3 |
2 |
3 |
| This screenshot shows the functionality of using PDFs with the 'Essay Expert' prompt template. At the top, the system leverages PDF content for detailed responses, while the lower section illustrates responses generated without PDFs. |
55 | | 
| This example demonstrates the advanced capabilities of the 'Essays Expert' prompt template. The screenshot highlights how the system utilises PDF content to generate comprehensive responses at the top, while the lower section shows the output generated without PDFs, illustrating the impact of including detailed content. |
56 | | 
| This screenshot reveals the limitations of the current app, indicating that it may struggle with queries beyond the scope of the provided documents. It underscores the need for further improvements and extensive testing to enhance the model's accuracy and robustness. |
57 | | 
| This screenshot demonstrates the PDF Management & Statistics Dashboard, showcasing how users can view detailed statistics related to the uploaded PDFs and documents within the system. |
58 |
59 | ## Installation Instructions ⚙️
60 |
61 | ### System Requirements 🖥️
62 |
63 | For Ollama3.1 8B, ensure your system has at least 16 GB of RAM, and reasonable disk space for a reasonable performance; a GPU is recommended for models with 70B parameters or higher.
64 |
65 | ### 1. Download and Install Ollama
66 |
67 | To use the `Ollama` model, follow these steps to download and install it based on your operating system:
68 |
69 | 1. **Visit the Ollama Download Page:**
70 | Go to [Ollama Download Page](https://ollama.com/download).
71 |
72 | 2. **Download the Installer:**
73 | Choose the appropriate installer for your operating system and download it.
74 |
75 | 3. **Install Ollama:**
76 | Follow the installation instructions provided on the download page for your specific operating system.
77 |
78 | 4. **Verify Installation:**
79 | After installation, verify that Ollama is installed correctly by running the following command in your terminal or command prompt:
80 |
81 | ```bash
82 | ollama --version
83 |
84 | 5. **Run the Llama3 Model:**
85 | Once Ollama is installed, start the llama3 model by running the following command in your terminal or command prompt:
86 |
87 | ```bash
88 | ollama run llama3.1
89 | ```
90 | When you run `ollama run llama3.1`, it defaults to using configuration 8b unless you specify otherwise.
91 |
92 | ### 2. **Clone the Repository**:
93 |
94 | git clone https://github.com/AbdArdati/PDFQueryAI.git
95 | cd PDFQueryAI
96 |
97 | ### 3. **Create a Virtual Environment**:
98 |
99 | python3 -m venv venv
100 | source venv/bin/activate # On Windows use `venv\Scripts\activate`
101 |
102 | ### 4. **Install Dependencies**:
103 |
104 | pip install -r requirements.txt
105 |
106 | ### 5. **Run the Application**:
107 |
108 | python app.py
109 |
110 | ### Endpoints and Their Functions
111 |
112 | - **`/`**: Serves the main HTML page for user interaction.
113 |
114 | - **`/ai`**:
115 | - **Method**: `POST`
116 | - **Function**: Accepts a JSON request containing a query, passes it to the AI model, and returns the generated response.
117 |
118 | - **`/ask_pdf`**:
119 | - **Method**: `POST`
120 | - **Function**: Processes queries against uploaded PDFs. It uses a vector store to find relevant documents, generates an answer based on the context, and returns the answer along with sources and usage statistics.
121 |
122 | - **`/clear_chat_history`**:
123 | - **Method**: `POST`
124 | - **Function**: Clears the chat history stored in the global `chat_history` list.
125 |
126 | - **`/clear_db`**:
127 | - **Method**: `POST`
128 | - **Function**: Deletes all stored vector data and PDF files from the filesystem, effectively resetting the application’s state.
129 |
130 | - **`/list_pdfs`**:
131 | - **Method**: `GET`
132 | - **Function**: Lists all PDF files currently available in the storage directory.
133 |
134 | - **`/pdf`**:
135 | - **Method**: `POST`
136 | - **Function**: Handles PDF uploads. The file is saved, processed, and split into chunks. These chunks are then stored in a vector database for later querying.
137 |
138 | - **`/list_documents`**:
139 | - **Method**: `GET`
140 | - **Function**: Lists all documents in the vector store.
