This script extracts tables from LCSO Reports (PDFs) and formats them as Markdown for Telegram messages (using <pre> tags for proper rendering).

lcso_report_parser.py

import pymupdf


def generate_markdown_table(rows):
    """Generates a Markdown table string wrapped in <pre> tags."""
    if not rows:
        return "No incidents reported"

    headers = ["Station", "Date / Time", "Location", "Incident"]

    # Build the Markdown table string
    table_md = "<pre>\n"
    table_md += "|" + "|".join(headers) + "|\n"
    table_md += "|" + "|".join("-" * len(header) for header in headers) + "|\n"
    for row in rows:
        table_md += "|" + "|".join(str(cell).replace("\n", " ") for cell in row) + "|\n"
    table_md += "</pre>"

    return table_md


def process_pdf(pdf_path):
    """Extracts data from the PDF and returns a list of rows."""
    doc = pymupdf.open(pdf_path)
    rows = []
    for page in doc:
        tabs = page.find_tables()
        if tabs.tables:
            for tab in tabs:
                table_data = tab.extract()
                if table_data:
                    station_info = table_data[0][0].split("\n")[0]
                    data = table_data[2:]

                    for row in data:
                        # Add station info to each row
                        row.insert(0, station_info)
                        rows.append(row)

    return rows


if __name__ == "__main__":
    pdf_path = "LCSO Report - 2025-02-18.pdf"
    rows = process_pdf(pdf_path)
    markdown_table = generate_markdown_table(rows)
    print(markdown_table)

    print()

    pdf_path = "LCSO Report - 2025-02-19.pdf"
    rows = process_pdf(pdf_path)
    markdown_table = generate_markdown_table(rows)
    print(markdown_table)

Output Example:

Station	Date / Time	Location	Incident
Eastern Loudoun Station	02/18/2025 5:00 p.m.	46000 block Woodstone Ter., Sterling	Vehicle Tampering: Complainant reported that her vehicle was damaged. SO250002847
Ashburn Station	02/10/2025- 02/17/2025 12:00 a.m.	43000 block Postrail Sq., Ashburn	Vehicle Tampering: Complainant reported that her vehicle was damaged due to being scratched on all sides. SO250002877
Ashburn Station	02/18/2025 7:30 a.m.	20000 block Trails End Ter., Ashburn	Larceny: Complainant reported that his vehicle was entered, and items were taken. SO250002842
Dulles South Station			No significant incidents
Western Loudoun Station			No significant incidents
Scams			No significant incidents