scai 0.1.110 → 0.1.111

package/dist/commands/AskCmd.js

@@ -1,13 +1,12 @@
 import fs from 'fs';
 import path from 'path';
 import readline from 'readline';
-import { searchFiles, queryFiles, getFunctionsForFiles } from '../db/fileIndex.js';
+import { searchFiles, queryFiles } from '../db/fileIndex.js';
 import { sanitizeQueryForFts } from '../utils/sanitizeQuery.js';
 import { generate } from '../lib/generate.js';
 import { buildContextualPrompt } from '../utils/buildContextualPrompt.js';
-import { generateFocusedFileTree } from '../utils/fileTree.js';
 import { log } from '../utils/log.js';
-import { PROMPT_LOG_PATH, SCAI_HOME, RELATED_FILES_LIMIT, MAX_SUMMARY_LINES, getIndexDir, MAX_FUNCTION_LINES } from '../constants.js';
+import { PROMPT_LOG_PATH, SCAI_HOME, RELATED_FILES_LIMIT, getIndexDir } from '../constants.js';
 import chalk from 'chalk';
 export async function runAskCommand(query) {
     if (!query) {
@@ -19,22 +18,21 @@ export async function runAskCommand(query) {
         return;
     }
     console.log(`📁 Using index root: ${getIndexDir()}`);
-    console.log(`🔍 Searching for: "${query}"\n`);
-    // 🟩 STEP 1: Semantic Search
+    // Semantic Search
     const start = Date.now();
-    const semanticResults = await searchFiles(query, RELATED_FILES_LIMIT); // RankedFile[]
+    const semanticResults = await searchFiles(query, RELATED_FILES_LIMIT);
     const duration = Date.now() - start;
     console.log(`⏱️ searchFiles took ${duration}ms and returned ${semanticResults.length} result(s)`);
     semanticResults.forEach((file, i) => {
         console.log(`  ${i + 1}. 📄 Path: ${file.path} | Score: ${file.score?.toFixed(3) ?? 'n/a'}`);
     });
-    // 🟩 STEP 1.5: Fallback FTS search
+    // Fallback FTS search
     const safeQuery = sanitizeQueryForFts(query);
-    const fallbackResults = queryFiles(safeQuery, 10); // FileRow[]
+    const fallbackResults = queryFiles(safeQuery, 10);
     fallbackResults.forEach((file, i) => {
         console.log(`  ${i + 1}. 🔎 Fallback Match: ${file.path}`);
     });
-    // 🟩 STEP 2: Merge results
+    // Merge results
     const seen = new Set();
     const combinedResults = [];
     for (const file of semanticResults) {
@@ -56,7 +54,20 @@ export async function runAskCommand(query) {
             });
         }
     }
-    // 🟩 STEP 3: Log results
+    // Exact match prioritization
+    const queryFilenameRaw = path.basename(query).toLowerCase();
+    const queryFilenameNoExt = queryFilenameRaw.replace(/\.[^/.]+$/, '');
+    const exactMatchIndex = combinedResults.findIndex(f => {
+        const base = path.basename(f.path).toLowerCase();
+        const baseNoExt = base.replace(/\.[^/.]+$/, '');
+        return base === queryFilenameRaw || baseNoExt === queryFilenameNoExt;
+    });
+    if (exactMatchIndex !== -1) {
+        const [exactMatch] = combinedResults.splice(exactMatchIndex, 1);
+        combinedResults.unshift(exactMatch);
+        console.log(`🎯 Exact match prioritized: ${exactMatch.path}`);
+    }
+    // Log combined results
     if (combinedResults.length) {
         console.log('\n📊 Final Related Files:');
         combinedResults.forEach((f, i) => {
@@ -66,79 +77,38 @@ export async function runAskCommand(query) {
     else {
         console.log('⚠️ No similar files found. Using query only.');
     }
-    // 🟩 STEP 4: Load top file code + metadata
-    const topFile = combinedResults[0];
-    const filepath = topFile?.path || '';
-    let code = '';
-    let topSummary = topFile.summary || '(No summary available)';
-    let topFunctions = [];
-    const fileFunctions = {};
-    // Truncate summary
-    topSummary = topSummary.split('\n').slice(0, MAX_SUMMARY_LINES).join('\n');
-    const allFileIds = combinedResults
-        .map(file => file.id)
-        .filter((id) => typeof id === 'number');
-    const allFunctionsMap = getFunctionsForFiles(allFileIds); // Record<number, Function[]>
-    try {
-        code = fs.readFileSync(filepath, 'utf-8');
-        const topFileId = topFile.id;
