@te-river/opencode-team-mode 1.5.11 → 1.5.13

package/dist/tm/search.d.ts

@@ -3,16 +3,27 @@
  *
  * One call, one query, clean results: the engine URL is built here, fetched
  * through the SAME pipeline as tm_webfetch (allowlist, manual redirects,
- * 20 s timeout, 2 MB cap) and collapsed into a numbered title+URL hit list
- * before governance — the agent never sees raw SERP chrome.  Two engines
- * return structured JSON (npm registry search, GitHub repo search API) and
- * render with versions/stars/descriptions.
+ * 20 s timeout, 2 MB cap) and collapsed into a numbered title+URL(+snippet)
+ * hit list before governance — the agent never sees raw SERP chrome.  The
+ * structured engines (npm registry, GitHub repo search, StackExchange
+ * question search, HN Algolia, MediaWiki) render with stars/score/tags/
+ * snippets.
  *
- * Engine set is chosen for mainland-China reachability without API keys:
- * bing CN + international (ensearch=1), sogou, 360 (so.com), baidu,
- * bilibili, plus the npm/GitHub data sources.  Baidu is the flakiest
- * (anti-bot) and stays in the table because it is sometimes the only
- * engine indexing CN-specific content — failures name the alternatives.
+ * T4 quality upgrade (live benchmark 2026-09-14, CN host): of the HTML
+ * SERPs only bing still returns real results — sogou/so/baidu serve
+ * anti-bot shells and bing-int (ensearch=1) is 100% dead — so they are
+ * REMOVED from the engine table (webfetch's domain whitelist keeps their
+ * hosts untouched; users can still fetch those URLs directly).  In their
+ * place two no-key JSON engines: stackoverflow (api.stackexchange.com,
+ * 300 req/day/IP — quota_remaining is tracked and auto-routing degrades to
+ * bing once exhausted) and hn (hn.algolia.com).
+ *
+ * `engine: "auto"` (the new default, TM_SEARCH_DEFAULT_ENGINE) classifies
+ * the query, fans out to the matching engine set IN PARALLEL, normalizes
+ * every engine's hits to SearchHit, dedupes by host+path and re-ranks
+ * with weighted RRF (stackoverflow 0.4 / bing 0.4 / others 0.2) into a
+ * top-10 list tagged with its source engine(s).  Explicit single-engine
+ * calls keep the classic per-engine render.
  *
  * tm_webfetch stays the "fetch a KNOWN URL" tool; tm_browser covers
  * JS-rendered pages.  All three share one domain allowlist and one
@@ -23,14 +34,79 @@ import { type TmConfig } from "./config.js";
 import type { TmPipelines } from "./pipelines.js";
 export interface SearchEngine {
     name: string;
-    /** html engines are extracted via extractSearchHits; json engines have
-     *  a dedicated renderer below. */
-    kind: "html" | "npm-json" | "github-json" | "wiki-json";
+    /** html engines are extracted via extractSearchHits; json engines have a
+     *  dedicated parser/renderer below. */
+    kind: "html" | "npm-json" | "github-json" | "wiki-json" | "so-json" | "hn-json";
     buildUrl: (encodedQuery: string) => string;
     /** Hint appended when the engine returns nothing usable. */
     alt: string;
+    /** Transform the RAW query before URL-encoding (CJK phrase protection,
+     *  GitHub qualifier folding).  Identity when absent. */
+    prepareQuery?: (query: string) => string;
 }
-/** The engine table — keys are the tm_search `engine` arg values. */
+/**
+ * CJK phrase protection: a multi-word Chinese query gets its CORE phrase
+ * (the token holding the most CJK chars, ≥2) wrapped in double quotes —
+ * CN SERPs otherwise free-match tokens across pages and drown the result
+ * list in tangential hits.  Single-word queries and already-quoted queries
+ * pass through untouched (never fabricate a phrase boundary, never fight
+ * the user's own quoting).  Applied to bing via prepareQuery; JSON query
+ * APIs are skipped because they treat quotes literally.
