@aj-archipelago/cortex 1.4.7 → 1.4.8

package/lib/fileUtils.js

@@ -983,10 +983,7 @@ async function generateFileMessageContent(fileParam, contextId, contextKey = nul
     if (!contextId) {
         // Without contextId, we can't look up in collection
         // Return a basic content object from the URL
-        return {
-            type: 'file',
-            url: fileParam
-        };
+        return null;
     }
 
     // Load file collection
@@ -1000,31 +997,14 @@ async function generateFileMessageContent(fileParam, contextId, contextKey = nul
         return null;
     }
 
-    // Determine file type based on filename extension or existing type
-    const filename = foundFile.filename || '';
-    const extension = filename.split('.').pop()?.toLowerCase() || '';
-    const isImage = ['jpg', 'jpeg', 'png', 'gif', 'bmp', 'webp', 'svg'].includes(extension);
-    const fileType = foundFile.type || (isImage ? 'image_url' : 'file');
-
-    // Create content object in the proper format for plugins (url and gcs)
-    if (fileType === 'image_url') {
-        return {
-            type: 'image_url',
-            image_url: { url: foundFile.url },
-            url: foundFile.url,
-            gcs: foundFile.gcs || null,
-            originalFilename: foundFile.filename || null,
-            hash: foundFile.hash || null
-        };
-    } else {
-        return {
-            type: 'file',
-            url: foundFile.url,
-            gcs: foundFile.gcs || null,
-            originalFilename: foundFile.filename || null,
-            hash: foundFile.hash || null
-        };
-    }
+    return {
+        type: 'image_url',
+        url: foundFile.url,
+        gcs: foundFile.gcs || null,
+        originalFilename: foundFile.filename || null,
+        hash: foundFile.hash || null
+    };
+
 }
 
 /**
@@ -1448,6 +1428,35 @@ function getMimeTypeFromExtension(extension, defaultMimeType = 'application/octe
     return mimeType || defaultMimeType;
 }
 
+/**
+ * Check if a MIME type represents a text-based file that can be read as text
+ * @param {string} mimeType - MIME type to check
+ * @returns {boolean} - Returns true if it's a text-based MIME type
+ */
+function isTextMimeType(mimeType) {
+    if (!mimeType || typeof mimeType !== 'string') {
+        return false;
+    }
+    
+    const mimeTypeLower = mimeType.toLowerCase();
+    
+    // All text/* types
+    if (mimeTypeLower.startsWith('text/')) {
+        return true;
+    }
+    
+    // Text-based application types (consistent with sys_tool_writefile and sys_tool_editfile)
+    const textApplicationTypes = [
+        'application/json',
+        'application/javascript',
+        'application/x-javascript',
+        'application/typescript',
+        'application/xml',
+    ];
+    
+    return textApplicationTypes.includes(mimeTypeLower);
+}
+
 export { 
     computeFileHash,
     computeBufferHash,
@@ -1476,6 +1485,7 @@ export {
     uploadImageToCloud,
     resolveFileHashesToContent,
     getMimeTypeFromFilename,
-    getMimeTypeFromExtension
+    getMimeTypeFromExtension,
+    isTextMimeType
 };
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@aj-archipelago/cortex",
-  "version": "1.4.7",
+  "version": "1.4.8",
   "description": "Cortex is a GraphQL API for AI. It provides a simple, extensible interface for using AI services from OpenAI, Azure and others.",
   "private": false,
   "repository": {

package/pathways/system/entity/tools/sys_tool_editfile.js

@@ -2,7 +2,7 @@
 // Entity tool that modifies existing files by replacing line ranges or exact string matches
 import logger from '../../../../lib/logger.js';
 import { axios } from '../../../../lib/requestExecutor.js';
-import { uploadFileToCloud, findFileInCollection, loadFileCollection, saveFileCollection, getMimeTypeFromFilename, resolveFileParameter, deleteFileByHash, modifyFileCollectionWithLock } from '../../../../lib/fileUtils.js';
+import { uploadFileToCloud, findFileInCollection, loadFileCollection, saveFileCollection, getMimeTypeFromFilename, resolveFileParameter, deleteFileByHash, modifyFileCollectionWithLock, isTextMimeType } from '../../../../lib/fileUtils.js';
 
 export default {
     prompt: [],
@@ -297,9 +297,7 @@ export default {
             let mimeType = getMimeTypeFromFilename(filename, 'text/plain');
             
             // Add charset=utf-8 for text-based MIME types
-            if (mimeType.startsWith('text/') || mimeType === 'application/json' || 
-                mimeType === 'application/javascript' || mimeType === 'application/typescript' ||
-                mimeType === 'application/xml') {
+            if (isTextMimeType(mimeType)) {
                 mimeType = `${mimeType}; charset=utf-8`;
             }
 

package/pathways/system/entity/tools/sys_tool_readfile.js

@@ -2,7 +2,32 @@
 // Tool pathway that reads text files with line number support
 import logger from '../../../../lib/logger.js';
 import { axios } from '../../../../lib/requestExecutor.js';
-import { resolveFileParameter } from '../../../../lib/fileUtils.js';
+import { resolveFileParameter, getMimeTypeFromFilename, isTextMimeType } from '../../../../lib/fileUtils.js';
+
+/**
+ * Check if a file is a text file type that can be read
+ * @param {string} url - File URL or path
+ * @returns {boolean} - Returns true if it's a text file, false otherwise
+ */
+function isTextFile(url) {
+    if (!url || typeof url !== 'string') {
+        return false;
+    }
+
+    // Extract filename from URL (remove query string and fragment)
+    const urlPath = url.split('?')[0].split('#')[0];
+    
+    // Use existing MIME utility to get MIME type
+    const mimeType = getMimeTypeFromFilename(urlPath);
+    
+    if (!mimeType || mimeType === 'application/octet-stream') {
+        // Unknown MIME type, reject to be safe
+        return false;
+    }
+
+    // Use shared utility function for consistency with other tools
+    return isTextMimeType(mimeType);
+}
 
