Add a tool for computing groundtruth for a runbook

diskann-benchmark-core/src/streaming/executors/bigann/runbook.rs

@@ -393,6 +393,21 @@ impl ScanDirectory {
 
         Ok(Self { directory, files })
     }
+
+    /// Returns the expected BigANN groundtruth file name for a stage.
+    ///
+    /// Files follow the `step<stage>.gt<suffix>` naming convention.
+    pub fn groundtruth_filename(stage: usize, suffix: &str) -> String {
+        format!("step{}.{}", stage, suffix)
+    }
+
+    /// Returns the expected BigANN groundtruth path for a stage.
+    ///
+    /// See also: [`Self::groundtruth_filename`].
+    pub fn groundtruth_path(&self, stage: usize, suffix: &str) -> PathBuf {
+        self.directory
+            .join(Self::groundtruth_filename(stage, suffix))
+    }
 }
 
 impl FindGroundtruth for ScanDirectory {
@@ -402,7 +417,7 @@ impl FindGroundtruth for ScanDirectory {
     ///
     /// Returns an error if no matching file is found or if multiple matches exist.
     fn find_groundtruth(&mut self, stage: usize) -> anyhow::Result<PathBuf> {
-        let prefix = format!("step{}.gt", stage);
+        let prefix = Self::groundtruth_filename(stage, "gt");
 
         enum Matches<'a> {
             None,
@@ -437,10 +452,10 @@ impl FindGroundtruth for ScanDirectory {
         match matches {
             Matches::One(m) => Ok(self.directory.join(m)),
             Matches::None => Err(anyhow::anyhow!(
-                "No groundtruth found for step {} in \"{}\", expected pattern: \"step{}.gt[0-9]*\"",
+                "No groundtruth found for step {} in \"{}\", expected pattern: \"{}[0-9]*\"",
                 stage,
                 self.directory.display(),
-                stage,
+                prefix,
             )),
             Matches::Many(matches) => Err(anyhow::anyhow!(
                 "Multiple groundtruth files found for step {} in \"{}\": {:?}",

diskann-tools/Cargo.toml

@@ -15,6 +15,7 @@ clap = { workspace = true, features = ["derive"] }
 diskann-providers = { workspace = true, default-features = false } # see `linalg/Cargo.toml`
 diskann-vector = { workspace = true }
 diskann-utils = { workspace = true }
+diskann-benchmark-core = { workspace = true, features = ["bigann"] }
 bytemuck.workspace = true
 num_cpus.workspace = true
 rayon.workspace = true