diff --git a/.dockerignore b/.dockerignore index f626766d..fd026e17 100644 --- a/.dockerignore +++ b/.dockerignore @@ -1,7 +1,6 @@ .venv .env .next -node_modules out build .git @@ -9,4 +8,6 @@ build tmp debug .fastembed_cache +servers/fastapi/tmp +servers/fastapi/debug servers/nextjs/node_modules \ No newline at end of file diff --git a/.gitignore b/.gitignore index 81923a11..33c06363 100644 --- a/.gitignore +++ b/.gitignore @@ -11,4 +11,5 @@ app_data tmp debug .fastembed_cache -my-doc.txt \ No newline at end of file +my-doc.txt +generated_models \ No newline at end of file diff --git a/Dockerfile b/Dockerfile index 9e2d6a5c..bb2193f1 100644 --- a/Dockerfile +++ b/Dockerfile @@ -6,7 +6,7 @@ RUN apt-get update && apt-get install -y \ curl \ redis-server - # Install Node.js 20 using NodeSource repository +# Install Node.js 20 using NodeSource repository RUN curl -fsSL https://deb.nodesource.com/setup_20.x | bash - && \ apt-get install -y nodejs @@ -45,18 +45,15 @@ RUN npm run build WORKDIR /app -# Copy FastAPI and start script +# Copy FastAPI COPY servers/fastapi/ ./servers/fastapi/ COPY start.js LICENSE NOTICE ./ # Copy nginx configuration COPY nginx.conf /etc/nginx/nginx.conf -# Copy start script -COPY docker-start.sh /app/docker-start.sh - # Expose the port EXPOSE 80 # Start the servers -CMD ["/bin/bash", "/app/docker-start.sh"] \ No newline at end of file +CMD ["node", "/app/start.js"] \ No newline at end of file diff --git a/Dockerfile.dev b/Dockerfile.dev index 65882ea4..7c843232 100644 --- a/Dockerfile.dev +++ b/Dockerfile.dev @@ -7,9 +7,9 @@ RUN apt-get update && apt-get install -y \ redis-server - # Install Node.js 20 using NodeSource repository +# Install Node.js 20 using NodeSource repository RUN curl -fsSL https://deb.nodesource.com/setup_20.x | bash - && \ - apt-get install -y nodejs + apt-get install -y nodejs # Change working directory @@ -47,4 +47,4 @@ COPY nginx.conf /etc/nginx/nginx.conf EXPOSE 80 # Start the servers -CMD ["/bin/bash", "/app/docker-dev-start.sh"] +CMD ["node", "/app/start.js", "--dev"] diff --git a/README.md b/README.md index 4daae4da..ec4ee82a 100644 --- a/README.md +++ b/README.md @@ -33,17 +33,22 @@ ## ✨ More Freedom with AI Presentations -* ✅ **Custom layouts/templates/themes** — Create custom layouts with HTML and Tailwind, support any presentation design -* ✅ **Bring Your Own Key** — Only pay for what you use. OpenAI, Gemini (More coming soon...) -* ✅ **API Presentation Generation** — Host as API to generate presentations over requests -* ✅ **Ollama Support** — Run open-source models locally with Ollama integration -* ✅ **OpenAI API Compatibility** — Use any OpenAI-compatible API endpoint with your own models -* ✅ **Versatile Image Generation** — Choose from DALL-E 3, Gemini Flash, Pexels, or Pixabay for your visuals -* ✅ **Runs Locally** — All code runs on your device -* ✅ **Privacy-First** — No tracking, no data stored by us -* ✅ **Flexible** — Generate presentations from prompts or outlines -* ✅ **Export Ready** — Save as PowerPoint (PPTX) and PDF -* ✅ **Fully Open-Source** — Apache 2.0 licensed +Presenton gives you complete control over your AI presentation workflow. Choose your models, customize your experience, and keep your data private. + +* ✅ **Custom Layouts & Themes** — Create unlimited presentation designs with HTML and Tailwind CSS +* ✅ **Flexible Generation** — Build presentations from prompts or uploaded documents +* ✅ **Export Ready** — Save as PowerPoint (PPTX) and PDF with professional formatting +* ✅ **Bring Your Own Key** — Use your own API keys for OpenAI, Google Gemini, Anthropic Claude, or any compatible provider. Only pay for what you use, no hidden fees or subscriptions. +* ✅ **Ollama Integration** — Run open-source models locally with full privacy +* ✅ **OpenAI API Compatible** — Connect to any OpenAI-compatible endpoint with your own models +* ✅ **Multi-Provider Support** — Mix and match text and image generation providers +* ✅ **Versatile Image Generation** — Choose from DALL-E 3, Gemini Flash, Pexels, or Pixabay +* ✅ **Rich Media Support** — Icons, charts, and custom graphics for professional presentations +* ✅ **Runs Locally** — All processing happens on your device, no cloud dependencies +* ✅ **Privacy-First** — Zero tracking, no data stored by us, complete data sovereignty +* ✅ **API Deployment** — Host as your own API service for your team +* ✅ **Fully Open-Source** — Apache 2.0 licensed, inspect, modify, and contribute +* ✅ **Docker Ready** — One-command deployment with GPU support for local models ## Deploy on Cloud (one click deployment) [![Deploy on Railway](https://railway.com/button.svg)](https://railway.com/deploy/presenton-ai-presentations?referralCode=ubp0kk) @@ -72,9 +77,13 @@ Open http://localhost:5000 on browser of your choice to use Presenton. You may want to directly provide your API KEYS as environment variables and keep them hidden. You can set these environment variables to achieve it. - **CAN_CHANGE_KEYS=[true/false]**: Set this to **false** if you want to keep API Keys hidden and make them unmodifiable. -- **LLM=[openai/google/ollama/custom]**: Select **LLM** of your choice. +- **LLM=[openai/google/anthropic/ollama/custom]**: Select **LLM** of your choice. - **OPENAI_API_KEY=[Your OpenAI API Key]**: Provide this if **LLM** is set to **openai** +- **OPENAI_MODEL=[OpenAI Model ID]**: Provide this if **LLM** is set to **openai** (default: "gpt-4.1") - **GOOGLE_API_KEY=[Your Google API Key]**: Provide this if **LLM** is set to **google** +- **GOOGLE_MODEL=[Google Model ID]**: Provide this if **LLM** is set to **google** (default: "models/gemini-2.0-flash") +- **ANTHROPIC_API_KEY=[Your Anthropic API Key]**: Provide this if **LLM** is set to **anthropic** +- **ANTHROPIC_MODEL=[Anthropic Model ID]**: Provide this if **LLM** is set to **anthropic** (default: "claude-3-5-sonnet-20241022") - **OLLAMA_URL=[Custom Ollama URL]**: Provide this if you want to custom Ollama URL and **LLM** is set to **ollama** - **OLLAMA_MODEL=[Ollama Model ID]**: Provide this if **LLM** is set to **ollama** - **CUSTOM_LLM_URL=[Custom OpenAI Compatible URL]**: Provide this if **LLM** is set to **custom** @@ -97,11 +106,21 @@ You can also set the following environment variables to customize the image gene docker run -it --name presenton -p 5000:80 -e LLM="openai" -e OPENAI_API_KEY="******" -e IMAGE_PROVIDER="dall-e-3" -e CAN_CHANGE_KEYS="false" -v "./app_data:/app_data" ghcr.io/presenton/presenton:latest ``` +### Using Google +```bash +docker run -it --name presenton -p 5000:80 -e LLM="google" -e GOOGLE_API_KEY="******" -e IMAGE_PROVIDER="gemini_flash" -e CAN_CHANGE_KEYS="false" -v "./app_data:/app_data" ghcr.io/presenton/presenton:latest +``` + ### Using Ollama ```bash docker run -it --name presenton -p 5000:80 -e LLM="ollama" -e OLLAMA_MODEL="llama3.2:3b" -e IMAGE_PROVIDER="pexels" -e PEXELS_API_KEY="*******" -e CAN_CHANGE_KEYS="false" -v "./app_data:/app_data" ghcr.io/presenton/presenton:latest ``` +### Using Anthropic +```bash +docker run -it --name presenton -p 5000:80 -e LLM="anthropic" -e ANTHROPIC_API_KEY="******" -e IMAGE_PROVIDER="pexels" -e PEXELS_API_KEY="******" -e CAN_CHANGE_KEYS="false" -v "./app_data:/app_data" ghcr.io/presenton/presenton:latest +``` + ### Using OpenAI Compatible API ```bash docker run -it -p 5000:80 -e CAN_CHANGE_KEYS="false" -e LLM="custom" -e CUSTOM_LLM_URL="http://*****" -e CUSTOM_LLM_API_KEY="*****" -e CUSTOM_MODEL="llama3.2:3b" -e IMAGE_PROVIDER="pexels" -e PEXELS_API_KEY="********" -v "./app_data:/app_data" ghcr.io/presenton/presenton:latest diff --git a/docker-compose.yml b/docker-compose.yml index 64798520..6a0dfbc2 100644 --- a/docker-compose.yml +++ b/docker-compose.yml @@ -13,13 +13,18 @@ services: - CAN_CHANGE_KEYS=${CAN_CHANGE_KEYS} - LLM=${LLM} - OPENAI_API_KEY=${OPENAI_API_KEY} + - OPENAI_MODEL=${OPENAI_MODEL} - GOOGLE_API_KEY=${GOOGLE_API_KEY} + - GOOGLE_MODEL=${GOOGLE_MODEL} + - ANTHROPIC_API_KEY=${ANTHROPIC_API_KEY} + - ANTHROPIC_MODEL=${ANTHROPIC_MODEL} - OLLAMA_URL=${OLLAMA_URL} - OLLAMA_MODEL=${OLLAMA_MODEL} - CUSTOM_LLM_URL=${CUSTOM_LLM_URL} - CUSTOM_LLM_API_KEY=${CUSTOM_LLM_API_KEY} - CUSTOM_MODEL=${CUSTOM_MODEL} - PEXELS_API_KEY=${PEXELS_API_KEY} + - EXTENDED_REASONING=${EXTENDED_REASONING} - DATABASE_URL=${DATABASE_URL} production-gpu: @@ -43,13 +48,18 @@ services: - CAN_CHANGE_KEYS=${CAN_CHANGE_KEYS} - LLM=${LLM} - OPENAI_API_KEY=${OPENAI_API_KEY} + - OPENAI_MODEL=${OPENAI_MODEL} - GOOGLE_API_KEY=${GOOGLE_API_KEY} + - GOOGLE_MODEL=${GOOGLE_MODEL} + - ANTHROPIC_API_KEY=${ANTHROPIC_API_KEY} + - ANTHROPIC_MODEL=${ANTHROPIC_MODEL} - OLLAMA_URL=${OLLAMA_URL} - OLLAMA_MODEL=${OLLAMA_MODEL} - CUSTOM_LLM_URL=${CUSTOM_LLM_URL} - CUSTOM_LLM_API_KEY=${CUSTOM_LLM_API_KEY} - CUSTOM_MODEL=${CUSTOM_MODEL} - PEXELS_API_KEY=${PEXELS_API_KEY} + - EXTENDED_REASONING=${EXTENDED_REASONING} - DATABASE_URL=${DATABASE_URL} development: @@ -62,17 +72,21 @@ services: - .:/app - ./app_data:/app_data environment: - - NODE_ENV=development - CAN_CHANGE_KEYS=${CAN_CHANGE_KEYS} - LLM=${LLM} - OPENAI_API_KEY=${OPENAI_API_KEY} + - OPENAI_MODEL=${OPENAI_MODEL} - GOOGLE_API_KEY=${GOOGLE_API_KEY} + - GOOGLE_MODEL=${GOOGLE_MODEL} + - ANTHROPIC_API_KEY=${ANTHROPIC_API_KEY} + - ANTHROPIC_MODEL=${ANTHROPIC_MODEL} - OLLAMA_URL=${OLLAMA_URL} - OLLAMA_MODEL=${OLLAMA_MODEL} - CUSTOM_LLM_URL=${CUSTOM_LLM_URL} - CUSTOM_LLM_API_KEY=${CUSTOM_LLM_API_KEY} - CUSTOM_MODEL=${CUSTOM_MODEL} - PEXELS_API_KEY=${PEXELS_API_KEY} + - EXTENDED_REASONING=${EXTENDED_REASONING} - DATABASE_URL=${DATABASE_URL} development-gpu: @@ -92,15 +106,19 @@ services: - .:/app - ./app_data:/app_data environment: - - NODE_ENV=development - CAN_CHANGE_KEYS=${CAN_CHANGE_KEYS} - LLM=${LLM} - OPENAI_API_KEY=${OPENAI_API_KEY} + - OPENAI_MODEL=${OPENAI_MODEL} - GOOGLE_API_KEY=${GOOGLE_API_KEY} + - GOOGLE_MODEL=${GOOGLE_MODEL} + - ANTHROPIC_API_KEY=${ANTHROPIC_API_KEY} + - ANTHROPIC_MODEL=${ANTHROPIC_MODEL} - OLLAMA_URL=${OLLAMA_URL} - OLLAMA_MODEL=${OLLAMA_MODEL} - CUSTOM_LLM_URL=${CUSTOM_LLM_URL} - CUSTOM_LLM_API_KEY=${CUSTOM_LLM_API_KEY} - CUSTOM_MODEL=${CUSTOM_MODEL} - PEXELS_API_KEY=${PEXELS_API_KEY} + - EXTENDED_REASONING=${EXTENDED_REASONING} - DATABASE_URL=${DATABASE_URL} \ No newline at end of file diff --git a/docker-dev-start.sh b/docker-dev-start.sh deleted file mode 100644 index bd6ae2ad..00000000 --- a/docker-dev-start.sh +++ /dev/null @@ -1,13 +0,0 @@ -#!/bin/bash - -echo "Starting development server..." - -if [ -d "/node_dependencies/node_modules" ]; then - rm -rf /app/servers/nextjs/node_modules - mv /node_dependencies/node_modules /app/servers/nextjs -fi - -ollama serve & -service nginx start -service redis-server start -node /app/start.js diff --git a/docker-start.sh b/docker-start.sh deleted file mode 100644 index 22c80b91..00000000 --- a/docker-start.sh +++ /dev/null @@ -1,8 +0,0 @@ -#!/bin/bash - -echo "Starting production server..." - -ollama serve & -service nginx start -service redis-server start -node /app/start.js diff --git a/servers/fastapi/api/main.py b/servers/fastapi/api/main.py index 93517b12..80eea709 100644 --- a/servers/fastapi/api/main.py +++ b/servers/fastapi/api/main.py @@ -1,10 +1,8 @@ from fastapi import FastAPI from fastapi.middleware.cors import CORSMiddleware -from fastapi.staticfiles import StaticFiles from api.lifespan import app_lifespan from api.middlewares import UserConfigEnvUpdateMiddleware from api.v1.ppt.router import API_V1_PPT_ROUTER -from utils.asset_directory_utils import get_exports_directory, get_images_directory, get_uploads_directory app = FastAPI(lifespan=app_lifespan) @@ -13,25 +11,6 @@ app = FastAPI(lifespan=app_lifespan) # Routers app.include_router(API_V1_PPT_ROUTER) -# Static files -app.mount("/static", StaticFiles(directory="static"), name="static") -app.mount( - "/app_data/images", - StaticFiles(directory=get_images_directory()), - name="app_data/images", -) -app.mount( - "/app_data/exports", - StaticFiles(directory=get_exports_directory()), - name="app_data/exports", -) -app.mount( - "/app_data/uploads", - StaticFiles(directory=get_uploads_directory()), - name="app_data/uploads", -) - - # Middlewares origins = ["*"] app.add_middleware( diff --git a/servers/fastapi/api/v1/ppt/endpoints/anthropic.py b/servers/fastapi/api/v1/ppt/endpoints/anthropic.py new file mode 100644 index 00000000..e3e1d569 --- /dev/null +++ b/servers/fastapi/api/v1/ppt/endpoints/anthropic.py @@ -0,0 +1,16 @@ +from typing import Annotated, List +from fastapi import APIRouter, Body, HTTPException + +from utils.available_models import list_available_anthropic_models + +ANTHROPIC_ROUTER = APIRouter(prefix="/anthropic", tags=["Anthropic"]) + + +@ANTHROPIC_ROUTER.post("/models/available", response_model=List[str]) +async def get_available_models( + api_key: Annotated[str, Body(embed=True)], +): + try: + return await list_available_anthropic_models(api_key) + except Exception as e: + raise HTTPException(status_code=500, detail=str(e)) diff --git a/servers/fastapi/api/v1/ppt/endpoints/custom_llm.py b/servers/fastapi/api/v1/ppt/endpoints/custom_llm.py deleted file mode 100644 index 8a44cb22..00000000 --- a/servers/fastapi/api/v1/ppt/endpoints/custom_llm.py +++ /dev/null @@ -1,14 +0,0 @@ -from typing import Annotated, List, Optional -from fastapi import APIRouter, Body - -from utils.custom_llm_provider import list_available_custom_models - -CUSTOM_LLM_ROUTER = APIRouter(prefix="/custom_llm", tags=["Custom LLM"]) - - -@CUSTOM_LLM_ROUTER.post("/models/available", response_model=List[str]) -async def get_available_models( - url: Annotated[Optional[str], Body()] = None, - api_key: Annotated[Optional[str], Body()] = None, -): - return await list_available_custom_models(url, api_key) diff --git a/servers/fastapi/api/v1/ppt/endpoints/google.py b/servers/fastapi/api/v1/ppt/endpoints/google.py new file mode 100644 index 00000000..4c83627f --- /dev/null +++ b/servers/fastapi/api/v1/ppt/endpoints/google.py @@ -0,0 +1,14 @@ +from typing import Annotated, List +from fastapi import APIRouter, Body, HTTPException + +from utils.available_models import list_available_google_models + +GOOGLE_ROUTER = APIRouter(prefix="/google", tags=["Google"]) + + +@GOOGLE_ROUTER.post("/models/available", response_model=List[str]) +async def get_available_models(api_key: Annotated[str, Body(embed=True)]): + try: + return await list_available_google_models(api_key) + except Exception as e: + raise HTTPException(status_code=500, detail=str(e)) diff --git a/servers/fastapi/api/v1/ppt/endpoints/openai.py b/servers/fastapi/api/v1/ppt/endpoints/openai.py new file mode 100644 index 00000000..8e2f00e6 --- /dev/null +++ b/servers/fastapi/api/v1/ppt/endpoints/openai.py @@ -0,0 +1,17 @@ +from typing import Annotated, List +from fastapi import APIRouter, Body, HTTPException + +from utils.available_models import list_available_openai_compatible_models + +OPENAI_ROUTER = APIRouter(prefix="/openai", tags=["OpenAI"]) + + +@OPENAI_ROUTER.post("/models/available", response_model=List[str]) +async def get_available_models( + url: Annotated[str, Body()], + api_key: Annotated[str, Body()], +): + try: + return await list_available_openai_compatible_models(url, api_key) + except Exception as e: + raise HTTPException(status_code=500, detail=str(e)) diff --git a/servers/fastapi/api/v1/ppt/endpoints/outlines.py b/servers/fastapi/api/v1/ppt/endpoints/outlines.py index fa1e45bd..b5ff9f5a 100644 --- a/servers/fastapi/api/v1/ppt/endpoints/outlines.py +++ b/servers/fastapi/api/v1/ppt/endpoints/outlines.py @@ -54,7 +54,6 @@ async def stream_outlines( presentation.outlines = [ each.model_dump() for each in presentation_content.slides ] - presentation.notes = presentation_content.notes sql_session.add(presentation) await sql_session.commit() diff --git a/servers/fastapi/api/v1/ppt/endpoints/presentation.py b/servers/fastapi/api/v1/ppt/endpoints/presentation.py index f7fa7865..d06eecb4 100644 --- a/servers/fastapi/api/v1/ppt/endpoints/presentation.py +++ b/servers/fastapi/api/v1/ppt/endpoints/presentation.py @@ -2,10 +2,11 @@ import asyncio import json import os import random +import importlib from typing import Annotated, List, Literal, Optional from fastapi import APIRouter, Body, Depends, File, HTTPException, UploadFile from fastapi.responses import StreamingResponse -from sqlalchemy import String, cast, delete +from sqlalchemy import delete from sqlalchemy.ext.asyncio import AsyncSession from sqlmodel import select from constants.documents import UPLOAD_ACCEPTED_FILE_TYPES @@ -19,7 +20,7 @@ from models.pptx_models import PptxPresentationModel from models.presentation_layout import PresentationLayoutModel from models.presentation_structure_model import PresentationStructureModel from models.presentation_with_slides import PresentationWithSlides -from services.get_layout_by_name import get_layout_by_name +from utils.get_layout_by_name import get_layout_by_name from services.icon_finder_service import IconFinderService from services.image_generation_service import ImageGenerationService from utils.dict_utils import deep_update @@ -217,9 +218,11 @@ async def stream_presentation( ).to_string() for i, slide_layout_index in enumerate(structure.slides): slide_layout = layout.slides[slide_layout_index] + slide_content = await get_slide_content_from_type_and_outline( slide_layout, outline.slides[i], presentation.language ) + slide = SlideModel( presentation=presentation_id, layout_group=layout.name, @@ -236,9 +239,6 @@ async def stream_presentation( ) ) - # Give control to the event loop - await asyncio.sleep(0) - yield SSEResponse( event="response", data=json.dumps({"type": "chunk", "chunk": slide.model_dump_json()}), @@ -377,7 +377,6 @@ async def generate_presentation_api( presentation_outline=PresentationOutlineModel( title=presentation_content.title, slides=outlines, - notes=presentation_content.notes, ), presentation_layout=layout_model, ) @@ -401,7 +400,6 @@ async def generate_presentation_api( title=presentation_content.title, summary=summary, outlines=[each.model_dump() for each in outlines], - notes=presentation_content.notes, layout=layout_model.model_dump(), structure=presentation_structure.model_dump(), ) @@ -475,7 +473,6 @@ async def from_template( new_slide_data = list(filter(lambda x: x.index == each_slide.index, data.data)) if new_slide_data: updated_content = deep_update(each_slide.content, new_slide_data[0].content) - print(f"Updated content for slide {each_slide.index}: {updated_content}") new_slides.append( each_slide.get_new_slide(new_presentation.id, updated_content) ) diff --git a/servers/fastapi/api/v1/ppt/endpoints/slide.py b/servers/fastapi/api/v1/ppt/endpoints/slide.py index 3a254453..a0c81107 100644 --- a/servers/fastapi/api/v1/ppt/endpoints/slide.py +++ b/servers/fastapi/api/v1/ppt/endpoints/slide.py @@ -1,3 +1,4 @@ +import importlib from typing import Annotated, Optional from fastapi import APIRouter, Body, Depends, HTTPException from sqlalchemy.ext.asyncio import AsyncSession @@ -13,6 +14,7 @@ from utils.llm_calls.edit_slide_html import get_edited_slide_html from utils.llm_calls.select_slide_type_on_edit import get_slide_layout_from_prompt from utils.process_slides import process_old_and_new_slides_and_fetch_assets from utils.randomizers import get_random_uuid +from utils.schema_utils import remove_fields_from_schema SLIDE_ROUTER = APIRouter(prefix="/slide", tags=["Slide"]) @@ -32,12 +34,12 @@ async def edit_slide( raise HTTPException(status_code=404, detail="Presentation not found") presentation_layout = presentation.get_layout() - slide_layout = await get_slide_layout_from_prompt( prompt, presentation_layout, slide ) + edited_slide_content = await get_edited_slide_content( - prompt, slide_layout, slide, presentation.language + prompt, slide, presentation.language, slide_layout ) image_generation_service = ImageGenerationService(get_images_directory()) diff --git a/servers/fastapi/api/v1/ppt/router.py b/servers/fastapi/api/v1/ppt/router.py index 865afc19..2f22e1b5 100644 --- a/servers/fastapi/api/v1/ppt/router.py +++ b/servers/fastapi/api/v1/ppt/router.py @@ -1,6 +1,8 @@ from fastapi import APIRouter -from api.v1.ppt.endpoints.custom_llm import CUSTOM_LLM_ROUTER +from api.v1.ppt.endpoints.anthropic import ANTHROPIC_ROUTER +from api.v1.ppt.endpoints.google import GOOGLE_ROUTER +from api.v1.ppt.endpoints.openai import OPENAI_ROUTER from api.v1.ppt.endpoints.files import FILES_ROUTER from api.v1.ppt.endpoints.icons import ICONS_ROUTER from api.v1.ppt.endpoints.images import IMAGES_ROUTER @@ -19,4 +21,6 @@ API_V1_PPT_ROUTER.include_router(SLIDE_ROUTER) API_V1_PPT_ROUTER.include_router(IMAGES_ROUTER) API_V1_PPT_ROUTER.include_router(ICONS_ROUTER) API_V1_PPT_ROUTER.include_router(OLLAMA_ROUTER) -API_V1_PPT_ROUTER.include_router(CUSTOM_LLM_ROUTER) +API_V1_PPT_ROUTER.include_router(OPENAI_ROUTER) +API_V1_PPT_ROUTER.include_router(ANTHROPIC_ROUTER) +API_V1_PPT_ROUTER.include_router(GOOGLE_ROUTER) diff --git a/servers/fastapi/chroma/chroma.sqlite3 b/servers/fastapi/chroma/chroma.sqlite3 index b39add05..838ef478 100644 Binary files a/servers/fastapi/chroma/chroma.sqlite3 and b/servers/fastapi/chroma/chroma.sqlite3 differ diff --git a/servers/fastapi/constants/llm.py b/servers/fastapi/constants/llm.py new file mode 100644 index 00000000..ac4bd527 --- /dev/null +++ b/servers/fastapi/constants/llm.py @@ -0,0 +1,6 @@ +OPENAI_URL = "https://api.openai.com/v1" + +# Default models +DEFAULT_OPENAI_MODEL = "gpt-4.1" +DEFAULT_GOOGLE_MODEL = "models/gemini-2.0-flash" +DEFAULT_ANTHROPIC_MODEL = "claude-3-5-sonnet-20240620" diff --git a/servers/fastapi/constants/supported_ollama_models.py b/servers/fastapi/constants/supported_ollama_models.py index a46b5774..6ae3ac4e 100644 --- a/servers/fastapi/constants/supported_ollama_models.py +++ b/servers/fastapi/constants/supported_ollama_models.py @@ -7,7 +7,6 @@ SUPPORTED_OLLAMA_MODELS = { value="llama3:8b", description="❌ Graphs not supported.", size="4.7GB", - supports_graph=False, icon="/static/icons/meta.png", ), "llama3:70b": OllamaModelMetadata( @@ -15,7 +14,6 @@ SUPPORTED_OLLAMA_MODELS = { value="llama3:70b", description="✅ Graphs supported.", size="40GB", - supports_graph=True, icon="/static/icons/meta.png", ), "llama3.1:8b": OllamaModelMetadata( @@ -23,7 +21,6 @@ SUPPORTED_OLLAMA_MODELS = { value="llama3.1:8b", description="❌ Graphs not supported.", size="4.9GB", - supports_graph=False, icon="/static/icons/meta.png", ), "llama3.1:70b": OllamaModelMetadata( @@ -31,7 +28,6 @@ SUPPORTED_OLLAMA_MODELS = { value="llama3.1:70b", description="✅ Graphs supported.", size="43GB", - supports_graph=True, icon="/static/icons/meta.png", ), "llama3.1:405b": OllamaModelMetadata( @@ -39,7 +35,6 @@ SUPPORTED_OLLAMA_MODELS = { value="llama3.1:405b", description="✅ Graphs supported.", size="243GB", - supports_graph=True, icon="/static/icons/meta.png", ), "llama3.2:1b": OllamaModelMetadata( @@ -47,7 +42,6 @@ SUPPORTED_OLLAMA_MODELS = { value="llama3.2:1b", description="❌ Graphs not supported.", size="1.3GB", - supports_graph=False, icon="/static/icons/meta.png", ), "llama3.2:3b": OllamaModelMetadata( @@ -55,7 +49,6 @@ SUPPORTED_OLLAMA_MODELS = { value="llama3.2:3b", description="❌ Graphs not supported.", size="2GB", - supports_graph=False, icon="/static/icons/meta.png", ), "llama3.3:70b": OllamaModelMetadata( @@ -63,7 +56,6 @@ SUPPORTED_OLLAMA_MODELS = { value="llama3.3:70b", description="✅ Graphs supported.", size="43GB", - supports_graph=True, icon="/static/icons/meta.png", ), "llama4:16x17b": OllamaModelMetadata( @@ -71,7 +63,6 @@ SUPPORTED_OLLAMA_MODELS = { value="llama4:16x17b", description="✅ Graphs supported.", size="67GB", - supports_graph=True, icon="/static/icons/meta.png", ), "llama4:128x17b": OllamaModelMetadata( @@ -79,7 +70,6 @@ SUPPORTED_OLLAMA_MODELS = { value="llama4:128x17b", description="✅ Graphs supported.", size="245GB", - supports_graph=True, icon="/static/icons/meta.png", ), } @@ -90,7 +80,6 @@ SUPPORTED_GEMMA_MODELS = { value="gemma3:1b", description="❌ Graphs not supported.", size="815MB", - supports_graph=False, icon="/static/icons/gemma.png", ), "gemma3:4b": OllamaModelMetadata( @@ -98,7 +87,6 @@ SUPPORTED_GEMMA_MODELS = { value="gemma3:4b", description="❌ Graphs not supported.", size="3.3GB", - supports_graph=False, icon="/static/icons/gemma.png", ), "gemma3:12b": OllamaModelMetadata( @@ -106,7 +94,6 @@ SUPPORTED_GEMMA_MODELS = { value="gemma3:12b", description="❌ Graphs not supported.", size="8.1GB", - supports_graph=False, icon="/static/icons/gemma.png", ), "gemma3:27b": OllamaModelMetadata( @@ -114,7 +101,6 @@ SUPPORTED_GEMMA_MODELS = { value="gemma3:27b", description="✅ Graphs supported.", size="17GB", - supports_graph=True, icon="/static/icons/gemma.png", ), } @@ -125,7 +111,6 @@ SUPPORTED_DEEPSEEK_MODELS = { value="deepseek-r1:1.5b", description="❌ Graphs not supported.", size="1.1GB", - supports_graph=False, icon="/static/icons/deepseek.png", ), "deepseek-r1:7b": OllamaModelMetadata( @@ -133,7 +118,6 @@ SUPPORTED_DEEPSEEK_MODELS = { value="deepseek-r1:7b", description="❌ Graphs not supported.", size="4.7GB", - supports_graph=False, icon="/static/icons/deepseek.png", ), "deepseek-r1:8b": OllamaModelMetadata( @@ -141,7 +125,6 @@ SUPPORTED_DEEPSEEK_MODELS = { value="deepseek-r1:8b", description="❌ Graphs not supported.", size="5.2GB", - supports_graph=False, icon="/static/icons/deepseek.png", ), "deepseek-r1:14b": OllamaModelMetadata( @@ -149,7 +132,6 @@ SUPPORTED_DEEPSEEK_MODELS = { value="deepseek-r1:14b", description="❌ Graphs not supported.", size="9GB", - supports_graph=False, icon="/static/icons/deepseek.png", ), "deepseek-r1:32b": OllamaModelMetadata( @@ -157,7 +139,6 @@ SUPPORTED_DEEPSEEK_MODELS = { value="deepseek-r1:32b", description="✅ Graphs supported.", size="20GB", - supports_graph=True, icon="/static/icons/deepseek.png", ), "deepseek-r1:70b": OllamaModelMetadata( @@ -165,7 +146,6 @@ SUPPORTED_DEEPSEEK_MODELS = { value="deepseek-r1:70b", description="✅ Graphs supported.", size="43GB", - supports_graph=True, icon="/static/icons/deepseek.png", ), "deepseek-r1:671b": OllamaModelMetadata( @@ -173,7 +153,6 @@ SUPPORTED_DEEPSEEK_MODELS = { value="deepseek-r1:671b", description="✅ Graphs supported.", size="404GB", - supports_graph=True, icon="/static/icons/deepseek.png", ), } @@ -184,7 +163,6 @@ SUPPORTED_QWEN_MODELS = { value="qwen3:0.6b", description="❌ Graphs not supported.", size="523MB", - supports_graph=False, icon="/static/icons/qwen.png", ), "qwen3:1.7b": OllamaModelMetadata( @@ -192,7 +170,6 @@ SUPPORTED_QWEN_MODELS = { value="qwen3:1.7b", description="❌ Graphs not supported.", size="1.4GB", - supports_graph=False, icon="/static/icons/qwen.png", ), "qwen3:4b": OllamaModelMetadata( @@ -200,7 +177,6 @@ SUPPORTED_QWEN_MODELS = { value="qwen3:4b", description="❌ Graphs not supported.", size="2.6GB", - supports_graph=False, icon="/static/icons/qwen.png", ), "qwen3:8b": OllamaModelMetadata( @@ -208,7 +184,6 @@ SUPPORTED_QWEN_MODELS = { value="qwen3:8b", description="❌ Graphs not supported.", size="5.2GB", - supports_graph=False, icon="/static/icons/qwen.png", ), "qwen3:14b": OllamaModelMetadata( @@ -216,7 +191,6 @@ SUPPORTED_QWEN_MODELS = { value="qwen3:14b", description="❌ Graphs not supported.", size="9.3GB", - supports_graph=False, icon="/static/icons/qwen.png", ), "qwen3:30b": OllamaModelMetadata( @@ -224,7 +198,6 @@ SUPPORTED_QWEN_MODELS = { value="qwen3:30b", description="✅ Graphs supported.", size="19GB", - supports_graph=True, icon="/static/icons/qwen.png", ), "qwen3:32b": OllamaModelMetadata( @@ -232,7 +205,6 @@ SUPPORTED_QWEN_MODELS = { value="qwen3:32b", description="✅ Graphs supported.", size="20GB", - supports_graph=True, icon="/static/icons/qwen.png", ), "qwen3:235b": OllamaModelMetadata( @@ -240,7 +212,6 @@ SUPPORTED_QWEN_MODELS = { value="qwen3:235b", description="✅ Graphs supported.", size="142GB", - supports_graph=True, icon="/static/icons/qwen.png", ), } diff --git a/servers/fastapi/enums/llm_provider.py b/servers/fastapi/enums/llm_provider.py index a935425c..049d365f 100644 --- a/servers/fastapi/enums/llm_provider.py +++ b/servers/fastapi/enums/llm_provider.py @@ -5,4 +5,5 @@ class LLMProvider(Enum): OLLAMA = "ollama" OPENAI = "openai" GOOGLE = "google" + ANTHROPIC = "anthropic" CUSTOM = "custom" diff --git a/servers/fastapi/models/llm_message.py b/servers/fastapi/models/llm_message.py new file mode 100644 index 00000000..51284173 --- /dev/null +++ b/servers/fastapi/models/llm_message.py @@ -0,0 +1,7 @@ +from typing import Literal +from pydantic import BaseModel + + +class LLMMessage(BaseModel): + role: Literal["user", "system"] + content: str diff --git a/servers/fastapi/models/ollama_model_metadata.py b/servers/fastapi/models/ollama_model_metadata.py index ce2abae3..766924db 100644 --- a/servers/fastapi/models/ollama_model_metadata.py +++ b/servers/fastapi/models/ollama_model_metadata.py @@ -7,4 +7,3 @@ class OllamaModelMetadata(BaseModel): description: str icon: str size: str - supports_graph: bool diff --git a/servers/fastapi/models/presentation_layout.py b/servers/fastapi/models/presentation_layout.py index 5e90dc85..784e41fc 100644 --- a/servers/fastapi/models/presentation_layout.py +++ b/servers/fastapi/models/presentation_layout.py @@ -1,4 +1,5 @@ from typing import List, Optional +from fastapi import HTTPException from pydantic import BaseModel, Field from models.presentation_structure_model import PresentationStructureModel @@ -12,10 +13,18 @@ class SlideLayoutModel(BaseModel): class PresentationLayoutModel(BaseModel): - name: Optional[str] = None + name: str ordered: bool = Field(default=False) slides: List[SlideLayoutModel] + def get_slide_layout_index(self, slide_layout_id: str) -> int: + for index, slide in enumerate(self.slides): + if slide.id == slide_layout_id: + return index + raise HTTPException( + status_code=404, detail=f"Slide layout {slide_layout_id} not found" + ) + def to_presentation_structure(self): return PresentationStructureModel( slides=[index for index in range(len(self.slides))] diff --git a/servers/fastapi/models/presentation_outline_model.py b/servers/fastapi/models/presentation_outline_model.py index ed06485b..39383390 100644 --- a/servers/fastapi/models/presentation_outline_model.py +++ b/servers/fastapi/models/presentation_outline_model.py @@ -15,7 +15,6 @@ class PresentationOutlineModel(BaseModel): title: str = Field( description="Title of the presentation in about 3 to 8 words", ) - notes: Optional[List[str]] = Field(default=None, description="Notes for the presentation") slides: List[SlideOutlineModel] = Field(description="List of slides") def to_string(self): @@ -25,8 +24,8 @@ class PresentationOutlineModel(BaseModel): message += f" - Title: {slide.title} \n" message += f" - Body: {slide.body} \n" - if self.notes: - message += f"# Notes: \n" - for note in self.notes: - message += f" - {note} \n" + # if self.notes: + # message += f"# Notes: \n" + # for note in self.notes: + # message += f" - {note} \n" return message diff --git a/servers/fastapi/models/sql/presentation.py b/servers/fastapi/models/sql/presentation.py index 3fa5f779..a6f26604 100644 --- a/servers/fastapi/models/sql/presentation.py +++ b/servers/fastapi/models/sql/presentation.py @@ -46,7 +46,7 @@ class PresentationModel(SQLModel, table=True): return PresentationOutlineModel( title=self.title, slides=[SlideOutlineModel(**each) for each in self.outlines], - notes=self.notes, + # notes=self.notes, ) def get_layout(self): diff --git a/servers/fastapi/models/user_config.py b/servers/fastapi/models/user_config.py index 930aa1e5..4f8f5e53 100644 --- a/servers/fastapi/models/user_config.py +++ b/servers/fastapi/models/user_config.py @@ -4,13 +4,32 @@ from pydantic import BaseModel class UserConfig(BaseModel): LLM: Optional[str] = None + + # OpenAI OPENAI_API_KEY: Optional[str] = None + OPENAI_MODEL: Optional[str] = None + + # Google GOOGLE_API_KEY: Optional[str] = None + GOOGLE_MODEL: Optional[str] = None + + # Anthropic + ANTHROPIC_API_KEY: Optional[str] = None + ANTHROPIC_MODEL: Optional[str] = None + + # Ollama OLLAMA_URL: Optional[str] = None OLLAMA_MODEL: Optional[str] = None + + # Custom LLM CUSTOM_LLM_URL: Optional[str] = None CUSTOM_LLM_API_KEY: Optional[str] = None CUSTOM_MODEL: Optional[str] = None - PEXELS_API_KEY: Optional[str] = None + + # Image Provider IMAGE_PROVIDER: Optional[str] = None + PEXELS_API_KEY: Optional[str] = None PIXABAY_API_KEY: Optional[str] = None + + # Reasoning + EXTENDED_REASONING: Optional[bool] = None diff --git a/servers/fastapi/requirements.txt b/servers/fastapi/requirements.txt index 9693177a..7ccb1000 100644 --- a/servers/fastapi/requirements.txt +++ b/servers/fastapi/requirements.txt @@ -4,12 +4,15 @@ aiomysql==0.2.0 aiosignal==1.4.0 aiosqlite==0.21.0 annotated-types==0.7.0 +anthropic==0.60.0 anyio==4.9.0 +argcomplete==3.6.2 async-timeout==5.0.1 asyncpg==0.30.0 attrs==25.3.0 backoff==2.2.1 bcrypt==4.3.0 +black==25.1.0 build==1.2.2.post1 cachetools==5.5.2 certifi==2025.7.14 @@ -31,6 +34,7 @@ filelock==3.18.0 flatbuffers==25.2.10 frozenlist==1.7.0 fsspec==2025.7.0 +genson==1.3.0 google-auth==2.40.3 google-genai==1.25.0 googleapis-common-protos==1.70.0 @@ -49,7 +53,9 @@ hyperframe==6.1.0 idna==3.10 importlib_metadata==8.7.0 importlib_resources==6.5.2 +inflect==7.5.0 iniconfig==2.1.0 +isort==6.0.1 Jinja2==3.1.6 jiter==0.10.0 jsonschema==4.25.0 @@ -61,8 +67,10 @@ markdown-it-py==3.0.0 MarkupSafe==3.0.2 mdurl==0.1.2 mmh3==5.1.0 +more-itertools==10.7.0 mpmath==1.3.0 multidict==6.6.3 +mypy_extensions==1.1.0 numpy==2.3.2 oauthlib==3.3.1 onnxruntime==1.22.1 @@ -76,10 +84,12 @@ opentelemetry-semantic-conventions==0.56b0 orjson==3.11.1 overrides==7.7.0 packaging==25.0 +pathspec==0.12.1 pathvalidate==3.3.1 pdfminer.six==20250506 pdfplumber==0.11.7 pillow==11.3.0 +platformdirs==4.3.8 pluggy==1.6.0 portalocker==3.2.0 posthog==5.4.0 @@ -122,7 +132,9 @@ starlette==0.47.1 sympy==1.14.0 tenacity==8.5.0 tokenizers==0.21.2 +tomli==2.2.1 tqdm==4.67.1 +typeguard==4.4.4 typer==0.16.0 typing-inspection==0.4.1 typing_extensions==4.14.1 diff --git a/servers/fastapi/server.py b/servers/fastapi/server.py index b03ba2ac..60712ed7 100644 --- a/servers/fastapi/server.py +++ b/servers/fastapi/server.py @@ -8,14 +8,15 @@ if __name__ == "__main__": "--port", type=int, required=True, help="Port number to run the server on" ) parser.add_argument( - "--reload", type=bool, default=False, help="Reload the server on code changes" + "--reload", type=str, default="false", help="Reload the server on code changes" ) args = parser.parse_args() + reload = args.reload == "true" uvicorn.run( "api.main:app", host="0.0.0.0", port=args.port, log_level="info", - reload=args.reload, + reload=reload, ) diff --git a/servers/fastapi/services/image_generation_service.py b/servers/fastapi/services/image_generation_service.py index 667425ac..a0159b68 100644 --- a/servers/fastapi/services/image_generation_service.py +++ b/servers/fastapi/services/image_generation_service.py @@ -3,12 +3,12 @@ import os import aiohttp from google import genai from google.genai.types import GenerateContentConfig +from openai import AsyncOpenAI from models.image_prompt import ImagePrompt from models.sql.image_asset import ImageAsset from utils.download_helpers import download_file from utils.get_env import get_pexels_api_key_env from utils.get_env import get_pixabay_api_key_env -from utils.llm_provider import get_llm_client from utils.image_provider import ( is_pixels_selected, is_pixabay_selected, @@ -80,7 +80,7 @@ class ImageGenerationService: return "/static/images/placeholder.jpg" async def generate_image_openai(self, prompt: str, output_directory: str) -> str: - client = get_llm_client() + client = AsyncOpenAI() result = await client.images.generate( model="dall-e-3", prompt=prompt, diff --git a/servers/fastapi/services/llm_client.py b/servers/fastapi/services/llm_client.py new file mode 100644 index 00000000..05b19acf --- /dev/null +++ b/servers/fastapi/services/llm_client.py @@ -0,0 +1,640 @@ +import asyncio +import json +from typing import List, Optional +from fastapi import HTTPException +from openai import AsyncOpenAI +from google import genai +from google.genai.types import GenerateContentConfig +from anthropic import AsyncAnthropic +from anthropic.types import Message as AnthropicMessage +from anthropic import MessageStreamEvent as AnthropicMessageStreamEvent +from enums.llm_provider import LLMProvider +from models.llm_message import LLMMessage +from utils.async_iterator import iterator_to_async +from utils.get_env import ( + get_anthropic_api_key_env, + get_custom_llm_api_key_env, + get_custom_llm_url_env, + get_google_api_key_env, + get_ollama_url_env, + get_openai_api_key_env, +) +from utils.llm_provider import get_llm_provider +from utils.schema_utils import ensure_strict_json_schema + + +class LLMClient: + def __init__(self): + self.llm_provider = get_llm_provider() + self._client = self._get_client() + + # Supports json_schema + def supports_json_schema(self, model: str) -> bool: + if model.startswith("deepseek"): + return False + if model.startswith("claude"): + return False + return True + + # ? Clients + def _get_client(self): + match self.llm_provider: + case LLMProvider.OPENAI: + return self._get_openai_client() + case LLMProvider.GOOGLE: + return self._get_google_client() + case LLMProvider.ANTHROPIC: + return self._get_anthropic_client() + case LLMProvider.OLLAMA: + return self._get_ollama_client() + case LLMProvider.CUSTOM: + return self._get_custom_client() + case _: + raise HTTPException( + status_code=400, + detail="LLM Provider must be either openai, google, anthropic, ollama, or custom", + ) + + def _get_openai_client(self): + if not get_openai_api_key_env(): + raise HTTPException( + status_code=400, + detail="OpenAI API Key is not set", + ) + return AsyncOpenAI() + + def _get_google_client(self): + if not get_google_api_key_env(): + raise HTTPException( + status_code=400, + detail="Google API Key is not set", + ) + return genai.Client() + + def _get_anthropic_client(self): + if not get_anthropic_api_key_env(): + raise HTTPException( + status_code=400, + detail="Anthropic API Key is not set", + ) + return AsyncAnthropic() + + def _get_ollama_client(self): + return AsyncOpenAI( + base_url=(get_ollama_url_env() or "http://localhost:11434") + "/v1", + api_key="ollama", + ) + + def _get_custom_client(self): + if not get_custom_llm_url_env(): + raise HTTPException( + status_code=400, + detail="Custom LLM URL is not set", + ) + return AsyncOpenAI( + base_url=get_custom_llm_url_env(), + api_key=get_custom_llm_api_key_env() or "null", + ) + + # ? Prompts + def _get_system_prompt(self, messages: List[LLMMessage]) -> str: + for message in messages: + if message.role == "system": + return message.content + return "" + + def _get_user_prompts(self, messages: List[LLMMessage]) -> List[str]: + return [message.content for message in messages if message.role == "user"] + + def _get_user_llm_messages(self, messages: List[LLMMessage]) -> List[LLMMessage]: + return [message for message in messages if message.role == "user"] + + # ? Generate Unstructured Content + async def _generate_openai( + self, + model: str, + messages: List[LLMMessage], + max_tokens: Optional[int] = None, + ): + client: AsyncOpenAI = self._client + response = await client.chat.completions.create( + model=model, + messages=[message.model_dump() for message in messages], + max_completion_tokens=max_tokens, + ) + return response.choices[0].message.content + + async def _generate_google( + self, + model: str, + messages: List[LLMMessage], + max_tokens: Optional[int] = None, + ): + client: genai.Client = self._client + response = await asyncio.to_thread( + client.models.generate_content, + model=model, + contents=self._get_user_prompts(messages), + config=GenerateContentConfig( + system_instruction=self._get_system_prompt(messages), + response_mime_type="text/plain", + max_output_tokens=max_tokens, + ), + ) + return response.text + + async def _generate_anthropic( + self, + model: str, + messages: List[LLMMessage], + max_tokens: Optional[int] = None, + ): + client: AsyncAnthropic = self._client + response: AnthropicMessage = await client.messages.create( + model=model, + system=self._get_system_prompt(messages), + messages=[ + message.model_dump() + for message in self._get_user_llm_messages(messages) + ], + max_tokens=max_tokens or 4000, + ) + text = "" + for content in response.content: + if content.type == "text" and isinstance(content.text, str): + text += content.text + if text == "": + return None + return text + + async def _generate_ollama( + self, model: str, messages: List[LLMMessage], max_tokens: Optional[int] = None + ): + return await self._generate_openai(model, messages, max_tokens) + + async def _generate_custom( + self, model: str, messages: List[LLMMessage], max_tokens: Optional[int] = None + ): + return await self._generate_openai(model, messages, max_tokens) + + async def generate( + self, + model: str, + messages: List[LLMMessage], + max_tokens: Optional[int] = None, + ): + content = None + match self.llm_provider: + case LLMProvider.OPENAI: + content = await self._generate_openai(model, messages, max_tokens) + case LLMProvider.GOOGLE: + content = await self._generate_google(model, messages, max_tokens) + case LLMProvider.ANTHROPIC: + content = await self._generate_anthropic(model, messages, max_tokens) + case LLMProvider.OLLAMA: + content = await self._generate_ollama(model, messages, max_tokens) + case LLMProvider.CUSTOM: + content = await self._generate_custom(model, messages, max_tokens) + if content is None: + raise HTTPException( + status_code=400, + detail="LLM did not return any content", + ) + return