Add logic to parse docx files

api/requirements.txt

@@ -34,4 +34,5 @@ setuptools
 
 lxml 
 beautifulsoup4 
-async-web-search
+async-web-search
+python-docx

api/src/utils/custom_sources/read_content.py

@@ -1,5 +1,6 @@
 from io import BytesIO
 
+from docx import Document
 from fastapi import UploadFile
 from pypdf import PdfReader
 
@@ -21,6 +22,10 @@ def _read_pdf(self, content: bytes) -> tuple[str, PdfReader]:
     def _read_txt(self, content: bytes) -> str:
         return content.decode()
 
+    def _read_docx(self, content: bytes) -> str:
+        doc = Document(BytesIO(content))
+        return "\n\n".join([p.text for p in doc.paragraphs])
+
     async def _read_file(self, file: UploadFile, preserve: bool):
         file_bytes = await file.read()
 
@@ -31,6 +36,8 @@ async def _read_file(self, file: UploadFile, preserve: bool):
             text_content, _ = self._read_pdf(file_bytes)
         elif file.content_type == "text/plain":
             text_content = self._read_txt(file_bytes)
+        elif file.content_type == "application/vnd.openxmlformats-officedocument.wordprocessingml.document":
+            text_content = self._read_docx(file_bytes)
         else:
             return BytesIO(file_bytes)
 

api/src/utils/custom_sources/save_uploaded_sources.py

@@ -8,10 +8,9 @@
 TEN_MB = 10 * 1024 * 1024
 
 
-class UploadedFiles:
+class UploadedFiles(ReadContent):
     def __init__(self, session_id: str):
         self.session_id = session_id
-        self.content_reader = ReadContent()
 
     async def _extract_content(self, file: UploadFile):
         file_bytes = await file.read()
@@ -20,15 +19,20 @@ async def _extract_content(self, file: UploadFile):
             return None
 
         if file.content_type == "application/pdf":
-            text_content, pdf_reader = self.content_reader._read_pdf(file_bytes)
+            text_content, pdf_reader = self._read_pdf(file_bytes)
 
             metadata = {**(pdf_reader.metadata or {}), "pages": pdf_reader.get_num_pages()}
             content_type = "application/pdf"
         elif file.content_type == "text/plain":
-            text_content = self.content_reader._read_txt(file_bytes)
+            text_content = self._read_txt(file_bytes)
 
             metadata = {}
             content_type = "text/plain"
+        elif file.content_type == "application/vnd.openxmlformats-officedocument.wordprocessingml.document":
+            text_content = self._read_docx(file_bytes)
+
+            metadata = {}
+            content_type = "application/vnd.openxmlformats-officedocument.wordprocessingml.document"
         else:
             return None
 

api/src/utils/summarize_custom_sources.py

@@ -30,7 +30,7 @@ def summarize_custom_sources_prompt(combined_content: str) -> str:
     """
 
 
-async def get_source_content(source_url: str) -> str:
+async def get_source_content(source_url: str):
     """
     Get the content of a source URL.
     """
@@ -45,8 +45,11 @@ async def get_source_content(source_url: str) -> str:
         text_content, _ = content_reader._read_pdf(content_byte)
     elif blob.content_type == "text/plain":
         text_content = content_reader._read_txt(content_byte)
+    elif blob.content_type == "application/vnd.openxmlformats-officedocument.wordprocessingml.document":
+        text_content = content_reader._read_docx(content_byte)
     else:
-        raise ValueError(f"Unsupported content type: {blob.content_type}")
+        print(f"Unsupported content type: {blob.content_type}")
+        return None
 
     return text_content
 

app/src/lib/components/ChatBoxAndWidget.svelte

@@ -6,7 +6,7 @@
 	import ChatBoxAttachment from './ChatBoxAttachment.svelte';
 	import ChatBoxAttachmentPreview from './ChatBoxAttachmentPreview.svelte';
 	import { getAttachmentsContext } from '@/stores/attachmentsContext.svelte';
-	import Spinner from './Spinner.svelte';
+	import Spinner from './Spinner2.svelte';
 
 	export let searchTerm = '';
 	export let loading = false;

app/src/lib/components/ChatBoxAttachment.svelte

@@ -52,6 +52,7 @@
 		const fileId = `${sessionId}_${slug(file.name)}`;
 
 		const formData = new FormData();
+		// const newFile = new File([file], fileId, { type: file.type });
 		formData.append('file', file);
 		formData.append('filename', fileId);
 
