rendered paste body/* * FXRssResource.java * * Created on January 26, 2006, 3:19 PM * * To change this template, choose Tools | Options and locate the template under * the Source Creation and Management node. Right-click the template and choose * Open. You can then make changes to the template in the Source Editor. */ package com.xxxx.fxopencms.xxx;import java.io.BufferedWriter;import java.io.FileWriter;import java.io.File; import java.text.ParsePosition;import java.text.SimpleDateFormat;import java.util.Date;import java.util.Iterator;import java.util.List;import java.util.Locale;import java.util.regex.Matcher;import java.util.regex.Pattern;import java.io.FileOutputStream;import java.io.OutputStreamWriter;import java.io.Writer;import java.io.UnsupportedEncodingException;import java.io.IOException; import java.util.ArrayList;import org.opencms.jsp.CmsJspActionElement;import org.opencms.file.CmsObject;import org.opencms.file.CmsProperty;import java.util.List;import org.opencms.file.CmsResourceFilter;import org.opencms.file.CmsProperty;import org.opencms.file.CmsResource;import com.xxxx.fxopencms.xxx.FXConstants;import java.text.SimpleDateFormat;import java.util.Date;import org.opencms.db.CmsDefaultUsers;import org.opencms.file.CmsFile;import org.opencms.file.CmsObject;import org.opencms.file.CmsProperty;import org.opencms.main.CmsException;import org.opencms.main.FXResource;import org.opencms.main.OpenCms;import org.opencms.report.CmsShellReport;import org.opencms.xml.content.CmsXmlContent;import org.opencms.xml.content.CmsXmlContentFactory;import org.opencms.xml.types.CmsXmlHtmlValue;import com.opencms.htmlconverter.CmsHtmlConverter;import org.w3c.tidy.Tidy;import org.w3c.tidy.Configuration;import org.w3c.tidy.Node;import org.w3c.tidy.Out;import org.w3c.tidy.OutImpl;import org.w3c.tidy.PPrint;/** * * This class provides Rss generating file access functionality which is specific for * project. */public class FXRssResource { public FXResource fxres; public CmsObject cmsObject; public FXRssResourceThread mThreadForGeneratingRss = null; //String sfSiteRoot = "/sites/xxx"; String sfSiteRoot = ""; public FXRssResource(String sIdToUseForCmsObj, String sPassToUseForCmsObj) { try { cmsObject = OpenCms.initCmsObject((new CmsDefaultUsers()).getUserGuest()); System.out.println("sIdToUseForCmsObj, sPassToUseForCmsOb =5 " + sIdToUseForCmsObj + "," + sPassToUseForCmsObj); String sUserLoginStatus = cmsObject.loginUser(sIdToUseForCmsObj, sPassToUseForCmsObj); System.out.println("sUserLoginStatus =5 " + sUserLoginStatus ); cmsObject.getRequestContext().setSiteRoot("/sites/xxx/"); cmsObject.getRequestContext().setCurrentProject(cmsObject.readProject("Offline")); fxres = new FXResource(cmsObject); } catch (Exception e) { e.printStackTrace(); } } /** * This method generates RSS file * @param cmsObj CmsObject cms object * @param siteRoot String specifies where the latest files are to be retrived from * @param numListed int specifies the number of files to be retrived * @param rss_path_ String specifies the directory path where to write the files * @param xml_name String specifies the name of the file to write * @return description_of_param */ public void generateRssAllThread() { try { mThreadForGeneratingRss = new FXRssResourceThread(); mThreadForGeneratingRss.start(); //this will ensure that we are starting the thread mThreadForHashUpdateFromDB only once. } catch (Exception e) { e.printStackTrace(); System.out.println(e.getMessage()); } } private static File parent(File f) { String dirname = f.getParent(); if (dirname == null) { if (f.isAbsolute()) return new File(File.separator); else return new File(System.getProperty("user.dir")); } return new File(dirname); } public String normalizeURLs(String sStringToProcess) { //SOME ARE MISSING //http://www.xxx.com/export/sites/xxx/ /2006/04/charting_center/Weekly_Chart_Analysis/04-14-06weekly1.gif //http://www.xxx.com/export/sites/xxx/story-images/2006/04/charting_center/Weekly_Chart_Analysis/04-14-06weekly1.gif //SOME ARE //http://www.xxx.com/export/sites/xxx/ 2006/04/charting_center/Weekly_Chart_Analysis/04-17-06cot1.gif //http://www.xxx.com/export/sites/xxx/story-images/2006/04/charting_center/Weekly_Chart_Analysis/04-17-06cot1.gif //WRONG FOUND http://www.xxx.com/export/sites/xxx/2006/04/special_report/Special_Reports/Outlook2006.1.gif // ADDED story-images/ // in xxx/story-images/2006/ // CORREWCTED http://www.xxx.com/export/sites/xxx/story-images/2006/04/special_report/Special_Reports/Outlook2006.1.gif sStringToProcess = sStringToProcess.replaceAll("src=\"/story-images/", "src=\"http://www.xxx.com/export/sites/xxx/story-images/"); sStringToProcess = sStringToProcess.replaceAll("SRC=\"/story-images/", "SRC=\"http://www.xxx.com/export/sites/xxx/story-images/"); sStringToProcess = sStringToProcess.replaceAll("href=\"/story-images/", "href=\"http://www.xxx.com/export/sites/xxx/story-images/"); sStringToProcess = sStringToProcess.replaceAll("HREF=\"/story-images/", "HREF=\"http://www.xxx.com/export/sites/xxx/story-images/"); sStringToProcess = sStringToProcess.replaceAll("src=\"/opencms/opencms/", "src=\"http://www.xxx.com/export/sites/xxx/story-images/"); sStringToProcess = sStringToProcess.replaceAll("SRC=\"/opencms/opencms/", "SRC=\"http://www.xxx.com/export/sites/xxx/story-images/"); sStringToProcess = sStringToProcess.replaceAll("href=\"/opencms/opencms/", "href=\"http://www.xxx.com/export/sites/xxx/"); sStringToProcess = sStringToProcess.replaceAll("HREF=\"/opencms/opencms/", "HREF=\"http://www.xxx.com/export/sites/xxx/"); //FOUND A BAD LINK: //APPEARS href="/story-images/2006/04/special_report/Special_Reports/2006_outlook_0406.pdf" //APPEARS href="http://www.xxx.com/export/sites/xxx/story-images/2006/04/special_report/Special_Reports/2006_outlook_0406.pdf" //CORRECT href="http://www.xxx.com/export/sites/xxx/story-images/2006/04/special_report/Special_Reports/2006_outlook_0406.pdf" sStringToProcess = sStringToProcess.replaceAll("story-images/story-images/", "story-images/"); //sStringToProcess = sStringToProcess.replaceAll("HREF=\"/opencms/opencms/", "HREF=\"http://www.xxx.com/export/sites/xxx/"); //http://www.xxx.com/export/sites/xxx/2006/04/currency/weeklies/CAD_04-14-06.gif//http://www.xxx.com/export/sites/xxx/story-images/2006/04/currency/weeklies/CAD_04-14-06.gif return sStringToProcess; } private String cleanHTML(String htmlInput) { /** regular expression for cleanup. */ String[] m_cleanupPatterns = { "<o:p>.*(\\r\\n)*.*</o:p>", "<o:p>.*(\\r\\n)*.*</O:p>", "<\\?xml:.*(\\r\\n).*/>", "<\\?xml:.*(\\r\\n).*(\\r\\n).*/\\?>", "<\\?xml:.*(\\r\\n).*(\\r\\n).*/>", "<\\?xml:(.*(\\r\\n)).*/\\?>", "<o:SmartTagType.*(\\r\\n)*.*/>", "<o:smarttagtype.*(\\r\\n)*.*/>"}; /** patterns for cleanup. */ Pattern[] m_clearStyle; /** the input encoding. */ String m_encoding; /** the operation mode. */ List m_mode; /** regular expression for replace. */ String[] m_replacePatterns = {" ", "(\\r\\n){2,}", ""}; /** patterns for replace. */ Pattern[] m_replaceStyle; /** values for replace. */ String[] m_replaceValues = {" ", "", "–"}; String parsedHtml = new String(); parsedHtml = htmlInput.trim(); // create the regexp for cleanup m_clearStyle = new Pattern[m_cleanupPatterns.length]; for (int i = 0; i < m_cleanupPatterns.length; i++) { m_clearStyle[i] = Pattern.compile(m_cleanupPatterns[i]); } // create the regexp for replace m_replaceStyle = new Pattern[m_replacePatterns.length]; for (int i = 0; i < m_replacePatterns.length; i++) { m_replaceStyle[i] = Pattern.compile(m_replacePatterns[i]); } // process all cleanup regexp for (int i = 0; i < m_cleanupPatterns.length; i++) { parsedHtml = m_clearStyle[i].matcher(parsedHtml).replaceAll(""); } // process all replace regexp for (int i = 0; i < m_replacePatterns.length; i++) { parsedHtml = m_replaceStyle[i].matcher(parsedHtml).replaceAll(m_replaceValues[i]); } return parsedHtml; } public String replaceSpecialCharacters(String sPassedStringToProcess) { String sStringToProcess = ""; if(sPassedStringToProcess != null){ sStringToProcess = sPassedStringToProcess; //http://www.xxx.com/story/strategy_pieces/trade_or_fade/The_Moment_of_Truth_1145248240833.html //cd /usr/datum/xxx_xmlfeed/ && mv *.xml /usr/datum/xxx_xmlfeed/archives/ // /usr/local/xxx/tomcat/webapps/ROOT/WEB-INF/classes/com/xxxx/fxopencms/xxx sStringToProcess = sStringToProcess.replaceAll(" & "," and "); //only safe way to replace & sStringToProcess = sStringToProcess.replaceAll("<img ", "<br><img "); sStringToProcess = sStringToProcess.replaceAll("<IMG ", "<br><img "); //sStringToProcess = sStringToProcess.replaceAll("","“"); //sStringToProcess = sStringToProcess.replaceAll("","”"); sStringToProcess = sStringToProcess.replaceAll("","\""); sStringToProcess = sStringToProcess.replaceAll("","\""); sStringToProcess = sStringToProcess.replaceAll("","-"); //was • sStringToProcess = sStringToProcess.replaceAll("","-"); sStringToProcess = sStringToProcess.replaceAll("¡","¡"); sStringToProcess = sStringToProcess.replaceAll("¢","¢"); sStringToProcess = sStringToProcess.replaceAll("£","£"); sStringToProcess = sStringToProcess.replaceAll("¤","¤"); sStringToProcess = sStringToProcess.replaceAll("¥","¥"); sStringToProcess = sStringToProcess.replaceAll("¦","¦"); sStringToProcess = sStringToProcess.replaceAll("§","§"); sStringToProcess = sStringToProcess.replaceAll("¨","¨"); sStringToProcess = sStringToProcess.replaceAll("©","©"); sStringToProcess = sStringToProcess.replaceAll("ª","ª"); sStringToProcess = sStringToProcess.replaceAll("«","«"); sStringToProcess = sStringToProcess.replaceAll("¬","¬"); sStringToProcess = sStringToProcess.replaceAll("","­"); sStringToProcess = sStringToProcess.replaceAll("®","®"); sStringToProcess = sStringToProcess.replaceAll("¯","¯"); sStringToProcess = sStringToProcess.replaceAll("°","°"); sStringToProcess = sStringToProcess.replaceAll("±","±"); sStringToProcess = sStringToProcess.replaceAll("²","²"); sStringToProcess = sStringToProcess.replaceAll("³","³"); sStringToProcess = sStringToProcess.replaceAll("´","´"); sStringToProcess = sStringToProcess.replaceAll("µ","µ"); sStringToProcess = sStringToProcess.replaceAll("¶","¶"); sStringToProcess = sStringToProcess.replaceAll("·","-"); //· sStringToProcess = sStringToProcess.replaceAll("¸","¸"); sStringToProcess = sStringToProcess.replaceAll("¹","¹"); sStringToProcess = sStringToProcess.replaceAll("º","º"); sStringToProcess = sStringToProcess.replaceAll("»","»"); sStringToProcess = sStringToProcess.replaceAll("¼","¼"); sStringToProcess = sStringToProcess.replaceAll("½","½"); sStringToProcess = sStringToProcess.replaceAll("¾","¾"); sStringToProcess = sStringToProcess.replaceAll("¿","¿"); sStringToProcess = sStringToProcess.replaceAll("À","À"); sStringToProcess = sStringToProcess.replaceAll("Á","Á"); sStringToProcess = sStringToProcess.replaceAll("Â","Â"); sStringToProcess = sStringToProcess.replaceAll("Ã","Ã"); sStringToProcess = sStringToProcess.replaceAll("Ä","Ä"); sStringToProcess = sStringToProcess.replaceAll("Å","Å"); sStringToProcess = sStringToProcess.replaceAll("Æ","Æ"); sStringToProcess = sStringToProcess.replaceAll("Ç","Ç"); sStringToProcess = sStringToProcess.replaceAll("È","È"); sStringToProcess = sStringToProcess.replaceAll("É","É"); sStringToProcess = sStringToProcess.replaceAll("Ê","Ê"); sStringToProcess = sStringToProcess.replaceAll("Ë","Ë"); sStringToProcess = sStringToProcess.replaceAll("Ì","Ì"); sStringToProcess = sStringToProcess.replaceAll("Í","Í"); sStringToProcess = sStringToProcess.replaceAll("Î","Î"); sStringToProcess = sStringToProcess.replaceAll("Ï","Ï"); sStringToProcess = sStringToProcess.replaceAll("Ð","Ð"); sStringToProcess = sStringToProcess.replaceAll("Ñ","Ñ"); sStringToProcess = sStringToProcess.replaceAll("Ò","Ò"); sStringToProcess = sStringToProcess.replaceAll("Ó","Ó"); sStringToProcess = sStringToProcess.replaceAll("Ô","Ô"); sStringToProcess = sStringToProcess.replaceAll("Õ","Õ"); sStringToProcess = sStringToProcess.replaceAll("Ö","Ö"); sStringToProcess = sStringToProcess.replaceAll("×","×"); sStringToProcess = sStringToProcess.replaceAll("Ø","Ø"); sStringToProcess = sStringToProcess.replaceAll("Ù","Ù"); sStringToProcess = sStringToProcess.replaceAll("Ú","Ú"); sStringToProcess = sStringToProcess.replaceAll("Û","Û"); sStringToProcess = sStringToProcess.replaceAll("Ü","Ü"); sStringToProcess = sStringToProcess.replaceAll("Ý","Ý"); sStringToProcess = sStringToProcess.replaceAll("Þ","Þ"); sStringToProcess = sStringToProcess.replaceAll("ß","ß"); sStringToProcess = sStringToProcess.replaceAll("à","à"); sStringToProcess = sStringToProcess.replaceAll("á","á"); sStringToProcess = sStringToProcess.replaceAll("â","â"); sStringToProcess = sStringToProcess.replaceAll("ã","ã"); sStringToProcess = sStringToProcess.replaceAll("ä","ä"); sStringToProcess = sStringToProcess.replaceAll("å","å"); sStringToProcess = sStringToProcess.replaceAll("æ","æ"); sStringToProcess = sStringToProcess.replaceAll("ç","ç"); sStringToProcess = sStringToProcess.replaceAll("è","è"); sStringToProcess = sStringToProcess.replaceAll("é","é"); sStringToProcess = sStringToProcess.replaceAll("ê","ê"); sStringToProcess = sStringToProcess.replaceAll("ë","ë"); sStringToProcess = sStringToProcess.replaceAll("ì","ì"); sStringToProcess = sStringToProcess.replaceAll("í","í"); sStringToProcess = sStringToProcess.replaceAll("î","î"); sStringToProcess = sStringToProcess.replaceAll("ï","ï"); sStringToProcess = sStringToProcess.replaceAll("ð","ð"); sStringToProcess = sStringToProcess.replaceAll("ñ","ñ"); sStringToProcess = sStringToProcess.replaceAll("ò","ò"); sStringToProcess = sStringToProcess.replaceAll("ó","ó"); sStringToProcess = sStringToProcess.replaceAll("ô","ô"); sStringToProcess = sStringToProcess.replaceAll("õ","õ"); sStringToProcess = sStringToProcess.replaceAll("ö","ö"); sStringToProcess = sStringToProcess.replaceAll("÷","÷"); sStringToProcess = sStringToProcess.replaceAll("ø","ø"); sStringToProcess = sStringToProcess.replaceAll("ù","ù"); sStringToProcess = sStringToProcess.replaceAll("ú","ú"); sStringToProcess = sStringToProcess.replaceAll("û","û"); sStringToProcess = sStringToProcess.replaceAll("ü","ü"); sStringToProcess = sStringToProcess.replaceAll("ý","ý"); sStringToProcess = sStringToProcess.replaceAll("þ","þ"); sStringToProcess = sStringToProcess.replaceAll("ÿ","ÿ"); //sStringToProcess = sStringToProcess.replaceAll("","’"); //sStringToProcess = sStringToProcess.replaceAll("","’"); //sStringToProcess = sStringToProcess.replaceAll("","‚"); sStringToProcess = sStringToProcess.replaceAll("","‚"); //sStringToProcess = sStringToProcess.replaceAll("","’"); sStringToProcess = sStringToProcess.replaceAll("","‘"); sStringToProcess = sStringToProcess.replaceAll("","&euro;"); } return sStringToProcess; } public void generateRssAll() { try { String sArticle = ""; CmsFile fileForAvantGo; System.out.println("FXRssResource - generateRssAll new v 5"); List resultList = fxres.getLatestFiles("/sites/xxx", 15); SimpleDateFormat inputDateFormat = new SimpleDateFormat(FXConstants.DATE_FORMAT); String sRssItem = ""; String sRssItemFullAndImages = ""; String sRssItemKeywords = ""; String sRssItemFull = ""; String sxxxxTR_news = ""; int iUniqueIDofFile; int iCountOfMaxArticlesForxxxxTR = 0; FXListItem item; Newfxitem: for (int i = 0; i < resultList.size(); i++) { item = (FXListItem) resultList.get(i); //MAKE THE RSS String sArticleXMLElementText1plainText = ""; String sArticleXMLElementText2plainText = ""; String sArticleXMLElementText1HTML = ""; String sArticleXMLElementText2HTML = ""; while( Pattern.matches("(.*?)2005(.*?)", item.getDate() ) ){ System.out.println("Found year 2005 in article given as item.getDate() = ["+item.getDate()+"]"); i++; item = (FXListItem) resultList.get(i); } System.out.println("About to get a new fileResource and file"); String fileResource = ""; CmsFile file = null; try { fileResource = item.getFile(); //"/story/charting_center/weekly_chart_analysis/5085_eur_gbp_large_triangle_setup.html"; file = cmsObject.readFile(sfSiteRoot + fileResource); if(existsResource(cmsObject, fileResource) ){ System.out.println("Resource read "); } else { System.out.println("Resource failed to be read, getting next one"); i++; item = (FXListItem) resultList.get(i); fileResource = item.getFile(); //"/story/charting_center/weekly_chart_analysis/5085_eur_gbp_large_triangle_setup.html"; file = cmsObject.readFile(sfSiteRoot + fileResource); } } catch(Exception exc) { System.out.println("exception occured in getFile or redFile!"); System.out.println("fileResource = " + fileResource); System.out.println("sfSiteRoot = " + sfSiteRoot); System.out.println("trying next"); i++; item = (FXListItem) resultList.get(i); fileResource = item.getFile(); //"/story/charting_center/weekly_chart_analysis/5085_eur_gbp_large_triangle_setup.html"; file = cmsObject.readFile(sfSiteRoot + fileResource); } //fileResource if (fileResource.contains("_xxxxtr")){ System.out.println("FOUND _xxxxtr in path of 'fileResource' = [" + fileResource + "] it is only for xxx+ news, therefore skipping for RSS