content + + # ? Generate Structured Content + async def _generate_openai_structured( + self, + model: str, + messages: List[LLMMessage], + response_format: dict, + strict: bool = False, + max_tokens: Optional[int] = None, + ): + client: AsyncOpenAI = self._client + supports_json_schema = self.supports_json_schema(model) + response_schema = response_format + if strict: + response_schema = ensure_strict_json_schema( + response_schema, + path=(), + root=response_schema, + ) + if supports_json_schema: + response = await client.chat.completions.create( + model=model, + messages=[message.model_dump() for message in messages], + response_format={ + "type": "json_schema", + "json_schema": ( + { + "name": "ResponseSchema", + "strict": strict, + "schema": response_schema, + } + ), + }, + max_completion_tokens=max_tokens, + ) + content = response.choices[0].message.content + else: + response = await client.chat.completions.create( + model=model, + messages=[message.model_dump() for message in messages], + tools=[ + { + "type": "function", + "function": { + "name": "ResponseSchema", + "description": "A response to the user's message", + "strict": strict, + "parameters": response_format, + }, + } + ], + tool_choice="required", + max_completion_tokens=max_tokens, + ) + tool_calls = response.choices[0].message.tool_calls + if tool_calls: + content = tool_calls[0].function.arguments + + if content: + return json.loads(content) + return None + + async def _generate_google_structured( + self, + model: str, + messages: List[LLMMessage], + response_format: dict, + max_tokens: Optional[int] = None, + ): + client: genai.Client = self._client + response = await asyncio.to_thread( + client.models.generate_content, + model=model, + contents=self._get_user_prompts(messages), + config=GenerateContentConfig( + system_instruction=self._get_system_prompt(messages), + response_mime_type="application/json", + response_json_schema=response_format, + max_output_tokens=max_tokens, + ), + ) + content = None + if response.text: + content = json.loads(response.text) + + return content + + async def _generate_anthropic_structured( + self, + model: str, + messages: List[LLMMessage], + response_format: dict, + max_tokens: Optional[int] = None, + ): + client: AsyncAnthropic = self._client + response: AnthropicMessage = await client.messages.create( + model=model, + system=self._get_system_prompt(messages), + messages=[ + message.model_dump() + for message in self._get_user_llm_messages(messages) + ], + max_tokens=max_tokens or 4000, + tools=[ + { + "name": "ResponseSchema", + "description": "A response to the user's message", + "input_schema": response_format, + } + ], + tool_choice={ + "type": "tool", + "name": "ResponseSchema", + }, + ) + content: dict | None = None + for content_block in response.content: + if content_block.type == "tool_use": + content = content_block.input + + return content + + async def _generate_ollama_structured( + self, + model: str, + messages: List[LLMMessage], + response_format: dict, + strict: bool = False, + max_tokens: Optional[int] = None, + ): + return await self._generate_openai_structured( + model, messages, response_format, strict, max_tokens + ) + + async def _generate_custom_structured( + self, + model: str, + messages: List[LLMMessage], + response_format: dict, + strict: bool = False, + max_tokens: Optional[int] = None, + ): + return await self._generate_openai_structured( + model, messages, response_format, strict, max_tokens + ) + + async def generate_structured( + self, + model: str, + messages: List[LLMMessage], + response_format: dict, + strict: bool = False, + max_tokens: Optional[int] = None, + ) -> dict: + content = None + match self.llm_provider: + case LLMProvider.OPENAI: + content = await self._generate_openai_structured( + model, messages, response_format, strict, max_tokens + ) + case LLMProvider.GOOGLE: + content = await self._generate_google_structured( + model, messages, response_format, max_tokens + ) + case LLMProvider.ANTHROPIC: + content = await self._generate_anthropic_structured( + model, messages, response_format, max_tokens + ) + case LLMProvider.OLLAMA: + content = await self._generate_ollama_structured( + model, messages, response_format, strict, max_tokens + ) + case LLMProvider.CUSTOM: + content = await self._generate_custom_structured( + model, messages, response_format, strict, max_tokens + ) + if content is None: + raise HTTPException( + status_code=400, + detail="LLM did not return any content", + ) + return content + + # ? Stream Unstructured Content + async def _stream_openai( + self, + model: str, + messages: List[LLMMessage], + max_tokens: Optional[int] = None, + ): + client: AsyncOpenAI = self._client + async with client.chat.completions.stream( + model=model, + messages=[message.model_dump() for message in messages], + max_completion_tokens=max_tokens, + ) as stream: + async for event in stream: + if event.type == "content.delta": + yield event.delta + + async def _stream_google( + self, + model: str, + messages: List[LLMMessage], + max_tokens: Optional[int] = None, + ): + client: genai.Client = self._client + async for event in iterator_to_async(client.models.generate_content_stream)( + model=model, + contents=self._get_user_prompts(messages), + config=GenerateContentConfig( + system_instruction=self._get_system_prompt(messages), + response_mime_type="text/plain", + max_output_tokens=max_tokens, + ), + ): + if event.text: + yield event.text + + async def _stream_anthropic( + self, + model: str, + messages: List[LLMMessage], + max_tokens: Optional[int] = None, + ): + client: AsyncAnthropic = self._client + async with client.messages.stream( + model=model, + system=self._get_system_prompt(messages), + messages=[ + message.model_dump() + for message in self._get_user_llm_messages(messages) + ], + max_tokens=max_tokens or 4000, + ) as stream: + async for event in stream: + event: AnthropicMessageStreamEvent = event + if event.type == "text" and isinstance(event.text, str): + yield event.text + + def _stream_ollama( + self, + model: str, + messages: List[LLMMessage], + max_tokens: Optional[int] = None, + ): + return self._stream_openai(model, messages, max_tokens) + + def _stream_custom( + self, + model: str, + messages: List[LLMMessage], + max_tokens: Optional[int] = None, + ): + return self._stream_openai(model, messages, max_tokens) + + def stream( + self, model: str, messages: List[LLMMessage], max_tokens: Optional[int] = None + ): + match self.llm_provider: + case LLMProvider.OPENAI: + return self._stream_openai(model, messages, max_tokens) + case LLMProvider.GOOGLE: + return self._stream_google(model, messages, max_tokens) + case LLMProvider.ANTHROPIC: + return self._stream_anthropic(model, messages, max_tokens) + case LLMProvider.OLLAMA: + return self._stream_ollama(model, messages, max_tokens) + case LLMProvider.CUSTOM: + return self._stream_custom(model, messages, max_tokens) + + # ? Stream Structured Content + async def _stream_openai_structured( + self, + model: str, + messages: List[LLMMessage], + response_format: dict, + strict: bool = False, + max_tokens: Optional[int] = None, + ): + client: AsyncOpenAI = self._client + supports_json_schema = self.supports_json_schema(model) + response_schema = response_format + if strict: + response_schema = ensure_strict_json_schema( + response_schema, + path=(), + root=response_schema, + ) + if supports_json_schema: + async with client.chat.completions.stream( + model=model, + messages=[message.model_dump() for message in messages], + max_completion_tokens=max_tokens, + response_format=( + { + "type": "json_schema", + "json_schema": { + "name": "ResponseSchema", + "strict": strict, + "schema": response_schema, + }, + } + ), + ) as stream: + async for event in stream: + if event.type == "content.delta": + yield event.delta + else: + async with client.chat.completions.stream( + model=model, + messages=[message.model_dump() for message in messages], + max_completion_tokens=max_tokens, + tools=[ + { + "type": "function", + "function": { + "name": "ResponseSchema", + "description": "A response to the user's message", + "strict": strict, + "parameters": response_format, + }, + } + ], + tool_choice="required", + ) as stream: + async for event in stream: + if event.type == "tool_calls.function.arguments.delta": + yield event.arguments_delta + + async def _stream_google_structured( + self, + model: str, + messages: List[LLMMessage], + response_format: dict, + max_tokens: Optional[int] = None, + ): + client: genai.Client = self._client + async for event in iterator_to_async(client.models.generate_content_stream)( + model=model, + contents=self._get_user_prompts(messages), + config=GenerateContentConfig( + system_instruction=self._get_system_prompt(messages), + response_mime_type="application/json", + response_json_schema=response_format, + max_output_tokens=max_tokens, + ), + ): + if event.text: + yield event.text + + async def _stream_anthropic_structured( + self, + model: str, + messages: List[LLMMessage], + response_format: dict, + max_tokens: Optional[int] = None, + ): + client: AsyncAnthropic = self._client + async with client.messages.stream( + model=model, + system=self._get_system_prompt(messages), + messages=[ + message.model_dump() + for message in self._get_user_llm_messages(messages) + ], + max_tokens=max_tokens or 4000, + tools=[ + { + "name": "ResponseSchema", + "description": "A response to the user's message", + "input_schema": response_format, + } + ], + tool_choice={ + "type": "tool", + "name": "ResponseSchema", + }, + ) as stream: + async for event in stream: + event: AnthropicMessageStreamEvent = event + if event.type == "input_json" and isinstance(event.partial_json, str): + yield event.partial_json + + def _stream_ollama_structured( + self, + model: str, + messages: List[LLMMessage], + response_format: dict, + strict: bool = False, + max_tokens: Optional[int] = None, + ): + return self._stream_openai_structured( + model, messages, response_format, strict, max_tokens + ) + + def _stream_custom_structured( + self, + model: str, + messages: List[LLMMessage], + response_format: dict, + strict: bool = False, + max_tokens: Optional[int] = None, + ): + return self._stream_openai_structured( + model, messages, response_format, strict, max_tokens + ) + + def stream_structured( + self, + model: str, + messages: List[LLMMessage], + response_format: dict, + strict: bool = False, + max_tokens: Optional[int] = None, + ): + match self.llm_provider: + case LLMProvider.OPENAI: + return self._stream_openai_structured( + model, messages, response_format, strict, max_tokens + ) + case LLMProvider.GOOGLE: + return self._stream_google_structured( + model, messages, response_format, max_tokens + ) + case LLMProvider.ANTHROPIC: + return self._stream_anthropic_structured( + model, messages, response_format, max_tokens + ) + case LLMProvider.OLLAMA: + return self._stream_ollama_structured( + model, messages, response_format, strict, max_tokens + ) + case LLMProvider.CUSTOM: + return self._stream_custom_structured( + model, messages, response_format, strict, max_tokens + ) diff --git a/servers/fastapi/services/temp_file_service.py b/servers/fastapi/services/temp_file_service.py index 544f9246..eaa09e97 100644 --- a/servers/fastapi/services/temp_file_service.py +++ b/servers/fastapi/services/temp_file_service.py @@ -9,8 +9,7 @@ class TempFileService: def __init__(self): self.base_dir = get_temp_directory_env() or "/tmp/presenton" - # TODO: Uncomment this when we want to cleanup the base dir on startup - # self.cleanup_base_dir() + self.cleanup_base_dir() os.makedirs(self.base_dir, exist_ok=True) def create_dir_in_dir(self, base_dir: str, dir_name: Optional[str] = None) -> str: diff --git a/servers/fastapi/static/images/placeholder.jpg b/servers/fastapi/static/images/placeholder.jpg index dec5c65c..9477065c 100644 Binary files a/servers/fastapi/static/images/placeholder.jpg and b/servers/fastapi/static/images/placeholder.jpg differ diff --git a/servers/fastapi/utils/available_models.py b/servers/fastapi/utils/available_models.py new file mode 100644 index 00000000..539533ad --- /dev/null +++ b/servers/fastapi/utils/available_models.py @@ -0,0 +1,21 @@ +from anthropic import AsyncAnthropic +from openai import AsyncOpenAI +from google import genai + + +async def list_available_openai_compatible_models(url: str, api_key: str) -> list[str]: + client = AsyncOpenAI(api_key=api_key, base_url=url) + models = (await client.models.list()).data + if models: + return list(map(lambda x: x.id, models)) + return [] + + +async def list_available_anthropic_models(api_key: str) -> list[str]: + client = AsyncAnthropic(api_key=api_key) + return list(map(lambda x: x.id, (await client.models.list(limit=50)).data)) + + +async def list_available_google_models(api_key: str) -> list[str]: + client = genai.Client(api_key=api_key) + return list(map(lambda x: x.name, client.models.list(config={"page_size": 50}))) diff --git a/servers/fastapi/utils/custom_llm_provider.py b/servers/fastapi/utils/custom_llm_provider.py deleted file mode 100644 index c2e2af79..00000000 --- a/servers/fastapi/utils/custom_llm_provider.py +++ /dev/null @@ -1,18 +0,0 @@ -from typing import Optional -from openai import AsyncOpenAI - -from utils.llm_provider import get_llm_client - - -async def list_available_custom_models( - url: Optional[str] = None, api_key: Optional[str] = None -) -> list[str]: - if not url: - client = get_llm_client() - else: - client = AsyncOpenAI(api_key=api_key or "null", base_url=url) - models = [] - async for model in client.models.list(): - print(model) - models.append(model.id) - return models diff --git a/servers/fastapi/utils/dict_utils.py b/servers/fastapi/utils/dict_utils.py index e7e860b5..72c69c53 100644 --- a/servers/fastapi/utils/dict_utils.py +++ b/servers/fastapi/utils/dict_utils.py @@ -78,3 +78,12 @@ def deep_update(original: dict, updates: dict) -> dict: if not isinstance(value, (dict, list)): original[key] = value return original + + +def has_more_than_n_keys(obj: dict[str, object], n: int) -> bool: + i = 0 + for _ in obj.keys(): + i += 1 + if i > n: + return True + return False diff --git a/servers/fastapi/utils/get_dynamic_models.py b/servers/fastapi/utils/get_dynamic_models.py index 9d49c497..4dfded99 100644 --- a/servers/fastapi/utils/get_dynamic_models.py +++ b/servers/fastapi/utils/get_dynamic_models.py @@ -22,12 +22,6 @@ def get_presentation_outline_model_with_n_slides(n_slides: int): min_length=10, max_length=50, ) - notes: Optional[List[str]] = Field( - default=None, - description="Important notes for the presentation styling and formatting", - min_length=0, - max_length=10, - ) slides: List[SlideOutlineModelWithValidation] = Field( description="List of slides", min_items=n_slides, max_items=n_slides ) diff --git a/servers/fastapi/utils/get_env.py b/servers/fastapi/utils/get_env.py index 85af41c9..f2301b69 100644 --- a/servers/fastapi/utils/get_env.py +++ b/servers/fastapi/utils/get_env.py @@ -25,6 +25,14 @@ def get_llm_provider_env(): return os.getenv("LLM") +def get_anthropic_api_key_env(): + return os.getenv("ANTHROPIC_API_KEY") + + +def get_anthropic_model_env(): + return os.getenv("ANTHROPIC_MODEL") + + def get_ollama_url_env(): return os.getenv("OLLAMA_URL") @@ -37,10 +45,18 @@ def get_openai_api_key_env(): return os.getenv("OPENAI_API_KEY") +def get_openai_model_env(): + return os.getenv("OPENAI_MODEL") + + def get_google_api_key_env(): return os.getenv("GOOGLE_API_KEY") +def get_google_model_env(): + return os.getenv("GOOGLE_MODEL") + + def get_custom_llm_api_key_env(): return os.getenv("CUSTOM_LLM_API_KEY") @@ -79,3 +95,7 @@ def get_redis_db_env(): def get_redis_password_env(): return os.getenv("REDIS_PASSWORD") + + +def get_extended_reasoning_env(): + return os.getenv("EXTENDED_REASONING") diff --git a/servers/fastapi/services/get_layout_by_name.py b/servers/fastapi/utils/get_layout_by_name.py similarity index 100% rename from servers/fastapi/services/get_layout_by_name.py rename to servers/fastapi/utils/get_layout_by_name.py diff --git a/servers/fastapi/utils/llm_calls/edit_slide.py b/servers/fastapi/utils/llm_calls/edit_slide.py index 9069a459..a8df598a 100644 --- a/servers/fastapi/utils/llm_calls/edit_slide.py +++ b/servers/fastapi/utils/llm_calls/edit_slide.py @@ -1,15 +1,8 @@ -import asyncio -import json - +from models.llm_message import LLMMessage from models.presentation_layout import SlideLayoutModel from models.sql.slide import SlideModel -from google.genai.types import GenerateContentConfig -from utils.llm_provider import ( - get_google_llm_client, - get_large_model, - get_llm_client, - is_google_selected, -) +from services.llm_client import LLMClient +from utils.llm_provider import get_model from utils.schema_utils import remove_fields_from_schema system_prompt = """ @@ -42,64 +35,40 @@ def get_user_prompt(prompt: str, slide_data: dict, language: str): """ -def get_prompt_to_edit_slide_content( +def get_messages( prompt: str, slide_data: dict, language: str, ): return [ - { - "role": "system", - "content": system_prompt, - }, - { - "role": "user", - "content": get_user_prompt(prompt, slide_data, language), - }, + LLMMessage( + role="system", + content=system_prompt, + ), + LLMMessage( + role="user", + content=get_user_prompt(prompt, slide_data, language), + ), ] async def get_edited_slide_content( prompt: str, - slide_layout: SlideLayoutModel, slide: SlideModel, language: str, + slide_layout: SlideLayoutModel, ): - model = get_large_model() + model = get_model() + response_schema = remove_fields_from_schema( slide_layout.json_schema, ["__image_url__", "__icon_url__"] ) - if is_google_selected(): - client = get_google_llm_client() - response = await asyncio.to_thread( - client.models.generate_content, - model=model, - contents=[get_user_prompt(prompt, slide.content, language)], - config=GenerateContentConfig( - system_instruction=system_prompt, - response_mime_type="application/json", - response_json_schema=response_schema, - ), - ) - slide_content_json = json.loads(response.text) - else: - client = get_llm_client() - response = await client.beta.chat.completions.parse( - model=model, - messages=get_prompt_to_edit_slide_content( - prompt, - slide.content, - language, - ), - response_format={ - "type": "json_schema", - "json_schema": { - "name": "slide_content", - "schema": response_schema, - }, - }, - ) - slide_content_json = json.loads(response.choices[0].message.content) - - return slide_content_json + client = LLMClient() + response = await client.generate_structured( + model=model, + messages=get_messages(prompt, slide.content, language), + response_format=response_schema, + strict=False, + ) + return response diff --git a/servers/fastapi/utils/llm_calls/edit_slide_html.py b/servers/fastapi/utils/llm_calls/edit_slide_html.py index b20f3cf7..a5e2dfad 100644 --- a/servers/fastapi/utils/llm_calls/edit_slide_html.py +++ b/servers/fastapi/utils/llm_calls/edit_slide_html.py @@ -1,12 +1,7 @@ -import asyncio from typing import Optional -from google.genai.types import GenerateContentConfig -from utils.llm_provider import ( - get_google_llm_client, - get_large_model, - is_google_selected, - get_llm_client, -) +from models.llm_message import LLMMessage +from services.llm_client import LLMClient +from utils.llm_provider import get_model system_prompt = """ You are an expert HTML slide editor. Your task is to modify slide HTML content based on user prompts while maintaining proper structure, styling, and functionality. @@ -52,35 +47,17 @@ def get_user_prompt(prompt: str, html: str): async def get_edited_slide_html(prompt: str, html: str): - model = get_large_model() - llm_response = None - if is_google_selected(): - client = get_google_llm_client() - response = await asyncio.to_thread( - client.models.generate_content, - model=model, - contents=[get_user_prompt(prompt, html)], - config=GenerateContentConfig( - system_instruction=system_prompt, - response_mime_type="text/plain", - ), - ) - llm_response = response.text - else: - client = get_llm_client() - response = await client.chat.completions.create( - model=model, - messages=[ - {"role": "system", "content": system_prompt}, - {"role": "user", "content": get_user_prompt(prompt, html)}, - ], - ) - llm_response = response.choices[0].message.content + model = get_model() - if not llm_response: - return html - - return extract_html_from_response(llm_response) or html + client = LLMClient() + response = await client.generate( + model=model, + messages=[ + LLMMessage(role="system", content=system_prompt), + LLMMessage(role="user", content=get_user_prompt(prompt, html)), + ], + ) + return extract_html_from_response(response) or html def extract_html_from_response(response_text: str) -> Optional[str]: diff --git