add text embedding

'

@@ -0,0 +1,10 @@
+basic tests
+# Please enter the commit message for your changes. Lines starting
+# with '#' will be ignored, and an empty message aborts the commit.
+#
+# On branch text_view
+# Changes to be committed:
+#	modified:   backend/danswer/server/query_and_chat/query_backend.py
+#	modified:   backend/tests/integration/common_utils/managers/file.py
+#	modified:   backend/tests/integration/tests/file/test_file_detection.py
+#

backend/Dockerfile

@@ -73,6 +73,7 @@ RUN apt-get update && \
     rm -rf /var/lib/apt/lists/* && \
     rm -f /usr/local/lib/python3.11/site-packages/tornado/test/test.key
 
+
 # Pre-downloading models for setups with limited egress
 RUN python -c "from tokenizers import Tokenizer; \
 Tokenizer.from_pretrained('nomic-ai/nomic-embed-text-v1')"

backend/danswer/auth/users.py

@@ -147,7 +147,7 @@ def verify_email_is_invited(email: str) -> None:
     whitelist = get_invited_users()
     if not whitelist:
         return
-
+    print("whitelist is", whitelist)
     if not email:
         raise PermissionError("Email must be specified")
 

backend/danswer/document_index/vespa/index.py

@@ -295,6 +295,9 @@ def register_multitenant_indices(
 
         zip_file = in_memory_zip_from_file_bytes(zip_dict)
 
+        # Reset the file pointer to the beginning for the subsequent POST request
+        zip_file.seek(0)
+
         headers = {"Content-Type": "application/zip"}
         response = requests.post(deploy_url, headers=headers, data=zip_file)
 

backend/danswer/file_store/file_store.py

@@ -59,6 +59,12 @@ def read_file(
             Contents of the file and metadata dict
         """
 
+    @abstractmethod
+    def read_file_record(self, file_name: str) -> PGFileStore:
+        """
+        Read the file record by the name
+        """
+
     @abstractmethod
     def delete_file(self, file_name: str) -> None:
         """

backend/danswer/server/query_and_chat/chat_backend.py

@@ -1,6 +1,7 @@
 import asyncio
 import io
 import json
+import mimetypes
 import uuid
 from collections.abc import Callable
 from collections.abc import Generator
@@ -707,14 +708,24 @@ def upload_files_for_chat(
     }
 
 
-@router.get("/file/{file_id}")
+@router.get("/file/{file_id:path}")
 def fetch_chat_file(
     file_id: str,
     db_session: Session = Depends(get_session),
     _: User | None = Depends(current_user),
 ) -> Response:
     file_store = get_default_file_store(db_session)
+    file_record = file_store.read_file_record(file_id)
+    if not file_record:
+        raise HTTPException(status_code=404, detail="File not found")
+
+    media_type: str | None = None
+    if file_record.file_type:
+        media_type = file_record.file_type
+    else:
+        media_type, __ = mimetypes.guess_type(file_record.display_name)
+        if not media_type:
+            media_type = "application/octet-stream"
+
     file_io = file_store.read_file(file_id, mode="b")
-    # NOTE: specifying "image/jpeg" here, but it still works for pngs
-    # TODO: do this properly
-    return Response(content=file_io.read(), media_type="image/jpeg")
+    return StreamingResponse(file_io, media_type=media_type)

