tdpeuter/open-webui
tdpeuter/open-webui
1
Fork 0
forked from open-webui/open-webui
Code Issues Pull requests 1 Projects 1 Releases 2 Packages Activity Actions

merged from main and added new translation keys

Browse source
This commit is contained in:
Jannik Streidl 2024-03-14 09:42:37 +01:00
commit aa3985e879
13 changed files with 368 additions and 172 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

18
CHANGELOG.md
View file

@ -5,6 +5,24 @@ All notable changes to this project will be documented in this file.
The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
## [0.1.111] - 2024-03-10
### Added
- 🛡️ **Model Whitelisting**: Admins now have the ability to whitelist models for users with the 'user' role.
- 🔄 **Update All Models**: Added a convenient button to update all models at once.
- 📄 **Toggle PDF OCR**: Users can now toggle PDF OCR option for improved parsing performance.
- 🎨 **DALL-E Integration**: Introduced DALL-E integration for image generation alongside automatic1111.
- 🛠️ **RAG API Refactoring**: Refactored RAG logic and exposed its API, with additional documentation to follow.
### Fixed
- 🔒 **Max Token Settings**: Added max token settings for anthropic/claude-3-sonnet-20240229 (Issue #1094).
- 🔧 **Misalignment Issue**: Corrected misalignment of Edit and Delete Icons when Chat Title is Empty (Issue #1104).
- 🔄 **Context Loss Fix**: Resolved RAG losing context on model response regeneration with Groq models via API key (Issue #1105).
- 📁 **File Handling Bug**: Addressed File Not Found Notification when Dropping a Conversation Element (Issue #1098).
- 🖱️ **Dragged File Styling**: Fixed dragged file layover styling issue.
## [0.1.110] - 2024-03-06
### Added

32
backend/apps/openai/main.py
View file

@ -179,20 +179,26 @@ def merge_models_lists(model_lists):
async def get_all_models():
print("get_all_models")
tasks = [
fetch_url(f"{url}/models", app.state.OPENAI_API_KEYS[idx])
for idx, url in enumerate(app.state.OPENAI_API_BASE_URLS)
]
responses = await asyncio.gather(*tasks)
responses = list(filter(lambda x: x is not None and "error" not in x, responses))
models = {
"data": merge_models_lists(
list(map(lambda response: response["data"], responses))
)
}
app.state.MODELS = {model["id"]: model for model in models["data"]}
return models
if len(app.state.OPENAI_API_KEYS) == 1 and app.state.OPENAI_API_KEYS[0] == "":
models = {"data": []}
else:
tasks = [
fetch_url(f"{url}/models", app.state.OPENAI_API_KEYS[idx])
for idx, url in enumerate(app.state.OPENAI_API_BASE_URLS)
]
responses = await asyncio.gather(*tasks)
responses = list(
filter(lambda x: x is not None and "error" not in x, responses)
)
models = {
"data": merge_models_lists(
list(map(lambda response: response["data"], responses))
)
}
app.state.MODELS = {model["id"]: model for model in models["data"]}
return models
@app.get("/models")

