path: root/docs/cpt.txt

			_______________________

			 CARBS PACKAGING TOOLS
			      User Manual

			       Cem Keylan
			_______________________


Table of Contents
_________________

1. Copying
2. Preface
3. Usage
4. Configuration
.. 1. Configuration directory
..... 1. CPT Base
..... 2. Systemwide hooks
.. 2. Environment Variables
..... 1. `CPT_PATH'
..... 2. `CPT_COMPRESS'
..... 3. `CPT_FORCE'
..... 4. `CPT_PID'
.. 3. Hooks
.. 4. Editing the build file during pre-build
5. Packaging System
.. 1. build
.. 2. sources
.. 3. checksums
.. 4. version
.. 5. depends
.. 6. meta
.. 7. post-install
.. 8. message
.. 9. test
6. Package Repositories
.. 1. Rsync Repositories
..... 1. Setting up an Rsync repository
.. 2. Fossil repositories
.. 3. Message of the Day
7. Comparison Between CPT and KISS
8. CPT Library
.. 1. Calling the library
.. 2. Variables
.. 3. Option parsing
..... 1. Defining a parser
..... 2. `global_options()'
.. 4. Message functions
..... 1. `out()'
..... 2. `log()'
..... 3. `die()'
..... 4. `warn()'
..... 5. `prompt()'
.. 5. Text functions
..... 1. `contains()'
..... 2. `regesc()'
..... 3. `pop()'
..... 4. `sepchar()'
.. 6. Portability functions
..... 1. `_seq()'
..... 2. `_stat()'
..... 3. `_readlinkf()'
.. 7. System Functions
..... 1. `as_root()'
.. 8. Package Functions
..... 1. `pkg_build()'
..... 2. `pkg_depends()'
..... 3. `pkg_order()'
..... 4. `pkg_owner()'
..... 5. `pkg_isbuilt()'
..... 6. `pkg_lint()'
..... 7. `pkg_find()'
..... 8. `pkg_get_base()'
..... 9. `pkg_gentree()'
..... 10. `pkg_query_meta()'


This is a reference document containing both the user-guide and the
development manual for *Carbs Packaging Tools* version Fossil. For
development logs see [the fossil repository].


[the fossil repository] <https://fossil.carbslinux.org/cpt>


1 Copying
=========

  Copyright (c) 2020-2021 Cem Keylan

        Permission is granted to copy, distribute and/or modify
        this document under the terms of the GNU Free
        Documentation License, Version 1.3 or any later version
        published by the Free Software Foundation; with no
        Invariant Sections, with no Front-Cover Texts and no
        Back-Cover Texts. A copy of the license is included in the
        section entitled "GNU Free Documentation License."


2 Preface
=========

  Carbs Linux uses its own package management toolchain named `cpt'
  which was initially forked from the [kiss] package manager. Unlike
  `kiss', however, its main goal is being easily extendable. Instead of
  being a single file package manager, it revolves around the shell
  library `cpt-lib', and many tools that wrap around it. This document
  aims to document both the usage of the distributed tools and document
  the library functions.

  If you happen to find something that is not properly covered by the
  documentation, or an area that can be improved, please feel free to
  submit a patch, or [open a ticket].


[kiss] <https://github.com/kisslinux/kiss>

[open a ticket] <https://fossil.carbslinux.org/cpt/tktnew>


3 Usage
=======

  `cpt' is formed of many tools combined in a single environment,
  similar to `git'. When you run `cpt' without any arguments, it will
  show all available tools and their explanations. Here is an example
  call with extra scripts on my system:

  ,----
  | -> Carbs Packaging Tool
  | -> add               Commit the current directory as a new package
  | -> alternatives      List and swap to alternatives
  | -> build             Build a package
  | -> bump              Commit the current directory as a version bump
  | -> cargo-urlgen      Create static cargo sources for Rust packages
  | -> cargolock-urlgen  Convert the given Cargo.lock file to sources
  | -> cat               Concatanate package files in the installed package database
  | -> changelog         Print the git log of the specific package
  | -> chbuild           Create/destroy temporary chroots
  | -> checkmissing      Verify package manifests
  | -> checksum          Generate checksums
  | -> chroot            Enter a chroot
  | -> commit            Commit a package without the prefix of 'package:'
  | -> depends           Display a package's dependencies
  | -> download          Download sources for the given package
  | -> exec              Execute a command inside the alternatives system
  | -> export            Turn an installed package into a CPT tarball
  | -> fork              Fork a package to the current directory
  | -> getchoice         Prints the full path to a file in the alternatives system.
  | -> install           Install a package
  | -> link              Link a forked package's files to the other repository
  | -> list              List installed packages
  | -> maintainer        Find the maintainer of a package
  | -> manifest          Display all files owned by a package
  | -> manifest-tree     Display all files owned by a package with a tree view
  | -> new               Create a boilerplate CPT package
  | -> orphans           List orphaned packages
  | -> owns              Check which package owns a file
  | -> rel               Bump the release number of a package
  | -> remove            Remove a package
  | -> repodepends       Display a package's dependencies in the repository
  | -> reporevdepends    Display packages on the repository which depend on package
  | -> reset             Remove all packages except for the base
  | -> revdepends        Display packages which depend on package
  | -> search            Search for a package
  | -> size              Show the size on disk for a package
  | -> source            Extract sources of a given package to the current directory
  | -> update            Check for updates
  `----

  The documentation of `cpt' aims to keep tool flags and related usage
  information on concise manual pages, while moving the rest of the
  information to this User Manual to avoid unwarranted duplication. To
  learn more on a specific usage of a tool, use the `man' program:

  ,----
  | man cpt-build
  `----


4 Configuration
===============

  The package manager does *NOT* have a configuration file, but there
  are a variety of ways in order to interact with and configure the
  package manager.


