@duckmind/dm-windows-x64 0.62.3 → 0.62.9

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

@@ -1,61 +0,0 @@
-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
-    };
-  }
-}

package/extensions/greedysearch-dm/src/search/simple-research.mjs

@@ -1,418 +0,0 @@
-import { ALL_ENGINES, RESEARCH_ENGINES } from "./constants.mjs";
-import {
-  buildSourceRegistry,
-  mergeFetchDataIntoSources,
-  normalizeUrl,
-  trimText
-} from "./sources.mjs";
-import {
-  auditCitations,
-  buildFinalReportPrompt,
-  buildSynthesisFromEvidencePrompt,
-  computeResearchFloor,
-  createQuestionLedger,
-  extractEvidenceFromSources,
-  reconcileQuestionsFromSynthesis,
-  runCitationUrlCheck,
-  writeResearchBundle
-} from "./research.mjs";
-import { parseStructuredJson } from "./synthesis.mjs";
-import { writeSourcesToFiles } from "./file-sources.mjs";
-import { fetchMultipleSources } from "./fetch-source.mjs";
-import { runGeminiPrompt } from "./synthesis-runner.mjs";
-import { createProgressTracker } from "./progress.mjs";
-import { spawn } from "node:child_process";
-import { nodeRuntimeCommand } from "../utils/node-runtime.mjs";
-import { join } from "node:path";
-import { fileURLToPath } from "node:url";
-const __dir = fileURLToPath(new URL(".", import.meta.url)).replace(/^\/([A-Z]:)/, "$1");
-const SEARCH_BIN = join(__dir, "..", "..", "bin", "search.mjs");
-function uniqueStrings(items, limit = 1 / 0) {
-  const seen = new Set;
-  const out = [];
-  for (const item of items || []) {
-    const clean = trimText(String(item || ""), 1000);
-    if (!clean || seen.has(clean))
-      continue;
-    seen.add(clean);
-    out.push(clean);
-    if (out.length >= limit)
-      break;
-  }
-  return out;
-}
-export function buildSearchAngles(query) {
-  const trimmed = String(query || "").trim();
-  if (!trimmed)
-    return [];
-  return [
-    `${trimmed} — definition and overview`,
-    `${trimmed} — how it works, mechanism, or key details`,
-    `${trimmed} — current usage, comparison, or best practices`
-  ];
-}
-export function mergeSourcesByUrl(existing, incoming) {
-  const urlMap = new Map;
-  for (const s of existing || []) {
-    const key = s?.canonicalUrl || s?.finalUrl || s?.url;
-    if (key)
-      urlMap.set(key, s);
-  }
-  for (const s of incoming || []) {
-    const key = s?.canonicalUrl || s?.finalUrl || s?.url;
-    if (!key)
-      continue;
-    if (urlMap.has(key)) {
-      const merged = {
-        ...urlMap.get(key),
-        angles: [
-          ...urlMap.get(key).angles || [urlMap.get(key).query || ""],
-          s.query || ""
-        ]
-      };
-      urlMap.set(key, merged);
-    } else {
-      urlMap.set(key, s);
-    }
-  }
-  return Array.from(urlMap.values());
-}
-const _enginePattern = ALL_ENGINES.join("|");
-const _engineRegex = new RegExp(`^\\[(${_enginePattern})\\]`);
-function shouldForwardChildStderr(line) {
-  return /^PROGRESS:/.test(line) || /^\[greedysearch\]/.test(line) || _engineRegex.test(line) || /^GreedySearch Chrome/.test(line) || /^Launching GreedySearch Chrome/.test(line) || /^Headless mode/.test(line) || /^Ready\.?$/.test(line);
-}
-async function runFastAllSearch(query, { locale = null, short = true } = {}) {
-  const args = [SEARCH_BIN, "all", "--inline", "--stdin", "--fast"];
-  if (!short)
-    args.push("--full");
-  if (locale)
-    args.push("--locale", locale);
-  return new Promise((resolve, reject) => {
-    const proc = spawn(nodeRuntimeCommand(), args, {
-      stdio: ["pipe", "pipe", "pipe"],
-      env: { ...process.env, GREEDY_SEARCH_RESEARCH_CHILD: "1" }
-    });
-    proc.stdin.write(query);
-    proc.stdin.end();
-    let out = "";
