Inconsistent behavior in Ruby gsub replacement?

Question

The two gsub's yield different outcomes. Can anybody explain why?

Code is also available at https://gist.github.com/franklsf95/6c0f8938f28706b5644d.

    ver = 9999
    str = "\t<key>CFBundleDevelopmentRegion</key>\n\t<string>en</string>\n\t<key>CFBundleVersion</key>\n\t<string>0.1.190</string>\n\t<key>AppID</key>\n\t<string>000000000000000</string>"
    puts str.gsub /(CFBundleVersion<\/key>\n\t.*\.).*(<\/string>)/, "#{$1}#{ver}#{$2}"
    puts '--------'
    puts str.gsub /(CFBundleVersion<\/key>\n\t.*\.).*(<\/string>)/, "#{$1}#{ver}#{$2}"

My ruby version is ruby 2.1.2p95 (2014-05-08 revision 45877) [x86_64-darwin13.0] (MRI). On my machine, the outcome is:

<key>CFBundleDevelopmentRegion</key>
<string>en</string>
<key>9999
<key>AppID</key>
<string>000000000000000</string>
--------
<key>CFBundleDevelopmentRegion</key>
<string>en</string>
<key>CFBundleVersion</key>
<string>0.1.9999</string>
<key>AppID</key>
<string>000000000000000</string>

The second one is the desired effect, but the first one is wrong.

Answer 1

It has to do with timing and how ruby regexes work.

gsub sets $1 and $2, but not until after it completes. So when you run the first time through, they're blank. When you run the second time, they were set by the previous gsub. If you want to do regex captures in place, you need \1 and \2, like this:

puts str.gsub /(CFBundleVersion<\/key>\n\t.*\.).*(<\/string>)/, '\1' + ver.to_s + '\2'

Answer 2

If you use the block form of gsub(), your code will work correctly:

ver = 9999

str = "\t<key>CFBundleDevelopmentRegion</key>\n\t<string>en</string>\n\t<key>CFBundleVersion</key>\n\t<string>0.1.190</string>\n\t<key>AppID</key>\n\t<string>000000000000000</string>"

puts str.gsub(/(CFBundleVersion<\/key>\n\t.*\.).*(<\/string>)/) {|match|
  "#{$1}#{ver}#{$2}"
}

puts '-' * 20

puts str.gsub(/(CFBundleVersion<\/key>\n\t.*\.).*(<\/string>)/) {|match|
  "#{$1}#{ver}#{$2}"
}

--output:--
    <key>CFBundleDevelopmentRegion</key>
    <string>en</string>
    <key>CFBundleVersion</key>
    <string>0.1.9999</string>
    <key>AppID</key>
    <string>000000000000000</string>
--------------------
    <key>CFBundleDevelopmentRegion</key>
    <string>en</string>
    <key>CFBundleVersion</key>
    <string>0.1.9999</string>
    <key>AppID</key>
    <string>000000000000000</string>

The docs describe this behavior:

If replacement is a String, ... However, within replacement the special match variables, such as $&, will not refer to the current match.

...

In the block form, the current match string is passed in as a parameter, and variables such as $1, $2, $`, $&, and $' will be set appropriately. The value returned by the block will be substituted for the match on each call.