Survive through a malformed installed.json file.

[bucefal91]

We have been using Kint and occasionally would run into the following concern.

Somehow, during the deployments (when the system is transitioning from "old" to the "new" state), we would encounter PHP warnings

PHP Warning:  Invalid argument supplied for foreach() in ....../vendor/kint-php/kint/src/Utils.php on line 87

I am unable to reproduce it "per se". But any simple change to the vendor/composer/installed.json will reproduce it, i.e. just edit this file and make it a "malformed" JSON. The point is that \json_decode(\file_get_contents($installed), true); would fail to parse it. Technically, composer could fail or people could manually tamper the file so I think "dealing" with the edge case where the file does not contain valid JSON is a worthwhile thing to have :)

With this patch/PR applied, even if you "mess up" your vendor/composer/installed.json, the warning will not be generated.

[jnvsor]

Hi @bucefal91 - nice catch!

The phar will need to be rebuilt, and I'd like an additional test case for this edge case too. Do you want to try your hand at it or shall I?

[bucefal91]

Hello, @jnvsor ,

Thank you for your prompt reply. I sure can try :) I hope to have it done by the end of today.

[jnvsor]

You should just be able to add a file_put_contents and assert at the end of UtilsTest::testComposerGetExtras and that'll do the trick

[bucefal91]

@jnvsor ,

I think I have applied your proposals/feedback:

• .phar file "built" & committed.
• I figured I missed a code style thing in the Utils class.
• Added the test coverage into UtilsTest::testComposerGetExtras() and I confirmed the amended test passes in this branch and fails in master.

[jnvsor]

Hi @bucefal91 sorry for the delay I didn't get a notification from gh

Looks good and I presume the tests will pass too (Looks like the build didn't match, did you format after build or before?)

Could you use assertSame([], ... instead of assertEmpty in the test? That way we can check the type too (It's essentially === vs empty())

[bucefal91]

@jnvsor ,

Yes, sir! :) I've changed it to the strict equality in the test and rebuilt the .phar file anew.

[jnvsor]

Passed, merged, tagged, pushed. Good job!

[bucefal91]

Thank you, I appreciate your support and desire to take a patch into the project. :)