breakpilot-compliance/scripts/apply_block_markers_022.py

#!/usr/bin/env python3
"""
Migration 022: Insert [BLOCK:ID]...[/BLOCK:ID] markers into DB templates.

Targets:
  - NDA DE + NDA EN:   wraps Vertragsstrafe section → NDA_PENALTY_BLOCK
  - Cookie Banner DE:  wraps Analyse section        → COOKIE_ANALYTICS_BLOCK
                       wraps Marketing section      → COOKIE_MARKETING_BLOCK

Usage:
  python3 apply_block_markers_022.py

  Or via Docker:
  docker cp apply_block_markers_022.py bp-compliance-backend:/tmp/
  docker exec bp-compliance-backend python3 /tmp/apply_block_markers_022.py
"""

import os
import re
import sys

# Allow running inside container where the app is at /app
sys.path.insert(0, '/app')

from sqlalchemy import create_engine, text

# ─────────────────────────────────────────────────────────────────────────────
# DB connection
# ─────────────────────────────────────────────────────────────────────────────

DATABASE_URL = os.environ.get(
    'DATABASE_URL',
    'postgresql://compliance_user:compliance_pass@bp-core-postgres:5432/breakpilot_db'
)

engine = create_engine(DATABASE_URL)


# ─────────────────────────────────────────────────────────────────────────────
# Marker helpers
# ─────────────────────────────────────────────────────────────────────────────

def wrap_block(content: str, block_id: str, pattern: str, flags: int = re.MULTILINE | re.DOTALL) -> tuple[str, int]:
    """
    Finds the first match of `pattern` in `content` and wraps it with
    [BLOCK:block_id]...[/BLOCK:block_id].

    Returns (new_content, match_count).
    """
    match_count = 0

    def replacer(m: re.Match) -> str:
        nonlocal match_count
        match_count += 1
        matched = m.group(0)
        # Avoid double-wrapping
        if f'[BLOCK:{block_id}]' in matched:
            return matched
        return f'[BLOCK:{block_id}]\n{matched}[/BLOCK:{block_id}]\n'

    new_content = re.sub(pattern, replacer, content, flags=flags)
    return new_content, match_count


# ─────────────────────────────────────────────────────────────────────────────
# Template-specific transformations
# ─────────────────────────────────────────────────────────────────────────────

def apply_nda_penalty_block(content: str) -> tuple[str, int]:
    """Wraps the Vertragsstrafe section in NDA templates.

    Matches bold-style heading: **N. Vertragsstrafe ...**
    up to the next bold-numbered section or end of string.
    """
    pattern = r'(\*\*\d+\.\s+[^\n]*[Vv]ertragsstrafe[^\n]*\*\*\n)(.*?)(?=\*\*\d+\.|\Z)'
    return wrap_block(content, 'NDA_PENALTY_BLOCK', pattern)


def apply_cookie_analytics_block(content: str) -> tuple[str, int]:
    """Wraps the Analyse section in Cookie Banner templates.

    Matches: **Abschnitt „Analyse":**
    up to the next **Abschnitt or end of string.
    """
    pattern = r'(\*\*Abschnitt\s+[^\n]*Analyse[^\n]*\*\*[^\n]*\n)(.*?)(?=\*\*Abschnitt\s+[^\n]*Marketing|\Z)'
    return wrap_block(content, 'COOKIE_ANALYTICS_BLOCK', pattern)


def apply_cookie_marketing_block(content: str) -> tuple[str, int]:
    """Wraps the Marketing section in Cookie Banner templates.

    Matches: **Abschnitt „Marketing":**
    up to the next double-newline section divider or ---  or end.
    """
    pattern = r'(\*\*Abschnitt\s+[^\n]*Marketing[^\n]*\*\*[^\n]*\n)(.*?)(?=\n---|\n\*\*[A-Z]\)|\Z)'
    return wrap_block(content, 'COOKIE_MARKETING_BLOCK', pattern)


# ─────────────────────────────────────────────────────────────────────────────
# Main
# ─────────────────────────────────────────────────────────────────────────────

TARGETS = [
    # (document_type_filter, language_filter, list of transform functions)
    ('nda', 'de', [apply_nda_penalty_block]),
    ('nda', 'en', [apply_nda_penalty_block]),
    ('cookie_banner', 'de', [apply_cookie_analytics_block, apply_cookie_marketing_block]),
]

def main() -> None:
    print('=== Migration 022: Block Markers ===\n')

    with engine.begin() as conn:
        for doc_type, lang, transforms in TARGETS:
            rows = conn.execute(
                text(
                    'SELECT id, title, content FROM public.compliance_legal_templates '
                    'WHERE document_type = :doc_type AND language = :lang'
                ),
                {'doc_type': doc_type, 'lang': lang}
            ).fetchall()

            if not rows:
                print(f'[SKIP] No templates found for {doc_type}/{lang}')
                continue

            for row in rows:
                tid, title, content = row.id, row.title, row.content
                if content is None:
                    print(f'[SKIP] {title} (id={tid}) — content is NULL')
                    continue

                original_len = len(content)
                new_content = content
                total_matches = 0

                for transform in transforms:
                    new_content, match_count = transform(new_content)
                    total_matches += match_count

                if new_content == content:
                    print(f'[NOOP] {title} ({doc_type}/{lang}) — no changes')
                    continue

                conn.execute(
                    text(
                        'UPDATE public.compliance_legal_templates '
                        'SET content = :content, updated_at = NOW() '
                        'WHERE id = :id'
                    ),
                    {'content': new_content, 'id': tid}
                )
                print(
                    f'[OK]   {title} ({doc_type}/{lang})'
                    f' | {original_len} → {len(new_content)} chars'
                    f' | {total_matches} block(s) wrapped'
                )

    print('\n=== Done ===')


if __name__ == '__main__':
    main()