Example usage for org.jsoup.parser Parser xmlParser

List of usage examples for org.jsoup.parser Parser xmlParser

  1. HOME
  2. Java
  3. org.jsoup
  4. org.jsoup.parser.*
  5. Parser
  6. xmlParser

Introduction

In this page you can find the example usage for org.jsoup.parser Parser xmlParser.

Prototype

public static Parser xmlParser()

Source Link

Document

Create a new XML parser.

Usage

From source file:org.structr.web.importer.Importer.java

/**
 * Parse the code previously read by {@link Importer#readPage()} and treat it as page fragment.
 *
 * @param fragment//from   w  ww .j  a  v  a2  s .c o m
 * @return
 * @throws FrameworkException
 */
public boolean parse(final boolean fragment) throws FrameworkException {

    init();

    if (StringUtils.isNotBlank(code)) {

        if (!isDeployment) {
            logger.info("##### Start parsing code for page {} #####", new Object[] { name });
        } else {

            // a trailing slash to all void/self-closing tags so the XML parser can parse it correctly
            code = code.replaceAll(
                    "<(area|base|br|col|command|embed|hr|img|input|keygen|link|meta|param|source|track|wbr)([^>]*)>",
                    "<$1$2/>");
        }

        if (fragment) {

            if (isDeployment) {

                final List<Node> nodeList = Parser.parseXmlFragment(code, "");
                parsedDocument = Document.createShell("");
                final Element body = parsedDocument.body();
                final Node[] nodes = nodeList.toArray(new Node[nodeList.size()]);

                for (int i = nodes.length - 1; i > 0; i--) {
                    nodes[i].remove();
                }

                for (Node node : nodes) {
                    body.appendChild(node);
                }

            } else {

                parsedDocument = Jsoup.parseBodyFragment(code);
            }

        } else {

            if (isDeployment) {

                parsedDocument = Jsoup.parse(code, "", Parser.xmlParser());

            } else {

                parsedDocument = Jsoup.parse(code);
            }

        }

    } else {

        if (!isDeployment) {
            logger.info("##### Start fetching {} for page {} #####", new Object[] { address, name });
        }

        code = HttpHelper.get(address);
        parsedDocument = Jsoup.parse(code);

    }

    return true;

}

From source file:org.tinymediamanager.scraper.anidb.AniDBMetadataProvider.java

@Override
public MediaMetadata getTvShowMetadata(MediaScrapeOptions options) throws Exception {
    MediaMetadata md = new MediaMetadata(providerInfo.getId());
    String id = "";
    String langu = options.getLanguage().name();

    // id from result
    if (options.getResult() != null) {
        id = options.getResult().getId();
    }//from   w  ww  .jav a2  s.  co m

    // do we have an id from the options?
    if (StringUtils.isEmpty(id)) {
        id = options.getId(providerInfo.getId());
    }

    if (StringUtils.isEmpty(id)) {
        return md;
    }

    trackConnections();

    // call API http://api.anidb.net:9001/httpapi?request=anime&client=tinymediamanager&clientver=2&protover=1&aid=4242
    String url = "http://api.anidb.net:9001/httpapi?request=anime&client=tinymediamanager&clientver=2&protover=1&aid="
            + id;
    Document doc = null;
    try {
        CachedUrl cachedUrl = new CachedUrl(url);

        doc = Jsoup.parse(cachedUrl.getInputStream(), "UTF-8", "", Parser.xmlParser());
    } catch (Exception e) {
        LOGGER.error("failed to get TV show metadata: " + e.getMessage());

        // clear cache
        CachedUrl.removeCachedFileForUrl(url);
    }

    if (doc == null || doc.children().size() == 0) {
        return md;
    }

    md.setId(providerInfo.getId(), id);

    Element anime = doc.child(0);

    for (Element e : anime.children()) {
        if ("startdate".equalsIgnoreCase(e.tagName())) {
            md.storeMetadata(MediaMetadata.RELEASE_DATE, e.text());
            try {
                Date date = org.tinymediamanager.scraper.util.StrgUtils.parseDate(e.text());
                md.storeMetadata(MediaMetadata.YEAR, new SimpleDateFormat("yyyy").format(date));
            } catch (Exception ex) {
            }
        }

        if ("titles".equalsIgnoreCase(e.tagName())) {
            parseTitle(md, langu, e);
        }

        if ("description".equalsIgnoreCase(e.tagName())) {
            md.storeMetadata(MediaMetadata.PLOT, e.text());
        }

        if ("ratings".equalsIgnoreCase(e.tagName())) {
            getRating(md, e);
        }

        if ("picture".equalsIgnoreCase(e.tagName())) {
            md.storeMetadata(MediaMetadata.POSTER_URL, IMAGE_SERVER + e.text());
        }

        if ("characters".equalsIgnoreCase(e.tagName())) {
            getActors(md, e);
        }

