public Chapter createChapter(int id, String page) {
Chapter chapter = new Chapter(id);
chapter.setUrl(Constants.BASE_URL + getVersion() + page);
String cache = getCachePath() + page;
try {
String html = client.requestWithCache(chapter.getUrl(), cache, client.METHOD_GET, null);
Document chapterDoc = Jsoup.parse(html);
// 取出内容
Elements tables = chapterDoc.select("table");
int tableIndexOfMainBody = 1;
if (tables.size() == 1) {
tableIndexOfMainBody = 0;
}
Element table = chapterDoc.select("table").get(tableIndexOfMainBody);
Elements sectionElements = table.select("td[class=v]");
logger.debug(sectionElements.size());
for (Element tdIndex : sectionElements) {
Element tdContent = tdIndex.nextElementSibling();
String section = tdContent.text();
logger.debug(section);
chapter.addSection(section);
}
} catch (IOException e) {
logger.error(e.getMessage());
} catch (IndexOutOfBoundsException e) {
logger.error(e.getMessage());
}
return chapter;
}
public BibleCrawler() {
client = UndHttpClient.getInstance();
}