ocrd_utils.logging.initLogging() - Code Metrics - Inspection of "Workflow endpoint of the Processing Server" - OCR-D/core - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Pull Request — master (#1083)

by Konstantin

created 2023-10-11 11:25 UTC

ocrd_utils.logging.initLogging() C

↳ Parent: ocrd_utils.logging

Complexity

Conditions

Size

Total Lines	58
Code Lines	29

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
eloc	29
dl	0
loc	58
rs	6.6666
c	0
b	0
f	0
cc	9
nop	2

How to fix Long Method

"""
Logging setup

By default: Log with lastResort logger, usually STDERR.

Logging can be overridden either programmatically in code using the library or by creating one or more of

- /etc/ocrd_logging.py
- $HOME/ocrd_logging.py
- $PWD/ocrd_logging.py

These files will be executed in the context of ocrd/ocrd_logging.py, with `logging` global set.

Changes as of 2023-08-20:

    - Try to be less intrusive with OCR-D specific logging conventions to
      make it easier and less surprising to define logging behavior when
      using OCR-D/core as a library
    - Change setOverrideLogLevel to only override the log level of the ``ocrd``
      logger and its descendants
    - initLogging will set exactly one handler, for the root logger or for the
      ``ocrd`` logger.
    - Child loggers should propagate to the ancestor logging (default
      behavior of the logging library - no more PropagationShyLogger)
    - disableLogging only removes any handlers from the ``ocrd`` logger
"""
# pylint: disable=no-member

from __future__ import absolute_import

from traceback import format_stack

import logging
import logging.config
from pathlib import Path
import sys

from .constants import LOG_FORMAT, LOG_TIMEFMT

__all__ = [
    'disableLogging',
    'getLevelName',
    'getLogger',
    'initLogging',
    'logging',
    'setOverrideLogLevel',
]

# These are the loggers we add handlers to
ROOT_OCRD_LOGGERS = [
    'ocrd',
    'ocrd_network'
]

LOGGING_DEFAULTS = {
    'ocrd': logging.INFO,
    'ocrd_network': logging.DEBUG,
    # 'ocrd.resolver': logging.INFO,
    # 'ocrd.resolver.download_to_directory': logging.INFO,
    # 'ocrd.resolver.add_files_to_mets': logging.INFO,
    # To cut back on the `Self-intersection at or near point` INFO messages
    'shapely.geos': logging.ERROR,
    'tensorflow': logging.ERROR,
    'PIL': logging.INFO,
    'paramiko.transport': logging.INFO,
    'uvicorn.access': logging.DEBUG,
    'uvicorn.error': logging.DEBUG,
    'uvicorn': logging.INFO
}

_initialized_flag = False

_ocrdLevel2pythonLevel = {
    'TRACE': 'DEBUG',
    'OFF': 'CRITICAL',
    'FATAL': 'ERROR',
}

def tf_disable_interactive_logs():
    try:
        from os import environ
        # This env variable must be set before importing from Keras
        environ['TF_CPP_MIN_LOG_LEVEL'] = '3'
        from tensorflow.keras.utils import disable_interactive_logging
        # Enabled interactive logging throws an exception
        # due to a call of sys.stdout.flush()
        disable_interactive_logging()
    except ImportError:
        # Nothing should be handled here if TF is not available
        pass

def getLevelName(lvl):
    """
    Get (string) python logging level for (string) spec-defined log level name.
    """
    lvl = _ocrdLevel2pythonLevel.get(lvl, lvl)
    return logging.getLevelName(lvl)

def getLogger(*args, **kwargs):
    """
    Wrapper around ``logging.getLogger`` that alls :py:func:`initLogging` if
    that wasn't explicitly called before.
    """
    logger = logging.getLogger(*args, **kwargs)
    return logger

def setOverrideLogLevel(lvl, silent=True):
    """
    Override the output log level of the handlers attached to the ``ocrd`` logger.

