feedforall_xmlparser.inc.php

RSS to html

    }
    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($attrs["REL"]) && ($attrs["REL"] == "alternate")) || !isset($attrs["REL"]))) || !isset($attrs["TYPE"]))) {
          $this->currentItem->link = $attrs["HREF"];
        }
      } else {
        if (isset($attrs["HREF"]) && ((isset($attrs["TYPE"]) && ($attrs["TYPE"] == "text/html") && ((isset($attrs["REL"]) && ($attrs["REL"] == "alternate")) || !isset($attrs["REL"]))) || !isset($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);
      //
      // Find the element that opened this all up.
      $endingString = '</'.substr($this->wholeString, $this->currentItem->ItemStartPos+1, 4);
      if ($endingString == substr($this->wholeString, $pos, 6)) {
        // We are good just where we are
      } else {
        for (;$pos > 0; $pos--) {
          /* Look for a closing angle */
          if ($endingString == substr($this->wholeString, $pos, 6)) break;
        }
      }
      $pos++;
      $hereToEnd = substr($this->wholeString, $pos);
      $closePos = strpos($hereToEnd, '>');
      $this->currentItem->itemFullText = substr($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) {
          if (($this->currentItem->pubDate_t - $this->noon_t) > 43200) {
            if (isset($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);
        $this->currentItem->rssMeshExtra1 = trim($this->currentItem->rssMeshExtra1);
        $this->currentItem->rssMeshExtra2 = trim($this->currentItem->rssMeshExtra2);
        $this->currentItem->rssMeshExtra3 = trim($this->currentItem->rssMeshExtra3);
        $this->currentItem->rssMeshFeedImageTitle = trim($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);
        $this->currentItem->rssMeshFeedImageWidth = trim($this->currentItem->rssMeshFeedImageWidth);
        if ($this->operateAs == "rss2html") {
          //
          // Escape any links
          $this->currentItem->link = FeedForAll_rss2html_EscapeLink($this->currentItem->link);
          $this->currentItem->guid = FeedForAll_rss2html_EscapeLink($this->currentItem->guid);
          $this->currentItem->creativeCommons = FeedForAll_rss2html_EscapeLink($this->currentItem->creativeCommons);
          $this->currentItem->sourceURL = FeedForAll_rss2html_EscapeLink($this->currentItem->sourceURL);
          $this->currentItem->enclosureURL = FeedForAll_rss2html_EscapeLink($this->currentItem->enclosureURL);
          $this->currentItem->comments = FeedForAll_rss2html_EscapeLink($this->currentItem->comments);
          $this->currentItem->rssMeshFeedImageUrl = FeedForAll_rss2html_EscapeLink($this->currentItem->rssMeshFeedImageUrl);
          $this->currentItem->rssMeshFeedImageLink = FeedForAll_rss2html_EscapeLink($this->currentItem->rssMeshFeedImageLink);
        }
        
        //
        if ($this->currentItem->contentEncodedUsed) {
          $this->contentEncodedUsed = 1;
        }
        if (FeedForAll_parseExtensions() === TRUE) {
          FeedForAll_parseExtensions_endElemend($parser, $this, $tagName);
        }
        if ($this->UseItem) {
          $this->Items[] = $this->currentItem;
          if (isset($debugLevel) && ($debugLevel >= 3)) {
            echo "DIAG: adding to items, count=".count($this->Items)."<br>\n";
          }
        }
      } else {
        unset($this->currentItem);
      }
      $this->insideItem = FALSE;
      $this->level_item = 0;
      return;
    }
    elseif ($this->insideAtomAuthor && ($tagName == "AUTHOR")) {
      $this->insideAtomAuthor = FALSE;
    }
    elseif (($tagName == "IMAGE") && ($this->insideChannelImage)) {
      $this->FeedImageTitle = trim($this->FeedImageTitle);
      $this->FeedImageURL = trim($this->FeedImageURL);
      $this->FeedImageLink = trim($this->FeedImageLink);
      $this->FeedImageDescription = trim($this->FeedImageDescription);
      $this->FeedImageHeight = trim($this->FeedImageHeight);
      $this->FeedImageWidth = trim($this->FeedImageWidth);
      if ($this->operateAs == "rss2html") {
        //
        // Escape any links
        $this->FeedImageURL = FeedForAll_rss2html_EscapeLink($this->FeedImageURL);
        $this->FeedImageLink = FeedForAll_rss2html_EscapeLink($this->FeedImageLink);
      }
      if (FeedForAll_parseExtensions() === TRUE) {
        FeedForAll_parseExtensions_endElemend($parser, $this, $tagName);
      }
      $this->insideChannelImage = FALSE;
      $this->level_channelImage = 0;
      return;
    }
    elseif ((($tagName == "CHANNEL") && ($this->feedTYPE != "FEE")) || (($tagName == "FEED") && ($this->feedTYPE == "FEE"))) {
      $this->FeedPubDate = trim($this->FeedPubDate);
      $this->FeedPubDateDC = trim($this->FeedPubDateDC);
      $this->FeedAtomUpdated = trim($this->FeedAtomUpdated);
      //
      // Get the pubDate from pubDate first and then dc:date
      if (trim($this->FeedPubDate) != "") {
        $this->FeedPubDate_t = strtotime($this->FeedPubDate);
      }
      elseif (($this->feedTYPE == "FEE") && ($this->FeedAtomUpdated != "")) {
        $this->FeedAtomUpdated = trim($this->FeedAtomUpdated);
        $this->FeedPubDate_t = FeedForAll_scripts_getRFDdate($this->FeedAtomUpdated);
        $this->FeedPubDate = date("D, d M Y H:i:s O", $this->FeedPubDate_t);
      }
      elseif (trim($this->FeedPubDateDC) != "") {
        $this->FeedPubDate_t = FeedForAll_scripts_getRFDdate($this->FeedPubDateDC);
        $this->FeedPubDate = date("D, d M Y H:i:s O", $this->FeedPubDate_t);
      }
      elseif (trim($this->FeedLastBuildDate) != "") {
        $this->FeedPubDate_t = strtotime($this->FeedLastBuildDate);
        $this->FeedPubDate = date("D, d M Y H:i:s O", $this->FeedPubDate_t);
      } else {
        $this->FeedPubDate_t = time();
        $this->FeedPubDate = date("D, d M Y H:i:s O", $this->FeedPubDate_t);
      }
      $this->FeedTitle = trim($this->FeedTitle);
      if ($this->feedTYPE == "FEE") {
        $this->FeedAtomContent = trim($this->FeedAtomContent);
        $this->FeedDescription = $this->FeedAtomContent;
      } else {
        $this->FeedDescription = $this->FeedDescription;
      }
      if (trim($this->FeedContentEncoded) == "") {
        $this->FeedContentEncoded = $this->FeedDescription;
      }
      $this->FeedLink = trim($this->FeedLink);
      if ($this->operateAs == "rss2html") {
        //
        // Escape any links
        $this->FeedLink = FeedForAll_rss2html_EscapeLink($this->FeedLink);
        $this->FeedCreativeCommons = FeedForAll_rss2html_EscapeLink($this->FeedCreativeCommons);
      }
      if (FeedForAll_parseExtensions() === TRUE) {
        FeedForAll_parseExtensions_endElemend($parser, $this, $tagName);
      }
      $this->insideChannel = FALSE;
      $this->level_channel = 0;
      return;
    }
    elseif ($this->level == $this->level_channel) {
      if ($tagName == "TITLE") {
        $this->FeedTitle = trim($this->FeedTitle);
      }