scripts/enrich_registry.py

from __future__ import annotations

"""
Targeted TGSC enrichment of the PINO aroma registry.

Processes a priority subset of single-chemical fragrance materials, fetches their
TGSC data sheets, and writes the extracted boiling point, vapor pressure
(mmHg -> Pa), and odor description back into the SQLite registry.
"""

import argparse
import logging
import re
import sys
from pathlib import Path
from typing import Any

sys.path.insert(0, str(Path(__file__).parent.parent / "src"))

from camoufox.sync_api import Camoufox
from pino.registry import AromaRegistry
from pino.scraper import search_tgsc, parse_tgsc_page_camofox

logging.basicConfig(
    level=logging.INFO,
    format="%(asctime)s %(levelname)s %(name)s: %(message)s",
)
logger = logging.getLogger("enrich_registry")


# Priority palette covering top, heart, and base notes. These are the materials
# that most strongly influence the VLE dry-down and the optimizer's search space.
PRIORITY_CAS = [
    "5989-27-5",  # d-Limonene (citrus top)
    "78-70-6",    # Linalool (floral/citrus top-heart)
    "5392-40-5",  # Citral (lemon/citrus top)
    "106-24-1",   # Geraniol (rosy floral heart)
    "106-25-2",   # Nerol (citrus-rose heart)
    "115-95-7",   # Linalyl acetate (bergamot/citrus top)
    "80-56-8",    # alpha-Pinene (pine/citrus top)
    "99-85-4",    # Gamma-terpinene (citrus peel)
    "91-64-5",    # Coumarin (sweet hay base)
    "121-33-5",   # Vanillin (sweet gourmand base)
    "105-95-3",   # Ethylene brassylate (musk base)
    "77-53-2",    # Cedrol (cedarwood base)
    "87-44-5",    # Caryophyllene (spicy woody heart)
    "97-53-0",    # Eugenol (spicy clove heart)
    "119-36-8",   # Methyl salicylate (wintergreen top)
    "118-58-1",   # Benzyl salicylate (floral musk base)
    "1222-05-5",  # Galaxolide (musk base)
    "21145-77-7", # Tonalid (musk base)
    "4602-84-0",  # Farnesol (floral woody base)
    "7212-44-4",  # Nerolidol (woody floral base)
    "104-67-6",   # Gamma-undecalactone (peach lactone base)
    "120-57-0",   # Anisaldehyde (anise top)
    "124-13-0",   # Octanal (citrus aldehyde top)
    "112-31-2",   # Decanal (waxy aldehyde top)
    "140-11-4",   # Benzyl acetate (jasmin floral heart)
    "141-12-8",   # Linalyl acetate (duplicate alias)
    "24851-98-7", # Methyl dihydrojasmonate (jasmine heart)
    "127-51-5",   # alpha-Isomethyl ionone (powdery floral heart)
    "14901-07-6", # Gamma-methyl ionone (floral heart)
    "16409-43-1", # Cashmeran (woody musk base)
    "33704-61-9", # Cashmeran (alternative) - skipped, duplicate
]


def _is_single_chemical(cas: str) -> bool:
    """Natural oils and absolutes have 8000/9000 CAS prefixes; skip them."""
    return not re.match(r"^(8\d{3}|9\d{3})", cas)


def enrich_priority(
    registry: AromaRegistry,
    priority: list[str],
    limit: int | None = None,
) -> dict[str, Any]:
    """
    Enrich a priority list of CAS entries from TGSC.

    Returns a summary dict with counts and the list of updated records.
    """
    rows = registry._conn.execute(
        "SELECT cas, name FROM aroma_chemicals WHERE cas IN ("
        + ",".join("?" * len(priority))
        + ") ORDER BY cas",
        tuple(priority),
    ).fetchall()

    if limit is not None:
        rows = rows[:limit]

    logger.info("Enriching %d priority materials from TGSC", len(rows))

    enriched: list[dict[str, Any]] = []
    skipped = 0
    failed = 0

    with Camoufox(headless=True) as browser:
        for cas, name in rows:
            if not _is_single_chemical(cas):
                skipped += 1
                continue

            try:
                candidates = search_tgsc(name, browser=browser)
            except Exception as exc:
                logger.warning("Search failed for %s (%s): %s", cas, name, exc)
                failed += 1
                continue

            if not candidates:
                logger.warning("No TGSC results for %s (%s)", cas, name)
                failed += 1
                continue

            scraped = None
            for candidate in candidates:
                try:
                    scraped = parse_tgsc_page_camofox(candidate["url"], browser=browser)
                except Exception as exc:
                    logger.debug("Scrape failed for %s: %s", candidate["url"], exc)
                    continue
                if scraped.get("vapor_pressure_pa") is not None:
                    scraped["url"] = candidate["url"]
                    scraped["name"] = candidate["name"]
                    break

            if scraped is None or scraped.get("vapor_pressure_pa") is None:
                failed += 1
                continue

            registry._conn.execute(
                """
                UPDATE aroma_chemicals
                SET boiling_point_k = COALESCE(?, boiling_point_k),
                    vapor_pressure_pa = COALESCE(?, vapor_pressure_pa),
                    odor_description = COALESCE(?, odor_description)
                WHERE cas = ?
                """,
                (
                    scraped.get("boiling_point_k"),
                    scraped.get("vapor_pressure_pa"),
                    scraped.get("odor_description", ""),
                    cas,
                ),
            )
            registry._conn.commit()
            enriched.append({"cas": cas, "name": name, **scraped})
            logger.info(
                "Enriched %s (%s): VP=%.4f Pa, BP=%.2f K, odor=%r",
                cas,
                name,
                scraped["vapor_pressure_pa"],
                scraped["boiling_point_k"] or 0.0,
                scraped.get("odor_description", "")[:60],
            )

    return {
        "enriched": len(enriched),
        "skipped": skipped,
        "failed": failed,
        "records": enriched,
    }


def main() -> int:
    parser = argparse.ArgumentParser(description="Targeted TGSC registry enrichment")
    parser.add_argument(
        "--registry",
        default="src/pino/registry.db",
        help="Path to the SQLite aroma registry",
    )
    parser.add_argument(
        "--limit",
        type=int,
        default=None,
        help="Process only the first N priority materials",
    )
    parser.add_argument(
        "--priority",
        default=None,
        help="Comma-separated list of CAS numbers to enrich instead of the default priority list",
    )
    parser.add_argument(
        "--log-level",
        default="INFO",
        help="Logging level",
    )
    args = parser.parse_args()

    logging.getLogger().setLevel(getattr(logging, args.log_level.upper(), logging.INFO))

    registry = AromaRegistry(args.registry)
    priority = [c.strip() for c in args.priority.split(",")] if args.priority else PRIORITY_CAS

    summary = enrich_priority(registry, priority, limit=args.limit)
    print(summary)
    return 0


if __name__ == "__main__":
    raise SystemExit(main())