chore(xmlupload): improve error handling of Timeout Error (DEV-2299)

src/dsp_tools/utils/shared.py

@@ -12,7 +12,7 @@
 import pandas as pd
 import regex
 from lxml import etree
-from requests import RequestException
+from requests import ReadTimeout, RequestException
 
 from dsp_tools.models.connection import Connection
 from dsp_tools.models.exceptions import BaseError, UserError
@@ -75,7 +75,7 @@ def try_network_action(
     Returns:
         the return value of action
     """
-
+    action_as_str = f"action='{action}', args='{args}', kwargs='{kwargs}'"
     for i in range(7):
         try:
             if args and not kwargs:
@@ -86,9 +86,16 @@ def try_network_action(
                 return action(*args, **kwargs)
             else:
                 return action()
+        except (TimeoutError, ReadTimeout):
+            msg = f"Timeout Error: Try reconnecting to DSP server, next attempt in {2 ** i} seconds..."
+            print(f"{datetime.now().isoformat()}: {msg}")
+            logger.error(f"{msg} {action_as_str} (retry-counter i={i})", exc_info=True)
+            time.sleep(2**i)
+            continue
         except (ConnectionError, RequestException):
-            print(f"{datetime.now().isoformat()}: Try reconnecting to DSP server, next attempt in {2 ** i} seconds...")
-            logger.error(f"Try reconnecting to DSP server, next attempt in {2 ** i} seconds...", exc_info=True)
+            msg = f"Network Error: Try reconnecting to DSP server, next attempt in {2 ** i} seconds..."
+            print(f"{datetime.now().isoformat()}: {msg}")
+            logger.error(f"{msg} {action_as_str} (retry-counter i={i})", exc_info=True)
             time.sleep(2**i)
             continue
         except BaseError as err:
@@ -97,9 +104,9 @@ def try_network_action(
                 in_500_range = 500 <= err.status_code < 600
             try_again_later = "try again later" in err.message
             if try_again_later or in_500_range:
-                msg = f"Try reconnecting to DSP server, next attempt in {2 ** i} seconds..."
+                msg = f"Transient Error: Try reconnecting to DSP server, next attempt in {2 ** i} seconds..."
                 print(f"{datetime.now().isoformat()}: {msg}")
-                logger.error(msg, exc_info=True)
+                logger.error(f"{msg} {action_as_str} (retry-counter i={i})", exc_info=True)
                 time.sleep(2**i)
                 continue
             else:

src/dsp_tools/utils/xml_upload.py

@@ -1027,7 +1027,7 @@ def _handle_upload_error(
     that all information about what is already in DSP
     is written into diagnostic files.
 
-    It then re-raises the original error.
+    It then quits the Python interpreter with exit code 1.
 
     Args:
         err: error that was the cause of the abort
@@ -1041,9 +1041,12 @@ def _handle_upload_error(
     Returns:
         None
     """
-
-    print("\n==========================================\nxmlupload must be aborted because of an error")
-    logger.info("xmlupload must be aborted because of an error")
+    print(
+        f"\n==========================================\n"
+        f"xmlupload must be aborted because of an error.\n"
+        f"Error message: '{err}'\n"
+    )
+    logger.error("xmlupload must be aborted because of an error", exc_info=err)
 
     # only stashed properties of resources that already exist in DSP are of interest
     stashed_xml_texts = _purge_stashed_xml_texts(
@@ -1056,7 +1059,7 @@ def _handle_upload_error(
     )
 
     if id2iri_mapping:
-        id2iri_mapping_file = f"{save_location}/{timestamp_str}_id2iri_mapping.json"
+        id2iri_mapping_file = f"{save_location}/{timestamp_str}_id2iri_mapping.json\n"
         with open(id2iri_mapping_file, "x", encoding="utf-8") as f:
             json.dump(id2iri_mapping, f, ensure_ascii=False, indent=4)
         print(f"The mapping of internal IDs to IRIs was written to {id2iri_mapping_file}")
@@ -1077,7 +1080,7 @@ def _handle_upload_error(
             )
         msg = (
             f"There are stashed text properties that could not be reapplied to the resources they were stripped from. "
-            f"They were saved to {xml_filename}."
+            f"They were saved to {xml_filename}.\n"
         )
         print(msg)
         logger.info(msg)
@@ -1096,7 +1099,7 @@ def _handle_upload_error(
             )
         msg = (
             f"There are stashed resptr properties that could not be reapplied "
-            f"to the resources they were stripped from. They were saved to {resptr_filename}"
+            f"to the resources they were stripped from. They were saved to {resptr_filename}\n"
         )
         print(msg)
         logger.info(msg)
@@ -1107,9 +1110,4 @@ def _handle_upload_error(
         print(msg)
         logger.info(msg)
 
-    if isinstance(err, KeyboardInterrupt):
-        sys.exit(1)
-    else:
-        print("The error will now be raised again:\n==========================================\n")
-        logger.info("The error will now be raised again")
-        raise err
+    sys.exit(1)

testdata/xml-data/test-data-systematic.xml

@@ -639,8 +639,6 @@
         </text-prop>
     </resource>
 
-    <!-- the API supports also MP4 files that are purely audio (MIME "audio/mp4), but there is no test file for it -->
-
     <resource label="Moviething"
               restype=":MovieThing"
               id="video_thing_1"