scripts/map_sections.py

#!/usr/bin/env python3
"""
Diagnostic: print paragraphs of target clauses from a TS DOCX with indices and styles.

Usage:
    python3 map_sections.py <ts_path.docx> "11.1.22.3.2" "14.5.6"
"""

import sys
from docx import Document

def main():
    if len(sys.argv) < 3:
        print("Usage: map_sections.py <ts.docx> <clause1> [clause2 ...]")
        sys.exit(1)

    ts_path = sys.argv[1]
    clauses = sys.argv[2:]

    doc = Document(ts_path)
    in_section = None

    for i, para in enumerate(doc.paragraphs):
        text  = para.text.strip()
        style = para.style.name

        matched = False
        for clause in clauses:
            if clause in text and ('Heading' in style or 'heading' in style.lower()):
                in_section = clause
                print(f'\n=== [{i}] SECTION {clause} | style={style!r} ===')
                print(f'  [{i}] style={style!r:16s} | "{text}"')
                matched = True
                break

        if not matched and in_section:
            if 'Heading' in style and text:
                print(f'  --- section ends at [{i}] style={style!r}: "{text[:60]}"')
                in_section = None
            elif text:
                print(f'  [{i}] style={style!r:16s} | "{text[:100]}"')

if __name__ == '__main__':
    main()