Implement MacRomanEncoding for real following their own vendor file https://www.unicode.org/Public/MAPPINGS/VENDORS/APPLE/ROMAN.TXT

src/PDF/Encoding/MacRoman.hs

@@ -2,7 +2,140 @@ module PDF.Encoding.MacRoman (
     macRomanEncoding
   ) where
 
+import Data.ByteString.Char8 (unpack)
+import Data.Text (pack)
 import PDF.Font (Font)
 
 macRomanEncoding :: Font
-macRomanEncoding = Right
+macRomanEncoding = Right . pack . fmap decode . unpack
+
+decode :: Char -> Char
+decode '\x80' = '\x00C4' -- LATIN CAPITAL LETTER A WITH DIAERESIS
+decode '\x81' = '\x00C5' -- LATIN CAPITAL LETTER A WITH RING ABOVE
+decode '\x82' = '\x00C7' -- LATIN CAPITAL LETTER C WITH CEDILLA
+decode '\x83' = '\x00C9' -- LATIN CAPITAL LETTER E WITH ACUTE
+decode '\x84' = '\x00D1' -- LATIN CAPITAL LETTER N WITH TILDE
+decode '\x85' = '\x00D6' -- LATIN CAPITAL LETTER O WITH DIAERESIS
+decode '\x86' = '\x00DC' -- LATIN CAPITAL LETTER U WITH DIAERESIS
+decode '\x87' = '\x00E1' -- LATIN SMALL LETTER A WITH ACUTE
+decode '\x88' = '\x00E0' -- LATIN SMALL LETTER A WITH GRAVE
+decode '\x89' = '\x00E2' -- LATIN SMALL LETTER A WITH CIRCUMFLEX
+decode '\x8A' = '\x00E4' -- LATIN SMALL LETTER A WITH DIAERESIS
+decode '\x8B' = '\x00E3' -- LATIN SMALL LETTER A WITH TILDE
+decode '\x8C' = '\x00E5' -- LATIN SMALL LETTER A WITH RING ABOVE
+decode '\x8D' = '\x00E7' -- LATIN SMALL LETTER C WITH CEDILLA
+decode '\x8E' = '\x00E9' -- LATIN SMALL LETTER E WITH ACUTE
+decode '\x8F' = '\x00E8' -- LATIN SMALL LETTER E WITH GRAVE
+decode '\x90' = '\x00EA' -- LATIN SMALL LETTER E WITH CIRCUMFLEX
+decode '\x91' = '\x00EB' -- LATIN SMALL LETTER E WITH DIAERESIS
+decode '\x92' = '\x00ED' -- LATIN SMALL LETTER I WITH ACUTE
+decode '\x93' = '\x00EC' -- LATIN SMALL LETTER I WITH GRAVE
+decode '\x94' = '\x00EE' -- LATIN SMALL LETTER I WITH CIRCUMFLEX
+decode '\x95' = '\x00EF' -- LATIN SMALL LETTER I WITH DIAERESIS
+decode '\x96' = '\x00F1' -- LATIN SMALL LETTER N WITH TILDE
+decode '\x97' = '\x00F3' -- LATIN SMALL LETTER O WITH ACUTE
+decode '\x98' = '\x00F2' -- LATIN SMALL LETTER O WITH GRAVE
+decode '\x99' = '\x00F4' -- LATIN SMALL LETTER O WITH CIRCUMFLEX
+decode '\x9A' = '\x00F6' -- LATIN SMALL LETTER O WITH DIAERESIS
+decode '\x9B' = '\x00F5' -- LATIN SMALL LETTER O WITH TILDE
+decode '\x9C' = '\x00FA' -- LATIN SMALL LETTER U WITH ACUTE
+decode '\x9D' = '\x00F9' -- LATIN SMALL LETTER U WITH GRAVE
+decode '\x9E' = '\x00FB' -- LATIN SMALL LETTER U WITH CIRCUMFLEX
+decode '\x9F' = '\x00FC' -- LATIN SMALL LETTER U WITH DIAERESIS