4.1 Configuration directory
~~~~~~~~~~~~~~~~~~~~~~~~~~~

  Some features of the package manager can be configured from the files
  found under `/etc/cpt/'. Even though this doesn't sound like the
  premise of "no configuration" files, these files are completely
  optional to the package manager, and still the majority of
  configuration is done through environment variables. The files on this
  directory are for configuration that don't have a big impact on how
  the package manager behaves, and are not feasible to be used inside
  simple environment variables (such as the base package list and
  package manager hooks).


4.1.1 CPT Base
--------------

  The file `/etc/cpt/base' can be used in order to define the base to
  the package manager. Base packages are the packages that receive
  special treatment by utilities such as `cpt-reset', and `cpt-orphans'.

  ,----
  | # This file defines the base packages of the system. You can add or remove
  | # package names in order to redefine the base. This file will be used by
  | # cpt-orphans and cpt-reset. If this file doesn't exist on /etc/cpt/base, both
  | # of the tools will assume that there is no defined base, so use with caution.
  | baselayout
  | binutils
  | byacc
  | busybox
  | bzip2
  | ca-certificates
  | curl
  | flex
  | gcc
  | git
  | rsync
  | gzip
  | cpt
  | bearssl
  | linux-headers
  | m4
  | make
  | musl
  | pkgconf
  | xz
  | zlib
  `----


4.1.2 Systemwide hooks
----------------------

  A collection of hooks can be installed under `/etc/cpt/hooks/'. All of
  the files installed under this directory will then be sourced by the
  package manager whenever a hook is called. Some examples for system
  hooks can be found under the `/usr/share/cpt/examples/hooks/'
  directory.


4.2 Environment Variables
~~~~~~~~~~~~~~~~~~~~~~~~~

  Since there is no configuration file for `cpt', the package manager is
  configured through environment variables. These can be set per
  operation, or be set to your shell configuration or `~/.profile'. Here
  are the environment variables that alter the behaviour of `cpt', some
  of them have separate sections to provide detailed information.

  `CPT_PATH'
        Set the locations of your repositories. It is similar to the
        `PATH' variable.

  `CPT_CACHE' 
        The cache directory for `cpt'. Default: `$XDG_CACHE_HOME/cpt'.

  `CPT_CHOICE'
        If this is set to 0, a package installation will be aborted on
        conflicts.

  `CPT_COLOR'
        If this is set to 1, `cpt' tools will be forced to display
        coloured output. If set to 0, they will be forced to display
        them without colours. Otherwise, `cpt' will output colour as
        long as it is outputting to a terminal.

  `CPT_DEBUG'
        If set to 1, temporary directories will not be removed after the
        operation.

  `CPT_DOWNLOADER'
        The tool to be used to download package sources. One of `curl',
        `wget', `wget2', `axel', `aria2c'. Defaults to the first one
        found in that order.

  `CPT_FETCH'
        If set to 0, `cpt-update' will not fetch repositories.

  `CPT_FORCE'
        If set to 1, `cpt' tools will force operation.

  `CPT_HOOK'
        Absolute path to the package manager hook file.

  `CPT_KEEPLOG'
        If set to 1, `cpt' will keep logs regardless of operation
        success.

  `CPT_PID'
        Set the temporary build directory name.

  `CPT_PROMPT'
        If set to 0, `cpt' will not prompt you for anything.

  `CPT_REPO_CACHE'
        If set to 0, `cpt' will not use or write repository information
        cache.

  `CPT_ROOT'
        If this variable is set, `cpt' will assume the given path as the
        system root.

  `CPT_TEST'
        If set to 1, `cpt-build' will run tests whenever available.

  `CPT_TMPDIR'
        The directory to create the temporary directories.

  `CPT_VERBOSE'
        If this variable is set to 1, the package manager will print
        more information.


4.2.1 `CPT_PATH'
----------------

  Similar to the `PATH' variable, `cpt' find repositories from the
  `CPT_PATH' variable. Here is an example:

  ,----
  | CPT_PATH=$HOME/repos/repo1:$HOME/repos/repo2:$HOME/repos/repo3
  `----

  This is a simplistic and a structured example for repository
  locations, but it doesn't necessarily need to be as tidy as the
  example above. Here is an example for something a little more complex.

  ,----
  | CPT_PATH=$HOME/repos/overrides:/var/db/cpt/repo/core:/var/db/cpt/repo/extra:$HOME/repos/personal
  `----

  This example brings us to the next section of this document.


