#!/usr/bin/env python3
"""
verify_ene_schema.py — Verifies the presence and structure of Braid Eigensolid Compressor tables
in the ENE substrate schema and writes a validation receipt.
"""

import os
import json
import re
import hashlib
from pathlib import Path

SCHEMA_FILE = Path(__file__).parent / "ene_substrate_schema.sql"
RECEIPT_FILE = Path(__file__).parent.parent.parent / "shared-data" / "data" / "stack_solidification" / "ene_schema_validation_receipt.json"

REQUIRED_TABLES = [
    "ene.packages",
    "ene.relations",
    "ene.receipts",
    "ene.scars",
    "ene.vectors",
    "ene.ingest_events",
    "ene.sessions",
    "ene.routes",
    "ene.nspace_kv",
    "ene.gossip_surface_nodes",
    "ene.gossip_surface_edges",
    "ene.dsp_nodes",
    "ene.prover_state",
    "ene.prover_instances",
    "ene.sidon_labels",
    "ene.crossing_weights",
    "ene.eigensolid_snapshots",
    "ene.braid_strands"
]

def verify_schema():
    if not SCHEMA_FILE.exists():
        print(f"Error: Schema file not found at {SCHEMA_FILE}")
        return False

    content = SCHEMA_FILE.read_text()
    
    # Calculate sha256 of the schema file
    schema_sha256 = hashlib.sha256(content.encode('utf-8')).hexdigest()

    # Perform syntax checks: count open/close parenthesis
    open_paren = content.count('(')
    close_paren = content.count(')')
    if open_paren != close_paren:
        print(f"Warning: Parenthesis imbalance! '(' = {open_paren}, ')' = {close_paren}")

    # Check for presence of required tables
    missing_tables = []
    found_tables = []
    for table in REQUIRED_TABLES:
        # Match "CREATE TABLE IF NOT EXISTS table_name" or "CREATE TABLE table_name"
        pattern = rf"CREATE\s+TABLE\s+(?:IF\s+NOT\s+EXISTS\s+)?{re.escape(table)}\b"
        if re.search(pattern, content, re.IGNORECASE):
            found_tables.append(table)
        else:
            missing_tables.append(table)

    # Check for alterations in ene.receipts
    has_theorem_id = "theorem_id" in content and "ene.receipts" in content
    has_dispatch_path = "dispatch_path" in content and "ene.receipts" in content

    # Check constraint check
    has_no_floats_check = "weight_raw" in content and "REAL" not in content.split("crossing_weights")[1].split(");")[0]

    status = "valid"
    errors = []

    if missing_tables:
        status = "invalid"
        errors.append(f"Missing tables: {', '.join(missing_tables)}")
    if not has_theorem_id:
        status = "invalid"
        errors.append("Missing theorem_id column alteration in ene.receipts")
    if not has_dispatch_path:
        status = "invalid"
        errors.append("Missing dispatch_path column alteration in ene.receipts")
    if not has_no_floats_check:
        status = "invalid"
        errors.append("Float found or incorrect integer constraints in crossing_weights table")

    receipt = {
        "schema": "ene_schema_validation_receipt_v1",
        "schema_file_path": str(SCHEMA_FILE.relative_to(SCHEMA_FILE.parent.parent.parent)),
        "schema_sha256": schema_sha256,
        "status": status,
        "found_tables": found_tables,
        "missing_tables": missing_tables,
        "alterations_verified": {
            "theorem_id": has_theorem_id,
            "dispatch_path": has_dispatch_path
        },
        "errors": errors
    }

    # Write the receipt
    RECEIPT_FILE.parent.mkdir(parents=True, exist_ok=True)
    RECEIPT_FILE.write_text(json.dumps(receipt, indent=2))
    print(f"Validation complete. Status: {status}. Receipt written to {RECEIPT_FILE}")
    
    return status == "valid"

if __name__ == "__main__":
    import sys
    success = verify_schema()
    sys.exit(0 if success else 1)