从Ruby中包含“ [”的字符串中提取子字符串

Question

I have a file containing data like this:

[date,ip]:string{[0.892838,1.28820,8.828823]}

and I want to extract the data 0.892838, 1.28820, 8.828823 to a string for a later processing.

I have used the pattern line = String ~= /\\\\[/ to get the position where the "[" occurs but for the above input, I get this error message:

premature end of char-class /\\[/

Answer 1

How is this ?

str = 'date,ip]:string{[0.892838,1.28820,8.828823]}'
str.scan(/\d+.\d+/)
# => ["0.892838", "1.28820", "8.828823"]

Answer 2

Using capturing group:

'[date,ip]:string{[0.892838,1.28820,8.828823]}' =~ /{\[(.*?)\]}/
# => 16
$1            # => "0.892838,1.28820,8.828823"
$1.split(',') # => ["0.892838", "1.28820", "8.828823"]

Answer 3

As I'm prone to do:

require 'fruity'

str = '[date,ip]:' + ('string' * 1) + '{[0.892838,1.28820,8.828823]}'
compare do
  arup { str.scan(/\d+.\d+/) }
  falsetrue { str =~ /{\[(.*?)\]}/; $1.split(',') }
  ttm { str[/\[([^\]]+)\]}$/, 1].split(',') }
end

# >> Running each test 2048 times. Test will take about 1 second.
# >> falsetrue is similar to ttm
# >> ttm is faster than arup by 2x ± 0.1

The longer the string section is, the more the various attempts vary in their run times:

require 'fruity'

str = '[date,ip]:' + ('string' * 1000) + '{[0.892838,1.28820,8.828823]}'
compare do
  arup { str.scan(/\d+.\d+/) }
  falsetrue { str =~ /{\[(.*?)\]}/; $1.split(',') }
  ttm { str[/\[([^\]]+)\]}$/, 1].split(',') }
end

# >> Running each test 512 times. Test will take about 2 seconds.
# >> ttm is faster than falsetrue by 60.00000000000001% ± 10.0%
# >> falsetrue is faster than arup by 13x ± 1.0

The reason the "ttm" result improves in speed is because of '$' . That anchor gives the regular expression engine the information it needs to know where to search immediately. Without it it'd start at the beginning of the string and search forward so the longer the 'string' component, the more time it takes to find the desired pattern.

Experiment with the expressions using benchmarks and you can find the best average speed and expression for a particular task.

If the "string" section is always short, the difference for a single pass is so small it won't really matter and then it's sensible to use the most easily read (and easily maintained) code, which would be str.scan(/\\d+.\\d+/) . If the code is in a loop and being run millions of times, then it starts to make a difference and one of the others might be more sensible.