#!/usr/bin/perl

$HOME = $ENV{HOME} || $ENV{LOGDIR} || (getpwuid($<))[7] || '/';
$CAP = "$HOME/etc/caps";

# if ($] >= 5) { require 'AnyDBM_File.pm'; }

dbmopen(CAP, $CAP, undef) && $dbm++;

while (<DATA>) {
    chop;
    # allow and blank lines comments in data file
    s/#.*//;
    next if /^\s*$/;

    # one or two tab-separated fields
    ($lhs, $rhs) = split(/\t+/);

    # if no rhs, is capitalized lhs
    ($rhs = $lhs) =~ s/(\w+)/\u$1/g 	unless $rhs;

    # be kind to typos
    $lhs =~ s/^\s+//g; $rhs =~ s/^\s+//g;
    $lhs =~ s/\s+$//g; $rhs =~ s/\s+$//g;

    warn "dup: $lhs $rhs $CAP{$lhs}\n" if defined($CAP{$lhs}) && $debug;

    # save in parallel arrays for later sorting
    push(@lhs, $lhs);
    push(@rhs, $rhs);
} 

# this sorting puts "tom christiansen" before "tom"
for $i ( sort {index($lhs[$b],$lhs[$a]) >= 0} 0..$#lhs) {
    # preserve the lhs's whitespace
    $lhs[$i] =~ s/\s+/(\\s+)/g;
    $reg = 0;
    $rhs[$i] =~ s/\s+/'$' . ++$reg/ge;

    # save code for later eval; lhs's at word boundaries only
    $code .= "s/\\b$lhs[$i]\\b/$rhs[$i]/g;\n"
}

print "CODE: $code\n" if $debug;

# read all input into pattern space
$/ = undef; # $_ = <>;

while (<>) {

    tr/A-Z/a-z/	if $squish;

    # find initial letters in sentences, which may have smileys
    s/([.!?](\s*:-\))?)(\n\s*|\s{2,})([a-z])/$1$3\u$4/g;

    # process special case words
    eval $code; die $@ if $@;

    # in case we missed the first word
    s/^(\s*[a-z])/\U$1/;  

    s/(\w+)/$CAP{$1} ? $CAP{$1} : $1/ge;

    print;

}


__END__
colorado springs	Colorado Springs
convex computer corporation	CONVEX Computer Corporation
i
unix	UNIX
tom christiansen	Tom Christiansen
larry wall	Larry Wall
limon
dfw	DFW
grand junction
long beach
san diego
cos	COS
co	CO
ca	CA
tx	TX
snmp	SNMP
cmu	CMU
mit	MIT
rpc	RPC
www	WWW
jeff polk
kolstad
new york
kelm
new jersey
faq	FAQ
robbie
santa clause
cfv	CFV
star trek
pc	PC
mark hess
bsdi BSDI
bsd BSD
sysv SysV
ibm	IBM
united states
brownsville
hill country 
afb	AFB
hondo     
del rio         
san antonio    
rio grande 
brownsville    
cotulla       
harlingen    
mcallen        McAllen
port isabel     
south padre 
corpus christi 
kingsville 
port aransas    
ellington 
houston hobby  
palacios      
port arthur  
dfw airport    
dallas love   
navy dallas  
fort worth 
greenville     
longview       
lufkin        
abilene        
mineral wells 
wichita falls
stephenville
midland      
san angelo  
el paso    
dalhart   
childress
new year's day
new year's eve
jeffy
christmas day
christmas eve
rob
alladin
drew
fred