From 5c4baa74e5b869ee607a54b8c88be5cb3fedf60b Mon Sep 17 00:00:00 2001
From: Reza Rahemtola <reza.rahemtola@epitech.eu>
Date: Wed, 11 Sep 2024 16:34:54 +0900
Subject: [PATCH] feat(kb): Support programming languages

---
 package-lock.json                | 14 ----------
 package.json                     |  1 -
 src/utils/knowledge/embedding.ts | 25 +++++++++++++++--
 src/utils/knowledge/parsing.ts   | 47 ++++++++++++++++++++++++++++----
 4 files changed, 64 insertions(+), 23 deletions(-)

diff --git a/package-lock.json b/package-lock.json
index 52aa93e..f5c633d 100644
--- a/package-lock.json
+++ b/package-lock.json
@@ -27,7 +27,6 @@
         "localforage": "^1.10.0",
         "marked": "^14.0.0",
         "marked-highlight": "^2.1.4",
-        "mime": "^4.0.4",
         "ml-distance": "^4.0.1",
         "pdfjs-dist": "^4.5.136",
         "pinia": "^2.2.1",
@@ -13538,19 +13537,6 @@
         "miller-rabin": "bin/miller-rabin"
       }
     },
-    "node_modules/mime": {
-      "version": "4.0.4",
-      "funding": [
-        "https://github.com/sponsors/broofa"
-      ],
-      "license": "MIT",
-      "bin": {
-        "mime": "bin/cli.js"
-      },
-      "engines": {
-        "node": ">=16"
-      }
-    },
     "node_modules/mime-db": {
       "version": "1.52.0",
       "license": "MIT",
diff --git a/package.json b/package.json
index 7f5eee9..fb6004e 100644
--- a/package.json
+++ b/package.json
@@ -42,7 +42,6 @@
     "localforage": "^1.10.0",
     "marked": "^14.0.0",
     "marked-highlight": "^2.1.4",
-    "mime": "^4.0.4",
     "ml-distance": "^4.0.1",
     "pdfjs-dist": "^4.5.136",
     "pinia": "^2.2.1",
diff --git a/src/utils/knowledge/embedding.ts b/src/utils/knowledge/embedding.ts
index b0cc0da..42e37d6 100644
--- a/src/utils/knowledge/embedding.ts
+++ b/src/utils/knowledge/embedding.ts
@@ -43,8 +43,29 @@ const getTextSplitter = (
   chunkOverlap: number,
 ): MarkdownTextSplitter | RecursiveCharacterTextSplitter => {
   switch (fileType) {
-    case 'text/markdown':
-      return new MarkdownTextSplitter();
+    case 'markdown':
+    case 'html':
+    case 'cpp':
+    case 'go':
+    case 'java':
+    case 'js':
+    case 'php':
+    case 'proto':
+    case 'rst':
+    case 'scala':
+    case 'swift':
+    case 'sol':
+      return RecursiveCharacterTextSplitter.fromLanguage(fileType);
+    case 'ts':
+      return RecursiveCharacterTextSplitter.fromLanguage('js');
+    case 'py':
+      return RecursiveCharacterTextSplitter.fromLanguage('python');
+    case 'rb':
+      return RecursiveCharacterTextSplitter.fromLanguage('ruby');
+    case 'rs':
+      return RecursiveCharacterTextSplitter.fromLanguage('rust');
+    case 'md':
+      return RecursiveCharacterTextSplitter.fromLanguage('markdown');
     default:
       return new RecursiveCharacterTextSplitter({
         chunkSize,
diff --git a/src/utils/knowledge/parsing.ts b/src/utils/knowledge/parsing.ts
index fc7279e..e2f9636 100644
--- a/src/utils/knowledge/parsing.ts
+++ b/src/utils/knowledge/parsing.ts
@@ -1,8 +1,27 @@
-import mime from 'mime';
 import * as pdfjs from 'pdfjs-dist';
 import { TextItem } from 'pdfjs-dist/types/src/display/api';
 
-export const supportedInputFiles = ['.txt', '.md', '.pdf'].join(',');
+export const supportedInputFiles = [
+  '.txt',
+  '.md',
+  '.markdown',
+  '.pdf',
+  '.html',
+  '.cpp',
+  '.go',
+  '.java',
+  '.js',
+  '.ts',
+  '.php',
+  '.proto',
+  '.py',
+  '.rst',
+  '.rb',
+  '.rs',
+  '.scala',
+  '.swift',
+  '.sol',
+].join(',');
 
 const extractTextFromPdfFile = async (file: File): Promise<string> => {
   const pdfUrl = URL.createObjectURL(file);
@@ -27,15 +46,31 @@ const extractTextFromPdfFile = async (file: File): Promise<string> => {
 
 export const extractFileContent = async (file: File): Promise<{ type: string; content: string }> => {
   let extractedText = '';
-  const fileType = mime.getType(file.name) ?? file.type;
+  const fileType = file.name.split('.').pop();
 
   try {
     switch (fileType) {
-      case 'application/pdf':
+      case 'pdf':
         extractedText = await extractTextFromPdfFile(file);
         break;
-      case 'text/markdown':
-      case 'text/plain':
+      case 'markdown':
+      case 'md':
+      case 'txt':
+      case 'html':
+      case 'cpp':
+      case 'go':
+      case 'java':
+      case 'js':
+      case 'ts':
+      case 'php':
+      case 'proto':
+      case 'py':
+      case 'rst':
+      case 'rb':
+      case 'rs':
+      case 'scala':
+      case 'swift':
+      case 'sol':
         extractedText = await new Promise((resolve, reject) => {
           const reader = new FileReader();
           reader.onload = (event) => resolve(event.target!.result as string);