`records`#

Implements the data model for conda packages.

A PackageRecord is the record of a package present in a channel. A PackageCache is the record of a downloaded and cached package. A PrefixRecord is the record of a package installed into a conda environment.

Object inheritance:

Inheritance diagram of PackageRecord, PackageCacheRecord, PrefixRecord

Classes#

`LinkTypeField`	Fields are doing something very similar to boxing and unboxing
`NoarchField`	Fields are doing something very similar to boxing and unboxing
`TimestampField`	Fields are doing something very similar to boxing and unboxing
`Link`
`_FeaturesField`	Fields are doing something very similar to boxing and unboxing
`ChannelField`	Fields are doing something very similar to boxing and unboxing
`SubdirField`	Fields are doing something very similar to boxing and unboxing
`FilenameField`	Fields are doing something very similar to boxing and unboxing
`PackageTypeField`	Fields are doing something very similar to boxing and unboxing
`PathData`
`PathDataV1`
`PathsData`
`PackageRecord`	Representation of a concrete package archive (tarball or .conda file).
`Md5Field`	Fields are doing something very similar to boxing and unboxing
`PackageCacheRecord`	Representation of a package that has been downloaded or unpacked in the local package cache.
`SolvedRecord`	Representation of a package that has been returned as part of a solver solution.
`PrefixRecord`	Representation of a package that is installed in a local conda environmnet.

Attributes#

EMPTY_LINK

class LinkTypeField(enum_class, default=NULL, required=True, validation=None, in_dump=True, default_in_dump=True, nullable=False, immutable=False, aliases=())#

Bases: conda.auxlib.entity.EnumField

Fields are doing something very similar to boxing and unboxing of c#/java primitives. __set__ should take a "primitive" or "raw" value and create a "boxed" or "programmatically usable" value of it. While __get__ should return the boxed value, dump in turn should unbox the value into a primitive or raw value.

Parameters:

types (primitive literal or type or sequence of types)
default (any, callable, optional) -- If default is callable, it's guaranteed to return a valid value at the time of Entity creation.
required (boolean, optional)
validation (callable, optional)
dump (boolean, optional)

box(instance, instance_type, val)#

class NoarchField(enum_class, default=NULL, required=True, validation=None, in_dump=True, default_in_dump=True, nullable=False, immutable=False, aliases=())#

Bases: conda.auxlib.entity.EnumField

Fields are doing something very similar to boxing and unboxing of c#/java primitives. __set__ should take a "primitive" or "raw" value and create a "boxed" or "programmatically usable" value of it. While __get__ should return the boxed value, dump in turn should unbox the value into a primitive or raw value.

Parameters:

types (primitive literal or type or sequence of types)
default (any, callable, optional) -- If default is callable, it's guaranteed to return a valid value at the time of Entity creation.
required (boolean, optional)
validation (callable, optional)
dump (boolean, optional)

box(instance, instance_type, val)#

class TimestampField#

Bases: conda.auxlib.entity.NumberField

Fields are doing something very similar to boxing and unboxing of c#/java primitives. __set__ should take a "primitive" or "raw" value and create a "boxed" or "programmatically usable" value of it. While __get__ should return the boxed value, dump in turn should unbox the value into a primitive or raw value.

Parameters:

types (primitive literal or type or sequence of types)
default (any, callable, optional) -- If default is callable, it's guaranteed to return a valid value at the time of Entity creation.
required (boolean, optional)
validation (callable, optional)
dump (boolean, optional)

static _make_seconds(val)#

static _make_milliseconds(val)#

box(instance, instance_type, val)#

dump(instance, instance_type, val)#

__get__(instance, instance_type)#

class Link(**kwargs)#

Bases: conda.auxlib.entity.DictSafeMixin, conda.auxlib.entity.Entity

source#

type#

EMPTY_LINK#

class _FeaturesField(**kwargs)#

Bases: conda.auxlib.entity.ListField

Fields are doing something very similar to boxing and unboxing of c#/java primitives. __set__ should take a "primitive" or "raw" value and create a "boxed" or "programmatically usable" value of it. While __get__ should return the boxed value, dump in turn should unbox the value into a primitive or raw value.

Parameters:

types (primitive literal or type or sequence of types)
default (any, callable, optional) -- If default is callable, it's guaranteed to return a valid value at the time of Entity creation.
required (boolean, optional)
validation (callable, optional)
dump (boolean, optional)

box(instance, instance_type, val)#

dump(instance, instance_type, val)#

class ChannelField(aliases=())#

Bases: conda.auxlib.entity.ComposableField

Fields are doing something very similar to boxing and unboxing of c#/java primitives. __set__ should take a "primitive" or "raw" value and create a "boxed" or "programmatically usable" value of it. While __get__ should return the boxed value, dump in turn should unbox the value into a primitive or raw value.

