X-Git-Url: http://wamblee.org/gitweb/?a=blobdiff_plain;f=crawler%2Fkiss%2Fsrc%2Forg%2Fwamblee%2Fcrawler%2Fkiss%2Fmain%2FKissCrawler.java;h=3191ff24e627cffdd5d0eab3321e106c090a3ba9;hb=778bc657ac86064418cd165e30b35d8b7e1b8455;hp=969c5b23749e8e500f052308af0ed48a44f09308;hpb=616d9c8927b015cd8b652460c6227b40ee1ecd2e;p=utils

diff --git a/crawler/kiss/src/org/wamblee/crawler/kiss/main/KissCrawler.java b/crawler/kiss/src/org/wamblee/crawler/kiss/main/KissCrawler.java
index 969c5b23..3191ff24 100644
--- a/crawler/kiss/src/org/wamblee/crawler/kiss/main/KissCrawler.java
+++ b/crawler/kiss/src/org/wamblee/crawler/kiss/main/KissCrawler.java
@@ -29,6 +29,7 @@ import java.util.regex.Pattern;
 import javax.mail.MessagingException;
 
 import org.apache.commons.httpclient.HttpClient;
+import org.apache.commons.httpclient.NameValuePair;
 import org.apache.commons.logging.Log;
 import org.apache.commons.logging.LogFactory;
 import org.wamblee.crawler.Action;
@@ -46,6 +47,7 @@ import org.wamblee.crawler.kiss.guide.Time;
 import org.wamblee.crawler.kiss.guide.TimeInterval;
 import org.wamblee.crawler.kiss.notification.NotificationException;
 import org.wamblee.crawler.kiss.notification.Notifier;
+import org.wamblee.general.BeanFactory;
 import org.wamblee.xml.ClasspathUriResolver;
 import org.wamblee.xml.XslTransformer;
 
@@ -60,7 +62,7 @@ public class KissCrawler {
     /**
      * Start URL of the electronic programme guide.
      */
-    private static final String START_URL = "http://epg.kml.kiss-technology.com/login_core.php";
+    private static final String START_URL = "http://epg.kml.kiss-technology.com/login.php";
     
     /**
      * Default socket timeout to use. 
@@ -71,7 +73,7 @@ public class KissCrawler {
      * Regular expression for matching time interval strings in the retrieved
      * pages.
      */
-    private static final String TIME_REGEX = "([0-9]{2}):([0-9]{2})[^0-9]*([0-9]{2}):([0-9]{2}).*";
+    private static final String TIME_REGEX = "[^0-9]*([0-9]{2}):([0-9]{2})[^0-9]*([0-9]{2}):([0-9]{2}).*";
 
     /**
      * Compiled pattern for the time regular expression.
@@ -89,7 +91,10 @@ public class KissCrawler {
     public static void main(String[] aArgs) throws Exception {
         String crawlerConfig = new File(aArgs[0]).getCanonicalPath(); 
         String programConfig = new File(aArgs[1]).getCanonicalPath(); 
-        new KissCrawler(START_URL, SOCKET_TIMEOUT, crawlerConfig, programConfig);
+
+        BeanFactory factory = new StandaloneCrawlerBeanFactory();
+        Notifier notifier = factory.find(Notifier.class);
+        new KissCrawler(START_URL, SOCKET_TIMEOUT, crawlerConfig, programConfig, notifier, new Report());
     }
     
     /**
@@ -101,14 +106,16 @@ public class KissCrawler {
      *            Configuration file for the crawler.
      * @param aProgramConfig
      *            Configuration file describing interesting shows.
+     * @param aNotifier Object used to send notifications of the results.           
+     * @param aReport Report to use. 
      * @throws IOException
      *             In case of problems reading files.
-     * @throws MessagingException
-     *             In case of problems sending a mail notification.
+     * @throws NotificationException In case notification fails.
+     * @throws PageException In case of problems retrieving the TV guide. 
      */
     public KissCrawler(String aCrawlerConfig,
-            String aProgramConfig) throws IOException, NotificationException, PageException {
-        this(START_URL, SOCKET_TIMEOUT, aCrawlerConfig, aProgramConfig);
+            String aProgramConfig, Notifier aNotifier, Report aReport) throws IOException, NotificationException, PageException {
+        this(START_URL, SOCKET_TIMEOUT, aCrawlerConfig, aProgramConfig, aNotifier, aReport);
     }
 
