/**
* A helper function to print out the contents of an entry to log.trace
*
* @param entry
*/
public static void printEntry(SyndEntry entry) {
if (!log.isTraceEnabled()) {
return;
}
StringBuffer pr =
new StringBuffer(
"URI: "
+ entry.getUri()
+ "\n"
+ "Title: "
+ entry.getTitle()
+ "\n"
+ "\n"
+ "Date: "
+ entry.getPublishedDate()
+ "\n"
+ "Modified: "
+ entry.getUpdatedDate()
+ "\n");
pr.append("Creators: \n");
for (SyndPerson author : FeedHelper.getAuthors(entry)) {
pr.append(" - " + author.getName() + "\n");
}
pr.append("Links: \n");
for (SyndLink link : FeedHelper.getLinks(entry)) {
pr.append(" - " + link.getTitle() + ": " + link.getHref() + "\n");
}
SyndContent description = entry.getDescription();
if (description != null) {
pr.append("\nDescription(" + description.getType() + "): " + description.getValue());
}
pr.append("Contents: \n");
for (SyndContent content : FeedHelper.getContents(entry)) {
pr.append(" Type: " + content.getType());
pr.append(" Body: " + content.getValue());
try {
pr.append(
" Body (Plain text): "
+ PlainTextExtractor.getPlainText(content.getType(), content.getValue()));
} catch (ParserException e) {
pr.append("Failed to parse content");
}
}
pr.append("Categories: \n");
for (SyndCategory category : FeedHelper.getCategories(entry)) {
pr.append(category.getName() + "(" + category.getTaxonomyUri() + ")");
}
log.trace(pr.toString());
}
@SuppressWarnings("unchecked")
private boolean findFeedEntry(SyndFeed feed, String title, String[] bodyPortions) {
List<SyndEntry> entries = feed.getEntries();
for (SyndEntry entry : entries) {
if (entry.getTitle().equals(title)) {
if (bodyPortions == null) {
return true;
}
boolean missingPortion = false;
SyndContent description = entry.getDescription();
String value = description.getValue();
for (int i = 0; i < bodyPortions.length; i++) {
if (!value.contains(bodyPortions[i])) {
missingPortion = true;
break;
}
}
if (!missingPortion) {
return true;
}
}
}
return false;
}
protected Item createRSSItem(SyndEntry sEntry) {
Item item = super.createRSSItem(sEntry);
SyndContent sContent = sEntry.getDescription();
if (sContent != null) {
item.setDescription(createItemDescription(sContent));
}
List contents = sEntry.getContents();
if (contents != null && contents.size() > 0) {
SyndContent syndContent = (SyndContent) contents.get(0);
Content cont = new Content();
cont.setValue(syndContent.getValue());
cont.setType(syndContent.getType());
item.setContent(cont);
}
return item;
}
private Article mapArticle(SyndEntry syndEntry) {
StringBuilder sb = new StringBuilder();
for (Object obj : syndEntry.getContents()) {
if (!(obj instanceof SyndContent)) {
continue;
}
SyndContent syndContent = (SyndContent) obj;
sb.append(syndContent.getValue());
}
return Article.builder()
.num(0)
.title(syndEntry.getTitle())
.content(sb.toString())
.description(syndEntry.getDescription().getValue())
.author(syndEntry.getAuthor())
.image("")
.writtenDate(syndEntry.getPublishedDate())
.build();
}
protected Description createItemDescription(SyndContent sContent) {
Description desc = new Description();
desc.setValue(sContent.getValue());
desc.setType(sContent.getType());
return desc;
}
/**
* Generate an ebook from an RSS DOM Document.
*
* @param url The URL from where the Document was fetched (used only to set the author metadata)
* @param doc The DOM Document of the feed.
* @return An ebook.
* @throws IllegalArgumentException
* @throws FeedException
* @throws IOException
*/
private static Book createBookFromFeed(URL url, Document doc, List<Keyword> keywords)
throws IllegalArgumentException, FeedException, IOException {
Book book = new Book();
// start parsing our feed and have the above onItem methods called
SyndFeedInput input = new SyndFeedInput();
SyndFeed feed = input.build(doc);
System.out.println(feed);
// Set the title
book.getMetadata().addTitle(feed.getTitle());
// Add an Author
String author = feed.getAuthor();
if (author == null || "".equals(author.trim())) {
author = url.getHost();
}
book.getMetadata().addAuthor(new Author(author));
if (feed.getPublishedDate() != null) {
book.getMetadata().addDate(new nl.siegmann.epublib.domain.Date(feed.getPublishedDate()));
}
if (feed.getDescription() != null) {
book.getMetadata().addDescription(feed.getDescription());
}
if (feed.getCopyright() != null) {
book.getMetadata().getRights().add(feed.getCopyright());
}
// Set cover image - This has never worked.