-    let err = "";
-    let stderrBuffer = "";
-    proc.stdout.on("data", (d) => out += d);
-    proc.stderr.on("data", (d) => {
-      err += d;
-      stderrBuffer += d.toString();
-      const lines = stderrBuffer.split(`
-`);
-      stderrBuffer = lines.pop() || "";
-      for (const line of lines) {
-        if (shouldForwardChildStderr(line)) {
-          process.stderr.write(`${line}
-`);
-        }
-      }
-    });
-    const t = setTimeout(() => {
-      proc.kill();
-      reject(new Error(`research child search timed out for: ${query}`));
-    }, 140000);
-    proc.on("close", (code) => {
-      clearTimeout(t);
-      if (code !== 0) {
-        reject(new Error(err.trim() || `search child exited with code ${code}`));
-        return;
-      }
-      try {
-        resolve(JSON.parse(out.trim()));
-      } catch {
-        reject(new Error(`Invalid JSON from research child: ${out.slice(0, 200)}`));
-      }
-    });
-  });
-}
-function annotateFetchedSourcesWithIds(fetchedSources, sources) {
-  const byUrl = new Map;
-  for (const source of sources || []) {
-    const key = normalizeUrl(source?.canonicalUrl || source?.finalUrl || source?.url);
-    if (key && source?.id)
-      byUrl.set(key, source.id);
-  }
-  return (fetchedSources || []).map((source, index) => {
-    const key = normalizeUrl(source?.finalUrl || source?.canonicalUrl || source?.url);
-    return {
-      ...source,
-      id: source?.id || byUrl.get(key) || `F${index + 1}`
-    };
-  });
-}
-function questionProgress(questions) {
-  const total = questions.length;
-  const closed = questions.filter((q) => q.status === "closed").length;
-  return { total, closed, open: Math.max(0, total - closed) };
-}
-export async function runSimpleResearchMode({
-  query,
-  locale = null,
-  maxSources = 5,
-  qualityThreshold = 8.5,
-  writeBundle = process.env.GREEDY_RESEARCH_BUNDLE !== "0",
-  researchOutDir = null
-} = {}) {
-  const startedAt = new Date().toISOString();
-  const startMs = Date.now();
-  const questions = createQuestionLedger(query);
-  const extractedSourceKeys = new Set;
-  process.stderr.write(`[greedysearch] Simple research mode: single-pass for "${trimText(query, 80)}"
-`);
-  const searchAnglesCount = 3;
-  const totalSteps = searchAnglesCount + 1 + 2;
-  const progressTracker = createProgressTracker({
-    totalActions: totalSteps,
-    totalRounds: 1,
-    totalFetches: 1,
-    silent: process.env.GREEDY_RESEARCH_QUIET === "1"
-  });
-  progressTracker.startRound(1);
-  let combinedSources = [];
-  let fetchedSources = [];
-  const searchAngles = buildSearchAngles(query);
-  const searchResults = [];
-  progressTracker.startAction("search", `${searchAngles.length} angles in parallel`);
-  const angleOutcomes = await Promise.allSettled(searchAngles.map((angle) => runFastAllSearch(angle, { locale, short: true })));
-  for (let i = 0;i < searchAngles.length; i++) {
-    const angle = searchAngles[i];
-    const outcome = angleOutcomes[i];
-    progressTracker.endAction();
-    if (outcome.status === "fulfilled") {
-      const result = outcome.value;
-      searchResults.push({ angle, result });
-      const sources = buildSourceRegistry(result, angle);
-      combinedSources = mergeSourcesByUrl(combinedSources, sources);
-    } else {
-      process.stderr.write(`[greedysearch] Simple search angle "${angle}" failed: ${outcome.reason.message}
-`);
-    }
-  }
-  process.stderr.write(`PROGRESS:research:simple:fetching
-`);
-  if (combinedSources.length > 0) {
-    try {
-      progressTracker.startFetch(`top ${Math.min(maxSources, combinedSources.length)} sources`);
