#!/usr/bin/env python3
# validate_padnext_2_12.py
# Validate PADneXt 2.12 ADL and AUF XML files against XSDs using lxml.
# Usage:
#   python validate_padnext_2_12.py <XML_FILE> <XSD_FILE> [<XML_FILE> <XSD_FILE> ...]

from lxml import etree
from pathlib import Path
import sys
import textwrap

def load_schema(xsd_path: Path) -> etree.XMLSchema:
    parser = etree.XMLParser(remove_blank_text=True)
    doc = etree.parse(str(xsd_path), parser)
    return etree.XMLSchema(doc)

def validate(xml_path: Path, schema: etree.XMLSchema):
    parser = etree.XMLParser(remove_blank_text=True)
    doc = etree.parse(str(xml_path), parser)
    ok = schema.validate(doc)
    return ok, schema.error_log

def main(argv):
    if len(argv) < 3 or len(argv[1:]) % 2 != 0:
        print("Usage: python validate_padnext_2_12.py <XML_FILE> <XSD_FILE> [<XML_FILE> <XSD_FILE> ...]")
        return 2

    # Process pairs
    for i in range(1, len(argv), 2):
        xml_path = Path(argv[i])
        xsd_path = Path(argv[i+1])

        print(f"— Validate —")
        print(f"  XML: {xml_path}")
        print(f"  XSD: {xsd_path}")

        try:
            schema = load_schema(xsd_path)
        except Exception as e:
            print("  ❌ Failed to load schema:")
            print(textwrap.indent(str(e), "    "))
            print()
            continue

        try:
            ok, log = validate(xml_path, schema)
        except Exception as e:
            print("  ❌ Exception during validation:")
            print(textwrap.indent(str(e), "    "))
            print()
            continue

        if ok:
            print("  ✅ VALID\n")
        else:
            print("  ❌ NOT VALID")
            for entry in log:
                loc = f"{Path(entry.filename).name if entry.filename else xml_path.name}:{entry.line}:{entry.column}"
                print(f"    - {loc}: {entry.message}")
            print()

    return 0

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