\h is a hex digit in Ruby and a blank in Perl

Level: 201 · anyone moving a regex between languages

One line: Onigmo reads \h as a hexadecimal digit, Perl reads it as horizontal whitespace, and Python rejects it — so one pattern matches letters in one language, blanks in the next, and raises in the third; [[:xdigit:]] and [[:blank:]] mean the same in Ruby and Perl.

Verified output of h_is_a_hex_digit_rb.rb — regenerated by tools/run_examples.py, never hand-typed.

line = "color: cafe42\tbg: 00ff00"
  line.scan(/\h+/)             ["c", "cafe42", "b", "00ff00"]
  line.scan(/[[:xdigit:]]+/)   ["c", "cafe42", "b", "00ff00"]
  line.scan(/[[:blank:]]/)     [" ", "\t", " "]
  line.scan(/\H+/).first(2)    ["olor: ", "\t"]

\h+ found cafe42 and 00ff00, and also the c of color and the b of bg: hex digits include six letters. [[:xdigit:]] is the same class, \H its complement, and [[:blank:]] — a space or a tab — is what Perl means by \h.

The same line in Perl

Verified output of h_is_a_hex_digit_pl.pl — regenerated by tools/run_examples.py, never hand-typed.

  $line =~ /\h+/g              [U+0020] [U+0009] [U+0020]
  $line =~ /[[:xdigit:]]+/g    c cafe42 b 00ff00
  $line =~ /[[:blank:]]/g      U+0020 U+0009 U+0020

\h+ matched the three blanks: space, tab, space. [[:xdigit:]] and [[:blank:]] agree with Ruby's.

The same line in Python

Verified output of h_is_a_hex_digit_py.py — regenerated by tools/run_examples.py, never hand-typed.

  re.findall('\\h+', line)           re.error: bad escape \h at position 0
  re.findall('[0-9A-Fa-f]+', line)   ['c', 'cafe42', 'b', '00ff00']

Python's re has no \h and says so, which is the most useful of the three behaviours.

What to do

Write [[:xdigit:]] or [0-9a-fA-F] for a hex digit, and [[:blank:]] or [ \t] for a blank, in any pattern that might be copied into another language. In Ruby-only code, \h is fine; it is the port that breaks.