protected Book bookInfo(Element bookIcon) {
Document document = null;
String productId = bookIcon.attr("data-product-id");
try {
URL url = new URL("http://www.labirint.ru/books/" + productId);
document = Jsoup.parse(url, 5000);
} catch (MalformedURLException ex) {
} catch (IOException ex) {
}
Book book = new Book();
Element bookElement = document.getElementById("product");
String[] strs;
String str;
book.setProductId(Integer.parseInt(productId));
try {
String name;
name = bookElement.getElementById("product-title").getElementsByTag("h1").text().trim();
book.setName((name.length() <= 150) ? name : name.substring(0, 149));
} catch (Exception ex) {
book.setName("");
}
try {
String author = "";
String editor = "";
Elements authorElements =
bookElement.getElementById("product-specs").getElementsByClass("authors");
Iterator<Element> iterator = authorElements.iterator();
while (iterator.hasNext()) {
str = iterator.next().text();
if (str.contains("Автор")) {
strs = str.split(":");
if (strs.length == 2) author = strs[1].replaceAll("\"", "").trim();
else author = strs[0].replaceAll("\"", "").trim();
}
if (str.contains("Редактор")) {
strs = str.split(":");
if (strs.length == 2) editor = strs[1].replaceAll("\"", "").trim();
else editor = strs[0].replaceAll("\"", "").trim();
}
}
book.setAuthor((author.length() <= 100) ? author : author.substring(0, 99));
book.setEditor((editor.length() <= 100) ? editor : editor.substring(0, 99));
} catch (Exception ex) {
book.setAuthor("");
book.setEditor("");
}
try {
String publisherAndYear = "";
str =
bookElement.getElementById("product-specs").getElementsByClass("publisher").get(0).text();
strs = str.split(":");
if (strs.length == 2) publisherAndYear = strs[1].replaceAll("\"", "").trim();
else publisherAndYear = strs[0].replaceAll("\"", "").trim();
book.setPublisherAndYear(
(publisherAndYear.length() <= 200)
? publisherAndYear
: publisherAndYear.substring(0, 199));
} catch (Exception ex) {
book.setPublisherAndYear("");
}
try {
if (bookElement
.getElementById("product-specs")
.getElementsByClass("buying-pricenew-val-number")
.size()
> 0)
str =
bookElement
.getElementById("product-specs")
.getElementsByClass("buying-pricenew-val-number")
.get(0)
.text();
else
str =
bookElement
.getElementById("product-specs")
.getElementsByClass("buying-price-val-number")
.get(0)
.text();
book.setPrice(Double.parseDouble(str.length() <= 10 ? str : str.substring(0, 9)));
} catch (Exception ex) {
book.setPrice(null);
}
try {
String isbn = "";
str = bookElement.getElementById("product-specs").getElementsByClass("isbn").get(0).text();
strs = str.split(":");
if (strs.length == 2) isbn = strs[1].replaceAll("\"", "").trim();
else isbn = strs[0].replaceAll("\"", "").trim();
book.setIsbn((isbn.length() <= 20) ? isbn : isbn.substring(0, 19));
} catch (Exception ex) {
book.setIsbn("");
}
try {
String pagesCount = "";
str = bookElement.getElementById("product-specs").getElementsByClass("pages2").get(0).text();
strs = str.split(":");
if (strs.length == 2) pagesCount = strs[1].replaceAll("\"", "").trim();
else pagesCount = strs[0].replaceAll("\"", "").trim();
book.setPagesCount((pagesCount.length() <= 30) ? pagesCount : pagesCount.substring(0, 29));
} catch (Exception ex) {
book.setPagesCount("");
}
try {
String decor = "";
Document decorHtml =
Jsoup.parse(new URL("http://www.labirint.ru/ajax/design/" + productId), 5000);
book.setDecor(
(decorHtml.text().length() <= 400)
? decorHtml.text()
: decorHtml.text().substring(0, 399));
} catch (Exception ex) {
book.setDecor("");
}
try {
String weight = "";
str = bookElement.getElementById("product-specs").getElementsByClass("weight").get(0).text();
strs = str.split(":");
if (strs.length == 2) weight = strs[1].replaceAll("\"", "").trim();
else weight = strs[0].replaceAll("\"", "").trim();
book.setWeight((weight.length() <= 20) ? weight : weight.substring(0, 19));
} catch (Exception ex) {
book.setWeight("");
}
try {
String dimensions = "";
str =
bookElement
.getElementById("product-specs")
.getElementsByClass("dimensions")
.get(0)
.text();
strs = str.split(":");
if (strs.length == 2) dimensions = strs[1].replaceAll("\"", "").trim();
else dimensions = strs[0].replaceAll("\"", "").trim();