@@ -84,10 +85,9 @@
 	<PaperclipIcon class="h-5 w-5" />
 </Button>
 
-<!-- TODO: add support for .docx -->
 <input
 	type="file"
-	accept=".pdf,.txt"
+	accept=".pdf,.txt,.docx"
 	multiple
 	class="hidden"
 	bind:this={fileInput}

app/src/lib/components/ChatBoxAttachmentPreview.svelte

@@ -2,7 +2,7 @@
 	import { FileIcon, XIcon } from 'lucide-svelte';
 	import { Button } from './ui/button';
 	import { getAttachmentsContext } from '@/stores/attachmentsContext.svelte';
-	import Spinner from './Spinner.svelte';
+	import Spinner from './Spinner2.svelte';
 
 	const { removeUploadItem, sessionUploadItems$ } = getAttachmentsContext();
 
@@ -32,7 +32,7 @@
 </script>
 
 <div class="p-2 flex flex-wrap gap-2 bg-zinc-800/30">
-	{#each validItems as { file, id, loading } (id)}
+	{#each validItems as { file, id, loading }, ix (id + ix)}
 		<div class="flex items-center w-56 gap-2 justify-between bg-zinc-700/30 rounded p-2">
 			<div class="p-1">
 				{#if loading}

app/src/lib/components/Spinner2.svelte

@@ -0,0 +1,33 @@
+<script lang="ts" context="module">
+	export type SpinnerSize = 'small' | 'medium' | 'large';
+	const sizeClasses: Record<SpinnerSize, string> = {
+		small: 'w-4 h-4',
+		medium: 'w-6 h-6',
+		large: 'w-10 h-10'
+	};
+</script>
+
+<script lang="ts">
+	export let size: SpinnerSize = 'medium';
+	export let color = 'text-gray-600';
+</script>
+
+<div class="flex w-full h-full justify-center items-center">
+	<div class={`${sizeClasses[size]} ${color}`}>
+		<svg class="animate-spin" viewBox="0 0 24 24">
+			<path
+				fill="none"
+				stroke="currentColor"
+				stroke-width="2"
+				stroke-linecap="round"
+				d="M12 2C6.477 2 2 6.477 2 12c0 1.821.487 3.53 1.338 5"
+			/>
+		</svg>
+	</div>
+</div>
+
+<style>
+	.animate-spin {
+		animation-duration: 0.5s;
+	}
+</style>

app/src/lib/components/custom-source/AddCustomSourceForm.svelte

@@ -1,5 +1,20 @@
-<script context="module">
+<script lang="ts" context="module">
 	const TEN_MB = 10 * 1024 * 1024;
+
+	function validFileType(file: File) {
+		// Supported file types and extensions
+		const supportedTypes = [
+			'application/pdf',
+			'application/vnd.openxmlformats-officedocument.wordprocessingml.document'
+		];
+		const supportedExtensions = ['.pdf', '.docx', '.txt'];
+
+		// Check MIME type
+		if (supportedTypes.includes(file.type)) return true;
+
+		// Fallback: Check file extension (case insensitive)
+		return supportedExtensions.some((ext) => file.name.toLowerCase().endsWith(ext));
+	}
 </script>
 
 <script lang="ts">
@@ -65,7 +80,7 @@
 				continue;
 			}
 
-			if (file.type !== 'application/pdf' && !file.name.endsWith('.txt')) {
+			if (!validFileType(file)) {
 				toast.info(`Unsupported file type for ${file.name}. Skipping...`);
 				continue;
 			}
@@ -113,14 +128,14 @@
 							type="file"
 							class="hidden"
 							multiple
-							accept=".pdf,.txt"
+							accept=".pdf,.txt,.docx"
 							bind:this={fileuploadEl}
 							on:change={(e) => handleFiles(e.currentTarget.files)}
 						/>
 					</p>
 				</div>
 				<p class="text-sm text-gray-500">
-					Supported file types: PDF, .txt, Markdown. (Max size: 10MB)
+					Supported file types: PDF, TXT, DOCX, Markdown. (Max size: 10MB)
 				</p>
 			</div>
 		</div>

app/src/lib/db/db.customSources.ts

@@ -19,7 +19,11 @@ export type UploadSource = {
 
 export type Sources = (LinkSource | CopyPasteSource | UploadSource) & {
 	id: string;
-	content_type: 'text/plain' | 'text/html' | 'application/pdf';
+	content_type:
+		| 'text/plain'
+		| 'text/html'
+		| 'application/pdf'
+		| 'application/vnd.openxmlformats-officedocument.wordprocessingml.document';
 	content: string;
 	title?: string;
 	created_at?: string;