@lorrylurui/code-intelligence-mcp 1.1.14 → 1.2.0

package/dist/indexer/babelParser.js

@@ -3,7 +3,9 @@
  */
 import * as babelParser from '@babel/parser';
 import * as bt from '@babel/types';
-import { getRelativePathForDisplay, inferCategoryFromPath } from './heuristics.js';
+import { getRelativePathForDisplay, inferCategoryFromPath, inferCategoryFromName, } from './heuristics.js';
+import { computeFileHash } from './tsAstNormalizer.js';
+import { computeSemanticHashJs } from './jsAstNormalizer.js';
 /** 从 JS 文件内容解析导出的代码块 */
 export function parseJsFile(filePath, content, projectRoot) {
     const out = [];
@@ -46,7 +48,23 @@ export function parseJsFile(filePath, content, projectRoot) {
             out.push(...rows);
         }
     }
-    return out;
+    // 第三轮：更新content、file_hash、semantic_hash 字段
+    const outWithTwoHash = out.map((o) => {
+        const [semantic_hash, stableStr] = computeSemanticHashJs(o);
+        const { name, type, description, meta, category, path } = o;
+        return {
+            name,
+            type,
+            category,
+            path,
+            description,
+            meta,
+            content: stableStr,
+            file_hash: computeFileHash(content),
+            semantic_hash,
+        };
+    });
+    return outWithTwoHash;
 }
 /** 处理导出的声明 */
 function processStatement(stmt, filePath, isJsx, projectRoot) {
@@ -70,10 +88,14 @@ function processStatement(stmt, filePath, isJsx, projectRoot) {
         }
         else if (bt.isVariableDeclaration(decl)) {
             for (const declarator of decl.declarations) {
-                if (bt.isVariableDeclarator(declarator) && declarator.id && bt.isIdentifier(declarator.id)) {
+                if (bt.isVariableDeclarator(declarator) &&
+                    declarator.id &&
+                    bt.isIdentifier(declarator.id)) {
                     const name = declarator.id.name;
                     const init = declarator.init;
-                    if (name && (bt.isArrowFunctionExpression(init) || bt.isFunctionExpression(init))) {
+                    if (name &&
+                        (bt.isArrowFunctionExpression(init) ||
+                            bt.isFunctionExpression(init))) {
                         const fnDecl = arrowToFunction(name, init);
                         out.push(createRowFromFunction(name, fnDecl, filePath, projectRoot, isJsx));
                     }
@@ -84,19 +106,26 @@ function processStatement(stmt, filePath, isJsx, projectRoot) {
     // 处理 module.exports = xxx
     else if (bt.isExpressionStatement(stmt)) {
         const expr = stmt.expression;
-        if (bt.isAssignmentExpression(expr) && bt.isMemberExpression(expr.left)) {
+        if (bt.isAssignmentExpression(expr) &&
+            bt.isMemberExpression(expr.left)) {
             const left = expr.left;
             // module.exports = xxx
-            if (bt.isIdentifier(left.object) && left.object.name === 'module' &&
-                bt.isIdentifier(left.property) && left.property.name === 'exports') {
+            if (bt.isIdentifier(left.object) &&
+                left.object.name === 'module' &&
+                bt.isIdentifier(left.property) &&
+                left.property.name === 'exports') {
                 const right = expr.right;
                 if (bt.isObjectExpression(right)) {
                     for (const prop of right.properties) {
-                        if (bt.isObjectProperty(prop) && bt.isIdentifier(prop.key)) {
+                        if (bt.isObjectProperty(prop) &&
+                            bt.isIdentifier(prop.key)) {
                             const name = prop.key.name;
                             const value = prop.value;
-                            if (bt.isFunctionExpression(value) || bt.isArrowFunctionExpression(value)) {