-        topFunctions = allFunctionsMap[topFileId]?.map(fn => {
-            const content = fn.content
-                ? fn.content.split('\n').slice(0, MAX_FUNCTION_LINES).join('\n')
-                : '(No content available)';
-            return {
-                name: fn.name,
-                content,
-            };
-        }) || [];
-    }
-    catch (err) {
-        console.warn(`⚠️ Failed to read or analyze top file (${filepath}):`, err);
+    // STEP 4+: Build contextual prompt using topFile + combinedResults
+    if (combinedResults.length === 0) {
+        throw new Error('❌ No search results found. Cannot build contextual prompt.');
     }
-    // 🟩 STEP 5: Build relatedFiles with functions and fileFunctions
-    const relatedFiles = combinedResults.slice(0, RELATED_FILES_LIMIT).map(file => {
-        const fileId = file.id;
-        let summary = file.summary || '(No summary available)';
-        if (summary) {
-            summary = summary.split('\n').slice(0, MAX_SUMMARY_LINES).join('\n');
-        }
-        const functions = allFunctionsMap[fileId]?.map(fn => {
-            const content = fn.content
-                ? fn.content.split('\n').slice(0, MAX_FUNCTION_LINES).join('\n')
-                : '(No content available)';
-            return {
-                name: fn.name,
-                content,
-            };
-        }) || [];
-        return {
-            path: file.path,
-            summary,
-            functions,
-        };
-    });
-    // 🟩 STEP 6: Generate file tree
-    let fileTree = '';
+    const file = combinedResults[0];
+    let code = "";
+    // STEP 4++: Add code to params
     try {
-        fileTree = generateFocusedFileTree(filepath, 2);
+        code = fs.readFileSync(file.path, 'utf-8');
+        if (!code) {
+            console.warn(`⚠️ No code loaded for top file: ${file.path}`);
+        }
     }
     catch (e) {
-        console.warn('⚠️ Could not generate file tree:', e);
+        console.log("Error reading code from selected file: ", e instanceof Error ? e.message : e);
     }
-    // 🟩 STEP 7: Build prompt
-    console.log(chalk.blueBright('\n📦 Building contextual prompt...'));
-    const promptContent = buildContextualPrompt({
-        baseInstruction: query,
+    const topFile = {
+        id: file.id,
+        path: file.path,
+        summary: file.summary ?? "",
         code,
-        summary: topSummary,
-        functions: topFunctions,
-        relatedFiles,
-        projectFileTree: fileTree || undefined,
-        fileFunctions,
-    });
+    };
+    const promptArgs = {
+        topFile,
+        relatedFiles: combinedResults,
+        query,
+    };
+    console.log(chalk.blueBright('\n📦 Building contextual prompt...'));
+    const promptContent = await buildContextualPrompt(promptArgs);
     console.log(chalk.greenBright('✅ Prompt built successfully.'));
     console.log(chalk.cyan(`[runAskCommand] Prompt token estimate: ~${Math.round(promptContent.length / 4)} tokens`));
-    // 🟩 STEP 8: Save prompt
+    // STEP 5: Save prompt
     try {
         if (!fs.existsSync(SCAI_HOME))
             fs.mkdirSync(SCAI_HOME, { recursive: true });
@@ -148,21 +118,21 @@ export async function runAskCommand(query) {
     catch (err) {
         log('❌ Failed to write prompt log:', err);
     }
-    // 🟩 STEP 9: Ask model
+    // STEP 6: Ask model
     try {
         console.log('\n🤖 Asking the model...');
         const input = {
             content: promptContent,
-            filepath,
+            filepath: topFile.path,
         };
-        const modelResponse = await generate(input, 'llama3');
+        const modelResponse = await generate(input);
         console.log(`\n🧠 Model Response:\n${modelResponse.content}`);
     }
     catch (err) {
         console.error('❌ Model request failed:', err);
     }
 }
-// 🟩 Helper: Prompt once
+// Helper: Prompt once
 function promptOnce(promptText) {
     return new Promise(resolve => {
         console.log(promptText);