    }

    // add static "Anime" genre
    md.addGenre(MediaGenres.ANIME);

    return md;
}

From source file:org.tinymediamanager.scraper.anidb.AniDBMetadataProvider.java

@Override
public MediaMetadata getEpisodeMetadata(MediaScrapeOptions options) throws Exception {
    MediaMetadata md = new MediaMetadata(providerInfo.getId());

    String id = "";
    String langu = options.getLanguage().name();

    // id from result
    if (options.getResult() != null) {
        id = options.getResult().getId();
    }//from  w w w.jav a 2s .c o m

    // do we have an id from the options?
    if (StringUtils.isEmpty(id)) {
        id = options.getId(providerInfo.getId());
    }

    if (StringUtils.isEmpty(id)) {
        return md;
    }

    // get episode number and season number
    int seasonNr = -1;
    int episodeNr = -1;

    try {
        seasonNr = Integer.parseInt(options.getId(MediaMetadata.SEASON_NR));
        episodeNr = Integer.parseInt(options.getId(MediaMetadata.EPISODE_NR));
    } catch (Exception e) {
        LOGGER.warn("error parsing season/episode number");
    }

    if (seasonNr == -1 || episodeNr == -1) {
        return md;
    }

    trackConnections();

    String url = "http://api.anidb.net:9001/httpapi?request=anime&client=tinymediamanager&clientver=2&protover=1&aid="
            + id;
    Document doc = null;
    try {
        CachedUrl cachedUrl = new CachedUrl(url);
        doc = Jsoup.parse(cachedUrl.getInputStream(), "UTF-8", "", Parser.xmlParser());
    } catch (Exception e) {
        LOGGER.error("failed to get episode metadata: " + e.getMessage());

        // clear cache
        CachedUrl.removeCachedFileForUrl(url);
    }

    if (doc == null || doc.children().size() == 0) {
        return md;
    }

    md.setId(providerInfo.getId(), id);

    List<Episode> episodes = parseEpisodes(doc);

    Episode episode = null;

    // filter out the episode
    for (Episode ep : episodes) {
        if (ep.season == seasonNr && ep.episode == episodeNr) {
            episode = ep;
            break;
        }
    }

    if (episode == null) {
        return md;
    }

    String title = episode.titles.get(langu);
    if (StringUtils.isBlank(title)) {
        title = episode.titles.get("en");
    }
    if (StringUtils.isBlank(title)) {
        title = episode.titles.get("x-jat");
    }
    md.storeMetadata(MediaMetadata.TITLE, title);
    md.storeMetadata(MediaMetadata.PLOT, episode.summary);
    md.storeMetadata(MediaMetadata.RATING, episode.rating);
    md.storeMetadata(MediaMetadata.RELEASE_DATE, episode.airdate);
    md.storeMetadata(MediaMetadata.RUNTIME, episode.runtime);
    md.setId(providerInfo.getId(), episode.id);

    return md;
}

From source file:org.tinymediamanager.scraper.anidb.AniDBMetadataProvider.java

@Override
public List<MediaEpisode> getEpisodeList(MediaScrapeOptions options) throws Exception {
    List<MediaEpisode> episodes = new ArrayList<MediaEpisode>();

    String id = "";
    String langu = options.getLanguage().name();

    // id from result
    if (options.getResult() != null) {
        id = options.getResult().getId();
    }//w  w w  .  j  a v a2s  . com

    // do we have an id from the options?
    if (StringUtils.isEmpty(id)) {
        id = options.getId(providerInfo.getId());
    }

    if (StringUtils.isEmpty(id)) {
        return episodes;
    }

    trackConnections();

    String url = "http://api.anidb.net:9001/httpapi?request=anime&client=tinymediamanager&clientver=2&protover=1&aid="
            + id;
    Document doc = null;
    try {
        CachedUrl cachedUrl = new CachedUrl(url);
        doc = Jsoup.parse(cachedUrl.getInputStream(), "UTF-8", "", Parser.xmlParser());
    } catch (Exception e) {
        LOGGER.error("error getting episode list: " + e.getMessage());

        // clear cache
        CachedUrl.removeCachedFileForUrl(url);
    }

    if (doc == null || doc.children().size() == 0) {
        return episodes;
    }

    // filter out the episode
    for (Episode ep : parseEpisodes(doc)) {
        MediaEpisode episode = new MediaEpisode(getProviderInfo().getId());
        episode.title = ep.titles.get(langu);
        episode.season = ep.season;
        episode.episode = ep.episode;
        if (StringUtils.isBlank(episode.title)) {
            episode.title = ep.titles.get("en");
        }
        if (StringUtils.isBlank(episode.title)) {
            episode.title = ep.titles.get("x-jat");
        }

        episode.plot = ep.summary;
        episode.rating = ep.rating;
        episode.firstAired = ep.airdate;
        episode.ids.put(providerInfo.getId(), ep.id);
    }

    return episodes;
}