基本解决加密表情包的解析问题

app/DataBase/output_pc.py

@@ -284,7 +284,11 @@ class ChildThread(QThread):
         timestamp = message[5]
         if self.output_type == Output.HTML:
             emoji_path = get_emoji(str_content, thumb=True, output_path=origin_docx_path + '/emoji')
-            emoji_path = './emoji/' + os.path.basename(emoji_path)
+            if emoji_path == "":
+                shutil.copy(f"{os.path.abspath('.')}/app/resources/icons/404.png", origin_docx_path + '/emoji/404.png')
+                emoji_path = "./emoji/404.png"
+            else:
+                emoji_path = './emoji/' + os.path.basename(emoji_path)
             if self.is_5_min(timestamp):
                 doc.write(
                     f'''{{ type:0, text: '{str_time}',is_send:0,avatar_path:''}},'''
@@ -463,7 +467,7 @@ class ChildThread(QThread):
                 self.emoji(f, message)
             elif type_ == 10000 and self.message_types.get(type_):
                 self.system_msg(f, message)
-            elif type_ == 49 and sub_type == 57:
+            elif type_ == 49 and sub_type == 57 and self.message_types.get(1):
                 self.refermsg(f,message)
         f.write(html_end)
         f.close()

app/util/emoji.py

@@ -11,7 +11,8 @@ emoji.py
 import os
 import traceback
 import xml.etree.ElementTree as ET
-
+import sqlite3
+import threading
 import requests
 
 from app.log import log, logger
@@ -43,7 +44,10 @@ def get_image_format(header):
 def parser_xml(xml_string):
     assert type(xml_string) == str
     # Parse the XML string
-    root = ET.fromstring(xml_string)
+    try:
+        root = ET.fromstring(xml_string)
+    except:
+        root = ET.fromstring(xml_string.replace("&", "&"))
     emoji = root.find('./emoji')
     # Accessing attributes of the 'emoji' element
     fromusername = emoji.get('fromusername')
@@ -61,10 +65,67 @@ def parser_xml(xml_string):
         'height': height,
         'cdnurl': cdnurl,
         'thumburl': thumburl if thumburl else cdnurl,
-        'md5': md5 if md5 else androidmd5,
+        'md5': (md5 if md5 else androidmd5).lower(),
     }
 
-@log
+def singleton(cls):
+    _instance = {}
+
+    def inner():
+        if cls not in _instance:
+            _instance[cls] = cls()
+        return _instance[cls]
+
+    return inner
+
+lock = threading.Lock()
+db_path = "./app/Database/Msg/Emotion.db"
+
+@singleton
+class Emotion:
+    def __init__(self):
+        self.DB = None
+        self.cursor: sqlite3.Cursor = None
+        self.open_flag = False
+        self.init_database()
+
+    def init_database(self):
+        if not self.open_flag:
+            if os.path.exists(db_path):
+                self.DB = sqlite3.connect(db_path, check_same_thread=False)
+                # '''创建游标'''
+                self.cursor = self.DB.cursor()
+                self.open_flag = True
+                if lock.locked():
+                    lock.release()
+
+    def get_emoji_url(self, md5: str):
+        sql = '''
+            select CDNUrl
+            from CustomEmotion
+            where md5 = ?
+        '''
+        try:
+            lock.acquire(True)
+            self.cursor.execute(sql, [md5])
+            return self.cursor.fetchone()[0]
+        except:
+            md5 = md5.upper()
+            sql = """
+                select Data
+                from EmotionItem
+                where md5 = ?
+            """
+            self.cursor.execute(sql, [md5])
+            try:
+                return self.cursor.fetchone()[0]
+            except:
+                return ""
+        finally:
+            lock.release()
+
+emoji_db = Emotion()
+
 def download(url, output_dir, name, thumb=False):
     if not url:
         return ':/icons/icons/404.png'
@@ -92,23 +153,44 @@ def get_emoji(xml_string, thumb=True, output_path=root_path) -> str:
             prefix = 'th_' if thumb else ''
             file_path = os.path.join(output_path, prefix + md5 + f)
             if os.path.exists(file_path):
-                print('表情包已存在')
+                # print('表情包已存在', file_path)
                 return file_path
         url = emoji_info['thumburl'] if thumb else emoji_info['cdnurl']
-        print("下载表情包ing:", url)
+        if not url or url == "":
-        emoji_path = download(url, output_path, md5, thumb)
+            url = emoji_db.get_emoji_url(md5)
-        return emoji_path
+        if type(url) == str and url != "":
+            print("下载表情包ing:", url)
+            emoji_path = download(url, output_path, md5, thumb)
+            return emoji_path
+        elif type(url) == bytes:
+            image_format = get_image_format(url[:8])
+            if image_format:
+                if thumb:
+                    output_path = os.path.join(output_path, 'th_' + md5 + '.' + image_format)
+                else:
+                    output_path = os.path.join(output_path, md5 + '.' + image_format)
+            else:
+                output_path = os.path.join(output_path, md5)
+            with open(output_path, 'wb') as f:
+                f.write(url)
+            print("表情包数据库加载", output_path)
+            return output_path
+        else:
+            print("！！！未知表情包数据，信息：", xml_string, emoji_info, url)
+            return ""
     except:
         logger.error(traceback.format_exc())
         return ""
 
