dsh-search-enhance 0.1.3 → 0.1.4

package/lib/documentation/service.d.ts

@@ -2,7 +2,7 @@ import { Service, type Context } from '@deepseek-ai/cordis';
 import type { Config } from '../config.js';
 import type { CanonicalSource, SourceRecordCandidate } from '../contracts/index.js';
 import { type ProviderAttemptRecord, type ProviderErrorKind } from '../provider-runtime/index.js';
-import { Context7RemoteClient, type Context7Library } from '../providers/context7.js';
+import { type Context7RemoteClient, type Context7Library } from '../providers/context7.js';
 import type { BoundedSourceProvider, DocumentationSnippet, SourceProviderSearchInput, SourceProviderSearchOutcome } from '../providers/types.js';
 import type { Context7DocsCacheEntry, Context7DocRef } from './cache-domain.js';
 import { type Context7CachedDocLookup, type Context7CachedDocMatch } from './cache.js';
@@ -125,7 +125,7 @@ export declare class DocumentationSearchInfrastructureError extends Error {
     constructor(attempts: readonly ProviderAttemptRecord[], warnings: readonly DocumentationWarning[], cache: Readonly<DocumentationCacheReport>, cause: unknown);
 }
 export interface DocumentationSearchDependencies {
-    readonly context7: Context7RemoteClient;
+    readonly context7: Pick<Context7RemoteClient, 'resolve' | 'docs'>;
     readonly context7Cache: Context7CachedOperations;
     readonly exa: BoundedSourceProvider;
     readonly getConfig: () => Config;

package/lib/documentation/service.js

@@ -1,10 +1,11 @@
 import { Service } from '@deepseek-ai/cordis';
 import { createProviderAttemptRecord, isAbortError, isProviderError, ProviderError, throwIfAborted, } from '../provider-runtime/index.js';
 import { boundSourceProviderResult } from '../providers/bounded-result.js';
-import { Context7RemoteClient, context7LibrarySource, selectContext7Library, } from '../providers/context7.js';
+import { context7LibrarySource, selectContext7Library, } from '../providers/context7.js';
 import { applySourceQuality } from '../search/index.js';
 import { CONTEXT7_CACHE_QUERY_MAX_SCAN_RECORDS, } from './cache.js';
 import { CONTEXT7_CACHE_STATES, Context7CachedOperations, Context7OperationFailure, isContext7LibraryId, normalizeContext7LibraryId, } from './context7-cache.js';
+const CONTEXT7_RESOLVE_CANDIDATE_LIMIT = 30;
 export const DOCUMENTATION_SEARCH_SERVICE_KEY = 'searchEnhanceDocumentation';
 export const DOCUMENTATION_SEARCH_PROVIDERS = ['auto', 'context7', 'exa', 'all'];
 export const DOCUMENTATION_RESULT_PROVIDERS = ['context7', 'exa'];
@@ -347,6 +348,7 @@ export class DocumentationSearchService extends Service {
         await Promise.allSettled([...this.active]);
     }
     async executeContext7Resolve(input) {
+        const candidateLimit = input.candidateLimit ?? input.maxResults;
         const startedAt = safeClockValue(this.now);
         let dispatches = 0;
         const resolved = await this.context7Cache.resolve({
@@ -354,7 +356,7 @@ export class DocumentationSearchService extends Service {
             forceRefresh: input.forceRefresh,
             load: () => this.context7.resolve({
                 config: input.config,
-                limit: input.maxResults,
+                limit: candidateLimit,
                 onDispatch: () => {
                     dispatches += 1;
                     input.onDispatch?.();
@@ -362,12 +364,12 @@ export class DocumentationSearchService extends Service {
                 query: input.libraryName,
                 signal: input.signal,
             }),
-            maxResults: input.maxResults,
+            maxResults: candidateLimit,
             query: input.libraryName,
             signal: input.signal,
         });
-        const candidates = Object.freeze([...resolved.entry.libraries]);
-        const selected = selectContext7Library(candidates, input.libraryName, input.query);
+        const candidates = Object.freeze(resolved.entry.libraries.slice(0, input.maxResults));
+        const selected = selectContext7Library(resolved.entry.libraries, input.libraryName, input.query);
         return Object.freeze({
             attempts: successfulCacheAttempts('resolve', resolved, dispatches, elapsed(startedAt, this.now)),
             cache: cachePath(resolved.cache, resolved.evictedEntries),
@@ -638,6 +640,7 @@ export class DocumentationSearchService extends Service {
                     forceRefresh: input.forceRefresh,
                     libraryName: input.query,
                     maxResults: input.maxResults,
+                    candidateLimit: Math.min(CONTEXT7_RESOLVE_CANDIDATE_LIMIT, input.config.retention.providerMaxSources),
                     onDispatch: () => { resolveDispatches += 1; },
                     query: input.query,
                     signal: input.signal,

package/lib/providers/context7.js

@@ -121,17 +121,33 @@ export function parseContext7Libraries(body) {
 function normalizedMatchText(value) {
     return (value ?? '').toLowerCase().replace(/^@/, '').replace(/[^a-z0-9]+/g, '');
 }
-function context7LibraryScore(item, libraryName, query) {
+function normalizedTerms(value) {
+    return value.toLowerCase()
+        .split(/[^a-z0-9@._-]+/)
+        .map(normalizedMatchText)
+        .filter(term => term.length > 0);
+}
+function queryMentionsName(queryTerms, value) {
+    const nameTerms = normalizedTerms(value ?? '');
+    if (nameTerms.length === 0)
+        return false;
+    return queryTerms.some((_term, index) => nameTerms.every((nameTerm, offset) => queryTerms[index + offset] === nameTerm));
+}
+function context7LibraryScore(item, libraryName, queryTerms) {
     const wanted = normalizedMatchText(libraryName);
     const rawText = `${item.id ?? ''} ${item.title ?? ''} ${item.description ?? ''}`.toLowerCase();
     const title = normalizedMatchText(item.title);
     const description = normalizedMatchText(item.description);
     const idSegments = (item.id ?? '').split('/').filter(Boolean).map(normalizedMatchText);
     let score = 0;
-    if (/official.*documentation|documentation.*official|official docs|react\.dev/.test(rawText))
+    if (queryMentionsName(queryTerms, item.title))
+        score += 240;
+    for (const segment of (item.id ?? '').split('/').filter(Boolean)) {
+        if (queryMentionsName(queryTerms, segment))
+            score += 220;
+    }
+    if (/official.*documentation|documentation.*official|official docs/.test(rawText))
         score += 260;
-    if (rawText.includes('react.dev'))
-        score += 120;
     if (wanted.length > 0) {
         if (title === wanted)
             score += 240;
@@ -146,9 +162,8 @@ function context7LibraryScore(item, libraryName, query) {
         if (description.includes(wanted))
             score += 8;
     }
-    for (const term of query.toLowerCase().split(/[^a-z0-9@.\-/]+/).filter(term => term.length > 1)) {
-        const normalizedTerm = normalizedMatchText(term);
-        if (normalizedTerm.length === 0)
+    for (const normalizedTerm of queryTerms) {
+        if (normalizedTerm.length <= 1)
             continue;
         if (title === normalizedTerm)
             score += 24;
@@ -174,11 +189,12 @@ function context7LibraryScore(item, libraryName, query) {
 }
 /** Stable selection by exact name, description relevance, trust, benchmark, and snippet coverage. */
 export function selectContext7Library(libraries, libraryName, query = libraryName) {
+    const queryTerms = normalizedTerms(query);
     return libraries
         .map((library, index) => ({
         index,
         library,
-        score: context7LibraryScore(library, libraryName, query),
+        score: context7LibraryScore(library, libraryName, queryTerms),
     }))
         .filter(candidate => isContext7LibraryId(candidate.library.id))
         .sort((left, right) => right.score - left.score || left.index - right.index)[0]?.library;

package/lib/search/parse.js

@@ -29,8 +29,7 @@ export const DEFAULT_SEARCH_RESPONSE_PARSE_LIMITS = Object.freeze({
     maxUrlCharacters: 8192,
 });
 const URL_PATTERN = /https?:\/\/[^\s<>"'`，。、；：！？》）】)]+/g;
-const MARKDOWN_LINK_PATTERN = /\[([^\]]+)]\((https?:\/\/[^)]+)\)/g;
-const MARKDOWN_LINK_LINE_PATTERN = /\[[^\]]+]\(https?:\/\/[^)]+\)/;
+const MARKDOWN_LINK_PATTERN = /(?<!\[)\[([^\[\]]+|\[\d+\])]\((https?:\/\/[^)]+)\)/g;
 const SOURCES_HEADING_PATTERN = /(?:^|\n)(?:#{1,6}\s*)?(?:\*\*|__)?\s*(?:sources?|references?|citations?|信源|参考资料|参考|引用|来源列表|来源)(?:\s*[（(][^)\n]*[)）])?\s*(?:(?:\*\*|__)\s*)?[:：]?\s*(?:(?:\*\*|__)\s*)?$/gim;
 const SOURCES_FUNCTION_PATTERN = /(^|\n)\s*(sources|source|citations|citation|references|reference|citation_card|source_cards|source_card)\s*\(/gim;
 function isRecord(value) {
@@ -337,7 +336,7 @@ function isLinkOnlyLine(line) {
     const stripped = line.replace(/^\s*(?:[-*]|\d+\.)\s*/, '').trim();
     return stripped.length > 0 && (stripped.startsWith('http://')
         || stripped.startsWith('https://')
-        || MARKDOWN_LINK_LINE_PATTERN.test(stripped));
+        || stripped.search(MARKDOWN_LINK_PATTERN) !== -1);
 }
 function splitTailLinkBlock(text, state, limits) {
     const lines = text.split(/\r?\n/);

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "dsh-search-enhance",
-  "version": "0.1.3",
+  "version": "0.1.4",
   "description": "Bounded search, documentation, extraction, progressively disclosed site mapping/research planning, and read-only diagnostics for DeepSeek Harness",
   "type": "module",
   "main": "./lib/index.js",