Friday, May 9, 2014

Finding illegal characters somewhere in your chef cookbook / code

Use case:
I cant upload a cookbook / run powershell script / render web.config because something somewhere throws up on an illegal character.

One route to fix it, is to look at most critical files - web.config, json.config, README.MD and hope that the error is somewhere in there.

Well, that's one approach. About a year ago I would fire up notepad++ or BeyondCompare and look for iffy characters. It works, but it's stupid slow.

In case of Chef, the whole cookbook is processed, so theoretically the error could be anywhere. So, going through it one file at a time is just not practical. On top of that, I advocate automation, so... automation it is:

I've been learning OSX so I actually dont know the equivalent for windows, but since you dont have grep anyway, install the version that comes with -P and all is good.

brew install pcre
CD to the folder where you're breaking
pcregrep --color='auto'  -r -n "[^\x00-\x7F]" .
pcregrep --color='auto'  -r -n "[\x80-\xFF]" .

-r                              is for recursive
--exclude-dir .git      is to exclude .git from your search. You'll get a ton of hits here otherwise (not used in example above)
-n                             is to specify ascii range. Both work.


Way to test:  Grab a broken Repo and run the command to make sure you get expected broken file
Fix is here (, so we're going to grab a pre fix commit

git clone
git checkout 5e1a812a38c19a45b774862345f552c41836606f
pcregrep --color='auto'  -r -n "[\x80-\xFF]" .


No comments:

Post a Comment

Comments are welcomed and appreciated.