package/dist/db/fileIndex.js

@@ -6,6 +6,7 @@ import * as sqlTemplates from './sqlTemplates.js';
 import { CANDIDATE_LIMIT } from '../constants.js';
 import { getDbForRepo } from './client.js';
 import { scoreFiles } from '../fileRules/scoreFiles.js'; // 👈 NEW
+import chalk from 'chalk';
 export function indexFile(filePath, summary, type) {
     const stats = fs.statSync(filePath);
     const lastModified = stats.mtime.toISOString();
@@ -44,7 +45,7 @@ export function queryFiles(safeQuery, limit = 10) {
   `).all(safeQuery, limit);
 }
 export async function searchFiles(query, topK = 5) {
-    console.log(`🧠 Searching for query: "${query}"`);
+    console.log(chalk.yellow(`🧠 Searching for query: "${query}"`));
     const embedding = await generateEmbedding(query);
     if (!embedding) {
         console.log('⚠️ Failed to generate embedding for query');

package/dist/lib/generate.js

@@ -1,8 +1,9 @@
 // File: lib/generate.ts
 import { Spinner } from './spinner.js';
-import { readConfig } from '../config.js';
+import { Config, readConfig } from '../config.js';
 import { startModelProcess } from '../utils/checkModel.js';
-export async function generate(input, model) {
+export async function generate(input) {
+    const model = Config.getModel();
     const contextLength = readConfig().contextLength ?? 8192;
     let prompt = input.content;
     if (prompt.length > contextLength) {

package/dist/pipeline/modules/changeLogModule.js

@@ -18,7 +18,7 @@ If no meaningful changes are present, return the text: "NO UPDATE".
 ${input.content}
 --- DIFF END ---
 `.trim();
-        const response = await generate({ content: prompt }, model);
+        const response = await generate({ content: prompt });
         // Check if we received a meaningful result or "NO UPDATE"
         const content = response?.content?.trim();
         if (content === 'NO UPDATE') {

package/dist/pipeline/modules/commentModule.js

@@ -53,7 +53,7 @@ Rules:
 ${input.content}
 
 `.trim();
-        const response = await generate({ content: prompt }, model);
+        const response = await generate({ content: prompt });
         const contentToReturn = (response.content && response.content !== 'NO UPDATE') ? response.content : input.content;
         return {
             content: contentToReturn,

package/dist/pipeline/modules/commitSuggesterModule.js

@@ -22,7 +22,7 @@ Format your response exactly as:
 Here is the diff:
 ${content}
 `.trim();
-        const response = await generate({ content: prompt }, model);
+        const response = await generate({ content: prompt });
         const lines = response.content
             .split('\n')
             .map(line => line.trim())

package/dist/pipeline/modules/generateTestsModule.js

@@ -45,7 +45,7 @@ describe('moduleUnderTest', () => {
 ${content}
 --- END MODULE CODE ---
 `.trim();
-        const response = await generate({ content: prompt }, model);
+        const response = await generate({ content: prompt });
         if (!response)
             throw new Error('⚠️ No test code returned from model');
         return {

package/dist/pipeline/modules/kgModule.js

@@ -38,7 +38,7 @@ Do NOT include raw content from the file. Only provide the structured JSON outpu
 ${content}
 --- FILE CONTENT END ---
     `.trim();
-        const response = await generate({ content: prompt, filepath: input.filepath }, model);
+        const response = await generate({ content: prompt, filepath: input.filepath });
         try {
             // Clean the model output first
             const cleaned = await cleanupModule.run({ content: response.content });

package/dist/pipeline/modules/refactorModule.js

@@ -19,7 +19,7 @@ Refactor the following code:
 ${input.content}
 --- CODE END ---
     `.trim();
-        const response = await generate({ content: prompt }, model);
+        const response = await generate({ content: prompt });
         if (!response) {
             throw new Error('❌ Model returned empty response for refactoring.');
         }

package/dist/pipeline/modules/repairTestsModule.js

@@ -26,7 +26,7 @@ Instructions:
 
