warnings —- Warning control

源代码:Lib/warnings.py


Warning messages are typically issued in situations where it is useful to alertthe user of some condition in a program, where that condition (normally) doesn'twarrant raising an exception and terminating the program. For example, onemight want to issue a warning when a program uses an obsolete module.

Python programmers issue warnings by calling the warn() function definedin this module. (C programmers use PyErr_WarnEx(); see异常处理 for details).

Warning messages are normally written to sys.stderr, but their dispositioncan be changed flexibly, from ignoring all warnings to turning them intoexceptions. The disposition of warnings can vary based on the warning category, the text of the warning message, and the source location where itis issued. Repetitions of a particular warning for the same source location aretypically suppressed.

There are two stages in warning control: first, each time a warning is issued, adetermination is made whether a message should be issued or not; next, if amessage is to be issued, it is formatted and printed using a user-settable hook.

The determination whether to issue a warning message is controlled by thewarning filter, which is a sequence of matching rules and actions. Rules can beadded to the filter by calling filterwarnings() and reset to its defaultstate by calling resetwarnings().

The printing of warning messages is done by calling showwarning(), whichmay be overridden; the default implementation of this function formats themessage by calling formatwarning(), which is also available for use bycustom implementations.

参见

logging.captureWarnings() allows you to handle all warnings withthe standard logging infrastructure.

警告类别

There are a number of built-in exceptions that represent warning categories.This categorization is useful to be able to filter out groups of warnings.

While these are technicallybuilt-in exceptions, they aredocumented here, because conceptually they belong to the warnings mechanism.

User code can define additional warning categories by subclassing one of thestandard warning categories. A warning category must always be a subclass ofthe Warning class.

The following warnings category classes are currently defined:

Class描述
WarningThis is the base class of all warningcategory classes. It is a subclass ofException.
UserWarningThe default category for warn().
DeprecationWarningBase category for warnings about deprecatedfeatures when those warnings are intended forother Python developers (ignored by default,unless triggered by code in main).
SyntaxWarningBase category for warnings about dubioussyntactic features.
RuntimeWarningBase category for warnings about dubiousruntime features.
FutureWarningBase category for warnings about deprecatedfeatures when those warnings are intended forend users of applications that are written inPython.
PendingDeprecationWarningBase category for warnings about featuresthat will be deprecated in the future(ignored by default).
ImportWarningBase category for warnings triggered duringthe process of importing a module (ignored bydefault).
UnicodeWarningBase category for warnings related toUnicode.
BytesWarningBase category for warnings related tobytes and bytearray.
ResourceWarningBase category for warnings related toresource usage.

在 3.7 版更改: Previously DeprecationWarning and FutureWarning weredistinguished based on whether a feature was being removed entirely orchanging its behaviour. They are now distinguished based on theirintended audience and the way they're handled by the default warningsfilters.

The Warnings Filter

The warnings filter controls whether warnings are ignored, displayed, or turnedinto errors (raising an exception).

Conceptually, the warnings filter maintains an ordered list of filterspecifications; any specific warning is matched against each filterspecification in the list in turn until a match is found; the filter determinesthe disposition of the match. Each entry is a tuple of the form (action,message, category, module, lineno), where:

  • action is one of the following strings:

处置

"default"

为发出警告的每个位置(模块+行号)打印第一个匹配警告

"error"

将匹配警告转换为异常

"ignore"

从不打印匹配的警告

"always"

总是打印匹配的警告

"module"

为发出警告的每个模块打印第一次匹配警告(无论行号如何)

"once"

无论位置如何,仅打印第一次出现的匹配警告

  • message is a string containing a regular expression that the start ofthe warning message must match. The expression is compiled to always becase-insensitive.

  • category is a class (a subclass of Warning) of which the warningcategory must be a subclass in order to match.

  • module is a string containing a regular expression that the module name mustmatch. The expression is compiled to be case-sensitive.

  • lineno is an integer that the line number where the warning occurred mustmatch, or 0 to match all line numbers.

Since the Warning class is derived from the built-in Exceptionclass, to turn a warning into an error we simply raise category(message).

If a warning is reported and doesn't match any registered filter then the"default" action is applied (hence its name).

Describing Warning Filters

The warnings filter is initialized by -W options passed to the Pythoninterpreter command line and the PYTHONWARNINGS environment variable.The interpreter saves the arguments for all supplied entries withoutinterpretation in sys.warnoptions; the warnings module parses thesewhen it is first imported (invalid options are ignored, after printing amessage to sys.stderr).

Individual warnings filters are specified as a sequence of fields separated bycolons:

  1. action:message:category:module:line

The meaning of each of these fields is as described in The Warnings Filter.When listing multiple filters on a single line (as forPYTHONWARNINGS), the individual filters are separated by commas andthe filters listed later take precedence over those listed before them (asthey're applied left-to-right, and the most recently applied filters takeprecedence over earlier ones).