    Args:
        lvl (string): Log level name.
        silent (boolean): Whether to log the override call
    """
    if not _initialized_flag:
        initLogging()
    ocrd_logger = logging.getLogger('ocrd')

    if lvl is None:
        if not silent:
            ocrd_logger.info('Reset log level override')
        ocrd_logger.setLevel(logging.NOTSET)
    else:
        if not silent:
            ocrd_logger.info('Overriding ocrd log level to %s', lvl)
        ocrd_logger.setLevel(lvl)

def initLogging(builtin_only=False, force_reinit=False):
    """
    Reset ``ocrd`` logger, read logging configuration if exists, otherwise use basicConfig

    initLogging is to be called by OCR-D/core once, i.e.
        -  for the ``ocrd`` CLI
        -  for the processor wrapper methods

    Other processes that use OCR-D/core as a library can, but do not have to, use this functionality.

    Keyword Args:
        - builtin_only (bool, False): Whether to search for logging configuration
                                      on-disk (``False``) or only use the
                                      hard-coded config (``True``). For testing
        - force_reinit (bool, False): Whether to ignore the module-level
                                      ``_initialized_flag``. For testing only.
    """
    global _initialized_flag
    if _initialized_flag and not force_reinit:
        return
    # disableLogging()

    # https://docs.python.org/3/library/logging.html#logging.disable
    # If logging.disable(logging.NOTSET) is called, it effectively removes this
    # overriding level, so that logging output again depends on the effective
    # levels of individual loggers.
    logging.disable(logging.NOTSET)

    # remove all handlers for the ocrd root loggers
    for logger_name in ROOT_OCRD_LOGGERS:
        for handler in logging.getLogger(logger_name).handlers[:]:
            logging.getLogger(logger_name).removeHandler(handler)

    config_file = None
    if not builtin_only:
        CONFIG_PATHS = [
            Path.cwd(),
            Path.home(),
            Path('/etc'),
        ]
        config_file = next((f for f \

                in [p / 'ocrd_logging.conf' for p in CONFIG_PATHS] \
                if f.exists()),
                None)
    if config_file:
        logging.config.fileConfig(config_file)
        logging.getLogger('ocrd.logging').debug("Picked up logging config at %s", config_file)
    else:
        # Default logging config
        ocrd_handler = logging.StreamHandler(stream=sys.stderr)
        ocrd_handler.setFormatter(logging.Formatter(fmt=LOG_FORMAT, datefmt=LOG_TIMEFMT))
        ocrd_handler.setLevel(logging.DEBUG)
        for logger_name in ROOT_OCRD_LOGGERS:
            logging.getLogger(logger_name).addHandler(ocrd_handler)
        for logger_name, logger_level in LOGGING_DEFAULTS.items():
            logging.getLogger(logger_name).setLevel(logger_level)

    _initialized_flag = True

def disableLogging(silent=True):
    """
    Disables all logging of the ``ocrd`` logger and descendants

    Keyword Args:
        silent (bool, True): Whether to log the call to disableLogging
    """
    global _initialized_flag # pylint: disable=global-statement
    if _initialized_flag and not silent:
        logging.getLogger('ocrd.logging').debug("Disabling logging")
    _initialized_flag = False
    # logging.basicConfig(level=logging.CRITICAL)
    # logging.disable(logging.ERROR)
    # remove all handlers for the ocrd logger
    for logger_name in ROOT_OCRD_LOGGERS:
        for handler in logging.getLogger(logger_name).handlers[:]:
            logging.getLogger(logger_name).removeHandler(handler)
    for logger_name in LOGGING_DEFAULTS:
        logging.getLogger(logger_name).setLevel(logging.NOTSET)

# Initializing stream handlers at module level
# would cause message output in all runtime contexts,
# including those which are already run for std output
# (--dump-json, --version, ocrd-tool, bashlib etc).
# So this needs to be an opt-in from the CLIs/decorators:
#initLogging()
# Also, we even have to block log output for libraries
# (like matplotlib/tensorflow) which set up logging
# themselves already:
disableLogging()


