passlib.context - CryptContext Hash Manager

This page provides a complete reference of all the methods and options supported by the CryptContext class and helper utilities.

See also

The CryptContext Class

class passlib.context.CryptContext(schemes=None, **kwds)

Helper for hashing passwords using different algorithms.

At its base, this is a proxy object that makes it easy to use multiple PasswordHash objects at the same time. Instances of this class can be created by calling the constructor with the appropriate keywords, or by using one of the alternate constructors, which can load directly from a string or a local file. Since this class has so many options and methods, they have been broken out into subsections:

Constructor Keywords

The CryptContext class accepts the following keywords, all of which are optional. The keywords are divided into two categories: context options, which affect the CryptContext itself; and algorithm options, which place defaults and limits on the algorithms used by the CryptContext.

Context Options

Options which directly affect the behavior of the CryptContext instance:

schemes

List of algorithms which the instance should support.

The most important option in the constructor, This option controls what hashes can be used by the hash() method, which hashes will be recognized by verify() and identify(), and other effects throughout the instance. It should be a sequence of names, drawn from the hashes in passlib.hash. Listing an unknown name will cause a ValueError. You can use the schemes() method to get a list of the currently configured algorithms. As an example, the following creates a CryptContext instance which supports the sha256_crypt and des_crypt schemes:

>>> from passlib.context import CryptContext
>>> myctx = CryptContext(schemes=["sha256_crypt", "des_crypt"])
>>> myctx.schemes()
("sha256_crypt", "des_crypt")

Note

The order of the schemes is sometimes important, as identify() will run through the schemes from first to last until an algorithm “claims” the hash. So plaintext algorithms and the like should be listed at the end.

See also

the Basic Usage example in the tutorial.

default

Specifies the name of the default scheme.

This option controls which of the configured schemes will be used as the default when creating new hashes. This parameter is optional; if omitted, the first non-deprecated algorithm in schemes will be used. You can use the default_scheme() method to retrieve the name of the current default scheme. As an example, the following demonstrates the effect of this parameter on the hash() method:

>>> from passlib.context import CryptContext
>>> myctx = CryptContext(schemes=["sha256_crypt", "md5_crypt"])

>>> # hash() uses the first scheme
>>> myctx.default_scheme()
'sha256_crypt'
>>> myctx.hash("password")
'$5$rounds=80000$R5ZIZRTNPgbdcWq5$fT/Oeqq/apMa/0fbx8YheYWS6Z3XLTxCzEtutsk2cJ1'

>>> # but setting default causes the second scheme to be used.
>>> myctx.update(default="md5_crypt")
>>> myctx.default_scheme()
'md5_crypt'
>>> myctx.hash("password")
'$1$Rr0C.KI8$Kvciy8pqfL9BQ2CJzEzfZ/'

See also

the Basic Usage example in the tutorial.

deprecated

List of algorithms which should be considered “deprecated”.

This has the same format as schemes, and should be a subset of those algorithms. The main purpose of this method is to flag schemes which need to be rehashed when the user next logs in. This has no effect on the Primary Methods; but if the special Hash Migration methods are passed a hash belonging to a deprecated scheme, they will flag it as needed to be rehashed using the default scheme.

This may also contain a single special value, ["auto"], which will configure the CryptContext instance to deprecate all supported schemes except for the default scheme.

New in version 1.6: Added support for the ["auto"] value.

See also

Deprecation & Hash Migration in the tutorial

truncate_error

By default, some algorithms will truncate large passwords (e.g. bcrypt truncates ones larger than 72 bytes). Such hashes accept a truncate_error=True option to make them raise a PasswordTruncateError instead.

This can also be set at the CryptContext level, and will passed to all hashes that support it.

New in version 1.7.

min_verify_time

If specified, unsuccessful verify() calls will be penalized, and take at least this may seconds before the method returns. May be an integer or fractional number of seconds.

Deprecated since version 1.6: This option has not proved very useful, is ignored by 1.7, and will be removed in version 1.8.

Changed in version 1.7: Per deprecation roadmap above, this option is now ignored.

harden_verify

Companion to min_verify_time, currently ignored.

New in version 1.7.

Deprecated since version 1.7.1: This option is ignored by 1.7.1, and will be removed in 1.8 along with min_verify_time.

Algorithm Options

All of the other options that can be passed to a CryptContext constructor affect individual hash algorithms. All of the following keys have the form scheme__key, where scheme is the name of one of the algorithms listed in schemes, and option one of the parameters below:

scheme__rounds

Set the number of rounds required for this scheme when generating new hashes (using hash()). Existing hashes which have a different number of rounds will be marked as deprecated.

