猿人学第8题-验证码图文点选

猿人学Web端爬虫攻防刷题平台/猿人学第8题-验证码图文点选/README.md

@@ -1,4 +1,4 @@
-# 知识点:图像识别、去除干扰线、降噪、机器学习
+# 知识点:图像识别、去除干扰线、降噪、机器学习、headers请求顺序
 
 ## 一、解决点选坐标问题
 
@@ -22,7 +22,7 @@
 
 ![debugger](./img/3.png)
 
-由此推断
+由此推断，坐标和对应的值
 
     第1个图片坐标值可以是（0～9）
     第2个图片坐标值可以是（10～19）
@@ -34,6 +34,22 @@
     第8个图片坐标值可以是（610～619）
     第9个图片坐标值可以是（620～629）
 
+这里就可以写一个python方法作为调用
+
+    def coordinate(num):
+        coordinate_dict = {
+            1: random.randint(0, 9),
+            2: random.randint(10, 19),
+            3: random.randint(20, 29),
+            4: random.randint(300, 309),
+            5: random.randint(310, 319),
+            6: random.randint(320, 329),
+            7: random.randint(600, 609),
+            8: random.randint(610, 619),
+            9: random.randint(620, 629)
+        }
+        return coordinate_dict.get(num)
+
 ## 二、图像去噪点处理
 
 ### 选出rgp颜色最多2个，进行降噪处理
@@ -145,3 +161,26 @@
 ![debugger](./img/f-2.jpg)
 
 ## 三、图像识别/机器学习
+
+测试百度ocr、ddddocr、还尝试了自己训练模型，效果都不是很好，菜是原罪啊！最后发现科大讯飞ocr对这些生僻字识别效果还是不错的。
+
+![debugger](./img/6.png)
+
+效果
+
+![debugger](./img/7.png)
+
+## headers请求顺序
+
+    HEADERS = {
+        'Proxy-Connection': 'keep-alive',
+        'Accept': 'application/json, text/javascript, */*; q=0.01',
+        'User-Agent': 'yuanrenxue.project',
+        'X-Requested-With': 'XMLHttpRequest',
+        'Referer': 'http://match.yuanrenxue.com/match/8',
+        'Accept-Language': 'zh-CN,zh;q=0.9'
+    }
+    SESSION = requests.session()
+    SESSION.headers = HEADERS
+
+    req = SESSION.get(self.url)

猿人学Web端爬虫攻防刷题平台/猿人学第8题-验证码图文点选/main.py

@@ -3,6 +3,17 @@ import requests
 import re
 import base64
 
+HEADERS = {
+    'Proxy-Connection': 'keep-alive',
+    'Accept': 'application/json, text/javascript, */*; q=0.01',
+    'User-Agent': 'yuanrenxue.project',
+    'X-Requested-With': 'XMLHttpRequest',
+    'Referer': 'http://match.yuanrenxue.com/match/8',
+    'Accept-Language': 'zh-CN,zh;q=0.9'
+}
+SESSION = requests.session()
+SESSION.headers = HEADERS
+
 
 class YuanrenXuan(object):
 
@@ -11,7 +22,7 @@ class YuanrenXuan(object):
         self.sum_value = 0
 
     def get_task(self):
-        req = requests.get(self.url)
+        req = SESSION.get(self.url)
         text = re.findall(r'请依次点击：---<p>(.*)</p>--- <br>提示', req.json().get('html'))[0]
         text_list = text.split('</p>---<p>')
         img = re.findall(r'<img src="(.*)" alt="">', req.json().get('html'))[0]
@@ -22,16 +33,18 @@ class YuanrenXuan(object):
         return text_list
 
     def get_match(self, page, answer):
-        url = f"https://match.yuanrenxue.cn/api/match/8?page={page}&answer={answer}"
+        url = f"https://match.yuanrenxue.cn/api/match/8"
-        payload = {}
+        print(answer)
-        headers = {
+        params = (
-            'cookie': 'sessionid=iikaj9bo7vzqv4mz1xvryl13o7z98l13;'
+            ('page', str(page)),
-        }
+            ('answer', answer)
-        response = requests.request("GET", url, headers=headers, data=payload)
+        )
-        print(response.json())
+        response = SESSION.get(url, params=params)
+        return response.json()
 
     def run(self):
         num = 1
+        data_num_list = []
         while True:
             text_list = self.get_task()
             print(text_list)
@@ -43,14 +56,14 @@ class YuanrenXuan(object):
             answer = '|'.join(answer_list)
             if 'None' in answer:
                 continue
-            print(num, answer)
+            data_list = self.get_match(num, answer).get('data')
-            self.get_match(num, answer)
+            for data in data_list:
+                data_num_list.append(int(data.get('value')))
             num += 1
 
             if num == 5:
                 break
-
+        print(data_num_list)
-            exit()
 
 
 if __name__ == '__main__':

猿人学Web端爬虫攻防刷题平台/猿人学第8题-验证码图文点选/xin_fei_ocr.py

@@ -106,9 +106,9 @@ def get_result(universal_ocr, file_path):
 
 
 def run_ocr(file_path):
-    appid = "xxxx"
+    appid = "zzzz"
     apisecret = "xxxx"
-    apikey = "xxxxx"
+    apikey = "xxxx"
     universal_ocr = UniversalOcr(appid, apikey, apisecret)
     res = get_result(universal_ocr, file_path)
     return res.get('whole_text')