58 lines
2.5 KiB
Python
58 lines
2.5 KiB
Python
|
"""Build a pinned isolated native worker; development VM builds use --backend cpu."""
|
|||
|
|
import argparse
|
|||
|
|
import os
|
|||
|
|
from pathlib import Path
|
|||
|
|
import shutil
|
|||
|
|
import subprocess
|
|||
|
|
import sys
|
|||
|
|
|
|||
|
|
ROOT = Path(__file__).resolve().parents[2]
|
|||
|
|
TAG = "b11461"
|
|||
|
|
COMMIT = "4d756bc72bf00a4aacf410ae15a2d315f3db400d"
|
|||
|
|
|
|||
|
|
|
|||
|
|
def run(arguments, cwd=None):
|
|||
|
|
subprocess.run([str(value) for value in arguments], cwd=cwd, check=True)
|
|||
|
|
|
|||
|
|
|
|||
|
|
def main():
|
|||
|
|
parser = argparse.ArgumentParser(description="Build Searchi's pinned Google embedding worker")
|
|||
|
|
parser.add_argument("--backend", choices=["cpu", "auto", "metal", "vulkan", "cuda"], default="cpu")
|
|||
|
|
parser.add_argument("--jobs", type=int, default=2)
|
|||
|
|
args = parser.parse_args()
|
|||
|
|
if not 1 <= args.jobs <= 16:
|
|||
|
|
parser.error("jobs must be 1–16")
|
|||
|
|
source = ROOT / ".searchi/references/llama.cpp"
|
|||
|
|
build = ROOT / ".searchi/runtime" / args.backend
|
|||
|
|
if not source.exists():
|
|||
|
|
source.parent.mkdir(parents=True, exist_ok=True)
|
|||
|
|
run(["git", "clone", "--depth", "1", "--branch", TAG,
|
|||
|
|
"https://github.com/ggml-org/llama.cpp.git", source])
|
|||
|
|
actual = subprocess.check_output(["git", "rev-parse", "HEAD"], cwd=source, text=True).strip()
|
|||
|
|
if actual != COMMIT:
|
|||
|
|
raise RuntimeError("Runtime reference checkout differs from pinned commit; refusing to reset it")
|
|||
|
|
patch = Path(__file__).with_name("no-logits.patch")
|
|||
|
|
check = subprocess.run(["git", "apply", "--check", str(patch)], cwd=source, capture_output=True)
|
|||
|
|
if check.returncode == 0:
|
|||
|
|
run(["git", "apply", patch], cwd=source)
|
|||
|
|
else:
|
|||
|
|
reverse = subprocess.run(["git", "apply", "--reverse", "--check", str(patch)], cwd=source, capture_output=True)
|
|||
|
|
if reverse.returncode:
|
|||
|
|
raise RuntimeError("Runtime memory patch cannot be applied or verified")
|
|||
|
|
cmake_exe = shutil.which("cmake")
|
|||
|
|
if not cmake_exe:
|
|||
|
|
try:
|
|||
|
|
import cmake
|
|||
|
|
cmake_exe = str(Path(cmake.CMAKE_BIN_DIR) / ("cmake.exe" if os.name == "nt" else "cmake"))
|
|||
|
|
except ImportError as error:
|
|||
|
|
raise RuntimeError("Install CMake first (python -m pip install cmake)") from error
|
|||
|
|
worker_source = ROOT / "crates/searchi-semantic/native"
|
|||
|
|
run([cmake_exe, "-S", worker_source, "-B", build,
|
|||
|
|
"-DCMAKE_BUILD_TYPE=Release", f"-DLLAMA_CPP_DIR={source}",
|
|||
|
|
f"-DSEARCHI_ACCELERATION={args.backend}"])
|
|||
|
|
run([cmake_exe, "--build", build, "--config", "Release", "--parallel", args.jobs])
|
|||
|
|
print(f"Runtime build complete: {build}")
|
|||
|
|
|
|||
|
|
|
|||
|
|
if __name__ == "__main__":
|
|||
|
|
main()
|