" +sArticleXMLElementText2HTML) + "
\n]]>\n" + " \n" + "
" +sArticleXMLElementText2HTML) + "
\n]]>\n" + " \n" + "
"); //getFile = /story/xxx_reports/Euro_Strengthens_Ahead_of_NFP_1141994436831.html Pattern pForFileNameToWrite = Pattern.compile("(.*)/(.*?).html"); Matcher mForFileNameToWrite = pForFileNameToWrite.matcher(item.getFile()); String sExtractedFileNameForXML = "empty - RegEx for content not found in file or file not loaded"; if (mForFileNameToWrite.find() == true) { sExtractedFileNameForXML = mForFileNameToWrite.group(2); } else { // comment and for debuging: System.outx.println("Problem occured in matching the file name - item.getFile() = " + item.getFile() + " "); } sArticle = normalizeURLs(sArticle); /* sArticle = sArticle.replaceAll("src=\"/story-images/", "src=\"http://www.xxx.com/export/sites/xxx/"); sArticle = sArticle.replaceAll("SRC=\"/story-images/", "SRC=\"http://www.xxx.com/export/sites/xxx/"); sArticle = sArticle.replaceAll("src=\"/opencms/opencms/", "src=\"http://www.xxx.com/export/sites/xxx/"); sArticle = sArticle.replaceAll("SRC=\"/opencms/opencms/", "SRC=\"http://www.xxx.com/export/sites/xxx/"); sArticle = sArticle.replaceAll("href=\"/opencms/opencms/", "href=\"http://www.xxx.com/export/sites/xxx/"); sArticle = sArticle.replaceAll("HREF=\"/opencms/opencms/", "HREF=\"http://www.xxx.com/export/sites/xxx/"); */ //-----[WRITING YAHOO FILE TO OS FS ------// //-----[WRITING YAHOO FILE TO OS FS ------// //-----[WRITING YAHOO FILE TO OS FS ------// //-----[WRITING YAHOO FILE TO OS FS ------// //-----[WRITING YAHOO FILE TO OS FS ------// //We also want to check if this directory is here and is writable: String sAbsolutePathToWriteYahooExportDirectoryOnUNIX = "/usr/local/xxx/resources/new_articles_for_yahoo/"; String sAbsolutePathToWriteYahooExportDirectoryOnWINDOWS = "C:\\Program Files\\netbeans-4.1\\enterprise1\\jakarta-tomcat-5.5.7\\bin\\new_articles_for_yahoo\\"; //String sNameOfYahooFile = sExtractedFileNameForXML + ".xml"; String sNameOfYahooFile = sExtractedFileNameForXML + ".xml"; String sEntirePathOfFileToWriteYahooExportTo = sAbsolutePathToWriteYahooExportDirectoryOnUNIX + sNameOfYahooFile; File destination_file = new File(sEntirePathOfFileToWriteYahooExportTo); //per http://javaalmanac.com/egs/java.io/WriteISOLatin1.html try { File parentdir = parent(destination_file); if( !(sArticleXMLElementText2HTML.equals("")) ) { if (parentdir.exists()){ if (parentdir.canWrite()){ //BufferedWriter out_fullForEachNewArticleForYahoo = new BufferedWriter(new FileWriter(sEntirePathOfFileToWriteYahooExportTo)); //this is export to raw OS FS for use by external means Writer out = new BufferedWriter(new OutputStreamWriter(new FileOutputStream(sEntirePathOfFileToWriteYahooExportTo), "8859_1")); out.write(sArticle); out.close(); //out_fullForEachNewArticleForYahoo.write(sArticle); //out_fullForEachNewArticleForYahoo.close(); } } else { //System.out.println("write to UNIX FX ("+sEntirePathOfFileToWriteYahooExportTo+") has failed"); //else { //write to UNIX system failed, because we might be on the windows env; attempting to write to windows