This essentially sets default_rounds, min_rounds, and max_rounds all at once. If any of those options are also specified, they will override the value specified by rounds.

New in version 1.7: Previous releases of Passlib treated this as an alias for default_rounds.

scheme__default_rounds

Sets the default number of rounds to use with this scheme when generating new hashes (using hash()).

If not set, this will fall back to the an algorithm-specific default_rounds. For hashes which do not support a rounds parameter, this option is ignored. As an example:

>>> from passlib.context import CryptContext

>>> # no explicit default_rounds set, so hash() uses sha256_crypt's default (80000)
>>> myctx = CryptContext(["sha256_crypt"])
>>> myctx.hash("fooey")
'$5$rounds=80000$60Y7mpmAhUv6RDvj$AdseAOq6bKUZRDRTr/2QK1t38qm3P6sYeXhXKnBAmg0'
           ^^^^^

>>> # but if a default is specified, it will be used instead.
>>> myctx = CryptContext(["sha256_crypt"], sha256_crypt__default_rounds=77123)
>>> myctx.hash("fooey")
'$5$rounds=77123$60Y7mpmAhUv6RDvj$AdseAOq6bKUZRDRTr/2QK1t38qm3P6sYeXhXKnBAmg0'
           ^^^^^

See also

the Using Default Settings example in the tutorial.

scheme__vary_rounds

Deprecated since version 1.7: This option has been deprecated as of Passlib 1.7, and will be removed in Passlib 2.0. The (very minimal) security benefit it provides was judged to not be worth code complexity it requires.

Instead of using a fixed rounds value (such as specified by default_rounds, above); this option will cause each call to hash() to vary the default rounds value by some amount.

This can be an integer value, in which case each call will use a rounds value within the range default_rounds +/- vary_rounds. It may also be a floating point value within the range 0.0 .. 1.0, in which case the range will be calculated as a proportion of the current default rounds (default_rounds +/- default_rounds*vary_rounds). A typical setting is 0.1 to 0.2.

As an example of how this parameter operates:

>>> # without vary_rounds set, hash() uses the same amount each time:
>>> from passlib.context import CryptContext
>>> myctx = CryptContext(schemes=["sha256_crypt"],
...                      sha256_crypt__default_rounds=80000)
>>> myctx.hash("fooey")
'$5$rounds=80000$60Y7mpmAhUv6RDvj$AdseAOq6bKUZRDRTr/2QK1t38qm3P6sYeXhXKnBAmg0'
>>> myctx.hash("fooey")
'$5$rounds=80000$60Y7mpmAhUv6RDvj$AdseAOq6bKUZRDRTr/2QK1t38qm3P6sYeXhXKnBAmg0'
           ^^^^^

>>> # but if vary_rounds is set, each one will be randomized
>>> # (in this case, within the range 72000 .. 88000)
>>> myctx = CryptContext(schemes=["sha256_crypt"],
...                      sha256_crypt__default_rounds=80000,
...                      sha256_crypt__vary_rounds=0.1)
>>> myctx.hash("fooey")
'$5$rounds=83966$bMpgQxN2hXo2kVr4$jL4Q3ov41UPgSbO7jYL0PdtsOg5koo4mCa.UEF3zan.'
>>> myctx.hash("fooey")
'$5$rounds=72109$43BBHC/hYPHzL69c$VYvVIdKn3Zdnvu0oJHVlo6rr0WjiMTGmlrZrrH.GxnA'
           ^^^^^

Note

This is not a needed security measure, but it lets some of the less-significant digits of the rounds value act as extra salt bits; and helps foil any attacks targeted at a specific number of rounds of a hash.

scheme__min_rounds, scheme__max_rounds

These options place a limit on the number of rounds allowed for a particular scheme.

For one, they limit what values are allowed for default_rounds, and clip the effective range of the vary_rounds parameter. More importantly though, they proscribe a minimum strength for the hash, and any hashes which don’t have sufficient rounds will be flagged as needing rehashing by the Hash Migration methods.

Note

These are configurable per-context limits. A warning will be issued if they exceed any hard limits set by the algorithm itself.

See also

the Settings Rounds Limitations example in the tutorial.

scheme__other-option

Finally, any other options are assumed to correspond to one of the that algorithm’s hash() settings, such as setting a salt_size.

See also

the Using Default Settings example in the tutorial.

Global Algorithm Options

all__option

The special scheme all permits you to set an option, and have it act as a global default for all the algorithms in the context. For instance, all__vary_rounds=0.1 would set the vary_rounds option for all the schemes where it was not overridden with an explicit scheme__vary_rounds option.

