@SuppressWarnings("finally")
public JSONObject getKeyWords(JSONObject semiData) {
JSONObject keyWords = new JSONObject();
try {
this.buildIndex(semiData);
this.getIndexInfo(this.indexDir, 4);
this.generateWekaFile(this.termList, this.maxDocNum, this.wekaFile);
JSONArray array = this.Cluster(this.wekaFile, 7);
int totalNum = 0;
for (int i = 0; i < array.length(); i++) {
totalNum += array.getJSONArray(i).length();
}
keyWords.put("maxFreq", this.maxFreq);
keyWords.put("totalNum", totalNum);
keyWords.put("WordList", array);
} catch (WeiboException e) {
System.out.print("getKeyWords++++++weibo\n");
System.out.print("error:" + e.getError() + "toString:" + e.toString());
keyWords.put("error", e.getError());
e.printStackTrace();
} catch (Exception e) {
System.out.print("getKeyWords++++++Exception");
keyWords.put("error", e.toString());
e.printStackTrace();
} finally {
try {
this.myDelete(this.indexDir);
this.myDelete(this.wekaFile);
} catch (Exception e) {
e.printStackTrace();
}
return keyWords;
}
}
private void process() {
// check start crawl time
if (!CrawlTool.checkStartCrawlTime(this.startCrawlTimeStamp)) { // wait
// to
// start
Long nowTimeStamp = new Date().getTime();
int waitTimeSecond =
CrawlTool.safeLongToInt((this.startCrawlTimeStamp - nowTimeStamp) / 1000);
System.out.println(
this.logSign
+ " Sleep a while to start: "
+ waitTimeSecond
+ "s, or:"
+ waitTimeSecond / 60
+ "m.");
CrawlTool.sleep(waitTimeSecond, this.logSign);
}
// the variables and flags
String currentKey = "";
int page = 0;
String currentUid = "";
boolean isResultEmpty = false;
boolean isResultTooOld = false;
boolean isError = false;
while (true) {
// check if it is time to stop
boolean stopCrawl = CrawlTool.checkStopCrawlTime(this.stopCrawlTimeStamp);
if (stopCrawl) {
// report this round
System.out.println("Time's up. Stop crawling now.");
break;
}
// check if it is the max times to crawl
if (!(this.requestedCrawlingTimes > 0 && this.requestedCrawlingTimes >= this.crawledTimes)) {
return;
}
// prepare the key
currentKey = CrawlTool.getNextKey(keyList, currentKey);
// prepare the uid, page, count
if (null == currentUid || "".equals(currentUid)) {
currentUid = getNextUidTillEnd(currentUid);
if (currentUid == "finish" || "finish".equals(currentUid)) { // all
// uids
// have
// been
// crawled
String line = "All uids have been crawled, now finish! " + new Date();
System.out.println(line);
ExpUtils.mylog(CrawlTool.splitFileNameByHour(logName), line);
break;
}
page = 0; // because later there will be page++
}
if (isResultEmpty || page > this.maxPageCount) {
// empty result get or page > max page, goto next id, page = 1
System.out.println("The result is empty.");
currentUid = getNextUidTillEnd(currentUid);
page = 0;
if (currentUid == "finish" || "finish".equals(currentUid)) { // all
// uids
// have
// been
// crawled
String line = "All uids have been crawled, now finish! " + new Date();
System.out.println(line);
ExpUtils.mylog(CrawlTool.splitFileNameByHour(logName), line);
break;
}
}
// if the posts got is older than the required date, then stop next
// page, and change to the next userid
// if (isResultTooOld) {
// System.out.println("The last post is too old.");
// currentUid = getNextUidTillEnd(currentUid);
// page = 0;
//
// if (currentUid == "finish" || "finish".equals(currentUid)) {//
// all
// // uids
// // have
// // been
// // crawled
// String line = "All uids have been crawled, now finish! "
// + new Date();
// System.out.println(line);
// ExpUtils.mylog(CrawlTool.splitFileNameByHour(logName), line);
// break;
// }
//
// }
page++;
// prepare the parameter map
Map<String, String> map = getParaMap(currentUid, page, count);
System.out.println(map.toString());
this.crawledTimes++;
System.out.println("Crawling for the times: " + this.crawledTimes);
// sleep interval
CrawlTool.sleep(this.interval, this.logSign);
// try to crawl
try {
String line =
"uid="
+ currentUid
+ ", page="
+ page
+ ", count="
+ count
+ ", Crawl times: "
+ this.crawledTimes;
ExpUtils.mylog(CrawlTool.splitFileNameByHour(logName), line);
// try to get the data
String result = ExpUtils.crawlData(api, map, currentKey);
// result = StringEscapeUtils.unescapeJava(result);
// init and set the flags
isResultEmpty = false;
isError = false;
if ("[]".equals(result)) {
isResultEmpty = true;
} else {
isResultEmpty = false;
}
if (result.startsWith("{\"error\"")) {
isError = true;
} else {
isError = false;
}
// if (isResultTooOld(result)) {
// isResultTooOld = true;
// } else {
// isResultTooOld = false;
// }
if (!isResultEmpty && !isError) {
ExpUtils.mylogJson(CrawlTool.splitFileNameByHour(JsonlogName), result);
} else {
// something wrong
ExpUtils.mylog(CrawlTool.splitFileNameByHour(logName), result);
}
} catch (WeiboException e) {
e.printStackTrace();
ExpUtils.mylog(CrawlTool.splitFileNameByHour(logName), e.getError());
continue;
}
}
}