+decode '\xA0' = '\x2020' -- DAGGER
+decode '\xA1' = '\x00B0' -- DEGREE SIGN
+decode '\xA2' = '\x00A2' -- CENT SIGN
+decode '\xA3' = '\x00A3' -- POUND SIGN
+decode '\xA4' = '\x00A7' -- SECTION SIGN
+decode '\xA5' = '\x2022' -- BULLET
+decode '\xA6' = '\x00B6' -- PILCROW SIGN
+decode '\xA7' = '\x00DF' -- LATIN SMALL LETTER SHARP S
+decode '\xA8' = '\x00AE' -- REGISTERED SIGN
+decode '\xA9' = '\x00A9' -- COPYRIGHT SIGN
+decode '\xAA' = '\x2122' -- TRADE MARK SIGN
+decode '\xAB' = '\x00B4' -- ACUTE ACCENT
+decode '\xAC' = '\x00A8' -- DIAERESIS
+decode '\xAD' = '\x2260' -- NOT EQUAL TO
+decode '\xAE' = '\x00C6' -- LATIN CAPITAL LETTER AE
+decode '\xAF' = '\x00D8' -- LATIN CAPITAL LETTER O WITH STROKE
+decode '\xB0' = '\x221E' -- INFINITY
+decode '\xB1' = '\x00B1' -- PLUS-MINUS SIGN
+decode '\xB2' = '\x2264' -- LESS-THAN OR EQUAL TO
+decode '\xB3' = '\x2265' -- GREATER-THAN OR EQUAL TO
+decode '\xB4' = '\x00A5' -- YEN SIGN
+decode '\xB5' = '\x00B5' -- MICRO SIGN
+decode '\xB6' = '\x2202' -- PARTIAL DIFFERENTIAL
+decode '\xB7' = '\x2211' -- N-ARY SUMMATION
+decode '\xB8' = '\x220F' -- N-ARY PRODUCT
+decode '\xB9' = '\x03C0' -- GREEK SMALL LETTER PI
+decode '\xBA' = '\x222B' -- INTEGRAL
+decode '\xBB' = '\x00AA' -- FEMININE ORDINAL INDICATOR
+decode '\xBC' = '\x00BA' -- MASCULINE ORDINAL INDICATOR
+decode '\xBD' = '\x03A9' -- GREEK CAPITAL LETTER OMEGA
+decode '\xBE' = '\x00E6' -- LATIN SMALL LETTER AE
+decode '\xBF' = '\x00F8' -- LATIN SMALL LETTER O WITH STROKE
+decode '\xC0' = '\x00BF' -- INVERTED QUESTION MARK
+decode '\xC1' = '\x00A1' -- INVERTED EXCLAMATION MARK
+decode '\xC2' = '\x00AC' -- NOT SIGN
+decode '\xC3' = '\x221A' -- SQUARE ROOT
+decode '\xC4' = '\x0192' -- LATIN SMALL LETTER F WITH HOOK
+decode '\xC5' = '\x2248' -- ALMOST EQUAL TO
+decode '\xC6' = '\x2206' -- INCREMENT
+decode '\xC7' = '\x00AB' -- LEFT-POINTING DOUBLE ANGLE QUOTATION MARK
+decode '\xC8' = '\x00BB' -- RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK
+decode '\xC9' = '\x2026' -- HORIZONTAL ELLIPSIS
+decode '\xCA' = '\x00A0' -- NO-BREAK SPACE
+decode '\xCB' = '\x00C0' -- LATIN CAPITAL LETTER A WITH GRAVE
+decode '\xCC' = '\x00C3' -- LATIN CAPITAL LETTER A WITH TILDE
+decode '\xCD' = '\x00D5' -- LATIN CAPITAL LETTER O WITH TILDE
+decode '\xCE' = '\x0152' -- LATIN CAPITAL LIGATURE OE
+decode '\xCF' = '\x0153' -- LATIN SMALL LIGATURE OE
+decode '\xD0' = '\x2013' -- EN DASH
+decode '\xD1' = '\x2014' -- EM DASH
+decode '\xD2' = '\x201C' -- LEFT DOUBLE QUOTATION MARK
