@angri450/zvec-grep 0.2.2

package/dist/engine/pipeline/search/index.js

@@ -0,0 +1,811 @@
+import { workspaceIndexDetail, detail, EngineError, errorDetails, } from "../../errors.js";
+import { TimingCollector } from "../../utils/timing.js";
+import { hasPathGlob, isAbsolutePathPattern, normalizePathForMatch, normalizePathPattern, pathPatternMatches, } from "../../utils/glob.js";
+import { matchesFileSelection, resolveFileTypePatterns, } from "../../utils/file-selection.js";
+const DEFAULT_LIMIT = 7;
+const RRF_K = 60;
+const RECALL_INITIAL_DEPTH = 200;
+const RECALL_MAX_DEPTH = 2000;
+const RECALL_GROWTH_FACTOR = 2;
+const RECALL_TARGET_FACTOR = 5;
+const RECALL_MIN_TARGET_CANDIDATES = 50;
+export async function searchWorkspaceIndex(plan, ctx) {
+    const timings = new TimingCollector();
+    const result = await timings.time("search_total", async () => {
+        const normalized = timings.timeSync("search_plan", () => validateSearchPlan(plan));
+        const limit = normalized.limit ?? DEFAULT_LIMIT;
+        const trace = normalized.trace === true || normalized.trackEntityId !== undefined;
+        const fileTypePatterns = await timings.time("search_file_types", () => resolveFileTypePatterns(normalized.fileTypes, normalized.excludedFileTypes));
+        const filter = timings.timeSync("search_filter", () => searchPlanToStorageFilter(normalized, ctx.storage, fileTypePatterns));
+        const hasSearchableFiles = !filterMatchesNoFiles(filter);
+        const candidates = new Map();
+        const vectorByRoute = hasSearchableFiles && planUsesVector(normalized)
+            ? await timings.time("query_embedding", () => embedVectorRoutes(normalized.routes, requireEmbeddingModel(ctx, "searchPlan")))
+            : new Map();
+        let recallDepth = RECALL_INITIAL_DEPTH;
+        if (hasSearchableFiles) {
+            recallDepth = timings.timeSync("recall", () => collectAdaptiveRecall({
+                routes: normalized.routes,
+                filter,
+                preferSymbol: normalized.preferSymbol === true,
+                vectorByRoute,
+                limit,
+                storage: ctx.storage,
+                candidates,
+            }));
+        }
+        if (normalized.trackEntityId) {
+            timings.timeSync("force_track", () => forceTrackEntity({
+                entityId: normalized.trackEntityId,
+                routes: normalized.routes,
+                vectorByRoute,
+                recallDepth,
+                filter,
+                storage: ctx.storage,
+                candidates,
+            }));
+        }
+        const fused = timings.timeSync("fusion", () => fuseCandidates(candidates));
+        const visible = fused.slice(0, limit);
+        const tracked = normalized.trackEntityId
+            ? fused.find((candidate) => candidate.id === normalized.trackEntityId)
+            : undefined;
+        if (tracked && !visible.some((candidate) => candidate.id === tracked.id)) {
+            visible.push(tracked);
+        }
+        const hits = timings.timeSync("materialize", () => visible.map((candidate) => candidateToHit(candidate, limit, trace)));
+        const trackedHit = normalized.trackEntityId
+            ? hits.find((hit) => hit.entity.id === normalized.trackEntityId)
+            : undefined;
+        return {
+            plan: normalized,
+            hits,
+            trackedHit,
+        };
+    });
+    return {
+        ...result,
+        timings: timings.entries(),
+    };
+}
+export async function diagnoseEntitySearch(query, entityId, ctx) {
+    const stored = ctx.storage.getEntity(entityId);
+    if (!stored) {
+        throw new Error(`Entity not found: ${entityId}`);
+    }
+    const search = await searchWorkspaceIndex({
+        routes: [
+            { mode: "fts", query },
+            { mode: "vector", query },
+        ],
+        trace: true,
+        trackEntityId: entityId,
+    }, ctx);
+    return {
+        query,
+        entityId,
+        file: stored.file,
+        entity: stored.entity,
+        search,
+    };
+}
+export async function diagnoseFileSearch(query, absolutePath, ctx) {
