# This is some code to get an entire web page. Just add
# "require 'webpage.pl';" near the start of your program.
# Then you can &TranslatePage('http://www.somepage.edu');
#
# By Patrick Kellogg (after David Mankin)
package webpage;
use LWP::UserAgent;
require "translation.pl";
$DEFAULT_CGI_METHOD="GET";
$DEFAULT_PROXY="http://cacheflow.cs.colorado.edu:3128";
$AGENT_NAME="FIT/0.1";
##
# TranslatePage(string);
##
sub TranslatePage {
my $TransLang = shift;
my $InputURL = shift;
$_ = $InputURL;
#Trim the white space
s/^\s*(.*?)\s*$/$1/;
if (substr($_,-1,1) ne "/") {
$_ = $_ . "/";
}
if (substr($_,0,3) eq "www") {
$_ = "http://" . $_;
}
$InputURL = $_;
$ua = new LWP::UserAgent;
$ua->agent($AGENT_NAME . " " . $ua->agent);
$ua->proxy('http', $DEFAULT_PROXY) if $DEFAULT_PROXY;
$req = new HTTP::Request 'GET' => $InputURL;
$req->header('Accept' => 'text/html');
$res = $ua->request($req);
if ($res->is_success) {
$NewPage = &FixContent($TransLang,$res->content);
} else {
print "Error: " . $res->code . " " . $res->message;
$NewPage = "";
}
#Do some final adjestments to NewPage
$_ = $NewPage;
s/>/>\n/g;
s/<[Hh][Ee][Aa][Dd]>/\n/g;
$NewPage =$_;
return $NewPage;
}
##
# FixContent(string);
##
sub FixContent {
my $NewLang = shift;
my $PageIn = shift;
#Loop through character-by-character and
#either pass the characters through or translate them
#$OutputPage = "\n";
$Pass = 0;
$FileLen = length $PageIn;
for ($i = 0; $i < $FileLen; $i++) {
$NextChar = substr($PageIn, $i, 1);
#Not translating
if ($Pass == 0) {
#See if we start a new block
if ($NextChar eq '>') {
#Start translating
$MyBuffer = "";
$Pass = 1;
}
$OutputPage = $OutputPage . $NextChar;
#Translating
} else {
#See if we've reached the end of the string
if ($NextChar eq '<') {
#See if anything was caught
$_ = $MyBuffer;
if (!(/^(\s*)$/)) {
#Remove tabs
s/\t//g;
#Remove newlines
s/\n//g;
#Remove carriage returns
s/\r//g;
#Remove formfeeds
s/\f//g;
#Remove any two spaces in a row
s/ / /g;
#Remove any spaces at the start or end
s/^\s*(.*?)\s*$/$1/;
#Don't do a translation
$OutputPage = $OutputPage . $_;
#Do the translation
#print $NewLang . " >" . $_ . "< ", "\n";
#$NewTrans = &translation::translate_lang($NewLang,$_);
#print "NewTrans >" . $NewTrans . "<", "\n";
#$OutputPage = $OutputPage . $NewTrans;
}
$Pass = 0;
$OutputPage = $OutputPage . "<";
} else {
#Add the next character to the buffer
$MyBuffer = $MyBuffer . $NextChar;
}
}
}
return $OutputPage;
}
1;