env. sEntirePathOfFileToWriteYahooExportTo = sAbsolutePathToWriteYahooExportDirectoryOnWINDOWS + sNameOfYahooFile; destination_file = new File(sEntirePathOfFileToWriteYahooExportTo); parentdir = parent(destination_file); if (parentdir.exists()){ if (parentdir.canWrite()){ Writer out = new BufferedWriter(new OutputStreamWriter(new FileOutputStream(sEntirePathOfFileToWriteYahooExportTo), "8859_1")); out.write(sArticle); out.close(); //BufferedWriter out_fullForEachNewArticleForYahoo = new BufferedWriter(new FileWriter(sEntirePathOfFileToWriteYahooExportTo)); //this is export to raw OS FS for use by external means //out_fullForEachNewArticleForYahoo.write(sArticle); //out_fullForEachNewArticleForYahoo.close(); } } } } } catch (UnsupportedEncodingException e) { } catch (IOException e) { } //ADD TO REGULAR RSS FEED ParsePosition pPaternForRssDataMatechUp = new ParsePosition(0); Date theDateForRssItem = inputDateFormat.parse(item.getDate(), pPaternForRssDataMatechUp); SimpleDateFormat outputDateFormatForRss = new SimpleDateFormat("EEE, dd MMM yyyy HH:mm:ss Z", Locale.US); //Mon, 23 Apr 2007 12:36:10 -0400 String outputDateForRssItem = outputDateFormatForRss.format(theDateForRssItem); SimpleDateFormat outputDateFormatForRssForKeywords = new SimpleDateFormat("yyyy-MM-dd HH:mm:ssZ", Locale.US); //2007-03-29T14:13:16-05:00 String outputDateForRssItemForKeywords = outputDateFormatForRssForKeywords.format(theDateForRssItem); outputDateForRssItemForKeywords = outputDateForRssItemForKeywords.replaceAll(" ", "T"); outputDateForRssItemForKeywords = outputDateForRssItemForKeywords.replaceAll("-0400", "-04:00"); //Hot fix for Google sitemaps format outputDateForRssItemForKeywords = outputDateForRssItemForKeywords.replaceAll("-0500", "-05:00"); //Hot fix for Google sitemaps format outputDateForRssItemForKeywords = outputDateForRssItemForKeywords.replaceAll("-0600", "-06:00"); //Hot fix for Google sitemaps format //sArticleXMLElementText1plainText = sArticleXMLElementText1plainText.replaceAll("&", "and"); sArticleXMLElementText1plainText = sArticleXMLElementText1plainText.replaceAll("\n", " "); sArticleXMLElementText1plainText = sArticleXMLElementText1plainText.replaceAll("<(.*?)>", " "); sArticleXMLElementText1plainText = sArticleXMLElementText1plainText.replaceAll("<[^>]+>"," "); sArticleXMLElementText1plainText = sArticleXMLElementText1plainText.replaceAll(" ", " "); sArticleXMLElementText1plainText = sArticleXMLElementText1plainText.replaceAll("^\\s+", " "); sArticleXMLElementText1plainText = sArticleXMLElementText1plainText.replaceAll("^ ", ""); sArticleXMLElementText1plainText = sArticleXMLElementText1plainText.replaceAll(" $", ""); sArticleXMLElementText1plainText = sArticleXMLElementText1plainText.replaceAll("\\s+$", ""); sArticleXMLElementText1plainText = sArticleXMLElementText1plainText.replaceAll("\\b\\s{2,}\\b", " "); //added here //sArticleXMLElementText2plainText = sArticleXMLElementText2plainText.replaceAll("&", "and"); sArticleXMLElementText2plainText = sArticleXMLElementText2plainText.replaceAll("\n", " "); sArticleXMLElementText2plainText = sArticleXMLElementText2plainText.replaceAll("<(.