chore: Add script to identify broken doc links

[ckadner]

Motivation

Broken links in documentation are making it harder for new comers to learn about ModelMesh and often it is not easy to find what the correct link should be. For project maintainers it is time consuming to revisit documentation to identify broken links manually and it certainly does not make the best impression on new comers if they click on links that return 404 errors.

Modifications

Add new script scripts/verify_doc_links.py and new Make target check-doc-links to identify broken links in any of the currently 211 links in 43 Markdown files.

Result

$ make check-doc-links

Checked 211 links (144 unique URLs) in 43 Markdown files.

docs/configuration/README.md:68: /docs/configuration/predictors/run-inference -> 404
docs/inference/data-types-mapping.md:5: https://github.com/triton-inference-server/server/blob/main/docs/model_configuration.md#datatypes -> 404
docs/model-formats/advanced-configuration.md:64: https://github.com/triton-inference-server/server/blob/main/docs/model_configuration.md#inputs-and-outputs -> 404
docs/model-formats/advanced-configuration.md:54: https://github.com/triton-inference-server/server/blob/main/docs/model_configuration.md#maximum-batch-size -> 404
docs/model-formats/advanced-configuration.md:48: https://github.com/triton-inference-server/server/blob/main/docs/model_configuration.md#scheduling-and-batching -> 404
docs/predictors/README.md:113: /docs/predictors/inferenceservice-cr.md.md#predictor-status -> 404
docs/runtimes/custom_runtimes.md:246: /docs/runtimes/configuration -> 404

ERROR: Found 7 invalid Markdown links
make: *** [check-doc-links] Error 1

And after fixing the broken links:

$ make check-doc-links

Verified 211 links (142 unique URLs) in 43 Markdown files.

check-doc-links: OK

How does the script work

• Find all Markdown files in the project
• Extract all Markdown-style links and plain URLs
• Create unique list of simplified URLs
• Use local files to check relative links, use cached HTTP requests for links outside project
• Run HTTP requests in parallel
• Retry GitHub rate-limited URLs after (at least) one minute
• Report broken links with source file and line number

Script pedigree: I contributed several iteration of this script to MLX and KFP-Tekton projects. This one differs in that it does not require any non-standard libraries like for Markdown or requests to be installed into a virtual environment prior to running the script.

Possible future enhancements

Add a new Pull Request builder check to run the check-doc-links frequently and fix broken links as soon as possible.

---

/cc @njhill @chinhuang007

[njhill]

Thanks @ckadner this is great! Maybe we should consider also adding it to the main kserve repo. That also has a lot of docs/inks.

[kserve-oss-bot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: ckadner, njhill

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

• OWNERS [njhill]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[ckadner]

Thanks @ckadner this is great! Maybe we should consider also adding it to the main kserve repo. That also has a lot of docs/inks.

Okay, good idea. Will do.

---

kserve/kserve#2423
kserve/website#186

[njhill]

/lgtm