yahoo_store_to_froogle.php

<?php

/* yahoo_store_export_to_froogle.pl port to PHP
# Author: Anatoly Kudinov <zz@rialabs.org>
# Original description:

# yahoo_store_export_to_froogle.pl
#
# This is a Perl script to convert Yahoo! Stores XML feed into
#   Froogle data feed, and FTP the file to Froogle's server.
# Note: Many stores are automatically indexed by Froogle.
#
# REQUIREMENTS
#  A merchant agreement/ftp account with Froogle.
#  A Yahoo! store account with store export enabled.
#
# INSTRUCTIONS
#   This script requires no command line parameters and creates no output.
#   It places the Froogle data feed into the system temporary directory.
#   Parameters for the store and Froogle user account
#     must be entered in the section labeled "CUSTOM INFORMATION".
#   The script can be tested against a saved Yahoo store's xml file
#     and alternate FTP server by setting the "DEBUG VARIABLES".
#
# FEATURES
#  Fully automated. The Froogle data format is the simple format, not
#  extended format.  Entire store product library is processed.
#  Product code is taken as the Yahoo! store created abbreviation.
#  Image URLs and Section names are handled correctly.
#  All HTML tags are removed from product captions.
#  All tabs, carraige returns, and new lines are replaced with spaces.
#  Only products with a price are sent. Orderable flag is ignored.
#
# LIMITATIONS
#  No special handling for books, music, dvd, etc.
#  No support for quantity pricing or alternate currency.
#  No support for item options.
#  No support for partial uploads.
#
# EXAMPLE CRONTAB ('crontab -e' to edit the crontab)
#] # Send Yahoo Store data to Froogle daily at 6 am
#] 0 6 * * * /home/u/user/bin/yahoo_store_export_to_froogle.pl
#
# LINKS
# Yahoo Store XML DTD
#    http://store.yahoo.com/lib/vw/StoreExport.dtd
# Store export overview
#    http://store.yahoo.com/storexport.html
# Froogle merchant info
#    http://froogle.google.com/froogle/merchants.html
# CPAN Perl documentation
# http://search.cpan.org/
# UTF8 bug under red hat when warning enabled
# http://archive.develooper.com/perl5-porters@perl.org/msg88085.html
#
# AUTHOR
#   Shailesh Humbad, March 21, 2003, http://www.somacon.com
#   This code is hereby granted to the public domain.
*/


// DEBUG VARIABLES
// set debug to 1 to enable debug mode, otherwise set to 0
$debug = 0;
// enter file containing xml feed to use in debug mode
$debug_xmlfilename = "storeexportdebug.xml";
// ftp server to use in debug mode
$debug_username = "";
$debug_password = "";
$debug_ftpserver = "";


// CUSTOM INFORMATION
$froogle_username = "";
$froogle_password = "";
$froogle_ftpserver = "";
$store_xmlfeed_url = "http://store.yahoo.com/storename/objinfo.xml";
$store_xmlfeed_url = "http://store.yahoo.com/ygear-test/objinfo.xml";




// DEFINE XML PARSER SUBROUTINES
$xmlparser = xml_parser_create();
xml_set_element_handler($xmlparser, "tag_start", "tag_end");
xml_set_character_data_handler($xmlparser,"handle_char");

// INITIALIZE VARIABLES
$product = Array();
$productarray = Array();
$context = Array();

if($debug) { echo "Yahoo Store Export to Froogle - Debug.\n"; }

// set the filename for the temporary froogle data file
$froogle_data_file = tmpfile();


// print column names
fwrite($froogle_data_file, "product_url\tname\tdescription\tprice\timage_url\tcategory\tcode\n");



// RETRIEVE THE ENTIRE XML FEED INTO A STRING
if (!$debug)
{
  // allow url fopen should be true, could be ported to curl
 
    $xmldata = file_get_contents($store_xmlfeed_url);
    if(!$xmldata)
      die ("Error while getting ".$store_xmlfeed_url.
	   "\nAborting");

}
else
{
  $xmldata = file_get_contents($debug_xml_filename);
}

// RUN THE XML PARSER (PARSING IS DONE IN THE SUBROUTINES)
xml_parse($xmlparser, $xmldata);


