trace.ai-cli/services/fileAnalyzer.js

A powerful AI-powered CLI tool

const path = require('path');
const fs = require('fs').promises;
const pdfParse = require('pdf-parse');
const { getFileType, readFileContent } = require('../utils/fileUtils');
const { processWithAI } = require('./aiService');
const { extractTextFromImage } = require('./imageService');

async function analyzeFile(filePath, query = '', mode = 2) {
    try {
        const fileType = getFileType(filePath);
        
        // Handle images - use vision API only for images
        if (fileType === 'image') {
            const result = await extractTextFromImage(filePath, query, mode);
            return { text: result };
        }
        
        // Handle documents (PDFs, Word, etc.)
        if (fileType === 'document') {
            const fileName = path.basename(filePath);
            const ext = path.extname(filePath).toLowerCase();
            
            // Extract text from PDF
            if (ext === '.pdf') {
                try {
                    const dataBuffer = await fs.readFile(filePath);
                    const pdfData = await pdfParse(dataBuffer);
                    const content = pdfData.text;
                    
                    if (!content || content.trim().length === 0) {
                        return {
                            text: `❌ **Unable to Extract Text from PDF**\n\n` +
                                  `The PDF "${fileName}" appears to be empty or contains only images/scanned content.\n\n` +
                                  `**Try using the /image command for scanned PDFs:**\n` +
                                  `\`/image "${filePath}" ${query || 'analyze this document'}\``
                        };
                    }
                    
                    // Build prompt for AI analysis
                    let prompt;
                    if (query) {
                        prompt = `PDF Document: ${fileName}
Content:
\`\`\`
${content}
\`\`\`

User Question: ${query}`;
                    } else {
                        prompt = `Analyze this PDF document (${fileName}):

\`\`\`
${content}
\`\`\`

Please provide a comprehensive summary and analysis of the document.`;
                    }
                    
                    const result = await processWithAI(prompt, '', mode);
                    return { text: result || 'No response generated' };
                    
                } catch (pdfError) {
                    return {
                        text: `❌ **Error Reading PDF**\n\n` +
                              `Failed to extract text from "${fileName}": ${pdfError.message}\n\n` +
                              `**Possible reasons:**\n` +
                              `- The PDF is corrupted or password-protected\n` +
                              `- The PDF contains only images (scanned document)\n\n` +
                              `**Try using /image command for scanned PDFs:**\n` +
                              `\`/image "${filePath}" ${query || 'analyze this document'}\``
                    };
                }
            }
            
            // For other document types (Word, Excel, etc.)
            return { 
                text: `❌ **Document Type Not Supported**\n\n` +
                      `The file "${fileName}" is a ${ext.toUpperCase()} document.\n\n` +
                      `**Supported formats:**\n` +
                      `- PDF files (.pdf) - Text extraction supported\n` +
                      `- Text files (.txt, .md, etc.)\n` +
                      `- Code files (.js, .py, .java, etc.)\n\n` +
                      `**For ${ext.toUpperCase()} files:**\n` +
                      `Please convert to PDF or text format first.`
            };
        }
        
        // Handle text-based files
        const content = await readFileContent(filePath);
        const fileName = path.basename(filePath);
        
        let prompt;
        if (query) {
            prompt = `File: ${fileName} (${fileType})
Content:
\`\`\`
${content}
\`\`\`

User Question: ${query}`;
        } else {
            if (fileType !== 'Unknown' && fileType !== 'Text' && fileType !== 'Markdown') {
                prompt = `Analyze this ${fileType} code file (${fileName}):

\`\`\`
${content}
\`\`\`

Please provide:
1. Code overview and purpose
2. Key functions/components
3. Potential issues or improvements
4. Code quality assessment
5. Suggestions for optimization`;
            } else {
                prompt = `Analyze this file content (${fileName}):

\`\`\`
${content}
\`\`\`

Please provide a summary and analysis of the content.`;
            }
        }
        
        const result = await processWithAI(prompt, '', mode);
        return { text: result || 'No response generated' };
    } catch (error) {
        throw error;
    }
}

module.exports = {
    analyzeFile
};