Update retry logic

zstash/extract.py

@@ -386,6 +386,28 @@ def multiprocess_extract(
     return failures
 
 
+def check_sizes_match(cur, tfname):
+    match: bool
+    if cur and tars_table_exists(cur):
+        logger.info(f"{tfname} exists. Checking expected size matches actual size.")
+        actual_size = os.path.getsize(tfname)
+        name_only = os.path.split(tfname)[1]
+        cur.execute(f"select size from tars where name is '{name_only}';")
+        expected_size: int = cur.fetchall()[0][0]
+        if expected_size != actual_size:
+            logger.info(
+                f"{name_only}: expected size={expected_size} != {actual_size}=actual_size"
+            )
+            match = False
+        else:
+            # Sizes match
+            match = True
+    else:
+        # Cannot access size information; assume the sizes match.
+        match = True
+    return match
+
+
 # FIXME: C901 'extractFiles' is too complex (33)
 def extractFiles(  # noqa: C901
     files: List[FilesRow],
@@ -445,35 +467,30 @@ def extractFiles(  # noqa: C901
                 hpss: str = config.hpss
             else:
                 raise TypeError("Invalid config.hpss={}".format(config.hpss))
-            tries = args.retries + 1
+            tries: int = args.retries + 1
             # Set to True to test the `--retries` option with a forced failure.
             # Then run `python -m unittest tests.test_extract.TestExtract.testExtractRetries`
-            test_retry = False
+            test_retry: bool = False
             while tries > 0:
                 tries -= 1
+                do_retrieve: bool
+
+                if not os.path.exists(tfname):
+                    do_retrieve = True
+                else:
+                    do_retrieve = not check_sizes_match(cur, tfname)
+
                 try:
                     if test_retry:
                         test_retry = False
                         raise RuntimeError
-                    if not os.path.exists(tfname):
-                        # Will need to retrieve from HPSS
+                    if do_retrieve:
                         hpss_get(hpss, tfname, cache)
-                    elif cur and tars_table_exists(cur):
-                        logger.info(
-                            f"{tfname} exists. Checking expected size matches actual size."
-                        )
-                        actual_size = os.path.getsize(tfname)
-                        name_only = os.path.split(tfname)[1]
-                        cur.execute(
-                            f"select size from tars where name is '{name_only}';"
-                        )
-                        expected_size: int = cur.fetchall()[0][0]
-                        if expected_size > actual_size:
-                            logger.info(
-                                f"{name_only}: expected size={expected_size} > {actual_size}=actual_size"
+                        if not check_sizes_match(cur, tfname):
+                            raise RuntimeError(
+                                f"{tfname} size does not match expected size."
                             )
-                            hpss_get(hpss, tfname, cache)
-                    # `get` successful or not needed: no more tries needed
+                    # `hpss_get` successful or not needed: no more tries needed
                     break
                 except RuntimeError as e:
                     if tries > 0: