public void setEpisodeAttributes(Episode episode, Podcast podcast, SyndEntry entry) {
// set DESCRIPTION for episode - used in search
if (null != entry.getDescription()) {
String episodeDesc = entry.getDescription().getValue();
// tags are removed from description
String descWithoutTabs = episodeDesc.replaceAll("\\<[^>]*>", "");
// carriage returns are removed from description - for player
String descWithoutEndOfLine = descWithoutTabs.replaceAll("\\n", "");
if (descWithoutEndOfLine.length() > MAX_LENGTH_DESCRIPTION) {
episode.setDescription(descWithoutEndOfLine.substring(0, MAX_LENGTH_DESCRIPTION));
} else {
episode.setDescription(descWithoutEndOfLine);
}
}
// set author
episode.setAuthor(entry.getAuthor());
// set title for episode - used in search
String episodeTitle = entry.getTitle();
if (episodeTitle != null) {
// removes quotes to display properly in player
episodeTitle = episodeTitle.replaceAll("\"", "");
if (episodeTitle.length() > MAX_PERMITTED_TITLE_LENGTH) {
episodeTitle = episodeTitle.substring(0, MAX_PERMITTED_TITLE_LENGTH);
}
episode.setTitle(episodeTitle);
String titleInUrl = episodeTitle.trim().replaceAll("[^a-zA-Z0-9\\-\\s\\.]", "");
titleInUrl = titleInUrl.replaceAll("[\\-| |\\.]+", "-");
if (titleInUrl.length() > TITLE_IN_URL_MAX_LENGTH) {
episode.setTitleInUrl(titleInUrl.substring(0, TITLE_IN_URL_MAX_LENGTH));
} else {
episode.setTitleInUrl(titleInUrl);
}
}
episode.setLink(entry.getLink());
// in the beginning inherit the media type from the podcast
episode.setMediaType(podcast.getMediaType());
// get the list of enclosures
@SuppressWarnings("unchecked")
List<SyndEnclosure> enclosures = (List<SyndEnclosure>) entry.getEnclosures();
List<String> audioMimeTypesList = Arrays.asList(audioMimeTypesArray);
List<String> videoMimeTypesList = Arrays.asList(videoMimeTypesArray);
// set media url for the episode - this will be played in the player
if (null != enclosures) {
// if in the enclosure list is a media type (either audio or video),
// this will set as the link of the episode
for (SyndEnclosure enclosure : enclosures) {
if (null != enclosure) {
episode.setMediaUrl(enclosure.getUrl());
if (enclosure.getLength() >= 0) episode.setLength(enclosure.getLength());
// when adding a new podcast media type is selected for the
// podcast based on an initial view, but it can be that is a
// mixed podcast so both audio
// and video should be considered and in that case PRIORITY
// has the type of the episode if any...
if (null != enclosure.getType()) {
episode.setEnclosureType(enclosure.getType().trim());
if (audioMimeTypesList.contains(enclosure.getType().trim())) {
episode.setMediaType(MediaType.Audio);
break;
}
if (videoMimeTypesList.contains(enclosure.getType().trim())) {
episode.setMediaType(MediaType.Video);
break;
}
}
}
}
} else {
episode.setMediaUrl("noMediaUrl");
}
if (episode.getMediaUrl() == null) {
episode.setMediaUrl("noMediaUrl");
}
if (episode.getMediaUrl() == null || episode.getMediaUrl().equals("noMediaUrl")) {
LOG.warn(
"PodcastId["
+ podcast.getPodcastId()
+ "] - "
+ "COULD NOT SET MEDIA URL - "
+ "epTitle["
+ entry.getTitle()
+ "]"
+ "feed["
+ podcast.getUrl()
+ "]");
}
// set link attribute
episode.setLink(entry.getLink());
episode.setPublicationDate(entry.getPublishedDate());
updatePodcastPublicationDateAndLastMediaUrl(episode, podcast);
if (episode.getPublicationDate() == null) {
LOG.warn(
"PodcastId["
+ podcast.getPodcastId()
+ "] - "
+ "COULD NOT SET publication date "
+ "epTitle["
+ entry.getTitle()
+ "]"
+ "feed["
+ podcast.getUrl()
+ "]");
}
}
private List<Outlink> parseFeed(String url, byte[] content, Metadata parentMetadata)
throws MalformedURLException {
List<Outlink> links = new ArrayList<>();
SyndFeed feed = null;
try (ByteArrayInputStream is = new ByteArrayInputStream(content)) {
SyndFeedInput input = new SyndFeedInput();
feed = input.build(new InputSource(is));
} catch (Exception e) {
LOG.error("Exception parsing feed from DOM {}", url);
return links;
}
URL sURL = new URL(url);
List<SyndEntry> entries = feed.getEntries();
for (SyndEntry entry : entries) {
String targetURL = entry.getLink();
// build an absolute URL
try {
targetURL = URLUtil.resolveURL(sURL, targetURL).toExternalForm();
} catch (MalformedURLException e) {
LOG.debug("MalformedURLException on {}", targetURL);
continue;
}
targetURL = urlFilters.filter(sURL, parentMetadata, targetURL);
if (StringUtils.isBlank(targetURL)) continue;
Outlink newLink = new Outlink(targetURL);
Metadata targetMD = metadataTransfer.getMetaForOutlink(targetURL, url, parentMetadata);
newLink.setMetadata(targetMD);
String title = entry.getTitle();
if (StringUtils.isNotBlank(title)) {
targetMD.setValue("feed.title", title.trim());
}
Date publishedDate = entry.getPublishedDate();
if (publishedDate != null) {
// filter based on the published date
if (filterHoursSincePub != -1) {
Calendar rightNow = Calendar.getInstance();
rightNow.add(Calendar.HOUR, -filterHoursSincePub);
if (publishedDate.before(rightNow.getTime())) {
LOG.info(
"{} has a published date {} which is more than {} hours old",
targetURL,
publishedDate.toString(),
filterHoursSincePub);
continue;
}
}
targetMD.setValue("feed.publishedDate", publishedDate.toString());
}
SyndContent description = entry.getDescription();
if (description != null && StringUtils.isNotBlank(description.getValue())) {
targetMD.setValue("feed.description", description.getValue());
}
links.add(newLink);
}
return links;
}