generation and broadcast"); continue Newfxitem; } if (fileResource.contains("guest_trade_ideas")){ System.out.println("FOUND guest_trade_ideas in path of 'fileResource' = [" + fileResource + "] it is only for xxx+ news, therefore skipping for RSS generation and broadcast"); continue Newfxitem; } iUniqueIDofFile = java.lang.Math.abs( file.getResourceId().hashCode() ); try { // get default locale from the new article Locale locale = (Locale)OpenCms.getLocaleManager().getDefaultLocales(cmsObject, file.getRootPath()).get(0); // load article text (anything in <Content>, should just be text1 & text2) CmsXmlContent fileContent = CmsXmlContentFactory.unmarshal(cmsObject, file, false); List fileList = fileContent.getValues("element", locale); if (fileList.size() > 0) { Iterator iter = fileList.iterator(); while (iter.hasNext()) { sArticleXMLElementText1plainText = ((CmsXmlHtmlValue)iter.next()).getPlainText(cmsObject); sArticleXMLElementText2plainText = ((CmsXmlHtmlValue)iter.next()).getPlainText(cmsObject); } } if (fileList.size() > 0) { Iterator iter = fileList.iterator(); while (iter.hasNext()) { sArticleXMLElementText1HTML = ((CmsXmlHtmlValue)iter.next()).getStringValue(cmsObject); sArticleXMLElementText2HTML = ((CmsXmlHtmlValue)iter.next()).getStringValue(cmsObject); } } sArticleXMLElementText1HTML = replaceSpecialCharacters(sArticleXMLElementText1HTML); sArticleXMLElementText2HTML = replaceSpecialCharacters(sArticleXMLElementText2HTML); sArticleXMLElementText1plainText = replaceSpecialCharacters(sArticleXMLElementText1plainText); sArticleXMLElementText2plainText = replaceSpecialCharacters(sArticleXMLElementText2plainText); } catch (CmsException e) {} String sUniqueIncimentedIdForArticle = ""; ParsePosition pp = new ParsePosition(0); Date theDate = inputDateFormat.parse(item.getDate(), pp); if (theDate == null) { continue; } SimpleDateFormat outputDateFormat = new SimpleDateFormat("yyyy-MM-dd HH:mm:ss"); //SimpleDateFormat outputDateFormat = new SimpleDateFormat("EEE', 'dd' 'MMM' 'yyyy' 'HH:mm:ss' 'Z", Locale.US); String outputDate = outputDateFormat.format(theDate); //dateformat for xxx keywords.xml feed SimpleDateFormat outputDateFormatForKeywords = new SimpleDateFormat("yyyy-MM-dd HH:mm:ss"); //SimpleDateFormat outputDateFormat = new SimpleDateFormat("EEE', 'dd' 'MMM' 'yyyy' 'HH:mm:ss' 'Z", Locale.US); String outputDateForKeywords = outputDateFormatForKeywords.format(theDate); //2006-05-03T17:54:30-04:00 old: Mon, 23 Apr 2007 13:52:34 -0400 //getFile = /story/xxx_reports/Euro_Strengthens_Ahead_of_NFP_1141994436831.html Pattern p = Pattern.compile("(.*)/(.*?).html"); Matcher m = p.matcher(item.getFile()); String sExtractedSecondsStringFromItem = "empty - RegEx for content not found in file or file not loaded"; if (m.find() == true) { sExtractedSecondsStringFromItem = m.group(2); } else { long totalMilliseconds = System.currentTimeMillis(); //sExtractedSecondsStringFromItem = (String) totalMilliseconds.toString(); sExtractedSecondsStringFromItem = String.valueOf(totalMilliseconds) + "_XA"; // comment and for debuging: System.outx.println("DEBUG FXRssResource - first match failed"); } Pattern p1 = Pattern.compile("(.*)_(.*?).html"); Matcher m1 = p1.matcher(item.getFile()); if (m1.find() == true) { sUniqueIncimentedIdForArticle = m1.group(2); } else { sUniqueIncimentedIdForArticle = String.valueOf(iUniqueIDofFile); } //String sForYahooFeedSummary = sArticleXMLElementText1plainText; //I decided to use Text - because all headers for stories on [http://biz.yahoo.com/xxxx/] are in plain text (bodies are in HTML) sArticleXMLElementText1HTML; //String sForYahooFeedBody = sArticleXMLElementText1HTML + "<br><br>" +sArticleXMLElementText2HTML;/* sForYahooFeedBody = sForYahooFeedBody.replaceAll("<o:","<"); //this addresses special Microsoft tags that get into the output from copy paste from MSFT docs by analysts sForYahooFeedBody = sForYahooFeedBody.replaceAll("<o:","<"); //this addresses special Microsoft tags that get into the output from copy paste from MSFT docs by analysts sForYahooFeedBody = sForYahooFeedBody.replaceAll("</o:","</"); //this addresses special Microsoft tags that get into the output from copy paste from MSFT docs by analysts sForYahooFeedBody = sForYahooFeedBody.replaceAll("<O:","<"); //this addresses special Microsoft tags that get into the output from copy paste from MSFT docs by analysts sForYahooFeedBody = sForYahooFeedBody.replaceAll("</O:","</"); //this addresses special Microsoft tags that get into the output from copy paste from MSFT docs by analysts*/ //sArticle = sArticle.replaceAll("src=\"/opencms/opencms/", "src=\"http://www.xxx.com/export/sites/xxx/"); sArticle = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n" + "<article>\n" + //" <id>" + iUniqueIDofFile + "</id>\n" + //" <id>" + sExtractedSecondsStringFromItem + "</id>\n" + " <id>" + sUniqueIncimentedIdForArticle + "</id>\n" + //" <id>" + sExtractedSecondsStringFromItem + "</id>\n" + " <timestamp>" + outputDate + "</timestamp>\n" + " <published>no</published>\n" + " <already_published_link></already_published_link>\n" + " <media_type>text</media_type>\n" + " <access_type>free</access_type>\n" + " <topics>\n" + " <topic>USD</topic>\n" + " </topics>\n" + " <title>" + replaceSpecialCharacters( item.getTitle() ) + "</title>\n" + " <summary><![CDATA[" + replaceSpecialCharacters( sArticleXMLElementText1plainText ) + "<br />\n]]></summary>\n" + " <author>" + replaceSpecialCharacters( item.getAuthor() ) + " strategist@xxx.com</author>\n" + " <industries>\n" + " <industry>Financial - Money Ctr. Banks</industry>\n" + " </industries>\n" + " <body><![CDATA[" + replaceSpecialCharacters(sArticleXMLElementText1HTML + "<br><br>" +sArticleXMLElementText2HTML) + "<br />\n]]>\n" + " </body>\n" + "</article>"; // comment and for debuging: System.outx.println("sArticle = <TEXTAREA>" + sArticle + "</TEXTAREA><HR>"); //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 += " <item>\n" + " <title>" + replaceSpecialCharacters( item.getTitle() ) + "</title>\n" + " <link>http://www.xxx.com" + item.getFile() + "?engine=rss&keyword=article</link>\n" + " <author>" + replaceSpecialCharacters( item.getAuthor() ) + " strategist@xxx.com</author>\n" + " <description>" + sDescriptionForRss + "</description>\n" + " <pubDate>" + outputDateForRssItem + "</pubDate>\n" + //<pubDate>2006-02-16 12:31:12</pubDate> " <guid>http://www.xxx.com" + item.getFile() + "</guid>\n" + //ideally should have tag isPermaLink="false" per http://feedvalidator.org/docs/error/InvalidHttpGUID.html " </item>\n"; CmsProperty propertyObject = cmsObject.readPropertyObject(sfSiteRoot + fileResource, "Keywords", false); String sPropertyKeywords = ""; if (propertyObject != null) { sPropertyKeywords = propertyObject.getValue(); } if (propertyObject == null) { sPropertyKeywords = ""; } if (sPropertyKeywords == null) { sPropertyKeywords = ""; } CmsProperty propertyObject2 = cmsObject.readPropertyObject(sfSiteRoot + fileResource, "Stockticker", false); String sPropertyStockTicker = ""; if (propertyObject2 != null) { sPropertyStockTicker = propertyObject2.getValue(); } if (propertyObject2 == null) { sPropertyStockTicker = ""; } if (sPropertyStockTicker == null) { sPropertyStockTicker = ""; } sRssItemKeywords += " <url>\n" + " <loc>http://www.xxx.com" + item.getFile() + "</loc>\n" + " <news:news>\n" + " <news:publication_date>" + outputDateForRssItemForKeywords + "</news:publication_date>\n" + //new: 2006-05-03T17:54:30-04:00 old: <pubDate>2006-02-16 12:31:12</pubDate> " <news:keywords>Business</news:keywords>\n" + " <news:stock_tickers>" + sPropertyStockTicker + "</news:stock_tickers>\n" + " </news:news>\n" + " </url>\n"; String sRssDescriptionForRssWoImages = sArticleXMLElementText1HTML + "<br /><br />" + sArticleXMLElementText2HTML; sRssDescriptionForRssWoImages = cleanHTML(sRssDescriptionForRssWoImages); sRssDescriptionForRssWoImages= sRssDescriptionForRssWoImages.replaceAll("&", "&"); sRssDescriptionForRssWoImages = sRssDescriptionForRssWoImages.replaceAll("<img(.*?)>", "<br />"); sRssDescriptionForRssWoImages = sRssDescriptionForRssWoImages.replaceAll("<IMG(.*?)