_CURRENT_STAGE=false
|
||
|
||
# Memoir：叙事前口述归一（segment 原文仍落库；仅 story 流水线派生输入）
|
||
# MEMOIR_ORAL_NORMALIZE_ENABLED=true
|
||
# off | rules | llm（llm 为先规则再 LLM 纠错，失败回退规则结果）
|
||
# MEMOIR_ORAL_NORMALIZE_MODE=llm
|
||
# MEMOIR_ORAL_NORMALIZE_LLM_MAX_TOKENS=512
|
||
# MEMOIR_ORAL_NORMALIZE_LLM_MAX_INPUT_CHARS=8000
|
||
|
||
# Chat：模型消费净稿（segment 原文仍落库；访谈编排层归一后注入 Agent / 记忆检索）
|
||
# CHAT_INPUT_NORMALIZE_ENABLED=true
|
||
# off | rules | llm（llm 为先规则再 LLM；失败回退规则；编排层已带 LLM 时不重复在 Agent 调）
|
||
# CHAT_INPUT_NORMALIZE_MODE=rules
|
||
# CHAT_INPUT_NORMALIZE_LLM_MAX_TOKENS=512
|
||
# CHAT_INPUT_NORMALIZE_LLM_MAX_INPUT_CHARS=8000
|
||
# True：仅 is_from_voice 时走 LLM 纠错；键盘输入仅规则归一
|
||
# CHAT_INPUT_NORMALIZE_LLM_VOICE_ONLY=true
|
||
|
||
# Memoir Phase1：True 时用一次「批量 JSON」做抽取+分类（单段或多段均可；失败自动回退逐段）。
|
||
# False 时始终逐段（与启用本开关前的行为一致，含防抖合并后的多段任务）。
|
||
# MEMOIR_PHASE1_BATCH_LLM_ENABLED=false
|
||
# MEMOIR_PHASE1_BATCH_LLM_MAX_TOKENS=4096
|
||
|
||
# =============================================================================
|
||
# Database
|
||
# =============================================================================
|
||
# 本地开发:
|
||
# DATABASE_URL=postgresql://postgres:postgres@localhost:5432/life_echo
|
||
# Docker / 服务端（主机名一般为 compose 服务名 postgres）:
|
||
# DATABASE_URL=postgresql://postgres:postgres@postgres:5432/life_echo
|
||
DATABASE_URL=postgresql://postgres:postgres@localhost:5432/life_echo
|
||
# 启动时 Alembic（main.py）；生产可设 ALEMBIC_STARTUP_FAIL_FAST=true，迁移失败则拒绝启动
|
||
# ALEMBIC_RUN_ON_STARTUP=true
|
||
# ALEMBIC_STARTUP_FAIL_FAST=false
|
||
# ALEMBIC_STARTUP_MAX_RETRIES=3
|
||
# ALEMBIC_STARTUP_RETRY_BASE_SECONDS=1.0
|
||
|
||
# =============================================================================
|
||
# Redis
|
||
# =============================================================================
|
||
# 本地开发:
|
||
# REDIS_URL=redis://localhost:6379/0
|
||
# Docker / 服务端:
|
||
# REDIS_URL=redis://redis:6379/0
|
||
REDIS_URL=redis://localhost:6379/0
|
||
REDIS_SESSION_TTL=86400
|
||
|
||
# =============================================================================
|
||
# Memory compaction（近重复 memory chunk 软排除；Celery + Redis 防抖）
|
||
# 模板统一默认开启；须同时运行 celery worker 与 celery-beat（docker-compose 已含 beat，负责 memory_compaction_sweep）。
|
||
# =============================================================================
|
||
MEMORY_COMPACTION_ENABLED=true
|
||
# MEMORY_COMPACTION_DEBOUNCE_SECONDS=105
|
||
# MEMORY_COMPACTION_LOCK_TTL_SECONDS=600
|
||
# MEMORY_COMPACTION_CHUNK_SIMILARITY_THRESHOLD=0.92
|
||
# MEMORY_COMPACTION_MIN_LAYERS_FOR_EXCLUDE=2
|
||
# MEMORY_COMPACTION_MAX_CHUNKS_PER_RUN=200
|
||
# MEMORY_COMPACTION_MAX_EXCLUDES_PER_RUN=50
|
||
# MEMORY_COMPACTION_MAX_NEIGHBORS_PER_CHUNK=25
|
||
# MEMORY_COMPACTION_TEXT_JACCARD_MIN=0.55
|
||
# MEMORY_COMPACTION_METADATA_EVENT_YEAR_WINDOW=1
|
||
# MEMORY_COMPACTION_SWEEP_RECENT_HOURS=24
|
||
|
||
# =============================================================================
|
||
# Story 流水线（post-commit、章节物化、append 上限、evidence 检索）
|
||
# =============================================================================
|
||
# STORY_IMAGE_ENQUEUE_DEDUP_TTL=300
|
||
# RECOMPOSE_CHAPTER_DELAY_SECONDS=8
|
||
# 与 Phase2 / 章节物化共用；应 ≥ 最长单次叙事+物化耗时
|
||
# CHAPTER_PIPELINE_LOCK_TTL_SECONDS=360
|
||
# STORY_APPEND_MAX_CANONICAL_CHARS=12000
|
