@duckmind/dm-windows-x64 0.60.6 → 0.60.8

package/extensions/greedysearch-dm/src/search/scale-aware.mjs

@@ -1,11 +1,61 @@
-function V(R="",E=240){let _=String(R).replaceAll(/\s+/g," ").trim();if(_.length<=E)return _;let D=_.slice(0,E),G=D.lastIndexOf(" ");return G>0?`${D.slice(0,G)}...`:`${D}...`}import{spawn as a}from"node:child_process";import{basename as p}from"node:path";function P(R=process.env,E=process.execPath){let _=R.GREEDY_SEARCH_NODE||R.NODE_BINARY||R.NODE;if(_?.trim())return _.trim();let D=p(E||"").toLowerCase();if(D==="node"||D==="node.exe")return E;return"node"}import{existsSync as H,mkdirSync as x,readFileSync as M,writeFileSync as y}from"node:fs";import{homedir as z}from"node:os";import{join as B}from"node:path";import{tmpdir as h}from"node:os";function l(R,E=9222){let _=Number.parseInt(String(R??""),10);return Number.isInteger(_)&&_>1024&&_<65535?_:E}var m=h().replaceAll("\\","/"),FE=l(process.env.GREEDY_SEARCH_PORT),$=(process.env.GREEDY_SEARCH_PROFILE_DIR||process.env.CDP_PROFILE_DIR||`${m}/greedysearch-chrome-profile`).replaceAll("\\","/"),CE=`${$}/DevToolsActivePort`,gE=process.env.GREEDY_SEARCH_PID_FILE||`${$}/browser.pid`,AE=process.env.CDP_PAGES_CACHE||`${$}/cdp-pages.json`,OE=process.env.GREEDY_SEARCH_MODE_FILE||`${$}/browser-mode`,HE=process.env.GREEDY_SEARCH_METADATA_FILE||`${$}/browser-metadata.json`,LE=process.env.GREEDY_SEARCH_LAUNCH_LOCK_FILE||`${$}/browser-launch.lock`,vE=process.env.GREEDY_SEARCH_ACTIVITY_FILE||`${$}/browser-last-activity`,VE=(process.env.CDP_SOCKET_DIR||`${$}/cdp-sockets`).replaceAll("\\","/"),ZE=`${$}/visible-recovery.jsonl`,K=B(z(),".dm"),j=B(K,"greedyconfig"),Q=["perplexity","google","chatgpt","gemini"],W="gemini";function u(){try{if(H(j)){let R=M(j,"utf8"),E=JSON.parse(R);if(Array.isArray(E.engines)&&E.engines.length>0&&E.engines.every((_)=>typeof _==="string")){let _=E.engines.filter((G)=>Z[G]),D=E.engines.filter((G)=>!Z[G]);if(D.length>0)process.stderr.write(`[greedysearch] Warning: ignoring unknown engine(s) in ${j}: ${D.join(", ")}
-[greedysearch] Available engines: ${Object.keys(Z).join(", ")}
-`);if(_.length>0)return _;process.stderr.write(`[greedysearch] Warning: no valid engines in ${j}, falling back to defaults: ${Q.join(", ")}
-`)}}}catch{}return Q}function n(){try{if(!H(K))x(K,{recursive:!0});if(!H(j))y(j,JSON.stringify({engines:Q,synthesizer:W},null,2)+`
-`,"utf8")}catch{}}n();var O=["gemini","chatgpt"];function i(){try{if(H(j)){let R=M(j,"utf8"),E=JSON.parse(R);if(typeof E.synthesizer==="string"){let _=E.synthesizer.toLowerCase();if(O.includes(_))return _;process.stderr.write(`[greedysearch] Warning: unknown synthesizer "${E.synthesizer}" in ${j}
-[greedysearch] Available synthesizers: ${O.join(", ")}
-[greedysearch] Falling back to default: ${W}
-`)}}}catch{}return W}var Z={perplexity:"perplexity.mjs",p:"perplexity.mjs",bing:"bing-copilot.mjs",b:"bing-copilot.mjs",google:"google-ai.mjs",g:"google-ai.mjs",gemini:"gemini.mjs",gem:"gemini.mjs",chatgpt:"chatgpt.mjs",gpt:"chatgpt.mjs","semantic-scholar":"semantic-scholar.mjs",semanticscholar:"semantic-scholar.mjs",s2:"semantic-scholar.mjs",logically:"logically.mjs",log:"logically.mjs"},d=u();var KE=i(),QE=Math.max(1,Number.parseInt(process.env.GREEDY_FETCH_CONCURRENCY||"5",10)||5);process.env.CDP_PROFILE_DIR=$;import{existsSync