37
backend/apps/rag/main.py
View file

@ -77,6 +77,7 @@ from constants import ERROR_MESSAGES
app = FastAPI()
app.state.PDF_EXTRACT_IMAGES = False
app.state.CHUNK_SIZE = CHUNK_SIZE
app.state.CHUNK_OVERLAP = CHUNK_OVERLAP
app.state.RAG_TEMPLATE = RAG_TEMPLATE
@ -184,12 +185,15 @@ async def update_embedding_model(
}
@app.get("/chunk")
async def get_chunk_params(user=Depends(get_admin_user)):
@app.get("/config")
async def get_rag_config(user=Depends(get_admin_user)):
return {
"status": True,
"chunk_size": app.state.CHUNK_SIZE,
"chunk_overlap": app.state.CHUNK_OVERLAP,
"pdf_extract_images": app.state.PDF_EXTRACT_IMAGES,
"chunk": {
"chunk_size": app.state.CHUNK_SIZE,
"chunk_overlap": app.state.CHUNK_OVERLAP,
},
}
@ -198,17 +202,24 @@ class ChunkParamUpdateForm(BaseModel):
chunk_overlap: int
@app.post("/chunk/update")
async def update_chunk_params(
form_data: ChunkParamUpdateForm, user=Depends(get_admin_user)
):
app.state.CHUNK_SIZE = form_data.chunk_size
app.state.CHUNK_OVERLAP = form_data.chunk_overlap
class ConfigUpdateForm(BaseModel):
pdf_extract_images: bool
chunk: ChunkParamUpdateForm
@app.post("/config/update")
async def update_rag_config(form_data: ConfigUpdateForm, user=Depends(get_admin_user)):
app.state.PDF_EXTRACT_IMAGES = form_data.pdf_extract_images
app.state.CHUNK_SIZE = form_data.chunk.chunk_size
app.state.CHUNK_OVERLAP = form_data.chunk.chunk_overlap
return {
"status": True,
"chunk_size": app.state.CHUNK_SIZE,
"chunk_overlap": app.state.CHUNK_OVERLAP,
"pdf_extract_images": app.state.PDF_EXTRACT_IMAGES,
"chunk": {
"chunk_size": app.state.CHUNK_SIZE,
"chunk_overlap": app.state.CHUNK_OVERLAP,
},
}
@ -364,7 +375,7 @@ def get_loader(filename: str, file_content_type: str, file_path: str):
]
if file_ext == "pdf":
loader = PyPDFLoader(file_path, extract_images=True)
loader = PyPDFLoader(file_path, extract_images=app.state.PDF_EXTRACT_IMAGES)
elif file_ext == "csv":
loader = CSVLoader(file_path)
elif file_ext == "rst":

86
backend/apps/rag/utils.py
View file

@ -95,3 +95,89 @@ def rag_template(template: str, context: str, query: str):
template = re.sub(r"\[query\]", query, template)
return template
def rag_messages(docs, messages, template, k, embedding_function):
print(docs)
last_user_message_idx = None
for i in range(len(messages) - 1, -1, -1):
if messages[i]["role"] == "user":
last_user_message_idx = i
break
user_message = messages[last_user_message_idx]
if isinstance(user_message["content"], list):
# Handle list content input
content_type = "list"
query = ""
for content_item in user_message["content"]:
if content_item["type"] == "text":
query = content_item["text"]
break
elif isinstance(user_message["content"], str):
# Handle text content input
content_type = "text"
query = user_message["content"]
else:
# Fallback in case the input does not match expected types
content_type = None
query = ""
relevant_contexts = []
for doc in docs:
context = None
try:
if doc["type"] == "collection":
context = query_collection(
collection_names=doc["collection_names"],
query=query,
k=k,
embedding_function=embedding_function,
)
else:
context = query_doc(
collection_name=doc["collection_name"],
query=query,
k=k,
embedding_function=embedding_function,
)
except Exception as e:
print(e)
context = None
relevant_contexts.append(context)
context_string = ""
for context in relevant_contexts:
if context:
context_string += " ".join(context["documents"][0]) + "\n"
ra_content = rag_template(
template=template,
context=context_string,
query=query,
)
if content_type == "list":
new_content = []
for content_item in user_message["content"]:
if content_item["type"] == "text":
# Update the text item's content with ra_content
new_content.append({"type": "text", "text": ra_content})
else:
# Keep other types of content as they are
new_content.append(content_item)
new_user_message = {**user_message, "content": new_content}
else:
new_user_message = {
**user_message,
"content": ra_content,
}
messages[last_user_message_idx] = new_user_message
return messages