>", "<br />"); sRssDescriptionForRssWoImages = sRssDescriptionForRssWoImages.replaceAll("\n", " "); sRssDescriptionForRssWoImages = sRssDescriptionForRssWoImages.replaceAll("<", "<"); sRssDescriptionForRssWoImages = sRssDescriptionForRssWoImages.replaceAll(">", ">"); /* sRssDescriptionForRssWoImages= sRssDescriptionForRssWoImages.replaceAll("’", "'"); sRssDescriptionForRssWoImages= sRssDescriptionForRssWoImages.replaceAll("–", "-"); sRssDescriptionForRssWoImages= sRssDescriptionForRssWoImages.replaceAll("·", "·"); //sRssDescriptionForRssWoImages= sRssDescriptionForRssWoImages.replaceAll("·", "·"); sRssDescriptionForRssWoImages= sRssDescriptionForRssWoImages.replaceAll(" ", " "); sRssDescriptionForRssWoImages= sRssDescriptionForRssWoImages.replaceAll("“", "\""); sRssDescriptionForRssWoImages= sRssDescriptionForRssWoImages.replaceAll("”", "\""); sRssDescriptionForRssWoImages= sRssDescriptionForRssWoImages.replaceAll("•", "·"); sRssDescriptionForRssWoImages= sRssDescriptionForRssWoImages.replaceAll("—", ""); */ //sRssDescriptionForRssWoImages= sRssDescriptionForRssWoImages.replaceAll("&(.*?);", " "); sRssItemFull += " <item>\n" + " <title>" + replaceSpecialCharacters( item.getTitle() ) + "</title>\n" + " <link>http://www.xxx.com" + item.getFile() + "?engine=rss&keyword=article</link>\n" + " <description>" + replaceSpecialCharacters( sRssDescriptionForRssWoImages ) + "</description>\n" + " <pubDate>" + outputDateForRssItem + "</pubDate>\n" + //<pubDate>2006-02-16 12:31:12</pubDate> " <guid>http://www.xxx.com" + item.getFile() + "</guid>\n" + " </item>\n"; //http://www.xxx.com/feeds/full-images.php String sRssDescriptionForRssWithImages = sArticleXMLElementText1HTML + "<br /><br />" + sArticleXMLElementText2HTML; sRssDescriptionForRssWithImages = sRssDescriptionForRssWithImages.replaceAll("&", "&"); sRssDescriptionForRssWithImages = cleanHTML(sRssDescriptionForRssWithImages); sRssDescriptionForRssWithImages = sRssDescriptionForRssWithImages.replaceAll("\n", " "); sRssDescriptionForRssWithImages = sRssDescriptionForRssWithImages.replaceAll("<o:","<"); //this addresses special Microsoft tags that get into the output from copy paste from MSFT docs by analysts sRssDescriptionForRssWithImages = sRssDescriptionForRssWithImages.replaceAll("</o:","</"); //this addresses special Microsoft tags that get into the output from copy paste from MSFT docs by analysts sRssDescriptionForRssWithImages = sRssDescriptionForRssWithImages.replaceAll("<O:","<"); //this addresses special Microsoft tags that get into the output from copy paste from MSFT docs by analysts sRssDescriptionForRssWithImages = sRssDescriptionForRssWithImages.replaceAll("</O:","</"); //this addresses special Microsoft tags that get into the output from copy paste from MSFT docs by analysts sRssDescriptionForRssWithImages = sRssDescriptionForRssWithImages.replaceAll("<", "<"); sRssDescriptionForRssWithImages = sRssDescriptionForRssWithImages.replaceAll(">", ">"); //sRssDescriptionForRssWithImages = sRssDescriptionForRssWithImages.replaceAll("src=\"/story-images/", "src=\"http://www.xxx.com/export/sites/xxx/"); //sRssDescriptionForRssWithImages = sRssDescriptionForRssWithImages.replaceAll("SRC=\"/story-images/", "SRC=\"http://www.xxx.com/export/sites/xxx/"); sRssDescriptionForRssWithImages = normalizeURLs(sRssDescriptionForRssWithImages); //src="http://www.xxx.com/export/sites/xxx/story-images/2006/04/currency_crosses/Crosses/04-13-06cross2.gif" //src="/opencms/opencms/story-images/2006/04/currency_crosses/Crosses/04-13-06cross2.gif" //sRssDescriptionForRssWithImages = sRssDescriptionForRssWithImages.replaceAll("src=\"/opencms/opencms/", "src=\"http://www.xxx.com/export/sites/xxx/"); //sRssDescriptionForRssWithImages = sRssDescriptionForRssWithImages.replaceAll("SRC=\"/opencms/opencms/", "SRC=\"http://www.xxx.com/export/sites/xxx/"); //sRssDescriptionForRssWithImages = sRssDescriptionForRssWithImages.replaceAll("href=\"/opencms/opencms/", "href=\"http://www.xxx.com/export/sites/xxx/"); //sRssDescriptionForRssWithImages = sRssDescriptionForRssWithImages.replaceAll("HREF=\"/opencms/opencms/", "HREF=\"http://www.xxx.com/export/sites/xxx/"); // http://xxx.opencms.prod.xxxx.com/export/sites/xxx/story-images/archive/daily_fundamentals/ssi_3_30_06_1.gif // http://xxx.opencms.prod.xxxx.com/export/sites/xxx/story-images/archive/market_brief/03-29-06MB.gif sRssItemFullAndImages += " <item>\n" + " <title>" + item.getTitle() + "</title>\n" + " <link>http://www.xxx.com" + item.getFile() + "?engine=rss&keyword=article</link>\n" + " <author>" + replaceSpecialCharacters( item.getAuthor() ) + " strategist@xxx.com</author>\n" + " <description>" + sRssDescriptionForRssWithImages + "</description>\n" + " <pubDate>" + outputDateForRssItem + "</pubDate>\n" + //<pubDate>2006-02-16 12:31:12</pubDate> " <guid>http://www.xxx.com" + item.getFile() + "</guid>\n" + " </item>\n"; Tidy tidy = new Tidy(); tidy.setWord2000(true); if (iCountOfMaxArticlesForxxxxTR++ < 7){ sxxxxTR_news += "<div class=\"headlinesDate\">" + item.getDate() + "</div>\n" + "<a name=\"#"+iUniqueIDofFile+"_b\"></a>" + "<div id=\"plus_sign\" class=\"plus_sign\">\n" + "<a style=\"font-family:Verdana, sans-serif; font-size:14px; font-weight:bold; color:#E86F0D\" class=\"effect10\" onClick=\"javascript:displayContent('"+iUniqueIDofFile+"')\" href=\"#"+iUniqueIDofFile+"_b\">\n" + "<div>" + item.getTitle() + "</div>\n" + "<img border=\"0\" height=\"1\" width=\"1\" src=\"/images/plus.gif\"></a>\n" + "</div>\n" + "<a name=\"#"+iUniqueIDofFile+"_b\"></a>\n" + "<div style=\"display:none;\" id=\""+iUniqueIDofFile+"\" class=\""+iUniqueIDofFile+"_b\">\n" + "<div id=\"minus_sign\" class=\"minus_sign\">\n" + "<a onClick=\"displayContent('"+iUniqueIDofFile+"');\" href=\"#"+iUniqueIDofFile+"_b\"><img alt=\"expand article\" border=\"0\" height=\"18\" width=\"96\" src=\"/images/minus.gif\"></a>\n" + "</div>\n" + "<div style=\"font-family:Georgia,serif;font-size:11px;font-weight:normal;color:#30303E;line-height:19px;text-align:left\">" + item.getAuthor() + "<br>\n" + sRssDescriptionForRssWithImages + "<hr>\n" + "</div>\n" + "</div>\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 = "<?xml version=\"1.0\" encoding=\"UTF-8\" ?> \n" + " <rss version=\"2.0\">\n" + " <channel>\n" + " <title>Currency Trading News by xxx</title> \n" + " <link>http://www.xxx.com</link> \n" + " <language>en-us</language>\n" + " <description>Forex news about foreign exchange rates and reports on world business events that affect the currency market provided by xxx.</description> \n" + " <lastBuildDate>" + sOutputDateTimeNowForRss + "</lastBuildDate> \n"; //Tue, 14 Feb 2006 20:57:44 +0100 String sRssEnding = " </channel>\n</rss>\n"; String sRssStartKeywords = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n" + "<urlset xmlns=\"http://www.sitemaps.org/schemas/sitemap/0.9\"\n" + " xmlns:news=\"http://www.google.com/schemas/sitemap-news/0.9\">\n"; String sRssEndingKeywords = "</urlset>\n"; String sRssRegularFeed = sRssStart + sRssItem + sRssEnding; // comment and for debuging: System.outx.println("sRssRegularFeed = <TEXTAREA COLS=70 ROWS=20>" + sRssRegularFeed + "</TEXTAREA><HR>"); String sRssRegularFeedFullWithoutImages = sRssStart + sRssItemFull + sRssEnding; // comment and for debuging: System.outx.println("sRssRegularFeed = <TEXTAREA COLS=70 ROWS=20>" + sRssRegularFeedFullWithoutImages + "</TEXTAREA><HR>"); String sRssRegularFeedFullWithImages = sRssStart + sRssItemFullAndImages + sRssEnding; // comment and for debuging: System.outx.println("sRssRegularFeed = <TEXTAREA COLS=70 ROWS=20>" + sRssRegularFeedFullWithImages + "</TEXTAREA><HR>"); 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("<img(.*?)>", "<br />"); //<img(.*?)>", "<br />"); //<img title="funda 5 5 06_1" alt="funda 5 5 06_1" src="http://www.xxx.com/export/sites/xxx/story-images/2006/05/xxx_reports sRssRegularFeedFullWithoutImages2 = sRssRegularFeedFullWithoutImages2.replaceAll("<IMG(.*?)