Translation(s): German - English

(!) ?Discussion

Command-line tools in Debian

Reviewing upstream packages to write debian/copyright files is tedious but important manual work. It is done during initial packaging and after every new upstream release.

Making initial copyright file construction and subsequent review/update easier will improve Debian's software quality.

Starting with Stretch (Debian 9) there are significantly improved tools over previous releases to help.

licensecheck

licensecheck from licensecheck (and older versions of devscripts) can scan source code and report found copyright holders and known licenses. Its approach is to detect licenses with a dataset (medium:~200 regexes) of regex patterns and key phrases (parts) and to reassemble these in detected licenses based on rules. In that sense this is somewhat similar to the combined approaches of Fossology/nomos and Ninka (see below for these tools). It also detects copyright statements. It output results in plain text (with customizable delimiter) or a Debian copyright file format. Written in Perl.

licensecheck --check '.*' --recursive --deb-machine --lines 0 -- *

licensecheck notably does not extract metadata from binary files yet, and a recommended workaround is to couple it with exiftool like this:

1>&2 exiftool '-textOut!' %d%f.%e:meta -short -short -recurse -ext ttf .
licensecheck --copyright --deb-machine --recursive --lines 0 --check '.*' --ignore '.*\.ttf$' -- *
find -type f -name '*.ttf' -delete

For more elaborate and actively maintained examples, see ghostscript and emscripten.

licensing

licensing from licenseutils is primarily for adding license boilerplate to new code but can also scan source code and report found licenses. Written in C.

licensing detect *

(2020-09-19 - bug 970580 created)

scan-copyrights

scan-copyrights from libconfig-model-dpkg-perl creates a copyright file from scratch. Written in Perl, using licensecheck. (note that copyright update is handled by cme. See below)

The main features of scan-copyrights are:

cme

Config::Model can update (or create from scratch) Debian copyright files using the cme command (from cme or libconfig-model-dpkg-perl less than 2.063). Written in Perl, using licensecheck.

The main features of scan-copyrights are:

# Install both packages. Latter one is "recommended" by cme, though

sudo apt install cme libconfig-model-dpkg-perl

# Run cme command

cme update dpkg-copyright

Usage is detailed in Config::Model wiki

licensecheck2dep5

A script from cdbs can create a copyright file by tidying output from licensecheck. Written in Perl, using #licensecheck.

licensecheck --check '.*' --recursive --copyright --deb-fmt --lines 0 * | /usr/lib/cdbs/licensecheck2dep5

/!\ This tool is discouraged - better use ?#licensecheck

license-miner

A script from cdbs can extract structured metadata embedded in binary content, for subsequent parsing by #licensecheck and suffix stripping by #licensecheck2dep5. Written in Perl, using Image::ExifTool and Font::TTF.

find -type f -name '*.png' -print0 | perl -0 /usr/lib/cdbs/license-miner
licensecheck --check '.*' --ignore '.+\.png$' --recursive --copyright --deb-fmt --lines 0 * | /usr/lib/cdbs/licensecheck2dep5
find -type f -name '*.png.metadata' -delete

CDBS

A makefile from cdbs can automate selection, mining, parsing, and cleanup, comparing previously autogenerated file debian/copyright_hints included with source package with freshly autogenerated instance and warning about newly introduced (but not disappearing) changes to discovered hints, using #license-miner and #licensecheck and #licensecheck2dep5 under the hood. Written in make.

Typical use is by shipping a package-specific script `debian/copyright-check with source package and executing that script manually (not as part of normal build) when sources change:

export DEB_COPYRIGHT_EXTRACT_EXTS="icc pdf png ttf"
export DEB_COPYRIGHT_EXTRACT_PATHS_EXIF="Resource/Font/"
export DEB_COPYRIGHT_CHECK_IGNORE_EXTS="cat ico xls pcl xps"
export DEB_COPYRIGHT_CHECK_IGNORE_PATHS="doc/.*\.htm"
export DEB_COPYRIGHT_CHECK_MERGE_SAME_LICENSE=yes

make -f /usr/share/cdbs/1/rules/utils.mk pre-build || true
make -f /usr/share/cdbs/1/rules/utils.mk clean DEB_COPYRIGHT_CHECK_STRICT=1

/!\ This tool is discouraged - better use ?#licensecheck

license-reconcile

license-reconcile compares the existing copyright with the source code and reports discrepancies. Written in Perl, using licensecheck. Partially replaced by licenserecon.

licenserecon

licenserecon parses a valid DEP-5 copyright file and notes the licenses of all files in the source tree. licensecheck is then run, and the results compared. Differences between licenses in debian/copyright and the output of licensecheck are reported.

It should be run in the top level of a cleaned Debian source tree, with a valid DEP-5 copyright file. The source tree should be clean, otherwise results may be contaminated by spurious reports on the build's generated files. It is advisable to run lintian first to ensure correct syntax of debian/copyright.

The results are indicative only, and not a substitute for manual checking. It is intended to report obvious errors. The design intends to minimise false positives as much as is practical. However, false positives will occur if the spelling of the license short-string is not identical between the file and debian/copyright. This is quite likely with complex licensing such as 'and'/'or' constructs and specific exceptions.

