Really do not consider 'resource' WARC record for all operations

openzim · Mar 18, 2024 · 4068d85 · 4068d85
1 parent 60d174b
commit 4068d85
Show file tree

Hide file tree

Showing 2 changed files with 10 additions and 0 deletions.
diff --git a/src/warc2zim/converter.py b/src/warc2zim/converter.py
@@ -292,6 +292,12 @@ def iter_all_warc_records(self):
     def gather_information_from_warc(self):
         main_page_found = False
         for record in iter_warc_records(self.inputs):
+
+            # only response records can be considered as main_path and as existing ZIM
+            # path
+            if record.rec_type not in ("response", "revisit"):
+                continue
+
             url = get_record_url(record)
             normalized_url = normalize(url)
 
@@ -500,6 +506,10 @@ def is_self_redirect(self, record, url):
         return normalize(url) == normalize(location)
 
     def add_items_for_warc_record(self, record):
+
+        if record.rec_type not in ("response", "revisit"):
+            return
+
         url = get_record_url(record)
         normalized_url = normalize(url)
         if not url:

diff --git a/tests/data/example-resource.warc.gz b/tests/data/example-resource.warc.gz