rss2html.php

RSS to html

//
// Revision 2.34  2006/07/21 12:23:36  housley
// * If there is no encoding, default to ISO-8859-1
// * Modify the XML to show the encoding we used
//
// Revision 2.33  2006/07/16 10:33:23  housley
// Force $useFopenURL if cURL is not installed
//
// Revision 2.32  2006/07/13 17:05:08  housley
// Remove space that causes problems
//
// Revision 2.31  2006/07/12 12:48:27  housley
// Try the iconv() conversion option if mb_string_encode() doesn't exist
//
// Revision 2.30  2006/05/28 17:52:10  housley
// Handle no encoding specified in the feed
//
// Revision 2.29  2006/05/28 17:51:28  housley
// Allow displaying of the Creative Commons License URL
//
// Revision 2.28  2006/05/28 14:21:47  housley
// Add additional capabilities to work with enclosures.  The 3 new fields are 
// ~~~ItemEnclosureType~~~, ~~~ItemEnclosureLength~~~ and ~~~ItemEnclosureLengthFormatted~~~
//
// Revision 2.27  2006/05/27 19:27:45  housley
// * Show a more universal TZ offset example
// * Make setting contentEncoded more robust
//
// Revision 2.26  2006/04/08 23:17:22  housley
// A "%" should not be encoded to "%2525", but to "%25"
//
// Revision 2.25  2006/04/08 23:16:17  housley
// Indicate that this is the last parse of XML
//
// Revision 2.24  2006/03/23 12:10:30  housley
// Add a simple way to change the timezone of produced times and dates
//
// Revision 2.23  2006/03/10 14:21:04  housley
// Update the licenses
//
// Revision 2.22  2006/03/06 15:01:57  housley
// Trim white space before and after the XML
//
// Revision 2.21  2006/03/05 15:15:11  housley
// Rename rss2html_CachingExtension.php to FeedForAll_Scripts_CachingExtension.php
//
// Revision 2.20  2006/03/05 14:43:59  housley
// Fix the testing for the character set conversion function
//
// Revision 2.19  2006/02/28 02:00:04  housley
// Add support for ~~~FeedXMLFilename~~~
//
// Revision 2.18  2006/02/26 15:24:15  housley
// Add the capability to limit the length of feed and item titles and descriptions
//
// Revision 2.17  2006/02/13 18:00:27  housley
// Fix the initialization of the item arrays
//
// Revision 2.16  2006/02/12 14:43:18  housley
// If possible convert the feed to UTF-8, for uniformity
//
// Revision 2.15  2006/02/12 00:21:13  housley
// Fix the offsetting of the time
//
// Revision 2.14  2006/01/26 15:52:37  housley
// Fix the error message for opening a feed, it was displaying the template filename.
//
// Revision 2.13  2006/01/08 23:25:44  housley
// Move all user configuration options at the top of the file to make them
// easier to find
//
// Revision 2.12  2005/12/12 16:27:26  housley
// Add an interface to allow FeedForAll_scripts_readFile() to be replaced
// by one that does caching of the XML files
//
// Revision 2.11  2005/12/09 19:08:26  housley
// Remove the first "banner" since IE barfs
//
// Revision 2.10  2005/10/22 18:51:47  housley
// Improve the formatting
//
// Revision 2.9  2005/10/22 14:27:57  housley
// Fix label in buildURL
//
// Revision 2.8  2005/10/22 14:20:31  housley
// Add buildURL to assist in creating properly encoded links.  Show proper
// include methods and contents of the files.
//
// Revision 2.7  2005/10/16 17:54:10  housley
// Improvements when using CURL:
// - Use the requested file as the REFERER, for sites that might require one
// - Allow to follow up to 10 redirects, some sites redirect to real content
//
// Revision 2.6  2005/10/16 17:32:27  housley
// Use lastBuildDate as another possible source if pubDate is empty at the
// <channel> level.
//
// Revision 2.5  2005/09/28 02:08:15  housley
// Fix the storage of pubDate at the feed level
//
// Revision 2.4  2005/09/12 18:56:31  housley
// Set a user agent for both fopen and curl transerfers
//
// Revision 2.3  2005/09/06 22:55:27  housley
// GUID doesn't need urlencode()
//
// Revision 2.2  2005/08/16 19:53:15  housley
// Add the ~~~ItemAuthor~~~ subsitution that uses first <author> and then
// <dc:creator> for its contents
//
// Revision 2.1  2005/08/15 14:49:24  housley
// Convert &apos; to ' since &apos; is not HTML
//
// Revision 2.0  2005/07/30 14:09:38  housley
// Allow "allow_url_fopen" to be sellected, incase CURL is not available.
//
//

//
// If using cURL, make sure it exists
if (($useFopenURL == 0) && !function_exists("curl_init")) {
  $useFopenURL = -1;
  if (isset($debugLevel) && ($debugLevel >= 3)) {
    echo "DIAG: setting \$useFopenURL=-1 because curl_init() doesn't exist<br>\n";
  }
}

if ($useFopenURL == 1) {
  ini_set("allow_url_fopen", "1");
  ini_set("user_agent", "FeedForAll rss2html scripts v3");
}

$FeedMaxItems = 10000;
$NoFutureItems = FALSE;

@include("FeedForAll_rss2html_pro.php");

if (function_exists("FeedForAll_rss2html_pro") === FALSE) {
  Function FeedForAll_rss2html_pro($source) {
    //
    // This is the place to do any processing that is desired
    return $source;
  }
}

if (function_exists("FeedForAll_parseExtensions") === FALSE) {
  Function FeedForAll_parseExtensions() {
    return FALSE;
  }
}

@include("FeedForAll_Scripts_CachingExtension.php");

