dep

utils/strings.py

@@ -0,0 +1,29 @@
+import re
+
+
+def clean_text(text):
+    # Удаляем табуляции, неразрывные пробелы, кавычки и обратные кавычки
+    text = re.sub(r'[\t\xa0"`]+', ' ', text)
+    text = re.sub(r'\*{2,}', '', text)
+    return re.sub(r'\s+', ' ', text).strip()
+
+
+
+def parse_post(text):
+    cleaned = clean_text(text)
+    results = []
+    pattern = re.compile(r'(?:логин|login)[\s:]*([^\s\n]+)[^\n\r]*(?:пароль|password)[\s:]*([^\s\n]+)', re.IGNORECASE)
+    alt_login = re.findall(r'(?:логин|login)[\s:]*([^\s\n]+)', cleaned, re.IGNORECASE)
+    alt_pass = re.findall(r'(?:пароль|password)[\s:]*([^\s\n]+)', cleaned, re.IGNORECASE)
+
+    matches = pattern.findall(cleaned)
+    for login, password in matches:
+        if 3 <= len(login) <= 50 and 4 <= len(password) <= 300:
+            results.append((login.strip(), password.strip()))
+
+    if not results and len(alt_login) == len(alt_pass):
+        for login, password in zip(alt_login, alt_pass):
+            if 3 <= len(login) <= 50 and 4 <= len(password) <= 300:
+                results.append((login.strip(), password.strip()))
+
+    return results