+    const file = ctx.storage.getFileByPath(absolutePath);
+    if (!file) {
+        return null;
+    }
+    const entityId = await chooseBestEntityInFile(query, file, ctx);
+    if (!entityId) {
+        return null;
+    }
+    return diagnoseEntitySearch(query, entityId, ctx);
+}
+function validateSearchPlan(plan) {
+    if (!Array.isArray(plan.routes) || plan.routes.length === 0) {
+        throw new EngineError("Search plan requires at least one route", {
+            code: "ZVEC_GREP.ENGINE.SEARCH_PLAN.EMPTY_ROUTES",
+        });
+    }
+    const routeIds = new Set();
+    const routeCounts = new Map();
+    const routes = plan.routes.map((route, index) => {
+        const query = typeof route.query === "string" ? route.query.trim() : "";
+        if (route.mode !== "fts" && route.mode !== "vector") {
+            throw new EngineError("Search plan route has an unsupported mode", {
+                code: "ZVEC_GREP.ENGINE.SEARCH_PLAN.UNSUPPORTED_ROUTE_MODE",
+                context: `routeIndex=${index} mode=${String(route.mode)}`,
+            });
+        }
+        const id = makeDefaultRouteId(route.mode, routeCounts, routeIds);
+        if (query.length === 0) {
+            throw new EngineError("Search plan route requires a non-empty query", {
+                code: "ZVEC_GREP.ENGINE.SEARCH_PLAN.EMPTY_ROUTE_QUERY",
+                context: `routeId=${id}`,
+            });
+        }
+        routeIds.add(id);
+        return {
+            id,
+            mode: route.mode,
+            query,
+        };
+    });
+    const modifiedAfter = normalizeModifiedTime(plan.modifiedAfter, "modifiedAfter");
+    const modifiedBefore = normalizeModifiedTime(plan.modifiedBefore, "modifiedBefore");
+    if (modifiedAfter !== undefined &&
+        modifiedBefore !== undefined &&
+        modifiedAfter > modifiedBefore) {
+        throw new EngineError("Search plan modified-after filter must not be later than modified-before", {
+            code: "ZVEC_GREP.ENGINE.SEARCH_PLAN.INVALID_MODIFIED_TIME_RANGE",
+            context: `modifiedAfter=${modifiedAfter} modifiedBefore=${modifiedBefore}`,
+        });
+    }
+    return {
+        ...plan,
+        routes,
+        includePaths: normalizePathFilters(plan.includePaths, "includePaths"),
+        excludePaths: normalizePathFilters(plan.excludePaths, "excludePaths"),
+        globs: normalizeStringFilters(plan.globs, "globs"),
+        insensitiveGlobs: normalizeStringFilters(plan.insensitiveGlobs, "insensitiveGlobs"),
+        fileTypes: normalizeStringFilters(plan.fileTypes, "fileTypes"),
+        excludedFileTypes: normalizeStringFilters(plan.excludedFileTypes, "excludedFileTypes"),
+        modifiedAfter,
+        modifiedBefore,
+    };
+}
+function makeDefaultRouteId(mode, routeCounts, usedRouteIds) {
+    let count = routeCounts.get(mode) ?? 0;
+    while (true) {
+        count++;
+        const id = count === 1 ? mode : `${mode}-${count}`;
+        if (!usedRouteIds.has(id)) {
+            routeCounts.set(mode, count);
+            return id;
+        }
+    }
+}
+function planUsesVector(plan) {
+    return plan.routes.some((route) => route.mode === "vector");
+}
+function requireEmbeddingModel(ctx, operation) {
+    if (!ctx.embeddingModel) {
+        throw new EngineError("Search operation requires an embedding model", {
+            code: "ZVEC_GREP.ENGINE.SEARCH.EMBEDDING_MODEL_REQUIRED",
+            context: errorDetails([
+                workspaceIndexDetail(ctx.workspaceIndex.name),
+                detail("operation", operation),
+            ]),
+        });
+    }
+    return ctx.embeddingModel;
+}
+function normalizePathFilters(value, field) {
+    if (value === undefined) {
+        return undefined;
+    }
+    if (!Array.isArray(value)) {
+        throw new EngineError("Search plan path filters must be arrays", {
+            code: "ZVEC_GREP.ENGINE.SEARCH_PLAN.INVALID_PATH_FILTERS",
+            context: `field=${field}`,
+        });
+    }
+    const patterns = [];
+    for (const [index, item] of value.entries()) {
+        if (typeof item !== "string") {
+            throw new EngineError("Search plan path filters must contain strings", {
+                code: "ZVEC_GREP.ENGINE.SEARCH_PLAN.INVALID_PATH_FILTER",
+                context: `field=${field} index=${index}`,
+            });
+        }
+        const pattern = normalizePathFilterPattern(item);
+        if (pattern.length > 0) {
+            patterns.push(pattern);
+        }
+    }
+    return patterns.length > 0 ? patterns : undefined;
+}
+function normalizeStringFilters(value, field) {
+    if (value === undefined) {
+        return undefined;
+    }
+    if (!Array.isArray(value)) {
+        throw new EngineError("Search plan filters must be arrays", {
+            code: "ZVEC_GREP.ENGINE.SEARCH_PLAN.INVALID_FILTERS",
+            context: `field=${field}`,
+        });
+    }
+    const values = value.map((item, index) => {
+        if (typeof item !== "string" || !item.trim()) {
+            throw new EngineError("Search plan filters must contain strings", {
+                code: "ZVEC_GREP.ENGINE.SEARCH_PLAN.INVALID_FILTER",
+                context: `field=${field} index=${index}`,
+            });
+        }
+        return item.trim();
+    });
+    return values.length > 0 ? values : undefined;
+}
+function normalizeModifiedTime(value, field) {
+    if (value === undefined) {
+        return undefined;
+    }
+    if (!Number.isFinite(value) || value < 0) {
+        throw new EngineError("Search plan modified time filters must be non-negative epoch milliseconds", {
+            code: "ZVEC_GREP.ENGINE.SEARCH_PLAN.INVALID_MODIFIED_TIME_FILTER",
+            context: `field=${field} value=${String(value)}`,
+        });
+    }
+    return value;
+}
+async function embedVectorRoutes(routes, model) {
+    const vectorRoutes = routes.filter((route) => route.mode === "vector");
+    const vectorsByRoute = new Map();
+    for (let start = 0; start < vectorRoutes.length; start += model.info.limits.maxBatchSize) {
+        const batch = vectorRoutes.slice(start, start + model.info.limits.maxBatchSize);
+        const { vectors } = await model.embed(batch.map((route) => ({
+            kind: "text",
+            text: route.query,
+        })), { purpose: "query" });
+        for (const [index, route] of batch.entries()) {
+            vectorsByRoute.set(route.id, vectors[index]);
+        }
+    }
+    return vectorsByRoute;
+}
+function addRecallHits(candidates, hits, route, storage, options = {}) {
+    const startIndex = Math.max(0, options.startIndex ?? 0);
+    for (let index = startIndex; index < hits.length; index++) {
+        const hit = hits[index];
+        if (!hit) {
+            continue;
+        }
+        const rank = index + 1;
+        const entityId = publicEntityId(hit.fragment);
+        const resolved = candidates.get(entityId) ?? resolveHitEntity(hit, storage);
+        if (!resolved) {
+            continue;
+        }
+        const candidate = candidates.get(entityId) ?? {
+            id: entityId,
+            entity: resolved.entity,
+            file: resolved.file,
+            sources: new Set(),
+            recall: [],
+            evidence: [],
+            score: 0,
+            rank: Number.POSITIVE_INFINITY,
+            forced: false,
+        };
+        candidate.sources.add(route.mode);
+        candidate.evidence.push({
+            fragment: hit.fragment,
+            path: route.mode,
+            routeId: route.id,
+            query: route.query,
+            rank,
+            score: hit.score,
+        });
+        addOrUpdateRecall(candidate, {
+            path: route.mode,
+            routeId: route.id,
+            query: route.query,
+            found: true,
+            rank,
+            score: hit.score,
+        });
+        candidates.set(candidate.id, candidate);
+    }
+}
+function collectAdaptiveRecall(input) {
+    const routes = buildRecallRoutes(input.routes, input.filter, input.preferSymbol);
+    const targetCandidates = recallTargetCandidateCount(input.limit);
+    let previousDepth = 0;
+    let depth = RECALL_INITIAL_DEPTH;
+    while (true) {
+        const saturated = collectRecallPass({
+            routes,
+            vectorByRoute: input.vectorByRoute,
+            depth,
+            previousDepth,
+            storage: input.storage,
+            candidates: input.candidates,
+        });
+        if (input.candidates.size >= targetCandidates ||
+            !saturated ||
+            depth >= RECALL_MAX_DEPTH) {
+            return depth;
+        }
+        previousDepth = depth;
+        depth = Math.min(depth * RECALL_GROWTH_FACTOR, RECALL_MAX_DEPTH);
+    }
+}
+function buildRecallRoutes(routes, filter, preferSymbol) {
+    const output = routes.map((route) => ({
+        ...route,
+        filter,
+        vectorRouteId: route.mode === "vector" ? route.id : undefined,
+    }));
+    if (!preferSymbol) {
+        return output;
+    }
+    const seen = new Set();
+    for (const route of routes) {
+        const symbolNames = extractSymbolNames(route.query);
+        if (symbolNames.length === 0) {
+            continue;
+        }
+        const key = `${route.id}\0${symbolNames.join("\0")}`;
+        if (seen.has(key)) {
+            continue;
+        }
+        seen.add(key);
+        output.push({
+            id: `${route.id}.prefer-symbol`,
+            mode: "fts",
+            query: route.query,
+            filter: {
+                ...filter,
+                symbolNames,
+            },
+        });
+    }
+    return output;
+}
+function collectRecallPass(input) {
+    let saturated = false;
+    for (const route of input.routes) {
+        const hits = recallRouteHits(route, input);
+        saturated = saturated || hits.length >= input.depth;
+        addRecallHits(input.candidates, hits, route, input.storage, {
+            startIndex: input.previousDepth,
+        });
+    }
+    return saturated;
+}
+function recallRouteHits(route, input) {
+    if (route.mode === "fts") {
+        return input.storage.searchFts(route.query, input.depth, route.filter);
+    }
+    const vector = input.vectorByRoute.get(route.vectorRouteId ?? route.id);
+    return vector
+        ? input.storage.searchVector(vector, input.depth, route.filter)
+        : [];
+}
+function recallTargetCandidateCount(limit) {
+    return Math.max(limit * RECALL_TARGET_FACTOR, RECALL_MIN_TARGET_CANDIDATES);
+}
+function resolveHitEntity(hit, storage) {
+    if (!hit.fragment.group || hit.fragment.group === hit.fragment.id) {
+        return {
+            entity: fragmentToEntity(hit.fragment),
+            file: hit.file,
+        };
+    }
+    return storage.getEntity(hit.fragment.group);
+}
+function fragmentToEntity(fragment) {
+    return {
+        id: publicEntityId(fragment),
+        fileId: fragment.fileId,
+        range: fragment.range,
+        content: fragment.content,
+        metadata: fragment.metadata,
+    };
+}
+function publicEntityId(fragment) {
+    return fragment.group ?? fragment.id;
+}
+function addOrUpdateRecall(candidate, recall) {
+    const existing = candidate.recall.find((item) => item.path === recall.path && item.routeId === recall.routeId);
+    if (!existing) {
+        candidate.recall.push(recall);
+        return;
+    }
+    if (!recall.found) {
+        return;
+    }
+    if (!existing.found ||
+        existing.rank === undefined ||
+        (recall.rank !== undefined && recall.rank < existing.rank)) {
+        Object.assign(existing, recall, {
+            forced: existing.forced || recall.forced || undefined,
+        });
+        return;
+    }
+    if (recall.forced) {
+        existing.forced = true;
+    }
+}
+function extractSymbolNames(query) {
+    const keywords = new Set([
+        "class",
+        "struct",
+        "enum",
+        "interface",
+        "function",
+        "method",
+        "type",
+        "const",
+        "let",
+        "var",
+        "namespace",
+        "where",
+        "find",
+        "explain",
+    ]);
+    const names = new Set();
+    for (const match of query.matchAll(/[A-Za-z_~][A-Za-z0-9_:~]*/g)) {
+        const token = match[0];
+        if (keywords.has(token.toLowerCase())) {
+            continue;
+        }
+        const name = symbolNameFromToken(token);
+        if (name) {
+            names.add(name);
+        }
+    }
+    return [...names];
+}
+function symbolNameFromToken(token) {
+    const parts = token.split("::").filter((part) => part.length > 0);
+    const name = parts[parts.length - 1] ?? token;
+    if (!/^[A-Za-z_~][A-Za-z0-9_~]*$/.test(name)) {
+        return null;
+    }
+    if (parts.length < 2) {
+        return name;
+    }
+    const owner = parts[parts.length - 2] ?? "";
+    if (/^[A-Z_~]/.test(owner)) {
+        return `${owner}::${name}`;
+    }
+    return name;
+}
+function forceTrackEntity(input) {
+    const tracked = input.storage.getEntity(input.entityId);
+    if (!tracked) {
+        return;
+    }
+    const candidate = input.candidates.get(input.entityId) ?? {
+        id: input.entityId,
+        entity: tracked.entity,
+        file: tracked.file,
+        sources: new Set(),
+        recall: [],
+        evidence: [],
+        score: 0,
+        rank: Number.POSITIVE_INFINITY,
+        forced: true,
+    };
+    const seenRoutes = new Set(candidate.recall.map((trace) => trace.routeId));
+    for (const route of input.routes) {
+        if (seenRoutes.has(route.id)) {
+            continue;
+        }
+        if (filterMatchesNoFiles(input.filter)) {
+            forceTrackNoMatchingFilesRoute(candidate, route);
+            continue;
+        }
+        if (filterExcludesFile(input.filter, tracked.file.id)) {
+            forceTrackPathExcludedRoute(candidate, route);
+            continue;
+        }
+        if (route.mode === "fts") {
+            forceTrackFtsRoute(candidate, {
+                ...input,
+                targetFileId: tracked.file.id,
+            }, route);
+        }
+        else {
+            forceTrackVectorRoute(candidate, {
+                ...input,
+                targetFileId: tracked.file.id,
+            }, route);
+        }
+    }
+    input.candidates.set(input.entityId, candidate);
+}
+function filterExcludesFile(filter, fileId) {
+    return filter?.fileIds !== undefined && !filter.fileIds.includes(fileId);
+}
+function forceTrackNoMatchingFilesRoute(candidate, route) {
+    candidate.recall.push({
+        path: route.mode,
+        routeId: route.id,
+        query: route.query,
+        found: false,
+        forced: true,
+        reason: "No files matched the path filters",
+    });
+}
+function forceTrackPathExcludedRoute(candidate, route) {
+    candidate.recall.push({
+        path: route.mode,
+        routeId: route.id,
+        query: route.query,
+        found: false,
+        forced: true,
+        reason: "Target entity file was excluded by the path filters",
+    });
+}
+function forceTrackFtsRoute(candidate, input, route) {
+    const hit = searchTrackedEntityFts(input, route);
+    if (hit) {
+        candidate.sources.add("fts");
+        candidate.evidence.push({
+            fragment: hit.fragment,
+            path: "fts",
+            routeId: route.id,
+            query: route.query,
+            rank: input.recallDepth + 1,
+            score: hit.score,
+            forced: true,
+        });
+        addOrUpdateRecall(candidate, {
+            path: "fts",
+            routeId: route.id,
+            query: route.query,
+            found: true,
+            forced: true,
+            rank: input.recallDepth + 1,
+            score: hit.score,
+        });
+        return;
+    }
+    candidate.recall.push({
+        path: "fts",
+        routeId: route.id,
+        query: route.query,
+        found: false,
+        forced: true,
+        reason: "Target entity did not match the FTS query",
+    });
+}
+function forceTrackVectorRoute(candidate, input, route) {
+    const vector = input.vectorByRoute.get(route.id);
+    if (!vector) {
+        candidate.recall.push({
+            path: "vector",
+            routeId: route.id,
+            query: route.query,
+            found: false,
+            forced: true,
+            reason: "Vector route was not available for this query",
+        });
+        return;
+    }