a/servers/fastapi/utils/llm_calls/generate_document_summary.py b/servers/fastapi/utils/llm_calls/generate_document_summary.py index 612359a9..044d28e8 100644 --- a/servers/fastapi/utils/llm_calls/generate_document_summary.py +++ b/servers/fastapi/utils/llm_calls/generate_document_summary.py @@ -1,8 +1,9 @@ import asyncio from typing import List -from openai.types.chat.chat_completion import ChatCompletion -from utils.llm_provider import get_llm_client, get_nano_model +from models.llm_message import LLMMessage +from services.llm_client import LLMClient +from utils.llm_provider import get_model sysmte_prompt = """ @@ -23,23 +24,21 @@ Maintain as much information as possible. async def generate_document_summary(documents: List[str]): - client = get_llm_client() - model = get_nano_model() + client = LLMClient() + model = get_model() coroutines = [] for document in documents: truncated_text = document[:200000] - coroutine = client.chat.completions.create( + coroutine = client.generate( model=model, messages=[ - {"role": "system", "content": sysmte_prompt}, - {"role": "user", "content": truncated_text}, + LLMMessage(role="system", content=sysmte_prompt), + LLMMessage(role="user", content=truncated_text), ], ) coroutines.append(coroutine) - completions: List[ChatCompletion] = await asyncio.gather(*coroutines) - combined = "\n\n\n\n".join( - [completion.choices[0].message.content for completion in completions] - ) + completions: List[str] = await asyncio.gather(*coroutines) + combined = "\n\n\n\n".join(completions) return combined diff --git a/servers/fastapi/utils/llm_calls/generate_presentation_outlines.py b/servers/fastapi/utils/llm_calls/generate_presentation_outlines.py index fb18d56f..54e41d61 100644 --- a/servers/fastapi/utils/llm_calls/generate_presentation_outlines.py +++ b/servers/fastapi/utils/llm_calls/generate_presentation_outlines.py @@ -1,16 +1,10 @@ +import asyncio from typing import Optional -from google.genai.types import GenerateContentConfig -from openai.types.chat.chat_completion_chunk import ChoiceDelta -from utils.async_iterator import iterator_to_async +from models.llm_message import LLMMessage +from services.llm_client import LLMClient from utils.get_dynamic_models import get_presentation_outline_model_with_n_slides -from utils.llm_provider import ( - get_google_llm_client, - get_large_model, - get_llm_client, - is_google_selected, -) -from pydantic import BaseModel +from utils.llm_provider import get_model system_prompt = """ You are an expert presentation creator. Generate structured presentations based on user requirements and format them according to the specified JSON schema with markdown content. @@ -62,61 +56,34 @@ def get_user_prompt(prompt: str, n_slides: int, language: str, content: str): """ -def get_prompt_template(prompt: str, n_slides: int, language: str, content: str): +def get_messages(prompt: str, n_slides: int, language: str, content: str): return [ - { - "role": "system", - "content": system_prompt, - }, - { - "role": "user", - "content": get_user_prompt(prompt, n_slides, language, content), - }, + LLMMessage( + role="system", + content=system_prompt, + ), + LLMMessage( + role="user", + content=get_user_prompt(prompt, n_slides, language, content), + ), ] -def get_response_format(response_model: BaseModel): - return { - "type": "json_schema", - "json_schema": { - "name": "PresentationOutlineModel", - "schema": response_model.model_json_schema(), - }, - } - - async def generate_ppt_outline( prompt: Optional[str], n_slides: int, language: Optional[str] = None, content: Optional[str] = None, ): - model = get_large_model() + model = get_model() response_model = get_presentation_outline_model_with_n_slides(n_slides) - if not is_google_selected(): - client = get_llm_client() - async for response in await client.chat.completions.create( - model=model, - messages=get_prompt_template(prompt, n_slides, language, content), - stream=True, - response_format=get_response_format(response_model), - ): - delta: ChoiceDelta = response.choices[0].delta - if delta.content: - yield delta.content + client = LLMClient() - else: - client = get_google_llm_client() - generate_stream = iterator_to_async(client.models.generate_content_stream) - async for event in generate_stream( - model=model, - contents=[get_user_prompt(prompt, n_slides, language, content)], - config=GenerateContentConfig( - system_instruction=system_prompt, - response_mime_type="application/json", - response_json_schema=response_model.model_json_schema(), - ), - ): - if event.text: - yield event.text + async for chunk in client.stream_structured( + model, + get_messages(prompt, n_slides, language, content), + response_model.model_json_schema(), + strict=True, + ): + yield chunk diff --git a/servers/fastapi/utils/llm_calls/generate_presentation_structure.py b/servers/fastapi/utils/llm_calls/generate_presentation_structure.py index f4d90fb2..47f47dba 100644 --- a/servers/fastapi/utils/llm_calls/generate_presentation_structure.py +++ b/servers/fastapi/utils/llm_calls/generate_presentation_structure.py @@ -1,24 +1,19 @@ +from models.llm_message import LLMMessage from models.presentation_layout import PresentationLayoutModel from models.presentation_outline_model import PresentationOutlineModel -from utils.llm_provider import ( - get_large_model, - get_llm_client, - get_nano_model, - get_small_model, -) -from utils.get_dynamic_models import ( - get_presentation_structure_model_with_n_slides, -) -from models.presentation_structure_model import ( - PresentationStructureModel, -) +from services.llm_client import LLMClient +from utils.llm_provider import get_model +from utils.get_dynamic_models import get_presentation_structure_model_with_n_slides +from models.presentation_structure_model import PresentationStructureModel -def get_prompt(presentation_layout: PresentationLayoutModel, n_slides: int, data: str): +def get_messages( + presentation_layout: PresentationLayoutModel, n_slides: int, data: str +): return [ - { - "role": "system", - "content": f""" + LLMMessage( + role="system", + content=f""" You're a professional presentation designer with creative freedom to design engaging presentations. {presentation_layout.to_string()} @@ -51,13 +46,13 @@ def get_prompt(presentation_layout: PresentationLayoutModel, n_slides: int, data Select layout index for each of the {n_slides} slides based on what will best serve the presentation's goals. """, - }, - { - "role": "user", - "content": f""" + ), + LLMMessage( + role="user", + content=f""" {data} """, - }, + ), ] @@ -66,20 +61,20 @@ async def generate_presentation_structure( presentation_layout: PresentationLayoutModel, ) -> PresentationStructureModel: - client = get_llm_client() - model = get_large_model() + client = LLMClient() + model = get_model() response_model = get_presentation_structure_model_with_n_slides( len(presentation_outline.slides) ) - response = await client.beta.chat.completions.parse( + response = await client.generate_structured( model=model, - messages=get_prompt( + messages=get_messages( presentation_layout, len(presentation_outline.slides), presentation_outline.to_string(), ), - response_format=response_model, + response_format=response_model.model_json_schema(), + strict=True, ) - print(response.choices[0].message.parsed) - return response.choices[0].message.parsed + return PresentationStructureModel(**response) diff --git a/servers/fastapi/utils/llm_calls/generate_slide_content.py b/servers/fastapi/utils/llm_calls/generate_slide_content.py index e4fe86b0..f75ed979 100644 --- a/servers/fastapi/utils/llm_calls/generate_slide_content.py +++ b/servers/fastapi/utils/llm_calls/generate_slide_content.py @@ -1,14 +1,8 @@ -import asyncio -import json -from google.genai.types import GenerateContentConfig +from models.llm_message import LLMMessage from models.presentation_layout import SlideLayoutModel from models.presentation_outline_model import SlideOutlineModel -from utils.llm_provider import ( - get_google_llm_client, - get_large_model, - get_llm_client, - is_google_selected, -) +from services.llm_client import LLMClient +from utils.llm_provider import get_model from utils.schema_utils import remove_fields_from_schema system_prompt = """ @@ -45,57 +39,38 @@ def get_user_prompt(title: str, outline: str, language: str): """ -def get_prompt_to_generate_slide_content(title: str, outline: str, language: str): +def get_messages(title: str, outline: str, language: str): return [ - { - "role": "system", - "content": system_prompt, - }, - { - "role": "user", - "content": get_user_prompt(title, outline, language), - }, + LLMMessage( + role="system", + content=system_prompt, + ), + LLMMessage( + role="user", + content=get_user_prompt(title, outline, language), + ), ] async def get_slide_content_from_type_and_outline( slide_layout: SlideLayoutModel, outline: SlideOutlineModel, language: str ): - model = get_large_model() + client = LLMClient() + model = get_model() response_schema = remove_fields_from_schema( slide_layout.json_schema, ["__image_url__", "__icon_url__"] ) - if not is_google_selected(): - client = get_llm_client() - response = await client.beta.chat.completions.parse( - model=model, - messages=get_prompt_to_generate_slide_content( - outline.title, - outline.body, - language, - ), - response_format={ - "type": "json_schema", - "json_schema": { - "name": "SlideContent", - "schema": response_schema, - }, - }, - ) - return json.loads(response.choices[0].message.content) - else: - client = get_google_llm_client() - response = await asyncio.to_thread( - client.models.generate_content, - model=model, - contents=[get_user_prompt(outline.title, outline.body, language)], - config=GenerateContentConfig( - system_instruction=system_prompt, - response_mime_type="application/json", - response_json_schema=response_schema, - ), - ) - return json.loads(response.text) + response = await client.generate_structured( + model=model, + messages=get_messages( + outline.title, + outline.body, + language, + ), + response_format=response_schema, + strict=False, + ) + return response diff --git a/servers/fastapi/utils/llm_calls/select_slide_type_on_edit.py b/servers/fastapi/utils/llm_calls/select_slide_type_on_edit.py index 861621d1..f3532b48 100644 --- a/servers/fastapi/utils/llm_calls/select_slide_type_on_edit.py +++ b/servers/fastapi/utils/llm_calls/select_slide_type_on_edit.py @@ -1,19 +1,21 @@ +from models.llm_message import LLMMessage from models.presentation_layout import PresentationLayoutModel, SlideLayoutModel from models.slide_layout_index import SlideLayoutIndex from models.sql.slide import SlideModel -from utils.llm_provider import get_large_model, get_llm_client +from services.llm_client import LLMClient +from utils.llm_provider import get_model -def get_prompt_to_select_slide_layout( +def get_messages( prompt: str, slide_data: dict, layout: PresentationLayoutModel, current_slide_layout: int, ): return [ - { - "role": "system", - "content": f""" + LLMMessage( + role="system", + content=f""" Select a Slide Layout index based on provided user prompt and current slide data. {layout.to_string()} @@ -23,15 +25,15 @@ def get_prompt_to_select_slide_layout( - If user prompt is not clear, select the layout that is most relevant to the slide data. **Go through all notes and steps and make sure they are followed, including mentioned constraints** """, - }, - { - "role": "user", - "content": f""" + ), + LLMMessage( + role="user", + content=f""" - User Prompt: {prompt} - Current Slide Data: {slide_data} - Current Slide Layout: {current_slide_layout} """, - }, + ), ] @@ -41,21 +43,21 @@ async def get_slide_layout_from_prompt( slide: SlideModel, ) -> SlideLayoutModel: - client = get_llm_client() - model = get_large_model() + client = LLMClient() + model = get_model() - slide_layout_ids = list(map(lambda x: x.id, layout.slides)) + slide_layout_index = layout.get_slide_layout_index(slide.layout) - response = await client.beta.chat.completions.parse( + response = await client.generate_structured( model=model, - temperature=0.2, - messages=get_prompt_to_select_slide_layout( + messages=get_messages( prompt, slide.content, layout, - slide_layout_ids.index(slide.layout), + slide_layout_index, ), - response_format=SlideLayoutIndex, + response_format=SlideLayoutIndex.model_json_schema(), + strict=True, ) - index = response.choices[0].message.parsed.index + index = SlideLayoutIndex(**response).index return layout.slides[index] diff --git a/servers/fastapi/utils/llm_provider.py b/servers/fastapi/utils/llm_provider.py index f4e801ea..aabc8f61 100644 --- a/servers/fastapi/utils/llm_provider.py +++ b/servers/fastapi/utils/llm_provider.py @@ -1,36 +1,31 @@ -import os from fastapi import HTTPException -from openai import AsyncOpenAI -from google import genai +from constants.llm import ( + DEFAULT_ANTHROPIC_MODEL, + DEFAULT_GOOGLE_MODEL, + DEFAULT_OPENAI_MODEL, +) from enums.llm_provider import LLMProvider from utils.get_env import ( - get_custom_llm_api_key_env, - get_custom_llm_url_env, + get_anthropic_model_env, get_custom_model_env, - get_google_api_key_env, + get_google_model_env, get_llm_provider_env, get_ollama_model_env, - get_ollama_url_env, - get_openai_api_key_env, + get_openai_model_env, ) - def get_llm_provider(): try: return LLMProvider(get_llm_provider_env()) except: raise HTTPException( status_code=500, - detail=f"Invalid LLM provider. Please select one of: openai, google, ollama, custom", + detail=f"Invalid LLM provider. Please select one of: openai, google, anthropic, ollama, custom", ) -def get_ollama_url(): - return get_ollama_url_env() or "http://localhost:11434" - - def is_openai_selected(): return get_llm_provider() == LLMProvider.OPENAI @@ -39,6 +34,10 @@ def is_google_selected(): return get_llm_provider() == LLMProvider.GOOGLE +def is_anthropic_selected(): + return get_llm_provider() == LLMProvider.ANTHROPIC + + def is_ollama_selected(): return get_llm_provider() == LLMProvider.OLLAMA @@ -47,85 +46,20 @@ def is_custom_llm_selected(): return get_llm_provider() == LLMProvider.CUSTOM -def get_model_base_url(): - selected_llm = get_llm_provider() - - if selected_llm == LLMProvider.OPENAI: - return "https://api.openai.com/v1" - elif selected_llm == LLMProvider.GOOGLE: - return "https://generativelanguage.googleapis.com/v1beta/openai" - elif selected_llm == LLMProvider.OLLAMA: - return os.path.join(get_ollama_url(), "v1") - elif selected_llm == LLMProvider.CUSTOM: - return get_custom_llm_url_env() - else: - raise HTTPException(f"LLM provider {selected_llm} is not supported") - - -def get_llm_api_key(): +def get_model(): selected_llm = get_llm_provider() if selected_llm == LLMProvider.OPENAI: - return get_openai_api_key_env() + return get_openai_model_env() or DEFAULT_OPENAI_MODEL elif selected_llm == LLMProvider.GOOGLE: - return get_google_api_key_env() + return get_google_model_env() or DEFAULT_GOOGLE_MODEL + elif selected_llm == LLMProvider.ANTHROPIC: + return get_anthropic_model_env() or DEFAULT_ANTHROPIC_MODEL elif selected_llm == LLMProvider.OLLAMA: - return "ollama" + return get_ollama_model_env() elif selected_llm == LLMProvider.CUSTOM: - return get_custom_llm_api_key_env() or "none" + return get_custom_model_env() else: - raise HTTPException(f"LLM provider {selected_llm} is not supported") - - -def get_llm_client(): - client = AsyncOpenAI( - base_url=get_model_base_url(), - api_key=get_llm_api_key(), + raise HTTPException( + status_code=500, + detail=f"Invalid LLM provider. Please select one of: openai, google, anthropic, ollama, custom", ) - return client - - -def get_google_llm_client(): - client = genai.Client(api_key=get_llm_api_key()) - return client - - -def get_large_model(): - selected_llm = get_llm_provider() - if selected_llm == LLMProvider.OPENAI: - return "gpt-4.1" - elif selected_llm == LLMProvider.GOOGLE: - return "gemini-2.0-flash" - elif selected_llm == LLMProvider.OLLAMA: - return get_ollama_model_env() - elif selected_llm == LLMProvider.CUSTOM: - return get_custom_model_env() - else: - raise ValueError(f"Invalid LLM model") - - -def get_small_model(): - selected_llm = get_llm_provider() - if selected_llm == LLMProvider.OPENAI: - return "gpt-4.1-mini" - elif selected_llm == LLMProvider.GOOGLE: - return "gemini-2.0-flash" - elif selected_llm == LLMProvider.OLLAMA: - return get_ollama_model_env() - elif selected_llm == LLMProvider.CUSTOM: - return get_custom_model_env() - else: - raise ValueError(f"Invalid LLM model") - - -def get_nano_model(): - selected_llm = get_llm_provider() - if selected_llm == LLMProvider.OPENAI: - return "gpt-4.1-nano" - elif selected_llm == LLMProvider.GOOGLE: - return "gemini-2.0-flash" - elif selected_llm == LLMProvider.OLLAMA: - return get_ollama_model_env() - elif selected_llm == LLMProvider.CUSTOM: - return get_custom_model_env() - else: - raise ValueError(f"Invalid LLM model") diff --git a/servers/fastapi/utils/model_availability.py b/servers/fastapi/utils/model_availability.py index 210fcdd2..4f942ac7 100644 --- a/servers/fastapi/utils/model_availability.py +++ b/servers/fastapi/utils/model_availability.py @@ -1,11 +1,19 @@ -import os from constants.supported_ollama_models import SUPPORTED_OLLAMA_MODELS +from constants.llm import OPENAI_URL from enums.image_provider import ImageProvider from enums.llm_provider import LLMProvider -from utils.custom_llm_provider import list_available_custom_models +from utils.available_models import ( + list_available_anthropic_models, + list_available_google_models, + list_available_openai_compatible_models, +) from utils.get_env import ( + get_anthropic_api_key_env, + get_anthropic_model_env, get_can_change_keys_env, + get_google_model_env, get_openai_api_key_env, + get_openai_model_env, get_pixabay_api_key_env, get_pexels_api_key_env, ) @@ -20,13 +28,7 @@ from utils.llm_provider import ( is_ollama_selected, ) from utils.ollama import pull_ollama_model -from utils.image_provider import ( - get_selected_image_provider, - is_pixels_selected, - is_pixabay_selected, - is_gemini_flash_selected, - is_dalle3_selected, -) +from utils.image_provider import get_selected_image_provider async def check_llm_and_image_provider_api_or_model_availability(): @@ -36,11 +38,41 @@ async def check_llm_and_image_provider_api_or_model_availability(): openai_api_key = get_openai_api_key_env() if not openai_api_key: raise Exception("OPENAI_API_KEY must be provided") + openai_model = get_openai_model_env() + if openai_model: + available_models = await list_available_openai_compatible_models( + OPENAI_URL, openai_api_key + ) + if openai_model not in available_models: + print("-" * 50) + print("Available models: ", available_models) + raise Exception(f"Model {openai_model} is not available") elif get_llm_provider() == LLMProvider.GOOGLE: google_api_key = get_google_api_key_env() if not google_api_key: raise Exception("GOOGLE_API_KEY must be provided") + google_model = get_google_model_env() + if google_model: + available_models = await list_available_google_models(google_api_key) + if google_model not in available_models: + print("-" * 50) + print("Available models: ", available_models) + raise Exception(f"Model {google_model} is not available") + + elif get_llm_provider() == LLMProvider.ANTHROPIC: + anthropic_api_key = get_anthropic_api_key_env() + if not anthropic_api_key: + raise Exception("ANTHROPIC_API_KEY must be provided") + anthropic_model = get_anthropic_model_env() + if anthropic_model: + available_models = await list_available_anthropic_models( + anthropic_api_key + ) + if anthropic_model not in available_models: + print("-" * 50) + print("Available models: ", available_models) + raise Exception(f"Model {anthropic_model} is not available") elif is_ollama_selected(): ollama_model = get_ollama_model_env() @@ -60,21 +92,16 @@ async def check_llm_and_image_provider_api_or_model_availability(): elif is_custom_llm_selected(): custom_model = get_custom_model_env() custom_llm_url = get_custom_llm_url_env() - custom_llm_api_key = get_custom_llm_api_key_env() if not custom_model: raise Exception("CUSTOM_MODEL must be provided") if not custom_llm_url: raise Exception("CUSTOM_LLM_URL must be provided") - if not custom_llm_api_key: - raise Exception("CUSTOM_LLM_API_KEY must be provided") - print("-" * 50) - print("Selecting model: ", custom_model) - models = await list_available_custom_models( - custom_llm_url, custom_llm_api_key + available_models = await list_available_openai_compatible_models( + custom_llm_url, get_custom_llm_api_key_env() or "null" ) - print("Available models: ", models) print("-" * 50) - if custom_model not in models: + print("Available models: ", available_models) + if custom_model not in available_models: raise Exception(f"Model {custom_model} is not available") # Check for Image Provider and API keys diff --git a/servers/fastapi/utils/ollama.py