||
# STORY_APPEND_MAX_VERSIONS=20
|
||
# EVIDENCE_TOP_K_DEFAULT=10
|
||
# EVIDENCE_TOP_K_LARGE_BATCH=5
|
||
# EVIDENCE_LARGE_BATCH_THRESHOLD=3
|
||
#
|
||
# Memoir 可靠性（叙事 faithful、标题 slots、证据渗漏、Phase1→2 追踪）
|
||
# MEMOIR_FIDELITY_FAIL_OPEN_ON_PARSE_ERROR=false
|
||
# MEMOIR_NARRATIVE_EVIDENCE_OVERLAP_MIN_CHARS=14
|
||
# MEMOIR_EVIDENCE_SCENE_ANCHOR_CHECK_ENABLED=true
|
||
# MEMOIR_TITLE_SLOTS_REQUIRE_BODY_OR_ORAL_MATCH=true
|
||
# MEMOIR_TITLE_HAY_GROUNDING_STRICT_PHRASES_ENABLED=true
|
||
# MEMOIR_RECOMPOSE_RETRY_ON_LOCK_CONTENTION=true
|
||
# MEMOIR_PHASE2_SINGLEFLIGHT_IMMEDIATE=true
|
||
#
|
||
# Memory：事实检索未命中时是否退回「最近 confirmed 事实」（默认关，易串台）
|
||
# MEMORY_FACT_SEARCH_USE_RECENT_FALLBACK=false
|
||
|
||
# =============================================================================
|
||
# Auth
|
||
# =============================================================================
|
||
# 建议使用: openssl rand -hex 32
|
||
SECRET_KEY=replace_with_a_strong_random_secret
|
||
ALGORITHM=HS256
|
||
ACCESS_TOKEN_EXPIRE_MINUTES=120
|
||
|
||
# =============================================================================
|
||
# Tencent Cloud — 短信
|
||
# =============================================================================
|
||
# 短信、一句话 ASR/TTS、COS 为不同产品；同一主账号可共用同一对 SecretId/SecretKey（分别填三处）。
|
||
TENCENT_SMS_SECRET_ID=your_tencent_sms_secret_id
|
||
TENCENT_SMS_SECRET_KEY=your_tencent_sms_secret_key
|
||
# 短信应用 SDK AppID
|
||
TENCENT_SMS_SDK_APP_ID=your_sms_sdk_app_id
|
||
# 短信签名内容（不包含【】符号）
|
||
TENCENT_SMS_SIGN_NAME=your_sms_sign_name
|
||
# 短信模板 ID
|
||
TENCENT_SMS_TEMPLATE_ID=your_sms_template_id
|
||
# 短信模板参数数量（1=仅验证码，2=验证码+过期时间）
|
||
# 若遇 TemplateParamSetNotMatchApprovedTemplate，请对照控制台模板配置
|
||
TENCENT_SMS_TEMPLATE_PARAM_COUNT=1
|
||
|
||
# =============================================================================
|
||
# ASR Provider（whisper | tencent）
|
||
# =============================================================================
|
||
ASR_PROVIDER=whisper
|
||
|
||
# =============================================================================
|
||
# Whisper ASR（ASR_PROVIDER=whisper 时使用）
|
||
# =============================================================================
|
||
ASR_MODEL_SIZE=small
|
||
ASR_DEVICE=cpu
|
||
ASR_COMPUTE_TYPE=int8
|
||
|
||
# GPU 环境（示例，按需启用）
|
||
# ASR_MODEL_SIZE=medium
|
||
# ASR_DEVICE=cuda
|
||
# ASR_COMPUTE_TYPE=float16
|
||
|
||
# =============================================================================
|
||
# Tencent Cloud — 一句话 ASR + TTS（ASR_PROVIDER=tencent 或 TTS_PROVIDER=tencent）
|
||
# =============================================================================
|
||
TENCENT_SECRET_ID=your_tencent_asr_secret_id
|
||
TENCENT_SECRET_KEY=your_tencent_asr_secret_key
|
||
|
||
# =============================================================================
|
||
# TTS（文字转语音，Agent 回复播音）— 与 ASR 独立
|
||
# =============================================================================
|
||
# ENABLE_TTS：仅控制是否合成并下发 TTS_AUDIO；不影响用户语音转写（ASR）
|
||
ENABLE_TTS=true
|
||
TTS_PROVIDER=tencent
|
||
# 仅 TTS_PROVIDER=openai 时需要
|
||
# OPENAI_API_KEY=your_openai_api_key
|
||
# 音色 ID 见 https://cloud.tencent.com/document/product/1073/92668
|
||
TTS_VOICE_TYPE=502001
|
||
TTS_CODEC=mp3
|
||
|
||
# =============================================================================
|
||
# WeChat Pay
|
||
# =============================================================================
|
||
WECHAT_PAY_APP_ID=your_wechat_pay_app_id
|
||
WECHAT_PAY_MCH_ID=your_wechat_mch_id
|
||
WECHAT_PAY_API_V3_KEY=your_wechat_api_v3_key
|
||
# 商户私钥：推荐使用文件路径，避免 .env 中长 PEM 转义问题
|
||
WECHAT_PAY_PRIVATE_KEY_PATH=certs/apiclient_key.pem
|
||
# 若不用文件，可配置 WECHAT_PAY_PRIVATE_KEY（PEM，换行用 \n）
|
||
# WECHAT_PAY_PRIVATE_KEY="-----BEGIN PRIVATE KEY-----\n...\n-----END PRIVATE KEY-----"
|
||
WECHAT_PAY_CERT_SERIAL_NO=your_wechat_cert_serial_no
|
||
WECHAT_PAY_NOTIFY_URL=https://your-domain.com/api/payment/notify/wechat
|
||
# 平台公钥模式（仅当无法走平台证书自动拉取时使用）；勿填商户私钥路径
|
||
# WECHAT_PAY_PLATFORM_PUBLIC_KEY_PATH=certs/wechat_platform_public_key.pem
|
||
# WECHAT_PAY_PLATFORM_PUBLIC_KEY_ID=your_wechat_platform_public_key_id
|
||
|
||
# =============================================================================
|
||
# Alipay
|
||
# =============================================================================
|
||
ALIPAY_APP_ID=your_alipay_app_id
|
||
ALIPAY_PRIVATE_KEY=your_alipay_private_key
|
||
ALIPAY_PUBLIC_KEY=your_alipay_public_key
|
||
ALIPAY_NOTIFY_URL=https://your-domain.com/api/payment/notify/alipay
|
||
|
||
# =============================================================================
|
||
# Misc
|
||
# =============================================================================
|
||
ENABLE_TEST_SUBSCRIPTION=0
|
||
|
||
# =============================================================================
|
||
# Memoir image generation（Story 主图等；轮询 Liblib 任务）
|
||
# =============================================================================
|
||
MEMOIR_IMAGE_ENABLED=false
|
||
MEMOIR_IMAGE_POLL_INTERVAL=3
|
||
MEMOIR_IMAGE_MAX_ATTEMPTS=20
|
||
MEMOIR_IMAGE_PROVIDER=liblib
|
||
MEMOIR_IMAGE_STYLE_DEFAULT=watercolor
|
||
MEMOIR_IMAGE_SIZE_DEFAULT=1280x720
|
||
# Story 正文至少多少字才生成主图 intent / 调图（0=不限制）
|
||
STORY_IMAGE_MIN_BODY_CHARS=400
|
||
# 叙事模型输出相对口述过短则回退为口述原文
|
||
MEMOIR_NARRATIVE_FALLBACK_BODY_RATIO=0.5
|
||
MEMOIR_NARRATIVE_FALLBACK_MIN_CHARS=20
|
||
# 回忆录 segment 入队：累计 strip 后字数未达此值则暂缓提交 Celery（0=关闭字数门闸，仅静默防抖后提交）
|
||
# MEMOIR_SEGMENT_BATCH_MIN_CHARS=50
|
||
# 本批首条入队起最长等待（秒），超时仍提交；测试可调低，生产可调高
|
||
# MEMOIR_SEGMENT_BATCH_MAX_WAIT_SECONDS=60
|
||
# 可选，Liblib 返回图片域名不在默认白名单时（逗号分隔）
|
||
# MEMOIR_IMAGE_DOWNLOAD_HOSTS=liblib.cloud,liblibai.cloud
|
||
|
||
# =============================================================================
|
||
# Liblib image provider
|
||
# =============================================================================
|
||
LIBLIB_ACCESS_KEY=your_liblib_access_key
|
||
LIBLIB_SECRET_KEY=your_liblib_secret_key
|
||
LIBLIB_BASE_URL=https://openapi.liblibai.cloud
|
||
LIBLIB_TEMPLATE_UUID=your_liblib_template_uuid
|
||
|
||
# =============================================================================
|
||
# Tencent Cloud — COS（回忆录图片存储）
|
||
# =============================================================================
|
||
TENCENT_COS_SECRET_ID=your_tencent_cos_secret_id
|
||
TENCENT_COS_SECRET_KEY=your_tencent_cos_secret_key
|
||
TENCENT_COS_REGION=ap-shanghai
|
||
TENCENT_COS_BUCKET=your_bucket_name
|
||
TENCENT_COS_BASE_URL=https://your_bucket_name.cos.ap-shanghai.myqcloud.com
|
||
# 可选临时凭证
|
||
# TENCENT_COS_TOKEN=
|