141 |
142 | - **`/delete_pdf`**:
143 | - **Method**: `POST`
144 | - **Function**: Deletes a specific PDF file and its associated documents from the vector store.
145 |
146 | - **`/delete_document`**:
147 | - **Method**: `POST`
148 | - **Function**: Deletes a specific document from the vector store by its ID.
149 |
150 | - **`/pdf_usage`**:
151 | - **Method**: `GET`
152 | - **Function**: Provides usage statistics on how often each PDF has been queried.
153 |
154 | ### Internal Functions 🔧
155 |
156 | - **`file_exists(filename)`**: Checks if a file with the given name exists in the directory.
157 |
158 | - **`compute_file_hash(file)`**: Computes the MD5 hash of a file to detect duplicates.
159 |
160 | - **`perform_ocr(pdf_path)`**: Placeholder function for performing Optical Character Recognition (OCR) on a PDF if it is not structured. This needs to be implemented with an actual OCR library like `pytesseract`.
161 |
162 | ### Error Handling ⚠️
163 |
164 | - The application includes basic error handling for missing data, file operations, and vector store interactions. For example, if a file is not found or an operation fails, the application returns an appropriate error message and HTTP status code.
165 |
166 | ### Release Information 🚀
167 |
168 | **Current Version:** `v1.0.0-beta`
169 |
170 | **Release Date:** `25/07/2024`
171 |
172 | **Description:** This is the beta release of AskItRight. It includes features for uploading, managing, and querying PDF documents using an AI model, as well as basic statistics and administrative functionalities.
173 |
174 | **Changelog:**
175 | - Initial beta release with core functionalities.
176 | - Added PDF upload, query, and management features.
177 | - Integrated AI model for querying PDF content.
178 | - Implemented basic statistics and database management features.
179 |
180 | ### Summary 📝
181 |
182 | Overall, the application provides a structured way to manage and interact with PDF documents using an AI model. It integrates file management, data processing, and querying capabilities into a Flask web service, allowing users to upload, query, and manage PDFs while also keeping track of usage statistics and providing administrative functionalities.
183 |
184 |
185 | ### Disclaimer 📝
186 | Please be aware that this application is provided "as is," without any guarantee of functionality or reliability. It's important to note that the codebase may not adhere to best practices in terms of tidiness and organisation, and there is significant room for improvement and bug fixes. While I try to enhance the application, I plan to release a new version in the future as time permits.
187 |
188 | The developers disclaim all responsibility for bugs or issues, stating use is at one's own risk. Users are encouraged to contribute but the developers are not liable for any damages from using the application.
189 |
190 | ## License 🛡️
191 |
192 | This project is licensed under the Apache 2.0 License. The core components of this application follow the work from [https://github.com/ThomasJay/RAG](https://github.com/ThomasJay/RAG) which is also licensed under Apache 2.0.
193 |
194 | ```
195 | Licensed under the Apache License, Version 2.0 (the "License");
196 | you may not use this file except in compliance with the License.
197 | You may obtain a copy of the License at
198 |
199 | http://www.apache.org/licenses/LICENSE-2.0
200 |
201 | Unless required by applicable law or agreed to in writing, software
202 | distributed under the License is distributed on an "AS IS" BASIS,
203 | WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
204 | See the License for the specific language governing permissions and
205 | limitations under the License.
206 | ```
207 |
208 | ## Code of Conduct
209 |
210 | We are committed to fostering a welcoming and inclusive community. Please read our [Code of Conduct](CODE_OF_CONDUCT.md) to understand our guidelines and expectations.
211 |
212 | ---
213 | **Developed by Abd Alsattar Ardati for the sake of exploring, learning, and sharing.**
214 | Visit my [website](https://abd.ardati.org) for more information or contact me at abd.alsattar.ardati @ gmail.
215 |
--------------------------------------------------------------------------------
/SECURITY.md:
--------------------------------------------------------------------------------
1 | # Security Policy
2 |
3 | ## Supported Versions
4 |
5 | As this project is primarily for personal experimentation and not intended for production deployment, I do not provide security updates or support for any versions. The project is in a developmental stage, and no security considerations have been incorporated. Use it at your own risk, as mentioned in the README.md.