* 4.2.1.1 Repository preferences

  When you are using multiple repositories from multiple vendors, you
  will find out that some repositories have the same packages. `cpt'
  doesn't care about conflicting packages. If you want to build a
  package that exists on multiple repositories, `cpt' will build the
  first matching package. This means that if `grep' package (for the
  sake of an example) exists on both `$HOME/repos/personal' and
  `$HOME/repos/carbs/extra', and you want to install from your personal
  repository, you must set `CPT_PATH' so that your personal repository
  is listed before the `extra' repository.

  ,----
  | CPT_PATH=$HOME/repos/personal:$HOME/repos/carbs/extra
  `----


* 4.2.1.2 Setting the `CPT_PATH'

  You can set the `CPT_PATH' variable on your shell configuration or
  your `.profile' file in a way that is easy to read.

  The below example sets `CPT_PATH' in a way that is easy to understand
  which repository comes first:

  ,----
  | CPT_PATH=$HOME/repos/overrides
  | CPT_PATH=$CPT_PATH:$HOME/repos/carbs/core
  | CPT_PATH=$CPT_PATH:$HOME/repos/carbs/extra
  | CPT_PATH=$CPT_PATH:$HOME/repos/carbs/xorg
  | CPT_PATH=$CPT_PATH:$HOME/repos/personal
  | export CPT_PATH
  `----


4.2.2 `CPT_COMPRESS'
--------------------

  When setting the `CPT_COMPRESS' value, you should set the name of the
  default suffixes for the program. Available values are:

  - `gz'
  - `zst'
  - `bz2'
  - `xz'
  - `lz'

  Defaults to `gz'.


4.2.3 `CPT_FORCE'
-----------------

  If this is set to 1, some of the `cpt' tools will continue regardless
  of errors or skip certain checks. Here are some examples:

  - `cpt-install' will install a package without verifying its manifest.
  - `cpt-install' will install a package even when there are missing
    dependencies.
  - `cpt-remove' will remove packages even when there are other packages
    that depend on the current package.

  Defaults to 0.


4.2.4 `CPT_PID'
---------------

  If this variable is set, the temporary files will be created with this
  variable as the suffix, instead of the PID of the `cpt' process. The
  advantage is that you can know exactly where the build directory is
  located, while the disadvantage is that there will be issues with
  multiple operations at the same time. So the best way to use this
  variable is during one-time `cpt' calls.

  ,----
  | CPT_PID=mesa cpt b mesa
  `----

  By running the above, you will know that the created build directories
  will end with the `*-mesa' suffix.


4.3 Hooks
~~~~~~~~~

  Hooks can be used in order to change the runtime behaviour of the
  package manager.  There are a variety of package hooks, mostly self
  explanatory:

  pre-build   
        Run just before the `build' script is run
  post-build  
        Run after the `build' script is run successfully
  build-fail  
        Run if the `build' script fails
  pre-test    
        Run before the `test' script is run
  test-fail   
        Run if the `test' script fails
  pre-install 
        Run before a package is installed for each package
  post-install
        Run after a package is installed for each package
  end-install 
        Run after all given packages are installed
  pre-remove  
        Run before a package is removed for each package
  post-remove 
        Run after a package is removed for each package
  end-remove  
        Run after all given packages are removed
  pre-fetch   
        Run before all repositories are fetched
  post-fetch  
        Run after all repositories are fetched
  post-package
        Run after a tarball for a package is created

  In order to use hooks, you will need to set the `CPT_HOOK' variable
  pointing to your hook file. Your hook file *MUST* be a POSIX shell
  script as its contents are sourced by the package manager.

  The hook is given 3 variables when it is executed. Those are:

  `$TYPE'
        The type of the hook, (`pre-build', `post-build', etc.)
  `$PKG' 
        The package that `cpt' is currently working on. Can be null.
  `$DEST'
        The destination of the operation. Can be null.


4.4 Editing the build file during pre-build
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

  You can edit the `build' file during pre-build. The file is copied
  from the repository to the build directory named as `.build.cpt'. You
  can use `sed' or any other tool to edit the build file. After the
  build is complete, a `diff' file will be placed to the package
  database named as `build.diff'. Here is an example `build' file
  manipulation during the pre-build hook.

  ,----
  | cat <<EOF> .build.cpt
  | #!/bin/sh -e
  | 
  | for patch in bash50-0??; do
  |     patch -p0 < "\$patch"
  | done
  | 
  | export LDFLAGS=-static
  | 
  | ./configure \
  |     --prefix=/usr \
  |     --without-bash-malloc \
  |     --disable-nls
  | 
  | export MAKEFLAGS="TERMCAP_LIB=/usr/lib/libncursesw.a $MAKEFLAGS"
  | 
  | make
  | make DESTDIR="\$1" install
  | 
  | ln -s bash "\$1/usr/bin/sh"
  | EOF
  `----


5 Packaging System
==================

  A package is a directory formed of several files, from these files,
  only `build', `checksums', and `version' files are mandatory.

  This section talks about files that are interpreted specially by the
  package manager. Any other file can be added to the package directory
  at the discretion of the package maintainer. Everything in the package
  directory will also be added to the package database that is located
  on `/var/db/cpt/installed'. These can be patches, configuration files,
  etc.

  Below is a table that provides a small summary for each file, see the
  relevant section to learn detailed information on each of them.

   File          Language                         Executable  Mandatory 
  ----------------------------------------------------------------------
   build         any                              yes         yes       
   checksums     generated by `cpt-checksum'      no          no        
   meta          key-value pairs as in RFC822[1]  no          no[2]     
   depends       custom format                    no          no        
   sources       custom format                    no          no        
   version       custom format                    no          yes       
   message       plaintext                        no          no        
   post-install  any                              yes         no        
   test          any                              yes         no        


5.1 build
~~~~~~~~~