Deprecated since version 1.7: This special scheme is deprecated as of Passlib 1.7, and will be removed in Passlib 2.0. It’s only legitimate use was for vary_rounds, which is also being removed in Passlib 2.0.

User Categories

category__context__option, category__scheme__option

Passing keys with this format to the CryptContext constructor allows you to specify conditional context and algorithm options, controlled by the category parameter supported by most CryptContext methods.

These options are conditional because they only take effect if the category prefix of the option matches the value of the category parameter of the CryptContext method being invoked. In that case, they override any options specified without a category prefix (e.g. admin__sha256_crypt__min_rounds would override sha256_crypt__min_rounds). The category prefix and the value passed into the category parameter can be any string the application wishes to use, the only constraint is that None indicates the default category.

Motivation: Policy limits such as default rounds values and deprecated schemes generally have to be set globally. However, it’s frequently desirable to specify stronger options for certain accounts (such as admin accounts), choosing to sacrifice longer hashing time for a more secure password. The user categories system allows for this. For example, a CryptContext could be set up as follows:

>>> # A context object can be set up as follows:
>>> from passlib.context import CryptContext
>>> myctx = CryptContext(schemes=["sha256_crypt"],
...                      sha256_crypt__default_rounds=77000,
...                      staff__sha256_crypt__default_rounds=88000)

>>> # In this case, calling hash() with ``category=None`` would result
>>> # in a hash that used 77000 sha256-crypt rounds:
>>> myctx.hash("password", category=None)
'$5$rounds=77000$sj3XI0AbKlEydAKt$BhFvyh4.IoxaUeNlW6rvQ.O0w8BtgLQMYorkCOMzf84'
           ^^^^^

>>> # But if the application passed in ``category="staff"`` when an administrative
>>> # account set their password, 88000 rounds would be used:
>>> myctx.hash("password", category="staff")
'$5$rounds=88000$w7XIdKfTI9.YLwmA$MIzGvs6NU1QOQuuDHhICLmDsdW/t94Bbdfxdh/6NJl7'
           ^^^^^

Primary Methods

The main interface to the CryptContext object deliberately mirrors the PasswordHash interface, since its central purpose is to act as a container for multiple password hashes. Most applications will only need to make use two methods in a CryptContext instance:

CryptContext.hash(secret, scheme=None, category=None, **kwds)

run secret through selected algorithm, returning resulting hash.

Parameters:
  • secret (unicode or bytes) – the password to hash.
  • scheme (str or None) –

    Optional scheme to use. Scheme must be one of the ones configured for this context (see the schemes option). If no scheme is specified, the configured default will be used.

    Deprecated since version 1.7: Support for this keyword is deprecated, and will be removed in Passlib 2.0.

  • category (str or None) – Optional user category. If specified, this will cause any category-specific defaults to be used when hashing the password (e.g. different default scheme, different default rounds values, etc).
  • **kwds – All other keyword options are passed to the selected algorithm’s PasswordHash.hash() method.
Returns:

The secret as encoded by the specified algorithm and options. The return value will always be a str.

Raises:

TypeError, ValueError

  • If any of the arguments have an invalid type or value. This includes any keywords passed to the underlying hash’s PasswordHash.hash() method.

See also

the Basic Usage example in the tutorial

CryptContext.encrypt(*args, **kwds)

Legacy alias for hash().

Deprecated since version 1.7: This method was renamed to hash() in version 1.7. This alias will be removed in version 2.0, and should only be used for compatibility with Passlib 1.3 - 1.6.

CryptContext.verify(secret, hash, scheme=None, category=None, **kwds)

verify secret against an existing hash.

If no scheme is specified, this will attempt to identify the scheme based on the contents of the provided hash (limited to the schemes configured for this context). It will then check whether the password verifies against the hash.

Parameters:
  • secret (unicode or bytes) – the secret to verify
  • hash (unicode or bytes) –

    hash string to compare to

    if None is passed in, this will be treated as “never verifying”

  • scheme (str) –

    Optionally force context to use specific scheme. This is usually not needed, as most hashes can be unambiguously identified. Scheme must be one of the ones configured for this context (see the schemes option).

    Deprecated since version 1.7: Support for this keyword is deprecated, and will be removed in Passlib 2.0.

  • category (str or None) – Optional user category string. This is mainly used when generating new hashes, it has little effect when verifying; this keyword is mainly provided for symmetry.
  • **kwds – All additional keywords are passed to the appropriate handler, and should match its context_kwds.
Returns:

True if the password matched the hash, else False.