 Output the repaired test file:
 `.trim();
-        const response = await generate({ content: prompt }, model);
+        const response = await generate({ content: prompt });
         if (!response)
             throw new Error("⚠️ No repaired test code returned from model");
         return {

package/dist/pipeline/modules/reviewModule.js

@@ -20,7 +20,7 @@ Format your response exactly as:
 Changes:
 ${content}
 `.trim();
-        const response = await generate({ content: prompt, filepath }, model);
+        const response = await generate({ content: prompt, filepath });
         // Parse response: only keep numbered lines
         const lines = response.content
             .split('\n')

package/dist/pipeline/modules/summaryModule.js

@@ -27,7 +27,7 @@ Extension: ${ext}
 ${content}
 --- FILE CONTENT END ---
     `.trim();
-        const response = await generate({ content: prompt, filepath }, model);
+        const response = await generate({ content: prompt, filepath });
         if (response.content) {
             response.summary = response.content;
             console.log('\n📝 Summary:\n');

package/dist/utils/buildContextualPrompt.js

@@ -1,74 +1,112 @@
-// File: src/utils/buildContextualPrompt.ts
-import chalk from 'chalk';
-import path from 'path';
-function estimateTokenCount(text) {
-    return Math.round(text.length / 4); // simple heuristic approximation
-}
-export function buildContextualPrompt({ baseInstruction, code, summary, functions, relatedFiles, projectFileTree, }) {
-    const parts = [baseInstruction];
-    if (summary) {
-        parts.push(`📄 File Summary:\n${summary}`);
-    }
-    if (functions?.length) {
-        const formattedFunctions = functions
-            .map(fn => `• ${fn.name}:\n${fn.content}`)
-            .join('\n\n');
-        parts.push(`🔧 Functions:\n${formattedFunctions}`);
+// src/utils/buildContextualPrompt.ts
+import { getDbForRepo } from "../db/client.js";
+import { generateFocusedFileTree } from "./fileTree.js";
+export async function buildContextualPrompt({ topFile, query, kgDepth = 3, }) {
+    const db = getDbForRepo();
+    const log = (...args) => console.log("[buildContextualPrompt]", ...args);
+    const promptSections = [];
+    const seenPaths = new Set();
+    function summarizeForPrompt(summary, maxLines = 5) {
+        if (!summary)
+            return undefined;
+        const lines = summary.split("\n").map(l => l.trim()).filter(Boolean);
+        if (lines.length <= maxLines)
+            return lines.join(" ");
+        return lines.slice(0, maxLines).join(" ") + " …";
     }
-    else {
-        console.log(chalk.yellow(` ⚠️ No functions found in top rated file.`));
+    // --- Step 1: Top file summary ---
+    if (topFile.summary) {
+        promptSections.push(`**Top file:** ${topFile.path}\n${topFile.summary}`);
+        seenPaths.add(topFile.path);
     }
-    if (relatedFiles?.length) {
-        const formattedRelatedFiles = relatedFiles
-            .map(f => {
-            const relatedFunctions = f.functions
-                .map(fn => `  • ${fn.name}:\n    ${fn.content}`)
-                .join('\n\n');
-            return `• ${f.path}: ${f.summary}\n${relatedFunctions}`;
-        })
-            .join('\n\n');
-        parts.push(`📚 Related Files:\n${formattedRelatedFiles}`);
+    // --- Step 2: KG entities/tags for top file ---
+    const topEntitiesStmt = db.prepare(`
+    SELECT et.entity_type, et.entity_id, tm.name AS tag
+    FROM entity_tags et
+    JOIN tags_master tm ON et.tag_id = tm.id
+    WHERE et.entity_id = ?
+  `);
+    const topEntitiesRows = topEntitiesStmt.all(topFile.id);
+    if (topEntitiesRows.length > 0) {
+        const tags = topEntitiesRows.map(r => `- **${r.entity_type}**: ${r.tag}`);
+        promptSections.push(`**Knowledge Graph context for ${topFile.path}:**\n${tags.join("\n")}`);
     }
-    if (projectFileTree) {
-        parts.push(`📁 Project File Structure:\n\`\`\`\n${projectFileTree.trim()}\n\`\`\``);
+    // --- Step 3: Recursive KG traversal ---
+    const kgRelatedStmt = db.prepare(`
+    SELECT DISTINCT f.id, f.path, f.summary
+    FROM edges e
+    JOIN files f ON e.target_id = f.id
+    WHERE e.source_type = 'file'
+      AND e.target_type = 'file'
+      AND e.source_id = ?
+  `);
+    function getRelatedKGFiles(fileId, visited = new Set()) {
+        if (visited.has(fileId)) {
+            log(`🔹 Already visited fileId ${fileId}, skipping`);
+            return [];
+        }
+        visited.add(fileId);
+        const rows = kgRelatedStmt.all(fileId);
+        if (rows.length === 0) {
+            log(`⚠️ No edges found for fileId ${fileId}`);
+        }
+        else {
