From db45c694b981e4edf9bc4e37462505add8a84cb1 Mon Sep 17 00:00:00 2001
From: Andreas Motl <andreas.motl@crate.io>
Date: Thu, 12 Sep 2024 17:07:36 +0200
Subject: [PATCH] MongoDB: Avoid URL object <-> string conversions on a few
 spots

---
 cratedb_toolkit/api/main.py        |  7 +++++--
 cratedb_toolkit/io/mongodb/api.py  | 22 +++++++++++++++-------
 cratedb_toolkit/io/mongodb/copy.py |  6 +++---
 3 files changed, 23 insertions(+), 12 deletions(-)

diff --git a/cratedb_toolkit/api/main.py b/cratedb_toolkit/api/main.py
index c68a016..32bcece 100644
--- a/cratedb_toolkit/api/main.py
+++ b/cratedb_toolkit/api/main.py
@@ -146,7 +146,7 @@ def load_table(self, resource: InputOutputResource, target: TableAddress, transf
                 return mongodb_relay_cdc(str(source_url_obj), target_url, progress=True)
             else:
                 return mongodb_copy_generic(
-                    str(source_url_obj),
+                    source_url_obj,
                     target_url,
                     transformation=transformation,
                     progress=True,
@@ -157,7 +157,10 @@ def load_table(self, resource: InputOutputResource, target: TableAddress, transf
 
 
 def mongodb_copy_generic(
-    source_url: str, target_url: str, transformation: t.Union[Path, None] = None, progress: bool = False
+    source_url: t.Union[str, URL],
+    target_url: t.Union[str, URL],
+    transformation: t.Union[Path, None] = None,
+    progress: bool = False,
 ):
     from cratedb_toolkit.io.mongodb.api import mongodb_copy
 
diff --git a/cratedb_toolkit/io/mongodb/api.py b/cratedb_toolkit/io/mongodb/api.py
index de8c31d..b201731 100644
--- a/cratedb_toolkit/io/mongodb/api.py
+++ b/cratedb_toolkit/io/mongodb/api.py
@@ -90,7 +90,12 @@ def mongodb_copy_migr8(source_url, target_url, transformation: Path = None, limi
     return True
 
 
-def mongodb_copy(source_url, target_url, transformation: t.Union[Path, None] = None, progress: bool = False):
+def mongodb_copy(
+    source_url: t.Union[str, URL],
+    target_url: t.Union[str, URL],
+    transformation: t.Union[Path, None] = None,
+    progress: bool = False,
+):
     """
     Transfer MongoDB collection using translator component.
 
@@ -102,6 +107,9 @@ def mongodb_copy(source_url, target_url, transformation: t.Union[Path, None] = N
 
     logger.info(f"mongodb_copy. source={source_url}, target={target_url}")
 
+    source_url = URL(source_url)
+    target_url = URL(target_url)
+
     # Optionally configure transformations.
     tm = None
     if transformation:
@@ -110,9 +118,9 @@ def mongodb_copy(source_url, target_url, transformation: t.Union[Path, None] = N
     tasks = []
 
     has_table = True
-    if "*" in source_url:
+    if "*" in source_url.path:
         has_table = False
-    mongodb_address = DatabaseAddress.from_string(source_url)
+    mongodb_address = DatabaseAddress(source_url)
     mongodb_uri, mongodb_collection_address = mongodb_address.decode()
     if mongodb_collection_address.table is None:
         has_table = False
@@ -129,8 +137,8 @@ def mongodb_copy(source_url, target_url, transformation: t.Union[Path, None] = N
         )
     else:
         logger.info(f"Inquiring collections at {source_url}")
-        mongodb_uri = URL(source_url)
-        cratedb_uri = URL(target_url)
+        mongodb_uri = source_url
+        cratedb_uri = target_url
         # What the hack?
         if (
             mongodb_uri.scheme.startswith("mongodb")
@@ -151,8 +159,8 @@ def mongodb_copy(source_url, target_url, transformation: t.Union[Path, None] = N
             cratedb_uri_effective = cratedb_uri.navigate(Path(collection_path).stem)
             tasks.append(
                 MongoDBFullLoad(
-                    mongodb_url=str(mongodb_uri_effective),
-                    cratedb_url=str(cratedb_uri_effective),
+                    mongodb_url=mongodb_uri_effective,
+                    cratedb_url=cratedb_uri_effective,
                     tm=tm,
                     progress=progress,
                 )
diff --git a/cratedb_toolkit/io/mongodb/copy.py b/cratedb_toolkit/io/mongodb/copy.py
index ec894c0..893a428 100644
--- a/cratedb_toolkit/io/mongodb/copy.py
+++ b/cratedb_toolkit/io/mongodb/copy.py
@@ -69,8 +69,8 @@ class MongoDBFullLoad:
 
     def __init__(
         self,
-        mongodb_url: str,
-        cratedb_url: str,
+        mongodb_url: t.Union[str, URL],
+        cratedb_url: t.Union[str, URL],
         tm: t.Union[TransformationManager, None],
         on_error: t.Literal["ignore", "raise"] = "raise",
         progress: bool = False,
@@ -83,7 +83,7 @@ def __init__(
         self.mongodb_adapter = mongodb_adapter_factory(self.mongodb_uri)
 
         # Decode database URL: CrateDB.
-        self.cratedb_address = DatabaseAddress.from_string(cratedb_url)
+        self.cratedb_address = DatabaseAddress(self.cratedb_uri)
         self.cratedb_sqlalchemy_url, self.cratedb_table_address = self.cratedb_address.decode()
         cratedb_table = self.cratedb_table_address.fullname