-                                const fnDecl = arrowToFunction(name, bt.isArrowFunctionExpression(value) ? value : value);
+                            if (bt.isFunctionExpression(value) ||
+                                bt.isArrowFunctionExpression(value)) {
+                                const fnDecl = arrowToFunction(name, bt.isArrowFunctionExpression(value)
+                                    ? value
+                                    : value);
                                 out.push(createRowFromFunction(name, fnDecl, filePath, projectRoot, isJsx));
                             }
                         }
@@ -112,11 +141,15 @@ function processStatement(stmt, filePath, isJsx, projectRoot) {
                 }
             }
             // exports.xxx = xxx
-            else if (bt.isIdentifier(left.object) && left.object.name === 'exports') {
-                const name = bt.isIdentifier(left.property) ? left.property.name : null;
+            else if (bt.isIdentifier(left.object) &&
+                left.object.name === 'exports') {
+                const name = bt.isIdentifier(left.property)
+                    ? left.property.name
+                    : null;
                 if (name) {
                     const right = expr.right;
-                    if (bt.isFunctionExpression(right) || bt.isArrowFunctionExpression(right)) {
+                    if (bt.isFunctionExpression(right) ||
+                        bt.isArrowFunctionExpression(right)) {
                         const fnDecl = arrowToFunction(name, bt.isArrowFunctionExpression(right) ? right : right);
                         out.push(createRowFromFunction(name, fnDecl, filePath, projectRoot, isJsx));
                     }
@@ -166,10 +199,14 @@ function scanAllDeclarations(stmt, filePath, isJsx, projectRoot) {
     // 变量声明: const foo = () => {}, const bar = function() {}
     else if (bt.isVariableDeclaration(stmt)) {
         for (const declarator of stmt.declarations) {
-            if (bt.isVariableDeclarator(declarator) && declarator.id && bt.isIdentifier(declarator.id)) {
+            if (bt.isVariableDeclarator(declarator) &&
+                declarator.id &&
+                bt.isIdentifier(declarator.id)) {
                 const name = declarator.id.name;
                 const init = declarator.init;
-                if (name && (bt.isArrowFunctionExpression(init) || bt.isFunctionExpression(init))) {
+                if (name &&
+                    (bt.isArrowFunctionExpression(init) ||
+                        bt.isFunctionExpression(init))) {
                     const fnDecl = arrowToFunction(name, init);
                     out.push(createRowFromFunction(name, fnDecl, filePath, projectRoot, isJsx));
                 }
@@ -197,24 +234,24 @@ function arrowToFunction(name, arrow) {
 }
 function createRowFromFunction(name, decl, filePath, projectRoot, isJsx) {
     const relPath = getRelativePathForDisplay(projectRoot, filePath);
-    const category = inferCategoryFromPath(filePath);
+    const category = inferCategoryFromPath(filePath) || inferCategoryFromName(name);
     // 检测是否有 JSX
     const hasJsx = isJsx || containsJsx(decl);
     // 判断类型：
     // 1. 有 JSX = component
-    // 2. 名字包含 selector = selector
+    // 2. 名字包含 use = hook
     // 3. 大写开头（JSX 组件约定）= component
-    // 4. 其他 = util
-    const type = hasJsx
-        ? 'component'
-        : name.toLowerCase().includes('selector')
-            ? 'selector'
-            : isJsx && /^[A-Z]/.test(name)
-                ? 'component'
-                : 'util';
+    // 4. 其他 = function
+    const type = name.toLowerCase().includes('use')
+        ? 'hook'
+        : isJsx || /^[A-Z]/.test(name)
+            ? 'component'
