private MajorModel getMajorModel(String major) {
for (MajorModel model : majors.get("one")) {
if (major.equals(model.getValue())) {
return model;
}
}
for (MajorModel model : majors.get("two")) {
if (major.equals(model.getValue())) {
return model;
}
}
return null;
}
// step1
// 计算各个分类的Jd数
public void handler() {
ClassLoader classLoader = (new Chi()).getClass().getClassLoader();
InputStream stream = classLoader.getResourceAsStream(SRC_JD);
Scanner scanner = new Scanner(stream);
String line = null;
String jdLine = null;
String[] arr = null;
Set<String> words = null;
Set<String> jdWords = null;
Set<String> jds = null;
String jd = null;
boolean next = false;
MajorModel model = null;
Splitter splitter = Splitter.on(" ").omitEmptyStrings().trimResults();
while (scanner.hasNextLine()) {
line = scanner.nextLine();
if (line == null || line.trim().equals("")) {
continue;
}
if (line.startsWith("jobid:")) {
jdLine = line;
} else if (line.startsWith("majorcalssify:")) {
arr = line.split(":")[1].split(",");
jd = jdLine.split(" ")[1];
jdWords = new HashSet<String>();
jd_words.put(jd, jdWords);
for (String m : arr) {
model = getMajorModel(m);
if (model != null) { // 对专业进行验证，确保专业存在
next = true;
if (major_words.containsKey(model.getId())) {
words = major_words.get(model.getId());
} else {
words = new HashSet<String>();
if (model.isMain()) {
major_words.put(model.getId(), words);
} else {
major_words.put(model.getParent().getId(), words);
}
}
if (major_jds.containsKey(model.getId())) {
jds = major_jds.get(model.getId());
} else {
jds = new HashSet<String>();
if (model.isMain()) {
major_jds.put(model.getId(), jds);
} else {
major_jds.put(model.getParent().getId(), jds);
}
}
jds.add(jd); // get job id
} else {
next = false;
}
}
} else if (line.startsWith("corp:")) {
// ignore
// } else if (line.startsWith("title:")) {
// ignore
} else {
if (next) {
for (String word : splitter.split(line)) {
words.add(word);
jdWords.add(word);
}
jdWords.remove("intro:");
jdWords.remove("title:");
words.remove("intro:");
words.remove("title:");
}
}
}
}