+    const hit = searchTrackedEntityVector(input, vector);
+    if (hit) {
+        candidate.sources.add("vector");
+        candidate.evidence.push({
+            fragment: hit.fragment,
+            path: "vector",
+            routeId: route.id,
+            query: route.query,
+            rank: input.recallDepth + 1,
+            score: hit.score,
+            forced: true,
+        });
+        addOrUpdateRecall(candidate, {
+            path: "vector",
+            routeId: route.id,
+            query: route.query,
+            found: true,
+            forced: true,
+            rank: input.recallDepth + 1,
+            score: hit.score,
+        });
+        return;
+    }
+    candidate.recall.push({
+        path: "vector",
+        routeId: route.id,
+        query: route.query,
+        found: false,
+        forced: true,
+        reason: "Target entity could not be scored by vector search",
+    });
+}
+function searchTrackedEntityFts(input, route) {
+    const [groupHit] = input.storage.searchFts(route.query, 1, {
+        ...input.filter,
+        groupIds: [input.entityId],
+    });
+    if (groupHit) {
+        return groupHit;
+    }
+    return input.storage
+        .searchFts(route.query, input.recallDepth, restrictFilterToFile(input.filter, input.targetFileId))
+        .find((hit) => publicEntityId(hit.fragment) === input.entityId);
+}
+function searchTrackedEntityVector(input, vector) {
+    const [groupHit] = input.storage.searchVector(vector, 1, {
+        ...input.filter,
+        groupIds: [input.entityId],
+    });
+    if (groupHit) {
+        return groupHit;
+    }
+    return input.storage
+        .searchVector(vector, input.recallDepth, restrictFilterToFile(input.filter, input.targetFileId))
+        .find((hit) => publicEntityId(hit.fragment) === input.entityId);
+}
+function restrictFilterToFile(filter, fileId) {
+    return {
+        ...filter,
+        fileIds: [fileId],
+    };
+}
+function searchPlanToStorageFilter(plan, storage, fileTypePatterns) {
+    const fileIds = resolveFilteredFileIds(plan, storage.listFiles(), fileTypePatterns);
+    const symbolTypes = plan.symbolTypes && plan.symbolTypes.length > 0
+        ? plan.symbolTypes
+        : undefined;
+    if (fileIds === undefined && symbolTypes === undefined) {
+        return undefined;
+    }
+    return {
+        ...(fileIds !== undefined ? { fileIds } : {}),
+        ...(symbolTypes !== undefined ? { symbolTypes } : {}),
+    };
+}
+function filterMatchesNoFiles(filter) {
+    return filter?.fileIds !== undefined && filter.fileIds.length === 0;
+}
+function resolveFilteredFileIds(plan, files, fileTypePatterns) {
+    const includeMatchers = (plan.includePaths ?? []).map(compilePathFilter);
+    const excludeMatchers = (plan.excludePaths ?? []).map(compilePathFilter);
+    const hasModifiedFilter = plan.modifiedAfter !== undefined || plan.modifiedBefore !== undefined;
+    const hasSharedSelection = (plan.globs?.length ?? 0) > 0 ||
+        (plan.insensitiveGlobs?.length ?? 0) > 0 ||
+        fileTypePatterns.include.length > 0 ||
+        fileTypePatterns.exclude.length > 0;
+    if (includeMatchers.length === 0 &&
+        excludeMatchers.length === 0 &&
+        !hasModifiedFilter &&
+        !hasSharedSelection) {
+        return undefined;
+    }
+    return files
+        .filter((file) => {
+        const included = includeMatchers.length === 0 ||
+            includeMatchers.some((matcher) => matcher(file));
+        const excluded = excludeMatchers.some((matcher) => matcher(file));
+        return (included &&
+            !excluded &&
+            matchesFileSelection(file.relativePath, plan, fileTypePatterns) &&
+            matchesModifiedTimeFilter(file, plan));
+    })
+        .map((file) => file.id);
+}
+function matchesModifiedTimeFilter(file, plan) {
+    if (plan.modifiedAfter !== undefined &&
+        file.lastModifiedTime < plan.modifiedAfter) {
+        return false;
+    }
+    if (plan.modifiedBefore !== undefined &&
+        file.lastModifiedTime > plan.modifiedBefore) {
+        return false;
+    }
+    return true;
+}