as q}from"node:fs";import{dirname as t,join as L}from"node:path";import{fileURLToPath as s}from"node:url";function T(R){return s(new URL(".",R)).replace(/^\/([A-Z]:)/,"$1")}function c(R,E=q){let _=new Set;for(let D of R.filter(Boolean)){if(_.has(D))continue;if(_.add(D),E(D))return D}return R.filter(Boolean).at(-1)}function b(R,{moduleDir:E,entrypoint:_=process.argv[1],env:D=process.env,exists:G=q}={}){let Y=_?t(_):null,I=D.GREEDY_SEARCH_EXTENSION_DIR?.trim()||null;return c([E?L(E,"..","..","extractors",R):null,E?L(E,"..","extractors",R):null,I?L(I,"extractors",R):null,Y?L(Y,"..","extractors",R):null],G)}function o(R){let E="",_=!1,D=!1;for(let G of String(R)){if(D){E+=G,D=!1;continue}if(G==="\\"){E+=G,D=!0;continue}if(G==='"'){_=!_,E+=G;continue}if(_&&G===`
-`)E+="\\n";else if(_&&G==="\r")E+="\\r";else if(_&&G==="\t")E+="\\t";else E+=G}return E}function J(R){if(!R)return null;let E=String(R).trim(),_=E.indexOf("BEGIN_JSON"),D=E.indexOf("END_JSON");if(_!==-1&&D!==-1&&_<D)E=E.slice(_+10,D).trim();else{let N=E.indexOf("{");if(N>0)E=E.slice(N)}let G=[E,E.replace(/^```json\s*/i,"").replace(/^```\s*/i,"").replace(/```$/i,"").trim()],Y=E.indexOf("{"),I=E.lastIndexOf("}");if(Y!==-1&&I!==-1&&Y<I)G.push(E.slice(Y,I+1));for(let N of[...G]){let F=o(N);if(F!==N)G.push(F)}for(let N of G)try{return JSON.parse(N)}catch{}return null}var r=T(import.meta.url),e={gemini:"gemini.mjs",chatgpt:"chatgpt.mjs"};function EE(R="gemini"){let E=String(R||"gemini").toLowerCase();if(E==="gem")return"gemini";if(E==="gpt")return"chatgpt";return E}async function _E(R,E,{tabPrefix:_=null,timeoutMs:D=180000,visible:G=null}={}){let Y=EE(R),I=e[Y];if(!I||!O.includes(Y))throw Error(`Unsupported synthesizer "${R}". Supported: ${O.join(", ")}`);return new Promise((N,F)=>{let w=b(I,{moduleDir:r}),f=_?["--tab",String(_)]:[],g={...process.env,CDP_PROFILE_DIR:$};if(G!==!0)delete g.GREEDY_SEARCH_VISIBLE,delete g.GREEDY_SEARCH_ALWAYS_VISIBLE;else g.GREEDY_SEARCH_VISIBLE="1",g.GREEDY_SEARCH_ALWAYS_VISIBLE="1";let C=a(P(),[w,"--stdin",...f],{stdio:["pipe","pipe","pipe"],env:g});C.stdin.write(E),C.stdin.end();let v="",X="";C.stdout.on("data",(A)=>v+=A),C.stderr.on("data",(A)=>X+=A);let S=setTimeout(()=>{C.kill(),F(Error(`${Y} prompt timed out after ${D/1000}s`))},D);C.on("close",(A)=>{if(clearTimeout(S),A!==0){F(Error(X.trim()||`${Y} extractor failed`));return}try{N(JSON.parse(v.trim()))}catch{F(Error(`bad JSON from ${Y}: ${v.slice(0,100)}`))}})})}async function U(R,E={}){return _E("gemini",R,E)}var RE=30000;function k(R,E,_,D){let G=Number.parseInt(String(R??""),10);if(!Number.isFinite(G))return D;return Math.min(_,Math.max(E,G))}async function uE(R){let E=["You are a research complexity classifier.","Classify the following query by research complexity.","","- simple: A narrow factual question (what is X, define X, how does X work).","  Answerable with 1-3 search queries and a short synthesis. No sub-questions.","- moderate: A focused comparison, recent change, or best-practice lookup.","  Needs 2-4 angles but stays within one domain.","- complex: Multi-faceted survey, landscape analysis, or cross-domain investigation.","  Benefits from parallel research directions and iterative deepening.","","Respond ONLY with JSON wrapped in BEGIN_JSON / END_JSON markers:","BEGIN_JSON",JSON.stringify({complexity:"simple",reasoning:"narrow factual question",suggestedBreadth:1,suggestedIterations:1,needsAcademicSources:!1},null,2),"END_JSON","","Query: "+R].join(`
