diff options
Diffstat (limited to 'normchr')
| -rwxr-xr-x | normchr | 9 |
1 files changed, 4 insertions, 5 deletions
@@ -3,10 +3,10 @@ # http://www.utf8-chartable.de/unicode-utf8-table.pl?start=8192&number=128&utf8=string-literal # https://www.cs.tut.fi/~jkorpela/chars/spaces.html -require 'htmlentities' +require "htmlentities" -STDIN.set_encoding 'utf-8' -STDOUT.set_encoding 'utf-8' +STDIN.set_encoding "utf-8" +STDOUT.set_encoding "utf-8" coder = HTMLEntities.new @@ -24,7 +24,7 @@ while line = STDIN.gets line.gsub! /[\u{e000}-\u{f8ff}]/, " " # UTF-8 PUA line.gsub! /[\u{f0000}-\u{ffffd}]/, " " line.gsub! /[\u{100000}-\u{10fffd}]/, " " - line.gsub! "\r", " " # carriage return + line.gsub! "\r", " " # carriage return line.gsub! /[\u{2000}-\u{200f}]/, " " # EN QUAD -- RIGHT-TO-LEFT MARK line.gsub! /[\u{2028}-\u{202f}]/, " " # LINE SEPARATOR -- NARROW NO-BREAK SPACE line.gsub! /[\u{205f}-\u{206f}]/, " " # MEDIUM MATHEMATICAL SPACE -- NOMINAL DIGIT SHAPES @@ -32,4 +32,3 @@ while line = STDIN.gets line.gsub! /[[:space:]]+/, " " # collapse space puts coder.decode(line) end - |