1			"""
2			Logging setup
3
4			By default: Log with lastResort logger, usually STDERR.
5
6			Logging can be overridden either programmatically in code using the library or by creating one or more of
7
8			- /etc/ocrd_logging.py
9			- $HOME/ocrd_logging.py
10			- $PWD/ocrd_logging.py
11
12			These files will be executed in the context of ocrd/ocrd_logging.py, with `logging` global set.
13
14			Changes as of 2023-08-20:
15
16			- Try to be less intrusive with OCR-D specific logging conventions to
17			make it easier and less surprising to define logging behavior when
18			using OCR-D/core as a library
19			- Change setOverrideLogLevel to only override the log level of the ``ocrd``
20			logger and its descendants
21			- initLogging will set exactly one handler, for the root logger or for the
22			``ocrd`` logger.
23			- Child loggers should propagate to the ancestor logging (default
24			behavior of the logging library - no more PropagationShyLogger)
25			- disableLogging only removes any handlers from the ``ocrd`` logger
26			"""
27			# pylint: disable=no-member
28
29			from __future__ import absolute_import
30
31			from traceback import format_stack
32
33			import logging
34			import logging.config
35			from pathlib import Path
36			import sys
37
38			from .constants import LOG_FORMAT, LOG_TIMEFMT
39
40			__all__ = [
41			'disableLogging',
42			'getLevelName',
43			'getLogger',
44			'initLogging',
45			'logging',
46			'setOverrideLogLevel',
47			]
48
49			# These are the loggers we add handlers to
50			ROOT_OCRD_LOGGERS = [
51			'ocrd',
52			'ocrd_network'
53			]
54
55			LOGGING_DEFAULTS = {
56			'ocrd': logging.INFO,
57			'ocrd_network': logging.DEBUG,
58			# 'ocrd.resolver': logging.INFO,
59			# 'ocrd.resolver.download_to_directory': logging.INFO,
60			# 'ocrd.resolver.add_files_to_mets': logging.INFO,
61			# To cut back on the `Self-intersection at or near point` INFO messages
62			'shapely.geos': logging.ERROR,
63			'tensorflow': logging.ERROR,
64			'PIL': logging.INFO,
65			'paramiko.transport': logging.INFO,
66			'uvicorn.access': logging.DEBUG,
67			'uvicorn.error': logging.DEBUG,
68			'uvicorn': logging.INFO
69			}
70
71			_initialized_flag = False
72
73			_ocrdLevel2pythonLevel = {
74			'TRACE': 'DEBUG',
75			'OFF': 'CRITICAL',
76			'FATAL': 'ERROR',
77			}
78
79			def tf_disable_interactive_logs():
80			try:
81			from os import environ
82			# This env variable must be set before importing from Keras
83			environ['TF_CPP_MIN_LOG_LEVEL'] = '3'
84			from tensorflow.keras.utils import disable_interactive_logging
85			# Enabled interactive logging throws an exception
86			# due to a call of sys.stdout.flush()
87			disable_interactive_logging()
88			except ImportError:
89			# Nothing should be handled here if TF is not available
90			pass
91
92			def getLevelName(lvl):
93			"""
94			Get (string) python logging level for (string) spec-defined log level name.
95			"""
96			lvl = _ocrdLevel2pythonLevel.get(lvl, lvl)
97			return logging.getLevelName(lvl)
98
99			def getLogger(args, *kwargs):
100			"""
101			Wrapper around ``logging.getLogger`` that alls :py:func:`initLogging` if
102			that wasn't explicitly called before.
103			"""
104			logger = logging.getLogger(args, *kwargs)
105			return logger
106
107			def setOverrideLogLevel(lvl, silent=True):
108			"""
109			Override the output log level of the handlers attached to the ``ocrd`` logger.
110
111			Args:
112			lvl (string): Log level name.
113			silent (boolean): Whether to log the override call
114			"""
115			if not _initialized_flag:
116			initLogging()
117			ocrd_logger = logging.getLogger('ocrd')
118
119			if lvl is None:
120			if not silent:
121			ocrd_logger.info('Reset log level override')
122			ocrd_logger.setLevel(logging.NOTSET)
123			else:
124			if not silent:
125			ocrd_logger.info('Overriding ocrd log level to %s', lvl)
126			ocrd_logger.setLevel(lvl)
127
128			def initLogging(builtin_only=False, force_reinit=False):
129			"""
