Java 爬取京东商品评论信息
public static void downJDProductComment() throws IOException { String url = "https://club.jd.com/comment/productPageComments.action?productId=100011199522&score=0&sortType=5&page=3&pageSize=10"; CloseableHttpClient httpclient = HttpClients.createDefault(); HttpGet httpGet = new HttpGet(url); httpGet.setHeader("user-agent", "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/70.0.3538.102 Safari/537.36"); CloseableHttpResponse response = httpclient.execute(httpGet); // 获取响应状态码 int statusCode = response.getStatusLine().getStatusCode(); try { HttpEntity entity = response.getEntity(); // 如果状态响应码为200,则获取html实体内容或者json文件 if (statusCode == 200) { String html = EntityUtils.toString(entity, Consts.UTF_8); JSONObject jsonObject = new JSONObject(html); String comments = jsonObject.getString("comments"); JSONArray jsonArray = new JSONArray(comments); for (int i = 0; i < jsonArray.length(); i++) { System.out.println(i + jsonArray.getJSONObject(i).getString("content")); } // 消耗掉实体 EntityUtils.consume(response.getEntity()); } else { // 消耗掉实体 EntityUtils.consume(response.getEntity()); } } catch (JSONException e) { e.printStackTrace(); } finally { response.close(); } }