11
backend/config.py
View file

@ -209,10 +209,6 @@ OLLAMA_API_BASE_URL = os.environ.get(
OLLAMA_BASE_URL = os.environ.get("OLLAMA_BASE_URL", "")
if ENV == "prod":
if OLLAMA_BASE_URL == "/ollama":
OLLAMA_BASE_URL = "http://host.docker.internal:11434"
if OLLAMA_BASE_URL == "" and OLLAMA_API_BASE_URL != "":
OLLAMA_BASE_URL = (
@ -221,6 +217,11 @@ if OLLAMA_BASE_URL == "" and OLLAMA_API_BASE_URL != "":
else OLLAMA_API_BASE_URL
)
if ENV == "prod":
if OLLAMA_BASE_URL == "/ollama":
OLLAMA_BASE_URL = "http://host.docker.internal:11434"
OLLAMA_BASE_URLS = os.environ.get("OLLAMA_BASE_URLS", "")
OLLAMA_BASE_URLS = OLLAMA_BASE_URLS if OLLAMA_BASE_URLS != "" else OLLAMA_BASE_URL
@ -234,8 +235,6 @@ OLLAMA_BASE_URLS = [url.strip() for url in OLLAMA_BASE_URLS.split(";")]
OPENAI_API_KEY = os.environ.get("OPENAI_API_KEY", "")
OPENAI_API_BASE_URL = os.environ.get("OPENAI_API_BASE_URL", "")
if OPENAI_API_KEY == "":
OPENAI_API_KEY = "none"
if OPENAI_API_BASE_URL == "":
OPENAI_API_BASE_URL = "https://api.openai.com/v1"

1
backend/data/config.json
View file

@ -1,4 +1,5 @@
{
"version": "0.0.1",
"ui": {
"prompt_suggestions": [
{

132
backend/main.py
View file

@ -28,7 +28,7 @@ from typing import List
from utils.utils import get_admin_user
from apps.rag.utils import query_doc, query_collection, rag_template
from apps.rag.utils import rag_messages
from config import (
WEBUI_NAME,
@ -60,19 +60,6 @@ app.state.MODEL_FILTER_LIST = MODEL_FILTER_LIST
origins = ["*"]
app.add_middleware(
CORSMiddleware,
allow_origins=origins,
allow_credentials=True,
allow_methods=["*"],
allow_headers=["*"],
)
@app.on_event("startup")
async def on_startup():
await litellm_app_startup()
class RAGMiddleware(BaseHTTPMiddleware):
async def dispatch(self, request: Request, call_next):
@ -91,98 +78,33 @@ class RAGMiddleware(BaseHTTPMiddleware):
# Example: Add a new key-value pair or modify existing ones
# data["modified"] = True # Example modification
if "docs" in data:
docs = data["docs"]
print(docs)
last_user_message_idx = None
for i in range(len(data["messages"]) - 1, -1, -1):
if data["messages"][i]["role"] == "user":
last_user_message_idx = i
break
user_message = data["messages"][last_user_message_idx]
if isinstance(user_message["content"], list):
# Handle list content input
content_type = "list"
query = ""
for content_item in user_message["content"]:
if content_item["type"] == "text":
query = content_item["text"]
break
elif isinstance(user_message["content"], str):
# Handle text content input
content_type = "text"
query = user_message["content"]
else:
# Fallback in case the input does not match expected types
content_type = None
query = ""
relevant_contexts = []
for doc in docs:
context = None
try:
if doc["type"] == "collection":
context = query_collection(
collection_names=doc["collection_names"],
query=query,
k=rag_app.state.TOP_K,
embedding_function=rag_app.state.sentence_transformer_ef,
)
else:
context = query_doc(
collection_name=doc["collection_name"],
query=query,
k=rag_app.state.TOP_K,
embedding_function=rag_app.state.sentence_transformer_ef,
)
except Exception as e:
print(e)
context = None
relevant_contexts.append(context)
context_string = ""
for context in relevant_contexts:
if context:
context_string += " ".join(context["documents"][0]) + "\n"
ra_content = rag_template(
template=rag_app.state.RAG_TEMPLATE,
context=context_string,
query=query,
data = {**data}
data["messages"] = rag_messages(
data["docs"],
data["messages"],
rag_app.state.RAG_TEMPLATE,
rag_app.state.TOP_K,
rag_app.state.sentence_transformer_ef,
)
if content_type == "list":
new_content = []
for content_item in user_message["content"]:
if content_item["type"] == "text":
# Update the text item's content with ra_content
new_content.append({"type": "text", "text": ra_content})
else:
# Keep other types of content as they are
new_content.append(content_item)
new_user_message = {**user_message, "content": new_content}
else:
new_user_message = {
**user_message,
"content": ra_content,
}
data["messages"][last_user_message_idx] = new_user_message
del data["docs"]
print(data["messages"])
modified_body_bytes = json.dumps(data).encode("utf-8")
# Create a new request with the modified body
scope = request.scope
scope["body"] = modified_body_bytes
request = Request(scope, receive=lambda: self._receive(modified_body_bytes))
# Replace the request body with the modified one
request._body = modified_body_bytes
# Set custom header to ensure content-length matches new body length
request.headers.__dict__["_list"] = [
(b"content-length", str(len(modified_body_bytes)).encode("utf-8")),
*[
(k, v)
for k, v in request.headers.raw
if k.lower() != b"content-length"
],
]
response = await call_next(request)
return response
@ -194,6 +116,15 @@ class RAGMiddleware(BaseHTTPMiddleware):
app.add_middleware(RAGMiddleware)
app.add_middleware(
CORSMiddleware,
allow_origins=origins,
allow_credentials=True,
allow_methods=["*"],
allow_headers=["*"],
)
@app.middleware("http")
async def check_url(request: Request, call_next):
start_time = int(time.time())
@ -204,6 +135,11 @@ async def check_url(request: Request, call_next):
return response
@app.on_event("startup")
async def on_startup():
await litellm_app_startup()
app.mount("/api/v1", webui_app)
app.mount("/litellm/api", litellm_app)