b/servers/fastapi/utils/ollama.py index 10c8500f..928ee895 100644 --- a/servers/fastapi/utils/ollama.py +++ b/servers/fastapi/utils/ollama.py @@ -1,7 +1,7 @@ -from http.client import HTTPException import json from typing import AsyncGenerator import aiohttp +from fastapi import HTTPException from models.ollama_model_status import OllamaModelStatus from utils.get_env import get_ollama_url_env diff --git a/servers/fastapi/utils/process_slides.py b/servers/fastapi/utils/process_slides.py index 6ffca66f..81a0a6b5 100644 --- a/servers/fastapi/utils/process_slides.py +++ b/servers/fastapi/utils/process_slides.py @@ -21,19 +21,19 @@ async def process_slide_and_fetch_assets( icon_paths = get_dict_paths_with_key(slide.content, "__icon_query__") for image_path in image_paths: - image_prompt_parent = get_dict_at_path(slide.content, image_path) + __image_prompt__parent = get_dict_at_path(slide.content, image_path) async_tasks.append( image_generation_service.generate_image( ImagePrompt( - prompt=image_prompt_parent["__image_prompt__"], + prompt=__image_prompt__parent["__image_prompt__"], ) ) ) for icon_path in icon_paths: - icon_query_parent = get_dict_at_path(slide.content, icon_path) + __icon_query__parent = get_dict_at_path(slide.content, icon_path) async_tasks.append( - icon_finder_service.search_icons(icon_query_parent["__icon_query__"]) + icon_finder_service.search_icons(__icon_query__parent["__icon_query__"]) ) results = await asyncio.gather(*async_tasks) diff --git a/servers/fastapi/utils/schema_utils.py b/servers/fastapi/utils/schema_utils.py index 0f2ad6d7..ae65f002 100644 --- a/servers/fastapi/utils/schema_utils.py +++ b/servers/fastapi/utils/schema_utils.py @@ -1,30 +1,25 @@ from copy import deepcopy -from typing import List +from typing import Any, List -from utils.dict_utils import get_dict_paths_with_key, get_dict_at_path +from openai import NOT_GIVEN +from utils.dict_utils import ( + get_dict_paths_with_key, + get_dict_at_path, + has_more_than_n_keys, +) -def resolve_refs(schema, defs): - if isinstance(schema, dict): - if "$ref" in schema: - ref_path = schema["$ref"] - if ref_path.startswith("#/$defs/"): - def_key = ref_path.replace("#/$defs/", "") - return resolve_refs(defs[def_key], defs) - else: - raise ValueError(f"Unsupported $ref path: {ref_path}") - else: - return {k: resolve_refs(v, defs) for k, v in schema.items()} - elif isinstance(schema, list): - return [resolve_refs(item, defs) for item in schema] - else: - return schema - - -def flatten_schema(schema): - schema = deepcopy(schema) - defs = schema.pop("$defs", {}) - return resolve_refs(schema, defs) +supported_string_formats = [ + "date-time", + "time", + "date", + "duration", + "email", + "hostname", + "ipv4", + "ipv6", + "uuid", +] def remove_fields_from_schema(schema: dict, fields_to_remove: List[str]): @@ -50,6 +45,138 @@ def remove_fields_from_schema(schema: dict, fields_to_remove: List[str]): return schema +# From OpenAI +def ensure_strict_json_schema( + json_schema: object, + *, + path: tuple[str, ...], + root: dict[str, object], +) -> dict[str, Any]: + """Mutates the given JSON schema to ensure it conforms to the `strict` standard + that the API expects. + """ + if not isinstance(json_schema, dict): + raise TypeError(f"Expected {json_schema} to be a dictionary; path={path}") + + defs = json_schema.get("$defs") + if isinstance(defs, dict): + for def_name, def_schema in defs.items(): + ensure_strict_json_schema( + def_schema, path=(*path, "$defs", def_name), root=root + ) + + definitions = json_schema.get("definitions") + if isinstance(definitions, dict): + for definition_name, definition_schema in definitions.items(): + ensure_strict_json_schema( + definition_schema, + path=(*path, "definitions", definition_name), + root=root, + ) + + typ = json_schema.get("type") + if typ == "object" and "additionalProperties" not in json_schema: + json_schema["additionalProperties"] = False + + # object types + # { 'type': 'object', 'properties': { 'a': {...} } } + properties = json_schema.get("properties") + if isinstance(properties, dict): + json_schema["required"] = [prop for prop in properties.keys()] + json_schema["properties"] = { + key: ensure_strict_json_schema( + prop_schema, path=(*path, "properties", key), root=root + ) + for key, prop_schema in properties.items() + } + + # arrays + # { 'type': 'array', 'items': {...} } + items = json_schema.get("items") + if isinstance(items, dict): + json_schema["items"] = ensure_strict_json_schema( + items, path=(*path, "items"), root=root + ) + + # unions + any_of = json_schema.get("anyOf") + if isinstance(any_of, list): + json_schema["anyOf"] = [ + ensure_strict_json_schema(variant, path=(*path, "anyOf", str(i)), root=root) + for i, variant in enumerate(any_of) + ] + + # intersections + all_of = json_schema.get("allOf") + if isinstance(all_of, list): + if len(all_of) == 1: + json_schema.update( + ensure_strict_json_schema( + all_of[0], path=(*path, "allOf", "0"), root=root + ) + ) + json_schema.pop("allOf") + else: + json_schema["allOf"] = [ + ensure_strict_json_schema( + entry, path=(*path, "allOf", str(i)), root=root + ) + for i, entry in enumerate(all_of) + ] + + # string + if typ == "string": + if "format" in json_schema: + if json_schema["format"] not in supported_string_formats: + del json_schema["format"] + + # strip `None` defaults as there's no meaningful distinction here + # the schema will still be `nullable` and the model will default + # to using `None` anyway + if json_schema.get("default", NOT_GIVEN) is None: + json_schema.pop("default") + + # we can't use `$ref`s if there are also other properties defined, e.g. + # `{"$ref": "...", "description": "my description"}` + # + # so we unravel the ref + # `{"type": "string", "description": "my description"}` + ref = json_schema.get("$ref") + if ref and has_more_than_n_keys(json_schema, 1): + assert isinstance(ref, str), f"Received non-string $ref - {ref}" + + resolved = resolve_ref(root=root, ref=ref) + if not isinstance(resolved, dict): + raise ValueError( + f"Expected `$ref: {ref}` to resolved to a dictionary but got {resolved}" + ) + + # properties from the json schema take priority over the ones on the `$ref` + json_schema.update({**resolved, **json_schema}) + json_schema.pop("$ref") + # Since the schema expanded from `$ref` might not have `additionalProperties: false` applied, + # we call `_ensure_strict_json_schema` again to fix the inlined schema and ensure it's valid. + return ensure_strict_json_schema(json_schema, path=path, root=root) + + return json_schema + + +def resolve_ref(*, root: dict[str, object], ref: str) -> object: + if not ref.startswith("#/"): + raise ValueError(f"Unexpected $ref format {ref!r}; Does not start with #/") + + path = ref[2:].split("/") + resolved = root + for key in path: + value = resolved[key] + assert isinstance( + value, dict + ), f"encountered non-dictionary entry while resolving {ref} - {resolved}" + resolved = value + + return resolved + + # ? Not used def generate_constraint_sentences(schema: dict) -> str: """ diff --git a/servers/fastapi/utils/set_env.py b/servers/fastapi/utils/set_env.py index 836c8caa..5566d800 100644 --- a/servers/fastapi/utils/set_env.py +++ b/servers/fastapi/utils/set_env.py @@ -25,10 +25,26 @@ def set_openai_api_key_env(value): os.environ["OPENAI_API_KEY"] = value +def set_openai_model_env(value): + os.environ["OPENAI_MODEL"] = value + + def set_google_api_key_env(value): os.environ["GOOGLE_API_KEY"] = value +def set_google_model_env(value): + os.environ["GOOGLE_MODEL"] = value + + +def set_anthropic_api_key_env(value): + os.environ["ANTHROPIC_API_KEY"] = value + + +def set_anthropic_model_env(value): + os.environ["ANTHROPIC_MODEL"] = value + + def set_custom_llm_api_key_env(value): os.environ["CUSTOM_LLM_API_KEY"] = value @@ -51,3 +67,7 @@ def set_image_provider_env(value): def set_pixabay_api_key_env(value): os.environ["PIXABAY_API_KEY"] = value + + +def set_extended_reasoning_env(value): + os.environ["EXTENDED_REASONING"] = value diff --git a/servers/fastapi/utils/user_config.py b/servers/fastapi/utils/user_config.py index b1065a2b..d1ddd20a 100644 --- a/servers/fastapi/utils/user_config.py +++ b/servers/fastapi/utils/user_config.py @@ -3,31 +3,41 @@ import json from models.user_config import UserConfig from utils.get_env import ( + get_anthropic_api_key_env, + get_anthropic_model_env, get_custom_llm_api_key_env, get_custom_llm_url_env, get_custom_model_env, get_google_api_key_env, + get_google_model_env, get_llm_provider_env, get_ollama_model_env, get_ollama_url_env, get_openai_api_key_env, + get_openai_model_env, get_pexels_api_key_env, get_user_config_path_env, get_image_provider_env, - get_pixabay_api_key_env + get_pixabay_api_key_env, + get_extended_reasoning_env, ) from utils.set_env import ( + set_anthropic_api_key_env, + set_anthropic_model_env, set_custom_llm_api_key_env, set_custom_llm_url_env, set_custom_model_env, + set_extended_reasoning_env, set_google_api_key_env, + set_google_model_env, set_llm_provider_env, set_ollama_model_env, set_ollama_url_env, set_openai_api_key_env, + set_openai_model_env, set_pexels_api_key_env, set_image_provider_env, - set_pixabay_api_key_env + set_pixabay_api_key_env, ) @@ -43,10 +53,21 @@ def get_user_config(): print("Error while loading user config") pass + new_extended_reasoning = ( + existing_config.EXTENDED_REASONING or get_extended_reasoning_env() + ) + if new_extended_reasoning is not None: + new_extended_reasoning = bool(new_extended_reasoning) + return UserConfig( LLM=existing_config.LLM or get_llm_provider_env(), OPENAI_API_KEY=existing_config.OPENAI_API_KEY or get_openai_api_key_env(), + OPENAI_MODEL=existing_config.OPENAI_MODEL or get_openai_model_env(), GOOGLE_API_KEY=existing_config.GOOGLE_API_KEY or get_google_api_key_env(), + GOOGLE_MODEL=existing_config.GOOGLE_MODEL or get_google_model_env(), + ANTHROPIC_API_KEY=existing_config.ANTHROPIC_API_KEY + or get_anthropic_api_key_env(), + ANTHROPIC_MODEL=existing_config.ANTHROPIC_MODEL or get_anthropic_model_env(), OLLAMA_URL=existing_config.OLLAMA_URL or get_ollama_url_env(), OLLAMA_MODEL=existing_config.OLLAMA_MODEL or get_ollama_model_env(), CUSTOM_LLM_URL=existing_config.CUSTOM_LLM_URL or get_custom_llm_url_env(), @@ -56,6 +77,7 @@ def get_user_config(): IMAGE_PROVIDER=existing_config.IMAGE_PROVIDER or get_image_provider_env(), PIXABAY_API_KEY=existing_config.PIXABAY_API_KEY or get_pixabay_api_key_env(), PEXELS_API_KEY=existing_config.PEXELS_API_KEY or get_pexels_api_key_env(), + EXTENDED_REASONING=new_extended_reasoning, ) @@ -65,8 +87,16 @@ def update_env_with_user_config(): set_llm_provider_env(user_config.LLM) if user_config.OPENAI_API_KEY: set_openai_api_key_env(user_config.OPENAI_API_KEY) + if user_config.OPENAI_MODEL: + set_openai_model_env(user_config.OPENAI_MODEL) if user_config.GOOGLE_API_KEY: set_google_api_key_env(user_config.GOOGLE_API_KEY) + if user_config.GOOGLE_MODEL: + set_google_model_env(user_config.GOOGLE_MODEL) + if user_config.ANTHROPIC_API_KEY: + set_anthropic_api_key_env(user_config.ANTHROPIC_API_KEY) + if user_config.ANTHROPIC_MODEL: + set_anthropic_model_env(user_config.ANTHROPIC_MODEL) if user_config.OLLAMA_URL: set_ollama_url_env(user_config.OLLAMA_URL) if user_config.OLLAMA_MODEL: @@ -83,3 +113,6 @@ def update_env_with_user_config(): set_pixabay_api_key_env(user_config.PIXABAY_API_KEY) if user_config.PEXELS_API_KEY: set_pexels_api_key_env(user_config.PEXELS_API_KEY) + if user_config.EXTENDED_REASONING: + if user_config.EXTENDED_REASONING: + set_extended_reasoning_env(str(user_config.EXTENDED_REASONING)) diff --git a/servers/nextjs/app/(presentation-generator)/upload/type.ts b/servers/nextjs/app/(presentation-generator)/upload/type.ts index 55d878d6..bf5927de 100644 --- a/servers/nextjs/app/(presentation-generator)/upload/type.ts +++ b/servers/nextjs/app/(presentation-generator)/upload/type.ts @@ -24,7 +24,8 @@ export enum LanguageType { Italian = "Italian (Italiano)", Dutch = "Dutch (Nederlands)", Russian = "Russian (Русский)", - Chinese = "Chinese (Simplified & Traditional - 中文, 汉语/漢語)", + ChineseSimplified = "Chinese (Simplified - 中文, 汉语)", + ChineseTraditional = "Chinese (Traditional - 中文, 漢語)", Japanese = "Japanese (日本語)", Korean = "Korean (한국어)", Arabic = "Arabic (العربية)", diff --git a/servers/nextjs/app/api/export-as-pdf/route.ts b/servers/nextjs/app/api/export-as-pdf/route.ts index afc5e378..10085f9c 100644 --- a/servers/nextjs/app/api/export-as-pdf/route.ts +++ b/servers/nextjs/app/api/export-as-pdf/route.ts @@ -8,7 +8,6 @@ import { NextResponse, NextRequest } from 'next/server'; export async function POST(req: NextRequest) { const { id, title } = await req.json(); - console.log('path', process.env.APP_DATA_DIRECTORY); if (!id) { return NextResponse.json({ error: "Missing Presentation ID" }, { status: 400 }); } @@ -69,7 +68,7 @@ export async function POST(req: NextRequest) { const sanitizedTitle = sanitizeFilename(title); const destinationPath = path.join(process.env.APP_DATA_DIRECTORY!, 'exports', `${sanitizedTitle}.pdf`); - console.log('destinationPath', destinationPath); + await fs.promises.mkdir(path.dirname(destinationPath), { recursive: true }); await fs.promises.writeFile(destinationPath, pdfBuffer); return NextResponse.json({ diff --git a/servers/nextjs/app/api/user-config/route.ts b/servers/nextjs/app/api/user-config/route.ts index 39ba481e..71219c0c 100644 --- a/servers/nextjs/app/api/user-config/route.ts +++ b/servers/nextjs/app/api/user-config/route.ts @@ -35,7 +35,11 @@ export async function POST(request: Request) { const mergedConfig: LLMConfig = { LLM: userConfig.LLM || existingConfig.LLM, OPENAI_API_KEY: userConfig.OPENAI_API_KEY || existingConfig.OPENAI_API_KEY, + OPENAI_MODEL: userConfig.OPENAI_MODEL || existingConfig.OPENAI_MODEL, GOOGLE_API_KEY: userConfig.GOOGLE_API_KEY || existingConfig.GOOGLE_API_KEY, + GOOGLE_MODEL: userConfig.GOOGLE_MODEL || existingConfig.GOOGLE_MODEL, + ANTHROPIC_API_KEY: userConfig.ANTHROPIC_API_KEY || existingConfig.ANTHROPIC_API_KEY, + ANTHROPIC_MODEL: userConfig.ANTHROPIC_MODEL || existingConfig.ANTHROPIC_MODEL, OLLAMA_URL: userConfig.OLLAMA_URL || existingConfig.OLLAMA_URL, OLLAMA_MODEL: userConfig.OLLAMA_MODEL || existingConfig.OLLAMA_MODEL, CUSTOM_LLM_URL: userConfig.CUSTOM_LLM_URL || existingConfig.CUSTOM_LLM_URL, @@ -50,6 +54,10 @@ export async function POST(request: Request) { userConfig.USE_CUSTOM_URL === undefined ? existingConfig.USE_CUSTOM_URL : userConfig.USE_CUSTOM_URL, + EXTENDED_REASONING: + userConfig.EXTENDED_REASONING === undefined + ? existingConfig.EXTENDED_REASONING + : userConfig.EXTENDED_REASONING, }; fs.writeFileSync(userConfigPath, JSON.stringify(mergedConfig)); return NextResponse.json(mergedConfig); diff --git a/servers/nextjs/app/settings/SettingPage.tsx b/servers/nextjs/app/settings/SettingPage.tsx index ece5f16b..b6e447ca 100644 --- a/servers/nextjs/app/settings/SettingPage.tsx +++ b/servers/nextjs/app/settings/SettingPage.tsx @@ -29,7 +29,6 @@ const SettingsPage = () => { const userConfigState = useSelector((state: RootState) => state.userConfig); const [llmConfig, setLlmConfig] = useState(userConfigState.llm_config); const canChangeKeys = userConfigState.can_change_keys; - const [isLoading, setIsLoading] = useState(false); const [buttonState, setButtonState] = useState({ isLoading: false, isDisabled: false, @@ -55,16 +54,13 @@ const SettingsPage = () => { const handleSaveConfig = async () => { try { - setIsLoading(true); setButtonState(prev => ({ ...prev, isLoading: true, isDisabled: true, text: "Saving Configuration..." })); - await handleSaveLLMConfig(llmConfig); - if (llmConfig.LLM === "ollama" && llmConfig.OLLAMA_MODEL) { const isPulled = await checkIfSelectedOllamaModelIsPulled(llmConfig.OLLAMA_MODEL); if (!isPulled) { @@ -72,24 +68,16 @@ const SettingsPage = () => { await handleModelDownload(); } } - toast.info("Configuration saved successfully"); - setIsLoading(false); setButtonState(prev => ({ ...prev, isLoading: false, isDisabled: false, text: "Save Configuration" })); - router.back(); + router.push("/upload"); } catch (error) { - console.error("Error:", error); - toast.info( - error instanceof Error - ? error.message - : "Failed to save configuration" - ); - setIsLoading(false); + toast.info(error instanceof Error ? error.message : "Failed to save configuration"); setButtonState(prev => ({ ...prev, isLoading: false, @@ -102,8 +90,8 @@ const SettingsPage = () => { const handleModelDownload = async () => { try { await pullOllamaModel(llmConfig.OLLAMA_MODEL!, setDownloadingModel); - } catch (error) { - console.error("Error downloading model:", error); + } + finally { setDownloadingModel(null); setShowDownloadModal(false); } diff --git a/servers/nextjs/app/storeInitializer.tsx b/servers/nextjs/app/storeInitializer.tsx index 44fd16f8..13ac3153 100644 --- a/servers/nextjs/app/storeInitializer.tsx +++ b/servers/nextjs/app/storeInitializer.tsx @@ -53,7 +53,7 @@ export function StoreInitializer({ children }: { children: React.ReactNode }) { } } if (llmConfig.LLM === 'custom') { - const isAvailable = await checkIfSelectedCustomModelIsAvailable(llmConfig.CUSTOM_MODEL); + const isAvailable = await checkIfSelectedCustomModelIsAvailable(llmConfig); if (!isAvailable) { router.push('/'); setLoadingToFalseAfterNavigatingTo('/'); @@ -83,16 +83,20 @@ export function StoreInitializer({ children }: { children: React.ReactNode }) { } - const checkIfSelectedCustomModelIsAvailable = async (customModel: string) => { + const checkIfSelectedCustomModelIsAvailable = async (llmConfig: LLMConfig) => { try { - const response = await fetch('/api/v1/ppt/custom_llm/models/available', { + const response = await fetch('/api/v1/ppt/openai/models/available', { method: 'POST', headers: { 'Content-Type': 'application/json', }, + body: JSON.stringify({ + url: llmConfig.CUSTOM_LLM_URL, + api_key: llmConfig.CUSTOM_LLM_API_KEY, + }), }); const data = await response.json(); - return data.includes(customModel); + return data.includes(llmConfig.CUSTOM_MODEL); } catch (error) { console.error('Error fetching custom models:', error); return false; diff --git a/servers/nextjs/components/AnthropicConfig.tsx b/servers/nextjs/components/AnthropicConfig.tsx new file mode 100644 index 00000000..567846b5 --- /dev/null +++ b/servers/nextjs/components/AnthropicConfig.tsx @@ -0,0 +1,231 @@ +"use client"; +import { useEffect, useState } from "react"; +import { Check, ChevronsUpDown, Loader2 } from "lucide-react"; +import { Button } from "./ui/button"; +import { + Command, + CommandEmpty, + CommandGroup, + CommandInput, + CommandItem, + CommandList, +} from "./ui/command"; +import { Popover, PopoverContent, PopoverTrigger } from "./ui/popover"; +import { cn } from "@/lib/utils"; +import { toast } from "sonner"; +import { Switch } from "./ui/switch"; + +interface AnthropicConfigProps { + anthropicApiKey: string; + anthropicModel: string; + extendedReasoning: boolean; + onInputChange: (value: string | boolean, field: string) => void; +} + + +export default function AnthropicConfig({ + anthropicApiKey, + anthropicModel, + extendedReasoning, + onInputChange, +}: AnthropicConfigProps) { + const [openModelSelect, setOpenModelSelect] = useState(false); + const [availableModels, setAvailableModels] = useState([]); + const [modelsLoading, setModelsLoading] = useState(false); + const [modelsChecked, setModelsChecked] = useState(false); + const [apiKey, setApiKey] = useState(anthropicApiKey); + + useEffect(() => { + setAvailableModels([]); + setModelsChecked(false); + onInputChange("", "anthropic_model"); + }, [apiKey]); + + const onApiKeyChange = (value: string) => { + setApiKey(value); + onInputChange(value, "anthropic_api_key"); + }; + + const fetchAvailableModels = async () => { + if (!anthropicApiKey) return; + + setModelsLoading(true); + try { + const response = await fetch('/api/v1/ppt/anthropic/models/available', { + method: 'POST', + headers: { + 'Content-Type': 'application/json', + }, + body: JSON.stringify({ + api_key: anthropicApiKey + }), + }); + + if (response.ok) { + const data = await response.json(); + setAvailableModels(data); + setModelsChecked(true); + onInputChange("claude-3-5-sonnet-20241022", "anthropic_model"); + } else { + console.error('Failed to fetch models'); + setAvailableModels([]); + setModelsChecked(true); + } + } catch (error) { + console.error('Error fetching models:', error); + toast.error('Error fetching models'); + setAvailableModels([]); + setModelsChecked(true); + } finally { + setModelsLoading(false); + } + }; + + return ( +
+ {/* API Key Input */} +
+ +
+ onApiKeyChange(e.target.value)} + className="w-full px-4 py-2.5 outline-none border border-gray-300 rounded-lg focus:ring-2 focus:ring-blue-500/20 focus:border-blue-500 transition-colors" + placeholder="Enter your Anthropic API key" + /> +
+