*?)>", " "); sArticleXMLElementText2plainText = sArticleXMLElementText2plainText.replaceAll("<[^>]+>"," "); sArticleXMLElementText2plainText = sArticleXMLElementText2plainText.replaceAll(" ", " "); sArticleXMLElementText2plainText = sArticleXMLElementText2plainText.replaceAll("^\\s+", " "); sArticleXMLElementText2plainText = sArticleXMLElementText2plainText.replaceAll("^ ", ""); sArticleXMLElementText2plainText = sArticleXMLElementText2plainText.replaceAll(" $", ""); sArticleXMLElementText2plainText = sArticleXMLElementText2plainText.replaceAll("\\s+$", ""); sArticleXMLElementText2plainText = sArticleXMLElementText2plainText.replaceAll("\\b\\s{2,}\\b", " "); String sDescriptionForRss = ""; if(sArticleXMLElementText1plainText != null && sArticleXMLElementText1plainText.indexOf(" ", 150) != -1){ sDescriptionForRss = sArticleXMLElementText1plainText.substring(0, sArticleXMLElementText1plainText.indexOf(" ", 150) ) + "..."; } else { sDescriptionForRss = sArticleXMLElementText2plainText; } try { if(sArticleXMLElementText1plainText != null && sArticleXMLElementText1plainText.indexOf(" ", 150) != -1){ sDescriptionForRss = sArticleXMLElementText1plainText.substring(0, sArticleXMLElementText1plainText.indexOf(" ", 150) ) + "..."; } else if (sArticleXMLElementText1plainText != null){ sDescriptionForRss = sArticleXMLElementText1plainText; } else{ //now this is in case description/summary is left blank entirely by analysts... when posting the data - we make use of the body/article data if(sArticleXMLElementText2plainText != null && sArticleXMLElementText2plainText.indexOf(" ", 150) != -1){ sDescriptionForRss = sArticleXMLElementText2plainText.substring(0, sArticleXMLElementText2plainText.indexOf(" ", 150) ) + "..."; } else{ sDescriptionForRss = "See the article..."; //todo, but this is in a very rare case that if summary is empty and body is under 150 characters... } } } catch ( StringIndexOutOfBoundsException e ){ // comment and for debuging: System.outx.println("StringIndexOutOfBoundsException occured "); } sDescriptionForRss = sDescriptionForRss.replaceAll("&", " and "); sDescriptionForRss = sDescriptionForRss.replaceAll("\n", " "); sDescriptionForRss = sDescriptionForRss.replaceAll("<(.*?)>", " "); sDescriptionForRss = sDescriptionForRss.replaceAll("<[^>]+>"," "); sDescriptionForRss = sDescriptionForRss.replaceAll(" ", " "); sDescriptionForRss = sDescriptionForRss.replaceAll("^\\s+", " "); sDescriptionForRss = sDescriptionForRss.replaceAll("^ ", ""); sDescriptionForRss = sDescriptionForRss.replaceAll(" $", ""); sDescriptionForRss = sDescriptionForRss.replaceAll("\\s+$", ""); sDescriptionForRss = sDescriptionForRss.replaceAll("\\b\\s{2,}\\b", " "); if(sDescriptionForRss.equals("")){ sDescriptionForRss = "See the article..."; } sRssItem += "
" + sArticleXMLElementText2HTML; sRssDescriptionForRssWoImages = cleanHTML(sRssDescriptionForRssWoImages); sRssDescriptionForRssWoImages= sRssDescriptionForRssWoImages.replaceAll("&", "&"); sRssDescriptionForRssWoImages = sRssDescriptionForRssWoImages.replaceAll("
