|
| 1 | +#!/usr/bin/env python |
| 2 | +"""Build script for mypyc compilation of graphql modules. |
| 3 | +
|
| 4 | +Usage: |
| 5 | + python build_mypyc.py # Build all configured modules |
| 6 | + python build_mypyc.py --clean # Remove compiled extensions |
| 7 | + python build_mypyc.py --bench # Run benchmark comparison |
| 8 | +
|
| 9 | +Benchmarks show ~18x speedup for recursive functions and ~16x for loops. |
| 10 | +""" |
| 11 | + |
| 12 | +from __future__ import annotations |
| 13 | + |
| 14 | +import argparse |
| 15 | +import os |
| 16 | +import shutil |
| 17 | +import sys |
| 18 | +from pathlib import Path |
| 19 | + |
| 20 | +# Modules to compile with mypyc. |
| 21 | +# Start small - add more as we verify they work correctly. |
| 22 | +MYPYC_MODULES = [ |
| 23 | + # Sentinel module for testing compilation |
| 24 | + "graphql_mypyc/_sentinel.py", |
| 25 | + # Phase 1: Type system (hot path during execution) |
| 26 | + # NOTE: definition.py has conditional class def - not mypyc compatible |
| 27 | + # "graphql/type/definition.py", |
| 28 | + "graphql/type/scalars.py", |
| 29 | + # Phase 2: Language (parsing hot path) |
| 30 | + "graphql/language/lexer.py", |
| 31 | + "graphql/language/parser.py", |
| 32 | + # Phase 3: Execution |
| 33 | + # NOTE: execute.py has complex async code that triggers mypyc code-gen bugs |
| 34 | + # "graphql/execution/execute.py", |
| 35 | +] |
| 36 | + |
| 37 | + |
| 38 | +def clean() -> None: |
| 39 | + """Remove all compiled .so files from the source tree.""" |
| 40 | + src = Path("src") |
| 41 | + if not src.exists(): |
| 42 | + src = Path(".") |
| 43 | + |
| 44 | + count = 0 |
| 45 | + for so_file in src.rglob("*.so"): |
| 46 | + print(f"Removing {so_file}") |
| 47 | + so_file.unlink() |
| 48 | + count += 1 |
| 49 | + |
| 50 | + # Also clean build artifacts |
| 51 | + for dirname in ["build", ".mypy_cache"]: |
| 52 | + dirpath = Path(dirname) |
| 53 | + if dirpath.exists(): |
| 54 | + print(f"Removing {dirpath}/") |
| 55 | + shutil.rmtree(dirpath) |
| 56 | + |
| 57 | + print(f"Cleaned {count} .so files") |
| 58 | + |
| 59 | + |
| 60 | +def build() -> int: |
| 61 | + """Compile modules with mypyc.""" |
| 62 | + if not MYPYC_MODULES: |
| 63 | + print("No modules configured for compilation") |
| 64 | + return 0 |
| 65 | + |
| 66 | + # Change to src directory for proper module resolution |
| 67 | + original_dir = os.getcwd() |
| 68 | + src_dir = Path("src") |
| 69 | + if src_dir.exists(): |
| 70 | + os.chdir(src_dir) |
| 71 | + |
| 72 | + try: |
| 73 | + # Verify all modules exist |
| 74 | + for module in MYPYC_MODULES: |
| 75 | + if not Path(module).exists(): |
| 76 | + print(f"Error: {module} not found") |
| 77 | + return 1 |
| 78 | + |
| 79 | + print(f"Compiling {len(MYPYC_MODULES)} modules with mypyc...") |
| 80 | + print("Modules:", MYPYC_MODULES) |
| 81 | + |
| 82 | + try: |
| 83 | + from mypyc.build import mypycify |
| 84 | + except ImportError: |
| 85 | + print("Error: mypyc not installed. Install with: pip install mypy") |
| 86 | + return 1 |
| 87 | + |
| 88 | + # Use mypycify to get extension modules |
| 89 | + try: |
| 90 | + ext_modules = mypycify( |
| 91 | + MYPYC_MODULES, |
| 92 | + opt_level="3", |
| 93 | + debug_level="0", |
| 94 | + ) |
| 95 | + except Exception as e: |
| 96 | + print(f"mypycify failed: {e}") |
| 97 | + return 1 |
| 98 | + |
| 99 | + if not ext_modules: |
| 100 | + print("No extension modules generated") |
| 101 | + return 1 |
| 102 | + |
| 103 | + # Build the extensions in-place |
| 104 | + from setuptools import Distribution |
| 105 | + from setuptools.command.build_ext import build_ext |
| 106 | + |
| 107 | + dist = Distribution({"ext_modules": ext_modules}) |
| 108 | + cmd = build_ext(dist) |
