Merge branch 'dev' into feature/cpu-torch

kzscisoft · kzscisoft · commit 1af543ef4b3a · 2024-04-04T08:51:08.000+01:00
diff --git a/poetry.lock b/poetry.lock
diff --git a/simvue/metrics.py b/simvue/metrics.py
@@ -1,3 +1,4 @@
+import contextlib
 import logging
 
 from .pynvml import (
@@ -20,10 +21,8 @@ def get_process_memory(processes):
     """
     rss = 0
     for process in processes:
-        try:
+        with contextlib.suppress(Exception):
             rss += process.memory_info().rss / 1024 / 1024
-        except:
-            pass
 
     return rss
 
@@ -34,10 +33,8 @@ def get_process_cpu(processes):
     """
     cpu_percent = 0
     for process in processes:
-        try:
+        with contextlib.suppress(Exception):
             cpu_percent += process.cpu_percent()
-        except:
-            pass
 
     return cpu_percent
 
@@ -64,7 +61,7 @@ def get_gpu_metrics(processes):
     """
     gpu_metrics = {}
 
-    try:
+    with contextlib.suppress(Exception):
         nvmlInit()
         device_count = nvmlDeviceGetCount()
         for i in range(device_count):
@@ -79,7 +76,5 @@ def get_gpu_metrics(processes):
                 gpu_metrics[f"resources/gpu.memory.percent.{i}"] = memory_percent
 
         nvmlShutdown()
-    except:
-        pass
 
     return gpu_metrics
diff --git a/simvue/pynvml.py b/simvue/pynvml.py
@@ -1049,7 +1049,7 @@ def _nvmlGetFunctionPointer(name):
     libLoadLock.acquire()
     try:
         # ensure library was loaded
-        if nvmlLib == None:
+        if nvmlLib is None:
             raise NVMLError(NVML_ERROR_UNINITIALIZED)
         try:
             _nvmlGetFunctionPointer_cache[name] = getattr(nvmlLib, name)
@@ -1310,7 +1310,7 @@ class nvmlClkMonStatus_t(Structure):
 # On Windows with the WDDM driver, usedGpuMemory is reported as None
 # Code that processes this structure should check for None, I.E.
 #
-# if (info.usedGpuMemory == None):
+# if (info.usedGpuMemory is None):
 #     # TODO handle the error
 #     pass
 # else:
@@ -2022,13 +2022,13 @@ def _LoadNvmlLibrary():
     """
     global nvmlLib
 
-    if nvmlLib == None:
+    if nvmlLib is None:
         # lock to ensure only one caller loads the library
         libLoadLock.acquire()
 
         try:
             # ensure the library still isn't loaded
-            if nvmlLib == None:
+            if nvmlLib is None:
                 try:
                     if sys.platform[:3] == "win":
                         # cdecl calling convention
@@ -2040,7 +2040,7 @@ def _LoadNvmlLibrary():
                                     "System32/nvml.dll",
                                 )
                             )
-                        except OSError as ose:
+                        except OSError:
                             # If nvml.dll is not found in System32, it should be in ProgramFiles
                             # load nvml.dll from %ProgramFiles%/NVIDIA Corporation/NVSMI/nvml.dll
                             nvmlLib = CDLL(
@@ -2052,9 +2052,9 @@ def _LoadNvmlLibrary():
                     else:
                         # assume linux
                         nvmlLib = CDLL("libnvidia-ml.so.1")
-                except OSError as ose:
+                except OSError:
                     _nvmlCheckReturn(NVML_ERROR_LIBRARY_NOT_FOUND)
-                if nvmlLib == None:
+                if nvmlLib is None:
                     _nvmlCheckReturn(NVML_ERROR_LIBRARY_NOT_FOUND)
         finally:
             # lock is always freed
diff --git a/simvue/system.py b/simvue/system.py
@@ -2,7 +2,9 @@
 import platform
 import socket
 import subprocess
+import shutil
 import sys
+import contextlib
 import typing
 
 
@@ -13,31 +15,28 @@ def get_cpu_info():
     model_name = ""
     arch = ""
 
-    try:
-        info = subprocess.check_output("lscpu").decode().strip()
-        for line in info.split("\n"):
-            if "Model name" in line:
-                model_name = line.split(":")[1].strip()
-            if "Architecture" in line:
-                arch = line.split(":")[1].strip()
-    except:
-        # TODO: Try /proc/cpuinfo
-        pass
+    if shutil.which("lscpu"):
+        with contextlib.suppress(subprocess.CalledProcessError):
+            info = subprocess.check_output("lscpu").decode().strip()
+            for line in info.split("\n"):
+                if "Model name" in line:
+                    model_name = line.split(":")[1].strip()
+                if "Architecture" in line:
+                    arch = line.split(":")[1].strip()
 
-    if arch == "":
-        arch = platform.machine()
+    # TODO: Try /proc/cpuinfo if process fails
 
