From 5c913a16fa955fbf5d94da8a82006bd2c1f89fd1 Mon Sep 17 00:00:00 2001
From: goutamadwant <workwithgoutam@gmail.com>
Date: Fri, 7 Aug 2026 23:08:53 -0700
Subject: [PATCH] feat(cli): add opt-in Spark function registration

---
 Cargo.lock                              |  1 +
 datafusion-cli/Cargo.toml               |  1 +
 datafusion-cli/src/main.rs              | 11 ++++++++-
 datafusion-cli/tests/cli_integration.rs | 30 +++++++++++++++++++++++++
 docs/source/user-guide/cli/usage.md     | 14 ++++++++++++
 5 files changed, 56 insertions(+), 1 deletion(-)

diff --git a/Cargo.lock b/Cargo.lock
index 3ddb32f60ffd5..4aeccccae8ef6 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -1866,6 +1866,7 @@ dependencies = [
  "ctor",
  "datafusion",
  "datafusion-common",
+ "datafusion-spark",
  "dirs",
  "env_logger",
  "futures",
diff --git a/datafusion-cli/Cargo.toml b/datafusion-cli/Cargo.toml
index 62eedafe798d4..50b4002541f89 100644
--- a/datafusion-cli/Cargo.toml
+++ b/datafusion-cli/Cargo.toml
@@ -56,6 +56,7 @@ datafusion = { workspace = true, features = [
     "unicode_expressions",
 ] }
 datafusion-common = { workspace = true }
+datafusion-spark = { workspace = true, features = ["core"] }
 dirs = "6.0.0"
 env_logger = { workspace = true }
 futures = { workspace = true }
diff --git a/datafusion-cli/src/main.rs b/datafusion-cli/src/main.rs
index 20a2537d7c10c..dd1a3940302d8 100644
--- a/datafusion-cli/src/main.rs
+++ b/datafusion-cli/src/main.rs
@@ -143,6 +143,12 @@ struct Args {
     #[clap(long, help = "Enables console syntax highlighting")]
     color: bool,
 
+    #[clap(
+        long,
+        help = "Enable Apache Spark-compatible functions, overriding DataFusion functions with the same name"
+    )]
+    spark: bool,
+
     #[clap(
         short = 'd',
         long,
@@ -244,8 +250,11 @@ async fn main_inner() -> Result<()> {
     let runtime_env = rt_builder.build_arc()?;
 
     // enable dynamic file query
-    let ctx = SessionContext::new_with_config_rt(session_config, runtime_env)
+    let mut ctx = SessionContext::new_with_config_rt(session_config, runtime_env)
         .enable_url_table();
+    if args.spark {
+        datafusion_spark::register_all(&mut ctx)?;
+    }
     ctx.refresh_catalogs().await?;
     // install dynamic catalog provider that can register required object stores
     ctx.register_catalog_list(Arc::new(DynamicObjectStoreCatalog::new(
diff --git a/datafusion-cli/tests/cli_integration.rs b/datafusion-cli/tests/cli_integration.rs
index 4dc244445a2eb..ec92fac190a3a 100644
--- a/datafusion-cli/tests/cli_integration.rs
+++ b/datafusion-cli/tests/cli_integration.rs
@@ -173,6 +173,36 @@ fn cli_quick_test<'a>(
     assert_cmd_snapshot!(cmd);
 }
 
+#[test]
+fn spark_functions_require_opt_in() {
+    let query = "SELECT next_day('2015-07-27'::DATE, 'Sun'::STRING);";
+
+    let output = cli()
+        .args(["-q", "--command", query])
+        .output()
+        .expect("failed to run datafusion-cli without --spark");
+    let stdout = String::from_utf8_lossy(&output.stdout);
+    assert!(!output.status.success(), "query unexpectedly succeeded");
+    assert!(
+        stdout.contains("Invalid function 'next_day'"),
+        "expected next_day to be unavailable without --spark, got:\n{stdout}"
+    );
+
+    let output = cli()
+        .args(["-q", "--spark", "--command", query])
+        .output()
+        .expect("failed to run datafusion-cli with --spark");
+    let stdout = String::from_utf8_lossy(&output.stdout);
+    assert!(
+        output.status.success(),
+        "query failed with --spark, got:\n{stdout}"
+    );
+    assert!(
+        stdout.contains("2015-08-02"),
+        "expected next_day result, got:\n{stdout}"
+    );
+}
+
 /// Read data piped into the CLI via the `/dev/stdin` pseudo-path.
 ///
 /// Unix-only: `/dev/stdin` does not exist on Windows. This drives the real
diff --git a/docs/source/user-guide/cli/usage.md b/docs/source/user-guide/cli/usage.md
index 75c6698f007a5..5cb5c55a69e80 100644
--- a/docs/source/user-guide/cli/usage.md
+++ b/docs/source/user-guide/cli/usage.md
@@ -52,6 +52,8 @@ Options:
           [possible values: numbers(0/10/...), inf(no limit)] [default: 40]
       --color
           Enables console syntax highlighting
+      --spark
+          Enable Apache Spark-compatible functions, overriding DataFusion functions with the same name
   -d, --disk-limit <DISK_LIMIT>
           Available disk space for spilling queries (e.g. '10g'), default to None (uses DataFusion's default value of '100g')
       --object-store-profiling <OBJECT_STORE_PROFILING>
@@ -63,6 +65,18 @@ Options:
           Print version
 ```
 
+## Apache Spark-compatible functions
+
+Use `--spark` to register the Apache Spark-compatible function library for the
+CLI session:
+
+```bash
+datafusion-cli --spark -c "SELECT next_day('2015-07-27'::DATE, 'Sun'::STRING);"
+```
+
+Spark-compatible functions are disabled by default. Enabling them overrides
+DataFusion functions with the same name for the duration of the session.
+
 ## Commands
 
 Available commands inside DataFusion CLI are: