a how-to on checking for broken links on GitLab merge requests

gitlab-mr-links-check.md

GitLab MR links check

how to test for broken links on incoming changes to the codebase (and not only validite URL syntax).

TL;DR

replace <1234> with the desired MR number, and <main> with the MR base branch name, and run

git mr upstream <1234> && git diff <main> mr-upstream-<1234> | grep '^+ ' | grep -Eo 'https?://\S+?\.html' | bash -c "$(curl -fsSL https://gist.githubusercontent.com/eliranmal/4c04e4fe93aea796e4d48ae1f1a6a5ac/raw/8ae356df273ea226de79c52f0cfb508e17ea4fc7/check-links.sh)"

step by step

• checkout the MR branch locally (using the git mr alias)

  git mr upstream 1234
  

• store changes to a patch file, diffing with the main branch

  git diff main mr-upstream-1234 > mr-changes.patch
  

• extract only the additions as input for the following step (in a patch file, these are simply lines that begin with a single plus sign)

  cat mr-changes.patch | grep '^+ ' > mr-additions.patch
  

note: the above steps can be automated/made obsolete/encapsulated by use of git hooks to act on relevant changes alone.

• further extract the links embedded in the diff text

  cat mr-additions.patch | grep -Eo 'https?://\S+?\.html' > mr-links.txt
  

• pass these links to a script that loops through them and invokes a network call for each, testing for successful response

  cat mr-links.txt | bash -c "$(curl -fsSL https://gist.githubusercontent.com/eliranmal/4c04e4fe93aea796e4d48ae1f1a6a5ac/raw/8ae356df273ea226de79c52f0cfb508e17ea4fc7/check-links.sh)"