Merge pull request #6 from AstraBert/december-2024

v1.0.0

.env.example

@@ -0,0 +1,3 @@
+PG_DB="postgres"
+PG_USER="pgql_usr"
+PG_PASSWORD="pgql_psw"

.gitignore

@@ -1,2 +1,11 @@
 model/
-__pycache__/
+lib/scripts/__pycache__/
+lib/docker/__pycache__/
+lib/scripts/.env
+lib/docker/.env
+.env
+virtualenv/
+qdrant_storage/
+lib/docker/florence-2/
+lib/docker/qwen/
+lib/docker/labse/

.gradio/flagged/dataset1.csv

@@ -0,0 +1,20 @@
+Search Query,Image Search Query,Maximum Number of Search Results,Enable RAG,Debug,Search Results,timestamp
+,.gradio\flagged\Image Search Query\41ddcc66f19720a7c3bd\Components-of-a-nuclear-power-plant-1400x803.png,3,false,true,"### Understanding Nuclear Power Plants and Reactors
+
+Nuclear power plants generate electricity through controlled nuclear fission in large, specialized facilities. These systems consist of multiple components including nuclear reactors which convert uranium or other fissile materials into usable energy.
+
+#### Key Components:
+- **Reactor Core**: The heart of any nuclear power plant where the actual reaction takes place.
+    - *Bangombe Deposit*: A notable location for potential future use as an additional source of fuel.
+
+- **Heat Transfer Fluids**: Various types such as heavy water (deuterium oxide) and light water (ordinary H₂O), depending on specific designs like RBMK reactors mentioned here.
+
+- **Power Outputs**: Units can produce significant amounts of electrical output; e.g., one reactor might be rated at around 150 MWe (Megawatts Electrical).
+
+- **Safety Measures & Regulations**: Strict guidelines ensure safety during operation by managing coolant flow rates carefully (""Follow Loads Reasonably Easily Without Burning"").
+
+- **Fuel Cycle Management**: Ensures smooth transition between different phases of operations using detailed guides available online.
+
+This type of technology has been extensively studied over decades leading up to today's advanced models. For instance, there have been numerous reports documenting its development process from early stages all the way to current state-of-the-art technologies.
+
+If interested further details about individual aspects could refer directly to World-Nuclear Organization resources linked within this summary. They provide comprehensive documentation covering everything from theoretical principles down to practical applications across diverse contexts worldwide.",2024-12-10 16:31:14.503594

CONTRIBUTING.md

@@ -1,4 +1,4 @@
-# Contributing to SearchPhi
+# Contributing to PrAIvateSearch
 
 Do you want to contribute to this project? Make sure to read this guidelines first :)
 
@@ -32,15 +32,15 @@ Do you want to contribute to this project? Make sure to read this guidelines fir
 3. Submit pull request (make sure to provide a thorough description of the changes)
 
 
-## Showcase your SearchPhi
+## Showcase your PrAIvateSearch
 
 **When to do it**:
 
-- You modified the base application with new features but you don't want/can't merge them with the original SearchPhi
+- You modified the base application with new features but you don't want/can't merge them with the original PrAIvateSearch
 
 **How to do it**:
 
-- Go to [_GitHub Discussions > Show and tell_](https://github.com/AstraBert/SearchPhi/discussions/categories/show-and-tell) page
-- Open a new discussion there, describing your SearchPhi application
+- Go to [_GitHub Discussions > Show and tell_](https://github.com/AstraBert/PrAIvateSearch/discussions/categories/show-and-tell) page
+- Open a new discussion there, describing your PrAIvateSearch application
 
 ### Thanks for contributing!

README.md

@@ -1,42 +1,52 @@
-<h1 align="center">SearchPhi</h1>
-<h2 align="center">Open source and AI-powered web search engine🌐</h2>
-
+<h1 align="center">PrAIvateSearch</h1>
+<h2 align="center">Own your AI, search the web with it🌐😎</h2>
 
