1use strict; 2use warnings; 3 4# A tiny private library routine which is a helper to several Perl core 5# modules, to allow a paradigm to be implemented in a single place. The name, 6# contents, or even the existence of this file may be changed at any time and 7# are NOT to be used by anything outside the Perl core. 8 9sub _meta_notation ($) { 10 11 # Returns a copy of the input string with the nonprintable characters 12 # below 0x100 changed into printables. Any ASCII printables or above 0xFF 13 # are unchanged. (XXX Probably above-Latin1 characters should be 14 # converted to \X{...}) 15 # 16 # \0 .. \x1F (which are "\c@" .. "\c_") are changed into ^@, ^A, ^B, ... 17 # ^Z, ^[, ^\, ^], ^^, ^_ 18 # \c? is changed into ^?. 19 # 20 # The above accounts for all the ASCII-range nonprintables. 21 # 22 # On ASCII platforms, the upper-Latin1-range characters are converted to 23 # Meta notation, so that \xC1 becomes 'M-A', \xE2 becomes 'M-b', etc. 24 # This is how it always has worked, so is continued that way for backwards 25 # compatibility. The range \x80 .. \x9F becomes M-^@ .. M-^A, M-^B, ... 26 # M-^Z, M-^[, M-^\, M-^], M-^, M-^_ 27 # 28 # On EBCDIC platforms, the upper-Latin1-range characters are converted 29 # into '\x{...}' Meta notation doesn't make sense on EBCDIC platforms 30 # because the ASCII-range printables are a mixture of upper bit set or 31 # not. [A-Za-Z0-9] all have the upper bit set. The underscore likely 32 # doesn't; and other punctuation may or may not. There's no simple 33 # pattern. 34 35 my $string = shift; 36 37 $string =~ s/([\0-\037])/ 38 sprintf("^%c",utf8::unicode_to_native(ord($1)^64))/xeg; 39 $string =~ s/\c?/^?/g; 40 if (ord("A") == 65) { 41 $string =~ s/([\200-\237])/sprintf("M-^%c",(ord($1)&0177)^64)/eg; 42 $string =~ s/([\240-\377])/sprintf("M-%c" ,ord($1)&0177)/eg; 43 } 44 else { 45 no warnings 'experimental::regex_sets'; 46 # Leave alone things above \xff 47 $string =~ s/( (?[ [\x00-\xFF] & [:^print:]])) / 48 sprintf("\\x{%X}", ord($1))/xaeg; 49 } 50 51 return $string; 52} 531 54