  Typically `build' files are shell scripts that run commands to prepare
  the source code to be installed on the target system. Even though we
  will be assuming that the `build' file is a POSIX shell script (for
  portability's sake), `build' files can be any executable program from
  binary programs to `perl' scripts.

  The contents of a build script do not need to follow a certain rule
  for the package manager, except for the fact that the user needs the
  permission to execute the file.

  An important advice is to append an '-e' to the shebang (#!/bin/sh -e)
  so that the build script exits on compilation error.

  Build is run with three arguments (`$#')

  - Location of the package directory (DESTDIR)
  - Package version
  - System architecture


5.2 sources
~~~~~~~~~~~

  `sources' file is a list of files and sources that will be put to the
  build directory during the build process. Those can be remote sources
  (such as tarballs), git repositories, and files that reside on the
  package directory.

  The syntax is pretty simple for the `soures' file; `src dest'. The
  `dest' parameter is optional. It is the directory that the source will
  be placed in.  Here is the `sources' file for the `gst-plugins'
  package:

  ,----
  | https://gstreamer.freedesktop.org/src/gst-plugins-good/gst-plugins-good-1.16.2.tar.xz good
  | https://gstreamer.freedesktop.org/src/gst-plugins-bad/gst-plugins-bad-1.16.2.tar.xz   bad
  | https://gstreamer.freedesktop.org/src/gst-plugins-ugly/gst-plugins-ugly-1.16.2.tar.xz ugly
  | https://gstreamer.freedesktop.org/src/gst-libav/gst-libav-1.16.2.tar.xz               libav
  `----

  This file is read from the package manager as space seperated. Files
  that begin with a `#' comment are ignored. The first value points to
  the location of the source.

  If it starts with a protcol url, (such as <ftp://> <http://>
  <https://>) it will be downloaded with `curl'.

  If the source is a git repository, it shall be prefixed with a `git+'
  git(1) will be used to do a shallow clone of the repository. If the
  commit is suffixed by a history pointer, git will checkout the
  relevant revision. So,

  `git+git://example.com/pub/repo@v1.2.3'
        will checkout the tag named "v1.2.3"
  `git+git://example.com/pub/repo#development'
        will checkout the branch named "development"
  `git+git://example.com/pub/repo#1a314s87'
        will checkout the commit named "1a314s87"

  Other files are assumed to be residing in the package directory. They
  should be added with their paths relative to the package directory.


5.3 checksums
~~~~~~~~~~~~~

  `checksums' file is generated by the `cpt c pkg' command. It is
  generated according to the order of the sources file. That's why you
  shouldn't be editing it manually. The checksums file is created with
  the digests of the files using the sha256 algorithm.


5.4 version
~~~~~~~~~~~

  The version file includes the version of the software and the release
  number of of the package on a space seperated format. The contents of
  the file should look like below.

  ,----
  | 1.3.2 1
  `----


5.5 depends
~~~~~~~~~~~

  This is a list of dependencies that must be installed before a package
  build. You can append "make" after a dependency to mark a package is
  only required during the build process of a package. Packages marked
  as a make dependency can be removed after the build. There are also
  "test" dependencies. These dependencies are only installed if either
  the `CPT_TEST' is set to 1, or the build is run with the `-t' or
  `--test' options. So, a package package could have the following
  `depends' file:

  ,----
  | linux-headers make
  | python        test
  | zlib
  `----


5.6 meta
~~~~~~~~

  `meta' is a non-mandatory package file that can be used to provide
  information otherwise non-relevant to the functions of the package
  manager. This file can later be queried with the [pkg_query_meta()]
  function. The file has a simple markup format, it must adhere to the
  `KEY: VAL' format. An example for the `cpt' package would be as
  follows:

  ,----
  | description: Carbs Packaging Tools
  | license: MIT
  | maintainer: Linux User <linux-user@example.com>
  `----

  Even though `meta' is not mandatory by the packaging system, it is a
  mandatory file for submitting packages to Carbs Linux repositories.


[pkg_query_meta()] See section 8.8.10


5.7 post-install
~~~~~~~~~~~~~~~~

  `post-install' files have the same requirements as the build
  script. They will be run after the package is installed as root (or as
  the user if the user has write permissions on `CPT_ROOT').