@include_once("FeedForAll_XMLParser.inc.php");

if (function_exists("FeedForAll_rss2html_limitLength") === FALSE) {
  Function FeedForAll_rss2html_limitLength($initialValue, $limit = 0) {
    if (($limit == 0) || (strlen($initialValue) <= $limit )) {
      // ZERO is for not limited
      return $initialValue;
    }

    // Cut the text at the exact point, ignoring if it is in a word.
    $result = substr($initialValue, 0, $limit);

    // Check to see if there are any space we can trim at and if it is not
    // too far from where we are
    $lastSpace = strrchr($result,' ');
    if (($lastSpace !== FALSE) && (strlen($lastSpace) < 20)) {
      // lose any incomplete word at the end
      $result = substr($result, 0, -(strlen($lastSpace)));

      // Append elipses, ... , to show it was truncated
      $result .= " ...";
    }

    return $result;
  }
}

if (function_exists("FeedForAll_rss2html_sizeToString") === FALSE) {
  Function FeedForAll_rss2html_sizeToString($filesize) {
    if ($filesize == "") {
      return "";
    }
    elseif ($filesize >= 1073741824) {
      return number_format($filesize/1073741824, 1, ".", ",")." GBytes";
    }
    elseif ($filesize >= 1048576) {
      return number_format($filesize/1048576, 1, ".", ",")." MBytes";
    }
    elseif ($filesize >= 1024) {
      return number_format($filesize/1024, 1, ".", ",")." KBytes";
    }
    else {
      return $filesize." Bytes";
    }
  }
}

if (function_exists("FeedForAll_rss2html_isTemplate") === FALSE) {
  Function FeedForAll_rss2html_isTemplate($templateData) {
    if ((strstr($templateData, "~~~Feed") !== FALSE) || (strstr($templateData, "~~~Item") !== FALSE)) {
      return TRUE;
    }
    return FALSE;
  }
}

if (function_exists("FeedForAll_rss2html_validExtension") === FALSE) {
  Function FeedForAll_rss2html_validExtension($filename, $extensions) {
    $foundValid = FALSE;
    foreach ($extensions as $value) {
      if (strtolower($value) == strtolower(substr($filename, -strlen($value)))) {
        $foundValid = TRUE;
        break;
      }
    }
    return $foundValid;
  }
}

if (function_exists("FeedForAll_rss2html_str_replace") === FALSE) {
  Function FeedForAll_rss2html_str_replace($search, $replace, $subject) {
    return str_replace($search, $replace, $subject);
  }
}

if (function_exists("FeedForAll_rss2html_encodeURL") === FALSE) {
  Function FeedForAll_rss2html_encodeURL($URLstring, $includePND = 0) {
    $result = "";
    for ($x = 0; $x < strlen($URLstring); $x++) {
      if ($URLstring[$x] == '%') {
        $result = $result."%25";
      }
      elseif ($URLstring[$x] == '?') {
        $result = $result."%3f";
      }
      elseif ($URLstring[$x] == '&') {
        $result = $result."%26";
      }
      elseif ($URLstring[$x] == '=') {
        $result = $result."%3d";
      }
      elseif ($URLstring[$x] == '+') {
        $result = $result."%2b";
      }
      elseif ($URLstring[$x] == ' ') {
        $result = $result."%20";
      }
      elseif ($includePND && ($URLstring[$x] == '#')) {
        $result = $result."%23";
      }else {
        $result = $result.$URLstring[$x];
      }
    }
    return $result;
  }
}

if (function_exists("FeedForAll_rss2html_CreateUniqueLink") === FALSE) {
  Function FeedForAll_rss2html_CreateUniqueLink($title, $description, $link, $guid, $XMLfilename, $itemTemplate) {
    GLOBAL $TEMPLATEfilename;
    $match = Array();
    
    while (preg_match("/~~~ItemUniqueLinkWithTemplate=.*~~~/", $itemTemplate, $match) !== FALSE) {
      if ((count($match) == 0) || ($match[0] == "")) {
        // All done
        return $itemTemplate;
      }
      
      $replace = "http://$_SERVER[SERVER_NAME]$_SERVER[SCRIPT_NAME]?XMLFILE=".FeedForAll_rss2html_encodeURL($XMLfilename)."&amp;TEMPLATE=".FeedForAll_rss2html_encodeURL($TEMPLATEfilename);
      $itemTemplate = FeedForAll_rss2html_str_replace($match[0], $replace, $itemTemplate);
    }
    if ($title);
    if ($description);
    if ($link);
    if ($guid);
    return $itemTemplate;
  }
}

if (function_exists("FeedForAll_rss2html_UseUniqueLink") === FALSE) {
  Function FeedForAll_rss2html_UseUniqueLink($title, $description, $link, $guid) {
    if ($title);
    if ($description);
    if ($link);
    if ($guid);
    return -1;
  }
}

if (function_exists("FeedForAll_rss2html_EscapeLink") === FALSE) {
  Function FeedForAll_rss2html_EscapeLink($link) {
    GLOBAL $escapeAmpInLinks;
    
    if ((strstr($link, "://") !== FALSE) && $escapeAmpInLinks) {
      // In HTML a link with an & must be converted to &amp;
      // And for here without :// it is not a link, since relative
      // URLs are not allowed
      $link = str_replace("&", "&amp;", $link);
    }
    return $link;
  }
}

if (function_exists("FeedForAll_rss2html_AddIdentity") === FALSE) {
  Function FeedForAll_rss2html_AddIdentity($itemString) {
    return "<!-- HTML generated from an RSS Feed by rss2html.php, http://www.FeedForAll.com/ a NotePage, Inc. product (http://www.notepage.com/) -->".$itemString;
  }
}

if (!isset($_REQUEST["buildURL"])) {
  //