 <div align="center">
-    <img src="https://img.shields.io/github/languages/top/AstraBert/SearchPhi" alt="GitHub top language">
-   <img src="https://img.shields.io/github/commit-activity/t/AstraBert/SearchPhi" alt="GitHub commit activity">
-   <img src="https://img.shields.io/badge/Status-stable_beta-green" alt="Static Badge">
-   <img src="https://img.shields.io/badge/Release-v0.0_beta.0-purple" alt="Static Badge">
-   <img src="https://img.shields.io/docker/image-size/astrabert/searchphi
-   " alt="Docker image size">
-   <img src="https://img.shields.io/badge/Supported_platforms-Windows/POSIX-brown" alt="Static Badge">
    <div>
-        <img src="./imgs/SearchPhi_logo.png" alt="Logo" align="center">
+        <img src="./imgs/PrAIvateSearch_logo.png" alt="Logo" align="center">
    </div>
 </div>
 
-## About SearchPhi
 
-SearchPhi is a Streamlit application that aims to implement similar features to SearchGPT, but in an open-source, local and private way. 
+## About PrAIvateSearch
+
+PrAIvateSearch is a Gradio application that aims to implement similar features to SearchGPT, but in an open-source, local and private way. 
+
+## Flowchart
+
+<div align="center">
+    <img src="./imgs/PrAIvateSearch_Flowchart.png" alt="Logo" align="center">
+    <p><i>Flowchart for PrAIvateSearch</i></p>
+</div>
+
+The process of creating and the functioning of PrAIvateSearch is explained in this [blog post on HuggingFace](https://huggingface.co/blog/as-cle-bert/build-an-ai-powered-search-engine-from-scratch).
 
 ## Installation and usage
 
-### Source code
 
 1. Clone the repository:
 
 ```bash
-git clone https://github.com/AstraBert/SearchPhi.git
-cd SearchPhi
+git clone https://github.com/AstraBert/PrAIvateSearch.git
+cd PrAIvateSearch
 ```
 
-2. Create a `model` folder, download [this GGUF file](https://huggingface.co/microsoft/Phi-3-mini-4k-instruct-gguf/blob/main/Phi-3-mini-4k-instruct-q4.gguf) and move the GGUF file in the `model` folder:
+2. Move `.env.example` to `.env`...
 
 ```bash
-mkdir model
-mv /path/to/Downloads/Phi-3-mini-4k-instruct-q4.gguf model/
+mv .env.example .env
 ```
 
+...and specify PostgreSQL related variables:
+
+```bash
+# .env file
+PG_DB="postgres"
+PG_USER="pgql_usr"
+PG_PASSWORD="pgql_psw"
+```
+
+
 3. Install necessary dependencies:
   - Linux:
 ```bash
@@ -56,59 +66,32 @@ source c:\path\to\SearchPhi\Scripts\activate  # For Git
 python3 -m pip install -r requirements.txt
 ```
 
-
-4. Run the application:
+4. Start third-party services:
 
 ```bash
-python3 -m streamlit run app.py
+docker compose up -d
 ```
 
-You'll see the application on `http://localhost:8501`.
-
-**PROs**: You can customize the application code (change the GGUF model, change CPU/GPU settings, change generation kwargs, modify the app interface...)
-
-**CONs**: Longer and more complex installation process
-
-### Docker
-
-1. Pull the image
-
-```bash
-docker pull astrabert/searchphi:latest
-```
-
-2. Run the container:
+5. Run the application:
 
 ```bash
-docker run -p 8501:8501 astrabert/searchphi:latest
+python3 scripts/app.py
 ```
 
-Shortly after you submit the `docker run` command, the container logs will tell you that the application is up and running on `http://localhost:8501`.
+Once the models will be downloaded and loaded on your hardware, you'll see the application on `http://localhost:7860`.
 
-**PROs**: Shorter and simpler installation process
+**PROs**: You can customize the application code (change the model, change CPU/GPU settings, change generation kwargs, modify the app interface...)
 