5.8 message
~~~~~~~~~~~

  This plaintext file will be outputted with `cat' after every package
  is installed.


5.9 test
~~~~~~~~

  Test files are mainly for the repository maintainer to test the
  packages, and will only run if the user has the `CPT_TEST' variable
  set, or the build is run with the `-t' or `--test' options. This
  script is run on the build directory. It is run right after the build
  script is finished.


6 Package Repositories
======================

  *cpt* has backends to support the use of a variety of distribution
  methods. You can currently use Git, Mercurial, Fossil, and Rsync to
  distribute a package repository. That, however, does not mean that you
  need to setup either of those, if you are simply going for a local
  repository on your system.

  In the broad sense, a package repository is any directory that
  contains packages that were described in 5. This means that as long as
  you can serve them, there is not much needed to do in order to
  distribute a repository.  The following subsections aim to detail the
  notes and the caveats of certain distribution methods.


6.1 Rsync Repositories
~~~~~~~~~~~~~~~~~~~~~~

  Rsync repositories are simple to serve and simple to use. In the
  repository directory, there needs to be a `.rsync' file that points to
  the remote of the repository. This is used in order to fetch changes
  from the upstream. `.rsync' file looks like this for the core
  repository:

  ,----
  | rsync://carbslinux.org/repo/core
  `----

  Rsync repositories have some few distinctions when it comes to
  fetching them.  They can be either synced individually or as a
  "root". There are 2 important files, those are `.rsync' and
  `.rsync_root'. Here is the Carbs Linux rsync repository structure.

  ,----
  |            /
  |    -----------------
  |   |                |
  | .rsync           core/
  |           ----------------
  |           |              |
  |         .rsync      .rsync_root
  `----

  Unlike git repositories, they don't have a defined "root"
  directory. This is both an advantage and a disadvantage. This way, we
  can sync individual repositories, but that also means we need extra
  files to define root directories and repository locations. Here is the
  content for each of these files:

  ,----
  | /.rsync:           rsync://carbslinux.org/repo
  | /core/.rsync:      rsync://carbslinux.org/repo/core
  | /core/.rsync_root: ..
  `----

  The `.rsync_root' file on the core repository points to the upper
  directory.  If a `.rsync' file exists on the upper directory, this
  means that is the whole repository and will sync the entire repository
  instead of each individual repository.

  If the upper directory doesn't have this `.rsync' file, this means
  that this is an individual repository, and the package manager will
  fetch accordingly.


6.1.1 Setting up an Rsync repository
------------------------------------

  Carbs Linux repositories automatically sync from the git repostitories
  and serve it through the rsync daemon. Here is a sample shell script
  that I use in order to sync repositories. Feel free to customize for
  your own use.

  ,----
  | #!/bin/sh
  | HOSTNAME="rsync://carbslinux.org/repo"
  | GITDIR="/pub/git/repo"
  | SHAREDIR="/pub/share/repo"
  | git -C "$GITDIR" pull
  | 
  | rsync -avcC --delete --include=core --exclude=.rsync,.rsync_root "$GITDIR/." "$SHAREDIR"
  | 
  | printf '%s\n' "$HOSTNAME" > "$GITDIR/.rsync"
  | for dir in "$GITDIR/"*; do
  |     [ -d "$dir" ] || continue
  |     [ -f "$dir/.rsync" ] ||
  |     printf '%s/%s\n' "$HOSTNAME" "${dir##*/}" > "$dir/.rsync"
  |     printf '..\n' > "$dir/.rsync_root"
  | done
  `----

  You can then create an *rsync* user for serving the repositories.

  ,----
  | $ adduser -SD rsync
  `----

  Create `/etc/rsyncd.conf' and a service configuration as well.

  ,----
  | uid = rsync
  | gid = rsync
  | address = example.com
  | max connections = 10
  | use chroot = yes
  | 
  | [repo]
  |     path = /pub/share/repo
  |     comment = My repository
  `----

  Create a service file at `/etc/sv/rsync/run' (runit):

  ,----
  | #!/bin/sh -e
  | exec rsync --daemon --no-detach
  `----


6.2 Fossil repositories
~~~~~~~~~~~~~~~~~~~~~~~

  Setting up a Fossil repository is no different than setting up any
  other repository. There are certainly many advantages of using Fossil
  as a means of distributing packages. You can create a Linux
  distribution and have your website, forum, documentation, and your
  package repository entirely contained inside a single Fossil
  repository. Fossil's built-in wiki and forum features make it the
  ultimate single-tool distribution software.

  However, the biggest caveat of Fossil is that it doesn't allow
  symlinks by default unless it's manually set by the user, and this
  feature cannot even be set globally. Symbolic links aren't quite
  common within distribution repositories, but they come in handy where
  there are two packages that use the same source files (`emacs' and
  `emacs-nox', or `libelf' and `libdw' from elfutils). If symbolic links
  are too big of a deal for your repository, this can be a huge issue
  for you.