Raises:
  • ValueError
    • if the hash did not match any of the configured schemes().
    • if any of the arguments have an invalid value (this includes any keywords passed to the underlying hash’s PasswordHash.verify() method).
  • TypeError
    • if any of the arguments have an invalid type (this includes any keywords passed to the underlying hash’s PasswordHash.verify() method).

See also

the Basic Usage example in the tutorial

CryptContext.identify(hash, category=None, resolve=False, required=False, unconfigured=False)

Attempt to identify which algorithm the hash belongs to.

Note that this will only consider the algorithms currently configured for this context (see the schemes option). All registered algorithms will be checked, from first to last, and whichever one positively identifies the hash first will be returned.

Parameters:
  • hash (unicode or bytes) – The hash string to test.
  • category (str or None) – Optional user category. Ignored by this function, this parameter is provided for symmetry with the other methods.
  • resolve (bool) – If True, returns the hash handler itself, instead of the name of the hash.
  • required (bool) – If True, this will raise a ValueError if the hash cannot be identified, instead of returning None.
Returns:

The handler which first identifies the hash, or None if none of the algorithms identify the hash.

CryptContext.dummy_verify(elapsed=0)

Helper that applications can call when user wasn’t found, in order to simulate time it would take to hash a password.

Runs verify() against a dummy hash, to simulate verification of a real account password.

Parameters:elapsed

Deprecated since version 1.7.1: this option is ignored, and will be removed in passlib 1.8.

New in version 1.7.

“crypt”-style methods

Additionally, the main interface offers wrappers for the two Unix “crypt” style methods provided by all the PasswordHash objects:

CryptContext.genhash(secret, config, scheme=None, category=None, **kwds)

Generate hash for the specified secret using another hash.

Deprecated since version 1.7: This method will be removed in version 2.0, and should only be used for compatibility with Passlib 1.3 - 1.6.

CryptContext.genconfig(scheme=None, category=None, **settings)

Generate a config string for specified scheme.

Deprecated since version 1.7: This method will be removed in version 2.0, and should only be used for compatibility with Passlib 1.3 - 1.6.

Hash Migration

Applications which want to detect and regenerate deprecated hashes will want to use one of the following methods:

CryptContext.verify_and_update(secret, hash, scheme=None, category=None, **kwds)

verify password and re-hash the password if needed, all in a single call.

This is a convenience method which takes care of all the following: first it verifies the password (verify()), if this is successfull it checks if the hash needs updating (needs_update()), and if so, re-hashes the password (hash()), returning the replacement hash. This series of steps is a very common task for applications which wish to update deprecated hashes, and this call takes care of all 3 steps efficiently.

Parameters:
  • secret (unicode or bytes) – the secret to verify
  • hash

    hash string to compare to.

    if None is passed in, this will be treated as “never verifying”

  • scheme (str) –

    Optionally force context to use specific scheme. This is usually not needed, as most hashes can be unambiguously identified. Scheme must be one of the ones configured for this context (see the schemes option).

    Deprecated since version 1.7: Support for this keyword is deprecated, and will be removed in Passlib 2.0.

  • category (str or None) – Optional user category. If specified, this will cause any category-specific defaults to be used if the password has to be re-hashed.
  • **kwds – all additional keywords are passed to the appropriate handler, and should match that hash’s PasswordHash.context_kwds.
Returns:

This function returns a tuple containing two elements: (verified, replacement_hash). The first is a boolean flag indicating whether the password verified, and the second an optional replacement hash. The tuple will always match one of the following 3 cases:

  • (False, None) indicates the secret failed to verify.
  • (True, None) indicates the secret verified correctly, and the hash does not need updating.
  • (True, str) indicates the secret verified correctly, but the current hash needs to be updated. The str will be the freshly generated hash, to replace the old one.

Raises:

TypeError, ValueError – For the same reasons as verify().

See also

the Deprecation & Hash Migration example in the tutorial.

CryptContext.needs_update(hash, scheme=None, category=None, secret=None)

Check if hash needs to be replaced for some reason, in which case the secret should be re-hashed.

This function is the core of CryptContext’s support for hash migration: This function takes in a hash string, and checks the scheme, number of rounds, and other properties against the current policy. It returns True if the hash is using a deprecated scheme, or is otherwise outside of the bounds specified by the policy (e.g. the number of rounds is lower than min_rounds configuration for that algorithm). If so, the password should be re-hashed using hash() Otherwise, it will return False.

Parameters:
  • hash (unicode or bytes) – The hash string to examine.
  • scheme (str or None) –

    Optional scheme to use. Scheme must be one of the ones configured for this context (see the schemes option). If no scheme is specified, it will be identified based on the value of hash.

    Deprecated since version 1.7: Support for this keyword is deprecated, and will be removed in Passlib 2.0.

