feat(rss): 适配refresh rss

hsuyelin · Jan 10, 2024 · 151b40c · 151b40c
1 parent 30c6def
commit 151b40c
Show file tree

Hide file tree

Showing 3 changed files with 90 additions and 22 deletions.
diff --git a/app/rsschecker.py b/app/rsschecker.py
@@ -1,3 +1,4 @@
+import re
 import json
 import time
 import traceback
@@ -478,6 +479,48 @@ def __parse_userrss_result(self, taskinfo):
                     ExceptionUtils.exception_traceback(err)
                     log.error(f"【RssChecker】任务 {task_name} RSS地址 {rss_url} 获取的订阅报文无法解析：{str(err)}")
                     continue
+            elif rss_parser.get("type") == "FRESH_XML":
+                try:
+                    namespaces = {"media": "http://search.yahoo.com/mrss/"}
+                    for prefix, uri in namespaces.items():
+                        etree.register_namespace(prefix, uri)
+                    result_tree = etree.XML(ret.text.replace('\n', '').replace('\t', '').encode("utf-8"))
+                    item_list = result_tree.xpath(rss_parser_format.get("list")) or []
+                    for item in item_list:
+                        rss_item = {}
+                        for key, attr in rss_parser_format.get("item", {}).items():
+                            item_html = etree.tostring(item, encoding='unicode')
+                            if key == "enclosure":
+                                enclosure_matches = re.findall(r'magnet:\?xt=urn:btih:(\w+)', item_html)
+                                if enclosure_matches and len(enclosure_matches) >= 1:
+                                    value = f"magnet:?xt=urn:btih:{enclosure_matches[0]}"
+                                    rss_item.update({key: value})
+                                else:
+                                    continue
+                            elif key == "size":
+                                size_matches = re.findall(r'\b(\d+(\.\d+)?)\s*([KMGTP]i?B)\b', item_html)
+                                if size_matches and len(size_matches[0]) >= 3:
+                                    size_value = size_matches[0][0]
+                                    size_unit = size_matches[0][2]
+                                    value = f"{size_value} {size_unit}"
+                                    rss_item.update({key: value})
+                                else:
+                                    continue
+                            else:
+                                if attr.get("path"):
+                                    values = item.xpath(attr.get("path", ""), namespaces=attr.get("namespaces", {}))
+                                    rss_item.update({key: values[0]})
+                                elif attr.get("value"):
+                                    values = attr.get("value")
+                                    rss_item.update({key: values[0]})
+                                else:
+                                    continue
+                        rss_item.update({"address_index": i + 1})
+                        rss_result.append(rss_item)
+                except Exception as err:
+                    ExceptionUtils.exception_traceback(err)
+                    log.error(f"【RssChecker】任务 {task_name} RSS地址 {rss_url} 获取的订阅报文无法解析：{str(err)}")
+                    continue
             elif rss_parser.get("type") == "JSON":
                 try:
                     result_json = json.loads(ret.text)

diff --git a/scripts/sqls/init_userrss_v3.sql b/scripts/sqls/init_userrss_v3.sql
@@ -75,26 +75,50 @@ INSERT OR REPLACE INTO "CONFIG_RSS_PARSER" ("ID", "NAME", "TYPE", "FORMAT", "PAR
     }
 }', 'api_key={TMDBKEY}&language=zh-CN', '', 'Y');
 INSERT OR REPLACE INTO "CONFIG_RSS_PARSER" ("ID", "NAME", "TYPE", "FORMAT", "PARAMS", "NOTE", "SYSDEF") VALUES ('5', 'Nyaa', 'XML', '{
-    "list": "//channel/item",
-    "item": {
-    "title": {
-        "path": ".//title/text()"
-    },
-    "enclosure": {
-         "path": ".//link/text()"
-    },
-    "link": {
-        "path": ".//guid/text()"
-    },
-    "date": {
-        "path": ".//pubDate/text()"
-    },
-    "description": {
-        "path": ".//description/text()"
-    },
-    "size": {
-        "path": "size/text()",
-        "namespaces": "https://nyaa.si/xmlns/nyaa"
+    "list":"//channel/item",
+    "item":{
+        "title":{
+            "path":".//title/text()"
+        },
+        "enclosure":{
+            "path":".//link/text()"
+        },
+        "link":{
+            "path":".//guid/text()"
+        },
+        "date":{
+            "path":".//pubDate/text()"
+        },
+        "description":{
+            "path":".//description/text()"
+        },
+        "size":{
+            "path":"size/text()",
+            "namespaces":"https://nyaa.si/xmlns/nyaa"
+        }
     }
+}', '', '', 'Y');
+INSERT OR REPLACE INTO "CONFIG_RSS_PARSER" ("ID", "NAME", "TYPE", "FORMAT", "PARAMS", "NOTE", "SYSDEF") VALUES ('6', 'FreshRSS', 'FRESH_XML', '{
+    "list":"//channel/item",
+    "item":{
+        "title":{
+            "path":".//title/text()"
+        },
+        "enclosure":{
+            "path":".//media:content[@type=''application/x-bittorrent'']/@url",
+            "namespaces": {"media": "http://search.yahoo.com/mrss/"}
+        },
+        "link":{
+            "path":".//link/text()"
+        },
+        "date":{
+            "path":".//pubDate/text()"
+        },
+        "description":{
+            "path":".//description/text()"
+        },
+        "size":{
+            "path":".//description/a[contains(@href, ''#'')]/following-sibling::text()[1]"
+        }
     }
 }', '', '', 'Y');
diff --git a/web/templates/rss/rss_parser.html b/web/templates/rss/rss_parser.html
@@ -108,8 +108,9 @@ <h5 class="modal-title" id="modal-filterrule-title">新增解析器</h5>
                                                    title="支持XML及JSON，需要分别按XPath及JsonPath的语法维护解析格式"
                                                    data-bs-toggle="tooltip">?</span></label>
               <select class="form-select" id="rssparser_type">
-                <option value="XML" selected>XML</option>
-                <option value="JSON">JSON</option>
+                  <option value="XML" selected>XML</option>
+                  <option value="JSON">JSON</option>
+                  <option value="FRESH_XML">FRESH_XML</option>
               </select>
             </div>
           </div>