>", "<br />"); //http://www.xxx.com/feeds/full.php writeStringAsFileToVFS(cmsObject, sRssRegularFeedFullWithoutImages2, sfSiteRoot + "/" + "feeds/full-noimages.xml"); writeStringAsFileToVFS(cmsObject, sRssRegularFeedFullWithoutImages2, sfSiteRoot + "/" + "feeds/full2.xml"); writeStringAsFileToVFS(cmsObject, sRssRegularFeedFullWithoutImages2, sfSiteRoot + "/" + "feeds/full.xml"); String sAvantGoDocument = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n" +"<site>\n" +" <page>\n" +" <pageid>1</pageid>\n" +" <content>\n" +" <a href="article.1." + avantGoGetArticleData("/sites/xxx", "", 0, "fileId", cmsObject, fxres) + ".php">" + avantGoGetArticleData("/sites/xxx", "", 0, "title", cmsObject, fxres) + "</a> " + avantGoGetArticleData("/sites/xxx", "", 0, "introHTML", cmsObject, fxres) + "&amp;nbsp;<a href="article.1." + avantGoGetArticleData("/sites/xxx", "", 0, "fileId", cmsObject, fxres) + ".php">Read More</a><br /><br />\n" +" <a href="article.1." + avantGoGetArticleData("/sites/xxx", "", 1, "fileId", cmsObject, fxres) + ".php">" + avantGoGetArticleData("/sites/xxx", "", 1, "title", cmsObject, fxres) + "</a> " + avantGoGetArticleData("/sites/xxx", "", 1, "introHTML", cmsObject, fxres) + "&amp;nbsp;<a href="article.1." + avantGoGetArticleData("/sites/xxx", "", 1, "fileId", cmsObject, fxres) + ".php">Read More</a><br /><br />\n" +" <a href="article.1." + avantGoGetArticleData("/sites/xxx", "", 2, "fileId", cmsObject, fxres) + ".php">" + avantGoGetArticleData("/sites/xxx", "", 2, "title", cmsObject, fxres) + "</a> " + avantGoGetArticleData("/sites/xxx", "", 2, "introHTML", cmsObject, fxres) + "&amp;nbsp;<a href="article.1." + avantGoGetArticleData("/sites/xxx", "", 2, "fileId", cmsObject, fxres) + ".php">Read More</a><br /><br />\n" +" <a href="article.1." + avantGoGetArticleData("/sites/xxx", "", 3, "fileId", cmsObject, fxres) + ".php">" + avantGoGetArticleData("/sites/xxx", "", 3, "title", cmsObject, fxres) + "</a> " + avantGoGetArticleData("/sites/xxx", "", 3, "introHTML", cmsObject, fxres) + "&amp;nbsp;<a href="article.1." + avantGoGetArticleData("/sites/xxx", "", 3, "fileId", cmsObject, fxres) + ".php">Read More</a><br /><br />\n" +" <a href="article.1." + avantGoGetArticleData("/sites/xxx", "", 4, "fileId", cmsObject, fxres) + ".php">" + avantGoGetArticleData("/sites/xxx", "", 4, "title", cmsObject, fxres) + "</a> " + avantGoGetArticleData("/sites/xxx", "", 4, "introHTML", cmsObject, fxres) + "&amp;nbsp;<a href="article.1." + avantGoGetArticleData("/sites/xxx", "", 4, "fileId", cmsObject, fxres) + ".php">Read More</a><br /><br />\n" +" </content>\n" +" </page>\n" + " <page>\n" +" <pageid>4</pageid>\n" +" <content>\n" +" <a href="article.4." + avantGoGetArticleData("/sites/xxx", "/story/currency/eur_fundamentals", 0, "fileId", cmsObject, fxres) + ".php">" + avantGoGetArticleData("/sites/xxx", "/story/currency/eur_fundamentals", 0, "title", cmsObject, fxres) + "</a> " + avantGoGetArticleData("/sites/xxx", "/story/currency/eur_fundamentals", 0, "introHTML", cmsObject, fxres) + "&amp;nbsp;<a href="article.4." + avantGoGetArticleData("/sites/xxx", "/story/currency/eur_fundamentals", 0, "fileId", cmsObject, fxres) + ".php">Read More</a><br /><br />\n" +" <a href="article.4." + avantGoGetArticleData("/sites/xxx", "/story/currency/eur_fundamentals", 1, "fileId", cmsObject, fxres) + ".php">" + avantGoGetArticleData("/sites/xxx", "/story/currency/eur_fundamentals", 1, "title", cmsObject, fxres) + "</a> " + avantGoGetArticleData("/sites/xxx", "/story/currency/eur_fundamentals", 1, "introHTML", cmsObject, fxres) + "&amp;nbsp;<a href="article.4." + avantGoGetArticleData("/sites/xxx", "/story/currency/eur_fundamentals", 1, "fileId", cmsObject, fxres) + ".php">Read More</a><br /><br />\n" +" </content>\n" +" </page>\n" +" <page>\n" +" <pageid>5</pageid>\n" +" <content>\n" +" <a href="article.5." + avantGoGetArticleData("/sites/xxx", "/story/currency/jpy_fundamentals", 0, "fileId", cmsObject, fxres) + ".php">" + avantGoGetArticleData("/sites/xxx", "/story/currency/jpy_fundamentals", 0, "title", cmsObject, fxres) + "</a> " + avantGoGetArticleData("/sites/xxx", "/story/currency/jpy_fundamentals", 0, "introHTML", cmsObject, fxres) + "&amp;nbsp;<a href="article.5." + avantGoGetArticleData("/sites/xxx", "/story/currency/jpy_fundamentals", 0, "fileId", cmsObject, fxres) + ".php">Read More</a><br /><br />\n" +" <a href="article.5." + avantGoGetArticleData("/sites/xxx", "/story/currency/jpy_fundamentals", 1, "fileId", cmsObject, fxres) + ".php">" + avantGoGetArticleData("/sites/xxx", "/story/currency/jpy_fundamentals", 1, "title", cmsObject, fxres) + "</a> " + avantGoGetArticleData("/sites/xxx", "/story/currency/jpy_fundamentals", 1, "introHTML", cmsObject, fxres) + "&amp;nbsp;<a href="article.5." + avantGoGetArticleData("/sites/xxx", "/story/currency/jpy_fundamentals", 1, "fileId", cmsObject, fxres) + ".php">Read More</a><br /><br />\n" +" </content>\n" +" </page>\n" +" <page>\n" +" <pageid>6</pageid>\n" +" <content>\n" +" <a href="article.6." + avantGoGetArticleData("/sites/xxx", "/story/currency/gbp_fundamentals", 0, "fileId", cmsObject, fxres) + ".php">" + avantGoGetArticleData("/sites/xxx", "/story/currency/gbp_fundamentals", 0, "title", cmsObject, fxres) + "</a> " + avantGoGetArticleData("/sites/xxx", "/story/currency/gbp_fundamentals", 0, "introHTML", cmsObject, fxres) + "&amp;nbsp;<a href="article.6." + avantGoGetArticleData("/sites/xxx", "/story/currency/gbp_fundamentals", 0, "fileId", cmsObject, fxres) + ".php">Read More</a><br /><br />\n" +" <a href="article.6." + avantGoGetArticleData("/sites/xxx", "/story/currency/gbp_fundamentals", 1, "fileId", cmsObject, fxres) + ".php">" + avantGoGetArticleData("/sites/xxx", "/story/currency/gbp_fundamentals", 1, "title", cmsObject, fxres) + "</a> " + avantGoGetArticleData("/sites/xxx", "/story/currency/gbp_fundamentals", 1, "introHTML", cmsObject, fxres) + "&amp;nbsp;<a href="article.6." + avantGoGetArticleData("/sites/xxx", "/story/currency/gbp_fundamentals", 1, "fileId", cmsObject, fxres) + ".php">Read More</a><br /><br />\n" +" </content>\n" +" </page>\n" +" <page>\n" +" <pageid>7</pageid>\n" +" <content>\n" +" <a href="article.7." + avantGoGetArticleData("/sites/xxx", "/story/currency/aud_fundamentals", 0, "fileId", cmsObject, fxres) + ".php">" + avantGoGetArticleData("/sites/xxx", "/story/currency/aud_fundamentals", 0, "title", cmsObject, fxres) + "</a> " + avantGoGetArticleData("/sites/xxx", "/story/currency/aud_fundamentals", 0, "introHTML", cmsObject, fxres) + "&amp;nbsp;<a href="article.7." + avantGoGetArticleData("/sites/xxx", "/story/currency/aud_fundamentals", 0, "fileId", cmsObject, fxres) + ".php">Read More</a><br /><br />\n" +" <a href="article.7." + avantGoGetArticleData("/sites/xxx", "/story/currency/aud_fundamentals", 1, "fileId", cmsObject, fxres) + ".php">" + avantGoGetArticleData("/sites/xxx", "/story/currency/aud_fundamentals", 1, "title", cmsObject, fxres) + "</a> " + avantGoGetArticleData("/sites/xxx", "/story/currency/aud_fundamentals", 1, "introHTML", cmsObject, fxres) + "&amp;nbsp;<a href="article.7." + avantGoGetArticleData("/sites/xxx", "/story/currency/aud_fundamentals", 1, "fileId", cmsObject, fxres) + ".php">Read More</a><br /><br />\n" +" </content>\n" +" </page>\n" +" <page>\n" +" <pageid>8</pageid>\n" +" <content>\n" +" <a href="article.8." + avantGoGetArticleData("/sites/xxx", "/story/currency/cad_fundamentals", 0, "fileId", cmsObject, fxres) + ".php">" + avantGoGetArticleData("/sites/xxx", "/story/currency/cad_fundamentals", 0, "title", cmsObject, fxres) + "</a> " + avantGoGetArticleData("/sites/xxx", "/story/currency/cad_fundamentals", 0, "introHTML", cmsObject, fxres) + "&amp;nbsp;<a href="article.8." + avantGoGetArticleData("/sites/xxx", "/story/currency/cad_fundamentals", 0, "fileId", cmsObject, fxres) + ".php">Read More</a><br /><br />\n" +" <a href="article.8." + avantGoGetArticleData("/sites/xxx", "/story/currency/cad_fundamentals", 1, "fileId", cmsObject, fxres) + ".php">" + avantGoGetArticleData("/sites/xxx", "/story/currency/cad_fundamentals", 1, "title", cmsObject, fxres) + "</a> " + avantGoGetArticleData("/sites/xxx", "/story/currency/cad_fundamentals", 1, "introHTML", cmsObject, fxres) + "&amp;nbsp;<a href="article.8." + avantGoGetArticleData("/sites/xxx", "/story/currency/cad_fundamentals", 1, "fileId", cmsObject, fxres) + ".php">Read More</a><br /><br />\n" +" </content>\n" +" </page>\n" +" <page>\n" +" <pageid>9</pageid>\n" +" <content>\n" +" <a href="article.9." + avantGoGetArticleData("/sites/xxx", "/story/currency/nzd_fundamentals", 0, "fileId", cmsObject, fxres) + ".php">" + avantGoGetArticleData("/sites/xxx", "/story/currency/nzd_fundamentals", 0, "title", cmsObject, fxres) + "</a> " + avantGoGetArticleData("/sites/xxx", "/story/currency/nzd_fundamentals", 0, "introHTML", cmsObject, fxres) + "&amp;nbsp;<a href="article.9." + avantGoGetArticleData("/sites/xxx", "/story/currency/nzd_fundamentals", 0, "fileId", cmsObject, fxres) + ".php">Read More</a><br /><br />\n" +" <a href="article.9." + avantGoGetArticleData("/sites/xxx", "/story/currency/nzd_fundamentals", 1, "fileId", cmsObject, fxres) + ".php">" + avantGoGetArticleData("/sites/xxx", "/story/currency/nzd_fundamentals", 1, "title", cmsObject, fxres) + "</a> " + avantGoGetArticleData("/sites/xxx", "/story/currency/nzd_fundamentals", 1, "introHTML", cmsObject, fxres) + "&amp;nbsp;<a href="article.9." + avantGoGetArticleData("/sites/xxx", "/story/currency/nzd_fundamentals", 1, "fileId", cmsObject, fxres) + ".php">Read More</a><br /><br />\n" +" </content>\n" +" </page>\n" +" <page>\n" +" <pageid>10</pageid>\n" +" <content>\n" +" <a href="article.10." + avantGoGetArticleData("/sites/xxx", "/story/currency/chf_fundamentals", 0, "fileId", cmsObject, fxres) + ".php">" + avantGoGetArticleData("/sites/xxx", "/story/currency/chf_fundamentals", 0, "title", cmsObject, fxres) + "</a> " + avantGoGetArticleData("/sites/xxx", "/story/currency/chf_fundamentals", 0, "introHTML", cmsObject, fxres) + "&amp;nbsp;<a href="article.10." + avantGoGetArticleData("/sites/xxx", "/story/currency/chf_fundamentals", 0, "fileId", cmsObject, fxres) + ".php">Read More</a><br /><br />\n" +" <a href="article.10." + avantGoGetArticleData("/sites/xxx", "/story/currency/chf_fundamentals", 1, "fileId", cmsObject, fxres) + ".php">" + avantGoGetArticleData("/sites/xxx", "/story/currency/chf_fundamentals", 1, "title", cmsObject, fxres) + "</a> " + avantGoGetArticleData("/sites/xxx", "/story/currency/chf_fundamentals", 1, "introHTML", cmsObject, fxres) + "&amp;nbsp;<a href="article.10." + avantGoGetArticleData("/sites/xxx", "/story/currency/chf_fundamentals", 1, "fileId", cmsObject, fxres) + ".php">Read More</a><br /><br />\n" +" </content>\n" +" </page>\n" +" <page>\n" +" <article>\n" +" <pageid>3</pageid>\n" +" <content>\n" +" " + avantGoGetArticleData("/sites/xxx", "calendar", 0, "firstElementOnly", cmsObject, fxres) + "Please visit xxx.com for detailed information and news about each event.<br />Data supplied by Briefing.com</content>\n" +" </article>\n" +" </page>\n" +" <article>\n" +" <id>" + avantGoGetArticleData("/sites/xxx", "", 0, "fileId", cmsObject, fxres) + "</id>\n" +" <pageid>1</pageid>\n" +" <title>" + avantGoGetArticleData("/sites/xxx", "", 0, "title", cmsObject, fxres) + "</title>\n" +" <date>" + avantGoGetArticleData("/sites/xxx", "", 0, "time", cmsObject, fxres) + "</date>\n" +" <author>" + avantGoGetArticleData("/sites/xxx", "", 0, "author", cmsObject, fxres) + "</author>\n" +" <content>" + avantGoGetArticleData("/sites/xxx", "", 0, "fullHTML", cmsObject, fxres) + "</content>\n" +" </article>\n" +" <article>\n" +" <id>" + avantGoGetArticleData("/sites/xxx", "", 1, "fileId", cmsObject, fxres) + "</id>\n" +" <pageid>1</pageid>\n" +" <title>" + avantGoGetArticleData("/sites/xxx", "", 1, "title", cmsObject, fxres) + "</title>\n" +" <date>" + avantGoGetArticleData("/sites/xxx", "", 1, "time", cmsObject, fxres) + "</date>\n" +" <author>" + avantGoGetArticleData("/sites/xxx", "", 1, "author", cmsObject, fxres) + "</author>\n" +" <content>" + avantGoGetArticleData("/sites/xxx", "", 1, "fullHTML", cmsObject, fxres) + "</content>\n" +" </article>\n" +" <article>\n" +" <id>" + avantGoGetArticleData("/sites/xxx", "", 2, "fileId", cmsObject, fxres) + "</id>\n" +" <pageid>1</pageid>\n" +" <title>" + avantGoGetArticleData("/sites/xxx", "", 2, "title", cmsObject, fxres) + "</title>\n" +" <date>" + avantGoGetArticleData("/sites/xxx", "", 2, "time", cmsObject, fxres) + "</date>\n" +" <author>" + avantGoGetArticleData("/sites/xxx", "", 2, "author", cmsObject, fxres) + "</author>\n" +" <content>" + avantGoGetArticleData("/sites/xxx", "", 2, "fullHTML", cmsObject, fxres) + "</content>\n" +" </article>\n" +" <article>\n" +" <id>" + avantGoGetArticleData("/sites/xxx", "", 3, "fileId", cmsObject, fxres) + "</id>\n" +" <pageid>1</pageid>\n" +" <title>" + avantGoGetArticleData("/sites/xxx", "", 3, "title", cmsObject, fxres) + "</title>\n" +" <date>" + avantGoGetArticleData("/sites/xxx", "", 3, "time", cmsObject, fxres) + "</date>\n" +" <author>" + avantGoGetArticleData("/sites/xxx", "", 3, "author", cmsObject, fxres) + "</author>\n" +" <content>" + avantGoGetArticleData("/sites/xxx", "", 3, "fullHTML", cmsObject, fxres) + "</content>\n" +" </article>\n" +" <article>\n" +" <id>" + avantGoGetArticleData("/sites/xxx", "", 4, "fileId", cmsObject, fxres) + "</id>\n" +" <pageid>1</pageid>\n" +" <title>" + avantGoGetArticleData("/sites/xxx", "", 4, "title", cmsObject, fxres) + "</title>\n" +" <date>" + avantGoGetArticleData("/sites/xxx", "", 4, "time", cmsObject, fxres) + "</date>\n" +" <author>" + avantGoGetArticleData("/sites/xxx", "", 4, "author", cmsObject, fxres) + "</author>\n" +" <content>" + avantGoGetArticleData("/sites/xxx", "", 4, "fullHTML", cmsObject, fxres) + "</content>\n" +" </article>\n" +" <article>\n" +" <id>0</id>\n" + //why is this 0 - shouldn't this be a uniqe id so that AvantGo would know that they need to update this aricle data to clients?" <pageid>1</pageid>\n" +" <title>" + avantGoGetArticleData("/sites/xxx", "/story/xxx_reports/daily_brief", 0, "title", cmsObject, fxres) + "</title>\n" +" <date>" + avantGoGetArticleData("/sites/xxx", "/story/xxx_reports/daily_brief", 0, "time", cmsObject, fxres) + "</date>\n" +" <author>" + avantGoGetArticleData("/sites/xxx", "/story/xxx_reports/daily_brief", 0, "author", cmsObject, fxres) + "</author>\n" +" <content>" + avantGoGetArticleData("/sites/xxx", "/story/xxx_reports/daily_brief", 0, "fullHTML", cmsObject, fxres) + "</content>\n" +" </article>\n" +" <article>\n" +" <id>" + avantGoGetArticleData("/sites/xxx", "/story/currency/eur_fundamentals", 0, "fileId", cmsObject, fxres) + "</id>\n" +" <pageid>4</pageid>\n" +" <title>" + avantGoGetArticleData("/sites/xxx", "/story/currency/eur_fundamentals", 0, "title", cmsObject, fxres) + "</title>\n" +" <date>" + avantGoGetArticleData("/sites/xxx", "/story/currency/eur_fundamentals", 0, "time", cmsObject, fxres) + "</date>\n" +" <author>" + avantGoGetArticleData("/sites/xxx", "/story/currency/eur_fundamentals", 0, "author", cmsObject, fxres) + "</author>\n" +" <content>" + avantGoGetArticleData("/sites/xxx", "/story/currency/eur_fundamentals", 0, "fullHTML", cmsObject, fxres) + "</content>\n" +" </article>\n" +" <article>\n" +" <id>" + avantGoGetArticleData("/sites/xxx", "/story/currency/eur_fundamentals", 1, "fileId", cmsObject, fxres) + "</id>\n" +" <pageid>4</pageid>\n" +" <title>" + avantGoGetArticleData("/sites/xxx", "/story/currency/eur_fundamentals", 1, "title", cmsObject, fxres) + "</title>\n" +" <date>" + avantGoGetArticleData("/sites/xxx", "/story/currency/eur_fundamentals", 1, "time", cmsObject, fxres) + "</date>\n" +" <author>" + avantGoGetArticleData("/sites/xxx", "/story/currency/eur_fundamentals", 1, "author", cmsObject, fxres) + "</author>\n" +" <content>" + avantGoGetArticleData("/sites/xxx", "/story/currency/eur_fundamentals", 1, "fullHTML", cmsObject, fxres) + "</content>\n" +" </article>\n" +" <article>\n" +" <id>" + avantGoGetArticleData("/sites/xxx", "/story/currency/jpy_fundamentals", 0, "fileId", cmsObject, fxres) + "</id>\n" +" <pageid>5</pageid>\n" +" <title>" + avantGoGetArticleData("/sites/xxx", "/story/currency/jpy_fundamentals", 0, "title", cmsObject, fxres) + "</title>\n" +" <date>" + avantGoGetArticleData("/sites/xxx", "/story/currency/jpy_fundamentals", 0, "time", cmsObject, fxres) + "</date>\n" +" <author>" + avantGoGetArticleData("/sites/xxx", "/story/currency/jpy_fundamentals", 0, "author", cmsObject, fxres) + "</author>\n" +" <content>" + avantGoGetArticleData("/sites/xxx", "/story/currency/jpy_fundamentals", 0, "fullHTML", cmsObject, fxres) + "</content>\n" +" </article>\n" +" <article>\n" +" <id>" + avantGoGetArticleData("/sites/xxx", "/story/currency/jpy_fundamentals", 1, "fileId", cmsObject, fxres) + "</id>\n" +" <pageid>5</pageid>\n" +" <title>" + avantGoGetArticleData("/sites/xxx", "/story/currency/jpy_fundamentals", 1, "title", cmsObject, fxres) + "</title>\n" +" <date>" + avantGoGetArticleData("/sites/xxx", "/story/currency/jpy_fundamentals", 1, "time", cmsObject, fxres) + "</date>\n" +" <author>" + avantGoGetArticleData("/sites/xxx", "/story/currency/jpy_fundamentals", 1, "author", cmsObject, fxres) + "</author>\n" +" <content>" + avantGoGetArticleData("/sites/xxx", "/story/currency/jpy_fundamentals", 1, "fullHTML", cmsObject, fxres) + "</content>\n" +" </article>\n" +" <article>\n" +" <id>" + avantGoGetArticleData("/sites/xxx", "/story/currency/gbp_fundamentals", 0, "fileId", cmsObject, fxres) + "</id>\n" +" <pageid>6</pageid>\n" +" <title>" + avantGoGetArticleData("/sites/xxx", "/story/currency/gbp_fundamentals", 0, "title", cmsObject, fxres) + "</title>\n" +" <date>" + avantGoGetArticleData("/sites/xxx", "/story/currency/gbp_fundamentals", 0, "time", cmsObject, fxres) + "</date>\n" +" <author>" + avantGoGetArticleData("/sites/xxx", "/story/currency/gbp_fundamentals", 