-      fetchedSources = await fetchMultipleSources(combinedSources, Math.min(maxSources, combinedSources.length), 8000, Math.min(3, maxSources));
-      progressTracker.endFetch(true);
-      combinedSources = mergeFetchDataIntoSources(combinedSources, fetchedSources);
-    } catch (error) {
-      progressTracker.endFetch(false);
-      process.stderr.write(`[greedysearch] Source fetching failed: ${error.message}
-`);
-    }
-  }
-  fetchedSources = annotateFetchedSourcesWithIds(fetchedSources, combinedSources);
-  process.stderr.write(`PROGRESS:research:simple:evidence
-`);
-  let evidenceItems = [];
-  try {
-    const evidenceRun = await extractEvidenceFromSources({
-      query,
-      questions,
-      fetchedSources,
-      extractedSourceKeys
-    });
-    evidenceItems = evidenceRun.evidence || [];
-    for (const evidence of evidenceRun.evidence) {
-      const answered = Array.isArray(evidence.answers) ? evidence.answers : [];
-      for (const ans of answered) {
-        const id = ans?.id || ans?.question;
-        if (id) {
-          const target = questions.find((q) => q.id === id);
-          if (target) {
-            target.status = "closed";
-            target.closedRound = 1;
-            if (ans.evidence)
-              target.evidence = uniqueStrings([...target.evidence || [], ans.evidence], 4);
-          }
-        }
-      }
-      const newQs = Array.isArray(evidence.newQuestions) ? evidence.newQuestions : [];
-      for (const q of newQs) {
-        const clean = trimText(String(q), 320);
-        if (clean && !questions.some((x) => x.question === clean)) {
-          questions.push({
-            id: `Q${questions.length + 1}`,
-            question: clean,
-            status: "open",
-            reason: "Discovered gap/follow-up",
-            createdRound: 1,
-            evidence: [],
-            sourceIds: []
-          });
-        }
-      }
-    }
-  } catch (error) {
-    process.stderr.write(`[greedysearch] Evidence extraction failed: ${error.message}
-`);
-  }
-  process.stderr.write(`PROGRESS:research:simple:synthesizing
-`);
-  let synthesis = {
-    answer: "",
-    agreement: { level: "mixed", summary: "Single-pass synthesis." },
-    differences: [],
-    caveats: [],
-    claims: [],
-    recommendedSources: combinedSources.slice(0, 4).map((s) => s.id),
-    synthesized: false
-  };
-  if (evidenceItems.length > 0) {
-    try {
-      progressTracker.startAction("synth-evidence", "from evidence");
-      const rawReport = await runGeminiPrompt(buildSynthesisFromEvidencePrompt(query, combinedSources, questions, evidenceItems), { timeoutMs: 120000 });
-      progressTracker.endAction();
-      synthesis = {
-        ...synthesis,
-        ...parseStructuredJson(rawReport?.answer || "") || {}
-      };
-      synthesis.synthesized = Array.isArray(synthesis.claims) && synthesis.claims.length > 0;
-    } catch (error) {
-      process.stderr.write(`[greedysearch] Evidence synthesis failed: ${error.message}
-`);
-    }
-  }
-  if (!synthesis.synthesized && combinedSources.length > 0) {
-    try {
-      progressTracker.startAction("synth-final", "fallback report");
-      const rawReport = await runGeminiPrompt(buildFinalReportPrompt(query, [{ round: 1, learnings: [], gaps: [], actions: [] }], combinedSources, questions, evidenceItems), { timeoutMs: 120000 });
-      progressTracker.endAction();
-      synthesis = {
-        ...synthesis,
-        ...parseStructuredJson(rawReport?.answer || "") || {}
-      };
-      synthesis.synthesized = Array.isArray(synthesis.claims) && synthesis.claims.length > 0;
-    } catch (error) {
-      process.stderr.write(`[greedysearch] Final synthesis failed: ${error.message}
