Content is not correctly censored

[alina-tuholukova-gg]

Environment

• ggshield version: latest
• Operating system (Linux, macOS, Windows): Linux
• Operating system version:
• Python version: 3.10

Describe the bug

Steps to reproduce:

1. create file secret.txt with the following content (note the starting empty line)

password = 123nlsdkjfsaodi09ufsdf

# known
secret = "sk_live_epISFDSkdeXmn5asTvb7RHAi"

# new
secret = "sk_live_epISFDSkdeXmn5bnKvb7RHBn"

2. Run command ggshield secret scan path secret.txt

Actual result:

Expected result:

Secret censored correctly

Note: when in commit (pre-commit, pre-push etc) the content is censored correctly. Same bug with ggshield secret scan archive ...

[MostlyGenius]

can you assign this issue to me?

[agateau-gg]

Sure, here you are!

[SuperAayush]

Hey @agateau-gg!!

I would like to contribute to this issue!!

[agateau-gg]

Hi @SuperAayush, great to hear! You can have a look at our CONTRIBUTING.md file to get started. Feel free to ask more questions about the task as you need to.

[SuperAayush]

Hey @agateau-gg!!

I wanted to clear some of my doubts regarding the issue and was searching for Slack or Discord server for gg, I was unable to find one.

Can you please let me know what is the best medium to have a conversion with you?
Thank You!!

[Walz]

Hello @SuperAayush,

I took a look at this bug and I found it's a server-side issue. It seems the document is stripped of its white-spaces by the API, rendering the indices of the match invalid. This can be seen from the secrets detection playground:

The value of index_start and index_end stay the same with spaces and newlines at the start of the document. This cause the bug describe in this issue, because it's based on the indices from the API.

I've opened an internal PR to fix this bug. This should be fixed in our next release.

Sorry for any wasted time, because the bug was caused by our API. Your involvement was still valuable. Thank you for dedicating your time to help with our open-source CLI. I encourage you to ask any questions directly on the GitHub issues as we currently use it as our primary communication channel.

[Walz]

The fix has been deployed.