+            log(`🔹 Found ${rows.length} related files for fileId ${fileId}:`, rows.map(r => r.path));
+        }
+        let results = [];
+        for (const row of rows) {
+            results.push(row);
+            results.push(...getRelatedKGFiles(row.id, visited));
+        }
+        return results;
     }
-    parts.push(`\n--- CODE START ---\n${code}\n--- CODE END ---`);
-    const prompt = parts.join('\n\n');
-    const tokenEstimate = estimateTokenCount(prompt);
-    // 🔵 Colorized diagnostic output
-    // 🔵 Colorized diagnostic output
-    const header = chalk.bgBlue.white.bold(' [SCAI] Prompt Overview ');
-    const labelColor = chalk.cyan;
-    const contentColor = chalk.gray;
-    console.log('\n' + header);
-    console.log(labelColor('🔢 Token Estimate:'), contentColor(`${tokenEstimate.toLocaleString()} tokens`));
-    // 📄 Summary
-    if (summary) {
-        console.log(labelColor('📄 Summary:'), contentColor(`${estimateTokenCount(summary).toLocaleString()} tokens`));
+    function buildFileTree(file, depth, visited = new Set()) {
+        log(`buildFileTree - file=${file.path}, depth=${depth}`);
+        if (depth === 0 || visited.has(file.id)) {
+            return { id: file.id.toString(), path: file.path, summary: summarizeForPrompt(file.summary) };
+        }
+        visited.add(file.id);
+        const relatedFiles = getRelatedKGFiles(file.id, visited)
+            .map(f => ({ id: f.id, path: f.path, summary: f.summary }))
+            .slice(0, 5); // limit max 5 children per node
+        log(`File ${file.path} has ${relatedFiles.length} related files`);
+        const relatedNodes = relatedFiles.map(f => buildFileTree(f, depth - 1, visited));
+        return {
+            id: file.id.toString(),
+            path: file.path,
+            summary: summarizeForPrompt(file.summary),
+            related: relatedNodes.length ? relatedNodes : undefined,
+        };
     }
-    // 🔧 Functions
-    if (functions?.length) {
-        const fnCount = functions.length;
-        const fnTokens = functions.reduce((sum, f) => sum + estimateTokenCount(f.content), 0);
-        console.log(labelColor(`🔧 Functions (${fnCount}):`), contentColor(`${fnTokens.toLocaleString()} tokens`));
+    const kgTree = buildFileTree({ id: topFile.id, path: topFile.path, summary: topFile.summary }, kgDepth);
+    const kgJson = JSON.stringify(kgTree, null, 2);
+    promptSections.push(`**KG-Related Files (JSON tree, depth ${kgDepth}):**\n\`\`\`json\n${kgJson}\n\`\`\``);
+    // --- Step 4: File tree (shallow, depth 2) ---
+    let fileTree = "";
+    try {
+        fileTree = generateFocusedFileTree(topFile.path, 2);
+        if (fileTree) {
+            promptSections.push(`**Focused File Tree (depth 2):**\n\`\`\`\n${fileTree}\n\`\`\``);
+        }
     }
-    // 📚 Related Files
-    if (relatedFiles?.length) {
-        const relCount = relatedFiles.length;
-        const relTokens = relatedFiles.reduce((sum, f) => sum + estimateTokenCount(f.summary), 0);
-        console.log(labelColor(`📚 Related Files (${relCount}):`), contentColor(`${relTokens.toLocaleString()} tokens`));
-        // Optional: Show top 3 file names
-        const fileList = relatedFiles.slice(0, 3).map(f => `- ${path.basename(f.path)}`).join('\n');
-        if (fileList)
-            console.log(contentColor(fileList + (relCount > 3 ? `\n  ...+${relCount - 3} more` : '')));
+    catch (e) {
+        console.warn("⚠️ Could not generate file tree:", e);
     }
-    // 📁 File Tree
-    if (projectFileTree) {
-        console.log(labelColor('📁 File Tree:'), contentColor(`${estimateTokenCount(projectFileTree).toLocaleString()} tokens`));
+    // --- Step 5: Code snippet ---
+    const MAX_LINES = 50;
+    if (topFile.code) {
+        const lines = topFile.code.split("\n").slice(0, MAX_LINES);
+        let snippet = lines.join("\n");
+        if (topFile.code.split("\n").length > MAX_LINES) {
+            snippet += "\n... [truncated]";
+        }
+        promptSections.push(`**Code Context (first ${MAX_LINES} lines):**\n\`\`\`\n${snippet}\n\`\`\``);
     }
-    // 📦 Code Section
-    console.log(labelColor('📦 Code:'), contentColor(`${estimateTokenCount(prompt).toLocaleString()} tokens`));
-    // 📌 Key
-    console.log(labelColor('🔍 Key:'), contentColor('[buildContextualPrompt]\n'));
-    return prompt;
+    // --- Step 6: User query ---
+    promptSections.push(`**Query:** ${query}`);
+    // --- Step 7: Combine prompt ---
+    const promptText = promptSections.join("\n\n---\n\n");
+    log("✅ Contextual prompt built for:", topFile.path);
+    log("📄 Prompt preview:\n", promptText + "\n");
+    return promptText;
 }