Parameters:

types (primitive literal or type or sequence of types)
default (any, callable, optional) -- If default is callable, it's guaranteed to return a valid value at the time of Entity creation.
required (boolean, optional)
validation (callable, optional)
dump (boolean, optional)

dump(instance, instance_type, val)#

__get__(instance, instance_type)#

class SubdirField#

Bases: conda.auxlib.entity.StringField

Fields are doing something very similar to boxing and unboxing of c#/java primitives. __set__ should take a "primitive" or "raw" value and create a "boxed" or "programmatically usable" value of it. While __get__ should return the boxed value, dump in turn should unbox the value into a primitive or raw value.

Parameters:

types (primitive literal or type or sequence of types)
default (any, callable, optional) -- If default is callable, it's guaranteed to return a valid value at the time of Entity creation.
required (boolean, optional)
validation (callable, optional)
dump (boolean, optional)

__get__(instance, instance_type)#

class FilenameField(aliases=())#

Bases: conda.auxlib.entity.StringField

Fields are doing something very similar to boxing and unboxing of c#/java primitives. __set__ should take a "primitive" or "raw" value and create a "boxed" or "programmatically usable" value of it. While __get__ should return the boxed value, dump in turn should unbox the value into a primitive or raw value.

Parameters:

types (primitive literal or type or sequence of types)
default (any, callable, optional) -- If default is callable, it's guaranteed to return a valid value at the time of Entity creation.
required (boolean, optional)
validation (callable, optional)
dump (boolean, optional)

__get__(instance, instance_type)#

class PackageTypeField#

Bases: conda.auxlib.entity.EnumField

Fields are doing something very similar to boxing and unboxing of c#/java primitives. __set__ should take a "primitive" or "raw" value and create a "boxed" or "programmatically usable" value of it. While __get__ should return the boxed value, dump in turn should unbox the value into a primitive or raw value.

Parameters:

types (primitive literal or type or sequence of types)
default (any, callable, optional) -- If default is callable, it's guaranteed to return a valid value at the time of Entity creation.
required (boolean, optional)
validation (callable, optional)
dump (boolean, optional)

__get__(instance, instance_type)#

class PathData(**kwargs)#

Bases: conda.auxlib.entity.Entity

_path#

prefix_placeholder#

file_mode#

no_link#

path_type#

property path#

class PathDataV1(**kwargs)#

Bases: PathData

sha256#

size_in_bytes#

inode_paths#

sha256_in_prefix#

class PathsData(**kwargs)#

Bases: conda.auxlib.entity.Entity

paths_version#

paths#

class PackageRecord(*args, **kwargs)#

Bases: conda.auxlib.entity.DictSafeMixin, conda.auxlib.entity.Entity

Representation of a concrete package archive (tarball or .conda file).

It captures all the relevant information about a given package archive, including its source, in the following attributes.

Note that there are three subclasses, SolvedRecord, PrefixRecord and PackageCacheRecord. These capture the same information, but are augmented with additional information relevant for these sources of packages.

Further note that PackageRecord makes use of its _pkey for comparison and hash generation. This means that for common operations, like comparisons between PackageRecord s and reference of PackageRecord s in mappings, _different_ objects appear identical. The fields taken into account are marked in the following list of attributes. The subclasses do not add further attributes to the _pkey.

name#

The name of the package.

Part of the _pkey.

version#

The version of the package.

Part of the _pkey.

build#

The build string of the package.

Part of the _pkey.

build_number#

The build number of the package.

Part of the _pkey.

channel#: The channel where the package can be found.

subdir#

The subdir, i.e. noarch or a platform (linux-64 or similar).

Part of the _pkey.

fn#

The filename of the package.

Only part of the _pkey if separate_format_cache is true (default: false).

md5#: The md5 checksum of the package.

legacy_bz2_md5#: If this is a .conda package and a corresponding .tar.bz2 package exists, this may contain the md5 checksum of that package.

legacy_bz2_size#: If this is a .conda package and a corresponding .tar.bz2 package exists, this may contain the size of that package.

url#: The download url of the package.

sha256#: The sha256 checksum of the package.

property channel_name: str | None#

The canonical name of the channel of this package.

Part of the _pkey.

Type:: str

property _pkey#

The components of the PackageRecord that are used for comparison and hashing.

The _pkey is a tuple made up of the following fields of the PackageRecord. Two PackageRecord s test equal if their respective _pkey s are equal. The hash of the PackageRecord (important for dictionary access) is the hash of the _pkey.

