#!/usr/bin/perl # web2elot.pl -- Version 0.02a 98.02.02 18:34 # (c) 1997,1998 Hellenic Resources Institute, Inc. while(<>) { # Get rid of ^M's and trailing spaces... s/\015$//g; s/\s*$//g; # No need for these to be HTML-compliant s/"/\"/g; s/&/&/g; s/ / /g; # ISO-8859-1 8bit codes s/À/\300/g; s/Á/\301/g; s/Â/\302/g; s/Ã/\303/g; s/Ä/\304/g; s/Å/\305/g; s/Æ/\306/g; s/Ç/\307/g; s/È/\310/g; s/É/\311/g; s/Ê/\312/g; s/Ë/\313/g; s/Ì/\314/g; s/Í/\315/g; s/Î/\316/g; s/Ï/\317/g; s/Ð/\320/g; s/Ñ/\321/g; s/Ò/\322/g; s/Ó/\323/g; s/Ô/\324/g; s/Õ/\325/g; s/Ö/\326/g; s/Ø/\330/g; s/Ù/\331/g; s/Ú/\332/g; s/Û/\333/g; s/Ü/\334/g; s/Ý/\335/g; s/Þ/\336/g; s/ß/\337/g; s/à/\340/g; s/á/\341/g; s/â/\342/g; s/ã/\343/g; s/ä/\344/g; s/å/\345/g; s/æ/\346/g; s/ç/\347/g; s/è/\350/g; s/é/\351/g; s/ê/\352/g; s/ë/\353/g; s/ì/\354/g; s/í/\355/g; s/î/\356/g; s/ï/\357/g; s/ð/\360/g; s/ñ/\361/g; s/ò/\362/g; s/ó/\363/g; s/ô/\364/g; s/õ/\365/g; s/ö/\366/g; s/ø/\370/g; s/ù/\371/g; s/ú/\372/g; s/û/\373/g; s/ü/\374/g; s/ý/\375/g; s/þ/\376/g; s/ÿ/\377/g; # Greek Chars without ISO-8859-1 equivalents s/×/\327/g; s/÷/\367/g; s/¢/\266/g; # Unicode s/Α/Á/g; s/Β/Â/g; s/Γ/Ø/g; s/Δ/Ä/g; s/Ε/Å/g; s/Ζ/Æ/g; s/Η/Ç/g; s/Θ/È/g; s/Ι/É/g; s/Κ/Ê/g; s/Λ/Ë/g; s/Μ/Ì/g; s/Ν/Í/g; s/Ξ/Î/g; s/Ο/Ï/g; s/Π/Ð/g; s/Ρ/Ñ/g; s/Σ/Ó/g; s/Σ/Ó/g; s/Τ/Ô/g; s/Υ/Õ/g; s/Φ/Ö/g; s/Χ/×/g; s/Ψ/Õ/g; s/Ω/Ù/g; s/Ά/¢/g; s/Έ/¸/g; s/Ή/¹/g; s/Ί/º/g; s/Ό/¼/g; s/Ύ/¾/g; s/Ώ/¿/g; s/Ϊ/Ú/g; s/Ϋ/Û/g; s/α/á/g; s/β/â/g; s/γ/ã/g; s/δ/ä/g; s/ε/å/g; s/ζ/æ/g; s/η/ç/g; s/θ/è/g; s/ι/é/g; s/κ/ê/g; s/λ/ë/g; s/μ/ì/g; s/ν/í/g; s/ξ/î/g; s/ο/ï/g; s/π/ð/g; s/ρ/ñ/g; s/ς/ò/g; s/σ/ó/g; s/τ/ô/g; s/υ/õ/g; s/φ/ö/g; s/χ/÷/g; s/ψ/ø/g; s/ω/ù/g; s/ά/Ü/g; s/έ/Ý/g; s/ή/Þ/g; s/ί/ß/g; s/ό/ü/g; s/ύ/ý/g; s/ώ/þ/g; s/ϊ/ú/g; s/ϋ/û/g; print; }