6 |
7 | | Version | Supported |
8 | | ------- | ------------------ |
9 | | All | :x: |
10 |
11 | ## Reporting a Vulnerability
12 |
13 | Since this project is for personal use and experimentation only, I am not actively managing or addressing security vulnerabilities. However, if you encounter any issues or potential vulnerabilities, you may report them through the project's [issue tracker](https://github.com/AbdArdati/PDFQueryAI/issues).
14 |
15 | Please be aware that due to the experimental nature of this project, there are no guarantees of updates or fixes for reported issues. Reports are reviewed on a case-by-case basis, and while I appreciate feedback, there is no commitment to address or resolve security concerns promptly.
16 |
17 | Thank you for understanding.
18 |
--------------------------------------------------------------------------------
/app.py:
--------------------------------------------------------------------------------
1 | from app import create_app
2 |
3 | if __name__ == "__main__":
4 | app = create_app()
5 | app.run(debug=True) # Turn on debugging mode for development
6 | print("Flask app is running on http://127.0.0.1:5000")
7 |
--------------------------------------------------------------------------------
/app/__init__.py:
--------------------------------------------------------------------------------
1 | from flask import Flask
2 | __version__ = '1.0.0-beta'
3 |
4 | def create_app():
5 | app = Flask(__name__, template_folder='templates')
6 |
7 | # Register blueprints
8 | from .routes import bp
9 | app.register_blueprint(bp)
10 |
11 | # Additional configuration if needed
12 |
13 | return app
14 |
--------------------------------------------------------------------------------
/app/prompts.py:
--------------------------------------------------------------------------------
1 | from langchain.prompts import PromptTemplate
2 |
3 | # Define all prompts
4 | PROMPTS = {
5 | "General AI Assistant": PromptTemplate.from_template(
6 | """
7 | ${await markdownToHTML(result.answer) || result.error}
`; 52 | } catch (error) { 53 | responseDiv.innerHTML = `An error occurred while processing the query: ${error.message}
`; 54 | } 55 | } 56 | 57 | // Function to ask about PDF 58 | async function askPDF() { 59 | console.log('askPDF called'); // Debug log 60 | const query = document.getElementById('queryPDF').value; 61 | const responseDiv = document.getElementById('queryResponse'); 62 | const promptType = document.getElementById('promptType').value; 63 | 64 | if (!responseDiv) { 65 | console.error('Element with ID "queryResponse" is missing.'); 66 | return; 67 | } 68 | 69 | if (!query) { 70 | alert('Please enter a query.'); 71 | return; 72 | } 73 | 74 | responseDiv.innerHTML = ''; 75 | 76 | try { 77 | console.log('Sending API request...'); // Debug log 78 | const result = await apiRequest('/ask_pdf', { 79 | method: 'POST', 80 | headers: { 'Content-Type': 'application/json' }, 81 | body: JSON.stringify({ query, promptType }), 82 | }); 83 | 84 | console.log('API response received:', result); // Debug log 85 | 86 | if (result.disclaimer) { 87 | responseDiv.innerHTML = `${result.disclaimer}
`; 88 | } else { 89 | responseDiv.innerHTML = `${await markdownToHTML(result.answer) || result.error}
`; 90 | } 91 | 92 | if (result.pdf_usage) displayPDFUsage(result.pdf_usage); 93 | if (result.query_usage) displayQueryUsage(result.query_usage); 94 | } catch (error) { 95 | responseDiv.innerHTML = `An error occurred while processing the PDF query: ${error.message}
`; 96 | console.error('Error in askPDF:', error); // Debug log 97 | } 98 | } 99 | 100 | // Function to display PDF usage statistics 101 | function displayPDFUsage(pdfUsage) { 102 | const usageDiv = document.getElementById('pdfUsageStats'); 103 | usageDiv.innerHTML = 'No usage statistics available.
'; 107 | return; 108 | } 109 | 110 | usageDiv.innerHTML += 'No usage statistics available.
