Reduce dependency footprint with feature-gated DuckDB and compression

Cargo.toml

@@ -29,6 +29,11 @@ exclude = [
 name = "sql-splitter"
 path = "src/main.rs"
 
+[features]
+default = ["duckdb-query", "compression"]
+duckdb-query = ["dep:duckdb", "compression"]
+compression = ["dep:flate2", "dep:bzip2", "dep:xz2", "dep:zstd"]
+
 [dependencies]
 clap = { version = "4", features = ["derive"] }
 clap_complete = "4"
@@ -38,10 +43,10 @@ memchr = "2"
 once_cell = "1"
 ahash = "0.8"
 anyhow = "1"
-flate2 = "1"
-bzip2 = "0.6"
-xz2 = "0.1"
-zstd = "0.13"
+flate2 = { version = "1", optional = true }
+bzip2 = { version = "0.6", optional = true }
+xz2 = { version = "0.1", optional = true }
+zstd = { version = "0.13", optional = true }
 indicatif = "0.18"
 rand = "0.10"
 smallvec = { version = "1.13", features = ["union"] }
@@ -53,7 +58,7 @@ serde_yaml_ng = "0.10"
 fake = { version = "5", features = ["derive"] }
 sha2 = "0.11"
 hex = "0.4"
-duckdb = { version = "1.10502", features = ["bundled"] }
+duckdb = { version = "1.10502", features = ["bundled"], optional = true }
 dirs = "6"
 rustyline = "17"
 schemars = "1"

README.md

@@ -22,6 +22,19 @@ Split large SQL dump files into individual table files. Fast, memory-efficient,
 cargo install sql-splitter
 ```
 
+### Cargo features (library consumers)
+
+`sql-splitter` enables all features by default. To reduce dependency footprint:
+
+```toml
+sql-splitter = { version = "1", default-features = false }
+```
+
+Optional features:
+
+- `compression` (gzip/bzip2/xz/zstd support)
+- `duckdb-query` (enables the `query` command and DuckDB integration)
+
 ### From source
 
 ```bash

src/cmd/mod.rs

@@ -5,6 +5,7 @@ mod glob_util;
 pub(crate) mod graph;
 pub(crate) mod merge;
 mod order;
+#[cfg(feature = "duckdb-query")]
 mod query;
 pub(crate) mod redact;
 pub(crate) mod sample;
@@ -700,6 +701,7 @@ pub enum Commands {
     },
 
     /// Query SQL dumps using DuckDB's analytical engine
+    #[cfg(feature = "duckdb-query")]
     #[command(visible_alias = "qy")]
     #[command(after_help = "\x1b[1mExamples:\x1b[0m
   sql-splitter query dump.sql \"SELECT COUNT(*) FROM users\"
@@ -1055,6 +1057,7 @@ pub fn run(cli: Cli) -> anyhow::Result<()> {
             dry_run,
             reverse,
         } => order::run(file, output, dialect, check, dry_run, reverse),
+        #[cfg(feature = "duckdb-query")]
         Commands::Query(args) => query::run(args),
         Commands::Schema {
             output,

src/duckdb/mod.rs

@@ -5,7 +5,7 @@
 //!
 //! # Features
 //!
-//! - **Zero dependencies**: DuckDB is bundled and compiled into sql-splitter
+//! - **Bundled engine**: DuckDB is embedded via the optional `duckdb-query` feature
 //! - **Multi-dialect support**: MySQL, PostgreSQL, and SQLite dumps
 //! - **Memory management**: Auto-switches to disk mode for large dumps
 //! - **Caching**: Optional persistent cache for repeated queries

src/lib.rs

@@ -2,6 +2,7 @@ pub mod analyzer;
 pub mod cmd;
 pub mod convert;
 pub mod differ;
+#[cfg(feature = "duckdb-query")]
 pub mod duckdb;
 pub mod graph;
 pub mod json_schema;

src/main.rs

@@ -5,6 +5,7 @@ mod analyzer;
 mod cmd;
 mod convert;
 mod differ;
+#[cfg(feature = "duckdb-query")]
 mod duckdb;
 mod graph;
 mod json_schema;

src/splitter/mod.rs

@@ -70,10 +70,21 @@ impl Compression {
     ) -> std::io::Result<Box<dyn Read + 'a>> {
         Ok(match self {
             Compression::None => reader,
+            #[cfg(feature = "compression")]
             Compression::Gzip => Box::new(flate2::read::GzDecoder::new(reader)),
+            #[cfg(feature = "compression")]
             Compression::Bzip2 => Box::new(bzip2::read::BzDecoder::new(reader)),
+            #[cfg(feature = "compression")]
             Compression::Xz => Box::new(xz2::read::XzDecoder::new(reader)),
+            #[cfg(feature = "compression")]
             Compression::Zstd => Box::new(zstd::stream::read::Decoder::new(reader)?),
+            #[cfg(not(feature = "compression"))]
+            _ => {
+                return Err(std::io::Error::new(
+                    std::io::ErrorKind::Unsupported,
+                    "compressed input requires the `compression` feature",
+                ))
+            }
         })
     }
 }

tests/splitter_unit_test.rs

@@ -110,6 +110,7 @@ fn test_splitter_data_only() {
 }
 
 #[test]
+#[cfg(feature = "compression")]
 fn test_splitter_gzip_compressed() {
     use flate2::write::GzEncoder;
     use flate2::Compression as GzCompression;
@@ -135,6 +136,29 @@ fn test_splitter_gzip_compressed() {
     assert!(output_dir.join("users.sql").exists());
 }
 
+#[test]
+#[cfg(not(feature = "compression"))]
+fn test_splitter_compressed_input_requires_feature() {
+    let temp_dir = TempDir::new().unwrap();
+    let input_file = temp_dir.path().join("input.sql.gz");
+    let output_dir = temp_dir.path().join("output");
+
+    std::fs::write(
+        &input_file,
+        b"\x1f\x8b\x08\x00\\\x06\x03j\x02\xffs\x0eru\x0cqU\x08qt\xf2qU(-N-*V\xd0\xc8LQ\xf0\xf4\x0b\xd1\xb4\xe6\x02\x00^\xb7Dc\x1d\x00\x00\x00",
+    )
+    .unwrap();
+
+    let splitter = Splitter::new(input_file, output_dir);
+    let err = match splitter.split() {
+        Ok(_) => panic!("expected compressed input to fail without `compression` feature"),
+        Err(err) => err,
+    };
+    assert!(err
+        .chain()
+        .any(|cause| cause.to_string() == "compressed input requires the `compression` feature"));
+}
+
 #[test]
 fn test_compression_detection() {
     assert_eq!(