  • category (str or None) – Optional user category. If specified, this will cause any category-specific defaults to be used when determining if the hash needs to be updated (e.g. is below the minimum rounds).
  • secret (unicode, bytes, or None) –

    Optional secret associated with the provided hash. This is not required, or even currently used for anything… it’s for forward-compatibility with any future update checks that might need this information. If provided, Passlib assumes the secret has already been verified successfully against the hash.

    New in version 1.6.

Returns:

True if hash should be replaced, otherwise False.

Raises:

ValueError – If the hash did not match any of the configured schemes().

New in version 1.6: This method was previously named hash_needs_update().

See also

the Deprecation & Hash Migration example in the tutorial.

CryptContext.hash_needs_update(hash, scheme=None, category=None)

Legacy alias for needs_update().

Deprecated since version 1.6: This method was renamed to needs_update() in version 1.6. This alias will be removed in version 2.0, and should only be used for compatibility with Passlib 1.3 - 1.5.

Disabled Hash Managment

New in version 1.7.

It’s frequently useful to disable a user’s ability to login by replacing their password hash with a standin that’s guaranteed to never verify, against any password. CryptContext offers some convenience methods for this through the following API.

CryptContext.disable(hash=None)

return a string to disable logins for user, usually by returning a non-verifying string such as "!".

Parameters:hash – Callers can optionally provide the account’s existing hash. Some disabled handlers (such as unix_disabled) will encode this into the returned value, so that it can be recovered via enable().
Raises:RuntimeError – if this function is called w/o a disabled hasher (such as unix_disabled) included in the list of schemes.
Returns:hash string which will be recognized as valid by the context, but is guaranteed to not validate against any password.
CryptContext.enable(hash)

inverse of disable() – attempts to recover original hash which was converted by a disable() call into a disabled hash – thus restoring the user’s original password.

Raises:ValueError – if original hash not present, or if the disabled handler doesn’t support encoding the original hash (e.g. django_disabled)
Returns:the original hash.
CryptContext.is_enabled(hash)

test if hash represents a usuable password – i.e. does not represent an unusuable password such as "!", which is recognized by the unix_disabled hash.

Raises:ValueError – if the hash is not recognized (typically solved by adding unix_disabled to the list of schemes).

Alternate Constructors

In addition to the main class constructor, which accepts a configuration as a set of keywords, there are the following alternate constructors:

classmethod CryptContext.from_string(source, section='passlib', encoding='utf-8')

create new CryptContext instance from an INI-formatted string.

Parameters:
  • source (unicode or bytes) – string containing INI-formatted content.
  • section (str) – option name of section to read from, defaults to "passlib".
  • encoding (str) – optional encoding used when source is bytes, defaults to "utf-8".
Returns:

new CryptContext instance, configured based on the parameters in the source string.

Usage example:

>>> from passlib.context import CryptContext
>>> context = CryptContext.from_string('''
... [passlib]
... schemes = sha256_crypt, des_crypt
... sha256_crypt__default_rounds = 30000
... ''')

New in version 1.6.

See also

to_string(), the inverse of this constructor.

classmethod CryptContext.from_path(path, section='passlib', encoding='utf-8')

create new CryptContext instance from an INI-formatted file.

this functions exactly the same as from_string(), except that it loads from a local file.

Parameters:
  • path (str) – path to local file containing INI-formatted config.
  • section (str) – option name of section to read from, defaults to "passlib".
  • encoding (str) – encoding used to load file, defaults to "utf-8".
Returns:

new CryptContext instance, configured based on the parameters stored in the file path.

New in version 1.6.

See also

from_string() for an equivalent usage example.

CryptContext.copy(**kwds)

Return copy of existing CryptContext instance.

This function returns a new CryptContext instance whose configuration is exactly the same as the original, with the exception that any keywords passed in will take precedence over the original settings. As an example:

>>> from passlib.context import CryptContext

>>> # given an existing context...
>>> ctx1 = CryptContext(["sha256_crypt", "md5_crypt"])

>>> # copy can be used to make a clone, and update
>>> # some of the settings at the same time...
>>> ctx2 = custom_app_context.copy(default="md5_crypt")

>>> # and the original will be unaffected by the change
>>> ctx1.default_scheme()
"sha256_crypt"
>>> ctx2.default_scheme()
"md5_crypt"

New in version 1.6: This method was previously named replace(). That alias has been deprecated, and will be removed in Passlib 1.8.

See also

update()

Changing the Configuration

CryptContext objects can have their configuration replaced or updated on the fly, and from a variety of sources (keywords, strings, files). This is done through three methods:

CryptContext.update(**kwds)

Helper for quickly changing configuration.

