feedforall_xmlparser.inc.php

RSS to SQL to store in Database

    }
    elseif ($elementName == "~~~ItemRssMeshFeedImageDescription~~~") {
      return $this->rssMeshFeedImageDescription;
    }
    elseif ($elementName == "~~~ItemRssMeshFeedImageHeight~~~") {
      return $this->rssMeshFeedImageHeight;
    }
    elseif ($elementName == "~~~ItemRssMeshFeedImageWidth~~~") {
      return $this->rssMeshFeedImageWidth;
    }
    return NULL;
  }

  Function getArrayOfFields() {
    $result = Array();

    $result[] = "~~~ItemTitle~~~";
    $result[] = "~~~ItemDescription~~~";
    $result[] = "~~~ItemContentEncoded~~~";
    $result[] = "~~~ItemLink~~~";
    $result[] = "~~~ItemPubDate~~~";
    $result[] = "~~~ItemPubDateAsNumber~~~";
    $result[] = "~~~ItemEnclosureUrl~~~";
    $result[] = "~~~ItemEnclosureType~~~";
    $result[] = "~~~ItemEnclosureLength~~~";
    $result[] = "~~~ItemGuid~~~";
    $result[] = "~~~ItemAuthor~~~";
    $result[] = "~~~ItemComments~~~";
    $result[] = "~~~ItemSource~~~";
    $result[] = "~~~ItemSourceUrl~~~";
    $result[] = "~~~ItemCategory~~~";
    $result[] = "~~~ItemCategoryDomain~~~";
    $result[] = "~~~ItemCreativeCommons~~~";
    $result[] = "~~~ItemRssMeshExtra~~~";
    $result[] = "~~~ItemRssMeshFeedImageTitle~~~";
    $result[] = "~~~ItemRssMeshFeedImageUrl~~~";
    $result[] = "~~~ItemRssMeshFeedImageLink~~~";
    $result[] = "~~~ItemRssMeshFeedImageDescription~~~";
    $result[] = "~~~ItemRssMeshFeedImageHeight~~~";
    $result[] = "~~~ItemRssMeshFeedImageWidth~~~";
    
    return $result;
  }
  
}

$startingClassName = "rootItemClass";
if (function_exists("rssFilter_extendClass")) {
  $startingClassName = rssFilter_extendClass($startingClassName);
}

if (function_exists("FeedForAll_parseExtensions_extendClass")) {
  $currentBaseClassName = FeedForAll_parseExtensions_extendClass($startingClassName);
} else {
  $currentBaseClassName = $startingClassName;
}
eval('class baseItemClassWithExtensions extends ' . $currentBaseClassName . ' {}');

class baseItemClass extends baseItemClassWithExtensions {
  Function baseItemClass($operateAs) {
    $parentClass = get_parent_class($this);
    $this->$parentClass($operateAs);
  }
}

class rootRSSParserClass {
  var $operateAs;
  var $gotROOT;
  var $feedTYPE;
  var $wholeString;
  var $level;
  var $tag;
  var $noFutureItems;
  
  var $currentItem;

  var $FeedTitle;
  var $FeedDescription;
  var $FeedContentEncoded;
  var $FeedLink;
  var $FeedPubDate;
  var $FeedPubDateDC;
  var $FeedPubDate_t;
  var $FeedLastBuildDate;
  var $FeedImageURL;
  var $FeedImageTitle;
  var $FeedImageLink;
  var $FeedImageDescription;
  var $FeedImageHeight;
  var $FeedImageWidth;
  var $FeedCreativeCommons;
  var $FeedAtomUpdated;
  var $FeedAtomContent;
  var $FeedAtomContentStartPos;
  var $FeedAtomAuthorEmail;
  
  var $contentEncodedUsed;

  var $Items;

  //
  var $insideChannel = FALSE;
  var $level_channel;
  var $insideChannelImage = FALSE;
  var $level_channelImage;
  var $insideItem = FALSE;
  var $level_item;
  var $insideAtomAuthor = FALSE;

  Function rootRSSParserClass($operateAs) {
    $this->operateAs = $operateAs;
    $this->gotROOT = 0;
    $this->feedTYPE = "RSS";
    $this->wholeString = "";
    $this->level = 0;
    $this->tag = "";
    $this->noFutureItems = 0;;
  
    $this->FeedImageURL = "";
    $this->FeedImageTitle = "";
    $this->FeedImageLink = "";
    $this->FeedImageDescription = "";
    $this->FeedImageHeight = "";
    $this->FeedImageWidth = "";
    $this->currentItem;