0, "author", cmsObject, fxres) + "</author>\n" +" <content>" + avantGoGetArticleData("/sites/xxx", "/story/currency/gbp_fundamentals", 0, "fullHTML", cmsObject, fxres) + "</content>\n" +" </article>\n" +" <article>\n" +" <id>" + avantGoGetArticleData("/sites/xxx", "/story/currency/gbp_fundamentals", 1, "fileId", cmsObject, fxres) + "</id>\n" +" <pageid>6</pageid>\n" +" <title>" + avantGoGetArticleData("/sites/xxx", "/story/currency/gbp_fundamentals", 1, "title", cmsObject, fxres) + "</title>\n" +" <date>" + avantGoGetArticleData("/sites/xxx", "/story/currency/gbp_fundamentals", 1, "time", cmsObject, fxres) + "</date>\n" +" <author>" + avantGoGetArticleData("/sites/xxx", "/story/currency/gbp_fundamentals", 1, "author", cmsObject, fxres) + "</author>\n" +" <content>" + avantGoGetArticleData("/sites/xxx", "/story/currency/gbp_fundamentals", 1, "fullHTML", cmsObject, fxres) + "</content>\n" +" </article>\n" +" <article>\n" +" <id>" + avantGoGetArticleData("/sites/xxx", "/story/currency/aud_fundamentals", 0, "fileId", cmsObject, fxres) + "</id>\n" +" <pageid>7</pageid>\n" +" <title>" + avantGoGetArticleData("/sites/xxx", "/story/currency/aud_fundamentals", 0, "title", cmsObject, fxres) + "</title>\n" +" <date>" + avantGoGetArticleData("/sites/xxx", "/story/currency/aud_fundamentals", 0, "time", cmsObject, fxres) + "</date>\n" +" <author>" + avantGoGetArticleData("/sites/xxx", "/story/currency/aud_fundamentals", 0, "author", cmsObject, fxres) + "</author>\n" +" <content>" + avantGoGetArticleData("/sites/xxx", "/story/currency/aud_fundamentals", 0, "fullHTML", cmsObject, fxres) + "</content>\n" +" </article>\n" +" <article>\n" +" <id>" + avantGoGetArticleData("/sites/xxx", "/story/currency/aud_fundamentals", 1, "fileId", cmsObject, fxres) + "</id>\n" +" <pageid>7</pageid>\n" +" <title>" + avantGoGetArticleData("/sites/xxx", "/story/currency/aud_fundamentals", 1, "title", cmsObject, fxres) + "</title>\n" +" <date>" + avantGoGetArticleData("/sites/xxx", "/story/currency/aud_fundamentals", 1, "time", cmsObject, fxres) + "</date>\n" +" <author>" + avantGoGetArticleData("/sites/xxx", "/story/currency/aud_fundamentals", 1, "author", cmsObject, fxres) + "</author>\n" +" <content>" + avantGoGetArticleData("/sites/xxx", "/story/currency/aud_fundamentals", 1, "fullHTML", cmsObject, fxres) + "</content>\n" +" </article>\n" +" <article>\n" +" <id>" + avantGoGetArticleData("/sites/xxx", "/story/currency/cad_fundamentals", 0, "fileId", cmsObject, fxres) + "</id>\n" +" <pageid>8</pageid>\n" +" <title>" + avantGoGetArticleData("/sites/xxx", "/story/currency/cad_fundamentals", 0, "title", cmsObject, fxres) + "</title>\n" +" <date>" + avantGoGetArticleData("/sites/xxx", "/story/currency/cad_fundamentals", 0, "time", cmsObject, fxres) + "</date>\n" +" <author>" + avantGoGetArticleData("/sites/xxx", "/story/currency/cad_fundamentals", 0, "author", cmsObject, fxres) + "</author>\n" +" <content>" + avantGoGetArticleData("/sites/xxx", "/story/currency/cad_fundamentals", 0, "fullHTML", cmsObject, fxres) + "</content>\n" +" </article>\n" +" <article>\n" +" <id>" + avantGoGetArticleData("/sites/xxx", "/story/currency/cad_fundamentals", 1, "fileId", cmsObject, fxres) + "</id>\n" +" <pageid>8</pageid>\n" +" <title>" + avantGoGetArticleData("/sites/xxx", "/story/currency/cad_fundamentals", 1, "title", cmsObject, fxres) + "</title>\n" +" <date>" + avantGoGetArticleData("/sites/xxx", "/story/currency/cad_fundamentals", 1, "time", cmsObject, fxres) + "</date>\n" +" <author>" + avantGoGetArticleData("/sites/xxx", "/story/currency/cad_fundamentals", 1, "author", cmsObject, fxres) + "</author>\n" +" <content>" + avantGoGetArticleData("/sites/xxx", "/story/currency/cad_fundamentals", 1, "fullHTML", cmsObject, fxres) + "</content>\n" +" </article>\n" +" <article>\n" +" <id>" + avantGoGetArticleData("/sites/xxx", "/story/currency/nzd_fundamentals", 0, "fileId", cmsObject, fxres) + "</id>\n" +" <pageid>9</pageid>\n" +" <title>" + avantGoGetArticleData("/sites/xxx", "/story/currency/nzd_fundamentals", 0, "title", cmsObject, fxres) + "</title>\n" +" <date>" + avantGoGetArticleData("/sites/xxx", "/story/currency/nzd_fundamentals", 0, "time", cmsObject, fxres) + "</date>\n" +" <author>" + avantGoGetArticleData("/sites/xxx", "/story/currency/nzd_fundamentals", 0, "author", cmsObject, fxres) + "</author>\n" +" <content>" + avantGoGetArticleData("/sites/xxx", "/story/currency/nzd_fundamentals", 0, "fullHTML", cmsObject, fxres) + "</content>\n" +" </article>\n" +" <article>\n" +" <id>" + avantGoGetArticleData("/sites/xxx", "/story/currency/nzd_fundamentals", 1, "fileId", cmsObject, fxres) + "</id>\n" +" <pageid>9</pageid>\n" +" <title>" + avantGoGetArticleData("/sites/xxx", "/story/currency/nzd_fundamentals", 1, "title", cmsObject, fxres) + "</title>\n" +" <date>" + avantGoGetArticleData("/sites/xxx", "/story/currency/nzd_fundamentals", 1, "time", cmsObject, fxres) + "</date>\n" +" <author>" + avantGoGetArticleData("/sites/xxx", "/story/currency/nzd_fundamentals", 1, "author", cmsObject, fxres) + "</author>\n" +" <content>" + avantGoGetArticleData("/sites/xxx", "/story/currency/nzd_fundamentals", 1, "fullHTML", cmsObject, fxres) + "</content>\n" +" </article>\n" +" <article>\n" +" <id>" + avantGoGetArticleData("/sites/xxx", "/story/currency/chf_fundamentals", 0, "fileId", cmsObject, fxres) + "</id>\n" +" <pageid>10</pageid>\n" +" <title>" + avantGoGetArticleData("/sites/xxx", "/story/currency/chf_fundamentals", 0, "title", cmsObject, fxres) + "</title>\n" +" <date>" + avantGoGetArticleData("/sites/xxx", "/story/currency/chf_fundamentals", 0, "time", cmsObject, fxres) + "</date>\n" +" <author>" + avantGoGetArticleData("/sites/xxx", "/story/currency/chf_fundamentals", 0, "author", cmsObject, fxres) + "</author>\n" +" <content>" + avantGoGetArticleData("/sites/xxx", "/story/currency/chf_fundamentals", 0, "fullHTML", cmsObject, fxres) + "</content>\n" +" </article>\n" +" <article>\n" +" <id>" + avantGoGetArticleData("/sites/xxx", "/story/currency/chf_fundamentals", 1, "fileId", cmsObject, fxres) + "</id>\n" +" <pageid>10</pageid>\n" +" <title>" + avantGoGetArticleData("/sites/xxx", "/story/currency/chf_fundamentals", 1, "title", cmsObject, fxres) + "</title>\n" +" <date>" + avantGoGetArticleData("/sites/xxx", "/story/currency/chf_fundamentals", 1, "time", cmsObject, fxres) + "</date>\n" +" <author>" + avantGoGetArticleData("/sites/xxx", "/story/currency/chf_fundamentals", 1, "author", cmsObject, fxres) + "</author>\n" +" <content>" + avantGoGetArticleData("/sites/xxx", "/story/currency/chf_fundamentals", 1, "fullHTML", cmsObject, fxres) + "</content>\n" +" </article>\n" +"</site>\n"; //http://www.xxx.com/avantgo_xml.php //writeStringAsFileToVFS(cmsObject, sAvantGoDocument, sfSiteRoot + "/" + "avantgo_xml.php"); sAvantGoDocument = sAvantGoDocument.replaceAll(">.", ">."); sAvantGoDocument = sAvantGoDocument.replaceAll("<?xml:namespace prefix = st1 ns = \"urn:schemas-microsoft-com:office:smarttags\" />", ""); sAvantGoDocument = sAvantGoDocument.replaceAll(":", ":"); sAvantGoDocument = sAvantGoDocument.replaceAll("null", ""); sAvantGoDocument = sAvantGoDocument.replaceAll(". .", ""); writeStringAsFileToVFS(cmsObject, sAvantGoDocument, sfSiteRoot + "/" + "feeds/avantgo_xml.xml"); } catch (Exception e) { System.out.println("Exception occured in FXRssResource"); e.printStackTrace(); System.out.println(e.getMessage()); } } public void testingGetFiles() { try { List resultListForAvantGo; resultListForAvantGo = (List) fxres.getFiles("", "/story/currency/chf_fundamentals", 2 ); System.out.println("avantGoGetArticleData - AFTER resultListForAvantGo.size() = " + resultListForAvantGo.size() ); } catch (Exception e) { e.printStackTrace(); //return "Error occured in importCalendar: " + e.getMessage(); } } public String avantGoGetArticleData(String sNameOfSite, String sPathOfDirectoryFiles, int iPositionOfArticleInReference, String sWhatIsRequested, CmsObject cmsObj, FXResource fxres) { String sResult = "--[inital declaration of sResult - SOMETHING SOMEWHERE NEEDS CORRECTION]--"; //AvantGo //objects declarations for AvantGo List resultListForAvantGo;// = fxres.getLatestFiles("/sites/xxx", 5); FXListItem itemForAvantGo; // = (FXListItem) resultList.get(1); String fileResourceForAvantGo = ""; Locale localeForAvantGo; CmsFile fileForAvantGo; int iUniqueIDofFileAvantGo; try { if (sPathOfDirectoryFiles.equals("")){ resultListForAvantGo = fxres.getLatestFiles(sNameOfSite, (iPositionOfArticleInReference+1) ); } else if (sPathOfDirectoryFiles.equals("calendar")){ resultListForAvantGo = fxres.getLatestFiles("/sites/xxx", 1 ); } else { resultListForAvantGo = (List) fxres.getFiles(sNameOfSite, sPathOfDirectoryFiles, (iPositionOfArticleInReference+1) ); if ( resultListForAvantGo == null ){ resultListForAvantGo = (List) fxres.getFiles(sNameOfSite, sPathOfDirectoryFiles, (iPositionOfArticleInReference) ); } } //itemForAvantGo = (FXListItem) resultListForAvantGo.get(0); itemForAvantGo = (FXListItem) resultListForAvantGo.get(iPositionOfArticleInReference); if (sPathOfDirectoryFiles.equals("calendar")){ fileResourceForAvantGo = "/calendar/briefing/calendar_data_ready_for_avant_go.html"; } else if ( !