This acts much like the dict.update() method: it updates the context’s configuration, replacing the original value(s) for the specified keys, and preserving the rest. It accepts any keyword accepted by the CryptContext constructor.

New in version 1.6.

See also

copy()

CryptContext.load(source, update=False, section='passlib', encoding='utf-8')

Load new configuration into CryptContext, replacing existing config.

Parameters:
  • source

    source of new configuration to load. this value can be a number of different types:

    • a dict object, or compatible Mapping
      the key/value pairs will be interpreted the same keywords for the CryptContext class constructor.
    • a unicode or bytes string
      this will be interpreted as an INI-formatted file, and appropriate key/value pairs will be loaded from the specified section.
    • another CryptContext object.
      this will export a snapshot of its configuration using to_dict().
  • update (bool) – By default, load() will replace the existing configuration entirely. If update=True, it will preserve any existing configuration options that are not overridden by the new source, much like the update() method.
  • section (str) – When parsing an INI-formatted string, load() will look for a section named "passlib". This option allows an alternate section name to be used. Ignored when loading from a dictionary.
  • encoding (str) –

    Encoding to use when source is bytes. Defaults to "utf-8". Ignored when loading from a dictionary.

    Deprecated since version 1.8: This keyword, and support for bytes input, will be dropped in Passlib 2.0

Raises:
  • TypeError
    • If the source cannot be identified.
    • If an unknown / malformed keyword is encountered.
  • ValueError – If an invalid keyword value is encountered.

Note

If an error occurs during a load() call, the CryptContext instance will be restored to the configuration it was in before the load() call was made; this is to ensure it is never left in an inconsistent state due to a load error.

New in version 1.6.

CryptContext.load_path(path, update=False, section='passlib', encoding='utf-8')

Load new configuration into CryptContext from a local file.

This function is a wrapper for load() which loads a configuration string from the local file path, instead of an in-memory source. Its behavior and options are otherwise identical to load() when provided with an INI-formatted string.

New in version 1.6.

Examining the Configuration

The CryptContext object also supports basic inspection of its current configuration:

CryptContext.schemes(resolve=False, category=None, unconfigured=False)

return schemes loaded into this CryptContext instance.

Parameters:resolve (bool) – if True, will return a tuple of PasswordHash objects instead of their names.
Returns:returns tuple of the schemes configured for this context via the schemes option.

New in version 1.6: This was previously available as CryptContext().policy.schemes()

See also

the schemes option for usage example.

CryptContext.default_scheme(category=None, resolve=False, unconfigured=False)

return name of scheme that hash() will use by default.

Parameters:
  • resolve (bool) – if True, will return a PasswordHash object instead of the name.
  • category (str or None) – Optional user category. If specified, this will return the catgory-specific default scheme instead.
Returns:

name of the default scheme.

See also

the default option for usage example.

New in version 1.6.

Changed in version 1.7: This now returns a hasher configured with any CryptContext-specific options (custom rounds settings, etc). Previously this returned the base hasher from passlib.hash.

CryptContext.handler(scheme=None, category=None, unconfigured=False)

helper to resolve name of scheme -> PasswordHash object used by scheme.

Parameters:
  • scheme – This should identify the scheme to lookup. If omitted or set to None, this will return the handler for the default scheme.
  • category – If a user category is specified, and no scheme is provided, it will use the default for that category. Otherwise this parameter is ignored.
  • unconfigured – By default, this returns a handler object whose .hash() and .needs_update() methods will honor the configured provided by CryptContext. See unconfigured=True to get the underlying handler from before any context-specific configuration was applied.
Raises:

KeyError – If the scheme does not exist OR is not being used within this context.

Returns:

PasswordHash object used to implement the named scheme within this context (this will usually be one of the objects from passlib.hash)

New in version 1.6: This was previously available as CryptContext().policy.get_handler()

Changed in version 1.7: This now returns a hasher configured with any CryptContext-specific options (custom rounds settings, etc). Previously this returned the base hasher from passlib.hash.

CryptContext.context_kwds

return set containing union of all contextual keywords supported by the handlers in this context.

New in version 1.6.6.

Saving the Configuration

More detailed inspection can be done by exporting the configuration using one of the serialization methods:

CryptContext.to_dict(resolve=False)

Return current configuration as a dictionary.

Parameters:resolve (bool) – if True, the schemes key will contain a list of a PasswordHash objects instead of just their names.

