Java SyndEntry.getPublishedDate方法代码示例

import com.rometools.rome.feed.synd.SyndEntry; //导入方法依赖的package包/类
/**
 * This Comparator assumes that SyndEntry returns a valid Date either for
 * getPublishedDate() or getUpdatedDate().
 */
private Comparator<SyndEntry> getEntryComparator() {
	return new Comparator<SyndEntry>() {
		@Override
		public int compare(SyndEntry e1, SyndEntry e2) {
			if (e1.getPublishedDate() == null &&
					e1.getUpdatedDate() == null) {
				// we will be ignoring such entries anyway
				return 0;
			}
			Date d1 =
					e1.getPublishedDate() != null ? e1.getPublishedDate() :
							e1.getUpdatedDate();
			Date d2 =
					e2.getPublishedDate() != null ? e2.getPublishedDate() :
							e2.getUpdatedDate();
			if (d1.after(d2)) return 1;
			if (d1.before(d2)) return -1;
			return 0;
		}
	};
}
 

import com.rometools.rome.feed.synd.SyndEntry; //导入方法依赖的package包/类
static JsonObject toJson(SyndEntry entry) {
	JsonObject json = new JsonObject();
	json.put("title", entry.getTitle());
	Date published = entry.getPublishedDate();
	if (published == null) {
		// TODO : log warning ? use another date ?
		LOG.warn("!!!!!! The RSS has no published date : this will lead to duplicates entry");
		published = new Date(); // FIXME : absolutely wrong...
	}
	json.put("published", toJson(published));
	json.put("score", (double)published.getTime());
	json.put("link", entry.getLink());
	SyndContent description = entry.getDescription();
	if (description != null) {
		json.put("description", description.getValue());
	}
	return json;
}
 

import com.rometools.rome.feed.synd.SyndEntry; //导入方法依赖的package包/类
private long postFeedEntries(Feed feed, List<SyndEntry> entries)
		throws DbException {

	long lastEntryTime = feed.getLastEntryTime();
	Transaction txn = db.startTransaction(false);
	try {
		Collections.sort(entries, getEntryComparator());
		for (SyndEntry entry : entries) {
			long entryTime;
			if (entry.getPublishedDate() != null) {
				entryTime = entry.getPublishedDate().getTime();
			} else if (entry.getUpdatedDate() != null) {
				entryTime = entry.getUpdatedDate().getTime();
			} else {
				// no time information available, ignore this entry
				LOG.warning("Entry has no date, ignored.");
				continue;
			}
			if (entryTime > feed.getLastEntryTime()) {
				postEntry(txn, feed, entry);
				if (entryTime > lastEntryTime) lastEntryTime = entryTime;
			}
		}
		db.commitTransaction(txn);
	} finally {
		db.endTransaction(txn);
	}
	return lastEntryTime;
}
 

import com.rometools.rome.feed.synd.SyndEntry; //导入方法依赖的package包/类
private Date getLastDate(SyndEntry entry) {
    if(entry.getUpdatedDate() != null) {
        return entry.getUpdatedDate();
    } else {
        return entry.getPublishedDate();
    }
}
 

import com.rometools.rome.feed.synd.SyndEntry; //导入方法依赖的package包/类
private String computeEntryTag(final SyndEntry entry) {

        // Following hash algorithm suggested by Robert Cooper needs to be
        // evaluated in future.
        // int hash = ( entry.getUri() != null ? entry.getUri().hashCode() :
        // entry.getLink().hashCode() ) ^
        // (entry.getUpdatedDate() != null ? entry.getUpdatedDate().hashCode() :
        // entry.getPublishedDate().hashCode()) ^
        // entry.getTitle().hashCode() ^
        // entry.getDescription().hashCode();

        final String id = entry.getUri();
        Date updateDate = entry.getUpdatedDate();
        final Date publishedDate = entry.getPublishedDate();
        if (updateDate == null) {
            if (publishedDate != null) {
                updateDate = publishedDate;
            } else {
                // For misbehaving feeds that do not set updateDate or
                // publishedDate we use current tiem which pretty mucg assures
                // that it will be viewed as
                // changed even when it is not
                updateDate = new Date();
            }
        }
        final String key = id + ":" + entry.getUpdatedDate();
        return computeDigest(key);
    }
 

import com.rometools.rome.feed.synd.SyndEntry; //导入方法依赖的package包/类
@Override
protected SyndEntry createSyndEntry(final Item item, final boolean preserveWireItem) {

    final SyndEntry syndEntry = super.createSyndEntry(item, preserveWireItem);

    final Date pubDate = item.getPubDate();
    final Date publishedDate = syndEntry.getPublishedDate();
    if (pubDate != null && publishedDate == null) {
        syndEntry.setPublishedDate(pubDate); // c
    }

    return syndEntry;
}
 

import com.rometools.rome.feed.synd.SyndEntry; //导入方法依赖的package包/类
private Date getEntryUpdateDate(SyndEntry item) {
	Date date = item.getUpdatedDate();
	if (date == null) {
		date = item.getPublishedDate();
	}
	if (date == null) {
		date = new Date();
	}
	return date;
}
 

