/**
* 提取网页中所有的IssueComment元素
*
* @param source
*/
private List<IssueCommentEvent> processComment(
NodeList nodeList, List<IssueCommentEvent> icList) {
SimpleNodeIterator sni = nodeList.elements();
while (sni.hasMoreNodes()) {
Node node = sni.nextNode();
if (node.getText().matches("div id=\"issuecomment-.*\".*+")) {
IssueCommentEvent i = new IssueCommentEvent();
// TODO 解析comment工作
Node actorNode = DownloadUtil.getSomeChild(node, "class=\"author\"");
i.setActor(actorNode.toPlainTextString());
Node contentNode = DownloadUtil.getSomeChild(node, "div class=\"comment-body");
i.setCommentBody(contentNode.toPlainTextString());
Node timeNode = DownloadUtil.getSomeChild(node, "datetime");
Pattern pattern = Pattern.compile("datetime=\".*\"");
Matcher matcher = pattern.matcher(timeNode.getText());
if (matcher.find()) {
String time = matcher.group().split("\"")[1];
i.setCreatedAt(time);
System.out.println(time);
}
icList.add(i);
} else {
// 得到该节点的子节点列表
NodeList childList = node.getChildren();
// 孩子节点为空，说明是值节点
if (null != childList) { // 如果孩子结点不为空则递归调用
processComment(childList, icList);
}
}
}
return icList;
}
public void getDataFromPullUrl(String url) {
/** 将数据填充到issueCommentEvent对象或IssuesEvent对象中，并用hibernate实现存储 */
NodeList nodeList = null;
String artifactId = "";
try {
String homePage = htmlAnalyzer.getResource(url);
nodeList = htmlAnalyzer.getNodeListByHtmlPage(homePage);
artifactId = processArtifactId(homePage);
if (!(artifactId).equals(url)) {
System.out.println("artifact 已经转换为其他形式");
return;
}
} catch (Exception e) {
System.out.println("出现异常");
return;
}
List<PullRequestEvent> pList = new ArrayList<PullRequestEvent>();
List<IssueCommentEvent> icList = new ArrayList<IssueCommentEvent>();
List<PullRequestReviewCommentEvent> prList = new ArrayList<PullRequestReviewCommentEvent>();
List<PushEvent> psList = new ArrayList<PushEvent>();
/**
* 具体处理流程,一个完整的pullrequest大概包括： 1、开启一个pullrequest请求； 2、代码的拥有者或协作者对代码进行评论；
* 3、代码的拥有者对代码进行codeReview并对某一段代码进行评论 4、代码的提交者对代码进行修改 5、代码被主干的拥有者认可，最后将代码merge(push)到主干上
*/
pList = this.processOpenPull(nodeList, pList);
pList = this.processClosePull(nodeList, pList);
pList = this.processMileStone(nodeList, pList);
pList = this.processRemoveMileStone(nodeList, pList);
pList = this.processReference(nodeList, pList);
pList = this.processLabled(nodeList, pList);
pList = this.processUnLabled(nodeList, pList);
icList = this.processComment(nodeList, icList);
/*prList = this.processPullRequestReviewComment(nodeList, prList);*/
pList = this.processAssigned(nodeList, pList);
pList = this.processUnassigned(nodeList, pList);
/** 解析所有对象中共有的信息，包括ArtifactId,网页URL，数据来源类型net,网页的repository */
for (IssueCommentEvent i : icList) {
i.setArtifactId(artifactId);
i.setHtmlUrl(url);
i.setSourceType("net");
i.setRepo(getRepo());
}
for (PullRequestEvent p : pList) {
p.setArtifactId(artifactId);
p.setPullrequestHtmlUrl(url);
p.setSourceType("net");
p.setRepo(getRepo());
}
for (PullRequestReviewCommentEvent pr : prList) {
pr.setArtifactId(artifactId);
pr.setHtmlUrl(url);
pr.setSourceType("net");
pr.setRepo(getRepo());
}
for (PushEvent ps : psList) {
ps.setArtifactId(artifactId);
ps.setHtmlUrl(url);
ps.setSourceType("net");
ps.setRepo(getRepo());
}
/** 向数据库中持久化数据 */
session.beginTransaction();
for (IssueCommentEvent ic : icList) {
session.save(ic);
}
for (PullRequestEvent p : pList) {
session.save(p);
if (p.getAction().equals("open")) {
ArtifactOwner artifactOwner = new ArtifactOwner();
artifactOwner.setArtifactId(artifactId);
artifactOwner.setOwner(p.getActor());
session.save(artifactOwner);
}
}
for (PullRequestReviewCommentEvent pr : prList) {
session.save(pr);
}
for (PushEvent ps : psList) {
session.save(ps);
}
session.getTransaction().commit();
}
