HC_APTBS/seed_aliases.py

"""One-shot seeder: inject <Aliases> blocks into pumps.xml from the legacy
KlineIDs comma-string at old_source/Herlic2.0/App.config.

Reads canonical -> [aliases] map (hard-coded below from the legacy default),
finds each <Pump id="..."> in pumps.xml, and inserts an <Aliases><KlineId>...
block as the first child of that <Pump> element.

Idempotent: if a <Pump> already has an <Aliases> child, the script merges new
aliases into it without creating duplicates.
"""
import sys
from pathlib import Path
import xml.etree.ElementTree as ET

# Source: old_source/Herlic2.0/App.config -> KlineIDs setting (default value).
# Format: "klineID:canonicalPumpID,..."
LEGACY_KLINEIDS = (
    "0470004007:0470004002,0470004008:0470004006,0470004012:0470004004,"
    "1093412060:0470504028,1093412050:0470504012,1093424041:0470504034,"
    "0986444003:0470504004,1093412071:0470504033,1093423001:0470504027,"
    "0986444012:0470504011,1093412070:0470504033,0986444005:0470504009,"
    "1093424051:0470504046,1093424025:1093424026,0986444002:0470504003,"
    "1093424027:1093424026,1093423002:0470504027,1093424024:1093424026,"
    "0986444006:0470506002,0986444011:0470504010,0986444044:0470506037,"
    "1093421004:0470504026,1093412072:0470504033,1093411003:0470504030,"
    "1093424040:0470504034,0986444019:0470504020,0986444004:0470504005,"
    "1093411022:0470504037,0986444042:0470506017,0986444014:0470504015,"
    "1093421006:0470504026,1093421005:0470504026,1093424080:0470504046,"
    "0986444026:0470506030,1093411004:0470504030,0986444010:0470506009,"
    "0986444008:0470506006,"
)

PUMPS_XML = Path.home() / ".HC_APTBS" / "config" / "pumps.xml"


def parse_legacy(s: str) -> dict[str, list[str]]:
    """Group legacy entries by canonical pump ID."""
    aliases_by_canonical: dict[str, list[str]] = {}
    for entry in s.split(","):
        entry = entry.strip()
        if not entry or ":" not in entry:
            continue
        kline_id, canonical = entry.split(":", 1)
        kline_id = kline_id.strip()
        canonical = canonical.strip()
        if not kline_id or not canonical:
            continue
        bucket = aliases_by_canonical.setdefault(canonical, [])
        if kline_id not in bucket:
            bucket.append(kline_id)
    return aliases_by_canonical


def main() -> int:
    aliases_by_canonical = parse_legacy(LEGACY_KLINEIDS)
    print(f"Parsed {sum(len(v) for v in aliases_by_canonical.values())} aliases "
          f"across {len(aliases_by_canonical)} canonical pumps.")

    if not PUMPS_XML.exists():
        print(f"ERROR: {PUMPS_XML} does not exist.", file=sys.stderr)
        return 1

    tree = ET.parse(PUMPS_XML)
    root = tree.getroot()

    pumps_by_id: dict[str, ET.Element] = {}
    for pump in root.iter("Pump"):
        pid = pump.attrib.get("id")
        if pid:
            pumps_by_id[pid] = pump

    missing = sorted(set(aliases_by_canonical) - set(pumps_by_id))
    if missing:
        print(f"WARNING: {len(missing)} canonical pumps from legacy data not in "
              f"pumps.xml — skipping: {missing}", file=sys.stderr)

    injected_pumps = 0
    injected_aliases = 0
    skipped_existing = 0

    for canonical, kline_ids in sorted(aliases_by_canonical.items()):
        pump = pumps_by_id.get(canonical)
        if pump is None:
            continue

        existing = pump.find("Aliases")
        if existing is None:
            existing = ET.Element("Aliases")
            pump.insert(0, existing)
            existing.text = "\n        "
            existing.tail = "\n      "

        already = {(c.tag, (c.text or "").strip())
                   for c in existing if c.tag in ("KlineId", "ModelRef")}

        added_here = 0
        for k in kline_ids:
            key = ("KlineId", k)
            if key in already:
                skipped_existing += 1
                continue
            kid = ET.SubElement(existing, "KlineId")
            kid.text = k
            kid.tail = "\n        "
            already.add(key)
            added_here += 1
            injected_aliases += 1

        if added_here:
            injected_pumps += 1
            # Fix the tail of the last child so the closing tag is on its own line.
            last = list(existing)[-1]
            last.tail = "\n      "

    tree.write(PUMPS_XML, encoding="utf-8", xml_declaration=True)
    print(f"Injected {injected_aliases} aliases into {injected_pumps} pumps.")
    if skipped_existing:
        print(f"  ({skipped_existing} aliases were already present, skipped.)")
    return 0


if __name__ == "__main__":
    sys.exit(main())