-`);
-    }
-  }
-  process.stderr.write(`PROGRESS:research:simple:audit
-`);
-  const citationAudit = auditCitations(synthesis.answer || "", combinedSources);
-  const citationUrls = await runCitationUrlCheck(combinedSources, citationAudit);
-  reconcileQuestionsFromSynthesis(questions, synthesis, citationAudit);
-  const allGaps = uniqueStrings(synthesis.caveats || []);
-  const floor = computeResearchFloor({
-    sources: combinedSources,
-    fetchedSources,
-    synthesis,
-    citationAudit,
-    gaps: allGaps,
-    questions,
-    rounds: [{ round: 1, actions: [], learnings: [], gaps: allGaps }],
-    qualityScore: synthesis.synthesized ? 8 : 5,
-    qualityThreshold,
-    maxSources
-  });
-  const finishedAt = new Date().toISOString();
-  const durationMs = Date.now() - startMs;
-  const baseManifest = {
-    startedAt,
-    finishedAt,
-    durationMs,
-    rounds: 1,
-    terminationReason: "simple_single_pass"
-  };
-  let bundle = null;
-  let fetchedFiles;
-  if (writeBundle) {
-    process.stderr.write(`PROGRESS:research:simple:bundle
-`);
-    try {
-      bundle = await writeResearchBundle({
-        query,
-        rounds: [
-          {
-            round: 1,
-            actions: [],
-            learnings: [],
-            gaps: allGaps,
-            evidence: evidenceItems
-          }
-        ],
-        sources: combinedSources,
-        fetchedSources,
-        evidenceItems,
-        synthesis,
-        citationAudit,
-        citationUrls,
-        floor,
-        manifest: {
-          ...baseManifest,
-          engines: RESEARCH_ENGINES,
-          synthesizer: "gemini",
-          actionsRun: 1,
-          searches: 1,
-          fetches: fetchedSources.length,
-          sourcesFetched: fetchedSources.filter((s) => s?.contentChars > 100).length,
-          engineFailures: [],
-          floorMet: floor.floorMet
-        },
-        allGaps,
-        questions,
-        outDir: researchOutDir
-      });
-      fetchedFiles = bundle.sourceFiles;
-      delete bundle.sourceFiles;
-    } catch (error) {
-      process.stderr.write(`[greedysearch] Research bundle write failed: ${error.message}
-`);
-      bundle = { error: error.message || String(error) };
-      fetchedFiles = await writeSourcesToFiles(fetchedSources);
-    }
-  } else {
-    fetchedFiles = await writeSourcesToFiles(fetchedSources);
-  }
-  process.stderr.write(`PROGRESS:research:done
-`);
-  progressTracker.endRound();
-  progressTracker.finish();
-  return {
-    query,
-    _research: {
-      mode: "simple",
-      breadth: 1,
-      iterations: 1,
-      maxSources,
-      rounds: [
-        {
-          round: 1,
-          actions: [],
-          learnings: [],
-          gaps: allGaps,
-          evidence: evidenceItems
-        }
-      ],
-      learnings: [],
-      gaps: allGaps,
-      evidence: evidenceItems,
-      questions,
-      questionProgress: questionProgress(questions),
-      qualityHistory: [synthesis.synthesized ? 8 : 5],
-      terminationReason: "simple_single_pass",
-      qualityThreshold,
-      floor,
-      bundle,
-      manifest: baseManifest
-    },
-    _citationAudit: citationAudit,
-    _citationUrls: citationUrls,
-    _sources: combinedSources,
-    _fetchedSources: fetchedFiles,
-    _synthesis: synthesis,
-    _confidence: {
-      sourcesCount: combinedSources.length,
-      fetchedSourceSuccessRate: fetchedSources.length > 0 ? fetchedSources.filter((source) => source.contentChars > 100).length / fetchedSources.length : 0,
-      agreementLevel: synthesis.agreement?.level || "mixed",
-      floorMet: floor.floorMet
-    }
-  };
-}