+            : 'function';
     const params = decl.params
         .filter((p) => bt.isIdentifier(p))
         .map((p) => p.name);
+    const hooks = extractHooksFromBody(decl);
+    const sideEffects = extractSideEffects(decl);
     return {
         name,
         type,
@@ -223,24 +260,34 @@ function createRowFromFunction(name, decl, filePath, projectRoot, isJsx) {
         description: null,
         content: `function ${decl.id?.name || 'anonymous'}(${params.join(', ')}) { ... }`,
         meta: {
+            kind: 'function',
             params,
             returnType: getReturnType(decl),
+            ...(hooks.length ? { hooks } : {}),
+            ...(sideEffects.length ? { sideEffects } : {}),
         },
+        file_hash: '',
+        semantic_hash: '',
+        node: decl,
     };
 }
 function createRowFromClass(name, _decl, filePath, projectRoot) {
     const relPath = getRelativePathForDisplay(projectRoot, filePath);
     const category = inferCategoryFromPath(filePath);
     // 大写开头的类视为组件
-    const type = /^[A-Z]/.test(name) ? 'component' : 'util';
+    const type = /^[A-Z]/.test(name) ? 'component' : 'function';
     return {
         name,
         type,
         category,
         path: relPath,
         description: null,
+        // content meta.kind 暂时废弃不用，
         content: null,
-        meta: { kind: 'class' },
+        meta: {},
+        file_hash: '',
+        semantic_hash: '',
+        node: _decl,
     };
 }
 /** 简单检测是否包含 JSX */
@@ -253,7 +300,17 @@ function containsJsx(node) {
             return;
         }
         // 只遍历常见的包含子节点的属性
-        const keys = ['body', 'declarations', 'arguments', 'callee', 'init', 'left', 'right', 'consequent', 'alternate'];
+        const keys = [
+            'body',
+            'declarations',
+            'arguments',
+            'callee',
+            'init',
+            'left',
+            'right',
+            'consequent',
+            'alternate',
+        ];
         for (const key of keys) {
             const val = n[key];
             if (Array.isArray(val)) {
@@ -294,3 +351,175 @@ function getReturnType(fn) {
     }
     return undefined;
 }
+/**
+ * 遍历 Babel AST 节点，收集所有满足条件的回调
+ */
+function visitNodes(node, callback) {
+    callback(node);
+    for (const key of Object.keys(node)) {
+        const val = node[key];
+        if (Array.isArray(val)) {
+            for (const v of val) {
+                if (v && typeof v === 'object' && 'type' in v) {
+                    visitNodes(v, callback);
+                }
+            }
+        }
+        else if (val && typeof val === 'object' && 'type' in val) {
+            visitNodes(val, callback);
+        }
+    }
+}
+/**
+ * 从函数体中提取 React Hooks（use 开头的函数调用）
+ */
+function extractHooksFromBody(fn) {
+    const seen = new Set();
+    const body = fn.body;
+    if (!body || !bt.isBlockStatement(body))
+        return [];
+    visitNodes(body, (n) => {
+        if (bt.isCallExpression(n)) {
+            const callee = n.callee;
+            if (bt.isIdentifier(callee) && callee.name.startsWith('use')) {
+                seen.add(callee.name);
+            }
+        }
+    });
+    return [...seen].sort();
+}
+/**
+ * 获取节点的文本表示（通过 AST 节点属性构建）
+ */
+function getNodeText(n) {
+    const node = n;
+    if (!node || typeof node !== 'object')
+        return '';
+    const type = node.type;
+    if (type === 'MemberExpression' || type === 'OptionalMemberExpression') {
+        const obj = getNodeText(node.object);
+        const propNode = node.property;
+        let prop = '';
+        if (propNode && typeof propNode === 'object') {
+            const propType = propNode.type;
+            if (propType === 'Identifier') {
+                prop =
+                    propNode.name ||
+                        '';