+ */
+export declare function protectCjkPhrase(query: string): string;
+export declare function foldGithubQualifiers(query: string): {
+    q: string;
+    qualifiers: string[];
+};
+export type QueryClass = "error-code" | "dev-ecosystem" | "cjk" | "general";
+export declare function classifyQuery(query: string): QueryClass;
+export declare const AUTO_ENGINE_KEY = "auto";
+/** RRF fusion constants.  bing LOST its 0.4 trust weight (2026-09-18): with
+ *  0.4-vs-0.2 weights and a k=60 damping range of only 61..70, bing's WORST
+ *  hit (0.4/70 = 0.0057) outranked every other engine's BEST (0.2/61 =
+ *  0.0033) — a weight ratio above the whole rank-damping range means the
+ *  under-weighted engine can never place a single hit.  bing is still in the
+ *  table (it is the only live CN HTML SERP); it now has to EARN rank by
+ *  overlapping the query.  Override per engine via TM_SEARCH_WEIGHTS. */
+export declare const RRF_K = 60;
+export declare const RRF_WEIGHTS: Record<string, number>;
+export declare const RRF_DEFAULT_WEIGHT = 0.2;
+/** One engine's fusion weight: a TM_SEARCH_WEIGHTS override first, then the
+ *  built-in table, then the default. */
+export declare function weightFor(engine: string, overrides?: Record<string, number>): number;
+/**
+ * Tokenize a query for the relevance floor.  CJK has no word boundaries and
+ * the plugin ships no segmenter, so CJK runs index as every BIGRAM: "关键
+ * 词" contributes 关键/键词/词… which a Chinese title contains verbatim.
+ * Substring matching against that is a cheap, deterministic overlap proxy —
+ * no model, no network, no fabricated relevance.
+ */
+export declare function queryTerms(query: string): string[];
+/** 0..1 overlap between the query and one hit (title + snippet + host). */
+export declare function relevanceOf(terms: readonly string[], hit: SearchHit): number;
+/** Mutable SO quota state (per process).  api.stackexchange answers every
+ *  payload with quota_remaining/quota_max; hitting 0 flips `exhausted`
+ *  and engine:"auto" swaps the SO leg for bing until the process restarts
+ *  (the daily quota rolls over on the server, not here). */
+export declare const SO_QUOTA: {
+    max: number;
+    remaining: number | null;
+    exhausted: boolean;
+};
+export declare function resetSoQuota(): void;
+/** Resolve the auto route set for a query (exported for tests + PTC).
+ *  `disabled` (TM_SEARCH_DISABLED_ENGINES) removes legs; if that empties a
+ *  route the caller still gets bing rather than an error — a search tool
+ *  that refuses to answer because of its own config is worse than a
+ *  single-engine answer. */
+export declare function resolveAutoRoutes(query: string, disabled?: readonly string[]): {
+    routes: string[];
+    queryClass: QueryClass;
+    notes: string[];
+};
