Since this thing is on the innerwebs I suppose it should have a

[python_utils.git] / bootstrap.py

diff --git a/bootstrap.py b/bootstrap.py
index 35747865b7aacc38fa61f6005af6c5bd5833ff69..117468623b94e578894561d0ff6fede8a9a33135 100644 (file)
--- a/bootstrap.py
+++ b/bootstrap.py
@@ -1,5 +1,15 @@
 #!/usr/bin/env python3
 
+# © Copyright 2021-2022, Scott Gasch
+
+"""This is a module for wrapping around python programs and doing some
+minor setup and tear down work for them.  With it, you can break into
+pdb on unhandled top level exceptions, profile your code by passing a
+commandline argument in, audit module import events, examine where
+memory is being used in your program, and so on.
+
+"""
+
 import functools
 import importlib
 import logging
@@ -101,6 +111,15 @@ def handle_uncaught_exception(exc_type, exc_value, exc_tb):
 
 
 class ImportInterceptor(importlib.abc.MetaPathFinder):
+    """An interceptor that always allows module load events but dumps a
+    record into the log and onto stdout when modules are loaded and
+    produces an audit of who imported what at the end of the run.  It
+    can't see any load events that happen before it, though, so move
+    bootstrap up in your __main__'s import list just temporarily to
+    get a good view.
+
+    """
+
     def __init__(self):
         import collect.trie
 
@@ -111,9 +130,9 @@ class ImportInterceptor(importlib.abc.MetaPathFinder):
 
     def repopulate_modules_by_filename(self):
         self.module_by_filename_cache.clear()
-        for mod in sys.modules:
-            if hasattr(sys.modules[mod], '__file__'):
-                fname = getattr(sys.modules[mod], '__file__')
+        for _, mod in sys.modules.copy().items():  # copy here because modules is volatile
+            if hasattr(mod, '__file__'):
+                fname = getattr(mod, '__file__')
             else:
                 fname = 'unknown'
             self.module_by_filename_cache[fname] = mod
@@ -226,10 +245,16 @@ def initialize(entry_point):
         # Try to figure out the name of the program entry point.  Then
         # parse configuration (based on cmdline flags, environment vars
         # etc...)
-        if '__globals__' in entry_point.__dict__ and '__file__' in entry_point.__globals__:
-            config.parse(entry_point.__globals__['__file__'])
-        else:
-            config.parse(None)
+        entry_filename = None
+        entry_descr = None
+        try:
+            entry_filename = entry_point.__code__.co_filename
+            entry_descr = entry_point.__code__.__repr__()
+        except Exception:
+            if '__globals__' in entry_point.__dict__ and '__file__' in entry_point.__globals__:
+                entry_filename = entry_point.__globals__['__file__']
+                entry_descr = entry_filename
+        config.parse(entry_filename)
 
         if config.config['trace_memory']:
             import tracemalloc
@@ -243,12 +268,15 @@ def initialize(entry_point):
 
         # Maybe log some info about the python interpreter itself.
         logger.debug(
-            'Platform: %s, maxint=0x%x, byteorder=%s',
-            sys.platform, sys.maxsize, sys.byteorder
+            'Platform: %s, maxint=0x%x, byteorder=%s', sys.platform, sys.maxsize, sys.byteorder
         )
         logger.debug('Python interpreter version: %s', sys.version)
         logger.debug('Python implementation: %s', sys.implementation)
         logger.debug('Python C API version: %s', sys.api_version)
+        if __debug__:
+            logger.debug('Python interpreter running in __debug__ mode.')
+        else:
+            logger.debug('Python interpreter running in optimized mode.')
         logger.debug('Python path: %s', sys.path)
 
         # Log something about the site_config, many things use it.
@@ -273,7 +301,7 @@ def initialize(entry_point):
         random.seed(random_seed)
 
         # Do it, invoke the user's code.  Pay attention to how long it takes.
-        logger.debug('Starting %s (program entry point)', entry_point.__name__)
+        logger.debug('Starting %s (program entry point)', entry_descr)
         ret = None
         import stopwatch
 
@@ -293,7 +321,7 @@ def initialize(entry_point):
             with stopwatch.Timer() as t:
                 ret = entry_point(*args, **kwargs)
 
-        logger.debug('%s (program entry point) returned %s.', entry_point.__name__, ret)
+        logger.debug('%s (program entry point) returned %s.', entry_descr, ret)
 
         if config.config['trace_memory']:
             snapshot = tracemalloc.take_snapshot()
@@ -320,7 +348,12 @@ def initialize(entry_point):
             'child system: %.4fs\n'
             'machine uptime: %.4fs\n'
             'walltime: %.4fs',
-            utime, stime, cutime, cstime, elapsed_time, walltime
+            utime,
+            stime,
+            cutime,
+            cstime,
+            elapsed_time,
+            walltime,
         )
 
         # If it doesn't return cleanly, call attention to the return value.