-    if model_name == "":
-        try:
+    arch = arch or platform.machine()
+
+    if not model_name and shutil.which("sysctl"):
+        with contextlib.suppress(subprocess.CalledProcessError):
             info = (
                 subprocess.check_output(["sysctl", "machdep.cpu.brand_string"])
                 .decode()
                 .strip()
             )
             if "machdep.cpu.brand_string:" in info:
                 model_name = info.split("machdep.cpu.brand_string: ")[1]
-        except:
-            pass
 
     return model_name, arch
 
@@ -46,16 +45,19 @@ def get_gpu_info():
     """
     Get GPU info
     """
-    try:
-        output = subprocess.check_output(
-            ["nvidia-smi", "--query-gpu=name,driver_version", "--format=csv"]
-        )
-        lines = output.split(b"\n")
-        tokens = lines[1].split(b", ")
-    except:
-        return {"name": "", "driver_version": ""}
+    _gpu_info: dict[str, str] = {"name": "", "driver_version": ""}
+
+    if shutil.which("nvidia-smi"):
+        with contextlib.suppress(subprocess.CalledProcessError, IndexError):
+            output = subprocess.check_output(
+                ["nvidia-smi", "--query-gpu=name,driver_version", "--format=csv"]
+            )
+            lines = output.split(b"\n")
+            tokens = lines[1].split(b", ")
+            _gpu_info["name"] = tokens[0].decode()
+            _gpu_info["driver_version"] = tokens[1].decode()
 
-    return {"name": tokens[0].decode(), "driver_version": tokens[1].decode()}
+    return _gpu_info
 
 
 def get_system() -> dict[str, typing.Any]:
diff --git a/simvue/utilities.py b/simvue/utilities.py
@@ -2,43 +2,45 @@
 import datetime
 import hashlib
 import logging
+import importlib.util
+import contextlib
 import os
 import typing
 
 import jwt
 
 CHECKSUM_BLOCK_SIZE = 4096
+EXTRAS: tuple[str, ...] = ("plot", "torch", "dataset")
 
 logger = logging.getLogger(__name__)
 
 
 def check_extra(extra_name: str) -> typing.Callable:
     def decorator(class_func: typing.Callable) -> typing.Callable:
         def wrapper(self, *args, **kwargs) -> typing.Any:
-            if extra_name == "plot":
-                try:
-                    import matplotlib
-                    import plotly
-                except ImportError:
-                    raise RuntimeError(
-                        f"Plotting features require the '{extra_name}' extension to Simvue"
-                    )
-            elif extra_name == "torch":
-                try:
-                    import torch
-                except ImportError:
-                    raise RuntimeError(
-                        "PyTorch features require the 'torch' module to be installed"
-                    )
-            elif extra_name == "dataset":
-                try:
-                    import numpy
-                    import pandas
-                except ImportError:
-                    raise RuntimeError(
-                        f"Dataset features require the '{extra_name}' extension to Simvue"
-                    )
-            else:
+            if extra_name == "plot" and not all(
+                [
+                    importlib.util.find_spec("matplotlib"),
+                    importlib.util.find_spec("plotly"),
+                ]
+            ):
+                raise RuntimeError(
+                    f"Plotting features require the '{extra_name}' extension to Simvue"
+                )
+            elif extra_name == "torch" and not importlib.util.find_spec("torch"):
+                raise RuntimeError(
+                    "PyTorch features require the 'torch' module to be installed"
+                )
+            elif extra_name == "dataset" and not all(
+                [
+                    importlib.util.find_spec("numpy"),
+                    importlib.util.find_spec("pandas"),
+                ]
+            ):
+                raise RuntimeError(
+                    f"Dataset features require the '{extra_name}' extension to Simvue"
+                )
+            elif extra_name not in EXTRAS:
                 raise RuntimeError(f"Unrecognised extra '{extra_name}'")
             return class_func(self, *args, **kwargs)
 
@@ -103,13 +105,11 @@ def get_auth():
         os.path.join(os.path.expanduser("~"), ".simvue.ini"),
         "simvue.ini",
     ):
-        try:
+        with contextlib.suppress(Exception):
             config = configparser.ConfigParser()
             config.read(filename)
             token = config.get("server", "token")
             url = config.get("server", "url")
-        except:
-            pass
 
     # Try environment variables
     token = os.getenv("SIMVUE_TOKEN", token)
@@ -128,12 +128,10 @@ def get_offline_directory():
         os.path.join(os.path.expanduser("~"), ".simvue.ini"),
         "simvue.ini",
     ):
-        try:
+        with contextlib.suppress(Exception):
             config = configparser.ConfigParser()
             config.read(filename)
             directory = config.get("offline", "cache")
-        except:
-            pass
 
     if not directory:
         directory = os.path.join(os.path.expanduser("~"), ".simvue")
@@ -168,10 +166,9 @@ def get_expiry(token):
     Get expiry date from a JWT token
     """
     expiry = 0
-    try:
+    with contextlib.suppress(jwt.DecodeError):
         expiry = jwt.decode(token, options={"verify_signature": False})["exp"]
-    except:
-        pass
+
     return expiry
 
 
diff --git a/tests/refactor/test_run_class.py b/tests/refactor/test_run_class.py