+ + Your API key will be stored locally and never shared +

+
+ + {/* Extended Reasoning Toggle */} + {/*
+
+ + onInputChange(checked, "extended_reasoning")} + /> +
+

+ + Enable extended reasoning for more detailed and thorough responses +

+
*/} + + {/* Check for available models button - show when no models checked or no models found */} + {(!modelsChecked || (modelsChecked && availableModels.length === 0)) && ( +
+ +
+ )} + + {/* Show message if no models found */} + {modelsChecked && availableModels.length === 0 && ( +
+

+ No models found. Please make sure your API key is valid and has access to Anthropic models. +

+
+ )} + + {/* Model Selection - only show if models are available */} + {modelsChecked && availableModels.length > 0 ? ( +
+ +
+ + + + + + + + + No model found. + + {availableModels.map((model, index) => ( + { + onInputChange(value, "anthropic_model"); + setOpenModelSelect(false); + }} + > + +
+
+
+ + {model} + +
+
+
+ + ))} + + + + + +
+
+ ) : null} +
+ ); +} \ No newline at end of file diff --git a/servers/nextjs/components/CustomConfig.tsx b/servers/nextjs/components/CustomConfig.tsx index f0a9cb4c..41e26fda 100644 --- a/servers/nextjs/components/CustomConfig.tsx +++ b/servers/nextjs/components/CustomConfig.tsx @@ -1,4 +1,5 @@ "use client"; +import { useState, useEffect } from "react"; import { Check, ChevronsUpDown, Loader2 } from "lucide-react"; import { Button } from "./ui/button"; import { @@ -11,34 +12,83 @@ import { } from "./ui/command"; import { Popover, PopoverContent, PopoverTrigger } from "./ui/popover"; import { cn } from "@/lib/utils"; +import { toast } from "sonner"; interface CustomConfigProps { customLlmUrl: string; customLlmApiKey: string; customModel: string; - customModels: string[]; - customModelsLoading: boolean; - customModelsChecked: boolean; - openModelSelect: boolean; onInputChange: (value: string, field: string) => void; - onOpenModelSelectChange: (open: boolean) => void; - onFetchCustomModels: () => void; } export default function CustomConfig({ customLlmUrl, customLlmApiKey, customModel, - customModels, - customModelsLoading, - customModelsChecked, - openModelSelect, onInputChange, - onOpenModelSelectChange, - onFetchCustomModels, }: CustomConfigProps) { + const [customModels, setCustomModels] = useState([]); + const [customModelsLoading, setCustomModelsLoading] = useState(false); + const [customModelsChecked, setCustomModelsChecked] = useState(false); + const [openModelSelect, setOpenModelSelect] = useState(false); + const [url, setUrl] = useState(customLlmUrl); + const [apiKey, setApiKey] = useState(customLlmApiKey); + + useEffect(() => { + setCustomModels([]); + setCustomModelsChecked(false); + onInputChange("", "custom_model"); + }, [url, apiKey]); + + const onUrlChange = (value: string) => { + setUrl(value); + onInputChange(value, "custom_llm_url"); + }; + + const onApiKeyChange = (value: string) => { + setApiKey(value); + onInputChange(value, "custom_llm_api_key"); + }; + + const fetchCustomModels = async () => { + if (!customLlmUrl) return; + + try { + setCustomModelsLoading(true); + const response = await fetch("/api/v1/ppt/openai/models/available", { + method: "POST", + headers: { + "Content-Type": "application/json", + }, + body: JSON.stringify({ + url: customLlmUrl, + api_key: customLlmApiKey, + }), + }); + + if (response.ok) { + const data = await response.json(); + setCustomModels(data); + setCustomModelsChecked(true); + } else { + console.error('Failed to fetch custom models'); + setCustomModels([]); + setCustomModelsChecked(true); + toast.error('Failed to fetch custom models'); + } + } catch (error) { + console.error('Error fetching custom models:', error); + toast.error('Error fetching custom models'); + setCustomModels([]); + setCustomModelsChecked(true); + } finally { + setCustomModelsLoading(false); + } + }; + return ( - <> +
+ {/* URL Input */}
+ + {/* API Key Input */}
+ {/* Check for available models button - show when no models checked or no models found */} + {(!customModelsChecked || (customModelsChecked && customModels.length === 0)) && ( +
+ +
+ )} + + {/* Show message if no models found */} + {customModelsChecked && customModels.length === 0 && ( +
+

+ No models found. Please make sure your API key is valid and has access to models. +

+
+ )} + {/* Model selection dropdown - only show if models are available */} {customModelsChecked && customModels.length > 0 && (
@@ -90,7 +170,7 @@ export default function CustomConfig({
)} - - {/* Check for available models button - show when no models checked or no models found */} - {(!customModelsChecked || - (customModelsChecked && customModels.length === 0)) && ( -
- -
- )} - - {/* Show message if no models found */} - {customModelsChecked && customModels.length === 0 && ( -
-

- No models found. Please make sure models are available. -

-
- )} - +
); } \ No newline at end of file diff --git a/servers/nextjs/components/GoogleConfig.tsx b/servers/nextjs/components/GoogleConfig.tsx index d595fbd4..6746f779 100644 --- a/servers/nextjs/components/GoogleConfig.tsx +++ b/servers/nextjs/components/GoogleConfig.tsx @@ -1,27 +1,210 @@ +"use client"; +import { useEffect, useState } from "react"; +import { Check, ChevronsUpDown, Loader2 } from "lucide-react"; +import { Button } from "./ui/button"; +import { + Command, + CommandEmpty, + CommandGroup, + CommandInput, + CommandItem, + CommandList, +} from "./ui/command"; +import { Popover, PopoverContent, PopoverTrigger } from "./ui/popover"; +import { cn } from "@/lib/utils"; +import { toast } from "sonner"; + interface GoogleConfigProps { googleApiKey: string; + googleModel: string; onInputChange: (value: string, field: string) => void; } -export default function GoogleConfig({ googleApiKey, onInputChange }: GoogleConfigProps) { +export default function GoogleConfig({ + googleApiKey, + googleModel, + onInputChange +}: GoogleConfigProps) { + const [openModelSelect, setOpenModelSelect] = useState(false); + const [availableModels, setAvailableModels] = useState([]); + const [modelsLoading, setModelsLoading] = useState(false); + const [modelsChecked, setModelsChecked] = useState(false); + const [apiKey, setApiKey] = useState(googleApiKey); + + useEffect(() => { + setAvailableModels([]); + setModelsChecked(false); + onInputChange("", "google_model"); + }, [apiKey]); + + const onApiKeyChange = (value: string) => { + setApiKey(value); + onInputChange(value, "google_api_key"); + }; + + const fetchAvailableModels = async () => { + if (!googleApiKey) return; + + setModelsLoading(true); + try { + const response = await fetch('/api/v1/ppt/google/models/available', { + method: 'POST', + headers: { + 'Content-Type': 'application/json', + }, + body: JSON.stringify({ + api_key: googleApiKey + }), + }); + + if (response.ok) { + const data = await response.json(); + setAvailableModels(data); + setModelsChecked(true); + onInputChange("models/gemini-2.0-flash", "google_model"); + } else { + console.error('Failed to fetch models'); + setAvailableModels([]); + setModelsChecked(true); + } + } catch (error) { + console.error('Error fetching models:', error); + toast.error('Error fetching models'); + setAvailableModels([]); + setModelsChecked(true); + } finally { + setModelsLoading(false); + } + }; + return ( -
- -
- onInputChange(e.target.value, "google_api_key")} - className="w-full px-4 py-2.5 outline-none border border-gray-300 rounded-lg focus:ring-2 focus:ring-blue-500/20 focus:border-blue-500 transition-colors" - placeholder="Enter your API key" - /> +
+ {/* API Key Input */} +
+ +
+ onApiKeyChange(e.target.value)} + className="w-full px-4 py-2.5 outline-none border border-gray-300 rounded-lg focus:ring-2 focus:ring-blue-500/20 focus:border-blue-500 transition-colors" + placeholder="Enter your API key" + /> +
+

+ + Your API key will be stored locally and never shared +

-

- - Your API key will be stored locally and never shared -

+ + {/* Check for available models button - show when no models checked or no models found */} + {(!modelsChecked || (modelsChecked && availableModels.length === 0)) && ( +
+ +
+ )} + + {/* Show message if no models found */} + {modelsChecked && availableModels.length === 0 && ( +
+

+ No models found. Please make sure your API key is valid and has access to Google models. +

+
+ )} + + {/* Model Selection - only show if models are available */} + {modelsChecked && availableModels.length > 0 ? ( +
+ +
+ + + + + + + + + No model found. + + {availableModels.map((model, index) => ( + { + onInputChange(value, "google_model"); + setOpenModelSelect(false); + }} + > + +
+
+
+ + {model} + +
+
+
+ + ))} + + + + + +
+
+ ) : null}
); } \ No newline at end of file diff --git a/servers/nextjs/components/Home.tsx b/servers/nextjs/components/Home.tsx index f3941d1d..059219b5 100644 --- a/servers/nextjs/components/Home.tsx +++ b/servers/nextjs/components/Home.tsx @@ -27,7 +27,6 @@ export default function Home() { const router = useRouter(); const config = useSelector((state: RootState) => state.userConfig); const [llmConfig, setLlmConfig] = useState(config.llm_config); - const [isLoading, setIsLoading] = useState(false); const [downloadingModel, setDownloadingModel] = useState<{ name: string; @@ -54,7 +53,6 @@ export default function Home() { const handleSaveConfig = async () => { try { - setIsLoading(true); setButtonState(prev => ({ ...prev, isLoading: true, @@ -70,7 +68,6 @@ export default function Home() { } } toast.info("Configuration saved successfully"); - setIsLoading(false); setButtonState(prev => ({ ...prev, isLoading: false, @@ -79,8 +76,7 @@ export default function Home() { })); router.push("/upload"); } catch (error) { - toast.info("Failed to save configuration"); - setIsLoading(false); + toast.info(error instanceof Error ? error.message : "Failed to save configuration"); setButtonState(prev => ({ ...prev, isLoading: false, @@ -93,8 +89,8 @@ export default function Home() { const handleModelDownload = async () => { try { await pullOllamaModel(llmConfig.OLLAMA_MODEL!, setDownloadingModel); - } catch (error) { - console.info("Error downloading model:", error); + } + finally { setDownloadingModel(null); setShowDownloadModal(false); } diff --git a/servers/nextjs/components/LLMSelection.tsx b/servers/nextjs/components/LLMSelection.tsx index 670d1c00..2a83e4c5 100644 --- a/servers/nextjs/components/LLMSelection.tsx +++ b/servers/nextjs/components/LLMSelection.tsx @@ -15,16 +15,13 @@ import { Popover, PopoverContent, PopoverTrigger } from "./ui/popover"; import { cn } from "@/lib/utils"; import OpenAIConfig from "./OpenAIConfig"; import GoogleConfig from "./GoogleConfig"; +import AnthropicConfig from "./AnthropicConfig"; import OllamaConfig from "./OllamaConfig"; import CustomConfig from "./CustomConfig"; import { - OllamaModel, LLMConfig, updateLLMConfig, changeProvider as changeProviderUtil, - fetchOllamaModelsWithConfig, - setOllamaConfig, - fetchCustomModels, } from "@/utils/providerUtils"; import { IMAGE_PROVIDERS, LLM_PROVIDERS } from "@/utils/providerConstants"; @@ -51,15 +48,6 @@ export default function LLMProviderSelection({ setButtonState, }: LLMProviderSelectionProps) { const [llmConfig, setLlmConfig] = useState(initialLLMConfig); - const [ollamaModels, setOllamaModels] = useState([]); - const [customModels, setCustomModels] = useState([]); - const [customModelsLoading, setCustomModelsLoading] = useState(false); - const [customModelsChecked, setCustomModelsChecked] = useState(false); - const [ollamaModelsLoading, setOllamaModelsLoading] = useState(false); - const [useCustomOllamaUrl, setUseCustomOllamaUrl] = useState( - initialLLMConfig.USE_CUSTOM_URL || false - ); - const [openModelSelect, setOpenModelSelect] = useState(false); const [openImageProviderSelect, setOpenImageProviderSelect] = useState(false); useEffect(() => { @@ -68,25 +56,31 @@ export default function LLMProviderSelection({ useEffect(() => { const needsModelSelection = + (llmConfig.LLM === "openai" && !llmConfig.OPENAI_MODEL) || + (llmConfig.LLM === "google" && !llmConfig.GOOGLE_MODEL) || (llmConfig.LLM === "ollama" && !llmConfig.OLLAMA_MODEL) || - (llmConfig.LLM === "custom" && !llmConfig.CUSTOM_MODEL); + (llmConfig.LLM === "custom" && !llmConfig.CUSTOM_MODEL) || + (llmConfig.LLM === "anthropic" && !llmConfig.ANTHROPIC_MODEL); const needsApiKey = ((llmConfig.IMAGE_PROVIDER === "dall-e-3" || llmConfig.LLM === "openai") && !llmConfig.OPENAI_API_KEY) || ((llmConfig.IMAGE_PROVIDER === "gemini_flash" || llmConfig.LLM === "google") && !llmConfig.GOOGLE_API_KEY) || + (llmConfig.LLM === "anthropic" && !llmConfig.ANTHROPIC_API_KEY) || (llmConfig.IMAGE_PROVIDER === "pexels" && !llmConfig.PEXELS_API_KEY) || (llmConfig.IMAGE_PROVIDER === "pixabay" && !llmConfig.PIXABAY_API_KEY); + const needsOllamaUrl = (llmConfig.LLM === "ollama" && !llmConfig.OLLAMA_URL); + setButtonState({ isLoading: false, - isDisabled: needsModelSelection || needsApiKey, - text: needsModelSelection ? "Please Select a Model" : needsApiKey ? "Please Enter API Key" : "Save Configuration", + isDisabled: needsModelSelection || needsApiKey || needsOllamaUrl, + text: needsModelSelection ? "Please Select a Model" : needsApiKey ? "Please Enter API Key" : needsOllamaUrl ? "Please Enter Ollama URL" : "Save Configuration", showProgress: false }); }, [llmConfig]); - const input_field_changed = (new_value: string, field: string) => { + const input_field_changed = (new_value: string | boolean, field: string) => { const updatedConfig = updateLLMConfig(llmConfig, field, new_value); setLlmConfig(updatedConfig); }; @@ -94,78 +88,33 @@ export default function LLMProviderSelection({ const handleProviderChange = (provider: string) => { const newConfig = changeProviderUtil(llmConfig, provider); setLlmConfig(newConfig); - if (provider === "ollama") { - fetchOllamaModels(); - } }; - const fetchOllamaModels = async () => { - try { - setOllamaModelsLoading(true); - const result = await fetchOllamaModelsWithConfig(llmConfig); - setOllamaModels(result.models); - if (result.updatedConfig) { - setLlmConfig(result.updatedConfig); + useEffect(() => { + if (!llmConfig.USE_CUSTOM_URL) { + setLlmConfig({ ...llmConfig, OLLAMA_URL: "http://localhost:11434" }); + } else { + if (!llmConfig.OLLAMA_URL) { + setLlmConfig({ ...llmConfig, OLLAMA_URL: "http://localhost:11434" }); } - } catch (error) { - console.error("Error fetching Ollama models:", error); - setOllamaModels([]); - } finally { - setOllamaModelsLoading(false); } - }; - - const fetchCustomModelsHandler = async () => { - try { - setCustomModelsLoading(true); - const models = await fetchCustomModels( - llmConfig.CUSTOM_LLM_URL || "", - llmConfig.CUSTOM_LLM_API_KEY || "" - ); - setCustomModels(models); - setCustomModelsChecked(true); - } catch (error) { - console.error("Error fetching custom models:", error); - setCustomModels([]); - } finally { - setCustomModelsLoading(false); - } - }; - - const setOllamaConfigHandler = () => { - const updatedConfig = setOllamaConfig(llmConfig, useCustomOllamaUrl); - setLlmConfig(updatedConfig); - }; - - useEffect(() => { - if (llmConfig.LLM === "ollama") { - setOllamaConfigHandler(); - fetchOllamaModels(); - } - }, [llmConfig.LLM]); - - useEffect(() => { - setOllamaConfigHandler(); - }, [useCustomOllamaUrl]); - - useEffect(() => { - if (llmConfig.LLM === "custom") { - setCustomModels([]); - setCustomModelsChecked(false); - setLlmConfig({ ...llmConfig, CUSTOM_MODEL: "" }); - } - }, [llmConfig.CUSTOM_LLM_URL, llmConfig.CUSTOM_LLM_API_KEY]); + }, [llmConfig.USE_CUSTOM_URL]); useEffect(() => { + let updates: any = {}; if (!llmConfig.IMAGE_PROVIDER) { if (llmConfig.LLM === "openai") { - setLlmConfig({ ...llmConfig, IMAGE_PROVIDER: "dall-e-3" }); + updates.IMAGE_PROVIDER = "dall-e-3"; } else if (llmConfig.LLM === "google") { - setLlmConfig({ ...llmConfig, IMAGE_PROVIDER: "gemini_flash" }); + updates.IMAGE_PROVIDER = "gemini_flash"; } else { - setLlmConfig({ ...llmConfig, IMAGE_PROVIDER: "pexels" }); + updates.IMAGE_PROVIDER = "pexels"; } } + if (!llmConfig.OLLAMA_URL) { + updates.OLLAMA_URL = "http://localhost:11434"; + } + setLlmConfig({ ...llmConfig, ...updates }); }, []); return ( @@ -177,9 +126,10 @@ export default function LLMProviderSelection({ onValueChange={handleProviderChange} className="w-full" > - + OpenAI Google + Anthropic Ollama Custom @@ -198,6 +148,7 @@ export default function LLMProviderSelection({ @@ -206,6 +157,17 @@ export default function LLMProviderSelection({ + + + {/* Anthropic Content */} + + @@ -215,16 +177,8 @@ export default function LLMProviderSelection({ { - input_field_changed(modelName, "ollama_model"); - }} /> @@ -234,19 +188,13 @@ export default function LLMProviderSelection({ customLlmUrl={llmConfig.CUSTOM_LLM_URL || ""} customLlmApiKey={llmConfig.CUSTOM_LLM_API_KEY || ""} customModel={llmConfig.CUSTOM_MODEL || ""} - customModels={customModels} - customModelsLoading={customModelsLoading} - customModelsChecked={customModelsChecked} - openModelSelect={openModelSelect} onInputChange={input_field_changed} - onOpenModelSelectChange={setOpenModelSelect} - onFetchCustomModels={fetchCustomModelsHandler} /> {/* Image Provider Selection */} -
+
@@ -388,7 +336,13 @@ export default function LLMProviderSelection({ ? llmConfig.OLLAMA_MODEL ?? "xxxxx" : llmConfig.LLM === "custom" ? llmConfig.CUSTOM_MODEL ?? "xxxxx" - : LLM_PROVIDERS[llmConfig.LLM!]?.model_label || "xxxxx"}{" "} + : llmConfig.LLM === "anthropic" + ? llmConfig.ANTHROPIC_MODEL ?? "xxxxx" + : llmConfig.LLM === "google" + ? llmConfig.GOOGLE_MODEL ?? "xxxxx" + : llmConfig.LLM === "openai" + ? llmConfig.OPENAI_MODEL ?? "xxxxx" + : "xxxxx"}{" "} for text generation and{" "} {llmConfig.IMAGE_PROVIDER && IMAGE_PROVIDERS[llmConfig.IMAGE_PROVIDER] diff --git a/servers/nextjs/components/OllamaConfig.tsx b/servers/nextjs/components/OllamaConfig.tsx index f0f09b76..aab9b880 100644 --- a/servers/nextjs/components/OllamaConfig.tsx +++ b/servers/nextjs/components/OllamaConfig.tsx @@ -1,5 +1,5 @@ "use client"; -import { useState } from "react"; +import { useState, useEffect } from "react"; import { Check, ChevronsUpDown, Loader2 } from "lucide-react"; import { Button } from "./ui/button"; import { @@ -13,6 +13,7 @@ import { import { Popover, PopoverContent, PopoverTrigger } from "./ui/popover"; import { cn } from "@/lib/utils"; import { Switch } from "./ui/switch"; +import { toast } from "sonner"; interface OllamaModel { label: string; @@ -26,30 +27,84 @@ interface OllamaConfigProps { ollamaModel: string; ollamaUrl: string; useCustomUrl: boolean; - ollamaModels: OllamaModel[]; - ollamaModelsLoading?: boolean; - onInputChange: (value: string, field: string) => void; - onUseCustomUrlChange: (checked: boolean) => void; - openModelSelect: boolean; - onOpenModelSelectChange: (open: boolean) => void; - onModelSelect?: (modelName: string) => void; + onInputChange: (value: string | boolean, field: string) => void; } export default function OllamaConfig({ ollamaModel, ollamaUrl, useCustomUrl, - ollamaModels, - ollamaModelsLoading = false, onInputChange, - onUseCustomUrlChange, - openModelSelect, - onOpenModelSelectChange, - onModelSelect, }: OllamaConfigProps) { + const [ollamaModels, setOllamaModels] = useState([]); + const [ollamaModelsLoading, setOllamaModelsLoading] = useState(false); + const [openModelSelect, setOpenModelSelect] = useState(false); + + const fetchOllamaModels = async () => { + try { + setOllamaModelsLoading(true); + const response = await fetch('/api/v1/ppt/ollama/models/supported'); + + if (response.ok) { + const data = await response.json(); + console.log(data); + setOllamaModels(data); + } else { + console.error('Failed to fetch Ollama models'); + setOllamaModels([]); + toast.error('Failed to fetch Ollama models'); + } + } catch (error) { + console.error('Error fetching Ollama models:', error); + toast.error('Error fetching Ollama models'); + setOllamaModels([]); + } finally { + setOllamaModelsLoading(false); + } + }; + + useEffect(() => { + fetchOllamaModels(); + }, []); + return ( - <> -
+
+ {/* URL Configuration */} +
+
+ + onInputChange(checked, "use_custom_url")} + /> +
+ {useCustomUrl && ( +
+ +
+ onInputChange(e.target.value, "ollama_url")} + /> +
+