 
 if __name__ == '__main__':
-    xml_string = '<msg><emoji fromusername = "wxid_0o18ef858vnu22" tousername = "wxid_27hqbq7vx5hf22" type="2" idbuffer="media:0_0" md5="71ce49ed3ce9e57e43e07f802983bf45" len = "352588" productid="com.tencent.xin.emoticon.person.stiker_1678703862259eb01f2ef4a313" androidmd5="71ce49ed3ce9e57e43e07f802983bf45" androidlen="352588" s60v3md5 = "71ce49ed3ce9e57e43e07f802983bf45" s60v3len="352588" s60v5md5 = "71ce49ed3ce9e57e43e07f802983bf45" s60v5len="352588" cdnurl = "http://wxapp.tc.qq.com/262/20304/stodownload?m=71ce49ed3ce9e57e43e07f802983bf45&amp;filekey=30350201010421301f020201060402535a041071ce49ed3ce9e57e43e07f802983bf45020305614c040d00000004627466730000000132&amp;hy=SZ&amp;storeid=263ffa00b000720d03274c5820000010600004f50535a1ca0c950b64287022&amp;bizid=1023" designerid = "" thumburl = "http://mmbiz.qpic.cn/mmemoticon/ajNVdqHZLLDSKTMRgM8agiadpFhKz9IJ3cD5Ra2sTROibOaShdt3D4z6PfE92WkjQY/0" encrypturl = "http://wxapp.tc.qq.com/262/20304/stodownload?m=cbaae1d847aac6389652b65562bacaa2&amp;filekey=30350201010421301f020201060402535a0410cbaae1d847aac6389652b65562bacaa20203056150040d00000004627466730000000132&amp;hy=SZ&amp;storeid=263ffa00b0008d8223274c5820000010600004f50535a17b82910b64764739&amp;bizid=1023" aeskey= "7051ab2a34442dec63434832463f45ce" externurl = "http://wxapp.tc.qq.com/262/20304/stodownload?m=960f68693454dfa64b9966ca5d70dbd3&amp;filekey=30340201010420301e020201060402535a0410960f68693454dfa64b9966ca5d70dbd3020221a0040d00000004627466730000000132&amp;hy=SZ&amp;storeid=26423dbe3000793a8720e40de0000010600004f50535a1d40c950b71be0a50&amp;bizid=1023" externmd5 = "41895664fc5a77878e2155fc96209a19" width= "240" height= "240" tpurl= "" tpauthkey= "" attachedtext= "" attachedtextcolor= "" lensid= "" emojiattr= "" linkid= "" desc= "ChEKB2RlZmF1bHQSBuWNlee6rw==" ></emoji> </msg>'
+    db_path = r"..\DataBase\Msg\Emotion.db"
-    res1 = parser_xml(xml_string)
+    xml_string = '<msg><emoji fromusername = "wxid_05rvkbftizq822" tousername = "wxid_lhbdvh3cnn4h22" type="2" androidmd5="882A3BC82838CBD4DD419F62A22EA244" androidlen="2469" aeskey="" encrypturl="" externurl="" externmd5=""></emoji></msg>'
-    print(res1, res1['md5'])
+    print(get_emoji(xml_string, thumb=False, output_path="."))
+    # ET.fromstring()
+    # print(res1, res1['md5'])
     # download(res1['cdnurl'], "./data/emoji/", res1['md5'])
     # download(res1['thumburl'], "./data/emoji/", res1['md5'], True)
-    print(get_emoji(xml_string, True))
+    # print(get_emoji(xml_string, True))
-    print(get_emoji(xml_string, False))
+    # print(get_emoji(xml_string, False))
 #     http://vweixinf.tc.qq.com/110/20403/stodownload?m=3a4d439aba02dce4834b2c54e9f15597&filekey=3043020101042f302d02016e0402534804203361346434333961626130326463653438333462326335346539663135353937020213f0040d00000004627466730000000131&hy=SH&storeid=323032313037323030373236313130303039653236646365316535316534383236386234306230303030303036653033303034666233&ef=3&bizid=1022