python: fix CalledProcessError on Intel Macs since v2.8.0 (#3045)

Signed-off-by: Jared Van Bortel <jared@nomic.ai>

gpt4all-bindings/python/CHANGELOG.md

@@ -12,6 +12,7 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/).
 ### Changed
 - Rebase llama.cpp on latest upstream as of September 26th ([#2998](https://github.com/nomic-ai/gpt4all/pull/2998))
 - Change the error message when a message is too long ([#3004](https://github.com/nomic-ai/gpt4all/pull/3004))
+- Fix CalledProcessError on Intel Macs since v2.8.0 ([#3045](https://github.com/nomic-ai/gpt4all/pull/3045))
 
 ## [2.8.2] - 2024-08-14
 

gpt4all-bindings/python/gpt4all/_pyllmodel.py

@@ -3,7 +3,6 @@ from __future__ import annotations
 import ctypes
 import os
 import platform
-import re
 import subprocess
 import sys
 import textwrap
@@ -28,17 +27,26 @@ if TYPE_CHECKING:
 
 EmbeddingsType = TypeVar('EmbeddingsType', bound='list[Any]')
 
+cuda_found: bool = False
+
+
+# TODO(jared): use operator.call after we drop python 3.10 support
+def _operator_call(obj, /, *args, **kwargs):
+    return obj(*args, **kwargs)
+
 
 # Detect Rosetta 2
-if platform.system() == "Darwin" and platform.processor() == "i386":
+@_operator_call
-    if subprocess.run(
+def check_rosetta() -> None:
-        "sysctl -n sysctl.proc_translated".split(), check=True, capture_output=True, text=True,
+    if platform.system() == "Darwin" and platform.processor() == "i386":
-    ).stdout.strip() == "1":
+        p = subprocess.run("sysctl -n sysctl.proc_translated".split(), capture_output=True, text=True)
+        if p.returncode == 0 and p.stdout.strip() == "1":
             raise RuntimeError(textwrap.dedent("""\
                 Running GPT4All under Rosetta is not supported due to CPU feature requirements.
                 Please install GPT4All in an environment that uses a native ARM64 Python interpreter.
             """).strip())
 
+
 # Check for C++ runtime libraries
 if platform.system() == "Windows":
     try:
@@ -53,7 +61,11 @@ if platform.system() == "Windows":
         """), file=sys.stderr)
 
 
-def _load_cuda(rtver: str, blasver: str) -> None:
+@_operator_call
+def find_cuda() -> None:
+    global cuda_found
+
+    def _load_cuda(rtver: str, blasver: str) -> None:
         if platform.system() == "Linux":
             cudalib   = f"lib/libcudart.so.{rtver}"
             cublaslib = f"lib/libcublas.so.{blasver}"
@@ -65,10 +77,8 @@ def _load_cuda(rtver: str, blasver: str) -> None:
         ctypes.CDLL(os.path.join(cuda_runtime.__path__[0], cudalib), mode=ctypes.RTLD_GLOBAL)
         ctypes.CDLL(os.path.join(cublas.__path__[0], cublaslib), mode=ctypes.RTLD_GLOBAL)
 
-
+    # Find CUDA libraries from the official packages
-# Find CUDA libraries from the official packages
+    if platform.system() in ("Linux", "Windows"):
-cuda_found = False
-if platform.system() in ("Linux", "Windows"):
         try:
             from nvidia import cuda_runtime, cublas
         except ImportError:
@@ -121,6 +131,7 @@ class LLModelPromptContext(ctypes.Structure):
         ("context_erase", ctypes.c_float),
     ]
 
+
 class LLModelGPUDevice(ctypes.Structure):
     _fields_ = [
         ("backend", ctypes.c_char_p),
@@ -131,6 +142,7 @@ class LLModelGPUDevice(ctypes.Structure):
         ("vendor", ctypes.c_char_p),
     ]
 
+
 # Define C function signatures using ctypes
 llmodel.llmodel_model_create.argtypes = [ctypes.c_char_p]
 llmodel.llmodel_model_create.restype = ctypes.c_void_p
@@ -540,7 +552,6 @@ class LLModel:
             ctypes.c_char_p(),
         )
 
-
     def prompt_model_streaming(
         self, prompt: str, prompt_template: str, callback: ResponseCallbackType = empty_response_callback, **kwargs
     ) -> Iterable[str]:

gpt4all-bindings/python/gpt4all/gpt4all.py

@@ -8,7 +8,6 @@ import os
 import platform
 import re
 import sys
-import time
 import warnings
 from contextlib import contextmanager
 from pathlib import Path