"""FDK reģistra pārbaude: XSD, ID unikalitāte, avotu esamība un pilnums, reģistrs = iestāžu failu apvienojums.

python3 tools/validate.py --nolikumi ../Nolikumi-as-Code
"""
import argparse
import glob
import os
import sys

from lxml import etree

NS = "{urn:pppa:vpk:nolikums:0.1}"
FNS = "{urn:pppa:vpk:fdk:0.1}"
FILES = ("funkcijas.xml", "darbi.xml", "kompetence.xml")


def main():
    ap = argparse.ArgumentParser()
    ap.add_argument("--nolikumi", required=True)
    a = ap.parse_args()
    schema = etree.XMLSchema(etree.parse("schemas/registrs-0.1.xsd"))
    errors, ids, by_file = [], {}, {}
    points, must = set(), set()
    for f in glob.glob(os.path.join(a.nolikumi, "data", "*.xml")):
        for p in etree.parse(f).getroot().iter(NS + "Point"):
            points.add(p.get("id"))
            t = (p.findtext(NS + "Text") or "").strip().lower()
            if p.get("kind") in ("function", "task", "right") and not t.startswith("(svītrots"):
                must.add(p.get("id"))
    used = {}
    for path in sorted(glob.glob("iestades/*/*.xml")) + sorted(glob.glob("registrs/*.xml")):
        doc = etree.parse(path)
        if not schema.validate(doc):
            errors += [f"{path}: {e.message}" for e in schema.error_log][:5]
        recs = [e for e in doc.getroot() if isinstance(e.tag, str)]
        by_file[path] = {e.get("id") for e in recs}
        if path.startswith("iestades/"):
            org = path.split("/")[1]
            for e in recs:
                i, av = e.get("id"), e.get("avots")
                if i in ids:
                    errors.append(f"{i}: divreiz ({ids[i]}, {path})")
                ids[i] = path
                if e.get("iestāde") != org or i[2:9] != org:
                    errors.append(f"{i}: iestāde nesakrīt ar mapi {org}")
                if av not in points:
                    errors.append(f"{i}: avots {av} nav Nolikumi-as-Code")
                if av in used:
                    errors.append(f"{av}: izmantots divreiz ({used[av]}, {i})")
                used[av] = i
                if not (e.findtext(FNS + "Formulējums") or "").strip():
                    errors.append(f"{i}: tukšs formulējums")
    built = {os.path.basename(d) for d in glob.glob("iestades/*")}
    must = {m for m in must if m[7:14] in built}  # tikai iestādes, kas ir uzbūvētas
    missing = sorted(must - set(used))
    errors += [f"{m}: funkcijas/uzdevuma/tiesību punkts nav reģistrā" for m in missing[:20]]
    for fn in FILES:
        inst = set().union(*[v for k, v in by_file.items() if k.startswith("iestades/") and k.endswith("/" + fn)] or [set()])
        if f"registrs/{fn}" in by_file and by_file[f"registrs/{fn}"] != inst:
            errors.append(f"registrs/{fn}: nesakrīt ar iestāžu failu apvienojumu")
    n = {fn: sum(len(v) for k, v in by_file.items() if k.startswith("iestades/") and k.endswith("/" + fn)) for fn in FILES}
    print(f"{len(by_file)} faili, {len(ids)} ieraksti {n}; obligātie punkti {len(must)}, iztrūkst {len(missing)}")
    if errors:
        print("\n".join(errors[:60]))
        sys.exit(1)
    print("OK")


if __name__ == "__main__":
    main()