+            }
+            else if (propType === 'Literal') {
+                prop = String(propNode.value ?? '');
+            }
+        }
+        const computed = node.computed;
+        return obj + (computed ? `[${prop}]` : `.${prop}`);
+    }
+    if (type === 'Identifier') {
+        return node.name || '';
+    }
+    if (type === 'Literal' || type === 'NullLiteral') {
+        const val = node.value;
+        return val === null ? 'null' : String(val);
+    }
+    if (type === 'CallExpression' || type === 'OptionalCallExpression') {
+        const callee = getNodeText(node.callee);
+        return callee + '(...)';
+    }
+    if (type === 'AssignmentExpression') {
+        const left = getNodeText(node.left);
+        return left + ' = ...';
+    }
+    return '';
+}
+/**
+ * 静态分析函数体的副作用
+ */
+function extractSideEffects(fn) {
+    const effects = new Set();
+    const body = fn.body;
+    if (!body || !bt.isBlockStatement(body))
+        return [];
+    const paramNames = new Set(fn.params
+        .filter((p) => bt.isIdentifier(p))
+        .map((p) => p.name));
+    visitNodes(body, (n) => {
+        // 1. 网络请求
+        if (bt.isCallExpression(n)) {
+            const calleeText = n.callee && 'name' in n.callee
+                ? n.callee.name
+                : '';
+            const calleeTextLower = calleeText.toLowerCase();
+            if (calleeTextLower === 'fetch' ||
+                calleeTextLower === 'axios' ||
+                calleeTextLower === 'xhr' ||
+                calleeTextLower === 'ajax' ||
+                calleeText.startsWith('axios.') ||
+                calleeTextLower.includes('request')) {
+                effects.add('network');
+            }
+            if (calleeTextLower.includes('xmlhttprequest')) {
+                effects.add('network');
+            }
+        }
+        // 2. 计时器
+        if (bt.isCallExpression(n)) {
+            const calleeName = n.callee && 'name' in n.callee
+                ? n.callee.name
+                : '';
+            if (calleeName === 'setTimeout' ||
+                calleeName === 'setInterval' ||
+                calleeName === 'requestAnimationFrame' ||
+                calleeName === 'setImmediate') {
+                effects.add('timer');
+            }
+        }
+        // 3. DOM/全局对象操作
+        if (bt.isExpressionStatement(n)) {
+            const text = getNodeText(n.expression);
+            if (/\bdocument\.\w+/.test(text) ||
+                /\bwindow\.\w+/.test(text) ||
+                /\bnavigator\.\w+/.test(text) ||
+                /\blocation\.\w+/.test(text)) {
+                if (/=/.test(text) &&
+                    !text.includes('===') &&
+                    !text.includes('==')) {
+                    effects.add('dom');
+                }
+            }
+        }
+        // 4. 存储操作
+        if (bt.isCallExpression(n)) {
+            const text = getNodeText(n);
+            if (text.includes('localStorage') ||
+                text.includes('sessionStorage') ||
+                text.includes('cookie')) {
+                effects.add('storage');
+            }
+        }
+        // 5. 入参修改
+        if (bt.isAssignmentExpression(n)) {
+            const leftText = getNodeText(n.left);
+            for (const param of paramNames) {
+                if (leftText.startsWith(`${param}.`) ||
+                    leftText.startsWith(`${param}[`)) {
+                    effects.add('mutation');
+                    break;
+                }
+            }
+        }
+        if (bt.isCallExpression(n) && n.callee) {
+            const calleeText = getNodeText(n.callee);
+            for (const param of paramNames) {
+                if (calleeText.startsWith(`${param}.`) ||