// WRITE THE PARSED DATA TO THE TEMP FILE
foreach($productarray as $product)
{
  fwrite($froogle_data_file, $product['product_url']."\t");
  fwrite($froogle_data_file, $product['name']."\t");
  fwrite($froogle_data_file, $product['description']."\t");
  fwrite($froogle_data_file, $product['price']."\t");
  fwrite($froogle_data_file, $product['image_url']."\t");
  fwrite($froogle_data_file, $product['category']."\t");
  fwrite($froogle_data_file, $product['code']."\t");
  fwrite($froogle_data_file, "\n");
}

if($debug)
{
    $froogle_ftpserver = $debug_ftpserver;
    $froogle_username = $debug_username;
    $froogle_password = $debug_password;
}

echo "Generated csv: \n";
fseek($froogle_data_file,0);
while($str = fgets($froogle_data_file))
  echo $str;
/*
# UPLOAD THE FROOGLE FORMAT DATA TO FROOGLE
$ftp = Net::FTP->new
(
    $froogle_ftpserver,
    Timeout => 30
) or die "Could not connect to FTP server: $froogle_ftpserver.\n";

$ftp->login($froogle_username, $froogle_password)
    or die "Could not log in to FTP server.\n";

$ftp->put($froogle_data_filename);

$ftp->quit();
*/
// END OF SCRIPT

// --------- XML ROUTINES ---------

// XML TAG START ROUTINE
// Start (Parser, Element [, Attr, Val [,...]])
function tag_start($parser, $tagvalue, $attrs) {
  global $product, $context;
  $tagvalue = strtolower($tagvalue);
  $tagvalue = ucfirst($tagvalue);

  if ($tagvalue ==  "Product") {
    // reset the product data
    $product['product_url'] = "";
    $product['name'] = "";
    $product['description'] = "";
    $product['price'] = "";
    $product['image_url'] = "";
    $product['orderable'] = "";
    $product['category'] = "";
    $product['code'] = $attrs['ID'];
  }
  $context[] = $tagvalue;
}

// XML CHARACTER DATA ROUTINE
function handle_char($parser, $tagvalue) {
  // Retrieve context
  global $context, $product;

  $_context = implode(" ", $context);
  // convert the XML feed to froogle format
  // froogle attributes
  if ($_context == "Storeexport Products Product Url") {
    $product['product_url'] .= $tagvalue;
  }
  if ($_context == "Storeexport Products Product Description") {
        $product['name'] .= $tagvalue;
  }
  if ($_context == "Storeexport Products Product Caption") {
        $product['description'] .= $tagvalue;
  }
  if ($_context == "Storeexport Products Product Pricing Baseprice") {
        $product['price'] .= $tagvalue;
  }
  if ($_context == "Storeexport Products Product Picture") {
    // must be concatenation due to &gt;
    $product['image_url'] .= $tagvalue;
  }
  // meta attributes
  if ($_context == "Storeexport Products Product Orderable") {
        $product['orderable'] .= $tagvalue;
  }
  if ($_context == "Storeexport Products Product Path Productref") {
    // must be concatenation to create category name
    // and to combine names with '&'
    $product['category'] .= $tagvalue;
  }
}

// XML TAG END ROUTINE
function tag_end($parser, $tagvalue) {
  global $product, $productarray, $context;
  $tagvalue = strtolower($tagvalue);
  $tagvalue = ucfirst($tagvalue);

  if($tagvalue == "Product") {
    // CLEAN UP THE PRODUCT DATA
        
    // trim off the trailing separator for the category
    if(strlen($product['category']) > 0) {
      $product['category'] = substr($product['category'], 0, 
                    strlen($product['category'])-3);
    }
    // extract the URL for the image
    $matches = Array();
    preg_match("/.*?src\=(.*?)\>/i", $product['image_url'], $matches);
    if(@$matches[1]) {
      $product['image_url'] = $matches[1];
    }
    
    // replace all cr, lf, and tab with spaces
    // in all fields of the hash
    foreach ($product as  $key=>$value) {
      if($product[$key]) {
	$product[$key] = str_replace(Array("\t","\n","\r"), " ",$product[$key]);
      }
    }
        
    // replace all html tags with empty string
    $product['description'] = strip_tags($product['description']);
    
    // PUSH THE PREVIOUS PRODUCT
    if($product['price']) {
      // do not push products without a price
      // these are either sections or unpriced items
      $productarray[]= Array(
                'code' => $product['code'],
                'name' => $product['name'],
                'description' => $product['description'],
                'product_url' => $product['product_url'],
                'image_url' => $product['image_url'],
                'price' => $product['price'],
                'orderable' => $product['orderable'],
                'category' => $product['category']);
	
        }
  }
    
  // add separator for each ProductRef to create section name
  if($tagvalue == "Productref") {
    $product['category'] .= " > ";
  }
  array_pop($context);
}