 
@@ -124,13 +131,15 @@ public class KissCrawler {
      *            Configuration file for the crawler.
      * @param aProgramConfig
      *            Configuration file describing interesting shows.
+     * @param aNotifier Object used to send notifications of the results.           
+     * @param aReport Report to use. 
      * @throws IOException
      *             In case of problems reading files.
-     * @throws MessagingException
-     *             In case of problems sending a mail notification.
+     * @throws NotificationException In case notification fails.
+     * @throws PageException In case of problems retrieving the TV guide.   
      */
     public KissCrawler(String aStartUrl, int aSocketTimeout, String aCrawlerConfig,
-            String aProgramConfig) throws IOException, NotificationException, PageException {
+            String aProgramConfig, Notifier aNotifier, Report aReport) throws IOException, NotificationException, PageException {
 
         _pattern = Pattern.compile(TIME_REGEX);
 
@@ -145,26 +154,23 @@ public class KissCrawler {
             Crawler crawler = createCrawler(aCrawlerConfig, client, transformer);
             InputStream programConfigFile = new FileInputStream(new File(
                     aProgramConfig));
-            ProgramConfigurationParser parser = new ProgramConfigurationParser(
-                    transformer);
+            ProgramConfigurationParser parser = new ProgramConfigurationParser();
             parser.parse(programConfigFile);
             List<ProgramFilter> programFilters = parser.getFilters();
 
-            Report report = new Report();
-
             try {
-                Page page = getStartPage(aStartUrl, crawler, report);
-                TVGuide guide = createGuide(page, report);
+                Page page = getStartPage(aStartUrl, crawler, aReport);
+                TVGuide guide = createGuide(page, aReport);
                 PrintVisitor printer = new PrintVisitor(System.out);
                 guide.accept(printer);
-                processResults(programFilters, guide, parser.getNotifier(),
-                        report);
+                processResults(programFilters, guide, aNotifier,
+                        aReport);
             } catch (PageException e) {
-                report.addMessage("Problem getting TV guide", e);
+                aReport.addMessage("Problem getting TV guide", e);
                 LOG.info("Problem getting TV guide", e);
                 throw e; 
             }
-            parser.getNotifier().send(report.asXml());
+            aNotifier.send(aReport.asXml());
         } finally {
             System.out.println("Crawler finished");
         }
@@ -232,7 +238,8 @@ public class KissCrawler {
     private Page getStartPage(String aStartUrl, Crawler aCrawler, Report aReport)
             throws PageException {
         try {
-            Page page = aCrawler.getPage(aStartUrl);
+            Page page = aCrawler.getPage(aStartUrl, new NameValuePair[0]);
+            page = page.getAction("login").execute();
             Action favorites = page.getAction("channels-favorites");
             if (favorites == null) {
                 String msg = "Channels favorites action not found on start page";
@@ -253,10 +260,15 @@ public class KissCrawler {
      * @param aReport
      *            Report to use.
      * @return TV guide.
+     * @throws PageException In case of problem getting the tv guide.
      */
-    private TVGuide createGuide(Page aPage, Report aReport) {
+    private TVGuide createGuide(Page aPage, Report aReport) throws PageException {
         LOG.info("Obtaining full TV guide");
         Action[] actions = aPage.getActions();
+        if ( actions.length == 0 ) { 
+            LOG.error("No channels found");
+            throw new PageException("No channels found");
+        }
         List<Channel> channels = new ArrayList<Channel>();
         for (Action action : actions) {
             try {