    $this->FeedTitle = "";
    $this->FeedDescription = "";
    $this->FeedContentEncoded = "";
    $this->FeedLink = "";
    $this->FeedPubDate = "";
    $this->FeedPubDateDC = "";
    $this->FeedPubDate_t = 0;
    $this->FeedLastBuildDate = "";
    $this->FeedImageURL = "";
    $this->FeedImageTitle = "";
    $this->FeedImageLink = "";
    $this->FeedImageDescription = "";
    $this->FeedImageHeight = "";
    $this->FeedImageWidth = "";
    $this->FeedCreativeCommons = "";
    $this->FeedAtomUpdated = "";
    $this->FeedAtomContent = "";
    $this->FeedAtomContentStartPos = 0;
    $this->FeedAtomAuthorEmail = "";

    $this->contentEncodedUsed = 0;
    
    $this->Items = Array();

    //
    $this->insideChannel = FALSE;
    $this->level_channel = 0;
    $this->insideChannelImage = FALSE;
    $this->level_channelImage = 0;
    $this->insideItem = FALSE;
    $this->level_item = 0;
  }

  function startElement($parser, $tagName, $attrs) {
    GLOBAL $debugLevel;
    
    $this->level++;
    $this->tag = $tagName;
    if ($this->gotROOT == 0) {
      $this->gotROOT = 1;
      if (strstr($tagName, "RSS")) {
        $this->feedTYPE = "RSS";
      }
      elseif (strstr($tagName, "RDF")) {
        $this->feedTYPE = "RDF";
      }
      elseif (strstr($tagName, "FEE")) {
        $this->feedTYPE = "FEE";
        $this->insideChannel = TRUE;
        $this->level_channel = 1;
      }
    }
    elseif ((($tagName == "ITEM") && ($this->feedTYPE != "FEE")) || (($tagName == "ENTRY") && ($this->feedTYPE == "FEE"))) {
      if (isset($debugLevel) && ($debugLevel >= 2)) {
        echo "DIAG: startElement(\$parser, $tagName, \$attrs)<br>\n";
      }
      
      $this->insideItem = TRUE;
      $this->level_item = $this->level;
      $this->currentItem = new baseItemClass($this->operateAs);

      //
      // Find the start of the <item> or <entry>
      $this->currentItem->ItemStartPos = xml_get_current_byte_index($parser);
      if ($this->wholeString[$this->currentItem->ItemStartPos] != "<") {
        $startToHere = substr($this->wholeString, 0, $this->currentItem->ItemStartPos);
        $this->currentItem->ItemStartPos = strrpos($startToHere, "<");
      }
    }
    elseif ($this->insideChannel && (($tagName == "AUTHOR") && ($this->feedTYPE == "FEE"))) {
      $this->insideAtomAuthor = TRUE;
    }
    elseif ($this->insideItem && (($tagName == "AUTHOR") && ($this->feedTYPE == "FEE"))) {
      $this->insideAtomAuthor = TRUE;
    }
    elseif (($this->insideItem) && ($tagName == "ENCLOSURE")) {
      if (isset($attrs["URL"])) {
        $this->currentItem->enclosureURL = $attrs["URL"];
      }
      if (isset($attrs["TYPE"])) {
        $this->currentItem->enclosureType = $attrs["TYPE"];
      }
      if (isset($attrs["LENGTH"])) {
        $this->currentItem->enclosureLength = $attrs["LENGTH"];
      }
    }
    elseif (($this->insideItem) && ($tagName == "SOURCE")) {
      if (isset($attrs["URL"])) {
        $this->currentItem->sourceURL = $attrs["URL"];
      }
    }
    elseif (($this->insideItem) && ($tagName == "CATEGORY")) {
      if (isset($attrs["DOMAIN"])) {
        $this->currentItem->categoryDomain = $attrs["DOMAIN"];
      }
    }
    elseif (($this->insideItem) && ($tagName == "GUID")) {
      if (isset($attrs["ISPERMALINK"])) {
        $this->currentItem->guidIsPermaLink = $attrs["ISPERMALINK"];
      }
    }
    elseif (($tagName == "LINK") && ($this->feedTYPE == "FEE")) {
      if ($this->insideItem) {
        if (isset($attrs["REL"]) && ($attrs["REL"] == "enclosure")) {
          $this->currentItem->enclosureURL = $attrs["HREF"];
          $this->currentItem->enclosureType = $attrs["TYPE"];
          $this->currentItem->enclosureLength = $attrs["LENGTH"];
        }
        elseif (isset($attrs["HREF"]) && ((isset($attrs["TYPE"]) && ($attrs["TYPE"] == "text/html")) || !isset($attrs["TYPE"]))) {
          $this->currentItem->link = $attrs["HREF"];
        }
      } else {
        if (isset($attrs["HREF"]) && ((isset($attrs["TYPE"]) && ($attrs["TYPE"] == "text/html")) || !isset($attrs["TYPE"]))) {
          $this->FeedLink = $attrs["HREF"];
        }
      }
    }
    elseif ($tagName == "CHANNEL") {
      $this->insideChannel = TRUE;
      $this->level_channel = $this->level;
    }
    elseif (($tagName == "IMAGE") && ($this->insideChannel == TRUE)) {
      $this->insideChannelImage = TRUE;
      $this->level_channelImage = $this->level;
    }
    elseif ($tagName == "CONTENT") {
      if ($this->insideItem == TRUE) {
        if (isset($attrs["TYPE"]) && ($attrs["TYPE"] == "xhtml")) {
          //
          // Find the start of the <content ... >
          $this->currentItem->atomContentStartPos = xml_get_current_byte_index($parser);
          if ($this->wholeString[$this->currentItem->atomContentStartPos] != "<") {
            $startToHere = substr($this->wholeString, 0, $this->currentItem->atomContentStartPos);
            $this->currentItem->atomContentStartPos = strrpos($startToHere, "<");
          }
        }
      } else {
        if (isset($attrs["TYPE"]) && ($attrs["TYPE"] == "xhtml")) {
          //
          // Find the start of the <content ... >
          $this->FeedAtomContentStartPos = xml_get_current_byte_index($parser);
          if ($this->wholeString[$this->FeedAtomContentStartPos] != "<") {
            $startToHere = substr($this->wholeString, 0, $this->FeedAtomContentStartPos);
            $this->FeedAtomContentStartPos = strrpos($startToHere, "<");
          }
        }
      }
    }
    if (FeedForAll_parseExtensions() === TRUE) {
      FeedForAll_parseExtensions_startElemend($parser, $this, $tagName, $attrs);
    }
  }

