Merge branch 'master' into remove_core_except

adbar · Oct 2, 2024 · 8cbffdd · 8cbffdd
2 parents f8807e6 + a882c61
commit 8cbffdd
Showing 1 changed file with 6 additions and 5 deletions.
diff --git a/trafilatura/core.py b/trafilatura/core.py
@@ -147,11 +147,6 @@ def bare_extraction(filecontent, url=None, no_fallback=False,  # fast=False,
 
     # load data
     try:
-        tree = load_html(filecontent)
-        if tree is None:
-            LOGGER.error('empty HTML tree: %s', url)
-            raise ValueError
-
         # regroup extraction options
         if not options or not isinstance(options, Extractor):
             options = Extractor(
@@ -164,6 +159,12 @@ def bare_extraction(filecontent, url=None, no_fallback=False,  # fast=False,
                           author_blacklist=author_blacklist, url_blacklist=url_blacklist,
                           date_params=date_extraction_params
                       )
+
+        # load the HTML tree
+        tree = load_html(filecontent)
+        if tree is None:
+            LOGGER.error('empty HTML tree: %s', url)
+            raise ValueError
 
         # quick and dirty HTML lang check
         if options.lang and (options.fast or not LANGID_FLAG):