feat(config,runtime): 上下文预算按模型窗口推导 + usage/缓存可观测 + sidecar 存全文（T1）
- 窗口声明：models.json 组级 contextWindow + env LLM_CONTEXT_WINDOW（per-group materialize 播种，0 不写盘、负数报错）。 - 预算三态：CONTEXT_MAX_TOKENS 未设置 ⇒ 按窗口推导；>0 显式覆盖；0|负 关闭压缩（不变）。推导 budget = min(窗口×比例, 窗口−输出保留)，比例默认 2/3（CONTEXT_BUDGET_RATIO，1M ⇒ 699050 tokens ≈ 683Ki）；输出保留 = max(32768, 该组 maxTokens) 且封顶窗口/2（防配置矛盾时退化成 1 token）；窗口未声明 ⇒ 回落 131072 + 恰好一条 WARN。接线在组合根 wire_context.go（唯一决策点）。 - usage/缓存可观测（此前**完全不可观测**）：Response 增 CacheHit/CacheMiss/Reasoning tokens + CacheHitRate；解析 OpenAI 兼容（prompt_cache_hit_tokens/prompt_cache_miss_tokens/cached_tokens/reasoning_tokens）与 anthropic（cache_read_input_tokens/cache_creation_input_tokens）两种形状；记录点在适配器 record（换入口不丢成本日志），日志无凭据；单次输入 ≥预算 90% 记 WARN（不阻断）。 - 根治清单「续读」假承诺：抽取期不再按注入预算截断（sidecar 存全文，只受 MaxSidecarBytes=8 MiB 安全上限，ExtractorVersion v1→v2 旧 sidecar 失效重抽）；注入预算改在装配期生效 ⇒ ref.Chars/Truncated 是注入侧事实、游标真是继续读的起点；document-read 的 limit 上限改为独立具名常量。 - 顺手修越界：config.go 505 行（超 500）拆出 llm.go。 - 主会话独立验收：隔离 worktree（HEAD + 本切片 39 文件，排除在途 gateway）build/vet 绿、全树 59 包 0 FAIL、linux+darwin 交叉编译绿；变异独立复现（比例改 0.25 ⇒ 推导用例红 262144 vs 699050）。
This commit is contained in:
AskaEth committed
1 parent
5b2b385524
commit
72b95df22b
39 files changed
+2496
-377
No files matched your search
@@ -132,94 +132,6 @@ type DBConfig struct {
|
||||
PGURL string `env:"DB_PG_URL"`
|
||||
}
|
||||
|
||||
// LLMConfig holds the multi-supplier LLM adapter settings. Provider selects the
|
||||
// wire protocol; every provider has its own independent base/key/model/fallback
|
||||
// so multiple vendors can coexist.
|
||||
type LLMConfig struct {
|
||||
// Provider declares which protocol to use: "openai" (OpenAI-compatible
|
||||
// /chat/completions) or "anthropic" (Anthropic Messages /v1/messages).
|
||||
Provider string `env:"LLM_PROVIDER" default:"openai"`
|
||||
|
||||
// OpenAI-compatible provider (default). Kept flat for backward compatibility
|
||||
// with the historic LLM_BASE_URL/API_KEY/MODEL/FALLBACK_MODEL env names.
|
||||
BaseURL string `env:"LLM_BASE_URL"`
|
||||
APIKey string `env:"LLM_API_KEY"`
|
||||
Model string `env:"LLM_MODEL"`
|
||||
Fallback string `env:"LLM_FALLBACK_MODEL"`
|
||||
|
||||
// VisionModel is the dedicated model behind the vision tool (image
|
||||
// understanding). It is the tool's model, not the main model's ability: the
|
||||
// per-persona/per-request routing decision belongs to the vision Router, and
|
||||
// the main model's own image input capability is declared by the active
|
||||
// models.json group's multimodal flag (裁决 14.4.1/14.4.5)。
|
||||
VisionModel string `env:"LLM_VISION_MODEL"`
|
||||
|
||||
// OCRModel is the dedicated model behind the OCR tool. Empty reuses
|
||||
// VisionModel, so OCR keeps working without a second endpoint (裁决 14.4.1
|
||||
// 的「独立视觉模型 + OCR 模型」)。
|
||||
OCRModel string `env:"LLM_OCR_MODEL"`
|
||||
|
||||
// Multimodal 声明**主模型自身**能否收图：models.json 组 `multimodal` 的 env
|
||||
// 等价物，只作用于 .env 合成的组，默认 false（语义与分工见 GroupFromEnv）。
|
||||
Multimodal bool `env:"LLM_MULTIMODAL" default:"false"`
|
||||
|
||||
// Local is the optional third degradation level (裁决 03.4.3): primary →
|
||||
// fallback → local. It defaults to off because the current machine cannot run
|
||||
// a local model comfortably; when off the chain keeps its two-level shape.
|
||||
Local LocalLLMConfig
|
||||
|
||||
// FallbackMaxTokens caps the last-resort fallback model in a group's chain
|
||||
// (primary → fallback). It is the final safety budget: it only applies when
|
||||
// the chain actually has a fallback step, that group sets no per-group
|
||||
// MaxTokens (models.json maxTokens) and the request carries no explicit
|
||||
// budget. Zero disables the cap (the adapter default then applies).
|
||||
FallbackMaxTokens int `env:"LLM_FALLBACK_MAX_TOKENS" default:"8192"`
|
||||
|
||||
// ReasoningKind declares this OpenAI-compatible endpoint's reasoning-switch
|
||||
// wire shape (裁决 03.4.2/08.4.1: 不同提供商的开关形状不同，必须逐 provider
|
||||
// 独立适配). Values mirror runtimeconfig.ReasoningKind*: "" or "auto" infers
|
||||
// the vendor from the model name/base URL and sends nothing when the shape is
|
||||
// ambiguous (the conservative default, so an existing .env is unchanged);
|
||||
// "deepseek" / "openai" / "vllm" / "dashscope" select that endpoint's shape
|
||||
// explicitly. It is the env-fallback twin of the models.json group's
|
||||
// `reasoning.kind`, which is the primary carrier (hot-reloadable, panel-edited).
|
||||
ReasoningKind string `env:"LLM_REASONING_KIND" default:""`
|
||||
|
||||
// Anthropic native provider, used when Provider=anthropic. Independent
|
||||
// config block so each vendor is a self-contained config.
|
||||
AnthropicBaseURL string `env:"ANTHROPIC_BASE_URL"`
|
||||
AnthropicAPIKey string `env:"ANTHROPIC_API_KEY"`
|
||||
AnthropicModel string `env:"ANTHROPIC_MODEL"`
|
||||
AnthropicFallback string `env:"ANTHROPIC_FALLBACK_MODEL"`
|
||||
}
|
||||
|
||||
// LocalLLMConfig configures the optional local model rung — the third level of
|
||||
// the primary → fallback → local degradation chain (裁决 03.4.3). It is opt-in
|
||||
// and defaults to off: a machine that cannot run a local model comfortably must
|
||||
// not silently pay for one. It becomes an active chain step only when Enabled is
|
||||
// true and both BaseURL and Model are set; anything less is reported and skipped
|
||||
// rather than producing a half-configured step.
|
||||
type LocalLLMConfig struct {
|
||||
// Enabled turns the local rung on. Default false (project red line: 本地模型可开关，默认关).
|
||||
Enabled bool `env:"LLM_LOCAL_ENABLED" default:"false"`
|
||||
// Provider is the local endpoint's wire protocol: "openai" (llama.cpp /
|
||||
// ollama / vLLM compatible endpoints) or "anthropic". Empty means openai.
|
||||
Provider string `env:"LLM_LOCAL_PROVIDER" default:"openai"`
|
||||
// BaseURL is the local endpoint's base URL (e.g. http://127.0.0.1:8080/v1).
|
||||
BaseURL string `env:"LLM_LOCAL_BASE_URL"`
|
||||
// APIKey is the local endpoint's key; empty is normal for a local server.
|
||||
APIKey string `env:"LLM_LOCAL_API_KEY"`
|
||||
// Model is the local model name. Without it the rung stays off.
|
||||
Model string `env:"LLM_LOCAL_MODEL"`
|
||||
}
|
||||
|
||||
// Configured reports whether the local rung has enough configuration to act as a
|
||||
// chain step. Enabled alone is not enough: a missing endpoint or model must not
|
||||
// become a step that can only fail.
|
||||
func (l LocalLLMConfig) Configured() bool {
|
||||
return l.Enabled && strings.TrimSpace(l.BaseURL) != "" && strings.TrimSpace(l.Model) != ""
|
||||
}
|
||||
|
||||
// EmbeddingConfig holds the optional semantic embedding API settings. When
|
||||
// unset, the runtime falls back to built-in n-gram/FTS vectors.
|
||||
type EmbeddingConfig struct {
|
||||
@@ -466,6 +378,13 @@ func Load(envPath string) (*Config, error) {
|
||||
if err := decodeValue(reflect.ValueOf(cfg), fromEnv); err != nil {
|
||||
return nil, err
|
||||
}
|
||||
// 「未设置」与「显式 0」必须分开（见 ContextConfig.MaxTokensSet）：反射解码只能
|
||||
// 给出值本身，因此这里补上「这个键到底有没有被配过」。
|
||||
cfg.Context.MaxTokensSet = envProvided(ContextMaxTokensEnv, fromEnv)
|
||||
if cfg.LLM.ContextWindow < 0 {
|
||||
return nil, fmt.Errorf("config: LLM_CONTEXT_WINDOW 不能为负（%d）：0 = 未声明（按回落窗口推导）",
|
||||
cfg.LLM.ContextWindow)
|
||||
}
|
||||
cfg.DataDir = resolveConfiguredDataDir(cfg.DataDir)
|
||||
|
||||
if errs := validateStruct(reflect.ValueOf(cfg).Elem()); len(errs) > 0 {
|
||||
|
||||
Reference in new issue
Block a user