2
package.json
View file

@ -1,6 +1,6 @@
{
"name": "open-webui",
"version": "0.1.110",
"version": "0.1.111",
"private": true,
"scripts": {
"dev": "vite dev --host",

21
src/lib/apis/rag/index.ts
View file

@ -1,9 +1,9 @@
import { RAG_API_BASE_URL } from '$lib/constants';
export const getChunkParams = async (token: string) => {
export const getRAGConfig = async (token: string) => {
let error = null;
const res = await fetch(`${RAG_API_BASE_URL}/chunk`, {
const res = await fetch(`${RAG_API_BASE_URL}/config`, {
method: 'GET',
headers: {
'Content-Type': 'application/json',
@ -27,18 +27,27 @@ export const getChunkParams = async (token: string) => {
return res;
};
export const updateChunkParams = async (token: string, size: number, overlap: number) => {
type ChunkConfigForm = {
chunk_size: number;
chunk_overlap: number;
};
type RAGConfigForm = {
pdf_extract_images: boolean;
chunk: ChunkConfigForm;
};
export const updateRAGConfig = async (token: string, payload: RAGConfigForm) => {
let error = null;
const res = await fetch(`${RAG_API_BASE_URL}/chunk/update`, {
const res = await fetch(`${RAG_API_BASE_URL}/config/update`, {
method: 'POST',
headers: {
'Content-Type': 'application/json',
Authorization: `Bearer ${token}`
},
body: JSON.stringify({
chunk_size: size,
chunk_overlap: overlap
...payload
})
})
.then(async (res) => {