'; 122 | return; 123 | } 124 | 125 | usageDiv.innerHTML += '') // Paragraphs
137 | .replace(/\n/g, '
') // New lines
138 | .replace(/\t/g, ' ') // Tabs
139 | .replace(/^(\d+)\.\s+(.*)$/gm, (match, number, text) => `
${number}. ${text}
`) // Numbered lists 140 | .replace(/^\*\s+(.*)$/gm, (match, text) => `${html.replace(/<\/p>\s*
/g, '
').replace(/
\s*<\/p>/g, '')}
`; 146 | } 147 | 148 | // Event listeners for form submission on Enter key press 149 | document.addEventListener('DOMContentLoaded', () => { 150 | console.log('DOMContentLoaded fired'); // Debug log 151 | 152 | const queryPDFInput = document.getElementById('queryPDF'); 153 | const queryPDFButton = document.getElementById('askPDFButton'); 154 | if (queryPDFInput && queryPDFButton) { 155 | queryPDFInput.addEventListener('keypress', function(event) { 156 | if (event.key === 'Enter') { 157 | event.preventDefault(); 158 | queryPDFButton.click(); 159 | } 160 | }); 161 | } else { 162 | console.log('queryPDFInput or queryPDFButton is missing'); // Debug log 163 | } 164 | 165 | const queryInput = document.getElementById('query'); 166 | const queryButton = document.getElementById('askAIButton'); 167 | if (queryInput && queryButton) { 168 | queryInput.addEventListener('keypress', function(event) { 169 | if (event.key === 'Enter') { 170 | event.preventDefault(); 171 | queryButton.click(); 172 | } 173 | }); 174 | } else { 175 | console.log('queryInput or queryButton is missing'); // Debug log 176 | } 177 | 178 | const copyPromptButton = document.getElementById('copyPromptButton'); 179 | if (copyPromptButton) { 180 | copyPromptButton.addEventListener('click', copyTextAndSubmit); 181 | } else { 182 | console.log('copyPromptButton is missing'); // Debug log 183 | } 184 | 185 | // Fetch prompts when the page is loaded 186 | fetchPrompts(); 187 | }); 188 | 189 | // Function to fetch prompts from the server 190 | async function fetchPrompts() { 191 | try { 192 | const prompts = await apiRequest('/prompts'); 193 | const selectElement = document.getElementById('promptType'); 194 | 195 | if (selectElement) { 196 | selectElement.innerHTML = ''; 197 | for (const [key] of Object.entries(prompts)) { 198 | const option = document.createElement('option'); 199 | option.value = key; 200 | option.textContent = key; 201 | selectElement.appendChild(option); 202 | } 203 | } else { 204 | console.error('Prompt type select element is missing in the DOM.'); 205 | } 206 | } catch (error) { 207 | console.error('Error fetching prompts:', error); 208 | } 209 | } 210 | -------------------------------------------------------------------------------- /app/static/js/script.js: -------------------------------------------------------------------------------- 1 | async function apiRequest(url, options = {}) { 2 | try { 3 | const response = await fetch(url, options); 4 | 5 | // Check if the response status is not OK 6 | if (!response.ok) { 7 | // Attempt to parse the error response 8 | let errorMessage = `Failed to fetch from ${url}. Status: ${response.status}`; 9 | 10 | try { 11 | // Try parsing the response as JSON 12 | const errorData = await response.json(); 13 | errorMessage += `. Response: ${errorData.message || errorData.error || 'Unknown error'}`; 14 | } catch { 15 | // If JSON parsing fails, fall back to plain text response 16 | errorMessage += `. Response: ${await response.text()}`; 17 | } 18 | 19 | throw new Error(errorMessage); 20 | } 21 | 22 | // Try parsing JSON, if applicable 23 | try { 24 | return await response.json(); 25 | } catch (jsonError) { 26 | throw new Error(`Failed to parse JSON response from ${url}. Response: ${await response.text()}`); 27 | } 28 | } catch (error) { 29 | console.error('API request error:', error.message); 30 | throw error; // Re-throw error to be handled by the calling function 31 | } 32 | } 33 | 34 | 35 | async function uploadPDF() { 36 | const fileInput = document.getElementById('pdfFile'); 37 | const file = fileInput.files[0]; 38 | 39 | if (!file) return showToast('Please select a PDF file to upload.'); 40 | 41 | const formData = new FormData(); 42 | formData.append('file', file); 43 | 44 | try { 45 | const result = await apiRequest('/pdf', { 46 | method: 'POST', 47 | body: formData, 48 | }); 49 | 50 | const { status, filename, doc_len, chunk_len, error } = result; 51 | if (status === 'Successfully Uploaded') { 52 | showToast(`Success: ${status}\nFilename: ${filename}\nLoaded ${doc_len} documents\nLoaded len=${chunk_len} chunks`, type = 'success'); 53 | 54 | // Call listPDFs function after successful upload 55 | listPDFs(); 56 | } else { 57 | showToast(error || 'An error occurred during the upload.', type = 'error'); 58 | } 59 | } catch (error) { 60 | // Display only the error message for 400 Bad Request 61 | const errorMessage = error.message.includes('Status: 400') 62 | ? `${error.message.split('Response: ')[1]}` 63 | : `An error occurred while uploading the PDF: ${error.message}`; 64 | 65 | showToast(errorMessage, type = 'error'); 66 | } 67 | } 68 | 69 | 70 | // Function to clear chat history 71 | async function clearChatHistory() { 72 | try { 73 | const data = await apiRequest('/clear_chat_history', { 74 | method: 'POST', 75 | headers: { 'Content-Type': 'application/json' }, 76 | }); 77 | 78 | const statusElement = document.getElementById('chatHistoryStatus'); 79 | statusElement.innerText = data.status === "Chat history cleared successfully" ? 'Chat history cleared successfully.' : 'Failed to clear chat history.'; 80 | statusElement.classList.remove('fade-out'); 81 | 82 | setTimeout(() => statusElement.classList.add('fade-out'), 1500); 83 | } catch (error) { 84 | const statusElement = document.getElementById('chatHistoryStatus'); 85 | statusElement.innerText = `An error occurred while clearing chat history: ${error.message}`; 86 | statusElement.classList.remove('fade-out'); 87 | } 88 | 89 | } 90 | // Function to list PDFs 91 | async function listPDFs() { 92 | try { 93 | const result = await apiRequest('/list_documents'); 94 | 95 | const pdfList = document.getElementById('pdfList'); 96 | pdfList.innerHTML = ''; 97 | 98 | if (result.documents && result.documents.length > 0) { 99 | const seenPDFs = new Set(); 100 | result.documents.forEach(doc => { 101 | const source = doc.source; 102 | if (!seenPDFs.has(source)) { 103 | seenPDFs.add(source); 104 | 105 | const listItem = document.createElement('div'); 106 | listItem.className = 'pdf-item'; // Apply the new class 107 | listItem.textContent = source; 108 | 109 | const buttonContainer = document.createElement('div'); 110 | buttonContainer.className = 'button-container'; 111 | 112 | buttonContainer.appendChild(createButton('View', 'btn view-button', () => window.open(`/pdfs/${source}`, '_blank').focus())); 113 | buttonContainer.appendChild(createButton('Delete', 'btn delete-button', () => deletePDF(source))); 114 | 115 | listItem.appendChild(buttonContainer); 116 | pdfList.appendChild(listItem); 117 | } 118 | }); 119 | } else { 120 | pdfList.innerHTML = 'Manage your PDFs, ask questions, and view usage statistics all in one place.
25 |Use the dropdown menu below to choose an appropriate prompt template...
46 | 49 |Submit your questions to Ollama3.1 AI model with no embedding from PDFs. I would use this mostly to compare answers
70 | 71 | 72 | 73 |View statistics on how frequently different queries are made to the AI and PDF content.
89 | 90 |Monitor how often each uploaded PDF is accessed or queried within the system.
94 | 95 |View statistics related to the uploaded PDFs and documents in the system.
26 |Upload a new PDF document to be processed and made available for querying.
48 | 49 | 50 |View and access the list of PDFs that have been uploaded to the system.
59 | 60 |Clear all stored data and uploaded files from the system. Use this carefully as it will delete all records.
70 | 71 |Total PDFs: {{ pdf_count }}
79 |Total Documents in Chroma DB: {{ doc_count }}
86 |