Sanitize the conn_id to disallow potential script execution

[andylamp]

---

This PR attempts to address an issue raised by @potiuk in #32770, which aims to sanitize the conn_id value. I have used a regex to check its validity during creation and have allowed any alphanumeric characters plus some special characters albeit the ones that could potentially be used to introduce scripts.

Further, I have also set the limit of characters for a valid connection to be up to 200, which I think is also a reasonable number but happy to adjust based on feedback.

closes: #32770

[boring-cyborg]

Congratulations on your first Pull Request and welcome to the Apache Airflow community! If you have any issues or are unsure about any anything please check our Contribution Guide (https://github.com/apache/airflow/blob/main/CONTRIBUTING.rst)
Here are some useful points:

• Pay attention to the quality of your code (ruff, mypy and type annotations). Our pre-commits will help you with that.
• In case of a new feature add useful documentation (in docstrings or in docs/ directory). Adding a new operator? Check this short guide Consider adding an example DAG that shows how users should use it.
• Consider using Breeze environment for testing locally, it's a heavy docker but it ships with a working Airflow and a lot of integrations.
• Be patient and persistent. It might take some time to get a review or get the final approval from Committers.
• Please follow ASF Code of Conduct for all communication including (but not limited to) comments on Pull Requests, Mailing list and Slack.
• Be sure to read the Airflow Coding style.
  Apache Airflow is a community-driven project and together we are making it better 🚀.
  In case of doubts contact the developers at:
  Mailing List: dev@airflow.apache.org
  Slack: https://s.apache.org/airflow-slack

[vincbeck]

Good job on your first PR 👏

[andylamp]

thanks a lot @vincbeck! hopefully one of many :)

[andylamp]

hmm @uranusjr, looking at this a bit further - I omitted a detail when reviewing your code. match returns a matched object which is a group of matches and not a str as is the desired outcome at this stage. I have tweaked it in order to still be valid and return the desired value.

[amoghrajesh]

One minor nit. Looks awesome otherwise +1
Good work!

[uranusjr]

Pending CI passing.

[andylamp]

@uranusjr thr failed CI tasks are not related to this PR, if am not mistaken... right?

[potiuk]

@uranusjr thr failed CI tasks are not related to this PR, if am not mistaken... right?

@andylamp -> you are 52 commits behind. It's hard to say. Certainly the main builds are not failing with the same error. But you will actually find out when you rebase to latest . Which is always the first thing to do if you see errors that might be unrelated.

[andylamp]

hi all, sorry for the delay - got a bit of time to get this fixed. @potiuk it turns out that the error FAB front-end was failing was due to the string message having an illegal character.

To further reduce any potential formatting (and security?) impact I have reduced the available characters that can comprise a valid conn_id. Now the allowed characters are allows alphanumeric plus the symbols #,!,-,_,.,:,,/ and () requiring at least one match. To allow a bit of freedom, I have put a parameter to configure the max length allowed and by default it is set to the same length keys should have, namely 250.

All tests that I tried pass and the connection error pops-up successfully without any errors in the backend of Airflow.

Happy for any comments regarding this but I think it is finally complete :-).

[andylamp]

addressing the pre-commit error and pushing a fix again with some typo fixes as well, hopefully that's the end of it :)

Modified multiple times.

[potiuk]

Hmm... Just thinking about that... What will happen in f someone already has conn_id with forbidden character ... I think - even if it's not the cleanest, maybe a better solution (if possible) we can simply escape what's there when passing it to form control?

[andylamp]

The issue was with the description text shown in the error message, not the regex itself - so in theory can I can use the old one but then perhaps the error message gets rather long due to not being able to show the concise message using special characters... What do you think @potiuk?

[potiuk]

The issue was with the description text shown in the error message, not the regex itself - so in theory can I can use the old one but then perhaps the error message gets rather long due to not being able to show the concise message using special characters... What do you think @potiuk?

Well ... in this case... we can also just escape AND shorten the connection id. It's not THAT important to show connection_id fully - we know what connection id we were just updating so showing all of it in the error message is probably good enough.

I have a feeling (now after looking at it and reallising that it's just error message) that sanitizing the id at entry in this case is far too invasive.

[andylamp]

Sorry, perhaps that was misunderstood - the error pops up when you try to edit a connection id and the new value is an updated one that includes an error. I do not think you can have a situation where an existing connection id was invalid...

As far as the error message goes; I think It is good practice to show the error but also what the rules of a valid connection id are...

Finally, just to be on the same page here - what do you mean by escape? Because, in terms of raw strings as far as Python goes it is escaped and parsed as a raw string but markdown safe parsing does not allow these characters from FAB itself.

edit: the sanitization happens to ensure validity and that the entry is valid - the concerns you had are relevant, it's just that the error message showing in my PR had an issue. That issue I rectified by tweaking the message shown :)

edit2: BTW, the connection ID is not shown in the error message. The error you encountered stemmed by the fact I showed the regex rules for an acceptable connection id; and that was causing an issue with the rendering. I understood based on what you said that perhaps this was not clear...

[potiuk]

Ah .. Stupid me.. It's been too long from the original discussion and I simply made some bad assumptions - I have forgotten that we already validate the connection_id in the first place and that all connection_ids in DB are already following it.

Sorry for the confusion.

It looks good now! LGTM

[andylamp]

I think these tests fail in the main branch as well and are unrelated to my changes :)

[potiuk]

The compat one - yes - I am looking at it now.

[boring-cyborg]

Awesome work, congrats on your first merged pull request! You are invited to check our Issue Tracker for additional contributions.

[potiuk]

Congrats on your firs PR merged 🎉

[potiuk]

(BTW the main bug is also fixed now).

[andylamp]

Awesome!! Thanks for the visibility!! 💪