book.setDimensions((dimensions.length() <= 30) ? dimensions : dimensions.substring(0, 29));
} catch (Exception ex) {
book.setDimensions("");
}
try {
if (bookElement.getElementById("fullannotation") != null)
book.setAnnotation(
(bookElement.getElementById("fullannotation").text().length() <= 2000)
? bookElement.getElementById("fullannotation").text()
: bookElement.getElementById("fullannotation").text().substring(0, 1999));
else if (bookElement.getElementById("product-about") != null)
book.setAnnotation(
(bookElement.getElementById("product-about").text().length() <= 2000)
? bookElement.getElementById("product-about").text()
: bookElement.getElementById("product-about").text().substring(0, 1999));
else
book.setAnnotation(
(bookElement.getElementById("smallannotation").text().length() <= 2000)
? bookElement.getElementById("smallannotation").text()
: bookElement.getElementById("smallannotation").text().substring(0, 1999));
} catch (Exception ex) {
book.setAnnotation("");
}
try {
String imageUrl = "";
if (document.getElementsByAttributeValue("property", "og:image") != null)
imageUrl =
document.getElementsByAttributeValue("property", "og:image").get(0).attr("content");
else imageUrl = "http://img.labirint.ru/design/emptycover.png";
book.setCoverImgUrl(imageUrl);
} catch (Exception ex) {
book.setCoverImgUrl("");
}
if (document.getElementById("product-comments") != null)
book.setComments(parseComments(document.getElementById("product-comments"), book));
return book;
}
private Element fromId(String id, Element doc) {
return doc.getElementById(id);
}
/**
* 新闻的 url 格式为 http://see.xidian.edu.cn/html/news/7928.html
*
* @param id 某个新闻页面的序号
* @return 爬取该页面上的新闻信息，提取相应的信息，存到新闻bean里。如果没有爬取到新闻返回null
* @throws Exception
*/
public static ArticleItem parseNewsItem(int id) throws Exception {
// 根据后缀的数字，拼接新闻 url
String urlStr = Constant.ARTICLE_BASE_URL + id + ".html";
// 利用get请求获取字符串再解析会有小部分乱码
// String htmlStr = HttpTool.doGet(urlStr);
// Document doc = Jsoup.parse(htmlStr);
// try {
Document doc = Jsoup.connect(urlStr).timeout(10000).get();
// 去掉jsoup对html字符串加的"\n"，方便json字符串返回
doc.outputSettings().prettyPrint(false);
Element articleEle = doc.getElementById("article");
// 标题
Element titleEle = articleEle.getElementById("article_title");
String titleStr = titleEle.text();
// article_detail包括了 2016-01-15 来源: 浏览次数:177
Element detailEle = articleEle.getElementById("article_detail");
Elements details = detailEle.getElementsByTag("span");
// 发布时间
String dateStr = details.get(0).text();
// 新闻来源
String sourceStr = details.get(1).text();
// 去掉"来源:"
if (SOURCE_PREFIX.equals(sourceStr.trim())) {
sourceStr = "SeeNews";
} else {
sourceStr = sourceStr.substring(3).trim();
}
// 访问这个新闻页面，浏览次数会+1，次数是 JS 渲染的
String jsStr = HttpTool.doGet(COUNT_BASE_URL + id);
int readTimes = Integer.parseInt(jsStr.replaceAll("\\D+", ""));
// 或者使用下面这个正则方法
// String readTimesStr = jsStr.replaceAll("[^0-9]", "");
Element contentEle = articleEle.getElementById("article_content");
// 新闻主体内容
String contentStr = contentEle.toString();
// 如果用 text()方法，新闻主体内容的 html 标签会丢失
// 为了在 Android 上用 WebView 显示 html，用toString()
// String contentStr = contentEle.text();
Elements images = contentEle.getElementsByTag("img");
String[] imageUrls = new String[images.size()];
// 图片上传到七牛
// 将body中的图片地址替换为七牛的地址
for (int i = 0; i < imageUrls.length; i++) {
String origin = images.get(i).attr("src");
imageUrls[i] = ImageTool.convertUrl(id, origin);
if (!origin.equals(imageUrls[i])) {
// 只有上传图片到七牛，url 才会变化
// 不相等，才替换为七牛的url
contentStr =
contentStr.replace(
Constant.SRC_PREFIX + origin,
Constant.SRC_PREFIX + Constant.BUCKET_HOST_NAME + imageUrls[i]);
}
}
// 处理相对路径 url，不和上面的 image url 冲突
Elements hrefs = contentEle.getElementsByTag("a");
for (int i = 0; i < hrefs.size(); i++) {
String origin = hrefs.get(i).attr("href");
if (Constant.DEBUG) {
System.out.println("原始 href=" + origin);
}
String newUrl = UrlTool.dealAttachmentUrl(id, origin);
// 防止页面的附件 重复出现，替换多次
// 出现这种
// http://see.xidian.edu.cnhttp://see.xidian.edu.cn/uploads/file
if (!origin.equals(newUrl)) {
// 不相等，才替换为新的url 且url未被替换过
contentStr =
contentStr.replace(Constant.HREF_PREFIX + origin, Constant.HREF_PREFIX + newUrl);
}
}
return new ArticleItem(id, imageUrls, titleStr, dateStr, readTimes, sourceStr, contentStr);
}