The included fields are:

channel_name
subdir
name
version
build_number
build
fn only if separate_format_cache is set to true (default: false)

Type:: tuple

__hash__()#

__eq__(other)#

dist_str(canonical_name: bool = True) → str#

dist_fields_dump()#

arch#

platform#

depends#

constrains#

track_features#

features#

noarch#

preferred_env#

python_site_packages_path#

license#

license_family#

package_type#

property is_unmanageable#

timestamp#

property combined_depends#

date#

size#

__str__()#

to_match_spec()#

to_simple_match_spec()#

property namekey#

property spec#

name=version=build

Type:: Return package spec

property spec_no_build#

name=version

Type:: Return package spec without build

record_id()#

metadata: set[str]#

classmethod feature(feature_name) → PackageRecord#

classmethod virtual_package(name: str, version: str | None = None, build_string: str | None = None) → PackageRecord#

Create a virtual package record.

Parameters:

name -- The name of the virtual package.
version -- The version of the virtual package, defaults to "0".
build_string -- The build string of the virtual package, defaults to "0".

Returns:

A PackageRecord representing the virtual package.

class Md5Field#

Bases: conda.auxlib.entity.StringField

Fields are doing something very similar to boxing and unboxing of c#/java primitives. __set__ should take a "primitive" or "raw" value and create a "boxed" or "programmatically usable" value of it. While __get__ should return the boxed value, dump in turn should unbox the value into a primitive or raw value.

Parameters:

types (primitive literal or type or sequence of types)
default (any, callable, optional) -- If default is callable, it's guaranteed to return a valid value at the time of Entity creation.
required (boolean, optional)
validation (callable, optional)
dump (boolean, optional)

__get__(instance, instance_type)#

class PackageCacheRecord(*args, **kwargs)#

Bases: PackageRecord

Representation of a package that has been downloaded or unpacked in the local package cache.

Specialization of PackageRecord that adds information for packages that exist in the local package cache, either as the downloaded package file, or unpacked in its own package dir, or both.

Note that this class does not add new fields to the PackageRecord._pkey so that a pure PackageRecord object that has the same _pkey fields as a different PackageCacheRecord object (or, indeed, a PrefixRecord object) will be considered equal and will produce the same hash.

package_tarball_full_path#: Full path to the local package file.

extracted_package_dir#: Full path to the local extracted package.

md5#

The md5 checksum of the package.

If the package file exists locally, this class can calculate a missing checksum on-the-fly.

property is_fetched#

Whether the package file exists locally.

Type:: bool

property is_extracted#

Whether the package has been extracted locally.

Type:: bool

property tarball_basename#

The basename of the local package file.

Type:: str

_calculate_md5sum()#

class SolvedRecord(*args, **kwargs)#

Bases: PackageRecord

Representation of a package that has been returned as part of a solver solution.

This sits between PackageRecord and PrefixRecord, simply adding requested_spec (and requested_specs) so they can be used in lockfiles without requiring the artifact on disk.

requested_spec#: The MatchSpec that the user requested or None if the package it was installed as a dependency.

_requested_specs#: The MatchSpec objects that the user requested or () if the package was installed as a dependency. See also .requested_specs property.

property requested_specs: tuple[str, Ellipsis]#: This property will use 'requested_spec' as the source for 'requested_specs' for old conda-meta/*.json files that did not define the plural version.

dump()#: We need to expose _requested_specs as its public counterpart, and also expose single specs as a plural list for backwards compatibility.

class PrefixRecord(*args, **kwargs)#

Bases: SolvedRecord

Representation of a package that is installed in a local conda environmnet.

Specialization of PackageRecord that adds information for packages that are installed in a local conda environment or prefix.

Note that this class does not add new fields to the PackageRecord._pkey so that a pure PackageRecord object that has the same _pkey fields as a different PrefixRecord object (or, indeed, a PackageCacheRecord object) will be considered equal and will produce the same hash.

Objects of this class are generally constructed from metadata in json files inside $prefix/conda-meta.

package_tarball_full_path#: The path to the originating package file, usually in the local cache.

extracted_package_dir#: The path to the extracted package directory, usually in the local cache.

files#: The list of all files comprising the package as relative paths from the prefix root.

paths_data#: List with additional information about the files, e.g. checksums and link type.

link#: Information about how the package was linked into the prefix.

auth#: Authentication information.

package_size(prefix_path: pathlib.Path) → int#

Compute the installed size of this package within a prefix.

This sums up the size_in_bytes of all non-softlink paths in paths_data, and stats the files on disk if size_in_bytes is missing and for the package's conda-meta JSON manifest.

Returns:: Total size in bytes of all files installed by this package.

records#

Classes#

Attributes#

`records`#