127
src/lib/components/chat/Settings/Models.svelte
View file

@ -14,6 +14,7 @@
import { splitStream } from '$lib/utils';
import { onMount, getContext } from 'svelte';
import { addLiteLLMModel, deleteLiteLLMModel, getLiteLLMModelInfo } from '$lib/apis/litellm';
import Tooltip from '$lib/components/common/Tooltip.svelte';
const i18n = getContext('i18n');
@ -39,6 +40,10 @@
let OLLAMA_URLS = [];
let selectedOllamaUrlIdx: string | null = null;
let updateModelId = null;
let updateProgress = null;
let showExperimentalOllama = false;
let ollamaVersion = '';
const MAX_PARALLEL_DOWNLOADS = 3;
@ -63,6 +68,71 @@
let deleteModelTag = '';
const updateModelsHandler = async () => {
for (const model of $models.filter(
(m) =>
m.size != null &&
(selectedOllamaUrlIdx === null ? true : (m?.urls ?? []).includes(selectedOllamaUrlIdx))
)) {
console.log(model);
updateModelId = model.id;
const res = await pullModel(localStorage.token, model.id, selectedOllamaUrlIdx).catch(
(error) => {
toast.error(error);
return null;
}
);
if (res) {
const reader = res.body
.pipeThrough(new TextDecoderStream())
.pipeThrough(splitStream('\n'))
.getReader();
while (true) {
try {
const { value, done } = await reader.read();
if (done) break;
let lines = value.split('\n');
for (const line of lines) {
if (line !== '') {
let data = JSON.parse(line);
console.log(data);
if (data.error) {
throw data.error;
}
if (data.detail) {
throw data.detail;
}
if (data.status) {
if (data.digest) {
updateProgress = 0;
if (data.completed) {
updateProgress = Math.round((data.completed / data.total) * 1000) / 10;
} else {
updateProgress = 100;
}
} else {
toast.success(data.status);
}
}
}
}
} catch (error) {
console.log(error);
}
}
}
}
updateModelId = null;
updateProgress = null;
};
const pullModelHandler = async () => {
const sanitizedModelTag = modelTag.trim();
if (modelDownloadStatus[sanitizedModelTag]) {
@ -389,7 +459,7 @@
return [];
});
if (OLLAMA_URLS.length > 1) {
if (OLLAMA_URLS.length > 0) {
selectedOllamaUrlIdx = 0;
}
@ -404,18 +474,51 @@
<div class="space-y-2 pr-1.5">
<div class="text-sm font-medium">{$i18n.t('Manage Ollama Models')}</div>
{#if OLLAMA_URLS.length > 1}
<div class="flex-1 pb-1">
<select
class="w-full rounded-lg py-2 px-4 text-sm dark:text-gray-300 dark:bg-gray-850 outline-none"
bind:value={selectedOllamaUrlIdx}
placeholder={$i18n.t('Select an Ollama instance')}
>
{#each OLLAMA_URLS as url, idx}
<option value={idx} class="bg-gray-100 dark:bg-gray-700">{url}</option>
{/each}
</select>
{#if OLLAMA_URLS.length > 0}
<div class="flex gap-2">
<div class="flex-1 pb-1">
<select
class="w-full rounded-lg py-2 px-4 text-sm dark:text-gray-300 dark:bg-gray-850 outline-none"
bind:value={selectedOllamaUrlIdx}
placeholder="Select an Ollama instance"
>
{#each OLLAMA_URLS as url, idx}
<option value={idx} class="bg-gray-100 dark:bg-gray-700">{url}</option>
{/each}
</select>
</div>
<div>
<div class="flex w-full justify-end">
<Tooltip content="Update All Models" placement="top">
<button
class="p-2.5 flex gap-2 items-center bg-gray-100 hover:bg-gray-200 text-gray-800 dark:bg-gray-850 dark:hover:bg-gray-800 dark:text-gray-100 rounded-lg transition"
on:click={() => {
updateModelsHandler();
}}
>
<svg
xmlns="http://www.w3.org/2000/svg"
viewBox="0 0 16 16"
fill="currentColor"
class="w-4 h-4"
>
<path
d="M7 1a.75.75 0 0 1 .75.75V6h-1.5V1.75A.75.75 0 0 1 7 1ZM6.25 6v2.94L5.03 7.72a.75.75 0 0 0-1.06 1.06l2.5 2.5a.75.75 0 0 0 1.06 0l2.5-2.5a.75.75 0 1 0-1.06-1.06L7.75 8.94V6H10a2 2 0 0 1 2 2v3a2 2 0 0 1-2 2H4a2 2 0 0 1-2-2V8a2 2 0 0 1 2-2h2.25Z"
/>
<path
d="M4.268 14A2 2 0 0 0 6 15h6a2 2 0 0 0 2-2v-3a2 2 0 0 0-1-1.732V11a3 3 0 0 1-3 3H4.268Z"
/>
</svg>
</button>
</Tooltip>
</div>
</div>
</div>
{#if updateModelId}
Updating "{updateModelId}" {updateProgress ? `(${updateProgress}%)` : ''}
{/if}
{/if}
<div class="space-y-2">

