List of usage examples for org.jdom2 Attribute getValue
public String getValue()
Attribute. From source file:de.nava.informa.parsers.OPMLParser.java
License:Open Source License
private static synchronized Collection<FeedIF> parse(Document doc) throws ParseException { logger.debug("start parsing."); // Get the root element (must be opml) Element root = doc.getRootElement(); String rootElement = root.getName().toLowerCase(); String opmlVersion;//from w w w . jav a 2 s.c o m // Decide which parser to use if (rootElement.startsWith("opml")) { Attribute attrOpmlVersion = root.getAttribute("version"); // there is no version information set if (attrOpmlVersion == null) { // there seems to be no opml version set, so we'll try to parse it with 1.1.. // TODO is it worth to implement also a opml 1.0 parser? are there markable differences between opml 1.1 and 1.0? logger.info( "Collection uses OPML root element (no version information available), trying to parse with 1.1 anyway."); return OPML_1_1_Parser.parse(root); } else { // version information seems to be set, so go get it opmlVersion = attrOpmlVersion.getValue(); } // version information is set if (opmlVersion.contains("1.1")) { // OPML 1.1 version information is set, so we'll parse it with 1.1! logger.info("Collection uses OPML root element (Version 1.1)."); return OPML_1_1_Parser.parse(root); } else if (opmlVersion.contains("1.0")) { // TODO is it worth to implement also a opml 1.0 parser? are there markable differences between opml 1.1 and 1.0? // OPML 1.0 version information is set, we'll try to parse it using 1.1 anyway.. logger.info("Collection uses OPML root element (Version 1.0), trying to parse with 1.1 anyway."); return OPML_1_1_Parser.parse(root); } else { // since it is neither a 1.0 nor 1.1 opml feed we maybe cannot handle it, so it's better to throw some exception.. throw new UnsupportedFormatException( MessageFormat.format("Unsupported OPML version information [{0}].", opmlVersion)); } } // did not match anything throw new UnsupportedFormatException( MessageFormat.format("Unsupported OPML root element [{0}].", rootElement)); }
From source file:de.nava.informa.parsers.OPML_1_1_Parser.java
License:Open Source License
static Collection<FeedIF> parse(Element root) { Collection<FeedIF> feedColl = new ArrayList<>(); Date dateParsed = new Date(); logger.debug("start parsing."); // Lower the case of these tags to simulate case-insensitive parsing ParserUtils.matchCaseOfChildren(root, "body"); // Get the head element (only one should occur) // Element headElem = root.getChild("head"); // String title = headElem.getChildTextTrim("title"); // Get the body element (only one occurs) Element bodyElem = root.getChild("body"); // 1..n outline elements ParserUtils.matchCaseOfChildren(bodyElem, "outline"); List feeds = bodyElem.getChildren("outline"); for (Object feed1 : feeds) { Element feedElem = (Element) feed1; // get title attribute Attribute attrTitle = feedElem.getAttribute("title"); String strTitle = "[No Title]"; if (attrTitle != null) { strTitle = attrTitle.getValue(); }// w w w .ja v a 2 s. c o m FeedIF feed = new Feed(strTitle); if (logger.isDebugEnabled()) { logger.debug("Feed element found (" + strTitle + ")."); } // get text attribute Attribute attrText = feedElem.getAttribute("text"); String strText = "[No Text]"; if (attrText != null) { strText = attrText.getValue(); } feed.setText(strText); // get attribute type (for example: 'rss') Attribute attrType = feedElem.getAttribute("type"); String strType = "text/xml"; if (attrType != null) { strType = attrType.getValue(); } feed.setContentType(strType); // TODO: handle attribute version (for example: 'RSS') // get attribute xmlUrl Attribute attrXmlUrl = feedElem.getAttribute("xmlUrl"); if (attrXmlUrl != null && attrXmlUrl.getValue() != null) { feed.setLocation(ParserUtils.getURL(attrXmlUrl.getValue())); } // get attribute htmllUrl Attribute attrHtmlUrl = feedElem.getAttribute("htmlUrl"); if (attrHtmlUrl != null && attrHtmlUrl.getValue() != null) { feed.setSite(ParserUtils.getURL(attrHtmlUrl.getValue())); } // set current date feed.setDateFound(dateParsed); // add feed to collection feedColl.add(feed); } return feedColl; }
From source file:de.nava.informa.parsers.RSS_0_91_Parser.java
License:Open Source License