-**CONs**: You cannot customize the application code
-
-### Run in cloud
-
-- **GitPod workspaces**: Click [here](https://gitpod.io/#https://github.com/AstraBert/SearchPhi) to open the GitPod workspace
-
-**PROs**: No local installation and you can exploit better hardwares
-
-**CONs**: Limited resources
+**CONs**: Longer and more complex installation process
 
 ### Usage note
 
-> ⚠️ _The Streamlit application was successfully developed and tested on a Windows 10.0.22631 machine, with 32GB RAM, 16 core CPU and Nvidia GEFORCE RTX4050 GPU (6GB, cuda version 12.3), python version 3.11.9_
-
-> ⚠️ _The Docker container was successfully tested on a Windows 10.0.22631 machine and on a Ubuntu 22.04.3 machine_ 
+> ⚠️ _The Gradio application was successfully developed and tested on a Windows 10.0.22631 machine, with 32GB RAM, 16 core CPU and Nvidia GEFORCE RTX4050 GPU (6GB, cuda version 12.3), python version 3.11.9_
 
 Although being at a good stage of development, the application is a `beta` and might still contain bugs and have OS/hardware/python version incompatibilities.
 
 ## Demo
 
-You can try out SearchPhi on [this HuggingFace Space](https://huggingface.co/spaces/as-cle-bert/SearchPhi). 
-
 Here's a video demo of what it can do:
 
 ![Video demo for SearechPhi](./imgs/demo.gif)

compose.yaml

@@ -0,0 +1,40 @@
+networks:
+  mynet:
+    driver: bridge
+
+services:
+  db:
+    image: postgres
+    restart: always
+    ports:
+      - "5432:5432"
+    networks:
+      - mynet
+    environment:
+      POSTGRES_DB: $PG_DB
+      POSTGRES_USER: $PG_USER
+      POSTGRES_PASSWORD: $PG_PASSWORD
+    volumes:
+      - pgdata:/var/lib/postgresql/data 
+
+  semantic_memory:
+    image: qdrant/qdrant
+    restart: always
+    ports:
+      - "6333:6333"
+      - "6334:6334"
+    networks:
+      - mynet
+    volumes:
+      - "./qdrant_storage:/qdrant/storage"
+
+  adminer:
+    image: adminer
+    restart: always
+    ports:
+      - "8080:8080"
+    networks:
+      - mynet
+
+volumes:
+  pgdata:

lib/scripts/app.py

@@ -0,0 +1,50 @@
+import warnings
+warnings.filterwarnings("ignore")
+
+import gradio as gr
+from text_inference import text_inference
+from image_gen import caption_image
+from PIL import Image
+from websearching import web_search, date_for_debug
+
+def reply(text_input, image_input=None, max_results=5, enable_rag=False, debug = True):
+    if debug:
+        print(f"[{date_for_debug()}] Started query processing...")
+    if image_input is None:
+        prompt, qdrant_success = web_search(text_input, max_results, enable_rag, debug)
+        if debug:
+            print(qdrant_success)
+        results = text_inference(prompt, debug)
+        results = results.replace("<|im_end|>","")
+        if debug:
+            print(f"[{date_for_debug()}] Finished query processing!")
+        return results
+    else:
+        if text_input:
+            img = Image.fromarray(image_input)
+            caption = caption_image(img)
+            full_query = caption +"\n\n"+text_input
+            prompt, qdrant_success = web_search(full_query, max_results, enable_rag)
+            if debug:
+                print(qdrant_success)
+            results = text_inference(prompt, debug)
+            results = results.replace("<|im_end|>","")
+            if debug:
+                print(f"[{date_for_debug()}] Finished query processing!")
+            return results
+        else:
+            img = Image.fromarray(image_input)
+            caption = caption_image(img)
+            prompt, qdrant_success = web_search(caption, max_results, enable_rag)
+            if debug:
+                print(qdrant_success)
+            results = text_inference(prompt, debug)
+            results = results.replace("<|im_end|>","")
+            if debug:
+                print(f"[{date_for_debug()}] Finished query processing!")
+            return results
+
+
+iface = gr.Interface(fn=reply, inputs=[gr.Textbox(value="",label="Search Query"), gr.Image(value=None, label="Image Search Query"), gr.Slider(1,10,value=5,label="Maximum Number of Search Results", step=1), gr.Checkbox(value=False, label="Enable RAG"), gr.Checkbox(value=True, label="Debug")], outputs=[gr.Markdown(value="Your output will be generated here", label="Search Results")], title="PrAIvateSearch")
+
+iface.launch(server_name="0.0.0.0", server_port=7860)