" + sArticleXMLElementText2HTML; sRssDescriptionForRssWithImages = sRssDescriptionForRssWithImages.replaceAll("&", "&"); sRssDescriptionForRssWithImages = cleanHTML(sRssDescriptionForRssWithImages); sRssDescriptionForRssWithImages = sRssDescriptionForRssWithImages.replaceAll("\n", " "); sRssDescriptionForRssWithImages = sRssDescriptionForRssWithImages.replaceAll("
" + item.getDate() + "
\n" +
"" +
"\n" +
"\n" +
"![](\"/images/plus.gif\")
\n" +
"
\n" +
"\n" +
"\n" +
"\n";
}
}
//WRTING xxxxTR AND REFCOFX NEWS EXPORT TO OS FS
//WRTING xxxxTR AND REFCOFX NEWS EXPORT TO OS FS
//WRTING xxxxTR AND REFCOFX NEWS EXPORT TO OS FS
//WRTING xxxxTR AND REFCOFX NEWS EXPORT TO OS FS
//WRTING xxxxTR AND REFCOFX NEWS EXPORT TO OS FS
//We also want to check if this directory is here and is writable:
String sAbsolutePathToWritexxxxTRExportDirectoryOnUNIX = "/usr/local/xxx/resources/new_articles_for_xxxxTR/";
String sAbsolutePathToWritexxxxTRExportDirectoryOnWINDOWS = "C:\\Program Files\\netbeans-4.1\\enterprise1\\jakarta-tomcat-5.5.7\\bin\\new_articles_for_xxxxTR\\";
String sNameOfxxxxTRFile = "All_latest_xxx_Articles_for_xxxxTR_ready.html";
String sEntirePathOfFileToWritexxxxTRExportTo = sAbsolutePathToWritexxxxTRExportDirectoryOnUNIX + sNameOfxxxxTRFile;
File destination_file = new File(sEntirePathOfFileToWritexxxxTRExportTo);
File parentdir = parent(destination_file);
if (parentdir.exists()){
if (parentdir.canWrite()){
//BufferedWriter out_fullForEachNewArticleForxxxxTR = new BufferedWriter(new FileWriter(sEntirePathOfFileToWritexxxxTRExportTo)); //this is export to raw OS FS for use by external means
//out_fullForEachNewArticleForxxxxTR.write(sxxxxTR_news);
//out_fullForEachNewArticleForxxxxTR.close();
Writer out = new BufferedWriter(new OutputStreamWriter(new FileOutputStream(sEntirePathOfFileToWritexxxxTRExportTo), "8859_1"));
out.write(sxxxxTR_news);
out.close();
}
}
else { //write to UNIX system failed, because we might be on the windows env; attempting to write to windows env.
sEntirePathOfFileToWritexxxxTRExportTo = sAbsolutePathToWritexxxxTRExportDirectoryOnWINDOWS + sNameOfxxxxTRFile;
destination_file = new File(sEntirePathOfFileToWritexxxxTRExportTo);
parentdir = parent(destination_file);
if (parentdir.exists()){
if (parentdir.canWrite()){
//BufferedWriter out_fullForEachNewArticleForxxxxTR = new BufferedWriter(new FileWriter(sEntirePathOfFileToWritexxxxTRExportTo)); //this is export to raw OS FS for use by external means
//out_fullForEachNewArticleForxxxxTR.write(sxxxxTR_news);
//out_fullForEachNewArticleForxxxxTR.close();
Writer out = new BufferedWriter(new OutputStreamWriter(new FileOutputStream(sEntirePathOfFileToWritexxxxTRExportTo), "8859_1"));
out.write(sxxxxTR_news);
out.close();
}
}
}
//BufferedWriter out_fullForxxxxTR_ready = new BufferedWriter(new FileWriter("All_latest_xxx_Articles_for_xxxxTR_ready.html")); //this is export to raw OS FS for use by external means
//out_fullForxxxxTR_ready.write(sxxxxTR_news);
//out_fullForxxxxTR_ready.close();
//BEGIN FORMING THE RSS FEEDS
//REGULAR RSS FEED
Date dCurrentDateNow = new Date();
SimpleDateFormat outputDateFormatForRssTop = new SimpleDateFormat("EEE, dd MMM yyyy HH:mm:ss Z", Locale.US);