| 109 | + cmd.inplace = True |
| 110 | + cmd.ensure_finalized() |
| 111 | + |
| 112 | + try: |
| 113 | + cmd.run() |
| 114 | + except Exception as e: |
| 115 | + print(f"Build failed: {e}") |
| 116 | + return 1 |
| 117 | + |
| 118 | + print("\nBuild successful!") |
| 119 | + # Show what was built |
| 120 | + for module in MYPYC_MODULES: |
| 121 | + path = Path(module) |
| 122 | + for so_file in path.parent.glob(f"{path.stem}*.so"): |
| 123 | + print(f" Built: {so_file}") |
| 124 | + |
| 125 | + return 0 |
| 126 | + |
| 127 | + finally: |
| 128 | + os.chdir(original_dir) |
| 129 | + |
| 130 | + |
| 131 | +def benchmark() -> int: |
| 132 | + """Run benchmark comparing interpreted vs compiled.""" |
| 133 | + import time |
| 134 | + import importlib.util |
| 135 | + |
| 136 | + print("=== Compiled (graphql_mypyc._sentinel) ===") |
| 137 | + |
| 138 | + from graphql_mypyc import _sentinel as compiled |
| 139 | + |
| 140 | + print(f"Module: {compiled.__file__}") |
| 141 | + print(f"is_compiled: {compiled.is_compiled()}") |
| 142 | + |
| 143 | + # Test fibonacci |
| 144 | + n = 30 |
| 145 | + start = time.perf_counter() |
| 146 | + result = compiled.fibonacci(n) |
| 147 | + elapsed_compiled_fib = time.perf_counter() - start |
| 148 | + print(f"fibonacci({n}) = {result} in {elapsed_compiled_fib:.4f}s") |
| 149 | + |
| 150 | + # Test sum_squares |
| 151 | + n = 1_000_000 |
| 152 | + start = time.perf_counter() |
| 153 | + result = compiled.sum_squares(n) |
| 154 | + elapsed_compiled_sq = time.perf_counter() - start |
| 155 | + print(f"sum_squares({n}) = {result} in {elapsed_compiled_sq:.6f}s") |
| 156 | + |
| 157 | + print() |
| 158 | + print("=== Interpreted (graphql._sentinel) ===") |
| 159 | + |
| 160 | + # Import the interpreted version directly by path |
| 161 | + spec = importlib.util.spec_from_file_location( |
| 162 | + "_sentinel_interp", "src/graphql/_sentinel.py" |
| 163 | + ) |
| 164 | + interp = importlib.util.module_from_spec(spec) |
| 165 | + spec.loader.exec_module(interp) # type: ignore[union-attr] |
| 166 | + print(f"Module: {interp.__file__}") |
| 167 | + print(f"is_compiled: {interp.is_compiled()}") |
| 168 | + |
| 169 | + # Test fibonacci |
| 170 | + n = 30 |
| 171 | + start = time.perf_counter() |
| 172 | + result = interp.fibonacci(n) |
| 173 | + elapsed_interp_fib = time.perf_counter() - start |
| 174 | + print(f"fibonacci({n}) = {result} in {elapsed_interp_fib:.4f}s") |
| 175 | + |
| 176 | + # Test sum_squares |
| 177 | + n = 1_000_000 |
| 178 | + start = time.perf_counter() |
| 179 | + result = interp.sum_squares(n) |
| 180 | + elapsed_interp_sq = time.perf_counter() - start |
| 181 | + print(f"sum_squares({n}) = {result} in {elapsed_interp_sq:.6f}s") |
| 182 | + |
| 183 | + print() |
| 184 | + print("=== Speedup ===") |
| 185 | + print(f"fibonacci: {elapsed_interp_fib/elapsed_compiled_fib:.1f}x faster") |
| 186 | + print(f"sum_squares: {elapsed_interp_sq/elapsed_compiled_sq:.1f}x faster") |
| 187 | + |
| 188 | + return 0 |
| 189 | + |
| 190 | + |
| 191 | +def main() -> int: |
| 192 | + """CLI entry point.""" |
| 193 | + parser = argparse.ArgumentParser(description="Build graphql modules with mypyc") |
| 194 | + parser.add_argument( |
| 195 | + "--clean", action="store_true", help="Remove compiled extensions" |
| 196 | + ) |
| 197 | + parser.add_argument( |
| 198 | + "--bench", action="store_true", help="Run benchmark comparison" |
| 199 | + ) |
| 200 | + args = parser.parse_args() |
| 201 | + |
| 202 | + if args.clean: |
| 203 | + clean() |
| 204 | + return 0 |
| 205 | + |
| 206 | + if args.bench: |
| 207 | + return benchmark() |
| 208 | + |
| 209 | + return build() |
| 210 | + |
| 211 | + |
| 212 | +if __name__ == "__main__": |
| 213 | + sys.exit(main()) |
0 commit comments