6.3 Message of the Day
~~~~~~~~~~~~~~~~~~~~~~

  If a file named `MOTD' (all uppercase) is found on the root directory
  of the package repository, its contents will be printed to the
  standard output when the users are updating their repositories. This
  method can be used to communicate messages to the users, such as
  package removals or otherwise important information.


7 Comparison Between CPT and KISS
=================================

  Lots of things have changed since `cpt' was forked from `kiss' in
  terms of functionalities and ideals. This section aims to describe the
  similarities and differences of both package managers as neutral as
  possible. Keep in mind that this is the `cpt' documentation, so it may
  be biased regardless.

  Package Manager
        While `kiss' aims to be a simple single file package manager,
        `cpt' aims to be an extendable package manager library. `kiss'
        has all of its features built-in, while `cpt' has all of its
        features separated into small tools.  These tools can be called
        from the main `cpt' tool (in order to keep `kiss'-like usage) or
        with their names directly (e.g `cpt-build').

  Configuration
        Neither `kiss' nor `cpt' use configuration files. Instead, they
        are configured through environment variables. Additionally, all
        `cpt' tools can receive flags that alter their
        functionality. `kiss' does not accept flags.

  Package Repositories
        In addition to git repositories, `cpt' also supports Rsync,
        Fossil, and Mercurial repositories.

  Package Sources
        In addition to git repositories for sources, `cpt' also supports
        Mercurial and Fossil repositories.

  Post-Installation Messages
        `kiss' and `cpt' interact with `post-install' messages
        differently. `kiss' does not differentiate between
        post-installation scripts and post-installation messages, and
        will save the output of all scripts named `post-install' to be
        printed after the installation of all packages are
        complete. `cpt' on the other hand, separates these with the
        `message' file. `cpt' runs `post-install' without saving the
        output to be printed a second time. It instead prints all
        `message' files after the installation is over.

  Portability
        `kiss' aims to be as portable as possible. `cpt' aims to be
        portable, but favours performance. `cpt' depends on `rsync' for
        package installation, while `kiss' has removed the dependency in
        favour of portability.


8 CPT Library
=============

  `cpt-lib' is the library of Carbs Packaging Tools which can be used to
  extend the functionality of the package manager. This is the API
  documentation of the package manager library.


8.1 Calling the library
~~~~~~~~~~~~~~~~~~~~~~~

  You can call the library on your scripts by adding the following line
  to your files:

  ,----
  | #!/bin/sh -e
  | . cpt-lib
  `----

  This will load the library inside your script, and will set some
  environment variables that are used inside the package manager.


8.2 Variables
~~~~~~~~~~~~~

  This section lists some of the variables defined by the package
  manager that can be used in scripts. These variables usually cannot be
  defined by the user, so they are not part of the [variables] section
  above.

  `$cpt_version'
        Package manager version.
  `$cpt_confdir'
        Location of the CPT system configuration directory. This is
        usually either `/etc/cpt' or `PREFIX/etc/cpt'.
  `$pkg_db'
        Location of the package database without the root
        (`var/db/cpt/installed').
  `$sys_db'
        Location of the package manager database, making use of the
        current `$CPT_ROOT' (`$CPT_ROOT/$pkg_db'). This is the database
        you probably want to use.
  `$cpt_base'
        Location of the file that defines the base packages.

  If for some reason, your script interacts with the directories created
  and managed by the package manager you should use the following
  variables instead of the user assigned variables such as `$CPT_CACHE'
  or `$CPT_TMPDIR'. The variables below are the ones used for package
  operations (which are assigned by using a combination of user-assigned
  values and their fallbacks).

  `$cac_dir'
        Cache directory used by the package manager.
  `$src_dir'
        Directory containing downloaded sources for packages.
  `$log_dir'
        Directory where logs are saved.
  `$bin_dir'
        Directory where built package tarballs are saved.
  `$tmp_dir'
        Temporary directory for the package manager operations.


[variables] See section 4.2


8.3 Option parsing
~~~~~~~~~~~~~~~~~~

  `cpt-lib' includes a POSIX-shell option parser inside named
  `getoptions'. You can see its own [documentation] for writing an
  option parser. The built-in version of the `getoptions' library is
  2.5.0 and there are no plans for updating it apart from bug fixes.


[documentation]
<https://github.com/ko1nksm/getoptions/blob/v2.5.0/README.md>

8.3.1 Defining a parser
-----------------------

  Some functions are called and set automatically when you call
  `cpt-lib', so you shouldn't define the option parser after calling the
  library, as some of the variables will already be set.

  If the function `parser_definition()' as defined when `cpt-lib' is
  called, cpt-lib will handle the option parsing itself by calling
  `getoptions' inside. Here is the proper way of doing it.

  ,----
  | #!/bin/sh -e
  | 
  | parser_definition() {
  |     # The rest arguments MUST be defined as 'REST'
  |     setup REST help:usage                  -- "usage: ${0##*/} [options] [pkg...]"
  |     msg                                    -- '' 'Options:'
  |     flag CPT_TEST -t export:1 init:@export -- "Enable tests"
  | 
  |     global_options
  | }
  | 
  | . cpt-lib
  `----


8.3.2 `global_options()'
------------------------

  The `global_options()' function is a simple convenience call to
  include flags that can be used inside most `cpt' tools. It defines the
  following flags:

   Flag  Long Option    Calls         
  ------------------------------------
   `-f'  `--force'      `CPT_FORCE'   
   `-y'  `--no-prompt'  `CPT_PROMPT'  
         `--root'       `CPT_ROOT'    
   `-h'  `--help'       `usage()'     
   `-v'  `--version'    `version()'   
         `--verbose'    `CPT_VERBOSE' 

  This function can take two arguments:

  `silent'
        If this argument is specified, the function does not print the
        usage information defined by its flags.
  `compact'
        If this argument is specified, the function only prints the help
        output of the `--help' and `--version' flags.