+decode '\xD3' = '\x201D' -- RIGHT DOUBLE QUOTATION MARK
+decode '\xD4' = '\x2018' -- LEFT SINGLE QUOTATION MARK
+decode '\xD5' = '\x2019' -- RIGHT SINGLE QUOTATION MARK
+decode '\xD6' = '\x00F7' -- DIVISION SIGN
+decode '\xD7' = '\x25CA' -- LOZENGE
+decode '\xD8' = '\x00FF' -- LATIN SMALL LETTER Y WITH DIAERESIS
+decode '\xD9' = '\x0178' -- LATIN CAPITAL LETTER Y WITH DIAERESIS
+decode '\xDA' = '\x2044' -- FRACTION SLASH
+decode '\xDB' = '\x20AC' -- EURO SIGN
+decode '\xDC' = '\x2039' -- SINGLE LEFT-POINTING ANGLE QUOTATION MARK
+decode '\xDD' = '\x203A' -- SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
+decode '\xDE' = '\xFB01' -- LATIN SMALL LIGATURE FI
+decode '\xDF' = '\xFB02' -- LATIN SMALL LIGATURE FL
+decode '\xE0' = '\x2021' -- DOUBLE DAGGER
+decode '\xE1' = '\x00B7' -- MIDDLE DOT
+decode '\xE2' = '\x201A' -- SINGLE LOW-9 QUOTATION MARK
+decode '\xE3' = '\x201E' -- DOUBLE LOW-9 QUOTATION MARK
+decode '\xE4' = '\x2030' -- PER MILLE SIGN
+decode '\xE5' = '\x00C2' -- LATIN CAPITAL LETTER A WITH CIRCUMFLEX
+decode '\xE6' = '\x00CA' -- LATIN CAPITAL LETTER E WITH CIRCUMFLEX
+decode '\xE7' = '\x00C1' -- LATIN CAPITAL LETTER A WITH ACUTE
+decode '\xE8' = '\x00CB' -- LATIN CAPITAL LETTER E WITH DIAERESIS
+decode '\xE9' = '\x00C8' -- LATIN CAPITAL LETTER E WITH GRAVE
+decode '\xEA' = '\x00CD' -- LATIN CAPITAL LETTER I WITH ACUTE
+decode '\xEB' = '\x00CE' -- LATIN CAPITAL LETTER I WITH CIRCUMFLEX
+decode '\xEC' = '\x00CF' -- LATIN CAPITAL LETTER I WITH DIAERESIS
+decode '\xED' = '\x00CC' -- LATIN CAPITAL LETTER I WITH GRAVE
+decode '\xEE' = '\x00D3' -- LATIN CAPITAL LETTER O WITH ACUTE
+decode '\xEF' = '\x00D4' -- LATIN CAPITAL LETTER O WITH CIRCUMFLEX
+decode '\xF0' = '\xF8FF' -- Apple logo
+decode '\xF1' = '\x00D2' -- LATIN CAPITAL LETTER O WITH GRAVE
+decode '\xF2' = '\x00DA' -- LATIN CAPITAL LETTER U WITH ACUTE
+decode '\xF3' = '\x00DB' -- LATIN CAPITAL LETTER U WITH CIRCUMFLEX
+decode '\xF4' = '\x00D9' -- LATIN CAPITAL LETTER U WITH GRAVE
+decode '\xF5' = '\x0131' -- LATIN SMALL LETTER DOTLESS I
+decode '\xF6' = '\x02C6' -- MODIFIER LETTER CIRCUMFLEX ACCENT
+decode '\xF7' = '\x02DC' -- SMALL TILDE
+decode '\xF8' = '\x00AF' -- MACRON
+decode '\xF9' = '\x02D8' -- BREVE
+decode '\xFA' = '\x02D9' -- DOT ABOVE
+decode '\xFB' = '\x02DA' -- RING ABOVE
+decode '\xFC' = '\x00B8' -- CEDILLA
+decode '\xFD' = '\x02DD' -- DOUBLE ACUTE ACCENT
+decode '\xFE' = '\x02DB' -- OGONEK
+decode '\xFF' = '\x02C7' -- CARON
+decode c = c             -- The rest is ASCII