6 from inspect import stack
9 # This module is commonly used by others in here and should avoid
10 # taking any unnecessary dependencies back on them.
12 from argparse_utils import ActionNoYes
16 logger = logging.getLogger(__name__)
18 args = config.add_commandline_args(
19 f'Bootstrap ({__file__})',
20 'Args related to python program bootstrapper and Swiss army knife')
22 '--debug_unhandled_exceptions',
25 help='Break into pdb on top level unhandled exceptions.'
31 help='Should we display (and log.debug) the global random seed?'
39 help='Override the global random seed with a particular number.'
45 help='Should we dump the Python import tree before main?'
48 '--audit_import_events',
51 help='Should we audit all import events?',
55 original_hook = sys.excepthook
58 def handle_uncaught_exception(exc_type, exc_value, exc_tb):
60 Top-level exception handler for exceptions that make it past any exception
61 handlers in the python code being run. Logs the error and stacktrace then
62 maybe attaches a debugger.
66 msg = f'Unhandled top level exception {exc_type}'
68 print(msg, file=sys.stderr)
69 if issubclass(exc_type, KeyboardInterrupt):
70 sys.__excepthook__(exc_type, exc_value, exc_tb)
74 not sys.stderr.isatty() or
75 not sys.stdin.isatty()
77 # stdin or stderr is redirected, just do the normal thing
78 original_hook(exc_type, exc_value, exc_tb)
80 # a terminal is attached and stderr is not redirected, maybe debug.
82 traceback.print_exception(exc_type, exc_value, exc_tb)
83 if config.config['debug_unhandled_exceptions']:
85 logger.info("Invoking the debugger...")
88 original_hook(exc_type, exc_value, exc_tb)
91 class ImportInterceptor(object):
94 self.module_by_filename_cache = {}
95 self.repopulate_modules_by_filename()
96 self.tree = collect.trie.Trie()
97 self.tree_node_by_module = {}
99 def repopulate_modules_by_filename(self):
100 self.module_by_filename_cache.clear()
101 for mod in sys.modules:
102 if hasattr(sys.modules[mod], '__file__'):
103 fname = getattr(sys.modules[mod], '__file__')
106 self.module_by_filename_cache[fname] = mod
108 def should_ignore_filename(self, filename: str) -> bool:
109 return 'importlib' in filename or 'six.py' in filename
111 def find_spec(self, loaded_module, path=None, target=None):
113 for x in range(3, len(s)):
114 filename = s[x].filename
115 if self.should_ignore_filename(filename):
118 loading_function = s[x].function
119 if filename in self.module_by_filename_cache:
120 loading_module = self.module_by_filename_cache[filename]
122 self.repopulate_modules_by_filename()
123 loading_module = self.module_by_filename_cache.get(filename, 'unknown')
125 path = self.tree_node_by_module.get(loading_module, [])
126 path.extend([loaded_module])
127 self.tree.insert(path)
128 self.tree_node_by_module[loading_module] = path
130 msg = f'*** Import {loaded_module} from {filename}:{s[x].lineno} in {loading_module}::{loading_function}'
134 msg = f'*** Import {loaded_module} from ?????'
138 def find_importer(self, module: str):
139 if module in self.tree_node_by_module:
140 node = self.tree_node_by_module[module]
145 # # TODO: test this with python 3.8+
146 # def audit_import_events(event, args):
147 # if event == 'import':
151 # sys_meta_path = args[3]
152 # sys_path_hooks = args[4]
157 # Audit import events? Note: this runs early in the lifetime of the
158 # process (assuming that import bootstrap happens early); config has
159 # (probably) not yet been loaded or parsed the commandline. Also,
160 # some things have probably already been imported while we weren't
161 # watching so this information may be incomplete.
163 # Also note: move bootstrap up in the global import list to catch
164 # more import events and have a more complete record.
165 import_interceptor = None
167 if arg == '--audit_import_events':
168 import_interceptor = ImportInterceptor()
169 sys.meta_path = [import_interceptor] + sys.meta_path
170 # if not hasattr(sys, 'frozen'):
172 # sys.version_info[0] == 3
173 # and sys.version_info[1] >= 8
175 # sys.addaudithook(audit_import_events)
178 def dump_all_objects() -> None:
179 global import_interceptor
181 all_modules = sys.modules
182 for obj in object.__subclasses__():
183 if not hasattr(obj, '__name__'):
186 if not hasattr(obj, '__module__'):
188 class_mod_name = obj.__module__
189 if class_mod_name in all_modules:
190 mod = all_modules[class_mod_name]
191 if not hasattr(mod, '__name__'):
192 mod_name = class_mod_name
194 mod_name = mod.__name__
195 if hasattr(mod, '__file__'):
196 mod_file = mod.__file__
199 if import_interceptor is not None:
200 import_path = import_interceptor.find_importer(mod_name)
202 import_path = 'unknown'
203 msg = f'{class_mod_name}::{klass} ({mod_file})'
204 if import_path != 'unknown' and len(import_path) > 0:
205 msg += f' imported by {import_path}'
206 messages[f'{class_mod_name}::{klass}'] = msg
207 for x in sorted(messages.keys()):
208 logger.debug(messages[x])
212 def initialize(entry_point):
214 Remember to initialize config, initialize logging, set/log a random
215 seed, etc... before running main.
218 @functools.wraps(entry_point)
219 def initialize_wrapper(*args, **kwargs):
220 # Hook top level unhandled exceptions, maybe invoke debugger.
221 if sys.excepthook == sys.__excepthook__:
222 sys.excepthook = handle_uncaught_exception
224 # Try to figure out the name of the program entry point. Then
225 # parse configuration (based on cmdline flags, environment vars
228 '__globals__' in entry_point.__dict__ and
229 '__file__' in entry_point.__globals__
231 config.parse(entry_point.__globals__['__file__'])
235 # Initialize logging... and log some remembered messages from
237 logging_utils.initialize_logging(logging.getLogger())
238 config.late_logging()
240 # Allow programs that don't bother to override the random seed
241 # to be replayed via the commandline.
243 random_seed = config.config['set_random_seed']
244 if random_seed is not None:
245 random_seed = random_seed[0]
247 random_seed = int.from_bytes(os.urandom(4), 'little')
249 if config.config['show_random_seed']:
250 msg = f'Global random seed is: {random_seed}'
253 random.seed(random_seed)
255 # Do it, invoke the user's code. Pay attention to how long it takes.
256 logger.debug(f'Starting {entry_point.__name__} (program entry point)')
259 with stopwatch.Timer() as t:
260 ret = entry_point(*args, **kwargs)
262 f'{entry_point.__name__} (program entry point) returned {ret}.'
265 if config.config['dump_all_objects']:
268 if config.config['audit_import_events']:
269 global import_interceptor
270 if import_interceptor is not None:
271 print(import_interceptor.tree)
274 (utime, stime, cutime, cstime, elapsed_time) = os.times()
277 f'system: {stime}s\n'
278 f'child user: {cutime}s\n'
279 f'child system: {cstime}s\n'
280 f'machine uptime: {elapsed_time}s\n'
281 f'walltime: {walltime}s')
283 # If it doesn't return cleanly, call attention to the return value.
284 if ret is not None and ret != 0:
285 logger.error(f'Exit {ret}')
287 logger.debug(f'Exit {ret}')
289 return initialize_wrapper