From d6587b933078b4d0e54fcdc4642333105a10c1f6 Mon Sep 17 00:00:00 2001
From: bacon <rik.slava@gmail.com>
Date: Tue, 4 Mar 2025 00:13:03 +0300
Subject: [PATCH] added rewrite guid link

---
 proxy/rss_proxy.py | 41 ++++++++++++++++++++++++++---------------
 1 file changed, 26 insertions(+), 15 deletions(-)

diff --git a/proxy/rss_proxy.py b/proxy/rss_proxy.py
index 5ea4aec..754ab03 100755
--- a/proxy/rss_proxy.py
+++ b/proxy/rss_proxy.py
@@ -3,9 +3,9 @@ import os
 import requests
 import redis
 import xml.etree.ElementTree as ET
+import re
 from flask import request, Response
 
-
 PROXY_URL = os.getenv("PROXY_URL")
 REDIS_URL = os.getenv("REDIS_URL", "redis://localhost:6379/0")
 CACHE_TTL = int(os.getenv("CACHE_TTL", 3600))
@@ -13,10 +13,16 @@ CACHE_TTL = int(os.getenv("CACHE_TTL", 3600))
 rdb = redis.from_url(REDIS_URL)
 
 
+def extract_viewtopic_link(description):
+    """Search viewtopic.php in description"""
+    match = re.search(r'href="(http://tapochek\.net/viewtopic\.php\?t=\d+)"', description)
+    return match.group(1) if match else None
+
+
 def init_proxy(app):
     @app.route("/proxy")
     def proxy():
-        """Proxy RSS feed with per-item caching."""
+        """Proxy RSS feed with per-item caching and GUID replacement."""
         raw_query = request.query_string.decode()
         if raw_query.startswith("url="):
             url = urllib.parse.unquote(raw_query[4:])
@@ -24,7 +30,6 @@ def init_proxy(app):
             return "Missing URL", 400
 
         try:
-            # Получаем ленту
             proxies = {"http": PROXY_URL, "https": PROXY_URL} if PROXY_URL else None
             r = requests.get(url, timeout=10, proxies=proxies)
             _encode = r.apparent_encoding.lower()
@@ -32,7 +37,6 @@ def init_proxy(app):
             xml_data = r.text.replace(f'<?xml version="1.0" encoding="{_encode}"?>',
                                       '<?xml version="1.0" encoding="UTF-8"?>')
 
-            # Разбираем XML
             root = ET.fromstring(xml_data)
             items = root.findall(".//item")
 
@@ -40,19 +44,26 @@ def init_proxy(app):
             new_items = []
 
             for item in items:
-                guid = item.find("guid").text if item.find("guid") is not None else None
-                if guid:
-                    cache_key = f"rss:item:{guid}"
-                    cached_item = rdb.get(cache_key)
+                guid = item.find("guid")
+                if guid is None or guid.get("isPermaLink") == "true":
+                    continue
 
-                    if cached_item:
-                        cached_items.append(cached_item.decode())
-                    else:
-                        item_str = ET.tostring(item, encoding="unicode")
-                        rdb.setex(cache_key, CACHE_TTL, item_str)
-                        new_items.append(item_str)
+                cache_key = f"rss:item:{guid.text}"
+                cached_item = rdb.get(cache_key)
+
+                if cached_item:
+                    cached_items.append(cached_item.decode())
+                else:
+                    description = item.find("description")
+                    if description is not None:
+                        new_guid = extract_viewtopic_link(description.text)
+                        if new_guid:
+                            guid.text = new_guid
+
+                    item_str = ET.tostring(item, encoding="unicode")
+                    rdb.setex(cache_key, CACHE_TTL, item_str)
+                    new_items.append(item_str)
 
-            # Собираем финальный RSS
             final_items = cached_items + new_items
             response_xml = f'<?xml version="1.0" encoding="UTF-8"?><rss><channel>{"".join(final_items)}</channel></rss>'