# Input: one or more lexical entries of the form
#
# lex(Word, CATEGORY).
#
# Word is a string; CATEGORY is actually a colon-separated pair of
# all-caps categories.
#
# We need to convert Word into a prolog atom: a string that starts
# with a lowercase letter and only contains lowercase, uppercase,
# numerals, and the underscore.
#
# In addition, I get rid of the umlauts, etc. at this point, mostly
# because I don't understand all of the encoding issues involved (i.e.
# which tools support which encodings to which degree).
#
# So all non-first-26-alphabetics/non-numerics (umlauted letters,
# hyphens, etc.) are converted into things like "_27_". Then, if the
# result doesn't start with a lowercase letter, the string "zz__" is
# prepended. As a result, we have a reversible transformation.
#
# The worst case for input is probably something like:
#
# lex(,, $,_NMC).
#
# which becomes
#
# lex(zz___44_, zz___36__44__nmc).

while (<>) {
    ($word, $cat) = /^lex\((.*), (.*)\).$/;
    $word =~ s/([^A-Za-z0-9])/"_".ord($1)."_"/eg;
    if (substr($word, 0, 1) =~ /[^a-z]/) {
	substr($word, 0, 0) = "zz__";
    }
    $cat =~ tr/A-Z/a-z/;
    $cat =~ s/([^A-Za-z0-9_])/"_".ord($1)."_"/eg;
    if (substr($cat, 0, 1) =~ /[^a-z]/) {
	substr($cat, 0, 0) = "zz__";
    }
    print "lex($word, $cat).\n";
}

# I've never figured out how to type umlauts in my version of emacs,
# but I can copy-and-paste them without problems. So these stay here
# in case of future need.
#
#    $word =~ s/Ä/Ae/g;
#    $word =~ s/Ö/Oe/g;
#    $word =~ s/Ü/Ue/g;
#    $word =~ s/ä/ae/g;
#    $word =~ s/ö/oe/g;
#    $word =~ s/ü/ue/g;
#    $word =~ s/ß/ss/g;