Standardize names and use kebab case in diskann-tools

diskann-tools/src/bin/compute_groundtruth.rs

@@ -108,40 +108,40 @@ fn main() -> CMDResult<()> {
 #[derive(Debug, Parser)]
 struct ComputeGroundTruthArgs {
     /// data type <int8/uint8/float / fp16> (required)
-    #[arg(long = "data_type", default_value = "float")]
+    #[arg(long = "data-type", default_value = "float")]
     pub data_type: DataType,
 
     /// Distance function to use.
-    #[arg(long = "dist_fn", default_value = "l2")]
+    #[arg(long = "dist-fn", default_value = "l2")]
     pub distance_function: Metric,
 
     /// File containing the base vectors in binary format
-    #[arg(long = "base_file", short, required = true)]
+    #[arg(long = "base-file", short, required = true)]
     pub base_file: String,
 
-    #[arg(long = "base_file_labels", default_value = None)]
+    #[arg(long = "base-file-labels", default_value = None)]
     pub base_file_labels: Option<String>,
 
     /// File containing the query vectors in binary format
-    #[arg(long = "query_file", short, required = true)]
+    #[arg(long = "query-file", short, required = true)]
     pub query_file: String,
 
-    #[arg(long = "query_file_labels", default_value = None)]
+    #[arg(long = "query-file-labels", default_value = None)]
     pub query_file_labels: Option<String>,
 
     /// Path of the file to write the ground truth to in binary format.  Please don't append .bin at the end if no filter_label or filter_label_file is provided.  It will save the file with '.bin' at the end.  Otherwise it will save the file as filename_label.bin.
-    #[arg(long = "gt_file", short, required = true)]
+    #[arg(long = "gt-file", short, required = true)]
     pub ground_truth_file: String,
 
     /// Filter bitmap file in the range ground truth format
-    #[arg(long = "filter_bitmap_file", short, default_value = None)]
+    #[arg(long = "filter-bitmap-file", short, default_value = None)]
     pub filter_bitmap_file: Option<String>,
 
     /// Number of ground truth nearest neigbhors to compute
-    #[arg(long = "recall_at", short = 'K', default_value = "10")]
+    #[arg(long = "recall-at", short = 'K', default_value = "10")]
     pub recall_at: u32,
 
     /// File containing the associated data in binary format
-    #[arg(long = "associated_data_file", required = false, default_value = None)]
+    #[arg(long = "associated-data-file")]
     pub associated_data_file: Option<String>,
 }

diskann-tools/src/bin/compute_multivec_groundtruth.rs

@@ -94,36 +94,36 @@ fn main() -> CMDResult<()> {
 #[derive(Debug, Parser)]
 struct ComputeMultivecGroundTruthArgs {
     /// data type <int8/uint8/float / fp16> (required)
-    #[arg(long = "data_type", default_value = "float")]
+    #[arg(long = "data-type", default_value = "float")]
     pub data_type: DataType,
 
     /// Distance function to use.
-    #[arg(long = "dist_fn", default_value = "l2")]
+    #[arg(long = "dist-fn", default_value = "l2")]
     pub distance_function: Metric,
 
     /// Whether to use average or min aggregation
     #[arg(long = "aggregation", default_value = "average")]
     pub aggregation: MultivecAggregationMethod,
 
     /// File containing the base vectors in binary format
-    #[arg(long = "base_file", short, required = true)]
+    #[arg(long = "base-file", short, required = true)]
     pub base_file: String,
 
-    #[arg(long = "base_file_labels", default_value = None)]
+    #[arg(long = "base-file-labels", default_value = None)]
     pub base_file_labels: Option<String>,
 
     /// File containing the query vectors in binary format
-    #[arg(long = "query_file", short, required = true)]
+    #[arg(long = "query-file", short, required = true)]
     pub query_file: String,
 
-    #[arg(long = "query_file_labels", default_value = None)]
+    #[arg(long = "query-file-labels")]
     pub query_file_labels: Option<String>,
 
     /// Path of the file to write the ground truth to in binary format.  Please don't append .bin at the end if no filter_label or filter_label_file is provided.  It will save the file with '.bin' at the end.  Otherwise it will save the file as filename_label.bin.
-    #[arg(long = "gt_file", short, required = true)]
+    #[arg(long = "gt-file", short, required = true)]
     pub ground_truth_file: String,
 