Commonly used warning filters apply to either all warnings, warnings in aparticular category, or warnings raised by particular modules or packages.Some examples:

  1. default # Show all warnings (even those ignored by default)
  2. ignore # Ignore all warnings
  3. error # Convert all warnings to errors
  4. error::ResourceWarning # Treat ResourceWarning messages as errors
  5. default::DeprecationWarning # Show DeprecationWarning messages
  6. ignore,default:::mymodule # Only report warnings triggered by "mymodule"
  7. error:::mymodule[.*] # Convert warnings to errors in "mymodule"
  8. # and any subpackages of "mymodule"

默认警告过滤器

By default, Python installs several warning filters, which can be overridden bythe -W command-line option, the PYTHONWARNINGS environmentvariable and calls to filterwarnings().

In regular release builds, the default warning filter has the following entries(in order of precedence):

  1. default::DeprecationWarning:__main__
  2. ignore::DeprecationWarning
  3. ignore::PendingDeprecationWarning
  4. ignore::ImportWarning
  5. ignore::ResourceWarning

In debug builds, the list of default warning filters is empty.

在 3.2 版更改: DeprecationWarning is now ignored by default in addition toPendingDeprecationWarning.

在 3.7 版更改: DeprecationWarning is once again shown by default when triggereddirectly by code in main.

在 3.7 版更改: BytesWarning no longer appears in the default filter list and isinstead configured via sys.warnoptions when -b is specifiedtwice.

Overriding the default filter

Developers of applications written in Python may wish to hide all Python levelwarnings from their users by default, and only display them when running testsor otherwise working on the application. The sys.warnoptions attributeused to pass filter configurations to the interpreter can be used as a marker toindicate whether or not warnings should be disabled:

  1. import sys
  2.  
  3. if not sys.warnoptions:
  4. import warnings
  5. warnings.simplefilter("ignore")

Developers of test runners for Python code are advised to instead ensure thatall warnings are displayed by default for the code under test, using codelike:

  1. import sys
  2.  
  3. if not sys.warnoptions:
  4. import os, warnings
  5. warnings.simplefilter("default") # Change the filter in this process
  6. os.environ["PYTHONWARNINGS"] = "default" # Also affect subprocesses

Finally, developers of interactive shells that run user code in a namespaceother than main are advised to ensure that DeprecationWarningmessages are made visible by default, using code like the following (whereuser_ns is the module used to execute code entered interactively):

  1. import warnings
  2. warnings.filterwarnings("default", category=DeprecationWarning,
  3. module=user_ns.get("__name__"))

暂时禁止警告

If you are using code that you know will raise a warning, such as a deprecatedfunction, but do not want to see the warning (even when warnings have beenexplicitly configured via the command line), then it is possible to suppressthe warning using the catch_warnings context manager:

  1. import warnings
  2.  
  3. def fxn():
  4. warnings.warn("deprecated", DeprecationWarning)
  5.  
  6. with warnings.catch_warnings():
  7. warnings.simplefilter("ignore")
  8. fxn()

While within the context manager all warnings will simply be ignored. Thisallows you to use known-deprecated code without having to see the warning whilenot suppressing the warning for other code that might not be aware of its useof deprecated code. Note: this can only be guaranteed in a single-threadedapplication. If two or more threads use the catch_warnings contextmanager at the same time, the behavior is undefined.

测试警告

To test warnings raised by code, use the catch_warnings contextmanager. With it you can temporarily mutate the warnings filter to facilitateyour testing. For instance, do the following to capture all raised warnings tocheck:

  1. import warnings
  2.  
  3. def fxn():
  4. warnings.warn("deprecated", DeprecationWarning)
  5.  
  6. with warnings.catch_warnings(record=True) as w:
  7. # Cause all warnings to always be triggered.
  8. warnings.simplefilter("always")
  9. # Trigger a warning.
  10. fxn()
  11. # Verify some things
  12. assert len(w) == 1
  13. assert issubclass(w[-1].category, DeprecationWarning)
  14. assert "deprecated" in str(w[-1].message)

One can also cause all warnings to be exceptions by using error instead ofalways. One thing to be aware of is that if a warning has already beenraised because of a once/default rule, then no matter what filters areset the warning will not be seen again unless the warnings registry related tothe warning has been cleared.

Once the context manager exits, the warnings filter is restored to its statewhen the context was entered. This prevents tests from changing the warningsfilter in unexpected ways between tests and leading to indeterminate testresults. The showwarning() function in the module is also restored toits original value. Note: this can only be guaranteed in a single-threadedapplication. If two or more threads use the catch_warnings contextmanager at the same time, the behavior is undefined.

When testing multiple operations that raise the same kind of warning, itis important to test them in a manner that confirms each operation is raisinga new warning (e.g. set warnings to be raised as exceptions and check theoperations raise exceptions, check that the length of the warning listcontinues to increase after each operation, or else delete the previousentries from the warnings list before each new operation).