+/** The engine table — keys are the tm_search `engine` arg values.  The
+ *  dead CN SERPs (sogou/so/baidu/bing-int) are GONE, not even manually
+ *  selectable (2026-09-14 benchmark: anti-bot shells / 100% empty). */
 export declare const SEARCH_ENGINES: Record<string, SearchEngine>;
 export declare const SEARCH_ENGINE_NAMES: string[];
 /** npm registry search JSON → numbered package list (name@version — desc). */
@@ -40,9 +116,45 @@ export declare function renderGithubResults(query: string, body: string): string
 /** MediaWiki search API JSON (moegirl) → numbered entry list with
  *  tag-stripped snippets and direct article URLs. */
 export declare function renderWikiResults(query: string, body: string): string | null;
-/** Collapse an HTML SERP into the compact hit list; null when the engine
+/** StackExchange search JSON → numbered question list; the per-hit snippet
+ *  is SYNTHESIZED from the engine's own metadata (score/answers/tags —
+ *  search/advanced carries no body text).  The quota footer reflects the
+ *  tracked quota_remaining. */
+export declare function renderSoResults(query: string, body: string): string | null;
+/** HN Algolia story search JSON → numbered post list; the per-hit snippet
+ *  is the engine's own points/comments/date metadata. */
+export declare function renderHnResults(query: string, body: string): string | null;
+/** Collapse an HTML SERP into the compact hit list; [] when the engine
  *  gave nothing extractable (anti-bot shell / markup change). */
-export declare function extractHtmlResults(html: string): SearchHit[];
+export declare function extractHtmlResults(html: string, max?: number): SearchHit[];
+/** host+path identity across engines (query/fragment/trailing-slash/www
+ *  ignored) — the fusion dedupe key. */
+export declare function dedupeKey(url: string): string;
+interface FusionGroup {
+    engine: string;
+    hits: SearchHit[];
+}
+/** Built-in zero-overlap multiplier (TmConfig.searchRelevanceFloor overrides). */
+export declare const RELEVANCE_FLOOR_DEFAULT = 0.35;
+/**
+ * Reciprocal Rank Fusion over the routed engines' hit lists: normalize →
+ * host+path dedupe → score = Σ (weight × relevance × fetchability)/(k+rank)
+ * → top-N, every hit tagged with its contributing engine(s).
+ *
+ * The relevance term is the fix for "auto returns a bing mirror": pure
+ * rank-position fusion lets a trusted engine's junk outrank another
+ * engine's best hit, so a hit that shares no token with the query now keeps
+ * only a floor fraction of its engine's weight.  Nothing is deleted — bing
+ * still carries useful data, and an over-filtered list is unfixable from
+ * the agent side.
+ */
+export declare function fuseRrf(groups: FusionGroup[], opts?: {
+    max?: number;
+    query?: string;
+    weights?: Record<string, number>;
+    floor?: number;