/** * @see de.nava.informa.core.ChannelParserIF#parse(de.nava.informa.core.ChannelBuilderIF, org.jdom2.Element) *//*from w w w . j a v a 2 s. c om*/ public ChannelIF parse(ChannelBuilderIF cBuilder, Element root) throws ParseException { if (cBuilder == null) { throw new RuntimeException("Without builder no channel can " + "be created."); } Date dateParsed = new Date(); logger.debug("start parsing."); // Get the channel element (only one occurs) ParserUtils.matchCaseOfChildren(root, "channel"); Element channel = root.getChild("channel"); if (channel == null) { logger.warn("Channel element could not be retrieved from feed."); throw new ParseException("No channel element found in feed."); } // --- read in channel information ParserUtils.matchCaseOfChildren(channel, new String[] { "title", "description", "link", "language", "item", "image", "textinput", "copyright", "rating", "pubDate", "lastBuildDate", "docs", "managingEditor", "webMaster", "cloud" }); // 1 title element ChannelIF chnl = cBuilder.createChannel(channel, channel.getChildTextTrim("title")); chnl.setFormat(ChannelFormat.RSS_0_91); // 1 description element chnl.setDescription(channel.getChildTextTrim("description")); // 1 link element chnl.setSite(ParserUtils.getURL(channel.getChildTextTrim("link"))); // 1 language element chnl.setLanguage(channel.getChildTextTrim("language")); // 1..n item elements List items = channel.getChildren("item"); Iterator i = items.iterator(); while (i.hasNext()) { Element item = (Element) i.next(); ParserUtils.matchCaseOfChildren(item, new String[] { "title", "link", "description", "source", "enclosure" }); // get title element Element elTitle = item.getChild("title"); String strTitle = "<No Title>"; if (elTitle != null) { strTitle = elTitle.getTextTrim(); } if (logger.isDebugEnabled()) { logger.debug("Item element found (" + strTitle + ")."); } // get link element Element elLink = item.getChild("link"); String strLink = ""; if (elLink != null) { strLink = elLink.getTextTrim(); } // get description element Element elDesc = item.getChild("description"); String strDesc = ""; if (elDesc != null) { strDesc = elDesc.getTextTrim(); } // generate new RSS item (link to article) ItemIF rssItem = cBuilder.createItem(item, chnl, strTitle, strDesc, ParserUtils.getURL(strLink)); rssItem.setFound(dateParsed); // get source element (an RSS 0.92 element) Element source = item.getChild("source"); if (source != null) { String sourceName = source.getTextTrim(); Attribute sourceAttribute = source.getAttribute("url"); if (sourceAttribute != null) { String location = sourceAttribute.getValue().trim(); ItemSourceIF itemSource = cBuilder.createItemSource(rssItem, sourceName, location, null); rssItem.setSource(itemSource); } } // get enclosure element (an RSS 0.92 element) Element enclosure = item.getChild("enclosure"); if (enclosure != null) { URL location = null; String type = null; int length = -1; Attribute urlAttribute = enclosure.getAttribute("url"); if (urlAttribute != null) { location = ParserUtils.getURL(urlAttribute.getValue().trim()); } Attribute typeAttribute = enclosure.getAttribute("type"); if (typeAttribute != null) { type = typeAttribute.getValue().trim(); } Attribute lengthAttribute = enclosure.getAttribute("length"); if (lengthAttribute != null) { try { length = Integer.parseInt(lengthAttribute.getValue().trim()); } catch (NumberFormatException e) { logger.warn(e); } } ItemEnclosureIF itemEnclosure = cBuilder.createItemEnclosure(rssItem, location, type, length); rssItem.setEnclosure(itemEnclosure); } } // 0..1 image element Element image = channel.getChild("image"); if (image != null) { ParserUtils.matchCaseOfChildren(image, new String[] { "title", "url", "link", "width", "height", "description" }); ImageIF rssImage = cBuilder.createImage(image.getChildTextTrim("title"), ParserUtils.getURL(image.getChildTextTrim("url")), ParserUtils.getURL(image.getChildTextTrim("link"))); Element imgWidth = image.getChild("width"); if (imgWidth != null) { try { rssImage.setWidth(Integer.parseInt(imgWidth.getTextTrim())); } catch (NumberFormatException e) { logger.warn(e); } } Element imgHeight = image.getChild("height"); if (imgHeight != null) { try { rssImage.setHeight(Integer.parseInt(imgHeight.getTextTrim())); } catch (NumberFormatException e) { logger.warn(e); } } Element imgDescr = image.getChild("description"); if (imgDescr != null) { rssImage.setDescription(imgDescr.getTextTrim()); } chnl.setImage(rssImage); } // 0..1 textinput element Element txtinp = channel.getChild("textinput"); if (txtinp != null) { ParserUtils.matchCaseOfChildren(txtinp, new String[] { "title", "description", "name", "link" }); TextInputIF rssTextInput = cBuilder.createTextInput(txtinp.getChild("title").getTextTrim(), txtinp.getChild("description").getTextTrim(), txtinp.getChild("name").getTextTrim(), ParserUtils.getURL(txtinp.getChild("link").getTextTrim())); chnl.setTextInput(rssTextInput); } // 0..1 copyright element Element copyright = channel.getChild("copyright"); if (copyright != null) { chnl.setCopyright(copyright.getTextTrim()); } // 0..1 rating element Element rating = channel.getChild("rating"); if (rating != null) { chnl.setRating(rating.getTextTrim()); } // 0..1 pubDate