Only files with a copyright header are checked. False negatives may occur if licensecheck cannot determine a file's license. Files named copyright, copying, readme etc. are not checked as they often specify the licenses of other files rather than their own.

debmake

debmake -k also compares the existing copyright with the source code and reports discrepancies.

debmake -cc generates a new copyright file from the source code.

license-detector

license-detector scans quickly for licenses within paths, and loosely summarizes a percentage covered by each SPDX license. Written in Go.

decopy

decopy is a tool that "automates creating and updating the debian/copyright files." It also "aims to detects as many licenses as possible" which makes it a tool for license detection too. It uses python-debian to handle Debian machine readable copyright files. Its approach to detect licenses is the same as license-checker. Written in Python, using python-debian.

licensee

licensee from ruby-licensee checks LICENSE files and returns known license names. This is the tool used by Github to provide a summary license indication on a repository main page. Its approach is to search for typical LICENSE file names or some package manifest (NPM, Bower, Gemfile, etc) and perform an exact or approximate license text matching against the set of common licenses texts as published at https://choosealicense.com (small: ~20). It output results in YAML format. Written in Ruby.

check-all-the-things

Wrapper for some of the other tools listed here.

check-all-the-things -f copyright

cargo-lichking

Automated license checking for rust. cargo lichking is a Cargo subcommand that checks licensing information for dependencies, based on David A. Wheeler's compatibility graph.

cargo lichking check

reuse

Tool for checking and helping with compliance with the REUSE Initiative recommendations. Uses a combination of SPDX license identifiers and Debian machine readable copyright files to document license in a project. Written in Python.

Libraries in Debian

python-debian

python-debian has support parsing and creating copyright files (and any Debian-style files such as description, control, Sources, Packages, etc.) Written in Python.

Command-line tools not in Debian

license_finder

LicenseFinder is a tool that "Find licenses for your project's dependencies." It does so by running application-specific package management tools and detecting package manifests to collect license-related metadata (e.g. Gemfile, etc) and detect licensing using regex against a set of common license texts (small: ~20) and license names. It outputs results in CSV, HTML and other report format. Written in Ruby.

licensed

licensed is used to check the licenses of the dependencies of a project. Modern language package managers (bower, bundler, cabal, go, npm, stack) are used to pull the dependency chain of a specific project. Licenses can be configured to be either accepted or rejected, easing the developer task of identifying problematic dependencies when importing a new third-party library. Use github/licensee for license detection. Written in Ruby.

scancode-toolkit

ScanCode is a tool "to scan code and detect licenses, copyrights and more".

Its approach is to detect licenses using a dataset of plain license texts (large:~1,760 texts) available as an online licensedb; and a comprehensive library of license notices, mentions and references (large:~30,000 notices and mentions). ?ScanCode finds exact and approximate matches in source and binaries using a combination of checksums, automatons, and full text alignments (e.g. diffs) as well as SPDX license identifiers.

It can return the exact matched text (and the parts of a text that are not matched e.g. added or removed). It detects and normalizes structured license tags in package manifests including the ability to parse, detect and normalize Debian copyright files, with special support for structured DEP-5 machine readable files using the debian-inspector library.

It can also output a Debian copyright format. And can collect license information for installed packages from processing the status file.

It also detects copyright statements and collects license metadata from package manifests (e.g Maven, npm, rpm, Debian, Cargo, Cocoapods, Bower, Composer, Pypi, Alpine, and many more).

It output results in JSON, YAML HTML, Debian copyright or SPDX format.

It is written in Python with some native C/C++ extensions.

See also 983640.

apache-rat

Apache Creadur rat is a "tool to improve accuracy and efficiency when checking releases." . Its goal is to help Apache Foundation projects to comply with the release policy including detecting licenses. Its approach is to use a key sentences dataset (small: ~20). Written in Java.

cargo-deny

cargo-deny is a plugin to Rust helper tool cargo, to recursively check project-wide licensing hints for all dependent Rust crates, and check that they match a set of allow/deny condidtions. Written in Rust.

gitlog2copyright

gitlog2copyright will print "Copyright:" lines based on git history, with output formatted in a reasonable way. Can be useful for comparing with debian/copyright when indirect copyright clauses like "Copyright (C) The Authors" are used.

Other tools that need further detailing and review

Applications

fossology

FOSSology is a open source license compliance software system and toolkit that can (in version 3.1) generate DEP5 copyright files. Its approach is to detect licenses with a either large (large:~2500 regexes) dataset of regex patterns (nomos) or a full string comparison against license full texts (large: ~400 text) (monk). It also detects copyright statements and does also integrate with Ninka (see below). This is a complete database-backed web application with some command line support written in C/C++ with a PHP frontend.

scancode.io

ScanCode.io is a "server to script and automate the process of Software Composition Analysis (SCA) to identify any open source components and their license compliance data in an application’s codebase." ?ScanCode.io can be used for various use cases, such as Docker container and VM composition analyses, among other applications." It embeds ?ScanCode as a primary detection tool. ScanCode.io can analyze a complete Debian installed system for license such as a Docker image or a VM image and provides a web UI, a JSON ReST API, a CLI interface, JSON and XLSX outputs, and a plugin API for extending and creating custom analysis pipelines. It is written in Python with Django and PostgreSQL.

Obsolete code

OSLC

OSLCv3 Open Source License Checker 3.0 is a "risk management tool for analyzing open source software licenses." It detects licenses using key sentences and diffs using a dataset of license texts (small: ~50). It is developed in Java and seems no longer under development since 2009.

ninka

Ninka is a "license identification tool for Source Code". Its approach is to detect licenses from text sentences using a dataset of key license sentences (large: ~600) and assemble the results based on the matched sentences. It output results in CSV format. Written in Perl. Unmaintained since 2017.

jninka

jninka is a port from Perl to Java of ninka. Written in Java. Unmaintained/retired project.

slic

gerv/slic "Speedy LIcense Checker and associated tools". Written in Python. No longer maintained since the death of its author.

dlt

dlt has support for parsing and creating Debian machine readable copyright files. Written in Python. Unmaintained/retired project.

jfrog/go-license-discovery

codeauroraforum/lid

See also


CategoryPackaging