+ + Change this if you are using a custom Ollama instance +

+
+ )} +
+ + {/* Model Selection */} +
@@ -64,7 +119,7 @@ export default function OllamaConfig({ ) : ollamaModels && ollamaModels.length > 0 ? (
-
-
- - -
- {useCustomUrl && ( - <> -
- -
- - onInputChange(e.target.value, "ollama_url") - } - /> -
-

- - Change this if you are using a custom Ollama instance -

-
- - )} -
- +
); } \ No newline at end of file diff --git a/servers/nextjs/components/OpenAIConfig.tsx b/servers/nextjs/components/OpenAIConfig.tsx index 6696493a..b73695e9 100644 --- a/servers/nextjs/components/OpenAIConfig.tsx +++ b/servers/nextjs/components/OpenAIConfig.tsx @@ -1,27 +1,215 @@ +"use client"; +import { useEffect, useState } from "react"; +import { Check, ChevronsUpDown, Loader2 } from "lucide-react"; +import { Button } from "./ui/button"; +import { + Command, + CommandEmpty, + CommandGroup, + CommandInput, + CommandItem, + CommandList, +} from "./ui/command"; +import { Popover, PopoverContent, PopoverTrigger } from "./ui/popover"; +import { cn } from "@/lib/utils"; +import { toast } from "sonner"; + interface OpenAIConfigProps { openaiApiKey: string; + openaiModel: string; onInputChange: (value: string, field: string) => void; } -export default function OpenAIConfig({ openaiApiKey, onInputChange }: OpenAIConfigProps) { +export default function OpenAIConfig({ + openaiApiKey, + openaiModel, + onInputChange +}: OpenAIConfigProps) { + const [openModelSelect, setOpenModelSelect] = useState(false); + const [availableModels, setAvailableModels] = useState([]); + const [modelsLoading, setModelsLoading] = useState(false); + const [modelsChecked, setModelsChecked] = useState(false); + const [apiKey, setApiKey] = useState(openaiApiKey); + + const openaiUrl = "https://api.openai.com/v1"; + + useEffect(() => { + setAvailableModels([]); + setModelsChecked(false); + onInputChange("", "openai_model"); + }, [apiKey]); + + const onApiKeyChange = (value: string) => { + setApiKey(value); + onInputChange(value, "openai_api_key"); + }; + + const fetchAvailableModels = async () => { + if (!openaiApiKey) return; + + setModelsLoading(true); + try { + const response = await fetch('/api/v1/ppt/openai/models/available', { + method: 'POST', + headers: { + 'Content-Type': 'application/json', + }, + body: JSON.stringify({ + url: openaiUrl, + api_key: openaiApiKey + }), + }); + + if (response.ok) { + const data = await response.json(); + setAvailableModels(data); + setModelsChecked(true); + onInputChange("gpt-4.1", "openai_model"); + } else { + console.error('Failed to fetch models'); + setAvailableModels([]); + setModelsChecked(true); + } + } catch (error) { + console.error('Error fetching models:', error); + toast.error('Error fetching models'); + setAvailableModels([]); + setModelsChecked(true); + } finally { + setModelsLoading(false); + } + }; + return ( -
- -
- onInputChange(e.target.value, "openai_api_key")} - className="w-full px-4 py-2.5 outline-none border border-gray-300 rounded-lg focus:ring-2 focus:ring-blue-500/20 focus:border-blue-500 transition-colors" - placeholder="Enter your API key" - /> +
+ {/* API Key Input */} +
+ +
+ onApiKeyChange(e.target.value)} + className="w-full px-4 py-2.5 outline-none border border-gray-300 rounded-lg focus:ring-2 focus:ring-blue-500/20 focus:border-blue-500 transition-colors" + placeholder="Enter your API key" + /> +
+

+ + Your API key will be stored locally and never shared +

-

- - Your API key will be stored locally and never shared -

+ + + + {/* Check for available models button - show when no models checked or no models found */} + {(!modelsChecked || (modelsChecked && availableModels.length === 0)) && ( +
+ +
+ )} + + {/* Show message if no models found */} + {modelsChecked && availableModels.length === 0 && ( +
+

+ No models found. Please make sure your API key is valid and has access to OpenAI models. +

+
+ )} + + {/* Model Selection - only show if models are available */} + {modelsChecked && availableModels.length > 0 ? ( +
+ +
+ + + + + + + + + No model found. + + {availableModels.map((model, index) => ( + { + onInputChange(value, "openai_model"); + setOpenModelSelect(false); + }} + > + +
+
+
+ + {model} + +
+
+
+ + ))} + + + + + +
+
+ ) : null}
); -} \ No newline at end of file +} \ No newline at end of file diff --git a/servers/nextjs/presentation-layouts/classic-dark/1-TitleSlide.tsx b/servers/nextjs/presentation-layouts/classic-dark/1-TitleSlide.tsx new file mode 100644 index 00000000..269466de --- /dev/null +++ b/servers/nextjs/presentation-layouts/classic-dark/1-TitleSlide.tsx @@ -0,0 +1,87 @@ +import React from 'react' +import * as z from "zod"; +import { ImageSchema } from '@/presentation-layouts/defaultSchemes'; + +export const layoutId = 'classic-dark-title-slide' +export const layoutName = 'Classic Dark Title Slide' +export const layoutDescription = 'A modern title slide with dark gradient background, gradient text, and geographical elements.' + +const titleSlideSchema = z.object({ + title: z.string().min(3).max(100).default('Nepal\'s Imports and\nExports: A Data-Driven\nOverview').meta({ + description: "Main title of the slide", + }), + subtitle: z.string().min(3).max(100).default('Key Trade Statistics and Trends (2022–2025)').meta({ + description: "Subtitle text", + }), + presenter: z.string().min(3).max(50).default('[Your Name]').meta({ + description: "Presenter name", + }), + date: z.string().min(3).max(50).default('April 13, 2025').meta({ + description: "Presentation date", + }), + image: ImageSchema.default({ + __image_url__: 'https://images.pexels.com/photos/9669089/pexels-photo-9669089.jpeg', + __image_prompt__: 'Map of Nepal with gradient coloring from orange to red-brown' + }).meta({ + description: "Image of the title slide of the presentation", + }), +}) + +export const Schema = titleSlideSchema + +export type TitleSlideData = z.infer + +interface TitleSlideLayoutProps { + data: Partial +} + +const TitleSlideLayout: React.FC = ({ data: slideData }) => { + const { title, subtitle, presenter, date, image } = slideData; + + return ( +
+ +
+ {/* Left side - Text content */} +
+ {/* Title */} + {title && ( +

+ {title} +

+ )} + + {/* Subtitle */} + {subtitle && ( +

+ {subtitle} +

+ )} + + {/* Presenter and date */} +
+ Presenter: {presenter} | {date} +
+
+ + {/* Right side - Visual elements */} +
+ {image && ( +
+ {image.__image_prompt__} +
+ )} +
+
+
+ ) +} + +export default TitleSlideLayout diff --git a/servers/nextjs/presentation-layouts/classic-dark/2-ChartAndMetrics.tsx b/servers/nextjs/presentation-layouts/classic-dark/2-ChartAndMetrics.tsx new file mode 100644 index 00000000..806dac67 --- /dev/null +++ b/servers/nextjs/presentation-layouts/classic-dark/2-ChartAndMetrics.tsx @@ -0,0 +1,160 @@ +import React from 'react' +import * as z from "zod"; +import { ChartContainer, ChartLegend, ChartLegendContent, ChartTooltip, ChartTooltipContent } from '@/components/ui/chart'; +import { PieChart, Pie, Cell } from "recharts"; + +export const layoutId = 'classic-dark-piechart-and-metrics' +export const layoutName = 'Classic Dark Pie Chart and Metrics' +export const layoutDescription = 'A modern slide with dark background, metrics on the left, and pie chart visualization on the right.' + +const chartDataSchema = z.object({ + name: z.string().meta({ description: "Data point name" }), + value: z.number().meta({ description: "Data point value" }), +}); + +const pieChartAndMetricsSchema = z.object({ + title: z.string().min(3).max(100).default('Introduction to Nepal\'s Trade').meta({ + description: "Main title of the slide", + }), + description: z.string().min(10).max(200).default('Nepal\'s landlocked geography heavily influences its trade, fostering reliance on India and China.').meta({ + description: "Description text", + }), + metrics: z.array(z.object({ + label: z.string().meta({ description: "Metric label" }), + value: z.string().meta({ description: "Metric value" }), + percentage: z.string().optional().meta({ description: "Optional percentage" }), + })).min(2).max(4).default([ + { label: 'Exports (2023)', value: '$2.85 billion', percentage: '6.76% of GDP' }, + { label: 'Imports (2023)', value: '$17.39 billion', percentage: '42.64% of GDP' }, + { label: 'GDP (2022)', value: '$40.83 billion' }, + { label: 'Trade Deficit (2022)', value: '-$12.44 billion' }, + ]).meta({ + description: "List of key metrics", + }), + chartData: z.array(chartDataSchema).min(2).max(4).default([ + { name: 'Imports', value: 42.64 }, + { name: 'Exports', value: 6.76 }, + { name: 'Other GDP', value: 50.6 }, + ]).meta({ + description: "Pie chart data", + }), + showLegend: z.boolean().default(true).meta({ + description: "Whether to show chart legend", + }), + showTooltip: z.boolean().default(true).meta({ + description: "Whether to show chart tooltip", + }), +}) + +const chartConfig = { + value: { + label: "Value", + }, + name: { + label: "Name", + }, +}; + +const CHART_COLORS = [ + '#8b5cf6', + '#3b82f6', + '#a855f7', +]; + +export const Schema = pieChartAndMetricsSchema + +export type PieChartAndMetricsData = z.infer + +interface PieChartAndMetricsLayoutProps { + data: Partial +} + +const PieChartAndMetricsLayout: React.FC = ({ data: slideData }) => { + const { title, description, metrics, chartData, showLegend = true, showTooltip = true } = slideData; + + const CustomLegend = () => ( +
+ {chartData?.map((entry, index) => ( +
+
+ {entry.name} +
+ ))} +
+ ); + + const renderPieChart = () => { + return ( + + {showTooltip && } />} + `${name} ${(percent * 100).toFixed(0)}%`} + labelLine={false} + fontSize={18} + > + {chartData?.map((entry, index) => ( + + ))} + + + ); + }; + + return ( +
+ +
+ {/* Left side - Text content and metrics */} +
+ {/* Title */} + {title && ( +

+ {title} +

+ )} + + {/* Description */} + {description && ( +

+ {description} +

+ )} + + {/* Metrics */} + {metrics && metrics.length > 0 && ( +
+ {metrics.map((metric, index) => ( +
+ • {metric.label}: + {metric.value} + {metric.percentage && ( + ({metric.percentage}) + )} +
+ ))} +
+ )} +
+ + {/* Right side - Chart */} +
+
+ + {renderPieChart()} + + {showLegend && } +
+
+
+
+ ) +} + +export default PieChartAndMetricsLayout \ No newline at end of file diff --git a/servers/nextjs/presentation-layouts/classic-dark/3-BarGraph.tsx b/servers/nextjs/presentation-layouts/classic-dark/3-BarGraph.tsx new file mode 100644 index 00000000..7f73ef69 --- /dev/null +++ b/servers/nextjs/presentation-layouts/classic-dark/3-BarGraph.tsx @@ -0,0 +1,152 @@ +import React from 'react' +import * as z from "zod"; +import { ChartContainer, ChartLegend, ChartLegendContent, ChartTooltip, ChartTooltipContent } from '@/components/ui/chart'; +import { BarChart, Bar, XAxis, YAxis, CartesianGrid, Cell } from "recharts"; + +export const layoutId = 'classic-dark-bar-graph' +export const layoutName = 'Classic Dark Bar Graph' +export const layoutDescription = 'A modern slide with dark background, gradient title, bar chart visualization, and footer text.' + +const barDataSchema = z.object({ + name: z.string().meta({ description: "Product name" }), + value: z.number().meta({ description: "Export value in millions" }), +}); + +const barGraphSchema = z.object({ + title: z.string().min(3).max(100).default('Export Overview: Key Products').meta({ + description: "Main title of the slide", + }), + description: z.string().min(10).max(150).default('Nepal\'s total exports were $1.3 billion in 2022, a 21% decrease from 2021, but showed a 47.5% YoY increase by Nov 2024.').meta({ + description: "Description text", + }), + chartData: z.array(barDataSchema).min(2).max(6).default([ + { name: 'Soybean Oil (non-crude)', value: 180 }, + { name: 'Palm Oil (non-crude)', value: 180 }, + { name: 'Carpets/Textile Floor...', value: 80 }, + { name: 'Cardamom', value: 50 }, + { name: 'Felt Products', value: 40 }, + ]).meta({ + description: "Bar chart data", + }), + showLegend: z.boolean().default(true).meta({ + description: "Whether to show chart legend", + }), + showTooltip: z.boolean().default(true).meta({ + description: "Whether to show chart tooltip", + }), +}) + +const chartConfig = { + value: { + label: "Value ($M)", + }, + name: { + label: "Product", + }, +}; + +const BAR_COLORS = [ + '#8b5cf6', // Dark purple for top products + '#8b5cf6', // Dark purple for top products + '#a855f7', // Light purple for other products + '#a855f7', // Light purple for other products + '#a855f7', // Light purple for other products +]; + +export const Schema = barGraphSchema + +export type BarGraphData = z.infer + +interface BarGraphLayoutProps { + data: Partial +} + +const BarGraphLayout: React.FC = ({ data: slideData }) => { + const { title, description, chartData, showLegend = false, showTooltip = true } = slideData; + + const CustomLegend = () => ( +
+ {chartData?.map((entry, index) => ( +
+
+ {entry.name} +
+ ))} +
+ ); + + const renderBarChart = () => { + return ( + + + { + if (value.length > 15) { + return value.substring(0, 15) + '...'; + } + return value; + }} + /> + `$${value.toFixed(0)}.00`} + /> + {showTooltip && } />} + + {chartData?.map((entry, index) => ( + + ))} + + + ); + }; + + return ( +
+ +
+ {/* Header section */} +
+ {/* Title */} + {title && ( +

+ {title} +

+ )} + + {/* Description */} + {description && ( +

+ {description} +

+ )} +
+ + {/* Chart section */} +
+
+ + {renderBarChart()} + + {showLegend && } +
+
+
+
+ ) +} + +export default BarGraphLayout \ No newline at end of file diff --git a/servers/nextjs/presentation-layouts/classic-dark/4-Comparison.tsx b/servers/nextjs/presentation-layouts/classic-dark/4-Comparison.tsx new file mode 100644 index 00000000..3fa9bec6 --- /dev/null +++ b/servers/nextjs/presentation-layouts/classic-dark/4-Comparison.tsx @@ -0,0 +1,120 @@ +import React from 'react' +import * as z from "zod"; +import { ImageSchema } from '@/presentation-layouts/defaultSchemes'; + +export const layoutId = 'classic-dark-comparison' +export const layoutName = 'Classic Dark Comparison' +export const layoutDescription = 'A modern slide with dark background, image on the left (2/5), and comparison content on the right (3/5).' + +const comparisonItemSchema = z.object({ + name: z.string().min(3).max(30).meta({ description: "Commodity name" }), + value: z.string().min(3).max(30).meta({ description: "Value" }), + when: z.string().min(3).max(30).meta({ description: "When the value was recorded" }), + details: z.string().min(3).max(50).optional().meta({ description: "Additional details" }), +}); + +const comparisonSectionSchema = z.object({ + title: z.string().min(3).max(50).meta({ description: "Section title" }), + items: z.array(comparisonItemSchema).min(1).max(3).meta({ description: "List of items in the section" }), +}); + +const comparisonSchema = z.object({ + title: z.string().min(3).max(100).default('Key Commodities in Focus').meta({ + description: "Main title of the slide", + }), + comparisonSections: z.array(comparisonSectionSchema).min(2).max(2).default([ + { + title: 'Exports', + items: [ + { name: 'Soybean Oil', value: '$186.91 million', when: '2022' }, + { name: 'Cardamom', value: '$46.64 million', when: '2022', details: 'primarily to India' }, + ] + }, + { + title: 'Imports', + items: [ + { name: 'Crude Soybean Oil', value: '$347.77 million', when: '2022' }, + { name: 'Petroleum Products', value: '$3.15 billion', when: '2022', details: '22% of total imports' }, + { name: 'Vehicles/Parts', value: '$526 million', when: '2022', details: 'down 45% from 2021' }, + ] + } + ]).meta({ + description: "Comparison sections with title and data items", + }), + image: ImageSchema.default({ + __image_url__: 'https://images.pexels.com/photos/9669089/pexels-photo-9669089.jpeg', + __image_prompt__: 'Map of South Asia showing Nepal and neighboring countries with trade routes highlighted' + }).meta({ + description: "Comparison visualization image", + }), +}) + +export const Schema = comparisonSchema + +export type ComparisonData = z.infer + +interface ComparisonLayoutProps { + data: Partial +} + +const ComparisonLayout: React.FC = ({ data: slideData }) => { + const { title, comparisonSections, image } = slideData; + + return ( +
+ +
+ {/* Left side - Image (2/5) */} +
+ {image && ( +
+ {image.__image_prompt__} +
+ )} +
+ + {/* Right side - Content (3/5) */} +
+ {/* Title */} + {title && ( +

+ {title} +

+ )} + + {/* Comparison Sections */} +
+ {comparisonSections && comparisonSections.map((section, sectionIndex) => ( +
+

+ {section.title} +

+ {section.items && section.items.length > 0 && ( +
+ {section.items.map((item, index) => ( +
+ • {item.name}: {item.value} ({item.when}) + {item.details && ( + , {item.details} + )} +
+ ))} +
+ )} +
+ ))} +
+
+
+
+ ) +} + +export default ComparisonLayout \ No newline at end of file diff --git a/servers/nextjs/presentation-layouts/classic-dark/5-Metrics.tsx b/servers/nextjs/presentation-layouts/classic-dark/5-Metrics.tsx new file mode 100644 index 00000000..7e7db991 --- /dev/null +++ b/servers/nextjs/presentation-layouts/classic-dark/5-Metrics.tsx @@ -0,0 +1,141 @@ +import React from 'react' +import * as z from "zod"; +import { IconSchema } from '@/presentation-layouts/defaultSchemes'; + +export const layoutId = 'classic-dark-metrics' +export const layoutName = 'Classic Dark Metrics' +export const layoutDescription = 'A modern slide with dark background, metric cards arranged in a grid with icons and data.' + +const metricItemSchema = z.object({ + title: z.string().min(3).max(50).meta({ description: "Metric title" }), + value: z.string().min(3).max(30).meta({ description: "Metric value" }), + percentage: z.string().min(3).max(30).meta({ description: "Percentage value" }), + icon: IconSchema.meta({ description: "Icon for the metric" }), +}); + +const metricsSchema = z.object({ + title: z.string().min(3).max(100).default('Top Export Destinations').meta({ + description: "Main title of the slide", + }), + description: z.string().min(10).max(200).default('Nepal exports 760 products to 132 countries, with a strong focus on regional trade.').meta({ + description: "Description text", + }), + metrics: z.array(metricItemSchema).min(2).max(6).default([ + { + title: 'India', + value: '$935 million', + percentage: '71.93%', + icon: { + __icon_url__: '/static/icons/placeholder.png', + __icon_query__: 'star rating' + } + }, + { + title: 'United States', + value: '$147 million', + percentage: '11.32%', + icon: { + __icon_url__: '/static/icons/placeholder.png', + __icon_query__: 'flag country' + } + }, + { + title: 'Germany', + value: '$33 million', + percentage: '2.51%', + icon: { + __icon_url__: '/static/icons/placeholder.png', + __icon_query__: 'user person' + } + }, + { + title: 'Turkey', + value: '$26 million', + percentage: '2.01%', + icon: { + __icon_url__: '/static/icons/placeholder.png', + __icon_query__: 'pen tool' + } + }, + { + title: 'United Kingdom', + value: '$24 million', + percentage: '1.83%', + icon: { + __icon_url__: '/static/icons/placeholder.png', + __icon_query__: 'message chat' + } + }, + ]).meta({ + description: "Metric cards data", + }), +}) + +export const Schema = metricsSchema + +export type MetricsData = z.infer + +interface MetricsLayoutProps { + data: Partial +} + +const MetricsLayout: React.FC = ({ data: slideData }) => { + const { title, description, metrics } = slideData; + + return ( +
+ +
+ {/* Header section */} +
+ {/* Title */} + {title && ( +

+ {title} +

+ )} + + {/* Description */} + {description && ( +

+ {description} +

+ )} +
+ + {/* Metrics Cards Grid */} +
+
+ {metrics && metrics.map((metric, index) => ( +
+ {/* Metric Card with overlapping icon */} +
+ {/* Icon overlapping the top */} +
+ {metric.icon.__icon_query__} +
+ + {/* Content with top padding for icon space */} +
+