element Element pubDate = channel.getChild("pubDate"); if (pubDate != null) { chnl.setPubDate(ParserUtils.getDate(pubDate.getTextTrim())); } // 0..1 lastBuildDate element Element lastBuildDate = channel.getChild("lastBuildDate"); if (lastBuildDate != null) { chnl.setLastBuildDate(ParserUtils.getDate(lastBuildDate.getTextTrim())); } // 0..1 docs element Element docs = channel.getChild("docs"); if (docs != null) { chnl.setDocs(docs.getTextTrim()); } // 0..1 managingEditor element Element managingEditor = channel.getChild("managingEditor"); if (managingEditor != null) { chnl.setCreator(managingEditor.getTextTrim()); } // 0..1 webMaster element Element webMaster = channel.getChild("webMaster"); if (webMaster != null) { chnl.setPublisher(webMaster.getTextTrim()); } // 0..1 cloud element Element cloud = channel.getChild("cloud"); if (cloud != null) { String _port = cloud.getAttributeValue("port"); int port = -1; if (_port != null) { try { port = Integer.parseInt(_port); } catch (NumberFormatException e) { logger.warn(e); } } chnl.setCloud( cBuilder.createCloud(cloud.getAttributeValue("domain"), port, cloud.getAttributeValue("path"), cloud.getAttributeValue("registerProcedure"), cloud.getAttributeValue("protocol"))); } chnl.setLastUpdated(dateParsed); // 0..1 skipHours element // 0..1 skipDays element return chnl; }
From source file:de.nava.informa.parsers.RSS_1_0_Parser.java
License:Open Source License
public ChannelIF parse(ChannelBuilderIF cBuilder, Element root) throws ParseException { if (cBuilder == null) { throw new RuntimeException("Without builder no channel can " + "be created."); }// w w w . j a va 2s . c o m Date dateParsed = new Date(); Namespace defNS = ParserUtils.getDefaultNS(root); if (defNS == null) { defNS = Namespace.NO_NAMESPACE; logger.info("No default namespace found."); } // RSS 1.0 Dublin Core Module namespace Namespace dcNS = ParserUtils.getNamespace(root, "dc"); // fall back to default name space (for retrieving descriptions) if (dcNS == null) { dcNS = defNS; } // RSS 1.0 Syndication Module namespace Namespace syNS = ParserUtils.getNamespace(root, "sy"); // RSS 1.0 Aggregation Module namespace Namespace agNS = ParserUtils.getNamespace(root, "ag"); // RSS 1.0 Administration Module namespace Namespace adminNS = ParserUtils.getNamespace(root, "admin"); // RSS 1.0 DCTerms Module namespace Namespace dctermsNS = ParserUtils.getNamespace(root, "dcterms"); // RSS 1.0 Annotation Module namespace Namespace annotateNS = ParserUtils.getNamespace(root, "annotate"); // RSS091 Module namespace Namespace rss091NS = ParserUtils.getNamespace(root, "rss091"); // Content namespace Namespace contentNS = ParserUtils.getNamespace(root, "content"); ParserUtils.matchCaseOfChildren(root, new String[] { "channel", "item", "image", "textinput" }); // Get the channel element (only one occurs) Element channel = root.getChild("channel", defNS); if (channel == null) { logger.warn("Channel element could not be retrieved from feed."); throw new ParseException("No channel element found in feed."); } // ----------------------- read in channel information ParserUtils.matchCaseOfChildren(channel, new String[] { "title", "description", "link", "creator", "managingEditor", "publisher", "errorReportsTo", "webMaster", "language", "rights", "copyright", "rating", "date", "issued", "pubdate", "lastBuildDate", "modified", "generatorAgent", "updatePeriod", "updateFrequency", "updateBase" }); // title element ChannelIF chnl = cBuilder.createChannel(channel, channel.getChildTextTrim("title", defNS)); // set channel format chnl.setFormat(ChannelFormat.RSS_1_0); // description element chnl.setDescription(channel.getChildTextTrim("description", defNS)); // link element chnl.setSite(ParserUtils.getURL(channel.getChildTextTrim("link", defNS))); // creator element Element creator = channel.getChild("creator", dcNS); if (creator == null) { creator = channel.getChild("managingEditor", rss091NS); } if (creator != null) { chnl.setCreator(creator.getTextTrim()); } // publisher element String publisher = channel.getChildTextTrim("publisher", dcNS); if (publisher == null) { Element elErrorReportsTo = channel.getChild("errorReportsTo", adminNS); if (elErrorReportsTo != null) { publisher = elErrorReportsTo.getAttributeValue("resource", ParserUtils.getNamespace(elErrorReportsTo, "rdf")); } } if (publisher == null) { publisher = channel.getChildTextTrim("webMaster", rss091NS); } chnl.setPublisher(publisher); // language element Element language = channel.getChild("language", dcNS); if (language == null) { language = channel.getChild("language", rss091NS); } if (language != null) { chnl.setLanguage(language.getTextTrim()); } // rights element Element copyright = channel.getChild("rights", dcNS); if (copyright == null) { copyright = channel.getChild("copyright", rss091NS); } if (copyright != null) { chnl.setCopyright(copyright.getTextTrim()); } // 0..1 Rating element Element rating = channel.getChild("rating", rss091NS); if (rating != null) { chnl.setRating(rating.getTextTrim()); } // 0..1 Docs element // use namespace URI chnl.setDocs(defNS.getURI()); // 0..1 pubDate element Element pubDate = channel.getChild("date", dcNS); if (pubDate == null) { pubDate = channel.getChild("issued", dctermsNS); } if (pubDate == null) { pubDate = channel.getChild("pubdate", rss091NS); } if (pubDate != null) { chnl.setPubDate(ParserUtils.getDate(pubDate.getTextTrim())); } // 0..1 lastBuildDate element Element lastBuildDate = channel.getChild("lastBuildDate"); if (lastBuildDate == null) { lastBuildDate = channel.getChild("modified", dctermsNS); } if (lastBuildDate == null) { lastBuildDate = channel.getChild("lastBuildDate", rss091NS); } if (lastBuildDate != null) { chnl.setLastBuildDate(ParserUtils.getDate(lastBuildDate.getTextTrim())); } // RSS 1.0 Administration Module support // 0..1 generator element Element elGenerator = channel.getChild("generatorAgent", adminNS); if (elGenerator != null) { Attribute generator = elGenerator.getAttribute("resource", ParserUtils.getNamespace(elGenerator, "rdf")); if (generator != null) { chnl.setGenerator(generator.getValue()); } } // RSS 1.0 Syndication Module support // 0..1 update period element Element updatePeriod = channel.getChild("updatePeriod", syNS); if (updatePeriod != null) { try { ChannelUpdatePeriod channelUpdatePeriod = ChannelUpdatePeriod .valueFromText(updatePeriod.getTextTrim()); chnl.setUpdatePeriod(channelUpdatePeriod); } catch (IllegalArgumentException ex) { logger.warn(updatePeriod.getTextTrim(), ex); } } // 0..1 update frequency element Element updateFrequency = channel.getChild("updateFrequency", syNS); if (updateFrequency != null) { chnl.setUpdateFrequency((new Integer(updateFrequency.getTextTrim())).intValue()); } // 0..1 update base element Element updateBase = channel.getChild("updateBase", syNS); if (updateBase != null) { chnl.setUpdateBase(ParserUtils.getDate(updateBase.getTextTrim())); } if ((updatePeriod != null) && updateFrequency != null) { int ttl = getTTL(chnl.getUpdatePeriod(), chnl.getUpdateFrequency()); chnl.setTtl(ttl); } // item elements List items = root.getChildren("item", defNS); Iterator i = items.iterator(); while (i.hasNext()) { Element item = (Element) i.next(); ParserUtils.matchCaseOfChildren(item, new String[] { "title", "link", "encoded", "description", "creator", "subject", "date", "sourceURL", "source", "timestamp", "reference" }); // get title element Element elTitle = item.getChild("title", defNS); String strTitle = "<No Title>"; if (elTitle != null) { strTitle = elTitle.getTextTrim(); } if (logger.isDebugEnabled()) { logger.debug("Item element found (" + strTitle + ")."); } // get link element Element elLink = item.getChild("link", defNS); String strLink = ""; if (elLink != null) { strLink = elLink.getTextTrim(); } // get description element Element elDesc = item.getChild("encoded", contentNS); if (elDesc == null) { elDesc = item.getChild("description", defNS); } if (elDesc == null) { elDesc = item.getChild("description", dcNS); } String strDesc = ""; if (elDesc != null) { strDesc = elDesc.getTextTrim(); } // generate new RSS item (link to article) ItemIF rssItem = cBuilder.createItem(item, chnl, strTitle, strDesc, ParserUtils.getURL(strLink)); rssItem.setFound(dateParsed); // get creator element Element elCreator = item.getChild("creator", dcNS); if (elCreator != null) { rssItem.setCreator(elCreator.getTextTrim()); } // get subject element Element elSubject = item.getChild("subject", dcNS); if (elSubject != null) { // TODO: Mulitple subject elements not handled currently rssItem.setSubject(elSubject.getTextTrim()); } // get date element Element elDate = item.getChild("date", dcNS); if (elDate != null) { rssItem.setDate(ParserUtils.getDate(elDate.getTextTrim())); } // get source element - default to Aggregation module, then try Dublin Core String sourceName = null; String sourceLocation = null; Date sourceTimestamp = null; Element elSourceURL = item.getChild("sourceURL", agNS); if (elSourceURL == null) { // No Aggregation module - try Dublin Core elSourceURL = item.getChild("source", dcNS); if (elSourceURL != null) { sourceLocation = elSourceURL.getTextTrim(); sourceName = "Source"; } } else { // Aggregation module sourceLocation = elSourceURL.getTextTrim(); Element elSourceName = item.getChild("source", agNS); if (elSourceName != null) { sourceName = elSourceName.getTextTrim(); } Element elSourceTimestamp = item.getChild("timestamp", agNS); if (elSourceTimestamp != null) { sourceTimestamp = ParserUtils.getDate(elSourceTimestamp.getTextTrim()); } } if (sourceLocation != null) { ItemSourceIF itemSource = cBuilder.createItemSource(rssItem, sourceName, sourceLocation, sourceTimestamp); rssItem.setSource(itemSource); } // comments element - use Annotation module Element elReference = item.getChild("reference", annotateNS); if (elReference != null) { Attribute resource = elReference.getAttribute("resource", ParserUtils.getNamespace(elReference, "rdf")); if (resource != null) { URL resourceURL = ParserUtils.getURL(resource.getValue()); if (resourceURL != null) { rssItem.setComments(resourceURL); } } } } // image element Element image = root.getChild("image", defNS); if (image != null) { ParserUtils.matchCaseOfChildren(image, new String[] { "title", "url", "link", "width", "height", "description" }); ImageIF rssImage = cBuilder.createImage(image.getChildTextTrim("title", defNS), ParserUtils.getURL(image.getChildTextTrim("url", defNS)), ParserUtils.getURL(image.getChildTextTrim("link", defNS))); Element imgWidth = image.getChild("width", defNS); if (imgWidth != null) { try { rssImage.setWidth(Integer.parseInt(imgWidth.getTextTrim())); } catch (NumberFormatException e) { logger.warn(e); } } Element imgHeight = image.getChild("height", defNS); if (imgHeight != null) { try { rssImage.setHeight(Integer.parseInt(imgHeight.getTextTrim())); } catch (NumberFormatException e) { logger.warn(e); } } Element imgDescr = image.getChild("description", defNS); if (imgDescr != null) { rssImage.setDescription(imgDescr.getTextTrim()); } chnl.setImage(rssImage); } // textinput element Element txtinp = root.getChild("textinput", defNS); if (txtinp != null) { ParserUtils.matchCaseOfChildren(image, new String[] { "title", "description", "name", "link" }); String tiTitle = null; if (txtinp.getChild("title", defNS) != null) { tiTitle = txtinp.getChild("title", defNS).getTextTrim(); } String tiDescr = null; if (txtinp.getChild("description", defNS) != null) { tiDescr = txtinp.getChild("description", defNS).getTextTrim(); } String tiName = null; if (txtinp.getChild("name", defNS) != null) { tiName = txtinp.getChild("name", defNS).getTextTrim(); } URL tiLink = null; if (txtinp.getChild("link", defNS) != null) { tiLink = ParserUtils.getURL(txtinp.getChild("link", defNS).getTextTrim()); } TextInputIF rssTextInput = cBuilder.createTextInput(tiTitle, tiDescr, tiName, tiLink); chnl.setTextInput(rssTextInput); } chnl.setLastUpdated(dateParsed); return chnl; }
From source file:de.nava.informa.parsers.RSS_2_0_Parser.java
License:Open Source License
/** * @see de.nava.informa.core.ChannelParserIF#parse(de.nava.informa.core.ChannelBuilderIF, org.jdom2.Element) *//*from ww w. j a va2 s . com*/ public ChannelIF parse(ChannelBuilderIF cBuilder, Element root) throws ParseException { if (cBuilder == null) { throw new RuntimeException("Without builder no channel can be created."); } Date dateParsed = new Date(); logger.debug("start parsing."); Namespace defNS = ParserUtils.getDefaultNS(root); if (defNS == null) { defNS = Namespace.NO_NAMESPACE; logger.info("No default namespace found."); } Namespace dcNS = ParserUtils.getNamespace(root, "dc"); // fall back to default name space if (dcNS == null) { dcNS = defNS; } // Content namespace Namespace contentNS = ParserUtils.getNamespace(root, "content"); // fall back to default name space if (contentNS == null) { contentNS = defNS; } ParserUtils.matchCaseOfChildren(root, "channel"); // Get the channel element (only one occurs) Element channel = root.getChild("channel", defNS); if (channel == null) { logger.warn("Channel element could not be retrieved from feed."); throw new ParseException("No channel element found in feed."); } // --- read in channel information ParserUtils.matchCaseOfChildren(channel, new String[] { "title", "description", "link", "language", "item", "image", "textinput", "copyright", "rating", "docs", "generator", "pubDate", "lastBuildDate", "category", "managingEditor", "webMaster", "cloud" }); // 1 title element ChannelIF chnl = cBuilder.createChannel(channel, channel.getChildTextTrim("title", defNS)); // set channel format chnl.setFormat(ChannelFormat.RSS_2_0); // 1 description element chnl.setDescription(channel.getChildTextTrim("description", defNS)); // 1 link element chnl.setSite(ParserUtils.getURL(channel.getChildTextTrim("link", defNS))); // 1 language element chnl.setLanguage(channel.getChildTextTrim("language", defNS)); // 1..n item elements List items = channel.getChildren("item", defNS); for (Object item1 : items) { Element item = (Element) item1; ParserUtils.matchCaseOfChildren(item, new String[] { "title", "link", "encoded", "description", "subject", "category", "pubDate", "date", "author", "creator", "comments", "guid", "source", "enclosure" }); // get