import com.rometools.rome.feed.synd.SyndEntry; //导入方法依赖的package包/类
private String getTitle(SyndEntry item) {
	String title = item.getTitle();
	if (StringUtils.isBlank(title)) {
		Date date = item.getPublishedDate();
		if (date != null) {
			title = DateFormat.getInstance().format(date);
		} else {
			title = "(no title)";
		}
	}
	return StringUtils.trimToNull(title);
}
 

import com.rometools.rome.feed.synd.SyndEntry; //导入方法依赖的package包/类
private List<Outlink> parseFeed(String url, byte[] content,
        Metadata parentMetadata) throws Exception {
    List<Outlink> links = new ArrayList<>();

    SyndFeed feed = null;
    try (ByteArrayInputStream is = new ByteArrayInputStream(content)) {
        SyndFeedInput input = new SyndFeedInput();
        feed = input.build(new InputSource(is));
    }

    URL sURL = new URL(url);

    List<SyndEntry> entries = feed.getEntries();
    for (SyndEntry entry : entries) {
        String targetURL = entry.getLink();
        // targetURL can be null?!?
        // e.g. feed does not use links but guid
        if (StringUtils.isBlank(targetURL)) {
            targetURL = entry.getUri();
            if (StringUtils.isBlank(targetURL)) {
                continue;
            }
        }
        Outlink newLink = filterOutlink(sURL, targetURL, parentMetadata);
        if (newLink == null)
            continue;

        String title = entry.getTitle();
        if (StringUtils.isNotBlank(title)) {
            newLink.getMetadata().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;
                }
            }
            newLink.getMetadata().setValue("feed.publishedDate",
                    publishedDate.toString());
        }

        SyndContent description = entry.getDescription();
        if (description != null
                && StringUtils.isNotBlank(description.getValue())) {
            newLink.getMetadata().setValue("feed.description",
                    description.getValue());
        }

        links.add(newLink);
    }

    return links;
}
 

import com.rometools.rome.feed.synd.SyndEntry; //导入方法依赖的package包/类
/**
 * RSSの解析処理。
 * @param rss １つのRSS情報
 * @return 解析後のLink情報
 * @throws FeedException Feedの解析失敗時の例外
 * @throws IOException IOException IO例外
 * @throws URISyntaxException URIの形式が不正な場合の例外
 * @throws CrawlerException Crawler共通例外
 */
@SuppressWarnings("deprecation")
public ArrayList<ArticleEntityBean> parseRss(RssEntityBean rss) throws IOException, FeedException,
                URISyntaxException, CrawlerException {
    log.info("start feed parse : " + rss.url);
    SyndFeed feed = null;
    feed = buildSyndFeed(new URI(rss.url));
    ArrayList<ArticleEntityBean> result = new ArrayList<ArticleEntityBean>();
    TimeZone tz = TimeZone.getTimeZone("GMT");
    if (feed != null) {
        List<SyndEntry> entries = feed.getEntries();
        log.info("feed entries count : " + entries.size());
        for (Object obj : entries) {
            SyndEntry entry = (SyndEntry) obj;
            String link = entry.getLink();
            DateTime createdAt = new DateTime(entry.getPublishedDate(), tz);
            String createdAtStr = Util.formatIsoDate(createdAt);
            String title = org.apache.commons.lang.StringEscapeUtils.unescapeXml(entry.getTitle());

            log.info("article : " + title + " / " + createdAtStr + " / " + link);
            if (isExistArticle(link, createdAt)) {
                log.info("->  article already exists");
                continue;
            }
            log.info("->  article not exists");

            ArticleEntityBean article = new ArticleEntityBean();
            article.link = link;
            article.title = title;
            article.auther = entry.getAuthor();
            article.url = entry.getUri();
            article.createdAt = createdAt;
            String fixHour = Conf.getValue("fix_time");
            Date now = new Date();
            Date fixDate = new Date(now.getYear(), now.getMonth(), now.getDate(), Integer.parseInt(fixHour), 0);
            if (now.getTime() > fixDate.getTime()) {
                article.publishedAt = tomorrow;
            } else {
                article.publishedAt = today;
            }
            article.site = rss.site;
            article.type = rss.type;
            article.description = null;
            SyndContent sc = entry.getDescription();
            if (sc != null) {
                article.description = sc.getValue();
                if (rss.replaceCR != null) {
                    article.rawHTML = article.description.replaceAll("\n", rss.replaceCR);
                } else {
                    article.rawHTML = article.description;
                }
            }

            article.tags = rss.defaultTag;
            result.add(article);
        }
    } else {
        log.warn("feed parse error url : " + rss.site);
        throw new CrawlerException();
    }
    return result;
}
 

import com.rometools.rome.feed.synd.SyndEntry; //导入方法依赖的package包/类
public Date getEntryPublishedDate(final Object o) {
    final SyndEntry e = (SyndEntry) o;
    return e.getPublishedDate();
}