(sPathOfDirectoryFiles.equals("calendar")) ) { fileResourceForAvantGo = itemForAvantGo.getFile(); //"/story/charting_center/weekly_chart_analysis/5085_eur_gbp_large_triangle_setup.html"; } //itemForAvantGo = (FXListItem) resultListForAvantGo.get(iPositionOfArticleInReference); //fileResourceForAvantGo = itemForAvantGo.getFile(); //"/story/charting_center/weekly_chart_analysis/5085_eur_gbp_large_triangle_setup.html"; //out.println("fileResourceForAvantGo (from /sites/xxx/story/currency/eur_fundamentals) = " + fileResourceForAvantGo + "<HR>"); 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 <Content>, should just be text1 & text2) CmsXmlContent fileContent = CmsXmlContentFactory.unmarshal(cmsObj, fileForAvantGo, false); List fileList = fileContent.getValues("element", localeForAvantGo); if (fileList.size() > 0) { Iterator iter = fileList.iterator(); while (iter.hasNext()) { sArticleXMLElementText1HTMLForAvantGo = ((CmsXmlHtmlValue)iter.next()).getPlainText(cmsObj); if ( !(sPathOfDirectoryFiles.equals("calendar")) ) { sArticleXMLElementText2HTMLForAvantGo = ((CmsXmlHtmlValue)iter.next()).getPlainText(cmsObj); } } } } catch (CmsException e) {} String sAllOfTextForAvantGo = sArticleXMLElementText1HTMLForAvantGo + "<br /><br />" + sArticleXMLElementText2HTMLForAvantGo; sAllOfTextForAvantGo = sAllOfTextForAvantGo.replaceAll("&", "&"); sAllOfTextForAvantGo = sAllOfTextForAvantGo.replaceAll("<img (.*)>", "<br />"); sAllOfTextForAvantGo = sAllOfTextForAvantGo.replaceAll("<IMG (.*)>", "<br />"); sAllOfTextForAvantGo = sAllOfTextForAvantGo.replaceAll("href=\"/opencms/opencms/", "href=\"http://www.xxx.com/export/sites/xxx/"); sAllOfTextForAvantGo = sAllOfTextForAvantGo.replaceAll("HREF=\"/opencms/opencms/", "HREF=\"http://www.xxx.com/export/sites/xxx/"); sAllOfTextForAvantGo = sAllOfTextForAvantGo.replaceAll("<?xml:namespace prefix = st1 ns = \"urn:schemas-microsoft-com:office:smarttags\" />", ""); sAllOfTextForAvantGo = sAllOfTextForAvantGo.replaceAll("\n", " "); sAllOfTextForAvantGo = sAllOfTextForAvantGo.replaceAll("<", "<"); sAllOfTextForAvantGo = sAllOfTextForAvantGo.replaceAll(">", ">"); /////////TRYING TO GET INTRO FOR TEXT String sInitialTextForAvantGo = ""; String sDescriptionForRss = ""; //sInitialTextForAvantGo = sAllOfTextForAvantGo.substring(0, (sAllOfTextForAvantGo.indexOf(">", 350)+3) ) + "..."; //will return the substring of first 0 - x charcters - where x is the first ">" (closing HTML tag) charcter after xth character if(sAllOfTextForAvantGo != null && sAllOfTextForAvantGo.indexOf(" ", 150) != -1){ sInitialTextForAvantGo = sAllOfTextForAvantGo.substring(0, sAllOfTextForAvantGo.indexOf(" ", 150) ) + "..."; } else { sInitialTextForAvantGo = " - please see article - "; } try { if(sAllOfTextForAvantGo != null && sAllOfTextForAvantGo.indexOf(" ", 150) != -1){ sInitialTextForAvantGo = sAllOfTextForAvantGo.substring(0, sAllOfTextForAvantGo.indexOf(" ", 150) ) + "..."; } else if (sAllOfTextForAvantGo != null){ sInitialTextForAvantGo = sAllOfTextForAvantGo; } 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(sAllOfTextForAvantGo != null && sAllOfTextForAvantGo.indexOf(" ", 150) != -1){ sInitialTextForAvantGo = sAllOfTextForAvantGo.substring(0, sAllOfTextForAvantGo.indexOf(" ", 150) ) + "..."; } else{ sInitialTextForAvantGo = "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 ){ System.out.println("Exception occurred in Avantgo section"); // comment and for debuging: System.outx.println("StringIndexOutOfBoundsException occured "); }//// //sInitialTextForAvantGo = sAllOfTextForAvantGo.substring(0, (sAllOfTextForAvantGo.indexOf(">", 350)+3) ) + "..."; //will return the substring of first 0 - x charcters - where x is the first ">" (closing HTML tag) charcter after xth character String sEntireHTMLTextForAvantGo = sAllOfTextForAvantGo; if(sWhatIsRequested.equals("title")) { return itemForAvantGo.getTitle(); } else if (sWhatIsRequested.equals("time")) { return itemForAvantGo.getDate(); } else if (sWhatIsRequested.equals("author")) { return itemForAvantGo.getAuthor(); } else if (sWhatIsRequested.equals("fileId")) { return String.valueOf(iUniqueIDofFileAvantGo); // (string) iUniqueIDofFileAvantGo; } else if (sWhatIsRequested.equals("introHTML")) { return sInitialTextForAvantGo; } else if (sWhatIsRequested.equals("fullHTML")) { return sEntireHTMLTextForAvantGo; } else if (sWhatIsRequested.equals("fullHTML")) { return sEntireHTMLTextForAvantGo; } else if (sPathOfDirectoryFiles.equals("calendar")) { //else if (sWhatIsRequested.equals("firstElementOnly")) { return sAllOfTextForAvantGo; //this will be only the first element of the item } else { return " - please see the article - "; } } catch (Exception e) { //if ( resultListForAvantGo == null ){ System.out.println("resultListForAvantGo == null"); //} e.printStackTrace(); //return "Error occured in importCalendar: " + e.getMessage(); } return sResult;} public boolean existsResource(CmsObject cms, String res) { try { cms.readFile(res); } catch (Exception e) { return false; } return true; } //delete delete delete public String readFileContent(String filename, CmsObject m_cms) { try { //Comment/optional debug info: System.out.println(m_cms.readFile(filename)); //Comment/optional debug info: System.out.println(new String(m_cms.readFile(filename).getContents())); return new String(m_cms.readFile(filename).getContents()); } catch(Exception exc) { System.out.println("exception occured in readFileContent for file: " + filename ); return "problem occured during read, filename " + filename; } //return "problem occured during read, filename " + filename; } public String writeStringAsFileToVFS(CmsObject cmsObject, String sDataRssDataPageToWrite, String sNameOfPathAndFileToWriteToOnVFS) { String sResult = ""; try { if (existsResource(cmsObject,sNameOfPathAndFileToWriteToOnVFS)) { //Comment/optional debug info: System.out.println("resource [" + rss_path_+xml_name+ "] exists."); cmsObject.lockResource(sNameOfPathAndFileToWriteToOnVFS); //cms_.replaceResource(rss_path_+xml_name,"plain",null,xml_content.getBytes()); cmsObject.replaceResource(sNameOfPathAndFileToWriteToOnVFS, 1, sDataRssDataPageToWrite.getBytes(), null); cmsObject.unlockResource(sNameOfPathAndFileToWriteToOnVFS); cmsObject.publishResource(sNameOfPathAndFileToWriteToOnVFS, true, new CmsShellReport( cmsObject.getRequestContext().getLocale()) ); //Comment/optional debug info: System.out.println("RSS file updated: "+rss_path_+xml_name); return "Calendar generation is done, RSS file updated"; } else { //Comment/optional debug info: System.out.println("resource [" + rss_path_+xml_name+ "] does NOT exist."); //cms_.createResource(rss_path_,xml_name,"plain",null,xml_content.getBytes()); cmsObject.createResource(sNameOfPathAndFileToWriteToOnVFS, 1, sDataRssDataPageToWrite.getBytes(), null); cmsObject.lockResource(sNameOfPathAndFileToWriteToOnVFS); cmsObject.unlockResource(sNameOfPathAndFileToWriteToOnVFS); cmsObject.publishResource(sNameOfPathAndFileToWriteToOnVFS, true, new CmsShellReport( cmsObject.getRequestContext().getLocale()) ); //Comment/optional debug info: System.out.println("RSS file created: "+rss_path_+xml_name); return "Calendar generation is done, RSS file created"; } } catch(Exception exc) { exc.printStackTrace(); System.out.println("exception occured in writeStringAsFileToVFS for file: " + sNameOfPathAndFileToWriteToOnVFS + " with data " + sDataRssDataPageToWrite ); return "exception occured in writeStringAsFileToVFS for file: " + sNameOfPathAndFileToWriteToOnVFS + " with data " + sDataRssDataPageToWrite; } }}/** * Class FXCalendarResourceRSSThread begins a thread to pull data from DB every x minutes * @version 1.0 */class FXRssResourceThread extends Thread { public void run() { try { FXRssResource fxresRss1 = new FXRssResource("Admin", "xxxx"); fxresRss1.generateRssAll(); } catch (Exception e) { System.out.println("Run terminated by exception"); e.printStackTrace(); } }}