title element Element elTitle = item.getChild("title", defNS); String strTitle = "<No Title>"; if (elTitle != null) { strTitle = elTitle.getTextTrim(); } if (logger.isDebugEnabled()) { logger.debug("Item element found (" + strTitle + ")."); } // get link element Element elLink = item.getChild("link", defNS); String strLink = ""; if (elLink != null) { strLink = elLink.getTextTrim(); } // get description element Element elDesc = item.getChild("encoded", contentNS); if (elDesc == null) { elDesc = item.getChild("description", defNS); } String strDesc = ""; if (elDesc != null) { strDesc = elDesc.getTextTrim(); } // generate new RSS item (link to article) ItemIF rssItem = cBuilder.createItem(item, chnl, strTitle, strDesc, ParserUtils.getURL(strLink)); // get subject element Element elSubject = item.getChild("subject", defNS); if (elSubject == null) { // fallback mechanism: get dc:subject element elSubject = item.getChild("subject", dcNS); } if (elSubject != null) { rssItem.setSubject(elSubject.getTextTrim()); } // get category list // get list of <category> elements List listCategory = item.getChildren("category", defNS); if (listCategory.size() < 1) { // fallback mechanism: get dc:category element listCategory = item.getChildren("category", dcNS); } if (listCategory.size() > 0) { RecursiveHashtable<String> catTable = new RecursiveHashtable<String>(); // for each category, parse hierarchy for (Object aListCategory : listCategory) { RecursiveHashtable<String> currTable = catTable; Element elCategory = (Element) aListCategory; // get contents of category element String[] titles = elCategory.getTextNormalize().split("/"); for (String title : titles) { // tokenize category string to extract out hierarchy if (!currTable.containsKey(title)) { // if token does not exist in current map, add it with child Hashtable currTable.put(title, new RecursiveHashtable<String>()); } // reset current Hashtable to child's Hashtable then iterate to next token currTable = currTable.get(title); } } ArrayList<CategoryIF> catList = new ArrayList<CategoryIF>(); // transform cat list & hierarchy into list of CategoryIF elements Enumeration<String> enumCategories = catTable.keys(); while (enumCategories.hasMoreElements()) { String key = enumCategories.nextElement(); // build category list: getCategoryList(parent, title, children) CategoryIF cat = getCategoryList(null, key, catTable.get(key)); catList.add(cat); } if (catList.size() > 0) { // if categories were actually created, then add list to item node rssItem.setCategories(catList); } } // get publication date Element elDate = item.getChild("pubDate", defNS); if (elDate == null) { // fallback mechanism: get dc:date element elDate = item.getChild("date", dcNS); } if (elDate != null) { rssItem.setDate(ParserUtils.getDate(elDate.getTextTrim())); } rssItem.setFound(dateParsed); // get Author element Element elAuthor = item.getChild("author", defNS); if (elAuthor == null) { // fallback mechanism: get dc:creator element elAuthor = item.getChild("creator", dcNS); } if (elAuthor != null) rssItem.setCreator(elAuthor.getTextTrim()); // get Comments element Element elComments = item.getChild("comments", defNS); String strComments = ""; if (elComments != null) { strComments = elComments.getTextTrim(); } rssItem.setComments(ParserUtils.getURL(strComments)); // get guid element Element elGuid = item.getChild("guid", defNS); if (elGuid != null) { String guidUrl = elGuid.getTextTrim(); if (guidUrl != null) { boolean permaLink = true; Attribute permaLinkAttribute = elGuid.getAttribute("isPermaLink", defNS); if (permaLinkAttribute != null) { String permaLinkStr = permaLinkAttribute.getValue(); if (permaLinkStr != null) { permaLink = Boolean.valueOf(permaLinkStr); } } ItemGuidIF itemGuid = cBuilder.createItemGuid(rssItem, guidUrl, permaLink); rssItem.setGuid(itemGuid); } } // get source element Element elSource = item.getChild("source", defNS); if (elSource != null) { String sourceName = elSource.getTextTrim(); Attribute sourceAttribute = elSource.getAttribute("url", defNS); if (sourceAttribute != null) { String sourceLocation = sourceAttribute.getValue().trim(); ItemSourceIF itemSource = cBuilder.createItemSource(rssItem, sourceName, sourceLocation, null); rssItem.setSource(itemSource); } } // get enclosure element Element elEnclosure = item.getChild("enclosure", defNS); if (elEnclosure != null) { URL location = null; String type = null; int length = -1; Attribute urlAttribute = elEnclosure.getAttribute("url", defNS); if (urlAttribute != null) { location = ParserUtils.getURL(urlAttribute.getValue().trim()); } Attribute typeAttribute = elEnclosure.getAttribute("type", defNS); if (typeAttribute != null) { type = typeAttribute.getValue().trim(); } Attribute lengthAttribute = elEnclosure.getAttribute("length", defNS); if (lengthAttribute != null) { try { length = Integer.parseInt(lengthAttribute.getValue().trim()); } catch (NumberFormatException e) { logger.warn(e); } } ItemEnclosureIF itemEnclosure = cBuilder.createItemEnclosure(rssItem, location, type, length); rssItem.setEnclosure(itemEnclosure); } } // 0..1 image element Element image = channel.getChild("image", defNS); if (image != null) { ParserUtils.matchCaseOfChildren(image, new String[] { "title", "url", "link", "width", "height", "description" }); ImageIF rssImage = cBuilder.createImage(image.getChildTextTrim("title", defNS), ParserUtils.getURL(image.getChildTextTrim("url", defNS)), ParserUtils.getURL(image.getChildTextTrim("link", defNS))); Element imgWidth = image.getChild("width", defNS); if (imgWidth != null) { try { rssImage.setWidth(Integer.parseInt(imgWidth.getTextTrim())); } catch (NumberFormatException e) { logger.warn("Error parsing width: " + e.getMessage()); } } Element imgHeight = image.getChild("height", defNS); if (imgHeight != null) { try { rssImage.setHeight(Integer.parseInt(imgHeight.getTextTrim())); } catch (NumberFormatException e) { logger.warn("Error parsing height: " + e.getMessage()); } } Element imgDescr = image.getChild("description", defNS); if (imgDescr != null) { rssImage.setDescription(imgDescr.getTextTrim()); } chnl.setImage(rssImage); } // 0..1 textinput element Element txtinp = channel.getChild("textinput", defNS); if (txtinp != null) { ParserUtils.matchCaseOfChildren(txtinp, new String[] { "title", "description", "name", "link" }); TextInputIF rssTextInput = cBuilder.createTextInput(txtinp.getChildTextTrim("title", defNS), txtinp.getChildTextTrim("description", defNS), txtinp.getChildTextTrim("name", defNS), ParserUtils.getURL(txtinp.getChildTextTrim("link", defNS))); chnl.setTextInput(rssTextInput); } // 0..1 copyright element Element copyright = channel.getChild("copyright", defNS); if (copyright != null) { chnl.setCopyright(copyright.getTextTrim()); } // 0..1 Rating element Element rating = channel.getChild("rating", defNS); if (rating != null) { chnl.setRating(rating.getTextTrim()); } // 0..1 Docs element Element docs = channel.getChild("docs", defNS); if (docs != null) { chnl.setDocs(docs.getTextTrim()); } // 0..1 Generator element Element generator = channel.getChild("generator", defNS); if (generator != null) { chnl.setGenerator(generator.getTextTrim()); } // 0..1 ttl element Element ttl = channel.getChild("ttl", defNS); if (ttl != null) { String ttlValue = ttl.getTextTrim(); try { chnl.setTtl(Integer.parseInt(ttlValue)); } catch (NumberFormatException e) { logger.warn("Invalid TTL format: '" + ttlValue + "'"); } } // 0..1 pubDate element Element pubDate = channel.getChild("pubDate", defNS); if (pubDate != null) { chnl.setPubDate(ParserUtils.getDate(pubDate.getTextTrim())); } // 0..1 lastBuildDate element Element lastBuildDate = channel.getChild("lastBuildDate", defNS); if (lastBuildDate != null) { chnl.setLastBuildDate(ParserUtils.getDate(lastBuildDate.getTextTrim())); } // get category list // get list of <category> elements List listCategory = channel.getChildren("category", defNS); if (listCategory.size() < 1) { // fallback mechanism: get dc:category element listCategory = channel.getChildren("category", dcNS); } if (listCategory.size() > 0) { RecursiveHashtable<String> catTable = new RecursiveHashtable<String>(); // for each category, parse hierarchy for (Object aListCategory : listCategory) { RecursiveHashtable<String> currTable = catTable; Element elCategory = (Element) aListCategory; // get contents of category element String[] titles = elCategory.getTextNormalize().split("/"); for (String title : titles) { // tokenize category string to extract out hierarchy if (!currTable.containsKey(title)) { // if token does not exist in current map, add it with child Hashtable currTable.put(title, new RecursiveHashtable<String>()); } // reset current Hashtable to child's Hashtable then iterate to next token currTable = currTable.get(title); } } ArrayList<CategoryIF> catList = new ArrayList<CategoryIF>(); // transform cat list & hierarchy into list of CategoryIF elements Enumeration<String> enumCategories = catTable.keys(); while (enumCategories.hasMoreElements()) { String key = enumCategories.nextElement(); // build category list: getCategoryList(parent, title, children) CategoryIF cat = getCategoryList(null, key, catTable.get(key)); catList.add(cat); } if (catList.size() > 0) { // if categories were actually created, then add list to item node chnl.setCategories(catList); } } // 0..1 managingEditor element Element managingEditor = channel.getChild("managingEditor", defNS); if (managingEditor != null) { chnl.setCreator(managingEditor.getTextTrim()); } // 