130			Reset ``ocrd`` logger, read logging configuration if exists, otherwise use basicConfig
131
132			initLogging is to be called by OCR-D/core once, i.e.
133			- for the ``ocrd`` CLI
134			- for the processor wrapper methods
135
136			Other processes that use OCR-D/core as a library can, but do not have to, use this functionality.
137
138			Keyword Args:
139			- builtin_only (bool, False): Whether to search for logging configuration
140			on-disk (``False``) or only use the
141			hard-coded config (``True``). For testing
142			- force_reinit (bool, False): Whether to ignore the module-level
143			``_initialized_flag``. For testing only.
144			"""
145			global _initialized_flag
146			if _initialized_flag and not force_reinit:
147			return
148			# disableLogging()
149
150			# https://docs.python.org/3/library/logging.html#logging.disable
151			# If logging.disable(logging.NOTSET) is called, it effectively removes this
152			# overriding level, so that logging output again depends on the effective
153			# levels of individual loggers.
154			logging.disable(logging.NOTSET)
155
156			# remove all handlers for the ocrd root loggers
157			for logger_name in ROOT_OCRD_LOGGERS:
158			for handler in logging.getLogger(logger_name).handlers[:]:
159			logging.getLogger(logger_name).removeHandler(handler)
160
161			config_file = None
162			if not builtin_only:
163			CONFIG_PATHS = [
164			Path.cwd(),
165			Path.home(),
166			Path('/etc'),
167			]
168			config_file = next((f for f \
			0 ignored issues – show Comprehensibility Best Practice introduced 2020-09-23 14:54 UTC by Report Bug Copy Issue Report The variable `f` does not seem to be defined. Loading history...
169			in [p / 'ocrd_logging.conf' for p in CONFIG_PATHS] \
170			if f.exists()),
171			None)
172			if config_file:
173			logging.config.fileConfig(config_file)
174			logging.getLogger('ocrd.logging').debug("Picked up logging config at %s", config_file)
175			else:
176			# Default logging config
177			ocrd_handler = logging.StreamHandler(stream=sys.stderr)
178			ocrd_handler.setFormatter(logging.Formatter(fmt=LOG_FORMAT, datefmt=LOG_TIMEFMT))
179			ocrd_handler.setLevel(logging.DEBUG)
180			for logger_name in ROOT_OCRD_LOGGERS:
181			logging.getLogger(logger_name).addHandler(ocrd_handler)
182			for logger_name, logger_level in LOGGING_DEFAULTS.items():
183			logging.getLogger(logger_name).setLevel(logger_level)
184
185			_initialized_flag = True
186
187			def disableLogging(silent=True):
188			"""
189			Disables all logging of the ``ocrd`` logger and descendants
190
191			Keyword Args:
192			silent (bool, True): Whether to log the call to disableLogging
193			"""
194			global _initialized_flag # pylint: disable=global-statement
195			if _initialized_flag and not silent:
196			logging.getLogger('ocrd.logging').debug("Disabling logging")
197			_initialized_flag = False
198			# logging.basicConfig(level=logging.CRITICAL)
199			# logging.disable(logging.ERROR)
200			# remove all handlers for the ocrd logger
201			for logger_name in ROOT_OCRD_LOGGERS:
202			for handler in logging.getLogger(logger_name).handlers[:]:
203			logging.getLogger(logger_name).removeHandler(handler)
204			for logger_name in LOGGING_DEFAULTS:
205			logging.getLogger(logger_name).setLevel(logging.NOTSET)
206
207			# Initializing stream handlers at module level
208			# would cause message output in all runtime contexts,
209			# including those which are already run for std output
210			# (--dump-json, --version, ocrd-tool, bashlib etc).
211			# So this needs to be an opt-in from the CLIs/decorators:
212			#initLogging()
213			# Also, we even have to block log output for libraries
214			# (like matplotlib/tensorflow) which set up logging
215			# themselves already:
216			disableLogging()
217

OCR-D / core

Pull Request — master (#1083)

ocrd_utils.logging.initLogging() C

Complexity

Size

Duplication

Importance

How to fix Long Method

Long Method

Duplication Side-by-Side

Filter issues like