String sOutputDateTimeNowForRss = outputDateFormatForRssTop.format(dCurrentDateNow);
String sRssStart =
" \n" +
" " + item.getTitle() + "
\n" +
"\n" +
"![\"expand](\"/images/minus.gif\")
"); String sRssRegularFeedFullWithoutImages = sRssStart + sRssItemFull + sRssEnding; // comment and for debuging: System.outx.println("sRssRegularFeed =
"); String sRssRegularFeedFullWithImages = sRssStart + sRssItemFullAndImages + sRssEnding; // comment and for debuging: System.outx.println("sRssRegularFeed =
"); String sRssRegularWithKeywords = sRssStartKeywords + sRssItemKeywords + sRssEndingKeywords; // IMPORTANT !!! DO NOT FORGET TO CHANGE THE PHP FILE TYPE TO XML mimetypes.properties to have .php render as text/xml instead of text/html what might upset certain poor RSS readers //TESTING //http://www.google.com/reader/view/feed/http%3A%2F%2Fwww.ingearx.info%2Fxxxx1%2Fxml_full1c.xml?q=http%3A%2F%2Fwww.ingearx.info%2Fxxxx1%2Fxml_full1c.xml //http://blogs.law.harvard.edu/tech/rss //http://www.xxx.com/index2.php?option=com_rss&feed=RSS2.0&no_html=1 writeStringAsFileToVFS(cmsObject, sRssRegularFeed, sfSiteRoot + "/" + "index2.php"); writeStringAsFileToVFS(cmsObject, sRssRegularFeed, sfSiteRoot + "/" + "feeds/rss.xml"); //http://www.xxx.com/feeds/full.php writeStringAsFileToVFS(cmsObject, sRssRegularFeedFullWithoutImages, sfSiteRoot + "/" + "feeds/full.php"); writeStringAsFileToVFS(cmsObject, sRssRegularFeedFullWithoutImages, sfSiteRoot + "/" + "feeds/full.xml"); //http://www.xxx.com/feeds/full-images.php writeStringAsFileToVFS(cmsObject, sRssRegularFeedFullWithImages, sfSiteRoot + "/" + "feeds/full-images.php"); writeStringAsFileToVFS(cmsObject, sRssRegularFeedFullWithImages, sfSiteRoot + "/" + "feeds/full-images.xml"); writeStringAsFileToVFS(cmsObject, sRssRegularWithKeywords, sfSiteRoot + "/" + "feeds/newsmap.xml"); String sRssRegularFeedFullWithoutImages2 = sRssRegularFeedFullWithImages; sRssRegularFeedFullWithoutImages2 = sRssRegularFeedFullWithoutImages2.replaceAll("\n", " "); sRssRegularFeedFullWithoutImages2 = sRssRegularFeedFullWithoutImages2.replaceAll("\r", " "); sRssRegularFeedFullWithoutImages2 = sRssRegularFeedFullWithoutImages2.replaceAll("\n\r", " "); sRssRegularFeedFullWithoutImages2 = sRssRegularFeedFullWithoutImages2.replaceAll(" ", " "); sRssRegularFeedFullWithoutImages2 = sRssRegularFeedFullWithoutImages2.replaceAll(" ", " "); sRssRegularFeedFullWithoutImages2 = sRssRegularFeedFullWithoutImages2.replaceAll("
"); fileForAvantGo = cmsObj.readFile(sfSiteRoot+fileResourceForAvantGo); iUniqueIDofFileAvantGo = java.lang.Math.abs( fileForAvantGo.getResourceId().hashCode() ); String sArticleXMLElementText1HTMLForAvantGo = ""; String sArticleXMLElementText2HTMLForAvantGo = ""; try { // get default locale from the new article localeForAvantGo = (Locale)OpenCms.getLocaleManager().getDefaultLocales(cmsObj, fileForAvantGo.getRootPath()).get(0); // load article text (anything in
" + sArticleXMLElementText2HTMLForAvantGo; sAllOfTextForAvantGo = sAllOfTextForAvantGo.replaceAll("&", "&"); sAllOfTextForAvantGo = sAllOfTextForAvantGo.replaceAll("