if (!publishDateFlag) {
String[] lastURLs = this.config.getUrlLevels()[(this.config.getUrlLevels().length - 1)].split("\\\n", -1);
if (lastURLs.length != 1) {
return;
}
RegexParser rpUrl = new RegexParser(lastURLs[0]);
boolean numberFlag = true;
for (int i = 0; i < set.size(); ++i) {
String url = set.get(i).getReferURL();
rpUrl.setText(url);
if (rpUrl.match()) {
String v = rpUrl.getMapx().getString("SortID");
set.get(i).setProp2(v);
if (!NumberUtil.isLong(v)) {
numberFlag = false;
}
}
}
set.sort("Prop2", "asc", numberFlag);
for (int i = set.size() - 1; i >= 0; --i) {
set.get(i).setOrderFlag(OrderUtil.getDefaultOrder());
set.get(i).setProp2(null);
}
set.update();
}
public void writeArticle()
{
prepareList();
if (this.config.getType() == 1) {
QueryBuilder imageQB = new QueryBuilder("select id from zccatalog where type=4 and siteid=?",
CatalogUtil.getSiteID(this.config.getCatalogID()));
String imageCatalogID = imageQB.executeString();
if (StringUtil.isEmpty(CatalogUtil.getSiteID(this.config.getCatalogID()))) {
LogUtil.getLogger().warn("文档采集的目的栏目不存在：ID=" + this.config.getCatalogID());
return;
}
String sitePath = SiteUtil.getAbsolutePath(CatalogUtil.getSiteID(this.config.getCatalogID()));
String imagePath = "upload/Image/" + CatalogUtil.getAlias(imageCatalogID) + "/";
RegexParser rp = this.config.getTemplate("Ref1");
RegexParser[] filters = this.config.getFilterBlocks();
this.list.moveFirst();
WebDocument doc = null;
int cSuccess = 0;
int cFailure = 0;
int cLost = 0;
boolean publishDateFlag = false;
ZCArticleSet set = new ZCArticleSet();
while ((doc = this.list.next()) != null) {
if (this.task.checkStop()) {
return;
}
if (doc.getLevel() != this.config.getUrlLevels().length - 1) {
continue;
}
int percent = (100 - this.task.getPercent()) * (cSuccess + cFailure + cLost) / this.list.size();
this.task.setPercent(this.task.getPercent() + percent);
if ((doc.isTextContent()) && (doc.getContent() != null)) {
String text = doc.getContentText();
rp.setText(text);
if (rp.match()) {
Mapx map = rp.getMapx();
Object[] ks = map.keyArray();
Object[] vs = map.valueArray();
for (int i = 0; i < map.size(); ++i) {
String key = ks[i].toString();
String value = vs[i].toString();
if (!key.equalsIgnoreCase("Content")) {
value = this.tagPattern.matcher(value).replaceAll("");
}
value = StringUtil.htmlDecode(value);
value = value.trim();
map.put(key, value);
}
String title = map.getString("Title");
String content = map.getString("Content");
String author = map.getString("Author");
String source = map.getString("Source");
String strDate = map.getString("PublishDate");
Date publishDate = doc.getLastmodifiedDate();
if ((StringUtil.isNotEmpty(strDate)) && (StringUtil.isNotEmpty(this.config.getPublishDateFormat())))
{
try {
strDate = DateUtil.convertChineseNumber(strDate);
publishDate = DateUtil.parse(strDate, this.config.getPublishDateFormat());
} catch (Exception e) {
this.task.addError("日期" + strDate + "不符合指定格式" + doc.getUrl());
}
publishDateFlag = true;
}
if (publishDate.getTime() > System.currentTimeMillis()) {
publishDate = new Date();
}
ArticleAPI api = new ArticleAPI();
try {
ZCArticleSchema article = new ZCArticleSchema();
if (StringUtil.isNotEmpty(title)) {
article.setTitle(title);
} else {
++cLost;
break label1209:
}
if (StringUtil.isNotEmpty(content)) {
content = content.trim();
while (rp.match()) {
String html = rp.getMapx().getString("Content");
content = content + html;
}
if (this.config.isCleanLinkFlag()) {
content = this.framePattern.matcher(content).replaceAll("");
content = this.stylePattern.matcher(content).replaceAll("");
content = this.scriptPattern.matcher(content).replaceAll("");
content = this.linkPattern.matcher(content).replaceAll("$1");
}
if (filters != null) {
for (int k = 0; k < filters.length; ++k) {
content = filters[k].replace(content, "");
}
}
String str = dealImage(content, doc.getUrl(), sitePath, imagePath, imageCatalogID);
article.setContent(str);
} else {
++cLost;
break label1209:
}
if (StringUtil.isNotEmpty(author)) {
article.setAuthor(author);
}
if (StringUtil.isNotEmpty(source)) {
article.setReferName(source);
}
article.setReferURL(doc.getUrl());
article.setPublishDate(publishDate);
article.setCatalogID(this.config.getCatalogID());
article.setBranchInnerCode("0001");
article.setProp2("FromWeb");
if (ExtendManager.hasAction("FromWeb.BeforeSave")) {
ExtendManager.executeAll("FromWeb.BeforeSave", new Object[] { article });
}
Date date = (Date)new QueryBuilder(
"select PublishDate from ZCArticle where ReferURL=? and CatalogID=?", doc.getUrl(),
this.config.getCatalogID()).executeOneValue();
if (date != null) {
if (date.getTime() < doc.getLastDownloadTime()) {
QueryBuilder qb = new QueryBuilder(
"update ZCArticle set Title=?,Content=? where CatalogID=? and ReferURL=?");
qb.add(article.getTitle());
qb.add(article.getContent());
qb.add(this.config.getCatalogID());
qb.add(doc.getUrl());
qb.executeNoQuery();
}
++cSuccess;
} else {
api.setSchema(article);
set.add(article);
if (api.insert() > 0L)
++cSuccess;
else
++cFailure;
}
}
catch (Exception e) {
++cFailure;
e.printStackTrace();
}
} else {
LogUtil.getLogger().info("未能匹配" + doc.getUrl());
this.task.addError("未能匹配" + doc.getUrl());
++cLost;
}
label1209: this.task.setCurrentInfo("正在转换文档, <font class='green'>" + cSuccess + "</font> 个成功, <font class='red'>" +
cFailure + "</font> 个失败, <font class='green'>" + cLost + "</font> 个未匹配");
}
}
