unescape items in text.split_html()

2021-03-29 02:12:29 +02:00
parent 36291176bc
commit 387fe415d5
4 changed files with 13 additions and 9 deletions
--- a/gallery_dl/text.py
+++ b/gallery_dl/text.py
@@ -27,11 +27,12 @@ def remove_html(txt, repl=" ", sep=" "):
    return txt.strip()


-def split_html(txt, sep=None):
-    """Split input string by html-tags"""
+def split_html(txt):
+    """Split input string by HTML tags"""
    try:
        return [
-            x.strip() for x in HTML_RE.split(txt)
+            unescape(x).strip()
+            for x in HTML_RE.split(txt)
            if x and not x.isspace()
        ]
    except TypeError: