Define a custom `User-Agent` in the `_retried_request` helper

arkindex_worker/__init__.py

+import importlib.metadata
 import logging
 
 logging.basicConfig(
@@ -5,3 +6,5 @@ logging.basicConfig(
     format="%(asctime)s %(levelname)s/%(name)s: %(message)s",
 )
 logger = logging.getLogger(__name__)
+
+VERSION = importlib.metadata.version("arkindex-base-worker")

arkindex_worker/image.py

@@ -27,7 +27,7 @@ from tenacity import (
     wait_exponential,
 )
 
-from arkindex_worker import logger
+from arkindex_worker import VERSION, logger
 from arkindex_worker.utils import pluralize
 from teklia_toolbox.requests import should_verify_cert
 
@@ -41,6 +41,8 @@ DOWNLOAD_TIMEOUT = (30, 60)
 
 BoundingBox = namedtuple("BoundingBox", ["x", "y", "width", "height"])
 
+# Specific User-Agent to bypass potential server limitations
+IIIF_USER_AGENT = f"Teklia/Workers {VERSION}"
 # To parse IIIF Urls
 IIIF_URL = re.compile(r"\w+:\/{2}.+\/.+\/.+\/.+\/(?P<size>.+)\/!?\d+\/\w+\.\w+")
 # Full size of the region
@@ -339,7 +341,12 @@ def _retry_log(retry_state, *args, **kwargs):
 )
 def _retried_request(url, *args, method=requests.get, **kwargs):
     resp = method(
-        url, *args, timeout=DOWNLOAD_TIMEOUT, verify=should_verify_cert(url), **kwargs
+        url,
+        *args,
+        headers={"User-Agent": IIIF_USER_AGENT},
+        timeout=DOWNLOAD_TIMEOUT,
+        verify=should_verify_cert(url),
+        **kwargs,
     )
     resp.raise_for_status()
     return resp