Queries and Subjects

class dnf.query.Query

Facilitates lookup of packages in a Sack based on given criteria. Query actually does not consult the information in the Sack until it is evaluated. The evaluation happens either explicitly using run() or by iterating the query, for example:

#!/usr/bin/python3
import dnf

base = dnf.Base()
base.fill_sack()

q = base.sack.query()
i = q.installed()
i = i.filter(name='dnf')

packages = list(i)  # i only gets evaluated here
print("Installed dnf package:")
for pkg in packages:
    print(pkg, pkg.reponame)

or:

#!/usr/bin/python3
import dnf

base = dnf.Base()
base.read_all_repos()
base.fill_sack()

q = base.sack.query()
a = q.available()
a = a.filter(name='dnf')

print("Available dnf packages:")
for pkg in a:  # a only gets evaluated here
    print('{} in repo {}'.format(pkg, pkg.reponame))

Notice that none of the filtering methods mutates the state of the Query but produces a new object instead.

available(): Returns a new query limiting the original query to the packages available from the repositories.

difference(other): Returns a new query that contains only those results of original query that are not in the results of the other query.

downgrades(): Returns a new query that limits the result only to packages that can be downgrade candidates to other packages in the current set. Downgrade candidate has the same name, lower EVR and the architecture of the original and the downgrade candidate are suitable for a downgrade. Specifically, the filtering does not take any steps to establish that the downgrade candidate can actually be installed.

duplicated(): Returns a new query that limits the result only to installed packages of same name and different version. Optional argument exclude accepts a list of package names that will be excluded from result.

extras(): Returns a new query that limits the result to installed packages that are not present in any repo

filter(\*\*kwargs)

Returns a new query limiting the original query to the key/value pairs from kwargs. Multiple kwargs can be passed, the filter then works by applying all of them together (logical AND). Values inside of list or query are cumulative (logical OR).

Allowed keys are:

key	value type	value meaning
arch	string	match against packages’ architecture
downgrades	boolean	see `downgrades()`. Defaults to `False`.
empty	boolean	`True` limits to empty result set. Defaults to `False`.
epoch	integer	match against packages’ epoch.
file	string	match against packages’ files
latest	integer	limit to all packages of number of versions
latest_per_arch	integer	see `latest()`.
name	string	match against packages’ names
release	string	match against packages’ releases
reponame	string	match against packages repositories’ names
version	string	match against packages’ versions
pkg	Query	match against packages in query
pkg*	list	match against hawkey.Packages in list
provides	string	match against packages’ provides
provides*	Hawkey.Reldep	match against packages’ provides
<DEP>	string	match against packages’ <DEP>
<DEP>*	Hawkey.Reldep	match a reldep against packages’ <DEP>
<DEP>*	Query	match the result of a query against packages’ <DEP>
<DEP>*	list(Package)	match the list of hawkey.Packages against packages’ <DEP>
sourcerpm	string	match against packages’ source rpm
upgrades	boolean	see `upgrades()`. Defaults to `False`.

<DEP> can be any of: requires, conflicts, obsoletes, enhances, recomments, suggests, supplements

* The key can also accept a list of values with specified type.

The key name can be supplemented with a relation-specifying suffix, separated by __:

key suffix	value type	semantics
eq	any	exact match; This is the default if no suffix is specified.
glob	string	shell-style wildcard match
gt	integer	the actual value is greater than specified
gte	integer	the actual value is greater than or equal to specified
lt	integer	the actual value is less than specified
lte	integer	the actual value is less than or equal to specified
neq	any	does not equal
substr	string	the specified value is contained in the actual value
eqg	string	exact match or the first higher, used with advisory filters
upgrade	string	skips advisory resolved by installed packages

For example, the following creates a query that matches all packages containing the string “club” in its name:

q = base.sack.query().filter(name__substr="club")

Note that using packages or queries for dependency filtering performs a more advanced resolution than using a string or a reldep. When a package list or a query is used, rich dependencies are resolved in a more precise way than what is possible when a string or a reldep is used.

filterm(\*\*kwargs): Similar to dnf.query.Query.filter() but it modifies the query in place.

installed(): Returns a new query that limits the result to the installed packages only.

intersection(other): Returns a new query where the result contains only packages that are found in both original and other queries.

latest(limit=1): Returns a new query that limits the result to limit highest version of packages per package name and per architecture. In case the limit is negative number, it excludes the number of latest versions according to limit.

run(): Evaluate the query. Returns a list of matching dnf.package.Package instances.

union(other): Returns a new query where the results of the other query are added to the results of the original query.

upgrades(): Returns a new query that limits the result only to packages that can be upgrade candidates to at least one package in the current set. Upgrade candidate has the same name, higher EVR and the architectures of the original and the upgrade candidate package are suitable for an upgrade. Specifically, the filtering does not take any steps to establish that the upgrade candidate can actually be installed.

class dnf.subject.Subject

As explained on the DNF man page, users of the CLI are able to select packages for an operation in different formats, leaving seemingly arbitrary parts out of the spec and even using globbing characters. This class implements a common approach to parsing such input and produce a Query listing all packages matching the input or a Selector selecting a single package that best matches the input given a transaction operation.

__init__(pkg_spec, ignore_case=False): Initialize the Subject with pkg_spec input string with following semantic. If ignore_case is True ignore the case of characters in pkg_spec.

get_best_query(sack, with_nevra=True, with_provides=True, with_filenames=True, forms=None): Returns a Query yielding packages matching the given input. The result of the returned query can be an empty set if no package matches. sack is the Sack that the returned query will search. with_nevra enable search by nevra, with_provides indicates whether besides package names also packages’ provides are searched for a match, and with_filenames indicates whether besides package provides also packages’ file provides are searched for a match. forms is a list of pattern forms from hawkey. Leaving the parameter to None results in using a reasonable default list of forms.

get_best_selector(sack, forms=None, obsoletes=True, reponame=None): Returns a Selector that will select a single best-matching package when used in a transaction operation. sack and forms have the same meaning as in get_best_query(). If obsoletes, selector will also contain packages that obsoletes requested packages (default is True). If reponame, the selection of available packages is limited to packages from that repo (default is None).

get_nevra_possibilities(self, forms=None)

Returns generator for every possible nevra. Each possible nevra is represented by NEVRA class (libdnf) that has attributes name, epoch, version, release, arch. forms have the same meaning as in get_best_query().

Example how to use it when it is known that string could be full NEVRA or NEVR:

#!/usr/bin/python3
import dnf
import hawkey

nevra_string = "dnf-0:4.2.2-2.fc30.noarch"
subject = dnf.subject.Subject(nevra_string)
possible_nevra = subject.get_nevra_possibilities(
    forms=[hawkey.FORM_NEVRA, hawkey.FORM_NEVR])

for i,nevra in enumerate(possible_nevra):
    print("Possibility {} for \"{}\":".format(i+1, nevra_string))
    print("name: {}".format(nevra.name))
    print("epoch: {}".format(nevra.epoch))
    print("version: {}".format(nevra.version))
    print("release: {}".format(nevra.release))
    print("architecture: {}".format(nevra.arch))
    print()