function arrayToXML($array, $passedViaEntry, $allowedContent)
{
$xmlString = "";
foreach (array_keys($array) as $arrayKey) {
// For each array item
$isEntry = false;
// Define as an entry
$linkType = $arrayKey;
// Set the arrayKey to the linkType
if (is_int($linkType)) {
// If a linkType is an INT, then we are manipulating an entry
$linkType = "entry";
$isEntry = true;
// Set as an entry
}
$arrayContent = $array[$arrayKey];
// Get the value of this particular array item
if (is_array($arrayContent)) {
// If the array item content is an array, meaning there is multiple of this type
if ($isEntry == false) {
// If this is NOT an entry
foreach ($arrayContent as $thisTagAttributes) {
// For each arrayIndex in the attributesArray
$tagContent = "";
if (array_key_exists("@attributes", $thisTagAttributes)) {
// If there is an @attributes in thisTagAttributes
$thisTagAttributes = $thisTagAttributes["@attributes"];
// Re-assign the tag attributes to this particular @attributes
}
foreach ($thisTagAttributes as $attribute => $attributeValue) {
// For each tag attribute
$tagContent = $tagContent . $attribute . '="' . $attributeValue . '" ';
}
$xmlString = $xmlString . "\n\t<" . $linkType . " " . $tagContent . "/>";
}
} else {
// If it IS an entry
if (filterContent($arrayContent["title"], $allowedContent) !== true) {
// If we are NOT going to be filtering this content
$tagContent = arrayToXML($arrayContent, true, $allowedContent);
// Recursively create the XML for the entry array content
$xmlString = $xmlString . "\n<" . $linkType . ">" . $tagContent . "\n</" . $linkType . ">";
}
}
} else {
// If it is NOT an array
if ($passedViaEntry == true) {
// If we are doing a recursive inner entry parsing
$tabChar = "\t";
// Allow the tab character so we can indent the tag
} else {
$tabChar = "";
// Disallow the tab character
}
$xmlString = $xmlString . "\n" . $tabChar . "<" . $linkType . ">" . $array[$arrayKey] . "</" . $linkType . ">";
}
}
return $xmlString;
}
function stringFilter($search, $contents)
{
$search = trim($search);
foreach ($contents as $key => $content) {
switch ($key) {
case "title":
case "subtitle":
// find content in search
// some cases, subtitle only contains 1 character
if (strlen($content) >= 3 && filterContent($content, $search)) {
return true;
}
default:
// find current search
if (filterContent($search, $content)) {
return true;
}
}
}
return false;
}
/**
* @param string $content
* @return string
*/
function prepareXmlContent($content)
{
$content = getPrettyXml(filterContent($content));
$content = preg_replace('/^<\\?xml.*\\n/', '', $content);
return $content;
}
// Content was obtained so update or add in mysql
$query = 'SELECT * from dnint_url_contents WHERE url="' . addslashes($url) . '"';
$stm2 = $dbh->execute($query);
if ($stmt->num_row() == 0) {
//this URL is new and its contents are not fetched yet
$query = "INSERT INTO dnint_url_contents \n SET url='" . addslashes($url) . "', \n url_content='" . addslashes($content) . "', \n fk_dnint_url_id=" . $id;
$dbh->execute($query);
} else {
//this URL exists so just overwrite its contents (and also its FK to make sure we are in sync)
$query = "UPDATE dnint_url_contents \n SET url_content='" . addslashes($content) . "', \n fk_dnint_url_id=" . addslashes($id) . " \n WHERE url='" . $url . "'";
$stmt = $dbh->execute($query);
}
// Semantic parsing phase
$sql = $dbh->execute("SELECT id FROM feed_url_contents ORDER BY id DESC LIMIT 1");
list($last_id) = $sql->fetch_array();
$filtered = filterContent($content, $wordsAfterFilter);
$numKeywords = 0;
$sumDimX = 0;
$sumDimY = 0;
$negAverage = 0;
evaluateFilteredContent($dbh, $filtered, $numKeywords, $sumDimX, $sumDimY);
$dimxAverage = round($sumDimX / $numKeywords, 2);
$dimyAverage = round($sumDimY / $numKeywords, 2);
$query = "INSERT INTO dnint_contents_parsed \n SET url='" . addslashes($url) . "', \n parsed_content='" . addslashes($filtered) . "',\n fk_dnint_url_contents_id=" . addslashes($last_id);
$stmt = $dbh->execute($query);
// Get the id of the inserted record
$newID = mysql_insert_id();
// Now also add a record in the feed_parsed_results table
$query = 'INSERT INTO dnint_parsed_results
SET dimx_avg=' . addslashes($dimxAverage) . ',
dimy_avg=' . addslashes($dimyAverage) . ' ,