Add http(s) support to the command line (#8753)

* Add http(s) support to the command line * fmt * Add documentation * Add a test * fmt --------- Co-authored-by: Andrew Lamb <[email protected]>
apache · Feb 4, 2024 · 840499f · 840499f
1 parent 86a2ab0
commit 840499f
Show file tree

Hide file tree

Showing 4 changed files with 50 additions and 11 deletions.
diff --git a/datafusion-cli/Cargo.lock b/datafusion-cli/Cargo.lock
diff --git a/datafusion-cli/Cargo.toml b/datafusion-cli/Cargo.toml
@@ -41,7 +41,7 @@ dirs = "4.0.0"
 env_logger = "0.9"
 futures = "0.3"
 mimalloc = { version = "0.1", default-features = false }
-object_store = { version = "0.9.0", features = ["aws", "gcp"] }
+object_store = { version = "0.9.0", features = ["aws", "gcp", "http"] }
 parking_lot = { version = "0.12" }
 parquet = { version = "50.0.0", default-features = false }
 regex = "1.8"

diff --git a/datafusion-cli/src/exec.rs b/datafusion-cli/src/exec.rs
@@ -42,6 +42,7 @@ use datafusion::physical_plan::{collect, execute_stream};
 use datafusion::prelude::SessionContext;
 use datafusion::sql::{parser::DFParser, sqlparser::dialect::dialect_from_str};
 
+use object_store::http::HttpBuilder;
 use object_store::ObjectStore;
 use rustyline::error::ReadlineError;
 use rustyline::Editor;
@@ -281,6 +282,11 @@ async fn create_external_table(
             let builder = get_gcs_object_store_builder(url, cmd)?;
             Arc::new(builder.build()?) as Arc<dyn ObjectStore>
         }
+        "http" | "https" => Arc::new(
+            HttpBuilder::new()
+                .with_url(url.origin().ascii_serialization())
+                .build()?,
+        ) as Arc<dyn ObjectStore>,
         _ => {
             // for other types, try to get from the object_store_registry
             ctx.runtime_env()
@@ -329,12 +335,24 @@ mod tests {
             return plan_err!("LogicalPlan is not a CreateExternalTable");
         }
 
+        // Ensure the URL is supported by the object store
         ctx.runtime_env()
             .object_store(ListingTableUrl::parse(location)?)?;
 
         Ok(())
     }
 
+    #[tokio::test]
+    async fn create_object_store_table_http() -> Result<()> {
+        // Should be OK
+        let location = "http://example.com/file.parquet";
+        let sql =
+            format!("CREATE EXTERNAL TABLE test STORED AS PARQUET LOCATION '{location}'");
+        create_external_table_test(location, &sql).await?;
+
+        Ok(())
+    }
+
     #[tokio::test]
     async fn create_object_store_table_s3() -> Result<()> {
         let access_key_id = "fake_access_key_id";

diff --git a/docs/source/user-guide/cli.md b/docs/source/user-guide/cli.md
@@ -260,6 +260,27 @@ STORED AS CSV
 LOCATION '/path/to/aggregate_test_100.csv';
 ```
 
+## Registering Remote Data Sources
+
+`datafusion-cli` can read from remote locations using a variety of protocols.
+For example to read from a remote parquet file via HTTP(S) you can use the following:
+
+```sql
+CREATE EXTERNAL TABLE hits
+STORED AS PARQUET
+LOCATION 'https://datasets.clickhouse.com/hits_compatible/athena_partitioned/hits_1.parquet';
+```
+
+```sql
+❯ select count(*) from hits;
++----------+
+| COUNT(*) |
++----------+
+| 1000000  |
++----------+
+1 row in set. Query took 0.344 seconds.
+```
+
 ## Registering S3 Data Sources
 
 [AWS S3](https://aws.amazon.com/s3/) data sources can be registered by executing a `CREATE EXTERNAL TABLE` SQL statement.