Install prerequisites

Depending on your operating system, you may have multiple options for installing the prerequisites. For a typical installation you will need:

• Python: linkmedic is only tested on officially supported Python versions.
• A Python package installer: For example, pip or pipx

Install linkmedic

You can install the linkmedic using your favorite Python package installer. For example, using pipx, you can download it from PyPI:

pipx install linkmedic

Run

To start a test web server with files at /var/www and crawl the pages and test all the links starting from the /var/www/index.html page, run:

linkmedic --root=/var/www

Container image

• Installation: Use a container engine like docker or podman to pull the image from Quay or MPCDF GitLab container registry:

podman pull quay.io/meisam/linkmedic:latest

podman pull gitlab-registry.mpcdf.mpg.de/tbz/linkmedic:latest

• Availability: All versions
• Content (latest): Alpine Linux Base OS with system packages, Python packages, and linkmedkit.
• Requirements: Container engine
• Compatibility: Works with open-source engines like podman (which support multiple operating systems such as Linux, macOS, Windows, and hardware architecture)
• Size (version 0.9.1.dev91+g7752c53): ~32MB
• Automatic update: Supported via tools like podman-auto-update
• Notes: Original distribution method. Includes bash shell with globstar enabled for CI environments. Uses musl library instead of GNU libc. Access specific versions using tags (e.g., linkmedic:v0.7.4). Available tags: Quay

Mount website files when using containers:

podman run --volume /www/public:/test quay.io/meisam/linkmedic:latest linkmedic --root=/test

The --volume flag maps /www/public to /test inside the container.

Static single binary

• Installation: Download the file with name ending in -static from this page, set execute permissions, and run.
• Availability: Latest version only (UNSTABLE)
• Content: linkmedic scripts, all dependencies, Python interpreter, and required system packages in compressed format
• Requirements: None
• Compatibility: Most Linux distributions (x86-64 only)
• Size (version 0.9.1.dev91+g7752c53): ~17 MB
• Automatic update: Not supported
• Notes: Minimal reproducible method. This distribution method may not suit packages with large data or short execution times.

Dynamically-linked single binary

• Installation: Download the non--static file from this page, set execute permissions, and run.
• Availability: Latest version only (UNSTABLE)
• Content: linkmedic scripts, dependencies, and Python interpreter
• Requirements: glibc and related system packages
• Compatibility: Linux with recent glibc (x86-64 only)
• Size (version 0.9.1.dev91+g7752c53): ~16 MB
• Automatic update: Not supported

Build from source

• Installation: Install directly from Git:

pipx install git+https://gitlab.mpcdf.mpg.de/tbz/linkmedic.git

• Availability: All versions
• Content: Git snapshot with submodules
• Requirements: Git, a python package manager (for example, pipx), and Python
• Compatibility: Any Python-compatible environment
• Size (version 0.9.1.dev91+g7752c53): ~2.1MB
• Automatic update: Supported
• Notes: Supports PEP 517 and PEP 621 builders. Tested with pdm-backend, flit-core, and hatchling.

Released Python package

• Installation: Install from PyPI or MPCDF GitLab:

pipx install linkmedic

pipx install linkmedic --index-url https://gitlab.mpcdf.mpg.de/api/v4/projects/5763/packages/pypi/simple

• Availability: All versions
• Content: Pure-Python wheel built by PDM using PDM-backend
• Requirements: a python package manager (for example, pipx) and Python
• Compatibility: Any Python-compatible environment
• Size (version 0.9.0): ~20KB
• Automatic update: Supported

Package with Optional Dependencies

To test OpenDocument files (.odt, .odp, .ods) or single XML files (.fodt, .fodp, .fods), install with:

pipx install linkmedic[odf]

CI/CD

You can also use the container image in your CI/CD pipelines. For example, for GitLab CI, in the .gitlab-ci.yml file:

test_internal_links:
  image: quay.io/meisam/linkmedic:latest
  script:
    - linkmedic --root=/var/www/ --entry=index.html --warn-http --with-badge
  after_script:
    - gitlab_badge_sticker.sh

or for Woodpecker CI in the .woodpecker.yml file:

test_internal_links:
  image: quay.io/meisam/linkmedic:latest
  commands:
    - linkmedic --root=/var/www/ --entry=index.html --warn-http

If you want to check the external links of your website in your CI pipeline, you must avoid running multiple tests in a short period of time, e.g., on each commit to the development branches. Otherwise, the IP address of your CI runners may get banned by external web servers. For example, in GitLab CI, you can limit the external link checks to only the default branch of your Git repository:

test_external_links:
  image: quay.io/meisam/linkmedic:latest
  rules:
    - if: $CI_COMMIT_BRANCH == $CI_DEFAULT_BRANCH
  script:
    - linkmedic --root=/var/www/ --ignore-local --with-badge
  after_script:
    - gitlab_badge_sticker.sh
  allow_failure: true

Please note that the gitlab_badge_sticker.sh script used in these examples requires an API access token CI_API_TOKEN with maintainer permission to modify the GitLab repository badges. See the linkmedkit documentation for more details.

