Merge pull request #4339 from inspirehep/fix-grobid

matcher: request explicitly XML format in grobid
inspirehep · Apr 17, 2024 · 7c9cbbd · 7c9cbbd
2 parents fab1c19 + 341d633
commit 7c9cbbd
Showing 1 changed file with 5 additions and 1 deletion.
diff --git a/inspirehep/modules/workflows/tasks/actions.py b/inspirehep/modules/workflows/tasks/actions.py
@@ -1091,7 +1091,11 @@ def post_pdf_to_grobid(obj, grobid_api_path, **kwargs):
         data.update(kwargs)
         grobid_url = current_app.config["GROBID_URL"]
         try:
-            response = requests.post(urljoin(grobid_url, grobid_api_path), files=data)
+            response = requests.post(
+                urljoin(grobid_url, grobid_api_path),
+                files=data,
+                headers={'Accept': 'application/xml'}
+            )
             response.raise_for_status()
         except GROBID_EXCEPTIONS:
             LOGGER.warning("(%s) Grobid request failed due to the GROBID serivce error!" % str(obj.id))