Upload WarOnline_Chat.py

WarOnline_Chat.py

@@ -6,9 +6,9 @@ import re
 from bs4 import BeautifulSoup
 import urllib.request as urllib
 import warnings
-import schedule
+#import schedule
 import time
-from tqdm import tqdm
+#from tqdm import tqdm
 warnings.filterwarnings("ignore")
 
 # Define the login URL and the thread URL
@@ -16,9 +16,6 @@ login_url = 'https://waronline.org/fora/index.php?login/login'
 thread_url = 'https://waronline.org/fora/index.php?threads/warbot-playground.17636/'
 post_url = "https://waronline.org/fora/index.php?threads/warbot-playground.17636/add-reply"
 
-# Sending the message (change that!)
-message = "Test"
-
 # Define the login credentials
 username = 'WarBot'
 password = 'naP2tion'
@@ -26,11 +23,15 @@ password = 'naP2tion'
 # Start a session to persist the login cookie across requests
 session = requests.Session()
 
-
 def compare_pages(url1, url2):
     #Compares 2 pages and returns True if they are the same
     return urllib.urlopen(url1).geturl() == urllib.urlopen(url2).geturl()
 
+def remove_non_english_russian_chars(s):
+    # Regular expression to match all characters that are not in English or Russian
+    pattern = '[^0-9A-Za-zА-Яа-яЁё(),.!?"\s-]'
+    # Replace all matched characters with an empty string
+    return re.sub(pattern, '', s)
 def login(username=username, password=password, thread_url=thread_url):
     # Log-In to the forum and redirect to thread
 
@@ -61,6 +62,7 @@ def post(message=message, thread_url=thread_url, post_url=post_url, quoted_by=""
 
     if quoted_by:
         message = f'[QUOTE="{quoted_by}, post: {quote_source}"]{quote_text}[/QUOTE]{message}'
+        #message = f'[QUOTE="{quoted_by}, data-source=post: {quote_source}"]{quote_text}[/QUOTE]{message}'
         # optionally add @{quoted_by} to indent the quoter
 
     # Retrieve the thread page HTML
@@ -92,7 +94,7 @@ def post(message=message, thread_url=thread_url, post_url=post_url, quoted_by=""
 
     print('Post submitted successfully.')
 
-def allQuotesFor(thread_url=thread_url, username=username, startingPage=1):
+def getMessages(thread_url=thread_url, quotedUser="", startingPage=1):
     # Returns all the quotes for #username in the specific multi-page thread url
     allquotes =[]
 
@@ -102,10 +104,13 @@ def allQuotesFor(thread_url=thread_url, username=username, startingPage=1):
     # Initial values for messangerName and the message ID
     messengerName = ""
     messageID = ""
+    quotedID = ""
 
     # Patterns to search in the last quote.
     namePattern = re.compile('data-lb-caption-desc="(.*?) ·')
     messageIDPattern = re.compile('data-lb-id="(.*?)"')
+    quotedIDPattern = re.compile('data-source="(.*?)"')
+    quotedNamePattern = re.compile('data-quote="(.*?)"')
 
     while not lastPage:
         response = requests.get(thread_url + 'page-' + str(page))
@@ -118,30 +123,44 @@ def allQuotesFor(thread_url=thread_url, username=username, startingPage=1):
             soup = BeautifulSoup(html_content, 'html.parser')
 
             # Find all the message in the thread page
-            messageData = soup.find_all('div', {'class': 'message-userContent'})
+            messageData = soup.find_all('div', {'class': 'message-userContent lbContainer js-lbContainer'})
 
             for data in messageData:
-                if username in data.text:
-                    try:
-                        # Get the messager username
-                        matchName = namePattern.search(str(data))
-                        if matchName:
-                            messengerName = matchName.group(1)
-
-                        # Get the message ID
-                        matchID = messageIDPattern.search(str(data))
-                        if matchID:
-                            messageID = matchID.group(1)
-
-                        if 'Click' in data.text:
-                            reply = data.text.split("Click to expand...")[-1].replace('\n', ' ').strip()
-                        elif "Нажмите" in data.text:
-                            reply = data.text.split("Нажмите для раскрытия...")[-1].replace('\n', ' ').strip()
-
-                        allquotes.append({'reply': reply, 'messengerName': messengerName, 'messageID': messageID})
-
-                    except:
-                        continue # There was no text in quote, move next
+                try:
+                    # Get the messager username
+                    matchName = namePattern.search(str(data))
+                    if matchName:
+                        messengerName = matchName.group(1)
+
+                    # Get the quoted ID
+                    matchID = quotedIDPattern.search(str(data))
+                    if matchID:
+                        quotedID = matchID.group(1)
+
+                    # Get the message ID
+                    matchID = messageIDPattern.search(str(data))
+                    if matchID:
+                        messageID = matchID.group(1)
+
+                    matchQuotedName = quotedNamePattern.search(str(data))
+                    if matchQuotedName:
+                        quotedName = matchQuotedName.group(1)
+                    if quotedUser and (quotedUser != quotedName):
+                        continue
+
+                    # Make sure that the messages have a quote inside
+                    blockquote = data.find('blockquote')
+                    if blockquote:
+                        # Extract the text
+                        text = data.find('div', {'class': 'bbWrapper'})
+                        for bq in text.find_all('blockquote'):
+                            bq.extract()
+                        reply = text.get_text().replace('\n', ' ').strip()
+
+                        allquotes.append({'reply': reply, 'messengerName': messengerName, 'messageID': messageID, 'quotedID': quotedID})
+
+                except:
+                    continue # There was no text in quote, move next
 
