manubot#337 Timeout for arxiv.

manubot/cite/arxiv.py

@@ -26,8 +26,8 @@ def inspect(self, citekey):
         if not self._get_pattern().fullmatch(citekey.accession):
             return "arXiv identifiers must conform to syntax described at https://arxiv.org/help/arxiv_identifier."
 
-    def get_csl_item(self, citekey):
-        return get_arxiv_csl_item(citekey.standard_accession)
+    def get_csl_item(self, citekey, timeout: int = 3):
+        return get_arxiv_csl_item(citekey.standard_accession, timeout=timeout)
 
 
 class CSL_Item_arXiv(CSL_Item):
@@ -77,15 +77,15 @@ def get_arxiv_csl_item(arxiv_id: str):
     return get_arxiv_csl_item_oai(arxiv_id)
 
 
-def query_arxiv_api(url, params):
+def query_arxiv_api(url, params, timeout: int = 3):
     headers = {"User-Agent": get_manubot_user_agent()}
-    response = requests.get(url, params, headers=headers)
+    response = requests.get(url, params, headers=headers, timeout=timeout)
     response.raise_for_status()
     xml_tree = xml.etree.ElementTree.fromstring(response.text)
     return xml_tree
 
 
-def get_arxiv_csl_item_export_api(arxiv_id):
+def get_arxiv_csl_item_export_api(arxiv_id, timeout: int = 3):
     """
     Return csl_item item for an arXiv record.
 
@@ -105,6 +105,7 @@ def get_arxiv_csl_item_export_api(arxiv_id):
     xml_tree = query_arxiv_api(
         url="https://export.arxiv.org/api/query",
         params={"id_list": arxiv_id, "max_results": 1},
+        timeout=timeout,
     )
 
     # XML namespace prefixes
@@ -157,7 +158,7 @@ def get_arxiv_csl_item_export_api(arxiv_id):
     return csl_item
 
 
-def get_arxiv_csl_item_oai(arxiv_id):
+def get_arxiv_csl_item_oai(arxiv_id, timeout: int = 3):
     """
     Generate a CSL Item for an unversioned arXiv identifier
     using arXiv's OAI_PMH v2.0 API <https://arxiv.org/help/oa>.
@@ -174,6 +175,7 @@ def get_arxiv_csl_item_oai(arxiv_id):
             "metadataPrefix": "arXiv",
             "identifier": f"oai:arXiv.org:{arxiv_id}",
         },
+        timeout=timeout,
     )
 
     # Create dictionary for CSL Item
@@ -237,10 +239,10 @@ def remove_newlines(text):
     return re.sub(pattern=r"\n(?!\s)", repl=" ", string=text)
 
 
-def get_arxiv_csl_item_zotero(arxiv_id):
+def get_arxiv_csl_item_zotero(arxiv_id, timeout: int = 3):
     """
     Generate CSL JSON Data for an arXiv ID using Zotero's translation-server.
     """
     from manubot.cite.zotero import get_csl_item
 
-    return get_csl_item(f"arxiv:{arxiv_id}")
+    return get_csl_item(f"arxiv:{arxiv_id}", timeout=timeout)