[squash] json.dumps for the logging

kralka · kralka · commit 4b73da722c71 · 2025-08-07T11:31:12.000+02:00
diff --git a/src/sedpack/io/dataset_base.py b/src/sedpack/io/dataset_base.py
@@ -314,7 +314,10 @@ def __init__(
             ]
 
             kept_metadata: set[str] = {
-                str(s.custom_metadata) for s in shard_list
+                json.dumps(
+                    s.custom_metadata,
+                    sort_keys=True,
+                ) for s in shard_list
             }
             self.dataset.logger.info(
                 "Filtered shards with custom metadata: %s from split: %s",

Original file line number	Diff line number	Diff line change
`@@ -314,7 +314,10 @@ def __init__(`
`314`	`314`	`]`
`315`	`315`
`316`	`316`	`kept_metadata: set[str] = {`
`317`		`- str(s.custom_metadata) for s in shard_list`
	`317`	`+ json.dumps(`
	`318`	`+ s.custom_metadata,`
	`319`	`+ sort_keys=True,`
	`320`	`+ ) for s in shard_list`
`318`	`321`	`}`
`319`	`322`	`self.dataset.logger.info(`
`320`	`323`	`"Filtered shards with custom metadata: %s from split: %s",`