From 4322ad0db0a1938a0f30ea43ac0bebe04289410e Mon Sep 17 00:00:00 2001
From: Tyler Murray <tylerm@allenai.org>
Date: Tue, 6 Aug 2024 16:01:38 -0700
Subject: [PATCH] Always use inferred extension (#183)

* Always use inferred extension

* fmt

* bump version
---
 pyproject.toml | 2 +-
 src/shard.rs   | 4 ++--
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/pyproject.toml b/pyproject.toml
index 950cca97..ca95fc6a 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -1,6 +1,6 @@
 [project]
 name = "dolma"
-version = "1.0.7"
+version = "1.0.8"
 description = "Data filters"
 license = { text = "Apache-2.0" }
 readme = "README.md"
diff --git a/src/shard.rs b/src/shard.rs
index 576ef167..a28dd873 100644
--- a/src/shard.rs
+++ b/src/shard.rs
@@ -113,8 +113,8 @@ impl Shard {
             }
             if !shard_inputs.is_empty() {
                 let output = format!(
-                    "{}/{}-{:04}.json.gz",
-                    stream_config.output.path, stream_config.name, stream_shard_count
+                    "{}/{}-{:04}.json{}",
+                    stream_config.output.path, stream_config.name, stream_shard_count, output_ext
                 );
                 let shard = Shard {
                     inputs: shard_inputs.clone(),