Patterns Specification

Description

Different pattern matching rules for arguments in *-spec form are defined for operation with each entity type in DNF5. In this section all rules are described and explained in detail, with examples.

Globs

Pattern specification supports the same glob pattern matching that shell does. The following patterns are supported:

*: Matches any number of characters.
?: Matches any single character.
[]: Matches any one of the enclosed characters. A pair of characters separated by a hyphen denotes a range expression; any character that falls between those two characters, inclusive, is matched. If the first character following the [ is a ! or a ^ then any character not enclosed is matched.

Note: Curly brackets ({}) are not supported. You can still use them in shells that support them and let the shell do the expansion, but if quoted or escaped, DNF5 will not expand them.

Packages

Many commands take a <package-spec*> parameter that selects a package for the operation. Depending on the command the <package-spec*> argument is matched against:

NEVRA [N] or case Insentive NEVRA [NI]
Provides [P]
File provides [F]
Binaries [B]

sequentially until a match is found.

For example <package-spec-NP> is first matched againts NEVRAs and if no matching NEVRA is found it is matched againts Provides.

When <package-spec> is a package name or a provide, the user can provide additional restriction rules for matching the arguments. Basic version comparisons can be used for this purpose (=, >, <, >=, <=), like this <package-spec> >= <version>, where the <version> argument is in a [EPOCH:]VERSION[-RELEASE] format as specified in the NEVRA matching section.

To build more complex expressions, a rich dependency feature is also supported, which is always enclosed in parentheses. Boolean operators and nesting can be used, f.e. (<spec1> or (<spec2> and <spec3>)). For more information, please see RPM boolean dependencies.

NEVRA matching

Each package can be uniquely identified by the NEVRA string. It consists of 5 parts of information:

Name: Package name.
Epoch: Epoch number.

It is not always included.

The epoch number overrides other version checking, so it can be used to force the package upgrade over some other one.
Version: Version string.

It is not strictly numeric.

It is intended to match the upstream software version.
Release: Edition string.

It is an information about the particular package build, usually a number increased with the newer build. It is not connected with the upstream software.
Architecture: Target architecture string.

Defines the processor type the package is intended to be installed on.

It can be also a package containing source files (src) or architecture-independent package (noarch).

When matching against NEVRAs, partial matching is supported. DNF5 tries to match the spec against the following list of NEVRA forms (in decreasing order of priority):

NAME-[EPOCH:]VERSION-RELEASE.ARCH

NAME.ARCH

NAME

NAME-[EPOCH:]VERSION-RELEASE

NAME-[EPOCH:]VERSION

Note that name can in general contain dashes (e.g. package-with-dashes).

The first form that matches any packages is used and the remaining forms are not tried. If none of the forms match any packages, an attempt is made to match the <package-spec> against full package NEVRAs. This is only relevant if globs are present in the <package-spec>.

You can specify globs as part of any of the five NEVRA components. You can also specify a glob pattern to match over multiple NEVRA components (in other words, to match across the NEVRA separators). In that case, however, you need to write the spec to match against full package NEVRAs, as it is not possible to split such spec into NEVRA forms.

File Provides matching

If a spec starts with either / or */, it is considered as a potential file provide.

Binaries matching

Whether to consider package binaries during matching. Checks if given spec is a binary in /usr/bin/ or /usr/sbin/.

Comps

<group-spec>``|``<environment-spec> allows one to select groups|environments a particular operation should work on. It is a case insensitive string (supporting globbing characters) that is matched against a group’s|environment’s ID, canonical name and name translated into the current LC_MESSAGES locale (if possible).

Comps specs are prefixed by @ for commands that also accept package specs. For example see dnf5-install(8), Install Command.

Modules

<module-spec> allows one to select modules or profiles a particular operation should work on.

Since NEVRA matching form is insufficient for modules, they are uniquely identified by the NSVCA format (NAME:STREAM:VERSION:CONTEXT:ARCH/PROFILE). Supported partial forms are the following:

NAME

NAME:STREAM

NAME:STREAM:VERSION

NAME:STREAM:VERSION:CONTEXT

All above combinations with ::ARCH (e.g. NAME::ARCH)

NAME:STREAM:VERSION:CONTEXT:ARCH

All above combinations with /PROFILE (e.g. NAME/PROFILE)

In case stream is not specified, the enabled or the default stream is used, in this order. In case profile is not specified, the system default profile or the ‘default’ profile is used.

Transactions

<transaction-spec> can be in one of several forms. If it is an integer, it specifies a transaction ID. Specifying last is the same as specifying the ID of the most recent transaction. The last form is last-<offset>, where <offset> is a positive integer. It specifies offset-th transaction preceding the most recent transaction.