8.4 Message functions
~~~~~~~~~~~~~~~~~~~~~

  `cpt' has various functions to print information to users.


8.4.1 `out()'
-------------

  `out()' is a really simple function that prints messages to the
  standard output. It prints every argument with a newline. It is not
  meant to communicate with the user, it just exists to have a simple
  function to interact with other functions.

  ,----
  | $ out "This is an example call" "How are you?"
  | This is an example call
  | How are you?
  `----


8.4.2 `log()'
-------------

  `log()' is the most commonly used message function in the package
  manager. It is used to pretty print messages with visual cues, so it
  is easier to read and understand for the users. It changes message
  output for each argument it receives (takes up to three arguments).

  - If it takes a single argument, it prints a yellow leading arrow
    followed by colorless text.
  - If it takes two arguments, it prints a yellow leading arrow followed
    by the first argument (colored blue), and then followed by colorless
    second argument.
  - If it takes three arguments, instead of a yellow arrow, it prints
    the third argument in yellow, followed by the same two arguments as
    above.


8.4.3 `die()'
-------------

  `die()' wraps the `log()' function and exits with an error (1). It
  takes one or two arguments, which are sent to the `log()'
  function. The third argument for `log()' is set as `!>'.


8.4.4 `warn()'
--------------

  `warn()' is another function that wraps `log()'. In place of the third
  argument, it uses the word `WARNING'.


8.4.5 `prompt()'
----------------

  `prompt()' is an interactive function that waits for user input to
  continue.  It takes a single argument string to print a message, and
  then asks the user whether they want to continue or not. Prompts can
  be disabled by the user if they use a flag to disable them or set
  `CPT_PROMPT' to 0.


8.5 Text functions
~~~~~~~~~~~~~~~~~~

  Following functions are used to manipulate, check, or interact with
  text.


8.5.1 `contains()'
------------------

  `contains' function can be used to check whether a list variable
  contains a given string. If the string is inside the list, it will
  return 0, otherwise 1.

  ,----
  | # Usage
  | contains "$LIST" foo
  | 
  | contains "foo bar" foo  # Returns 0
  | contains "foo bar" baz  # Returns 1
  `----


8.5.2 `regesc()'
----------------

  `regesc()' can be used to escape regular expression characters that
  are defined in POSIX BRE. Those characters are, `$', `.', `*', `[',
  `\\', and `^'.

  ,----
  | regesc '^[$\'  # Returns \^\[\$\\
  `----


8.5.3 `pop()'
-------------

  `pop()' can be used to remove a word from a "string list" without a
  `sed' call. Word splitting is intentional when using this function.

  ,----
  | # Usage
  | pop foo from $LIST
  | 
  | pop foo from foo baz bar # Returns baz bar
  `----


8.5.4 `sepchar()'
-----------------

  This function can be used to separate characters from the given string
  without resorting to external resources.

  ,----
  | sepchar mystring
  | # Prints:
  | # m
  | # y
  | # s
  | # t
  | # r
  | # i
  | # n
  | # g
  `----


8.6 Portability functions
~~~~~~~~~~~~~~~~~~~~~~~~~

  These helper functions are used so that we don't depend on non-POSIX
  programs for certain functionality. They are prefixed with the `_'
  character.


8.6.1 `_seq()'
--------------

  This function is similar to `seq(1)' except that it only takes a
  single argument and doesn't print any newlines. It is suitable to be
  used in `for' loops.

  ,----
  | _seq 5
  | # Prints:
  | # 1 2 3 4 5
  `----


8.6.2 `_stat()'
---------------

  This function imitates `stat %U'. `stat' isn't defined by POSIX, and
  this is also a GNU extension. This function returns the owner of a
  file. If the owner cannot be found, it will return `root'.


8.6.3 `_readlinkf()'
--------------------

  This function was taken from [POSIX sh readlinkf library by Koichi
  Nakashima].  `readlink' is also not defined by POSIX, so this function
  uses `ls' to follow symbolic links until it reaches the actual file.


[POSIX sh readlinkf library by Koichi Nakashima]
<https://github.com/ko1nksm/readlinkf>


8.7 System Functions
~~~~~~~~~~~~~~~~~~~~

8.7.1 `as_root()'
-----------------

  `as_root()' calls the rest of the arguments as a different
  user. Unless a `$user' environment variable is set, it will call the
  following arguments as the root user. It supports the following
  programs for privilege escalation with the following order:

  1. `ssu'
  2. `sudo'
  3. `doas'
  4. `su'

  The program called for this operation can be overridden using the
  `$CPT_SU' variable.