backend/shared_configs/configs.py

@@ -163,47 +163,92 @@ async def async_return_default_schema(*args: Any, **kwargs: Any) -> str:
         dim=1024,
         index_name="danswer_chunk_cohere_embed_english_v3_0",
     ),
+    SupportedEmbeddingModel(
+        name="cohere/embed-english-v3.0",
+        dim=1024,
+        index_name="danswer_chunk_embed_english_v3_0",
+    ),
     SupportedEmbeddingModel(
         name="cohere/embed-english-light-v3.0",
         dim=384,
         index_name="danswer_chunk_cohere_embed_english_light_v3_0",
     ),
+    SupportedEmbeddingModel(
+        name="cohere/embed-english-light-v3.0",
+        dim=384,
+        index_name="danswer_chunk_embed_english_light_v3_0",
+    ),
     SupportedEmbeddingModel(
         name="openai/text-embedding-3-large",
         dim=3072,
         index_name="danswer_chunk_openai_text_embedding_3_large",
     ),
+    SupportedEmbeddingModel(
+        name="openai/text-embedding-3-large",
+        dim=3072,
+        index_name="danswer_chunk_text_embedding_3_large",
+    ),
     SupportedEmbeddingModel(
         name="openai/text-embedding-3-small",
         dim=1536,
         index_name="danswer_chunk_openai_text_embedding_3_small",
     ),
+    SupportedEmbeddingModel(
+        name="openai/text-embedding-3-small",
+        dim=1536,
+        index_name="danswer_chunk_text_embedding_3_small",
+    ),
     SupportedEmbeddingModel(
         name="google/text-embedding-004",
         dim=768,
         index_name="danswer_chunk_google_text_embedding_004",
     ),
+    SupportedEmbeddingModel(
+        name="google/text-embedding-004",
+        dim=768,
+        index_name="danswer_chunk_text_embedding_004",
+    ),
     SupportedEmbeddingModel(
         name="google/textembedding-gecko@003",
         dim=768,
         index_name="danswer_chunk_google_textembedding_gecko_003",
     ),
+    SupportedEmbeddingModel(
+        name="google/textembedding-gecko@003",
+        dim=768,
+        index_name="danswer_chunk_textembedding_gecko_003",
+    ),
     SupportedEmbeddingModel(
         name="voyage/voyage-large-2-instruct",
         dim=1024,
         index_name="danswer_chunk_voyage_large_2_instruct",
     ),
+    SupportedEmbeddingModel(
+        name="voyage/voyage-large-2-instruct",
+        dim=1024,
+        index_name="danswer_chunk_large_2_instruct",
+    ),
     SupportedEmbeddingModel(
         name="voyage/voyage-light-2-instruct",
         dim=384,
         index_name="danswer_chunk_voyage_light_2_instruct",
     ),
+    SupportedEmbeddingModel(
+        name="voyage/voyage-light-2-instruct",
+        dim=384,
+        index_name="danswer_chunk_light_2_instruct",
+    ),
     # Self-hosted models
     SupportedEmbeddingModel(
         name="nomic-ai/nomic-embed-text-v1",
         dim=768,
         index_name="danswer_chunk_nomic_ai_nomic_embed_text_v1",
     ),
+    SupportedEmbeddingModel(
+        name="nomic-ai/nomic-embed-text-v1",
+        dim=768,
+        index_name="danswer_chunk_nomic_embed_text_v1",
+    ),
     SupportedEmbeddingModel(
         name="intfloat/e5-base-v2",
         dim=768,

backend/tests/integration/common_utils/managers/file.py

@@ -0,0 +1,62 @@
+import mimetypes
+from typing import cast
+from typing import IO
+from typing import List
+from typing import Tuple
+
+import requests
+
+from danswer.file_store.models import FileDescriptor
+from tests.integration.common_utils.constants import API_SERVER_URL
+from tests.integration.common_utils.constants import GENERAL_HEADERS
+from tests.integration.common_utils.test_models import DATestUser
+
+
+class FileManager:
+    @staticmethod
+    def upload_files(
+        files: List[Tuple[str, IO]],
+        user_performing_action: DATestUser | None = None,
+    ) -> Tuple[List[FileDescriptor], str]:
+        headers = (
+            user_performing_action.headers
+            if user_performing_action
+            else GENERAL_HEADERS
+        )
+        headers.pop("Content-Type", None)
+
+        files_param = []
+        for filename, file_obj in files:
+            mime_type, _ = mimetypes.guess_type(filename)
+            if mime_type is None:
+                mime_type = "application/octet-stream"
+            files_param.append(("files", (filename, file_obj, mime_type)))
+
+        response = requests.post(
+            f"{API_SERVER_URL}/chat/file",
+            files=files_param,
+            headers=headers,
+        )
+
+        if not response.ok:
+            return (
+                cast(List[FileDescriptor], []),
+                f"Failed to upload files - {response.json().get('detail', 'Unknown error')}",
+            )
+
+        response_json = response.json()
+        return response_json.get("files", cast(List[FileDescriptor], [])), ""
+
+    @staticmethod
+    def fetch_uploaded_file(
+        file_id: str,
+        user_performing_action: DATestUser | None = None,
+    ) -> bytes:
+        response = requests.get(
+            f"{API_SERVER_URL}/chat/file/{file_id}",
+            headers=user_performing_action.headers
+            if user_performing_action
+            else GENERAL_HEADERS,
+        )
+        response.raise_for_status()
+        return response.content

package.json

@@ -0,0 +1 @@
+{}

web/src/app/chat/ChatPage.tsx

@@ -106,8 +106,10 @@ import { NoAssistantModal } from "@/components/modals/NoAssistantModal";
 import { useAssistants } from "@/components/context/AssistantsContext";
 import { Separator } from "@/components/ui/separator";
 import AssistantBanner from "../../components/assistants/AssistantBanner";
+import TextView from "@/components/chat_search/TextView";
 import AssistantSelector from "@/components/chat_search/AssistantSelector";
 import { Modal } from "@/components/Modal";
+import { DocumentSidebar } from "./documentSidebar/DocumentSidebar";
 