             #check if that is not a last page
             if not compare_pages(thread_url + 'page-' + str(page), thread_url + 'page-' + str(page + 1)):
@@ -153,45 +172,67 @@ def allQuotesFor(thread_url=thread_url, username=username, startingPage=1):
 
     return allquotes
 
+# Core Engine of the Client
 def WarOnlineBot():
-    # Get All Quotes
-    quotes = allQuotesFor(thread_url=thread_url, username=username, startingPage=1)
-    repliedMessageIDs = []
-    for quote in quotes:
-        replies = allQuotesFor(thread_url=thread_url, username=quote['messengerName'], startingPage=1)
-        for reply in replies: # Remove all non-relevant and previously answered replies
-
-            if reply['messengerName']!=username or reply['messageID'] in repliedMessageIDs:
-                replies.remove(reply)
-            else:
-                repliedMessageIDs.append(reply['messageID'])
-                repliedMessageIDs = list(set(repliedMessageIDs))
-            # There is still some strange error. Will take a look later
 
-        message = ""
-        if not replies: # Meaning that there are no answers
+    login(username=username, password=password, thread_url=thread_url)
+    #print("logged in")
+
+    # All messages (with quotes) by ALL users:
+    allMessages = getMessages(thread_url=thread_url, quotedUser='', startingPage=1)
+
+    # IDs of the quoted messages, replied by the bot:
+    messages_by_bot_IDs = []
+
+    for msg in allMessages:
+        # Set a list of replied messages IDs
+        if msg['messengerName'] == username: #message posted by the WarBot
+            messages_by_bot_IDs.append(msg['quotedID'].split(': ')[-1])
+    # remove empty and repeated elements
+    messages_by_bot_IDs = list(set([elem for elem in messages_by_bot_IDs if elem]))
+
+    # All messages (with quotes) sent _FOR_ the Bot:
+    messagesForBot = getMessages(thread_url=thread_url, quotedUser=username, startingPage=1)
+
+    # IDs of the messages, quoting the bot:
+    messages_for_bot_IDs = []
+
+    for msg in messagesForBot:
+        # Set a list of posted message IDs
+        messages_for_bot_IDs.append(msg['messageID'].split('-')[-1])
+    # remove empty elements
+    messages_for_bot_IDs = [elem for elem in messages_for_bot_IDs if elem]
+
+    # Filter to leave just the unanswered messages IDs:
+    messages_for_bot_IDs = [ID for ID in messages_for_bot_IDs if ID not in messages_by_bot_IDs]
+
+    # Reply the unanswered messages:
+    for msg in messagesForBot:
+        if msg['messageID'].split('-')[-1] in messages_for_bot_IDs:
+
+            quote = remove_non_english_russian_chars(msg['reply'])
+            message = ""
+
             while not message:
-                message = WarClient.getReply(message=quote['reply'])
-            print('Quote: ', quote['reply'])
-            print('Reply: ',message)
-            post(message=message, thread_url=thread_url, post_url=post_url, quoted_by=quote['messengerName'],
-                 quote_text=quote['reply'])
-            print('posted the message to the forum')
-            time.sleep(5) # Standby time for server load release
+                message = WarClient.getReply(message=quote)
+
+            print('Quote: ', quote)
+            print('Reply: ', message)
+
+            login(username=username, password=password, thread_url=thread_url)
+            time.sleep(1)
+            post(message=message, thread_url=thread_url, post_url=post_url, quoted_by=msg['messengerName'], quote_text=quote, quote_source=msg['messageID'])
+
+            time.sleep(10)  # Standby time for server load release
 
 
 if __name__ == '__main__':
-    timeout = 2 # min
-    #post(message=message, thread_url=thread_url, post_url=post_url,quoted_by='Василий Пупкин',quote_text='Testing the XenForo response mechanism')
+    timeout = 5 # min
 
     # Start the scheduler
     while True:
-        login(username=username, password=password, thread_url=thread_url)
-        print("logged in")
         WarOnlineBot()
-        p_bar = tqdm(range(60 * timeout))
 
-        for i in p_bar:
-            p_bar.update(1)
-            p_bar.refresh()
-            time.sleep(1)
+        timer = range(60 * timeout)
+        for t in timer:
+            time.sleep(1)