Embed spark

Cargo.toml

@@ -48,6 +48,7 @@ dirs = "5"
 pathdiff = "0.2.3"
 glob = "0.3"
 flate2 = "1.1.2"
+tar = "0.4"
 tempfile = "3"
 uuid = { version = "1.21.0", features = ["v4"] }
 

build.rs

@@ -1,4 +1,6 @@
 use std::env;
+use std::fs;
+use std::path::{Path, PathBuf};
 use std::process::Command;
 
 fn non_empty_env(name: &str) -> Option<String> {
@@ -66,4 +68,93 @@ fn main() {
     } else {
         println!("cargo:rerun-if-changed=.git/HEAD");
     }
+
+    stage_spark_assets();
+}
+
+fn stage_spark_assets() {
+    println!("cargo:rerun-if-env-changed=BT_SPARK_DIR");
+    println!("cargo:rerun-if-env-changed=BT_SPARK_EMBED");
+
+    let out_dir = PathBuf::from(env::var_os("OUT_DIR").expect("OUT_DIR is set by cargo"));
+    let spark_out = out_dir.join("spark");
+    fs::create_dir_all(&spark_out).expect("create OUT_DIR/spark");
+
+    let cli_dest = spark_out.join("cli.mjs");
+    let harness_dest = spark_out.join("harness.tgz");
+    let hash_dest = spark_out.join("asset_hash");
+    let embedded_marker = spark_out.join("embedded");
+
+    let embed_enabled = non_empty_env("BT_SPARK_EMBED")
+        .map(|value| !matches!(value.as_str(), "0" | "false" | "no" | "off"))
+        .unwrap_or(true);
+
+    if !embed_enabled {
+        write_placeholder_assets(&cli_dest, &harness_dest, &hash_dest, &embedded_marker);
+        println!("cargo:warning=BT_SPARK_EMBED disabled; `bt spark` will be a stub");
+        return;
+    }
+
+    let manifest_dir = PathBuf::from(
+        env::var_os("CARGO_MANIFEST_DIR").expect("CARGO_MANIFEST_DIR is set by cargo"),
+    );
+    let spark_dir = non_empty_env("BT_SPARK_DIR")
+        .map(PathBuf::from)
+        .unwrap_or_else(|| manifest_dir.join("..").join("spark"));
+
+    let cli_src = spark_dir.join("packages/spark/dist/cli.mjs");
+    let harness_src = spark_dir.join("dist-sea-build/harness.tgz");
+
+    if !cli_src.exists() || !harness_src.exists() {
+        write_placeholder_assets(&cli_dest, &harness_dest, &hash_dest, &embedded_marker);
+        println!(
+            "cargo:warning=spark assets not found under {}; build spark with `pnpm build:sea` to embed (looked for {} and {})",
+            spark_dir.display(),
+            cli_src.display(),
+            harness_src.display(),
+        );
+        return;
+    }
+
+    println!("cargo:rerun-if-changed={}", cli_src.display());
+    println!("cargo:rerun-if-changed={}", harness_src.display());
+
+    fs::copy(&cli_src, &cli_dest).unwrap_or_else(|err| panic!("copy spark cli.mjs: {err}"));
+    fs::copy(&harness_src, &harness_dest)
+        .unwrap_or_else(|err| panic!("copy spark harness.tgz: {err}"));
+
+    let hash = hash_files(&[cli_dest.as_path(), harness_dest.as_path()]);
+    fs::write(&hash_dest, &hash).expect("write asset_hash");
+    fs::write(&embedded_marker, b"1").expect("write embedded marker");
+}
+
+fn write_placeholder_assets(cli: &Path, harness: &Path, hash: &Path, marker: &Path) {
+    fs::write(cli, b"").expect("write empty cli.mjs placeholder");
+    fs::write(harness, b"").expect("write empty harness.tgz placeholder");
+    fs::write(hash, b"unembedded").expect("write asset_hash placeholder");
+    if marker.exists() {
+        fs::remove_file(marker).expect("remove stale embedded marker");
+    }
+}
+
+fn hash_files(paths: &[&Path]) -> String {
+    use std::io::Read;
+
+    // Tiny FNV-1a 64-bit, ample for cache-key purposes (no crypto needs here).
+    let mut hash: u64 = 0xcbf29ce484222325;
+    for path in paths {
+        let mut file = fs::File::open(path).expect("open asset for hashing");
+        let mut buf = [0u8; 8192];
+        loop {
+            let n = file.read(&mut buf).expect("read asset for hashing");
+            if n == 0 {
+                break;
+            }
+            for byte in &buf[..n] {
+                hash ^= u64::from(*byte);
+                hash = hash.wrapping_mul(0x100000001b3);
+            }
+        }
+    }
+    format!("{hash:016x}")
 }

src/main.rs

@@ -23,6 +23,7 @@ mod scorers;
 mod self_update;
 mod setup;
 mod source_language;
+mod spark;
 mod sql;
 mod status;
 mod switch;
@@ -79,6 +80,7 @@ Additional
   docs         Manage workflow docs for coding agents
   self         Self-management commands
   setup        Configure Braintrust setup flows
+  spark        Run the embedded spark wizard
   status       Show current org and project context
 
 Flags
@@ -161,6 +163,8 @@ enum Commands {
     Switch(CLIArgs<switch::SwitchArgs>),
     /// Show current org and project context
     Status(CLIArgs<status::StatusArgs>),
+    /// Run the embedded spark wizard
+    Spark(CLIArgs<spark::SparkArgs>),
     // /// View and modify config
     // Config(CLIArgs<config::ConfigArgs>),
 }
@@ -189,6 +193,7 @@ impl Commands {
             Commands::Util(cmd) => &cmd.base,
             Commands::Switch(cmd) => &cmd.base,
             Commands::Status(cmd) => &cmd.base,
+            Commands::Spark(cmd) => &cmd.base,
         }
     }
 
@@ -215,6 +220,7 @@ impl Commands {
             Commands::Util(cmd) => &mut cmd.base,
             Commands::Switch(cmd) => &mut cmd.base,
             Commands::Status(cmd) => &mut cmd.base,
+            Commands::Spark(cmd) => &mut cmd.base,
         }
     }
 
@@ -293,6 +299,7 @@ fn try_main() -> Result<()> {
             Commands::SelfCommand(cmd) => self_update::run(cmd.base, cmd.args).await?,
             Commands::Switch(cmd) => switch::run(cmd.base, cmd.args).await?,
             Commands::Status(cmd) => status::run(cmd.base, cmd.args).await?,
+            Commands::Spark(cmd) => spark::run(cmd.base, cmd.args).await?,
         }
         Ok(())
     });