+    fetchable?: (url: string) => boolean;
+}): SearchHit[];
 /** Build the tm_search ToolDefinition over the SHARED main pipelines
  *  instance.  `fetchImpl` is injectable for tests. */
 export declare function buildTmSearchTool(deps: {
@@ -55,4 +167,5 @@ export declare function buildTmSearchTool(deps: {
     args: Record<string, unknown>;
     execute: (rawArgs: Record<string, unknown>, ctx: unknown) => Promise<import("../types.js").ToolResult>;
 };
+export {};
 //# sourceMappingURL=search.d.ts.map

package/dist/tm/search.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"search.d.ts","sourceRoot":"","sources":["../../src/tm/search.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;GAmBG;AAEH,OAAO,EAAkE,KAAK,SAAS,EAAE,KAAK,SAAS,EAAE,MAAM,eAAe,CAAA;AAE9H,OAAO,EAAW,KAAK,QAAQ,EAAE,MAAM,aAAa,CAAA;AAGpD,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,gBAAgB,CAAA;AAEjD,MAAM,WAAW,YAAY;IAC3B,IAAI,EAAE,MAAM,CAAA;IACZ;sCACkC;IAClC,IAAI,EAAE,MAAM,GAAG,UAAU,GAAG,aAAa,GAAG,WAAW,CAAA;IACvD,QAAQ,EAAE,CAAC,YAAY,EAAE,MAAM,KAAK,MAAM,CAAA;IAC1C,4DAA4D;IAC5D,GAAG,EAAE,MAAM,CAAA;CACZ;AAED,qEAAqE;AACrE,eAAO,MAAM,cAAc,EAAE,MAAM,CAAC,MAAM,EAAE,YAAY,CAuDvD,CAAA;AAED,eAAO,MAAM,mBAAmB,UAA8B,CAAA;AAI9D,8EAA8E;AAC9E,wBAAgB,gBAAgB,CAAC,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,GAAG,MAAM,GAAG,IAAI,CAmB3E;AAED,2EAA2E;AAC3E,wBAAgB,mBAAmB,CAAC,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,GAAG,MAAM,GAAG,IAAI,CAsB9E;AAED;qDACqD;AACrD,wBAAgB,iBAAiB,CAAC,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,GAAG,MAAM,GAAG,IAAI,CA2B5E;AAED;iEACiE;AACjE,wBAAgB,kBAAkB,CAAC,IAAI,EAAE,MAAM,GAAG,SAAS,EAAE,CAE5D;AAUD;sDACsD;AACtD,wBAAgB,iBAAiB,CAAC,IAAI,EAAE;IACtC,SAAS,EAAE,WAAW,CAAA;IACtB,GAAG,EAAE,QAAQ,CAAA;IACb,IAAI,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAAA;IAC9B,SAAS,CAAC,EAAE,SAAS,CAAA;CACtB,GAAG;IACF,WAAW,EAAE,MAAM,CAAA;IACnB,IAAI,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAAA;IAC7B,OAAO,EAAE,CAAC,OAAO,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,EAAE,GAAG,EAAE,OAAO,KAAK,OAAO,CAAC,OAAO,aAAa,EAAE,UAAU,CAAC,CAAA;CACvG,CA8FA"}
+{"version":3,"file":"search.d.ts","sourceRoot":"","sources":["../../src/tm/search.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA8BG;AAEH,OAAO,EAML,KAAK,SAAS,EACd,KAAK,SAAS,EACf,MAAM,eAAe,CAAA;AAEtB,OAAO,EAAW,KAAK,QAAQ,EAAE,MAAM,aAAa,CAAA;AAGpD,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,gBAAgB,CAAA;AAEjD,MAAM,WAAW,YAAY;IAC3B,IAAI,EAAE,MAAM,CAAA;IACZ;2CACuC;IACvC,IAAI,EAAE,MAAM,GAAG,UAAU,GAAG,aAAa,GAAG,WAAW,GAAG,SAAS,GAAG,SAAS,CAAA;IAC/E,QAAQ,EAAE,CAAC,YAAY,EAAE,MAAM,KAAK,MAAM,CAAA;IAC1C,4DAA4D;IAC5D,GAAG,EAAE,MAAM,CAAA;IACX;4DACwD;IACxD,YAAY,CAAC,EAAE,CAAC,KAAK,EAAE,MAAM,KAAK,MAAM,CAAA;CACzC;AAID;;;;;;;;GAQG;AACH,wBAAgB,gBAAgB,CAAC,KAAK,EAAE,MAAM,GAAG,MAAM,CAkBtD;AAQD,wBAAgB,oBAAoB,CAAC,KAAK,EAAE,MAAM,GAAG;IAAE,CAAC,EAAE,MAAM,CAAC;IAAC,UAAU,EAAE,MAAM,EAAE,CAAA;CAAE,CASvF;AAID,MAAM,MAAM,UAAU,GAAG,YAAY,GAAG,eAAe,GAAG,KAAK,GAAG,SAAS,CAAA;AAe3E,wBAAgB,aAAa,CAAC,KAAK,EAAE,MAAM,GAAG,UAAU,CAKvD;AAcD,eAAO,MAAM,eAAe,SAAS,CAAA;AAErC;;;;;;yEAMyE;AACzE,eAAO,MAAM,KAAK,KAAK,CAAA;AACvB,eAAO,MAAM,WAAW,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAA0B,CAAA;AACzE,eAAO,MAAM,kBAAkB,MAAM,CAAA;AAErC;wCACwC;AACxC,wBAAgB,SAAS,CAAC,MAAM,EAAE,MAAM,EAAE,SAAS,GAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAM,GAAG,MAAM,CAIxF;AAED;;;;;;GAMG;AACH,wBAAgB,UAAU,CAAC,KAAK,EAAE,MAAM,GAAG,MAAM,EAAE,CAelD;AAED,2EAA2E;AAC3E,wBAAgB,WAAW,CAAC,KAAK,EAAE,SAAS,MAAM,EAAE,EAAE,GAAG,EAAE,SAAS,GAAG,MAAM,CAM5E;AAED;;;4DAG4D;AAC5D,eAAO,MAAM,QAAQ,EAAE;IAAE,GAAG,EAAE,MAAM,CAAC;IAAC,SAAS,EAAE,MAAM,GAAG,IAAI,CAAC;IAAC,SAAS,EAAE,OAAO,CAAA;CAIjF,CAAA;AAED,wBAAgB,YAAY,IAAI,IAAI,CAInC;AAED;;;;4BAI4B;AAC5B,wBAAgB,iBAAiB,CAC/B,KAAK,EAAE,MAAM,EACb,QAAQ,GAAE,SAAS,MAAM,EAAO,GAC/B;IAAE,MAAM,EAAE,MAAM,EAAE,CAAC;IAAC,UAAU,EAAE,UAAU,CAAC;IAAC,KAAK,EAAE,MAAM,EAAE,CAAA;CAAE,CAgB/D;AAID;;uEAEuE;AACvE,eAAO,MAAM,cAAc,EAAE,MAAM,CAAC,MAAM,EAAE,YAAY,CA8CvD,CAAA;AAED,eAAO,MAAM,mBAAmB,UAA8B,CAAA;AA0F9D,8EAA8E;AAC9E,wBAAgB,gBAAgB,CAAC,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,GAAG,MAAM,GAAG,IAAI,CAmB3E;AAED,2EAA2E;AAC3E,wBAAgB,mBAAmB,CAAC,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,GAAG,MAAM,GAAG,IAAI,CAsB9E;AAED;qDACqD;AACrD,wBAAgB,iBAAiB,CAAC,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,GAAG,MAAM,GAAG,IAAI,CA2B5E;AAED;;;+BAG+B;AAC/B,wBAAgB,eAAe,CAAC,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,GAAG,MAAM,GAAG,IAAI,CAa1E;AAED;yDACyD;AACzD,wBAAgB,eAAe,CAAC,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,GAAG,MAAM,GAAG,IAAI,CAW1E;AAED;iEACiE;AACjE,wBAAgB,kBAAkB,CAAC,IAAI,EAAE,MAAM,EAAE,GAAG,SAAkB,GAAG,SAAS,EAAE,CAEnF;AAsED;wCACwC;AACxC,wBAAgB,SAAS,CAAC,GAAG,EAAE,MAAM,GAAG,MAAM,CAS7C;AAED,UAAU,WAAW;IACnB,MAAM,EAAE,MAAM,CAAA;IACd,IAAI,EAAE,SAAS,EAAE,CAAA;CAClB;AAED,kFAAkF;AAClF,eAAO,MAAM,uBAAuB,OAAO,CAAA;AAE3C;;;;;;;;;;;GAWG;AACH,wBAAgB,OAAO,CACrB,MAAM,EAAE,WAAW,EAAE,EACrB,IAAI,GAAE;IACJ,GAAG,CAAC,EAAE,MAAM,CAAA;IACZ,KAAK,CAAC,EAAE,MAAM,CAAA;IACd,OAAO,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAA;IAChC,KAAK,CAAC,EAAE,MAAM,CAAA;IACd,SAAS,CAAC,EAAE,CAAC,GAAG,EAAE,MAAM,KAAK,OAAO,CAAA;CAChC,GACL,SAAS,EAAE,CA2Cb;AA+BD;sDACsD;AACtD,wBAAgB,iBAAiB,CAAC,IAAI,EAAE;IACtC,SAAS,EAAE,WAAW,CAAA;IACtB,GAAG,EAAE,QAAQ,CAAA;IACb,IAAI,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAAA;IAC9B,SAAS,CAAC,EAAE,SAAS,CAAA;CACtB,GAAG;IACF,WAAW,EAAE,MAAM,CAAA;IACnB,IAAI,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAAA;IAC7B,OAAO,EAAE,CAAC,OAAO,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,EAAE,GAAG,EAAE,OAAO,KAAK,OAAO,CAAC,OAAO,aAAa,EAAE,UAAU,CAAC,CAAA;CACvG,CA0LA"}