Command line perl regex to find dangling Javascript commas - regex

Question

Hello I'm seeking a Perl one-liner if possible, to scan all of our Javascript files, to find so-called "rogue commas". That is, commas that come at the end of an array or object data structure, and therefore commas that come immediately before either an ']' or '}' character.
The main challenge I'm encountering is how to make the regex that checks for ] or } non-greedy. The regex needs to span multiple lines, since the comma could end one line, followed by the } or ] on the next line, but I've figured out how to do that with the help of the book Minimal Perl.
Also, I'd like to be able to pipe a number of files to this Perl regex (via find/xargs), and so I'd like to print the name of the input file, and the line number within that file.
Below are various attempts of mine that are not particularly close to working straight from my bash history. Thanks in advance:
find winhome/workspace/SsuExt4Zoura/quotetool/js
-name "*.js" | xargs perl -00 -wnl -e '/,\s+$/ and print $_;' find winhome/workspace/SsuExt4Zoura/quotetool/js
-name "*.js" | xargs perl -00 -wnl -e '/,\s+/ and print $_;' find winhome/workspace/SsuExt4Zoura/quotetool/js
-name "*.js" | xargs perl -00 -wnl -e '/,\s+\]/ and print $_;' find winhome/workspace/SsuExt4Zoura/quotetool/js
-name "*.js" | xargs perl -00 -wnl -e '/,\s+[\]\}]/ and print $_;' find winhome/workspace/SsuExt4Zoura/quotetool/js
-name "*.js" | xargs perl -00 -wnl -e '/,\s+[\]\}]/ and print $_;' | wc -l find winhome/workspace/SsuExt4Zoura/quotetool/js
-name "*.js" | xargs perl -00 -wnl -e '/,\s+[\]\}]/ and print $_;' | wc -l find winhome/workspace/SsuExt4Zoura/quotetool/js
-name "*.js" | xargs perl -00 -wnl -e '/,\s+}/ and print $_;' | wc -l find winhome/workspace/SsuExt4Zoura/quotetool/js
-name "*.js" | xargs perl -00 -wnl -e '/,\s+}?/ and print $_;' | wc -l find winhome/workspace/SsuExt4Zoura/quotetool/js
-name "*.js" | xargs perl -00 -wnl -e '/,\s+}+?/ and print $_;' | wc -l find winhome/workspace/SsuExt4Zoura/quotetool/js
-name "*.js" | xargs perl -00 -wnl -e '/,$/' and print $_;' find winhome/workspace/SsuExt4Zoura/quotetool/js
-name "*.js" | xargs perl -00 -wnl -e '/,$/ and print $_;' find winhome/workspace/SsuExt4Zoura/quotetool/js
-name "*.js" | xargs perl -00 -wnl -e '/\,$/ and print $_;'

Answer 1

With the -00 switch, you change the record separator, and (probably) get the whole file in one line, which allows you to find multi-line trailing commas. However, it also makes the print $_ print the whole line. What you probably want is printing the file name:
print $ARGV if /,\s*[\]\}]/;

Answer 2

Most of these look like a decent approach to the problem, with one small issue. You probably want ,\s*(?:$|[\]\}]) rather than ,\s+(?:$|[\]\}]) as there may not be even one space. Your + quantifier might miss forms like ,].
Having said that, JavaScript can be pretty subtle, and you might well encounter comments and other stuff, which might legitimately end with a comma before something unexpected, like the end of the file or a }. A cheap solution might be to use a perl s/// form to simply remove all the comments before applying your tests.
If you're handling JSON, JSON::XS can enforce validity with its relaxed option.
If you need real validation, something like JSLint is probably the way to go. I've had a lot of success with using Rhino to embed JavaScript (a bit less using Perl with SpiderMonkey) and using this as a set of tests against JavaScript code would be a nice way to ensure reliability over time.

Answer 3

An easy solution to this problem is to use comma-first style. Since commas never come at the end of a line, there is never a 'trailing comma'.
For example:
var myObj = { foo: 1
, bar: 2
, baz: 4
}
You can easily detect if a comma is missing, it's obvious which elements belong to what set of braces, and there's never a 'trailing comma problem'.
See also https://gist.github.com/357981