DocuCheck

---

🧠 Overview

DocuCheck is a Python-based tool that extracts factual claims from documents (like PDFs) and performs automated fact-checking using a generative AI model (Google’s Gemini).
It produces a human-readable HTML report (report.html) summarizing extracted claims, internal consistency checks, and external verification results.

---

⚙️ Features

• Structural Text Extraction: Parses PDFs to understand document structure (headings vs. paragraphs) for better context.
• AI-Powered Claim Extraction: Uses a generative model to identify and extract factual claims from text.
• Internal Consistency Analysis: Detects contradictions within the document’s own claims.
• External Fact-Checking: Verifies claims against the model’s external knowledge to identify outdated or invalid information.
• Modern HTML Reporting: Generates a clean, single-page HTML report with a visual summary dashboard.
• Result Caching: Caches analysis results to avoid re-processing and repeated API calls.

---

🗂️ Project Structure

Docucheck/
├── main.py           # Main CLI entry point with argument parsing
├── extractor.py      # Handles PDF parsing and claim extraction
├── verifier.py       # Handles internal/external fact-checking
├── reporter.py       # Generates the final HTML report
├── caching.py        # Manages file hashing and result caching
└── utils.py          # Shared utilities (e.g., JSON parsing)

run.py                # The main script to execute the package
requirements.txt      # Python dependencies
.env.example          # Example environment variables
.gitignore            # Ignores .env, .venv, pycache, etc.
LICENSE               # MIT License
README.md             # This file

---

🚀 Quick Start

1. Installation

Clone the repository and navigate into the project directory:

git clone https://github.com/<your-username>/docucheck.git
cd docucheck

---

2. Set Up Virtual Environment

Windows (PowerShell):

python -m venv .venv
.\.venv\Scripts\Activate.ps1

macOS/Linux (bash):

python3 -m venv .venv
source .venv/bin/activate

---

3. Install Dependencies

pip install -r requirements.txt

---

4. Set Up Environment

Copy the example .env file and add your Gemini API key.

Windows (PowerShell):

copy .env.example .env
# Now edit the .env file with Notepad or VS Code

macOS/Linux (bash):

cp .env.example .env
# Now edit the .env file with nano, vim, or VS Code

Your .env file should look like this:

GEMINI_API_KEY=your-api-key-goes-here

---

🧩 Usage

Run the application using run.py, passing the path to the document you want to analyze.

Basic Example

python run.py "path/to/your/document.pdf"

This will analyze the PDF and save the report as report.html in the same directory.

---

Command-Line Arguments

Argument	Description	Required	Default
input_file	Path to the input file (.pdf, .txt, etc.)	✅	—
-o, --output	Path to save the output HTML report	❌	report.html
-l, --limit	Limit the number of claims to externally fact-check (0 = all)	❌	0
--force	Force re-analysis and bypass cached results	❌	False

---

Full Example

python run.py "my_research.pdf" -o "MyReport.html" -l 5 --force

This command:

• Analyzes my_research.pdf
• Saves the report as MyReport.html
• Only fact-checks the first 5 claims
• Bypasses the cache

---

🧠 How It Works

1. run.py executes the main() function in Docucheck/__main__.py.
2. The script parses command-line arguments.
3. caching.py generates a SHA-256 hash of the input file and checks for a cached result.
4. If cache exists (and --force not used), analysis is skipped and report generation begins.
5. If no cache:
   • extractor.py uses PyMuPDF to extract structured text.
   • Text is sent to Gemini API for claim extraction.
6. verifier.py:
   • Checks for internal contradictions.
   • Performs external fact-checking using Gemini’s knowledge.
7. reporter.py compiles all results into a single HTML report.
8. caching.py saves all results (claims, contradictions, checks) to a .Docucheck_Cache JSON file.

---

📄 License

Distributed under the MIT License.
See LICENSE for more information.