  function endElement($parser, $tagName) {
    GLOBAL $debugLevel;

    $this->tag = "";
    $this->level--;
    if (($this->insideItem) && ($tagName == "CATEGORY")) {
      $this->currentItem->categoryArray[] = Array("Category" => $this->currentItem->category, "Domain" => $this->currentItem->categoryDomain);
      $this->currentItem->category = "";
      $this->currentItem->categoryDomain = "";
    }
    elseif ((($tagName == "ITEM") && ($this->feedTYPE != "FEE")) || (($tagName == "ENTRY") && ($this->feedTYPE == "FEE"))) {
      if (isset($debugLevel) && ($debugLevel >= 2)) {
        echo "DIAG: endElement(\$parser, $tagName)<br>\n";
      }
      
      $this->UseItem = TRUE;

      //
      // Do any special processing to convert ATOM to RSS 2.0
      if ($this->feedTYPE == "FEE") {
        if ($this->currentItem->guid == "") {
          // There was no GUID, use ID
          $this->currentItem->guid = $this->currentItem->atomID;
          $this->currentItem->guidIsPermaLink = "false";
        }
      }
      
      //
      // The the whole item string
      $pos = xml_get_current_byte_index($parser) - 2;
      if ($this->wholeString[$pos] != ">") {
        $hereToEnd = substr($this->wholeString, $pos);
        $closePos = strpos($hereToEnd, ">");
      } else {
        $closePos = 0;
      }
      $this->currentItem->itemFullText = substr($this->wholeString, $this->currentItem->ItemStartPos, $pos + $closePos - $this->currentItem->ItemStartPos+1);

      //
      // Get the pubDate from pubDate first and then dc:date
      if (trim($this->currentItem->pubDate) != "") {
        $this->currentItem->pubDate = trim($this->currentItem->pubDate);
        $this->currentItem->pubDate_t = strtotime($this->currentItem->pubDate);
      }
      elseif (($this->feedTYPE == "FEE") && (trim($this->currentItem->atomUpdated) != "")) {
        $this->currentItem->atomUpdated = trim($this->currentItem->atomUpdated);
        $this->currentItem->pubDate_t = FeedForAll_scripts_getRFDdate($this->currentItem->atomUpdated);
        $this->currentItem->pubDate = date("D, d M Y H:i:s O", $this->currentItem->pubDate_t);
      }
      elseif (trim($this->currentItem->pubDateDC) != "") {
        $this->currentItem->pubDate_t = FeedForAll_scripts_getRFDdate($this->currentItem->pubDateDC);
        $this->currentItem->pubDate = date("D, d M Y H:i:s O", $this->currentItem->pubDate_t);
      } else {
        $this->currentItem->pubDate_t = time();
        $this->currentItem->pubDate = date("D, d M Y H:i:s O", $this->currentItem->pubDate_t);
      }

      if (($this->operateAs == "rssFilter") && function_exists("rssFilter_useItem")) {
        GLOBAL $_REQUEST;