+                    calleeText.startsWith(`${param}[`)) {
+                    // 检测 push/pop/splice 等 mutations
+                    if (/\.(push|pop|shift|unshift|splice|sort|reverse|fill)\(/.test(calleeText)) {
+                        effects.add('mutation');
+                        break;
+                    }
+                }
+            }
+        }
+    });
+    return [...effects].sort();
+}

package/dist/indexer/categoryClassifier.js

@@ -0,0 +1,129 @@
+/**
+ * category 分类器：三层融合（规则 + embedding + LLM）
+ */
+import { createHash } from 'node:crypto';
+import { env } from '../config/env.js';
+import { createEmbeddingClient } from '../services/embeddingClient.js';
+import { cosineSimilarity } from '../services/vectorMath.js';
+import { inferCategoryFromPath, inferCategoryFromName } from './heuristics.js';
+// 1. 定义 category label space
+const CATEGORIES = [
+    'network',
+    'form',
+    'validation',
+    'format',
+    'state',
+    'ui',
+    'storage',
+    'math',
+    'hook',
+    'utils',
+    'component',
+    'service',
+];
+// 2. 预计算 category embeddings（缓存）
+let categoryEmbeddingsCache = null;
+export async function initCategoryEmbeddings() {
+    if (!env.embeddingServiceUrl)
+        return;
+    const client = createEmbeddingClient(env.embeddingServiceUrl);
+    const embeddings = await Promise.all(CATEGORIES.map(async (c) => {
+        const [vec] = await client.embed([c]);
+        return { name: c, embedding: vec };
+    }));
+    categoryEmbeddingsCache = embeddings;
+}
+// 3. embedding 层
+const EMBEDDING_THRESHOLD = 0.5;
+// TODO: 这里有问题，embedding是语义模板向量，categoryEmbeddingsCache是单个词的向量，相似度必然是<0.3
+function categoryFromEmbedding(embedding) {
+    if (!categoryEmbeddingsCache)
+        return null;
+    let best = 'utils';
+    let maxScore = -Infinity;
+    for (const c of categoryEmbeddingsCache) {
+        const score = cosineSimilarity(embedding, c.embedding);
+        if (score > maxScore) {
+            maxScore = score;
+            best = c.name;
+        }
+    }
+    return maxScore < EMBEDDING_THRESHOLD ? null : best;
+}
+// 4. LLM 层（带缓存）
+const LLM_CACHE_TTL = 24 * 60 * 60 * 1000;
+const OLLAMA_URL = 'http://127.0.0.1:11434/v1/chat/completions';
+// 可根据本地实际情况调整模型名称，例如：'llama3.2:3b'、'llama3.1:8b'
+const OLLAMA_MODEL = 'llama3.2:3b';
+const llmCategoryCache = new Map();
+async function categoryFromLLM(stableStr) {
+    if (!stableStr)
+        return null;
+    const cacheKey = createHash('sha256')
+        .update(stableStr)
+        .digest('hex')
+        .slice(0, 16);
+    const cached = llmCategoryCache.get(cacheKey);
+    if (cached && Date.now() - cached.timestamp < LLM_CACHE_TTL)
+        return cached.category;
+    const prompt = `给一段代码语义说明，在下面列出来的所有类别中找到最相符的类别，请直接返回命中的原始字符串，没有命中不用返回。
+    代码语义:${stableStr}
+    类别合集: ${CATEGORIES.join(', ')}`;
+    try {
+        const body = {
+            model: OLLAMA_MODEL,
+            messages: [{ role: 'user', content: prompt }],
+            temperature: 0.4,
+            stream: false,
+        };
+        const res = await fetch(OLLAMA_URL, {
+            method: 'POST',
+            headers: { 'Content-Type': 'application/json' },
+            body: JSON.stringify(body),
+        });
+        const data = (await res.json());
+        const content = data.choices?.[0]?.message?.content
+            ?.trim()
+            ?.toLowerCase();
+        const category = CATEGORIES.find((c) => content.includes(c));
+        if (category) {
+            llmCategoryCache.set(cacheKey, { category, timestamp: Date.now() });
+            return category;
+        }
+    }
+    catch (e) {
+        console.error('[categoryFromLLM]', e);
+    }
+    return null;
+}
+// 5. 三层融合
+export async function resolveCategory(rows, vecs) {
+    const pros = rows.map(async (r, i) => {