8.8 Package Functions
~~~~~~~~~~~~~~~~~~~~~

  Obviously, package functions are the most important ones for
  `cpt-lib', those are the ones you will use to build, to query, to
  manipulate, or to otherwise interact with packages.


8.8.1 `pkg_build()'
-------------------

  This function builds all given packages. It resolves dependencies for
  the given packages, lints the package, extracts its sources and runs
  the `build' script.

  ,----
  | # Example
  | 
  | # Create the cache directories first, this is where the package will be built.
  | create_cache
  | 
  | # Build the package(s) you want to build.
  | pkg_build cpt
  `----


8.8.2 `pkg_depends()'
---------------------

  This function calculates the dependencies for the requested package,
  returning the variable `$deps'. This variable can then be passed to
  [pkg_order()] in order to generate an ordered list for building
  packages.


[pkg_order()] See section 8.8.3


8.8.3 `pkg_order()'
-------------------

  This function receives package names and returns `$order' and `$redro'
  variables that can be used for building and removing packages.


8.8.4 `pkg_owner()'
-------------------

  This function can be used to determine the owner of a package. The
  first argument is used for flags that will be passed to `grep', and
  the second one is for the file query. Rest of the arguments can be
  used in order to specify the manifests to be used, but it is
  optional. `pkg_owner()' will search for all the installed packages if
  no other arguments are given.

  ,----
  | # Example
  | pkg_owner -lFx /usr/bin/grep  # Returns 'busybox'
  | 
  | # An example call made by `pkg_fix_deps()` to figure out whether the built
  | # package contains the file it depends.
  | pkg_owner -l "/${dep#/}\$" "$PWD/manifest" >/dev/null && continue
  | pkg_owner -l "/${dep#/}\$" "$@" ||:
  `----


8.8.5 `pkg_isbuilt()'
---------------------

  This function returns with success when the given package has a built
  tarball with the matching version and release strings from the
  repository.


8.8.6 `pkg_lint()'
------------------

  This function checks whether a given package fits the proper package
  specification. This function *does not return with failure, it exits
  outright* if it fails.


8.8.7 `pkg_find()'
------------------

  `pkg_find()' is the tool for searching packages. It accepts up to 3
  arguments.

  $1: Query
        This is the only mandatory argument. It accepts globbing,
        meaning that shell wildcards can be used in the query.
  $2: Match
        If this exists `pkg_find()' will print every single match found
        in the search path. If it doesn't, `pkg_find()' will print the
        first match and exit.
  $3: Type
        This is the argument to be passed to the `test' function. Unless
        this argument is given, it defaults to `-d', which tests for
        directories.

  ,----
  | # Returns the first match of cpt
  | pkg_find cpt
  | 
  | # Returns all matches of cpt
  | pkg_find cpt all
  | 
  | # Returns all globbed matches for cpt* (e.g. cpt and cpt-extra)
  | pkg_find 'cpt*' all
  | 
  | # Returns all matching cpt-* executables on user's PATH
  | SEARCH_PATH=$PATH pkg_find 'cpt-*' all -x
  `----


8.8.8 `pkg_get_base()'
----------------------

  This function returns the base packages as defined in the base
  file. If an optional argument is present, it will print all package
  names in a single line.  If it is not given any arguments, it will
  return one package per line. See 4.1.1 for more information on base
  packages.


8.8.9 `pkg_gentree()'
---------------------

  This function generates a dependency tree for the given package. The
  output and the dependency generation can be configured through a
  series of keys given as the second argument. Those keys are:

  `b'
        Include the base packages to the dependency tree.
  `f'
        Include the given package itself to the generated tree.
  `x'
        Do not include make dependencies of the package.
  `r'
        Print the generated tree in reverse order.
  `n'
        Print all packages in a single line instead of a package per
        line.


* 8.8.9.1 Examples

  This example uses the `cpt' package for Carbs Linux. The package
  itself is listed to depend on `curl' and `rsync'. Here is the output
  of calling the function for cpt directly:

  ,----
  | $ pkg_gentree cpt
  | bearssl
  | ca-certificates
  | zlib
  | curl
  | rsync
  `----

  The example above shows that even though `cpt' itself only depends on
  `curl' and `rsync', we also indirectly need `zlib', `bearssl' (for
  `curl'), and `ca-certificates' (for `bearssl').

  ,----
  | # Print the dependency tree reverse sorted in a single line:
  | $ pkg_gentree cpt rn
  | rsync curl zlib ca-certificates bearssl
  `----


8.8.10 `pkg_query_meta()'
-------------------------

  This function is used to query the [meta file] inside package
  directories. It can be used to retrieve information on a package that
  is otherwise irrelevant to the package manager itself. It takes two
  arguments, first being the package (or the full path to a package
  directory) and the second being the key to be retrieved.  If the
  package does not have a `meta' file or the file does not contain the
  requested key, the function will return with 1.

  ,----
  | $ pkg_query_meta cpt description
  | Carbs Packaging Tools
  | 
  | $ pkg_query_meta /path/to/cpt license
  | MIT
  `----


[meta file] See section 5.6



Footnotes
_________

[1] <https://datatracker.ietf.org/doc/html/rfc822#section-3.2>

[2] Not mandatory for the packaging system, but mandatory for
inclusion in the repositories