63
src/lib/components/documents/Settings/General.svelte
View file

@ -1,10 +1,10 @@
<script lang="ts">
import { getDocs } from '$lib/apis/documents';
import {
getChunkParams,
getRAGConfig,
updateRAGConfig,
getQuerySettings,
scanDocs,
updateChunkParams,
updateQuerySettings
} from '$lib/apis/rag';
import { documents } from '$lib/stores';
@ -19,6 +19,7 @@
let chunkSize = 0;
let chunkOverlap = 0;
let pdfExtractImages = true;
let querySettings = {
template: '',
@ -37,16 +38,24 @@
};
const submitHandler = async () => {
const res = await updateChunkParams(localStorage.token, chunkSize, chunkOverlap);
const res = await updateRAGConfig(localStorage.token, {
pdf_extract_images: pdfExtractImages,
chunk: {
chunk_overlap: chunkOverlap,
chunk_size: chunkSize
}
});
querySettings = await updateQuerySettings(localStorage.token, querySettings);
};
onMount(async () => {
const res = await getChunkParams(localStorage.token);
const res = await getRAGConfig(localStorage.token);
if (res) {
chunkSize = res.chunk_size;
chunkOverlap = res.chunk_overlap;
pdfExtractImages = res.pdf_extract_images;
chunkSize = res.chunk.chunk_size;
chunkOverlap = res.chunk.chunk_overlap;
}
querySettings = await getQuerySettings(localStorage.token);
@ -163,6 +172,22 @@
</div>
</div>
<div>
<div class="flex justify-between items-center text-xs">
<div class=" text-xs font-medium">{$i18n.t('PDF Extract Images (OCR)')}</div>
<button
class=" text-xs font-medium text-gray-500"
type="button"
on:click={() => {
pdfExtractImages = !pdfExtractImages;
}}>{pdfExtractImages ? $i18n.t('On') : $i18n.t('Off')}</button
>
</div>
</div>
</div>
<div>
<div class=" text-sm font-medium">{$i18n.t('Query Params')}</div>
<div class=" flex">
@ -182,19 +207,19 @@
</div>
<!-- <div class="flex w-full">
<div class=" self-center text-xs font-medium min-w-fit">Chunk Overlap</div>
<div class="self-center p-3">
<input
class="w-full rounded py-1.5 px-4 text-sm dark:text-gray-300 dark:bg-gray-800 outline-none border border-gray-100 dark:border-gray-600"
type="number"
placeholder="Enter Chunk Overlap"
bind:value={chunkOverlap}
autocomplete="off"
min="0"
/>
</div>
</div> -->
<div class=" self-center text-xs font-medium min-w-fit">Chunk Overlap</div>
<div class="self-center p-3">
<input
class="w-full rounded py-1.5 px-4 text-sm dark:text-gray-300 dark:bg-gray-800 outline-none border border-gray-100 dark:border-gray-600"
type="number"
placeholder="Enter Chunk Overlap"
bind:value={chunkOverlap}
autocomplete="off"
min="0"
/>
</div>
</div> -->
</div>
<div>

5
src/lib/i18n/locales/de/translation.json
View file

@ -236,7 +236,7 @@
"Prompts": "Prompts",
"Pull a model from Ollama.com": "Ein Modell von Ollama.com abrufen",
"Pull Progress": "Fortschritt abrufen",
"Query Params": "",
"Query Params": "Query Parameter",
"RAG Template": "RAG-Vorlage",
"Raw Format": "Rohformat",
"Record voice": "Stimme aufnehmen",
@ -346,5 +346,6 @@
"Write a summary in 50 words that summarizes [topic or keyword].": "Schreibe eine kurze Zusammenfassung in 50 Wörtern, die [Thema oder Schlüsselwort] zusammenfasst.",
"You": "Du",
"You're a helpful assistant.": "Du bist ein hilfreicher Assistent.",
"You're now logged in.": "Du bist nun eingeloggt."
"You're now logged in.": "Du bist nun eingeloggt.",
"PDF Extract Images (OCR)": "Text von Bilder aus PDFs extrahieren (OCR)"
}

5
src/lib/i18n/locales/en/translation.json
View file

@ -236,7 +236,7 @@
"Prompts": "Prompts",
"Pull a model from Ollama.com": "Pull a model from Ollama.com",
"Pull Progress": "Pull Progress",
"Query Params": "",
"Query Params": "Query Params",
"RAG Template": "RAG Template",
"Raw Format": "Raw Format",
"Record voice": "Record voice",
@ -346,5 +346,6 @@
"Write a summary in 50 words that summarizes [topic or keyword].": "Write a summary in 50 words that summarizes [topic or keyword].",
"You": "You",
"You're a helpful assistant.": "You're a helpful assistant.",
"You're now logged in.": "You're now logged in."
"You're now logged in.": "You're now logged in.",
"PDF Extract Images (OCR)": "PDF Extract Images (OCR)"
}