package/dist/utils/sanitizeQuery.js

@@ -2,18 +2,26 @@
 import { STOP_WORDS } from '../fileRules/stopWords.js';
 export function sanitizeQueryForFts(input) {
     input = input.trim().toLowerCase();
-    // If it's a single filename-like string (includes dots or slashes), quote it
+    // If the whole input looks like a filename/path, quote it
     if (/^[\w\-./]+$/.test(input) && !/\s/.test(input)) {
-        // Escape quotes and wrap with double-quotes for FTS safety
         return `"${input.replace(/"/g, '""')}"*`;
     }
-    // Otherwise, treat it as a natural language prompt
     const tokens = input
         .split(/\s+/)
         .map(token => token.toLowerCase())
+        .map(token => {
+        // If the token looks like a filename/path, keep it quoted
+        if (/[\w]+\.[a-z0-9]+$/.test(token)) {
+            return `"${token.replace(/"/g, '""')}"`;
+        }
+        // Otherwise, clean it like normal
+        return token
+            .replace(/[^a-z0-9_*"]/gi, '') // remove all invalid FTS5 chars
+            .replace(/'/g, "''");
+    })
         .filter(token => token.length > 2 &&
-        !STOP_WORDS.has(token) &&
-        /^[a-z0-9]+$/.test(token))
-        .map(token => token.replace(/[?*\\"]/g, '').replace(/'/g, "''") + '*');
+        !STOP_WORDS.has(token.replace(/[*"]/g, '')) // check unquoted
+    )
+        .map(token => (token.startsWith('"') ? token : token + '*'));
     return tokens.length > 0 ? tokens.join(' OR ') : '*';
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "scai",
-  "version": "0.1.110",
+  "version": "0.1.111",
   "type": "module",
   "bin": {
     "scai": "./dist/index.js"