// if (feed.getImage() != null) {
// System.out.println("There is an image for the feed");
// Promise<HttpResponse> futureImgResponse =
// WS.url(feed.getImage().getUrl()).getAsync();
// HttpResponse imgResponse = await(futureImgResponse);
// System.out.println("Content-type: " + imgResponse.getContentType());
// if (imgResponse.getContentType().startsWith("image/")) {
// String extension =
// imgResponse.getContentType().substring("image/".length());
// InputStream imageStream = imgResponse.getStream();
// book.getMetadata().setCoverImage(new Resource(imageStream, "cover." +
// extension));
// System.out.println("Using default cover");
// imageStream =
// VirtualFile.fromRelativePath("assets/cover.png").inputstream();
// if (imageStream != null) {
// System.out.println("Using default cover");
// book.getMetadata().setCoverImage(new Resource(imageStream,
// "cover.png"));
// } else {
// System.out.println("Could not load default cover");
// }
// }
// }
int entryNumber = 0;
List<SyndEntry> entries = feed.getEntries();
for (SyndEntry entry : entries) {
if (matchesKeyword(entry, keywords)) {
StringBuilder title = new StringBuilder(100);
if (entry.getTitle() != null) {
title.append(entry.getTitle());
}
if (entry.getAuthor() != null) {
title.append(" - ").append(entry.getAuthor());
}
StringBuilder content = new StringBuilder();
// Add title inside text
content.append("<h2>").append(title).append("</h2>");
if (entry.getDescription() != null) {
SyndContent syndContent = (SyndContent) entry.getDescription();
if (!syndContent.getType().contains("html")) {
content.append("<pre>\n");
}
content.append(syndContent.getValue());
if (!syndContent.getType().contains("html")) {
content.append("\n</pre>");
}
content.append("<hr/>");
}
if (entry.getContents().size() > 0) {
SyndContent syndContent = (SyndContent) entry.getContents().get(0);
if (!syndContent.getType().contains("html")) {
content.append("<pre>\n");
}
content.append(syndContent.getValue());
if (!syndContent.getType().contains("html")) {
content.append("\n</pre>");
}
}
String strContent = clean(content.toString());
// Add Chapter
try {
entryNumber++;
book.addSection(
title.toString(),
new Resource(new StringReader(strContent), "entry" + entryNumber + ".xhtml"));
} catch (IOException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
}
}
return book;
}
// build a SubscriptionEntry from Rome SyndEntry and SyndFeed
private SubscriptionEntry buildEntry(SyndEntry romeEntry) {
// if we don't have a permalink then we can't continue
if (romeEntry.getLink() == null) {
return null;
}
SubscriptionEntry newEntry = new SubscriptionEntry();
newEntry.setTitle(romeEntry.getTitle());
newEntry.setPermalink(romeEntry.getLink());
// Play some games to get the author
DCModule entrydc = (DCModule) romeEntry.getModule(DCModule.URI);
if (romeEntry.getAuthor() != null) {
newEntry.setAuthor(romeEntry.getAuthor());
} else {
newEntry.setAuthor(entrydc.getCreator()); // use <dc:creator>
}
// Play some games to get the updated date
if (romeEntry.getUpdatedDate() != null) {
newEntry.setUpdateTime(new Timestamp(romeEntry.getUpdatedDate().getTime()));
}
// TODO: should we set a default update time here?
// And more games getting publish date
if (romeEntry.getPublishedDate() != null) {
newEntry.setPubTime(new Timestamp(romeEntry.getPublishedDate().getTime())); // use <pubDate>
} else if (entrydc != null && entrydc.getDate() != null) {
newEntry.setPubTime(new Timestamp(entrydc.getDate().getTime())); // use <dc:date>
} else {
newEntry.setPubTime(newEntry.getUpdateTime());
}
// get content and unescape if it is 'text/plain'
if (romeEntry.getContents().size() > 0) {
SyndContent content = (SyndContent) romeEntry.getContents().get(0);
if (content != null && content.getType().equals("text/plain")) {
newEntry.setText(StringEscapeUtils.unescapeHtml(content.getValue()));
} else if (content != null) {
newEntry.setText(content.getValue());
}
}
// no content, try summary
if (newEntry.getText() == null || newEntry.getText().trim().length() == 0) {
if (romeEntry.getDescription() != null) {
newEntry.setText(romeEntry.getDescription().getValue());
}
}
// copy categories
if (romeEntry.getCategories().size() > 0) {
List list = new ArrayList();
Iterator cats = romeEntry.getCategories().iterator();
while (cats.hasNext()) {
SyndCategory cat = (SyndCategory) cats.next();
list.add(cat.getName());
}
newEntry.setCategoriesString(list);
}
return newEntry;
}