 export default {
     prompt: [],
@@ -11,8 +36,8 @@ export default {
         type: "function",
         icon: "📖",
         function: {
-            name: "ReadFile",
-            description: "Read text content from a file. Can read the entire file or specific line ranges. Use this to access and analyze text files from your file collection. Supports text files, markdown files, html, csv, and other document formats that can be converted to text, but not images, videos, or audio files or pdfs.",
+            name: "ReadTextFile",
+            description: "Read text content from a file. Can read the entire file or specific line ranges. Use this to access and analyze text files from your file collection. Supports text files, markdown files, html, csv, and other document formats that can be converted to text.  DOES NOT support binary files, images, videos, or audio files or pdfs.",
             parameters: {
                 type: "object",
                 properties: {
@@ -77,6 +102,19 @@ export default {
                 return JSON.stringify(errorResult);
             }
 
+            // Check if file is a text type before attempting to read
+            if (!isTextFile(cloudUrl)) {
+                const mimeType = getMimeTypeFromFilename(cloudUrl.split('?')[0].split('#')[0]);
+                const detectedType = mimeType || 'unknown type';
+                
+                const errorResult = {
+                    success: false,
+                    error: `This tool only supports text files. The file appears to be a non-text file (MIME type: ${detectedType}). For images, PDFs, videos, or other non-text files, please use the AnalyzeImage, AnalyzePDF, or AnalyzeVideo tools instead.`
+                };
+                resolver.tool = JSON.stringify({ toolUsed: "ReadFile" });
+                return JSON.stringify(errorResult);
+            }
+
             if (startLine !== undefined) {
                 if (typeof startLine !== 'number' || !Number.isInteger(startLine) || startLine < 1) {
                 const errorResult = {
@@ -130,6 +168,17 @@ export default {
                 throw new Error(`Failed to download file content: ${response.status}`);
             }
 
+            // Secondary check: verify content-type header if available
+            const contentType = response.headers['content-type'] || response.headers['Content-Type'];
+            if (contentType && !isTextMimeType(contentType)) {
+                const errorResult = {
+                    success: false,
+                    error: `This tool only supports text files. The file appears to be a non-text file (Content-Type: ${contentType}). For images, PDFs, videos, or other non-text files, please use the AnalyzeImage, AnalyzePDF, or AnalyzeVideo tools instead.`
+                };
+                resolver.tool = JSON.stringify({ toolUsed: "ReadFile" });
+                return JSON.stringify(errorResult);
+            }
+
             // Explicitly decode as UTF-8 to prevent mojibake (encoding corruption)
             const textContent = Buffer.from(response.data).toString('utf8');
             const allLines = textContent.split(/\r?\n/);

package/pathways/system/entity/tools/sys_tool_writefile.js

@@ -1,7 +1,7 @@
 // sys_tool_writefile.js
 // Entity tool that writes content to a file and uploads it to cloud storage
 import logger from '../../../../lib/logger.js';
-import { uploadFileToCloud, addFileToCollection, getMimeTypeFromFilename } from '../../../../lib/fileUtils.js';
+import { uploadFileToCloud, addFileToCollection, getMimeTypeFromFilename, isTextMimeType } from '../../../../lib/fileUtils.js';
 
 // Helper function to format file size
 function formatFileSize(bytes) {
@@ -129,9 +129,7 @@ export default {
             let mimeType = getMimeTypeFromFilename(filename, 'text/plain');
             
             // Add charset=utf-8 for text-based MIME types to ensure proper encoding
-            if (mimeType.startsWith('text/') || mimeType === 'application/json' || 
-                mimeType === 'application/javascript' || mimeType === 'application/typescript' ||
-                mimeType === 'application/xml') {
+            if (isTextMimeType(mimeType)) {
                 mimeType = `${mimeType}; charset=utf-8`;
             }
 

package/tests/integration/features/tools/fileCollection.test.js

@@ -518,7 +518,7 @@ test('generateFileMessageContent should find file by ID', async t => {
         const result = await generateFileMessageContent(fileId, contextId);
         
         t.truthy(result);
-        t.is(result.type, 'file');
+        t.is(result.type, 'image_url');
         t.is(result.url, 'https://example.com/test.pdf');
         t.is(result.gcs, 'gs://bucket/test.pdf');
         t.is(result.originalFilename, 'test.pdf');
@@ -607,8 +607,7 @@ test('generateFileMessageContent should detect image type', async t => {
         
         t.truthy(result);
         t.is(result.type, 'image_url');
-        t.truthy(result.image_url);
-        t.is(result.image_url.url, 'https://example.com/image.jpg');
+        t.is(result.url, 'https://example.com/image.jpg');
     } finally {
         await cleanup(contextId);
     }

package/tests/unit/core/util.test.js

@@ -239,12 +239,10 @@ test('generateFileMessageContent should return null for invalid input', async t
     t.is(await generateFileMessageContent(123, 'context-1'), null);
 });
 
-test('generateFileMessageContent should return basic object when no contextId', async t => {
+test('generateFileMessageContent should return null when no contextId', async t => {
     const result = await generateFileMessageContent('https://example.com/file.pdf', null);
     
-    t.truthy(result);
-    t.is(result.type, 'file');
-    t.is(result.url, 'https://example.com/file.pdf');
+    t.is(result, null);
 });
 
 test('generateFileMessageContent should return null for file not in collection', async t => {