Tl;dr: Use a regex approach. It is 39x faster than the rescue approach in the accepted answer and also handles cases like "1,000"
def regex_is_number? string
no_commas = string.gsub(',', '')
matches = no_commas.match(/-?\d+(?:\.\d+)?/)
if !matches.nil? && matches.size == 1 && matches[0] == no_commas
true
else
false
end
end
--
The accepted answer by @Jakob S works for the most part, but catching exceptions can be really slow. In addition, the rescue approach fails on a string like "1,000".
Let's define the methods:
def rescue_is_number? string
true if Float(string) rescue false
end
def regex_is_number? string
no_commas = string.gsub(',', '')
matches = no_commas.match(/-?\d+(?:\.\d+)?/)
if !matches.nil? && matches.size == 1 && matches[0] == no_commas
true
else
false
end
end
And now some test cases:
test_cases = {
true => ["5.5", "23", "-123", "1,234,123"],
false => ["hello", "99designs", "(123)456-7890"]
}
And a little code to run the test cases:
test_cases.each do |expected_answer, cases|
cases.each do |test_case|
if rescue_is_number?(test_case) != expected_answer
puts "**rescue_is_number? got #{test_case} wrong**"
else
puts "rescue_is_number? got #{test_case} right"
end
if regex_is_number?(test_case) != expected_answer
puts "**regex_is_number? got #{test_case} wrong**"
else
puts "regex_is_number? got #{test_case} right"
end
end
end
Here is the output of the test cases:
rescue_is_number? got 5.5 right
regex_is_number? got 5.5 right
rescue_is_number? got 23 right
regex_is_number? got 23 right
rescue_is_number? got -123 right
regex_is_number? got -123 right
**rescue_is_number? got 1,234,123 wrong**
regex_is_number? got 1,234,123 right
rescue_is_number? got hello right
regex_is_number? got hello right
rescue_is_number? got 99designs right
regex_is_number? got 99designs right
rescue_is_number? got (123)456-7890 right
regex_is_number? got (123)456-7890 right
Time to do some performance benchmarks:
Benchmark.ips do |x|
x.report("rescue") { test_cases.values.flatten.each { |c| rescue_is_number? c } }
x.report("regex") { test_cases.values.flatten.each { |c| regex_is_number? c } }
x.compare!
end
And the results:
Calculating -------------------------------------
rescue 128.000 i/100ms
regex 4.649k i/100ms
-------------------------------------------------
rescue 1.348k (±16.8%) i/s - 6.656k
regex 52.113k (± 7.8%) i/s - 260.344k
Comparison:
regex: 52113.3 i/s
rescue: 1347.5 i/s - 38.67x slower