Set User-Agent header

src/dapla_metadata/_shared/utils.py

@@ -0,0 +1,14 @@
+from importlib import metadata
+
+DAPLA_TOOLBELT_METADATA_PACKAGE_NAME = "dapla-toolbelt-metadata"
+TEAM_METADATA_CONTACT_EMAIL = "metadata@ssb.no"
+
+
+def get_app_version() -> str:
+    """Get the version of the this package."""
+    return metadata.distribution(DAPLA_TOOLBELT_METADATA_PACKAGE_NAME).version
+
+
+def get_user_agent() -> str:
+    """Get the value for the Use-Agent header."""
+    return f"{DAPLA_TOOLBELT_METADATA_PACKAGE_NAME}/{get_app_version()} ({TEAM_METADATA_CONTACT_EMAIL})"

src/dapla_metadata/datasets/statistic_subject_mapping.py

@@ -9,6 +9,7 @@
 from bs4 import BeautifulSoup
 from bs4 import ResultSet
 
+from dapla_metadata._shared.utils import get_user_agent
 from dapla_metadata.datasets.external_sources.external_sources import GetExternalSource
 from dapla_metadata.datasets.utility.enums import SupportedLanguages
 
@@ -121,7 +122,11 @@ def _fetch_data_from_external_source(self) -> ResultSet | None:
             return None
 
         try:
-            response = requests.get(str(self.source_url), timeout=30)
+            response = requests.get(
+                str(self.source_url),
+                headers={"User-Agent": get_user_agent()},
+                timeout=30,
+            )
             response.encoding = "utf-8"
             logger.debug("Got response %s from %s", response, self.source_url)
             soup = BeautifulSoup(response.text, features="xml")

src/dapla_metadata/variable_definitions/_utils/_client.py

@@ -1,3 +1,4 @@
+from dapla_metadata._shared.utils import get_user_agent
 from dapla_metadata.variable_definitions._generated.vardef_client.api_client import (
     ApiClient,
 )
@@ -40,5 +41,6 @@ def get_client(cls) -> ApiClient:
             if not cls._config:
                 cls._config = get_vardef_client_configuration()
             cls._client = ApiClient(cls._config)
+            cls._client.user_agent = get_user_agent()
         cls._client.configuration.access_token = refresh_access_token()
         return cls._client