main.py

import argparse
import json
from math import ceil

import queryTools
import physicalQueryTools
import index
import executionEngine

def main():
    parser = argparse.ArgumentParser()
    parser.add_argument("--r", required=False, default="./inputs/relations.json", help="path to relations json.")
    parser.add_argument("--s", required=False, default="./inputs/statistics.json", help="path to statistics json.")
    parser.add_argument("--q", required=False, default="./queries/Q1.json", help="path to query json.")
    args = parser.parse_args()

    print(f"\nrelations from {args.r}")
    print(f"statistics from {args.s}")
    print(f"query from {args.q}")

    # load data
    relations = loadFile(args.r)
    statistics = loadFile(args.s)
    logicalPlan = loadFile(args.q)
    print("\nORIGINAL LOGICAL PLAN")
    print(json.dumps(logicalPlan, indent=4))

    # create indexes
    indexes = index.loadIndexes(relations)

    # Query Rewrite
    logicalPlan = queryTools.rewriteLogicalPlan(logicalPlan, relations, statistics)
    # Get Cardonality and Join Order
    cardinality, finalLogicalPlan = queryTools.getCardinality(logicalPlan, statistics)
    print("\nREWRITTEN LOGICAL PLAN")
    print(json.dumps(finalLogicalPlan, indent=4))
    print(f"\nFINAL ESTIMATED CARDINALITY: {ceil(cardinality)}")

    # Physical Plans and Estimates
    physPlans = physicalQueryTools.createPhysicalPlans(finalLogicalPlan, indexes, statistics)
    ioEstimates = physicalQueryTools.estimatePhysPlans(physPlans, statistics)
    print("\nPHYSICAL PLANS")
    print(f"Number of plans: {len(physPlans)}")
    for num, physPlan in enumerate(physPlans):
        print(f"\nPhysical plan: {num+1}")
        print(f"Estimated cost: {ioEstimates[num]}")
        print(json.dumps(physPlan, indent=4))

    # Choose Physical Plan
    indexOfLowest = min(enumerate(ioEstimates), key=lambda x: x[1])[0]
    print(f"\nCHOSEN PHYSICAL PLAN: plan number {indexOfLowest+1}")
    
    # Execute Physical Plan
    ioCost, outputTuples, schema = executionEngine.execute(physPlans[indexOfLowest], indexes, relations)
    print(f"\nACTUAL IO COST: {ioCost}")

    print(f"\nOUTPUT TUPLES")
    print(f"schema: {schema}")
    for tuple in outputTuples:
        print(tuple)

def loadFile(directory: str):
    file = open(directory)
    result = json.load(file)
    file.close()
    return result

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