ajsbsd/occrp_trump_pages

OCCRP Donald Trump Coverage Archive

This dataset is independently curated and not affiliated with or endorsed by OCCRP.

A curated collection of raw HTML snapshots from the Organized Crime and Corruption Reporting Project (OCCRP) Donald Trump person page.

📋 Dataset Overview

Field	Value
Source	https://www.occrp.org/en/people/donald-trump/page/[1-17]
Pages Included	1–17 (as of January 2026)
Total Size	~8.51 MB
Format	Raw HTML files (page_1.html through page_17.html)
License	MIT
Language	English

🔍 What's Included

• Full HTML source for each paginated listing page
• Embedded article metadata (titles, dates, URLs, summaries)
• Project tags (e.g., "The FinCEN Files", "Paradise Papers")
• Image alt text and visual context clues
• Navigation and pagination structure

🎯 Intended Use Cases

• Journalistic research: Track evolution of OCCRP's Trump coverage over time
• Media analysis: Study framing, tone, and topic emphasis in investigative reporting
• NLP/ML training: Extract structured data for text classification, entity recognition, or timeline modeling
• Educational purposes: Teach web scraping, HTML parsing, or data journalism workflows

⚠️ Ethical Guidelines

1. Attribution: Always credit OCCRP as the original source
2. Non-commercial use: This archive is for research/education; contact OCCRP for commercial licensing
3. Respect robots.txt: Do not re-scrape without checking current site policies
4. Context matters: HTML structure may change; verify parsed data against live pages when possible

📦 File Structure

occrp_trump_pages/
├── page_1.html ... page_17.html # Raw HTML snapshots
├── parse_occrp.py # Extracts structured data from HTML
├── query_occrp.py # CLI tool to search parsed data
├── articles.json # Generated: parsed article data
├── articles.csv # Generated: flat format for spreadsheets
└── README.md # This file

🛠️ Getting Started

Step 1: Set Up Environment

# Create and activate virtual environment
python3 -m venv occrp-env
source occrp-env/bin/activate

Install dependencies

pip install beautifulsoup4 lxml pandas

Step 2: Parse HTML to Structured Data

Run the parser (generates articles.json + articles.csv)

python parse_occrp.py --input occrp_trump_pages/ --output articles

Step 3: Query the Data

Basic keyword search

python query_occrp.py --search "Ukraine" --limit 10

Filter by date range

python query_occrp.py --date-from "2020-01-01" --date-to "2020-12-31"

Filter by investigation project

python query_occrp.py --project "FinCEN Files" --format json

Export to CSV for spreadsheet analysis

python query_occrp.py --search "sanctions" --format csv > results.csv

🔄 Loading in Python

Load parsed JSON directly

import json
with open("articles.json") as f:
    articles = json.load(f)

Or use Hugging Face datasets (if articles.json is in repo)

from datasets import load_dataset
ds = load_dataset("ajsbsd/occrp_trump_pages", data_files="articles.json", split="train")
df = ds.to_pandas()
print(df[["title", "date", "url"]].head())

📤 Upload to Hugging Face

cd occrp_trump_pages/
git add articles.json projects.json timeline.json metadata.json
git commit -m "Add parsed JSON datasets"
git push

🔍 Quick Preview Command

from datasets import load_dataset
import json

# Load main articles
ds = load_dataset("json", data_files="articles.json", split="train")
print(f"Loaded {len(ds)} articles")

# Load projects summary
with open("projects.json") as f:
    projects = json.load(f)
print(f"Found {len(projects)} investigation projects")