Merge pull request #641 from flairNLP/handle-bsz-bug

Handle author dict Bug
flairNLP · Oct 22, 2024 · 346aae8 · 346aae8
2 parents 05cc97d + 9a997d1
commit 346aae8
Showing 1 changed file with 3 additions and 6 deletions.
diff --git a/src/fundus/publishers/de/braunschweiger_zeitung.py b/src/fundus/publishers/de/braunschweiger_zeitung.py
@@ -8,14 +8,15 @@
 from fundus.parser.utility import (
     apply_substitution_pattern_over_list,
     extract_article_body_with_selector,
+    generic_author_parsing,
     generic_date_parsing,
     generic_topic_parsing,
 )
 
 
 class BSZParser(ParserProxy):
     class V1(BaseParser):
-        _author_substitution_pattern: Pattern[str] = re.compile(r"FUNKE Mediengruppe")
+        _author_substitution_pattern: Pattern[str] = re.compile(r"FUNKE Mediengruppe|.*dpa(:|-infocom).*|^red$")
         _paragraph_selector = XPath(
             "//div[@class='article-body']//p[not(not(text()) or @rel='author' or em[@class='print'] or contains(@class, 'font-sans'))]"
         )
@@ -56,12 +57,8 @@ def topics(self) -> List[str]:
 
         @attribute
         def authors(self) -> List[str]:
-            authors = []
-            for author in self.precomputed.ld.bf_search("author", default=[]):
-                name_string = author.get("name")
-                authors.extend(re.split(r"und|,", name_string))
             return apply_substitution_pattern_over_list(
-                [author.strip() for author in authors], self._author_substitution_pattern
+                generic_author_parsing(self.precomputed.ld.bf_search("author")), self._author_substitution_pattern
             )
 
         @attribute