├── static
└── favicon.png
├── requirements.txt
├── README.md
├── LICENSE
├── app.py
└── templates
└── index.html
/static/favicon.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LexiestLeszek/sova_ollama/HEAD/static/favicon.png
--------------------------------------------------------------------------------
/requirements.txt:
--------------------------------------------------------------------------------
1 | beautifulsoup4==4.11.1
2 | Flask==3.0.2
3 | google==3.0.0
4 | gunicorn==21.2.0
5 | langchain==0.1.6
6 | langchain_community==0.0.19
7 | ollama==0.1.6
8 | Requests==2.31.0
9 | translators==5.8.9
10 |
--------------------------------------------------------------------------------
/README.md:
--------------------------------------------------------------------------------
1 | ### Sova
2 |
3 | ## Description
4 |
5 | Sova is a web search engine that utilizes power of Large Language Models (LLM) to return full text answer instead of bunch of links like traditional search engines do.
6 |
7 | ## Installation
8 |
9 | Before installing Sova, ensure you have Python installed on your system. You will also need pip, which is typically included with Python.
10 |
11 | To install Sova, follow these steps:
12 |
13 | 1. Clone the repository using this command in terminal: `git clone https://github.com/LexiestLeszek/sova_ollama.git`
14 | 2. Install the required dependencies using this command in terminal: `pip install -r requirements.txt`
15 | 3. Install Ollama from their website https://ollama.com/
16 | 4. Download the local LLM (in our case - qwen:1.8b) using this command in terminal: `ollama pull qwen:1.8b`
17 | 5. Run the gunicorn server using this command in terminal: `gunicorn -b 0.0.0.0:8080 app:app`
18 | 6. Use the search-bar to input your search query or or question.
19 |
20 |
--------------------------------------------------------------------------------
/LICENSE:
--------------------------------------------------------------------------------
1 | MIT License
2 |
3 | Copyright (c) 2024 Leszek Mielnikow
4 |
5 | Permission is hereby granted, free of charge, to any person obtaining a copy
6 | of this software and associated documentation files (the "Software"), to deal
7 | in the Software without restriction, including without limitation the rights
8 | to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
9 | copies of the Software, and to permit persons to whom the Software is
10 | furnished to do so, subject to the following conditions:
11 |
12 | The above copyright notice and this permission notice shall be included in all
13 | copies or substantial portions of the Software.
14 |
15 | THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
16 | IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17 | FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
18 | AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
19 | LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
20 | OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
21 | SOFTWARE.
22 |
--------------------------------------------------------------------------------
/app.py:
--------------------------------------------------------------------------------
1 | from flask import Flask, render_template, request
2 | from googlesearch import search
3 | import re
4 | import translators as ts
5 | import gunicorn
6 | import requests
7 | import ollama
8 | from googlesearch import search
9 | import requests
10 | from bs4 import BeautifulSoup
11 | from langchain_community.embeddings import HuggingFaceEmbeddings
12 | from langchain.text_splitter import RecursiveCharacterTextSplitter
13 | from langchain_community.vectorstores import Chroma
14 |
15 | app = Flask(__name__)
16 |
17 | def get_context(question):
18 | print("Init Search")
19 | search_results = search(question, num=5, stop=5, pause=2)
20 | top_links = list(search_results)
21 | exclude_list = ["google", "facebook", "twitter", "instagram", "youtube", "tiktok", "kremlin"]
22 | top_links = [link for link in top_links if not any(domain in link for domain in exclude_list)]
23 | scraped_texts = []
24 | for link in top_links:
25 | print(link)
26 | try:
27 | page = requests.get(link)
28 | soup = BeautifulSoup(page.content, 'html.parser')
29 | text = ' '.join([p.get_text() for p in soup.find_all('p')])
30 | print(text)
31 | except Exception as e:
32 | print(f"Failed to scrape the link: {link}\nError: {e}")
33 | scraped_texts.append(text)
34 |
35 | all_scraped_text = '.\n'.join(scraped_texts)
36 |
37 | print(all_scraped_text)
38 |
39 | text_splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=50)
40 | documents = text_splitter.split_text(all_scraped_text)
41 | embeddings = HuggingFaceEmbeddings(model_name = 'sentence-transformers/all-MiniLM-L6-v2')
42 |
43 | db = Chroma.from_texts(documents, embedding=embeddings)
44 | retriever = db.as_retriever(search_kwargs={"k": 3})
45 |
46 | context = retriever.get_relevant_documents(question)
47 |
48 | return context
49 |
50 | # Function to generate output using the LLM model
51 | def generate_output(question,context):
52 | # Prepare the prompt with the input text
53 | # Define a prompt template
54 | prompt = f"""Use the following context to answer the question at the end.
55 | If you don't know the answer, just say that you don't know, don't try to make up an answer.
56 | Answer only factual information based on the context below.
57 | Context: {context}.\n
58 | Question: {question}\n
59 | Helpful Answer:"""
60 |
61 | response = ollama.chat(model='qwen:1.8b', messages=[
62 |
63 | {
64 | 'role': 'system',
65 | 'content': 'You are a Helpful AI assistant that answers based on the information from the user prompt Context and nothing else.',
66 |
67 | 'role': 'user',
68 | 'content': f'{prompt}',
69 | },
70 | ])
71 |
72 | output = response['message']['content']
73 |
74 | # Return the generated text
75 | return output
76 |
77 | @app.route('/', methods=['GET', 'POST'])
78 | def index():
79 | query = ""
80 | answer = ""
81 |
82 | if request.method == 'POST':
83 | query = request.form['query']
84 |
85 | if query:
86 |
87 | context = get_context(query)
88 |
89 | # Generate the output using the LLM model
90 | answer = generate_output(query,context)
91 |
92 | print(answer)
93 |
94 | return render_template('index.html', query=query, answer=answer)
95 |
96 | #if __name__ == '__main__':
97 | # app.run(debug=True, port="8021")
98 |
99 | # to run on ngrok:
100 | # 1. in main Mac terminal: "ngrok http 8080"
101 | # 2. run this app in IDE terminal: "gunicorn -b 0.0.0.0:8080 app:app"
102 |
103 | # # Find the process running on port 8081
104 | # lsof -ti tcp:PORT
105 |
106 | # Kill the process (replace PID with the actual process ID)
107 | # kill PID
108 |
--------------------------------------------------------------------------------
/templates/index.html:
--------------------------------------------------------------------------------
1 |
2 |
3 |
4 |
5 |
6 |
7 | Sova Search
8 |
154 |
155 |
156 |