X-Git-Url: https://git.mxchange.org/?a=blobdiff_plain;f=application%2Fhub%2Fmain%2Fsource%2Furls%2Fclass_CrawlerUploadedListUrlSource.php;fp=application%2Fhub%2Fmain%2Fsource%2Furls%2Fclass_CrawlerUploadedListUrlSource.php;h=5b1814594a3b565105a4a83f92c4d48c866637cb;hb=0d4b1675b69d4f54e11e80af16027e15b8da4eb1;hp=ec77631f0c64191c4ca852d3974b6e1f97ca667c;hpb=c6d1d880b6ee6289f5cdedc7b3eefdc1ae92cda2;p=hub.git

diff --git a/application/hub/main/source/urls/class_CrawlerUploadedListUrlSource.php b/application/hub/main/source/urls/class_CrawlerUploadedListUrlSource.php
index ec77631f0..5b1814594 100644
--- a/application/hub/main/source/urls/class_CrawlerUploadedListUrlSource.php
+++ b/application/hub/main/source/urls/class_CrawlerUploadedListUrlSource.php
@@ -73,32 +73,16 @@ class CrawlerUploadedListUrlSource extends BaseUrlSource implements UrlSource, R
 	}
 
 	/**
-	 * Creates an instance of this class
-	 *
-	 * @return	$sourceInstance		An instance of a Source class
-	 */
-	public final static function createCrawlerUploadedListUrlSource () {
-		// Get new instance
-		$sourceInstance = new CrawlerUploadedListUrlSource();
-
-		// Init source
-		$sourceInstance->initSource('crawler', 'uploaded_list');
-
-		// Return the prepared instance
-		return $sourceInstance;
-	}
-
-	/**
-	 * Checks whether a CSV file is found
+	 * Checks whether a CSV file is found in configured path
 	 *
 	 * @return	$isFound	Whether a CSV file is found
 	 */
 	private function isCsvFileFound () {
 		//* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__)->debugOutput('CRAWLER-SOURCE [' . __METHOD__ . ':' . __LINE__ . ']: CALLED!');
 
-		// Is the instance valid?
+		// Is it valid?
 		if (!$this->getDirectoryInstance()->getDirectoryIteratorInstance()->valid()) {
-			// Then rewind it
+			// Rewind to start
 			$this->getDirectoryInstance()->getDirectoryIteratorInstance()->rewind();
 		} // END - if
 
@@ -106,9 +90,13 @@ class CrawlerUploadedListUrlSource extends BaseUrlSource implements UrlSource, R
 		/* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__)->debugOutput('CRAWLER-SOURCE [' . __METHOD__ . ':' . __LINE__ . ']: this->csvFileImported=' . print_r($this->csvFileImported, TRUE));
 		$directoryEntry = $this->getDirectoryInstance()->readDirectoryExcept(array_merge(array('.htaccess', '.', '..'), $this->csvFileImported));
 
-		// The read entry has not to be empty and extension must be '.csv'
+		// Debug message
+		/* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__)->debugOutput('SOURCE[' . __METHOD__ . ':' . __LINE__ . '] directoryEntry(' . strlen($directoryEntry) . ')=' . $directoryEntry);
+
+		// Is it empty or wrong file extension?
 		if ((empty($directoryEntry)) || (substr($directoryEntry, -4, 4) != '.csv')) {
 			// Skip further processing
+			/* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__)->debugOutput('SOURCE[' . __METHOD__ . ':' . __LINE__ . '] directoryEntry(' . strlen($directoryEntry) . ')=' . $directoryEntry . ' - SKIPPED!');
 			return FALSE;
 		} // END - if
 
@@ -120,6 +108,22 @@ class CrawlerUploadedListUrlSource extends BaseUrlSource implements UrlSource, R
 		return TRUE;
 	}
 
+	/**
+	 * Creates an instance of this class
+	 *
+	 * @return	$sourceInstance		An instance of a Source class
+	 */
+	public final static function createCrawlerUploadedListUrlSource () {
+		// Get new instance
+		$sourceInstance = new CrawlerUploadedListUrlSource();
+
+		// Init source
+		$sourceInstance->initSource('crawler', 'uploaded_list');
+
+		// Return the prepared instance
+		return $sourceInstance;
+	}
+
 	/**
 	 * Initializes the import of the CSV file which is being processed by other task
 	 *
@@ -165,10 +169,17 @@ class CrawlerUploadedListUrlSource extends BaseUrlSource implements UrlSource, R
 	public function processStack () {
 		// Does the stack have some entries left?
 		if (!$this->isUrlStackEmpty()) {
-			// Handle next entry
+			/*
+			 * Handle next entry. This method will be called very often, so need
+			 * to process more than one entry at a time.
+			 */
 			$this->processNextEntry();
 		} elseif ($this->isCsvFileFound()) {
-			// A CSV file has been found and can maybe be imported.
+			/*
+			 * A file containing an URL list is found. Please note the format is
+			 * CSV-like as you may wish to provide meta data such as crawl
+			 * depth, handling of 3rd-party URLs and such.
+			 */
 			$this->importCsvFile();
 		}