+        const { name } = r;
+        const ruleCategory = inferCategoryFromPath(r.path) || inferCategoryFromName(name);
+        console.error(`===from ruleCategory`, name, ruleCategory);
+        if (ruleCategory) {
+            return {
+                ...r,
+                category: ruleCategory,
+            };
+        }
+        // TODO: 这里有问题，embedding是语义模板向量，categoryEmbeddingsCache是单个词的向量，相似度必然是<0.3
+        const emd = categoryFromEmbedding(vecs[i]);
+        console.error(`===from categoryFromEmbedding`, name, emd);
+        if (emd) {
+            return {
+                ...r,
+                category: emd,
+            };
+        }
+        const cateLlm = await categoryFromLLM(r.content);
+        console.error(`===from categoryFromLLM`, name, cateLlm);
+        return {
+            ...r,
+            category: cateLlm,
+        };
+    });
+    const newRows = await Promise.all(pros);
+    return newRows;
+}

package/dist/indexer/embedText.js

@@ -1,15 +1,17 @@
+// 仅js类型使用，后续会删掉
+const MAX_CONTENT_LENGTH = 1200;
 function briefMeta(meta) {
-    const keys = ["props", "params", "properties", "hooks"];
+    const keys = ['props', 'params', 'properties', 'hooks'];
     const parts = [];
     for (const k of keys) {
         const v = meta[k];
         if (Array.isArray(v)) {
-            const strs = v.filter((x) => typeof x === "string");
+            const strs = v.filter((x) => typeof x === 'string');
             if (strs.length)
-                parts.push(`${k}: ${strs.slice(0, 24).join(", ")}`);
+                parts.push(`${k}: ${strs.slice(0, 24).join(', ')}`);
         }
     }
-    return parts.join("; ");
+    return parts.join('; ');
 }
 /**
  * 拼成一段供向量模型编码的文本（名称、路径、注释、meta 摘要、源码片段）。
@@ -19,10 +21,10 @@ export function indexedRowToEmbedText(row) {
     return [
         `${row.type} ${row.name}`,
         row.path,
-        row.description ?? "",
+        row.description ?? '',
         metaBit,
-        (row.content ?? "").slice(0, 1200)
+        (row.content ?? '').slice(0, MAX_CONTENT_LENGTH),
     ]
         .filter((s) => s.length > 0)
-        .join("\n");
+        .join('\n');
 }

package/dist/indexer/extractMeta.js

@@ -87,6 +87,7 @@ export function extractFunctionMeta(fn) {
     const returnType = extractReturnTypeText(fn);
     const sideEffects = extractSideEffects(fn);
     return {
+        kind: 'function',
         params,
         ...(paramTypeFields.length ? { paramTypeFields } : {}),
         ...(hooks.length ? { hooks } : {}),
@@ -116,6 +117,7 @@ export function extractInterfaceOrTypeMeta(node) {
     if (Node.isTypeAliasDeclaration(node)) {
         return { kind: 'typeAlias' };
     }
+    // 其他类型（如 enum）暂不处理，标记为 unknown 以供后续扩展。
     return { kind: 'unknown' };
 }
 /**
@@ -166,7 +168,9 @@ export function extractSideEffects(node) {
                 /\bnavigator\.\w+/.test(text) ||
                 /\blocation\.\w+/.test(text)) {
                 // 区分读取和写入
-                if (/=/.test(text) && !text.includes('===') && !text.includes('==')) {
+                if (/=/.test(text) &&
+                    !text.includes('===') &&
+                    !text.includes('==')) {
                     effects.add('dom');
                 }
             }
@@ -199,7 +203,8 @@ export function extractSideEffects(node) {
                 const text = n.getText();
                 // 检测 param.x = ... 或 param.push/pop/splice 等
                 for (const param of paramNames) {
-                    if (text.includes(`${param}.`) || text.startsWith(`${param} =`)) {
+                    if (text.includes(`${param}.`) ||
+                        text.startsWith(`${param} =`)) {
                         effects.add('mutation');
                         break;
                     }