Updating Code For New Versions of Dependencies

Warning categories that are primarily of interest to Python developers (ratherthan end users of applications written in Python) are ignored by default.

Notably, this "ignored by default" list includes DeprecationWarning(for every module except main), which means developers should make sureto test their code with typically ignored warnings made visible in order toreceive timely notifications of future breaking API changes (whether in thestandard library or third party packages).

In the ideal case, the code will have a suitable test suite, and the test runnerwill take care of implicitly enabling all warnings when running tests(the test runner provided by the unittest module does this).

In less ideal cases, applications can be checked for use of deprecatedinterfaces by passing -Wd to the Python interpreter (this isshorthand for -W default) or setting PYTHONWARNINGS=default inthe environment. This enables default handling for all warnings, including thosethat are ignored by default. To change what action is taken for encounteredwarnings you can change what argument is passed to -W (e.g.-W error). See the -W flag for more details on what ispossible.

Available Functions

  • warnings.warn(message, category=None, stacklevel=1, source=None)
  • Issue a warning, or maybe ignore it or raise an exception. The category_argument, if given, must be a warning category class; itdefaults to UserWarning. Alternatively, _message can be a Warning instance,in which case category will be ignored and message.class will be used.In this case, the message text will be str(message). This function raises anexception if the particular warning issued is changed into an error by thewarnings filter. The stacklevel argument can be used by wrapperfunctions written in Python, like this:
  1. def deprecation(message):
  2. warnings.warn(message, DeprecationWarning, stacklevel=2)

This makes the warning refer to deprecation()'s caller, rather than to thesource of deprecation() itself (since the latter would defeat the purposeof the warning message).

source, if supplied, is the destroyed object which emitted aResourceWarning.

在 3.6 版更改: Added source parameter.

  • warnings.warnexplicit(_message, category, filename, lineno, module=None, registry=None, module_globals=None, source=None)
  • This is a low-level interface to the functionality of warn(), passing inexplicitly the message, category, filename and line number, and optionally themodule name and the registry (which should be the warningregistrydictionary of the module). The module name defaults to the filename with.py stripped; if no registry is passed, the warning is never suppressed.message must be a string and category a subclass of Warning ormessage may be a Warning instance, in which case category will beignored.

module_globals, if supplied, should be the global namespace in use by the codefor which the warning is issued. (This argument is used to support displayingsource for modules found in zipfiles or other non-filesystem importsources).

source, if supplied, is the destroyed object which emitted aResourceWarning.

在 3.6 版更改: Add the source parameter.

  • warnings.showwarning(message, category, filename, lineno, file=None, line=None)
  • Write a warning to a file. The default implementation callsformatwarning(message, category, filename, lineno, line) and writes theresulting string to file, which defaults to sys.stderr. You may replacethis function with any callable by assigning to warnings.showwarning.line is a line of source code to be included in the warningmessage; if line is not supplied, showwarning() willtry to read the line specified by filename and lineno.

  • warnings.formatwarning(message, category, filename, lineno, line=None)

  • Format a warning the standard way. This returns a string which may containembedded newlines and ends in a newline. line is a line of source code tobe included in the warning message; if line is not supplied,formatwarning() will try to read the line specified by filename andlineno.

  • warnings.filterwarnings(action, message='', category=Warning, module='', lineno=0, append=False)

  • Insert an entry into the list of warnings filter specifications. The entry is inserted at the front by default; ifappend is true, it is inserted at the end. This checks the types of thearguments, compiles the message and module regular expressions, andinserts them as a tuple in the list of warnings filters. Entries closer tothe front of the list override entries later in the list, if both match aparticular warning. Omitted arguments default to a value that matcheseverything.

  • warnings.simplefilter(action, category=Warning, lineno=0, append=False)

  • Insert a simple entry into the list of warnings filter specifications. The meaning of the function parameters is as forfilterwarnings(), but regular expressions are not needed as the filterinserted always matches any message in any module as long as the category andline number match.

  • warnings.resetwarnings()

  • Reset the warnings filter. This discards the effect of all previous calls tofilterwarnings(), including that of the -W command line optionsand calls to simplefilter().

Available Context Managers

  • class warnings.catchwarnings(*, _record=False, module=None)
  • A context manager that copies and, upon exit, restores the warnings filterand the showwarning() function.If the record argument is False (the default) the context managerreturns None on entry. If record is True, a list isreturned that is progressively populated with objects as seen by a customshowwarning() function (which also suppresses output to sys.stdout).Each object in the list has attributes with the same names as the arguments toshowwarning().

The module argument takes a module that will be used instead of themodule returned when you import warnings whose filter will beprotected. This argument exists primarily for testing the warningsmodule itself.

注解

The catch_warnings manager works by replacing andthen later restoring the module'sshowwarning() function and internal list of filterspecifications. This means the context manager is modifyingglobal state and therefore is not thread-safe.