-`);try{let _=await U(E,{timeoutMs:RE}),D=J(_?.answer||"")||{},G=["simple","moderate","complex"].includes(D.complexity)?D.complexity:"moderate";return{complexity:G,reasoning:V(D.reasoning||"",200),suggestedBreadth:k(D.suggestedBreadth,1,5,G==="simple"?1:3),suggestedIterations:k(D.suggestedIterations,1,3,G==="simple"?1:2),needsAcademicSources:D.needsAcademicSources===!0}}catch(_){return process.stderr.write(`[greedysearch] Complexity classification failed, defaulting to moderate: ${_.message}
-`),{complexity:"moderate",reasoning:"classification failed",suggestedBreadth:3,suggestedIterations:2,needsAcademicSources:!1}}}export{uE as classifyResearchComplexity};
+import { trimText } from "./sources.mjs";
+import { runGeminiPrompt } from "./synthesis-runner.mjs";
+import { parseStructuredJson } from "./synthesis.mjs";
+const COMPLEXITY_PROMPT_TIMEOUT_MS = 30000;
+function clampInt(value, min, max, fallback) {
+  const n = Number.parseInt(String(value ?? ""), 10);
+  if (!Number.isFinite(n))
+    return fallback;
+  return Math.min(max, Math.max(min, n));
+}
+export async function classifyResearchComplexity(query) {
+  const prompt = [
+    "You are a research complexity classifier.",
+    "Classify the following query by research complexity.",
+    "",
+    "- simple: A narrow factual question (what is X, define X, how does X work).",
+    "  Answerable with 1-3 search queries and a short synthesis. No sub-questions.",
+    "- moderate: A focused comparison, recent change, or best-practice lookup.",
+    "  Needs 2-4 angles but stays within one domain.",
+    "- complex: Multi-faceted survey, landscape analysis, or cross-domain investigation.",
+    "  Benefits from parallel research directions and iterative deepening.",
+    "",
+    "Respond ONLY with JSON wrapped in BEGIN_JSON / END_JSON markers:",
+    "BEGIN_JSON",
+    JSON.stringify({
+      complexity: "simple",
+      reasoning: "narrow factual question",
+      suggestedBreadth: 1,
+      suggestedIterations: 1,
+      needsAcademicSources: false
+    }, null, 2),
+    "END_JSON",
+    "",
+    "Query: " + query
+  ].join(`
+`);
+  try {
+    const raw = await runGeminiPrompt(prompt, {
+      timeoutMs: COMPLEXITY_PROMPT_TIMEOUT_MS
+    });
+    const parsed = parseStructuredJson(raw?.answer || "") || {};
+    const complexity = ["simple", "moderate", "complex"].includes(parsed.complexity) ? parsed.complexity : "moderate";
+    return {
+      complexity,
+      reasoning: trimText(parsed.reasoning || "", 200),
+      suggestedBreadth: clampInt(parsed.suggestedBreadth, 1, 5, complexity === "simple" ? 1 : 3),
+      suggestedIterations: clampInt(parsed.suggestedIterations, 1, 3, complexity === "simple" ? 1 : 2),
+      needsAcademicSources: parsed.needsAcademicSources === true
+    };
+  } catch (error) {
+    process.stderr.write(`[greedysearch] Complexity classification failed, defaulting to moderate: ${error.message}
+`);
+    return {
+      complexity: "moderate",
+      reasoning: "classification failed",
+      suggestedBreadth: 3,
+      suggestedIterations: 2,
+      needsAcademicSources: false
+    };
+  }
+}