Bug #8560: CSV, skip_lines option causes error when passing a string

• Status: Closed
• Assignee: JEG2 (James Gray)
• Target version: -
• ruby -v: ruby 2.0.0p0 (2013-02-24 revision 39474) [x86_64-linux]
• Backport: 1.9.3: UNKNOWN, 2.0.0: UNKNOWN

[kstevens715 (Kyle Stevens)]

There seems to be a bug in the CSV class when using the skip_lines option.This option is currently undocumented, but according to the GitHub PR, it accepts any object that responds to match. String responds to match, so one would imagine a string can be used. However, String#match can take either a Regexp or another String. If the argument passed is a string, it will first be converted to a Regexp.

So if you pass a string to the skip_lines option, it will attempt to convert the row in the csv file to a Regexp. This doesn't make sense, and it can also lead to exceptions. For example, if the csv contains a row with the data, "# )", you will get an error, "RegexpError: unmatched close parenthesis: /# )/".

My particular use case when I found this problem was I was trying to ignore lines beginning with a "#".

This was my first, unsuccessful attempt:
csv = CSV.open(FILE_NAME, skip_lines: "#", encoding: "ISO8859-1")

What I ended up having to do was:
csv = CSV.open(FILE_NAME, skip_lines: /\A#/, encoding: "ISO8859-1")

This isn't a huge problem, since there's a perfectly acceptable work-around. However, it would be a very easy mistake to make and it could be a difficult problem for someone to debug. It could lead to quite strange behavior if each row in the csv is converted to a Regexp.

I think the skip_lines option should be converted to a Regexp if it's a string, because the alternative is the CSV row is going to be converted to a Regexp.

My proposal is if a string is passed to skip_lines, it should be converted to a regular expression to match the beginning of a line, excluding whitespace:
"#" => /\A\s+#/

I'd be willing to work on a pull request to implement a fix, but I'd love to hear some feedback first. I definitely think this should be fixed, but I'm not positive my solution is the best option...

Here is the original pull request that implemented this option:
https://github.com/ruby/ruby/pull/161

Thank you

• Assignee set to JEG2 (James Gray)

[zzak (zzak _)]

@Kyle you might want to try to ping JEG2 on that ticket as well. He
may not see this.

[kstevens715 (Kyle Stevens)]

• File csv-string-skip-lines.patch csv-string-skip-lines.patch added

Attached is a patch that converts skip_lines to a Regexp if it's a string. Also see pull request on GitHub: https://github.com/ruby/ruby/pull/455

[JEG2 (James Gray)]

• Category set to lib
• Status changed from Open to Closed
• % Done changed from 0 to 100