        $this->UseItem = rssFilter_useItem($this->currentItem);

        if (isset($_REQUEST["testScript"])) {
          if ($this->UseItem) {
            echo "USING Item: ".htmlentities($this->currentItem->title)."<br>\n";
          } else {
            echo "NOT Using: ".htmlentities($this->currentItem->title)."<br>\n";
          }
        }
      }

      if ($this->operateAs == "rssMesh") {
        if (($this->itemLimit >= 0) && (count($this->Items) > $this->itemLimit)) {
          $this->UseItem = FALSE;
        }
      }
      elseif ($this->operateAs == "rss2html") {
        if (($useUniq = FeedForAll_rss2html_UseUniqueLink($this->currentItem->title, $this->currentItem->description, $this->currentItem->link, $this->currentItem->guid)) != -1) {
          if ($useUniq == 0) {
            if (isset($debugLevel) && ($debugLevel >= 2)) {
              echo "DIAG: FeedForAll_rss2html_UseUniqueLink() => 0, Not using<br>\n";
            }

            $this->UseItem = FALSE;
          }
        }
        if ($this->noFutureItems) {
          $noon = strtotime("today at 12:00");
          if (($this->currentItem->pubDate_t - $noon) > 43200) {
            if (isset($debugLevel) && ($debugLevel >= 2)) {
              echo "DIAG: future pubdate, Not using<br>\n";
            }
            $this->UseItem = FALSE;
          }
        }
      }

      if ($this->UseItem) {
        if (isset($debugLevel) && ($debugLevel >= 2)) {
          echo "DIAG: Using item \"".$this->currentItem->title."\"<br>\n";
        }
        
        //
        // Clean up some of the values
        $this->currentItem->title = trim($this->currentItem->title);
        $this->currentItem->description = trim($this->currentItem->description);
        if ($this->feedTYPE == "FEE") {
          $this->currentItem->atomContent = trim($this->currentItem->atomContent);
          $this->currentItem->description = $this->currentItem->atomContent;
        } else {
          $this->currentItem->description = $this->currentItem->description;
        }
        if (trim($this->currentItem->contentEncoded) == "") {
          if  ($this->operateAs != "rssMesh") {
            $this->currentItem->contentEncoded = $this->currentItem->description;
          }
        } else {
          $this->currentItem->contentEncoded = trim($this->currentItem->contentEncoded);
        }
        if (trim($this->currentItem->description) == "") {
          $this->currentItem->description = trim($this->currentItem->contentEncoded);
        }
        $this->currentItem->link = trim($this->currentItem->link);
        $this->currentItem->guid = trim($this->currentItem->guid);
        $this->currentItem->guidIsPermaLink = trim($this->currentItem->guidIsPermaLink);
        if ($this->feedTYPE == "FEE") {
          $this->currentItem->atomAuthorEmail = trim($this->currentItem->atomAuthorEmail);
          $this->currentItem->author = trim($this->currentItem->atomAuthorEmail);
        }
        $this->currentItem->author = trim($this->currentItem->author);
        if ($this->currentItem->creativeCommons == "") {
          $this->currentItem->creativeCommons = trim($this->FeedCreativeCommons);
        } else {
          $this->currentItem->creativeCommons = trim($this->currentItem->creativeCommons);
        }
        if ($this->operateAs == "rss2sql") {
          if (($this->currentItem->source == "") && ($this->sourceFeedURL != "")) {
            $this->currentItem->source = $this->FeedTitle;
            $this->currentItem->sourceURL = $this->sourceFeedURL;
          }
        }
        $this->currentItem->source = trim($this->currentItem->source);
        $this->currentItem->sourceURL = trim($this->currentItem->sourceURL);
        $this->currentItem->enclosureURL = trim($this->currentItem->enclosureURL);
        $this->currentItem->enclosureLength = trim($this->currentItem->enclosureLength);
        $this->currentItem->enclosureType = trim($this->currentItem->enclosureType);
        $this->currentItem->comments = trim($this->currentItem->comments);
        $this->currentItem->rssMeshExtra = trim($this->currentItem->rssMeshExtra);
        $this->currentItem->rssMeshFeedImageTitle = trim($this->currentItem->rssMeshFeedImageTitle);
        $this->currentItem->rssMeshFeedImageUrl = trim($this->currentItem->rssMeshFeedImageUrl);
        $this->currentItem->rssMeshFeedImageLink = trim($this->currentItem->rssMeshFeedImageLink);
        $this->currentItem->rssMeshFeedImageDescription = trim($this->currentItem->rssMeshFeedImageDescription);
        $this->currentItem->rssMeshFeedImageHeight = trim($this->currentItem->rssMeshFeedImageHeight);