phantom_url_check.js

// 
//  phantom_test.js
//  bit_spam
//  
//  Created by gregory tomlinson on 2011-04-23.
//  Copyright 2011 the public domain. All rights reserved.
// 
// requires phantomjs (sudo port -v install phantomjs) 
//      -- requires qt4-(os) aka qt4-mac


/*
    Phantom Notes
        Currently, there is no 'great' way to maintain state across
        page reloads, this makes it difficult to track items
        
    The 'magic' phantom.state
        is allowed to maintain a state over multiple loads, however, the datatype of it MAY be limited to strings

*/
var document_active=false,locations_lst=[], page_wait_ms=20,
    last_known_url,
    error_status=false;

function open_connection() {
    
    var args = phantom.args;
    if( args ) {
        
        phantom.state = last_known_url= args[0];
        phantom.userAgent="bitly";

        console.log("Open Connection: " +  last_known_url)
        phantom.open(last_known_url);
    } else {
        phantom.exit(0);
    }
    
    
}
function print_message() {
    console.log("Page Redirect. URL is : " + document.location.href);
}
function on_page_unload(e) {
    document_active=true; // set 'global' var
    // phantom.state.count+=1;
    error_status=true;
    // use document_active as a lazy check for redirects
}

function close_connection() {
    phantom.sleep(page_wait_ms);
    if(!document_active) {
        console.log("Close Connection");
        print_message();
        if(phantom.state !== document.location.href) {
            error_status=true;
        }
        phantom.exit( (phantom.state > 0 || error_status ) ? 1 : 0  );
    }    
}

function main() {
    if (phantom.state.length === 0) {
        console.log("open connection")
        open_connection();
    } else {
        if(phantom.loadStatus == "success") {
            document_active=false;
            window.onbeforeunload=on_page_unload;
            
            console.log("Title: " + document.title);
            console.log("Document: " + document.location.href);
            // todo, search for metarefresh tags
            
            close_connection();
            
        } 
    }
}

main();