build_opcode_pack.py

import csv
import os
import zipfile
import json


LANGUAGES = ["python", "javascript", "c", "rust", "go", "lua", "sql"]

BASE_OPCODES = [
    "LOAD", "STORE", "MOVE", "ADD", "SUB", "MUL", "DIV",
    "CMP", "JMP", "JZ", "JNZ", "CALL", "RET", "PRINT"
]


def expand_opcodes(n=1000):
    ops = []

    for i in range(n):
        base = BASE_OPCODES[i % len(BASE_OPCODES)]
        ops.append(f"{base}_{i}")

    return ops


def generate_language_map(lang, opcodes):
    filename = f"{lang}_map.csv"

    with open(filename, "w", newline="") as f:
        writer = csv.writer(f)

        writer.writerow([
            "opcode",
            "variant_simple",
            "variant_function",
            "variant_native",
            "notes"
        ])

        for op in opcodes:
            base = op.split("_")[0]

            writer.writerow([
                op,
                f"{lang}_simple_{op}()",
                f"{base.lower()}_{lang}()",
                generate_native(lang, base),
                "auto-generated mapping"
            ])

    return filename


def generate_native(lang, op):
    if lang == "python":
        return f"# python_{op}"
    if lang == "javascript":
        return f"// js_{op}"
    if lang == "c":
        return f"/* c_{op} */"
    if lang == "rust":
        return f"// rust_{op}"
    if lang == "go":
        return f"// go_{op}"
    if lang == "sql":
        return f"-- sql_{op}"
    return f"{lang}_{op}"


def build_zip():
    opcodes = expand_opcodes(1000)

    files = []

    for lang in LANGUAGES:
        files.append(generate_language_map(lang, opcodes))

    # IR master file
    with open("ir_opcodes.csv", "w", newline="") as f:
        writer = csv.writer(f)
        writer.writerow(["opcode"])
        for op in opcodes:
            writer.writerow([op])

    files.append("ir_opcodes.csv")

    # meta file
    meta = {
        "languages": LANGUAGES,
        "opcode_count": len(opcodes),
        "type": "LLM cross-language IR compiler dataset"
    }

    with open("meta.json", "w") as f:
        json.dump(meta, f, indent=2)

    files.append("meta.json")

    # zip
    with zipfile.ZipFile("opcode_pack.zip", "w") as z:
        for file in files:
            z.write(file)

    print("Built opcode_pack.zip")


if __name__ == "__main__":
    build_zip()