This method dumps the current configuration of the CryptContext instance. The key/value pairs should be in the format accepted by the CryptContext class constructor, in fact CryptContext(**myctx.to_dict()) will create an exact copy of myctx. As an example:

>>> # you can dump the configuration of any crypt context...
>>> from passlib.apps import ldap_nocrypt_context
>>> ldap_nocrypt_context.to_dict()
{'schemes': ['ldap_salted_sha1',
'ldap_salted_md5',
'ldap_sha1',
'ldap_md5',
'ldap_plaintext']}

New in version 1.6: This was previously available as CryptContext().policy.to_dict()

See also

the Loading & Saving a CryptContext example in the tutorial.

CryptContext.to_string(section='passlib')

serialize to INI format and return as unicode string.

Parameters:section – name of INI section to output, defaults to "passlib".
Returns:CryptContext configuration, serialized to a INI unicode string.

This function acts exactly like to_dict(), except that it serializes all the contents into a single human-readable string, which can be hand edited, and/or stored in a file. The output of this method is accepted by from_string(), from_path(), and load(). As an example:

>>> # you can dump the configuration of any crypt context...
>>> from passlib.apps import ldap_nocrypt_context
>>> print ldap_nocrypt_context.to_string()
[passlib]
schemes = ldap_salted_sha1, ldap_salted_md5, ldap_sha1, ldap_md5, ldap_plaintext

New in version 1.6: This was previously available as CryptContext().policy.to_string()

See also

the Loading & Saving a CryptContext example in the tutorial.

Configuration Errors

The following errors may be raised when creating a CryptContext instance via any of its constructors, or when updating the configuration of an existing instance:

raises ValueError:
 
  • If a configuration option contains an invalid value (e.g. all__vary_rounds=-1).
  • If the configuration contains valid but incompatible options (e.g. listing a scheme as both default and deprecated).
raises KeyError:
 
  • If the configuration contains an unknown or forbidden option (e.g. scheme__salt).
  • If the schemes, default, or deprecated options reference an unknown hash scheme (e.g. schemes=['xxx'])
raises TypeError:
 
  • If a configuration value has the wrong type (e.g. schemes=123).

Note that this error shouldn’t occur when loading configurations from a file/string (e.g. using CryptContext.from_string()).

Additionally, a PasslibConfigWarning may be issued if any invalid-but-correctable values are encountered (e.g. if sha256_crypt__min_rounds is set to less than sha256_crypt ‘s minimum of 1000).

Changed in version 1.6: Previous releases used Python’s builtin UserWarning instead of the more specific passlib.exc.PasslibConfigWarning.

Other Helpers

class passlib.context.LazyCryptContext([schemes=None, ]**kwds[, onload=None])

CryptContext subclass which doesn’t load handlers until needed.

This is a subclass of CryptContext which takes in a set of arguments exactly like CryptContext, but won’t import any handlers (or even parse its arguments) until the first time one of its methods is accessed.

Parameters:
  • schemes – The first positional argument can be a list of schemes, or omitted, just like CryptContext.
  • onload

    If a callable is passed in via this keyword, it will be invoked at lazy-load time with the following signature: onload(**kwds) -> kwds; where kwds is all the additional kwds passed to LazyCryptContext. It should perform any additional deferred initialization, and return the final dict of options to be passed to CryptContext.

    New in version 1.6.

  • create_policy

    Deprecated since version 1.6: This option will be removed in Passlib 1.8, applications should use onload instead.

  • kwds – All additional keywords are passed to CryptContext; or to the onload function (if provided).

This is mainly used internally by modules such as passlib.apps, which define a large number of contexts, but only a few of them will be needed at any one time. Use of this class saves the memory needed to import the specified handlers until the context instance is actually accessed. As well, it allows constructing a context at module-init time, but using onload() to provide dynamic configuration at application-run time.

Note

This class is only useful if you’re referencing handler objects by name, and don’t want them imported until runtime. If you want to have the config validated before your application runs, or are passing in already-imported handler instances, you should use CryptContext instead.

New in version 1.4.

The CryptPolicy Class (deprecated)

class passlib.context.CryptPolicy(*args, **kwds)

Deprecated since version 1.6: This class has been deprecated, and will be removed in Passlib 1.8. All of its functionality has been rolled into CryptContext.

This class previously stored the configuration options for the CryptContext class. In the interest of interface simplification, all of this class’ functionality has been rolled into the CryptContext class itself. The documentation for this class is now focused on documenting how to migrate to the new api. Additionally, where possible, the deprecation warnings issued by the CryptPolicy methods will list the replacement call that should be used.

Constructors

CryptPolicy objects can be constructed directly using any of the keywords accepted by CryptContext. Direct uses of the CryptPolicy constructor should either pass the keywords directly into the CryptContext constructor, or to CryptContext.update() if the policy object was being used to update an existing context object.

In addition to passing in keywords directly, CryptPolicy objects can be constructed by the following methods:

classmethod from_path(path, section='passlib', encoding='utf-8')

create a CryptPolicy instance from a local file.

Deprecated since version 1.6.

Creating a new CryptContext from a file, which was previously done via CryptContext(policy=CryptPolicy.from_path(path)), can now be done via CryptContext.from_path(path). See CryptContext.from_path() for details.

Updating an existing CryptContext from a file, which was previously done context.policy = CryptPolicy.from_path(path), can now be done via context.load_path(path). See CryptContext.load_path() for details.

classmethod from_string(source, section='passlib', encoding='utf-8')

create a CryptPolicy instance from a string.

Deprecated since version 1.6.

Creating a new CryptContext from a string, which was previously done via CryptContext(policy=CryptPolicy.from_string(data)), can now be done via CryptContext.from_string(data). See CryptContext.from_string() for details.

Updating an existing CryptContext from a string, which was previously done context.policy = CryptPolicy.from_string(data), can now be done via context.load(data). See CryptContext.load() for details.

classmethod from_source(source, _warn=True)

create a CryptPolicy instance from some source.

this method autodetects the source type, and invokes the appropriate constructor automatically. it attempts to detect whether the source is a configuration string, a filepath, a dictionary, or an existing CryptPolicy instance.

Deprecated since version 1.6.

Create a new CryptContext, which could previously be done via CryptContext(policy=CryptPolicy.from_source(source)), should now be done using an explicit method: the CryptContext constructor itself, CryptContext.from_path(), or CryptContext.from_string().

Updating an existing CryptContext, which could previously be done via context.policy = CryptPolicy.from_source(source), should now be done using an explicit method: CryptContext.update(), or CryptContext.load().

classmethod from_sources(sources, _warn=True)

create a CryptPolicy instance by merging multiple sources.

each source is interpreted as by from_source(), and the results are merged together.

Deprecated since version 1.6: Instead of using this method to merge multiple policies together, a CryptContext instance should be created, and then the multiple sources merged together via CryptContext.load().

replace(*args, **kwds)

create a new CryptPolicy, optionally updating parts of the existing configuration.

Deprecated since version 1.6: Callers of this method should CryptContext.update() or CryptContext.copy() instead.

Introspection

All of the informational methods provided by this class have been deprecated by identical or similar methods in the CryptContext class:

has_schemes()

return True if policy defines any schemes for use.

Deprecated since version 1.6: applications should use bool(context.schemes()) instead. see CryptContext.schemes().

schemes(resolve=False)

return list of schemes defined in policy.

Deprecated since version 1.6: applications should use CryptContext.schemes() instead.

iter_handlers()

return iterator over handlers defined in policy.

Deprecated since version 1.6: applications should use context.schemes(resolve=True)) instead. see CryptContext.schemes().

get_handler(name=None, category=None, required=False)

return handler as specified by name, or default handler.

Deprecated since version 1.6: applications should use CryptContext.handler() instead, though note that the required keyword has been removed, and the new method will always act as if required=True.

get_options(name, category=None)

return dictionary of options specific to a given handler.

Deprecated since version 1.6: this method has no direct replacement in the 1.6 api, as there is not a clearly defined use-case. however, examining the output of CryptContext.to_dict() should serve as the closest alternative.

handler_is_deprecated(name, category=None)

check if handler has been deprecated by policy.

Deprecated since version 1.6: this method has no direct replacement in the 1.6 api, as there is not a clearly defined use-case. however, examining the output of CryptContext.to_dict() should serve as the closest alternative.

get_min_verify_time(category=None)

get min_verify_time setting for policy.

Deprecated since version 1.6: min_verify_time option will be removed entirely in passlib 1.8

Changed in version 1.7: this method now always returns the value automatically calculated by CryptContext.min_verify_time(), any value specified by policy is ignored.

Exporting

iter_config(ini=False, resolve=False)

iterate over key/value pairs representing the policy object.

Deprecated since version 1.6: applications should use CryptContext.to_dict() instead.

to_dict(resolve=False)

export policy object as dictionary of options.

Deprecated since version 1.6: applications should use CryptContext.to_dict() instead.

to_file(stream, section='passlib')

export policy to file.

Deprecated since version 1.6: applications should use CryptContext.to_string() instead, and then write the output to a file as desired.

to_string(section='passlib', encoding=None)

export policy to file.

Deprecated since version 1.6: applications should use CryptContext.to_string() instead.

Note

CryptPolicy are immutable. Use the replace() method to mutate existing instances.

Deprecated since version 1.6.