CLI reference

• Display help: This will show all the command-line options and their default values.

linkmedic -h

• Start the web server with the current directory as the root path of the server. Starting from index.html, crawl the pages and test all the links.

linkmedic

• Start the web server with ./tests/public1/ as the root path of the server. Starting from index.html, crawl the pages and test all the links.

linkmedic --root=./tests/public1/

• Start the web server with ./tests/public1/ as the root path of the server. Starting from index2.html, crawl the pages and test all the links. The entry point should be relative to the server root. (In the example, index2.html should be accessible at ./tests/public1/index2.html)

linkmedic --root=./tests/public1/ --entry=index2.html

• Configure the test web server not to redirect missing local pages (e.g., from /directory/page to /directory/page.html).

linkmedic --no-local-redirect

• Check links to external websites.

  ⚠️ IMPORTANT: You must avoid running the link checker on external links multiple times in a short period, e.g., on each commit to the development branch. Otherwise, the IP address of your machine (or CI runners) may get banned by the CDN or the DoS mitigation solution of the external web servers. See the CI/CD section for a possible solution.

linkmedic --check-external

• Do not follow external link redirections and consider the link to be alive. Depending on the configuration of external web servers, this option can result in some dead links not being detetcted when instead of returning 404 page directly, the webserver is asking the client to load another page.

linkmedic --no-external-redirects

• Ignore local dead links and activates external link checking.

linkmedic --ignore-local

• Do not consider external links that return HTTP status codes 403 and 503 as dead links.

linkmedic --ignore-status 403 503

• Check links in an OpenDocument file (e.g., .odt, .odp, .ods), or a single OpenDocument XML file (e.g., .fodt, .fodp, .fods).

linkmedic --entry=./presentation.odp

• Show warning for HTTP links.

linkmedic --warn-http

• If any link to mydomain.com is encountered, treat it as an internal link and resolve it locally.

linkmedic --domain=mydomain.com

• Start the web server on port 3000. If the web server cannot be started on the requested port, the initializer will automatically try the next available ports.

linkmedic --port=3000

• Generate badge information file. Depending on the type of diagnosis, this file will be named badge.dead_internal_links.json, badge.dead_external_links.json, or badge.dead_links.json. If the --warn-http flag is used, a badge file for the number of discovered HTTP links will also be written to the badge.http_links.json file. These files can be used to generate badges (see linkmedkit scripts) or to serve as a response for the shields.io endpoint.

linkmedic --with-badge

• Check the links but always exit with code 0.

linkmedic --exit-zero

• Log the output at a different level of verbosity. If more than one of these flags is defined, the most restrictive one will be in effect.
  • --verbose : log debug information
  • --quiet : log only errors
  • --silent : completely silence the output logs
• Read guidelines overrides from linkmedic.guides.ini instead of the default path (.linkmedic.ini)

linkmedic --guidelines-override-file=linkmedic.guides.ini

• Dump all guidelines to log.linkmedic.ini and exit. You can use this file as a template to write overrides for the default values.

linkmedic --guidelines-dump-file=log.linkmedic.ini

• Dump the crawler links list to the a file. The filename is set in guidelines ([crawler] -> links_dump_file). If the --domain flag has not been set, local links will be referenced from the website root as /your/path/page.html.

linkmedic --dump-links

Guidelines

You can override the internal guidelines (configuration) of linkmedic by adding a .linkmedic.ini file with your desired values. This file is parsed using Python's internal configparser module. You can choose a different name for this file using the --guidelines-override-file flag. The default values will be used for any options that are missing in the override file.

The guidelines are logged to the output while running in verbose mode and can be saved to a file using the --guidelines-dump-file flag.

.linkignore

Each line in the .linkignore file specifies a regex pattern for addresses that should be ignored during link checks. Note that regex matches . to any character (use \. for matching only to .) and the leading / is considered when matching local links.

/ignore/.*/this
/invalidfile\.tar\.gz
/will_add/later\.html
https://not\.accessible\.com