[topstarnews] Fix article extraction

bradenhilton · Oct 17, 2024 · 3d6ae89 · 3d6ae89
1 parent 67c0d65
commit 3d6ae89
Show file tree

Hide file tree

Showing 2 changed files with 5 additions and 25 deletions.
diff --git a/extractor/topstarnews.py b/extractor/topstarnews.py
@@ -62,28 +62,10 @@ def metadata(self, page):
                         '"',
                     )[0],
                 )
-                or text.parse_datetime(
-                    text.extr(
-                        page,
-                        '<i class="fa fa-clock-o fa-fw"></i>',
-                        "</li>",
-                    )
-                    .strip()
-                    .split(" ", maxsplit=1)[1],
-                    format="%Y.%m.%d %H:%M",
-                    utcoffset=9,
-                )
-            ),
-            "author": text.extr(
-                page,
-                '<i class="fa fa-user-o fa-fw"></i>',
-                "</li>",
-            ).strip(),
-            "views": text.parse_int(
-                text.extr(page, '<i class="fa fa-desktop fa-fw"></i>', "</li>").strip().split(" ", maxsplit=1)[1],
             ),
+            "author": text.extr(page, ' name="author" content="', '"').strip().replace(" 기자", ""),
             "post_id": self.post_id,
-            "post_url": self.url,
+            "post_url": self.post_url,
         }
         if ' name="keywords" content="' in page:
             data["tags"] = text.extr(page, ' name="keywords" content="', '"').split(",")
@@ -97,7 +79,7 @@ def items(self):
 
         yield Message.Directory, data
 
-        article_body = text.extr(page, ' itemprop="articleBody">', '<div id="article-sns2"')
+        article_body = text.extr(page, 'itemprop="articleBody"', "</article>")
 
         images = [
             text.extr(figure, "<img", ">")

diff --git a/test/results/topstarnews.py b/test/results/topstarnews.py
@@ -15,8 +15,7 @@
         "date": "dt:2024-09-11 10:54:00",
         "title": "레드벨벳 웬디, ‘리본 하트 해달랬더니 근육 몽몽이가 돼버린 와니’ (웬디의 영스트리트 출근길)",
         "tags": ["웬디", "WENDY", "영스트리트", "출근", "퇴근", "프리뷰"],
-        "author": "최규석 기자",
-        "views": int,
+        "author": "최규석",
         "post_id": "15543685",
         "post_url": "https://www.topstarnews.net/news/articleView.html?idxno=15543685",
     },
@@ -29,8 +28,7 @@
         "#count": 1,
         "date": "dt:2012-04-24 06:45:00",
         "title": "걸스데이(Girls Day) 혜리, '남자들은 다 똑같아!' 깜찍한 무대 …MBC MUSIC 쇼 챔피언 생방송 현장",
-        "author": "최규석 기자",
-        "views": int,
+        "author": "최규석",
         "post_id": "30789",
         "post_url": "https://www.topstarnews.net/news/articleView.html?idxno=30789",
     },