It isn't really conform to the specifications but Acrobat is working like that...
charCodeToGlyphId
differences
type1FontGlyphMapping
XRef.indexObjects
PartialEvaluator._buildSimpleFontToUnicode