0..1 webMaster element Element webMaster = channel.getChild("webMaster", defNS); if (webMaster != null) { chnl.setPublisher(webMaster.getTextTrim()); } // 0..1 cloud element Element cloud = channel.getChild("cloud", defNS); if (cloud != null) { String _port = cloud.getAttributeValue("port", defNS); int port = -1; if (_port != null) { try { port = Integer.parseInt(_port); } catch (NumberFormatException e) { logger.warn(e); } } chnl.setCloud(cBuilder.createCloud(cloud.getAttributeValue("domain", defNS), port, cloud.getAttributeValue("path", defNS), cloud.getAttributeValue("registerProcedure", defNS), cloud.getAttributeValue("protocol", defNS))); } chnl.setLastUpdated(dateParsed); // 0..1 skipHours element // 0..1 skipDays element return chnl; }
From source file:de.openVJJ.imagePublisher.XuggleVideoFileInput.java
License:Open Source License
@Override public void setConfig(Element element) { Attribute inputFileName = element.getAttribute("inputFileName"); if (inputFileName != null) { setInputFileName(inputFileName.getValue()); }/*from www . j av a 2s.c o m*/ }
From source file:de.relaunch64.popelganda.database.Settings.java
License:Open Source License
/** * // w w w .j a v a2 s . c o m * @param nr * @return */ public Assembler getRecentDocAssembler(int nr) { // checl for valid parameter if (nr < 0) return Assemblers.ASM_KICKASSEMBLER; // retrieve element Element el = root.getChild(SETTING_RECENT_DOC + String.valueOf(nr)); // if we have any valid document if (el != null) { // retrieve compiler attribute Attribute comp = el.getAttribute(REC_DOC_ASSEMBLER); // if we have any valid attribute if (comp != null) { try { return Assemblers.byID(Integer.parseInt(comp.getValue())); } catch (NumberFormatException ex) { return Assemblers.ASM_KICKASSEMBLER; } } } // else return null return Assemblers.ASM_KICKASSEMBLER; }
From source file:de.relaunch64.popelganda.database.Settings.java
License:Open Source License
/** * /*from w ww.ja v a 2 s. c o m*/ * @param nr * @return */ public int getRecentDocScript(int nr) { // checl for valid parameter if (nr < 0) return 0; // retrieve element Element el = root.getChild(SETTING_RECENT_DOC + String.valueOf(nr)); // if we have any valid document if (el != null) { // retrieve compiler attribute Attribute comp = el.getAttribute(REC_DOC_SCRIPT); // if we have any valid attribute if (comp != null) { try { return Integer.parseInt(comp.getValue()); } catch (NumberFormatException ex) { return 0; } } } // else return null return 0; }
From source file:de.smartics.maven.alias.domain.AliasesProcessor.java
License:Apache License
/** * Applies the alias information from the XML file to the given * {@code builders}.//from w w w .j av a 2 s. co m * * @param builders the builders to create alias scripts. */ public void process(final AliasCollector... builders) { final Element root = doc.getRootElement(); for (final Element extensionElement : root.getChildren("extension", nsAlias)) { final AliasExtension extension = createExtension(extensionElement); final ExtensionGroup extensionGroup = new ExtensionGroup(extension); extensionGroups.add(extensionGroup); } for (final Element groupElement : root.getChildren("group", nsAlias)) { final Attribute groupName = groupElement.getAttribute("name"); final String comment = readComment(groupElement); final AliasGroup group = new AliasGroup(groupName.getValue(), comment); for (final Element aliasElement : groupElement.getChildren("alias", nsAlias)) { final Alias alias = createAlias(aliasElement); for (final ExtensionGroup extensionGroup : extensionGroups) { extensionGroup.addAlias(group.getName(), alias); } group.addAlias(alias); } for (final AliasCollector builder : builders) { builder.addAliases(group); } } for (final AliasCollector builder : builders) { builder.setExtensionGroups(extensionGroups); } }
From source file:de.smartics.maven.alias.domain.AliasesProcessor.java
License:Apache License
private AliasExtension createExtension(final Element extensionElement) { final AliasExtension.Builder builder = new AliasExtension.Builder(); final Attribute env = extensionElement.getAttribute("env"); if (env != null) { builder.withEnv(env.getValue()); }/*from ww w. ja va2s. co m*/ final String name = extensionElement.getChildTextNormalize("name", nsAlias); final String template = extensionElement.getChildTextNormalize("template", nsAlias); final String comment = readComment(extensionElement); final Element commentElement = extensionElement.getChild("comment", nsAlias); if (commentElement != null) { final String mnemonic = commentElement.getAttributeValue("mnemonic"); builder.withMnemonic(mnemonic); } builder.withName(name).withTemplate(template).withComment(comment); appendApplyTos(extensionElement, builder); return builder.build(); }