+ {metric.title} +

+

+ {metric.value} ({metric.percentage}) +

+
+
+
+ ))} +
+
+
+
+ ) +} + +export default MetricsLayout \ No newline at end of file diff --git a/servers/nextjs/presentation-layouts/classic-dark/6-BulletPointWithDescription.tsx b/servers/nextjs/presentation-layouts/classic-dark/6-BulletPointWithDescription.tsx new file mode 100644 index 00000000..cfcb6b32 --- /dev/null +++ b/servers/nextjs/presentation-layouts/classic-dark/6-BulletPointWithDescription.tsx @@ -0,0 +1,103 @@ +import React from 'react' +import * as z from "zod"; +import { ImageSchema } from '@/presentation-layouts/defaultSchemes'; + +export const layoutId = 'classic-dark-bullet-point-with-description' +export const layoutName = 'Classic Dark Bullet Point with Description' +export const layoutDescription = 'A modern slide with dark background, image on the left (2/5), and bullet points with descriptions in boxes on the right (3/5).' + +const bulletPointSchema = z.object({ + title: z.string().min(3).max(80).meta({ description: "Bullet point title" }), + content: z.string().min(10).max(150).meta({ description: "Bullet point content (max 150 characters)" }), +}); + +const bulletPointWithDescriptionSchema = z.object({ + title: z.string().min(3).max(100).default('Trade Policies and Challenges').meta({ + description: "Main title of the slide", + }), + bulletPoints: z.array(bulletPointSchema).min(2).max(3).default([ + { + title: 'Tariffs', + content: 'Effectively Applied Tariff (2022): 11.59%. Most Favored Nation Tariff (2022): 12.87%. Duty-free imports: $412.11 million.' + }, + { + title: 'Forex Reserves', + content: '$8.18 billion in 2019, covering 8 months of imports.' + }, + { + title: 'Import Ban Impact', + content: 'Luxury goods ban (Apr-Dec 2022) cut deficit by 15.45% but reduced export earnings by 21.44%.' + } + ]).meta({ + description: "Bullet points with descriptions (max 3 items)", + }), + image: ImageSchema.default({ + __image_url__: 'https://images.pexels.com/photos/9669089/pexels-photo-9669089.jpeg', + __image_prompt__: 'Stylized mountainous landscape with trade arrows and network connections, dark gradient background with orange sun and purple mountains' + }).meta({ + description: "Visual representation image", + }), +}) + +export const Schema = bulletPointWithDescriptionSchema + +export type BulletPointWithDescriptionData = z.infer + +interface BulletPointWithDescriptionLayoutProps { + data: Partial +} + +const BulletPointWithDescriptionLayout: React.FC = ({ data: slideData }) => { + const { title, bulletPoints, image } = slideData; + + return ( +
+ +
+ {/* Left side - Image (2/5) */} +
+ {image && ( +
+ {image.__image_prompt__} +
+ )} +
+ + {/* Right side - Content (3/5) */} +
+ {/* Title */} + {title && ( +

+ {title} +

+ )} + + {/* Bullet Points */} + {bulletPoints && bulletPoints.length > 0 && ( +
+ {bulletPoints.map((point, index) => ( +
+

+ {point.title} +

+

+ {point.content} +

+
+ ))} +
+ )} +
+
+
+ ) +} + +export default BulletPointWithDescriptionLayout \ No newline at end of file diff --git a/servers/nextjs/presentation-layouts/classic-dark/settings.json b/servers/nextjs/presentation-layouts/classic-dark/settings.json new file mode 100644 index 00000000..2e888eab --- /dev/null +++ b/servers/nextjs/presentation-layouts/classic-dark/settings.json @@ -0,0 +1,5 @@ +{ + "description": "Classic dark layout for presentations", + "ordered": false, + "default": false +} \ No newline at end of file diff --git a/servers/nextjs/types/global.d.ts b/servers/nextjs/types/global.d.ts index 2dee9ebb..4b1caa32 100644 --- a/servers/nextjs/types/global.d.ts +++ b/servers/nextjs/types/global.d.ts @@ -15,17 +15,36 @@ interface TextFrameProps { interface LLMConfig { LLM?: string; + + // OpenAI OPENAI_API_KEY?: string; + OPENAI_MODEL?: string; + + // Google GOOGLE_API_KEY?: string; + GOOGLE_MODEL?: string; + + // Anthropic + ANTHROPIC_API_KEY?: string; + ANTHROPIC_MODEL?: string; + + // Ollama OLLAMA_URL?: string; OLLAMA_MODEL?: string; + + // Custom LLM CUSTOM_LLM_URL?: string; CUSTOM_LLM_API_KEY?: string; CUSTOM_MODEL?: string; + + // Image providers IMAGE_PROVIDER?: string; PIXABAY_API_KEY?: string; PEXELS_API_KEY?: string; + // Extended reasoning + EXTENDED_REASONING?: boolean; + // Only used in UI settings USE_CUSTOM_URL?: boolean; } \ No newline at end of file diff --git a/servers/nextjs/utils/providerConstants.ts b/servers/nextjs/utils/providerConstants.ts index 5fb28a7e..0dd53af9 100644 --- a/servers/nextjs/utils/providerConstants.ts +++ b/servers/nextjs/utils/providerConstants.ts @@ -67,16 +67,17 @@ export const LLM_PROVIDERS: Record = { openai: { value: "openai", label: "OpenAI", - description: "OpenAI's latest image generation model", - model_value: "gpt-4.1", - model_label: "GPT-4.1" + description: "OpenAI's latest text generation model", }, google: { value: "google", label: "Google", - description: "Google's primary image generation model", - model_value: "gemini-2.0-flash", - model_label: "Gemini 2.0 Flash" + description: "Google's primary text generation model", + }, + anthropic: { + value: "anthropic", + label: "Anthropic", + description: "Anthropic's Claude models", }, ollama: { value: "ollama", diff --git a/servers/nextjs/utils/providerUtils.ts b/servers/nextjs/utils/providerUtils.ts index 37e27b92..0d1ac62c 100644 --- a/servers/nextjs/utils/providerUtils.ts +++ b/servers/nextjs/utils/providerUtils.ts @@ -1,5 +1,3 @@ -import { toast } from "sonner"; - export interface OllamaModel { label: string; value: string; @@ -18,16 +16,37 @@ export interface DownloadingModel { export interface LLMConfig { LLM?: string; + + // OpenAI OPENAI_API_KEY?: string; + OPENAI_MODEL?: string; + + // Google GOOGLE_API_KEY?: string; + GOOGLE_MODEL?: string; + + // Anthropic + ANTHROPIC_API_KEY?: string; + ANTHROPIC_MODEL?: string; + + // Ollama OLLAMA_URL?: string; OLLAMA_MODEL?: string; + + // Custom LLM CUSTOM_LLM_URL?: string; CUSTOM_LLM_API_KEY?: string; CUSTOM_MODEL?: string; + + // Image providers + IMAGE_PROVIDER?: string; PEXELS_API_KEY?: string; PIXABAY_API_KEY?: string; - IMAGE_PROVIDER?: string; + + // Extended reasoning + EXTENDED_REASONING?: boolean; + + // Only used in UI settings USE_CUSTOM_URL?: boolean; } @@ -42,11 +61,15 @@ export interface OllamaModelsResult { export const updateLLMConfig = ( currentConfig: LLMConfig, field: string, - value: string + value: string | boolean ): LLMConfig => { const fieldMappings: Record = { openai_api_key: "OPENAI_API_KEY", + openai_model: "OPENAI_MODEL", google_api_key: "GOOGLE_API_KEY", + google_model: "GOOGLE_MODEL", + anthropic_api_key: "ANTHROPIC_API_KEY", + anthropic_model: "ANTHROPIC_MODEL", ollama_url: "OLLAMA_URL", ollama_model: "OLLAMA_MODEL", custom_llm_url: "CUSTOM_LLM_URL", @@ -55,6 +78,8 @@ export const updateLLMConfig = ( pexels_api_key: "PEXELS_API_KEY", pixabay_api_key: "PIXABAY_API_KEY", image_provider: "IMAGE_PROVIDER", + extended_reasoning: "EXTENDED_REASONING", + use_custom_url: "USE_CUSTOM_URL", }; const configKey = fieldMappings[field]; @@ -86,53 +111,6 @@ export const changeProvider = ( return newConfig; }; -/** - * Fetches supported Ollama models - */ -export const fetchOllamaModels = async (): Promise => { - try { - const response = await fetch("/api/v1/ppt/ollama/models/supported"); - const models = await response.json(); - return models || []; - } catch (error) { - console.error("Error fetching ollama models:", error); - return []; // Ensure we always return an empty array on error - } -}; - -/** - * Fetches Ollama models and validates current selection - * Returns models and updated config if needed - */ -export const fetchOllamaModelsWithConfig = async ( - config: LLMConfig -): Promise => { - try { - const models = await fetchOllamaModels(); - - // Check if currently selected model is still available - let updatedConfig: LLMConfig | undefined; - if (config.OLLAMA_MODEL && models && models.length > 0) { - const isModelAvailable = models.some( - (model: OllamaModel) => model.value === config.OLLAMA_MODEL - ); - if (!isModelAvailable) { - updatedConfig = { ...config, OLLAMA_MODEL: "" }; - } - } - - return { - models, - updatedConfig - }; - } catch (error) { - console.error("Error fetching ollama models:", error); - return { - models: [], - updatedConfig: { ...config, OLLAMA_MODEL: "" } - }; - } -}; export const checkIfSelectedOllamaModelIsPulled = async (ollamaModel: string) => { try { @@ -146,37 +124,6 @@ export const checkIfSelectedOllamaModelIsPulled = async (ollamaModel: string) => } } -/** - * Fetches available custom models - */ -export const fetchCustomModels = async ( - url: string, - apiKey: string -): Promise => { - try { - const response = await fetch("/api/v1/ppt/custom_llm/models/available", { - method: "POST", - headers: { - "Content-Type": "application/json", - }, - body: JSON.stringify({ - url: url || "", - api_key: apiKey || "", - }), - }); - - if (!response.ok) { - throw new Error(`HTTP error! status: ${response.status}`); - } - - const data = await response.json(); - return data; - } catch (error) { - toast.info("Could not fetch custom models"); - console.error("Error fetching custom models:", error); - throw error; - } -}; /** * Resets downloading model state @@ -234,51 +181,4 @@ export const pullOllamaModel = async ( } }, 1000); }); -}; - -/** - * Sets Ollama configuration based on custom URL preference - */ -export const setOllamaConfig = ( - currentConfig: LLMConfig, - useCustomUrl: boolean -): LLMConfig => { - let customUrl = "http://localhost:11434"; - if (!useCustomUrl) { - return { - ...currentConfig, - OLLAMA_URL: customUrl, - USE_CUSTOM_URL: false, - }; - } else { - return { ...currentConfig, USE_CUSTOM_URL: true, OLLAMA_URL: customUrl }; - } -}; - -/** - * Handles saving configuration with error handling - */ -export const handleSaveConfiguration = async ( - llmConfig: LLMConfig, - handleSaveLLMConfig: (config: LLMConfig) => Promise, - pullOllamaModels?: () => Promise -): Promise => { - try { - await handleSaveLLMConfig(llmConfig); - if (llmConfig.LLM === "ollama" && pullOllamaModels) { - await pullOllamaModels(); - } - toast.success("Configuration saved successfully"); - } catch (error) { - console.error("Error:", error); - toast.error( - error instanceof Error - ? error.message - : "Failed to save configuration", - { - description: "Failed to save configuration", - } - ); - throw error; - } -}; \ No newline at end of file +}; \ No newline at end of file diff --git a/servers/nextjs/utils/storeHelpers.ts b/servers/nextjs/utils/storeHelpers.ts index 3ed7d0f3..90c07dff 100644 --- a/servers/nextjs/utils/storeHelpers.ts +++ b/servers/nextjs/utils/storeHelpers.ts @@ -16,8 +16,30 @@ export const handleSaveLLMConfig = async (llmConfig: LLMConfig) => { export const hasValidLLMConfig = (llmConfig: LLMConfig) => { if (!llmConfig.LLM) return false; if (!llmConfig.IMAGE_PROVIDER) return false; - const OPENAI_API_KEY = llmConfig.OPENAI_API_KEY; - const GOOGLE_API_KEY = llmConfig.GOOGLE_API_KEY; + + const isOpenAIConfigValid = + llmConfig.OPENAI_MODEL !== "" && + llmConfig.OPENAI_MODEL !== null && + llmConfig.OPENAI_MODEL !== undefined && + llmConfig.OPENAI_API_KEY !== "" && + llmConfig.OPENAI_API_KEY !== null && + llmConfig.OPENAI_API_KEY !== undefined; + + const isGoogleConfigValid = + llmConfig.GOOGLE_MODEL !== "" && + llmConfig.GOOGLE_MODEL !== null && + llmConfig.GOOGLE_MODEL !== undefined && + llmConfig.GOOGLE_API_KEY !== "" && + llmConfig.GOOGLE_API_KEY !== null && + llmConfig.GOOGLE_API_KEY !== undefined; + + const isAnthropicConfigValid = + llmConfig.ANTHROPIC_MODEL !== "" && + llmConfig.ANTHROPIC_MODEL !== null && + llmConfig.ANTHROPIC_MODEL !== undefined && + llmConfig.ANTHROPIC_API_KEY !== "" && + llmConfig.ANTHROPIC_API_KEY !== null && + llmConfig.ANTHROPIC_API_KEY !== undefined; const isOllamaConfigValid = llmConfig.OLLAMA_MODEL !== "" && @@ -42,9 +64,9 @@ export const hasValidLLMConfig = (llmConfig: LLMConfig) => { case "pixabay": return llmConfig.PIXABAY_API_KEY && llmConfig.PIXABAY_API_KEY !== ""; case "dall-e-3": - return OPENAI_API_KEY && OPENAI_API_KEY !== ""; + return llmConfig.OPENAI_API_KEY && llmConfig.OPENAI_API_KEY !== ""; case "gemini_flash": - return GOOGLE_API_KEY && GOOGLE_API_KEY !== ""; + return llmConfig.GOOGLE_API_KEY && llmConfig.GOOGLE_API_KEY !== ""; default: return false; } @@ -52,18 +74,16 @@ export const hasValidLLMConfig = (llmConfig: LLMConfig) => { const isLLMConfigValid = llmConfig.LLM === "openai" - ? OPENAI_API_KEY !== "" && - OPENAI_API_KEY !== null && - OPENAI_API_KEY !== undefined + ? isOpenAIConfigValid : llmConfig.LLM === "google" - ? GOOGLE_API_KEY !== "" && - GOOGLE_API_KEY !== null && - GOOGLE_API_KEY !== undefined - : llmConfig.LLM === "ollama" - ? isOllamaConfigValid - : llmConfig.LLM === "custom" - ? isCustomConfigValid - : false; + ? isGoogleConfigValid + : llmConfig.LLM === "anthropic" + ? isAnthropicConfigValid + : llmConfig.LLM === "ollama" + ? isOllamaConfigValid + : llmConfig.LLM === "custom" + ? isCustomConfigValid + : false; return isLLMConfigValid && isImageConfigValid(); }; diff --git a/start.js b/start.js index 092b8af1..93ef1592 100644 --- a/start.js +++ b/start.js @@ -1,37 +1,59 @@ /* This script starts the FastAPI and Next.js servers, setting up user configuration if necessary. It reads environment variables to configure API keys and other settings, ensuring that the user configuration file is created if it doesn't exist. The script also handles the starting of both servers and keeps the Node.js process alive until one of the servers exits. */ -const path = require("path"); -const { spawn } = require("child_process"); -const fs = require("fs"); +import { join, dirname } from 'path'; +import { fileURLToPath } from 'url'; +import { spawn } from 'child_process'; +import { existsSync, mkdirSync, rmSync, cpSync, readFileSync, writeFileSync } from 'fs'; -const fastapiDir = path.join(__dirname, "servers/fastapi"); -const nextjsDir = path.join(__dirname, "servers/nextjs"); +const __filename = fileURLToPath(import.meta.url); +const __dirname = dirname(__filename); -const isDev = process.env.NODE_ENV === "development"; -const canChangeKeys = process.env.CAN_CHANGE_KEYS !== "false"; +const fastapiDir = join(__dirname, 'servers/fastapi'); +const nextjsDir = join(__dirname, 'servers/nextjs'); + +const args = process.argv.slice(2); +const hasDevArg = args.includes('--dev') || args.includes('-d'); +const isDev = hasDevArg; +const canChangeKeys = process.env.CAN_CHANGE_KEYS !== 'false'; const fastapiPort = 8000; const nextjsPort = 3000; const appmcpPort = 8001; -const userConfigPath = path.join( - process.env.APP_DATA_DIRECTORY, - "userConfig.json", -); -const userDataDir = path.dirname(userConfigPath); + +const userConfigPath = join(process.env.APP_DATA_DIRECTORY, 'userConfig.json'); +const userDataDir = dirname(userConfigPath); // Create user_data directory if it doesn't exist -if (!fs.existsSync(userDataDir)) { - fs.mkdirSync(userDataDir, { recursive: true }); +if (!existsSync(userDataDir)) { + mkdirSync(userDataDir, { recursive: true }); } +// Setup node_modules for development +const setupNodeModules = () => { + console.log('Setting up node_modules for development...'); + const nodeDependenciesPath = '/node_dependencies/node_modules'; + const nextjsNodeModulesPath = join(nextjsDir, 'node_modules'); + + if (existsSync(nodeDependenciesPath)) { + if (existsSync(nextjsNodeModulesPath)) { + rmSync(nextjsNodeModulesPath, { recursive: true, force: true }); + } + + cpSync(nodeDependenciesPath, nextjsNodeModulesPath, { recursive: true }); + console.log('Copied node_modules from /node_dependencies to Next.js server directory'); + } +}; + + process.env.USER_CONFIG_PATH = userConfigPath; //? UserConfig is only setup if API Keys can be changed const setupUserConfigFromEnv = () => { let existingConfig = {}; - if (fs.existsSync(userConfigPath)) { - existingConfig = JSON.parse(fs.readFileSync(userConfigPath, "utf8")); + + if (existsSync(userConfigPath)) { + existingConfig = JSON.parse(readFileSync(userConfigPath, 'utf8')); } if (!["ollama", "openai", "google"].includes(existingConfig.LLM)) { @@ -41,9 +63,13 @@ const setupUserConfigFromEnv = () => { const userConfig = { LLM: process.env.LLM || existingConfig.LLM, OPENAI_API_KEY: process.env.OPENAI_API_KEY || existingConfig.OPENAI_API_KEY, + OPENAI_MODEL: process.env.OPENAI_MODEL || existingConfig.OPENAI_MODEL, GOOGLE_API_KEY: process.env.GOOGLE_API_KEY || existingConfig.GOOGLE_API_KEY, + GOOGLE_MODEL: process.env.GOOGLE_MODEL || existingConfig.GOOGLE_MODEL, OLLAMA_URL: process.env.OLLAMA_URL || existingConfig.OLLAMA_URL, OLLAMA_MODEL: process.env.OLLAMA_MODEL || existingConfig.OLLAMA_MODEL, + ANTHROPIC_API_KEY: process.env.ANTHROPIC_API_KEY || existingConfig.ANTHROPIC_API_KEY, + ANTHROPIC_MODEL: process.env.ANTHROPIC_MODEL || existingConfig.ANTHROPIC_MODEL, CUSTOM_LLM_URL: process.env.CUSTOM_LLM_URL || existingConfig.CUSTOM_LLM_URL, CUSTOM_LLM_API_KEY: process.env.CUSTOM_LLM_API_KEY || existingConfig.CUSTOM_LLM_API_KEY, @@ -52,11 +78,14 @@ const setupUserConfigFromEnv = () => { PIXABAY_API_KEY: process.env.PIXABAY_API_KEY || existingConfig.PIXABAY_API_KEY, IMAGE_PROVIDER: process.env.IMAGE_PROVIDER || existingConfig.IMAGE_PROVIDER, + EXTENDED_REASONING: process.env.EXTENDED_REASONING || existingConfig.EXTENDED_REASONING, USE_CUSTOM_URL: process.env.USE_CUSTOM_URL || existingConfig.USE_CUSTOM_URL, }; - fs.writeFileSync(userConfigPath, JSON.stringify(userConfig)); -}; + + writeFileSync(userConfigPath, JSON.stringify(userConfig)); +} + const startServers = async () => { const fastApiProcess = spawn( "python", @@ -100,17 +129,75 @@ const startServers = async () => { console.error("Next.js process failed to start:", err); }); + const ollamaProcess = spawn( + "ollama", + ["serve"], + { + cwd: "/", + stdio: "inherit", + env: process.env, + } + ); + + ollamaProcess.on("error", err => { + console.error("Ollama process failed to start:", err); + }); + + const redisProcess = spawn( + "redis-server", + [], + { + cwd: "/", + stdio: "inherit", + env: process.env, + } + ); + + redisProcess.on("error", err => { + console.error("Redis process failed to start:", err); + }); + // Keep the Node process alive until both servers exit const exitCode = await Promise.race([ - new Promise((resolve) => fastApiProcess.on("exit", resolve)), - new Promise((resolve) => nextjsProcess.on("exit", resolve)), + + new Promise(resolve => fastApiProcess.on("exit", resolve)), + new Promise(resolve => nextjsProcess.on("exit", resolve)), + new Promise(resolve => ollamaProcess.on("exit", resolve)), + new Promise(resolve => redisProcess.on("exit", resolve)), ]); console.log(`One of the processes exited. Exit code: ${exitCode}`); process.exit(exitCode); }; +// Start nginx service +const startNginx = () => { + const nginxProcess = spawn('service', ['nginx', 'start'], { + stdio: 'inherit', + env: process.env, + }); + + nginxProcess.on('error', err => { + console.error('Nginx process failed to start:', err); + }); + + nginxProcess.on('exit', (code) => { + if (code === 0) { + console.log('Nginx started successfully'); + } else { + console.error(`Nginx failed to start with exit code: ${code}`); + } + }); +}; + +if (isDev) { + setupNodeModules(); +} + if (canChangeKeys) { setupUserConfigFromEnv(); } + startServers(); + +startNginx(); \ No newline at end of file