+function compilePathFilter(pattern) {
+    const pathTarget = isAbsolutePathPattern(pattern)
+        ? "absolutePath"
+        : "relativePath";
+    if (hasPathGlob(pattern)) {
+        return (file) => pathPatternMatches(pattern, normalizePathForMatch(file[pathTarget]));
+    }
+    return (file) => {
+        const path = normalizePathForMatch(file[pathTarget]);
+        return pathPatternMatches(pattern, path);
+    };
+}
+function normalizePathFilterPattern(pattern) {
+    return normalizePathPattern(pattern);
+}
+function fuseCandidates(candidates) {
+    for (const candidate of candidates.values()) {
+        candidate.score = 0;
+        candidate.forced = candidate.recall.some((trace) => trace.forced);
+        for (const recall of candidate.recall) {
+            if (recall.found && recall.rank !== undefined) {
+                candidate.score += 1 / (RRF_K + recall.rank);
+            }
+        }
+    }
+    const fused = [...candidates.values()].sort((left, right) => {
+        if (right.score !== left.score) {
+            return right.score - left.score;
+        }
+        return left.id.localeCompare(right.id);
+    });
+    for (const [index, candidate] of fused.entries()) {
+        candidate.rank = index + 1;
+    }
+    return fused;
+}
+function candidateToHit(candidate, limit, trace) {
+    return {
+        entity: candidate.entity,
+        file: candidate.file,
+        evidence: sortEvidence(candidate.evidence).map(evidenceToSearchHitEvidence),
+        rank: candidate.rank,
+        score: candidate.score,
+        matchedBy: deriveMatchedBy(candidate.sources),
+        trace: trace ? candidateToTrace(candidate, limit) : undefined,
+    };
+}
+function evidenceToSearchHitEvidence(evidence) {
+    return {
+        range: evidence.fragment.range,
+        content: evidence.fragment.content,
+        metadata: evidence.fragment.metadata,
+        isEntity: evidence.fragment.id === publicEntityId(evidence.fragment),
+        path: evidence.path,
+        routeId: evidence.routeId,
+        query: evidence.query,
+        rank: evidence.rank,
+        score: evidence.score,
+        forced: evidence.forced,
+    };
+}
+function sortEvidence(evidence) {
+    return [...evidence].sort((left, right) => {
+        const leftRank = left.rank ?? Number.POSITIVE_INFINITY;
+        const rightRank = right.rank ?? Number.POSITIVE_INFINITY;
+        if (leftRank !== rightRank) {
+            return leftRank - rightRank;
+        }
+        if (left.path !== right.path) {
+            return left.path.localeCompare(right.path);
+        }
+        return left.fragment.id.localeCompare(right.fragment.id);
+    });
+}
+function candidateToTrace(candidate, limit) {
+    const final = {
+        returnedByLimit: candidate.rank <= limit,
+        cutoffRank: limit,
+    };
+    return {
+        recall: candidate.recall,
+        fusion: {
+            rank: candidate.rank,
+            score: candidate.score,
+            forced: candidate.forced || undefined,
+        },
+        final,
+    };
+}
+async function chooseBestEntityInFile(query, file, ctx) {
+    const candidates = new Map();
+    addRecallHits(candidates, ctx.storage.searchFts(query, 10, {
+        fileIds: [file.id],
+    }), {
+        id: "fts",
+        mode: "fts",
+        query,
+    }, ctx.storage);
+    const { vectors: [queryVector], } = await requireEmbeddingModel(ctx, "diagnose").embed([{ kind: "text", text: query }], { purpose: "query" });
+    addRecallHits(candidates, ctx.storage.searchVector(queryVector, 10, {
+        fileIds: [file.id],
+    }), {
+        id: "vector",
+        mode: "vector",
+        query,
+    }, ctx.storage);
+    const [best] = fuseCandidates(candidates);
+    if (best) {
+        return best.id;
+    }
+    const [first] = ctx.storage.listEntitiesByFile(file.id, { limit: 1 });
+    return first?.entity.id ?? null;
+}
+function deriveMatchedBy(sources) {
+    if (sources.has("fts") && sources.has("vector")) {
+        return "fts+vector";
+    }
+    return sources.has("vector") ? "vector" : "fts";
+}
+//# sourceMappingURL=index.js.map