 const TEMP_USER_MESSAGE_ID = -1;
 const TEMP_ASSISTANT_MESSAGE_ID = -2;
@@ -241,10 +243,10 @@ export function ChatPage({
           (assistant) => assistant.id === existingChatSessionAssistantId
         )
       : defaultAssistantId !== undefined
-        ? availableAssistants.find(
-            (assistant) => assistant.id === defaultAssistantId
-          )
-        : undefined
+      ? availableAssistants.find(
+          (assistant) => assistant.id === defaultAssistantId
+        )
+      : undefined
   );
   // Gather default temperature settings
   const search_param_temperature = searchParams.get(
@@ -254,12 +256,12 @@ export function ChatPage({
   const defaultTemperature = search_param_temperature
     ? parseFloat(search_param_temperature)
     : selectedAssistant?.tools.some(
-          (tool) =>
-            tool.in_code_tool_id === "SearchTool" ||
-            tool.in_code_tool_id === "InternetSearchTool"
-        )
-      ? 0
-      : 0.7;
+        (tool) =>
+          tool.in_code_tool_id === "SearchTool" ||
+          tool.in_code_tool_id === "InternetSearchTool"
+      )
+    ? 0
+    : 0.7;
 
   const setSelectedAssistantFromId = (assistantId: number) => {
     // NOTE: also intentionally look through available assistants here, so that
@@ -279,6 +281,9 @@ export function ChatPage({
   const [alternativeAssistant, setAlternativeAssistant] =
     useState<Persona | null>(null);
 
+  const [presentingDocument, setPresentingDocument] =
+    useState<DanswerDocument | null>(null);
+
   const {
     visibleAssistants: assistants,
     recentAssistants,
@@ -490,6 +495,7 @@ export function ChatPage({
           clientScrollToBottom(true);
         }
       }
+
       setIsFetchingChatMessages(false);
 
       // if this is a seeded chat, then kick off the AI message generation
@@ -1152,8 +1158,8 @@ export function ChatPage({
     const currentAssistantId = alternativeAssistantOverride
       ? alternativeAssistantOverride.id
       : alternativeAssistant
-        ? alternativeAssistant.id
-        : liveAssistant.id;
+      ? alternativeAssistant.id
+      : liveAssistant.id;
 
     resetInputBar();
     let messageUpdates: Message[] | null = null;
@@ -1649,7 +1655,6 @@ export function ChatPage({
     scrollDist,
     endDivRef,
     debounceNumber,
-    waitForScrollRef,
     mobile: settings?.isMobile,
     enableAutoScroll: autoScrollEnabled,
   });
@@ -1946,6 +1951,7 @@ export function ChatPage({
       {popup}
 
       <ChatPopup />
+
       {currentFeedback && (
         <FeedbackModal
           feedbackType={currentFeedback[0]}
@@ -2024,6 +2030,13 @@ export function ChatPage({
         />
       )}
 
+      {presentingDocument && (
+        <TextView
+          presentingDocument={presentingDocument}
+          onClose={() => setPresentingDocument(null)}
+        />
+      )}
+
       {stackTraceModalContent && (
         <ExceptionTraceModal
           onOutsideClick={() => setStackTraceModalContent(null)}
@@ -2424,6 +2437,9 @@ export function ChatPage({
                                     }
                                   >
                                     <AIMessage
+                                      setPresentingDocument={
+                                        setPresentingDocument
+                                      }
                                       index={i}
                                       selectedMessageForDocDisplay={
                                         selectedMessageForDocDisplay
@@ -2811,6 +2827,20 @@ export function ChatPage({
         </div>
         {/* Right Sidebar - DocumentSidebar */}
       </div>
+      <DocumentSidebar
+        initialWidth={350}
+        ref={innerSidebarElementRef}
+        closeSidebar={() => setDocumentSelection(false)}
+        selectedMessage={aiMessage}
+        selectedDocuments={selectedDocuments}
+        toggleDocumentSelection={toggleDocumentSelection}
+        clearSelectedDocuments={clearSelectedDocuments}
+        selectedDocumentTokens={selectedDocumentTokens}
+        maxTokens={maxTokens}
+        isLoading={isFetchingChatMessages}
+        isOpen={documentSelection}
+        setPresentingDocument={setPresentingDocument}
+      />
     </>
   );
 }