Create a more efficient regex

[shanempope]

require 'benchmark'

# Original regular expressions
original = Regexp.union(
  %r{\A(?:/\*.*?\*/)?\s*ROLLBACK}i,
  %r{\A(?:/\*.*?\*/)?\s*COMMIT}i,
  %r{\A(?:/\*.*?\*/)?\s*RELEASE\s+SAVEPOINT}i,
)

# Combined regular expression
combined = %r{\A(?:/\*.*?\*/)?\s*(ROLLBACK|COMMIT|RELEASE\s+SAVEPOINT)}i

# Test string
sql = "/* comment */ COMMIT"

n = 1_000_000
Benchmark.bm do |x|
  x.report("original:") { n.times { original.match?(sql) } }
  x.report("combined:") { n.times { combined.match?(sql) } }
end

More complicated benchmark:

[rafaelfranca]

Pro tip, always use benchmark-ips to write benchmarks:

require "bundler/setup"
require "benchmark/ips"

# Original regular expressions
original = Regexp.union(
  %r{\A(?:/\*.*?\*/)?\s*ROLLBACK}i,
  %r{\A(?:/\*.*?\*/)?\s*COMMIT}i,
  %r{\A(?:/\*.*?\*/)?\s*RELEASE\s+SAVEPOINT}i,
)

# Combined regular expression
combined = %r{\A(?:/\*.*?\*/)?\s*(ROLLBACK|COMMIT|RELEASE\s+SAVEPOINT)}i

# Test string
sql = "/* comment */ COMMIT"

Benchmark.ips do |x|
  x.report("original:") { original.match?(sql) }
  x.report("combined:") { combined.match?(sql) }
  x.compare!
end

Warming up --------------------------------------
           original:   111.550k i/100ms
           combined:   387.956k i/100ms
Calculating -------------------------------------
           original:      1.062M (±12.4%) i/s -      5.243M in   5.049497s
           combined:      3.675M (± 9.6%) i/s -     18.234M in   5.029335s

Comparison:
           combined::  3674598.6 i/s
           original::  1061797.2 i/s - 3.46x  (± 0.00) slower