     /// Number of ground truth nearest neighbors to compute
-    #[arg(long = "recall_at", short = 'K', default_value = "10")]
+    #[arg(long = "recall-at", short = 'K', default_value = "10")]
     pub recall_at: u32,
 }

diskann-tools/src/bin/compute_range_groundtruth.rs

@@ -93,35 +93,35 @@ fn main() -> CMDResult<()> {
 #[derive(Debug, Parser)]
 struct ComputeRangeGroundTruthArgs {
     /// data type <int8/uint8/float/fp16>
-    #[arg(long = "data_type", default_value = "float")]
+    #[arg(long = "data-type", default_value = "float")]
     pub data_type: DataType,
 
     /// Distance function to use.
-    #[arg(long = "dist_fn", default_value = "l2")]
+    #[arg(long = "dist-fn", default_value = "l2")]
     pub distance_function: Metric,
 
     /// File containing the base vectors in binary format
-    #[arg(long = "base_file", short, required = true)]
+    #[arg(long = "base-file", short, required = true)]
     pub base_file: String,
 
     /// Optional labels file for base vectors
-    #[arg(long = "base_file_labels", default_value = None)]
+    #[arg(long = "base-file-labels")]
     pub base_file_labels: Option<String>,
 
     /// File containing the query vectors in binary format
-    #[arg(long = "query_file", short, required = true)]
+    #[arg(long = "query-file", short, required = true)]
     pub query_file: String,
 
     /// Optional labels file for query vectors
-    #[arg(long = "query_file_labels", default_value = None)]
+    #[arg(long = "query-file-labels", default_value = None)]
     pub query_file_labels: Option<String>,
 
     /// Path of the file to write range ground truth to in binary format
-    #[arg(long = "gt_file", short, required = true)]
+    #[arg(long = "gt-file", short, required = true)]
     pub ground_truth_file: String,
 
     /// Filter bitmap file in range ground truth format
-    #[arg(long = "filter_bitmap_file", short, default_value = None)]
+    #[arg(long = "filter-bitmap-file", short, default_value = None)]
     pub filter_bitmap_file: Option<String>,
 
     /// Radius threshold used to include neighbors in range-groundtruth

diskann-tools/src/bin/compute_specificities.rs

@@ -18,15 +18,15 @@ use std::process;
 )]
 struct Args {
     /// File containing the base labels
-    #[arg(long = "base_label_file", short = 'b')]
+    #[arg(long = "base-file-labels", short = 'b')]
     pub base_label_file: String,
 
     /// File containing the query labels
-    #[arg(long = "query_label_file", short = 'q')]
+    #[arg(long = "query-file-labels", short = 'q')]
     pub query_label_file: String,
 
     /// Output file for specificities (optional)
-    #[arg(long = "specificity_output_file", short = 'o')]
+    #[arg(long = "specificity-output-file", short = 'o')]
     pub specificity_output_file: Option<String>,
 }
 

diskann-tools/src/bin/gen_associated_data_from_range.rs

@@ -20,7 +20,7 @@ fn main() -> CMDResult<()> {
 
 #[derive(Debug, Parser)]
 struct GenAssociatedDataFromRangeArgs {
-    #[arg(long = "associated_data_path")]
+    #[arg(long = "associated-data-file")]
     pub associated_data_path: String,
 
     #[arg(long = "start")]

diskann-tools/src/bin/generate_minmax.rs

@@ -27,11 +27,11 @@ use rand::{rngs::StdRng, SeedableRng};
 #[command(author, version, about, long_about = None)]
 struct Args {
     /// Input binary file path containing vector data
-    #[arg(short, long)]
+    #[arg(short, long = "input-file")]
     input: String,
 
     /// Output binary file path for quantized vectors
-    #[arg(short, long)]
+    #[arg(short, long = "output-file")]
     output: String,
 
     /// Number of bits for quantization (1, 2, 4, or 8)

diskann-tools/src/bin/generate_pq.rs

@@ -52,30 +52,30 @@ fn main() -> Result<(), CMDToolError> {
 #[derive(Debug, Parser)]
 struct BuildPQArgs {
     /// data type <int8/uint8/float / fp16> (required)
-    #[arg(long = "data_type", default_value = "float")]
+    #[arg(long = "data-type", default_value = "float")]
     pub data_type: DataType,
 
     /// Distance function to use.
-    #[arg(long = "dist_fn", default_value = "l2")]
+    #[arg(long = "dist-fn", default_value = "l2")]
     pub dist_fn: Metric,
 
     /// Path to the data file. The file should be in the format specified by the `data_type` argument.
-    #[arg(long = "data_path", short, required = true)]
+    #[arg(long = "data-file", short, required = true)]
     pub data_path: String,
 
     /// Path to the index file. The index will be saved to this prefixed name.
-    #[arg(long = "index_path_prefix", short, required = true)]
+    #[arg(long = "index-path-prefix", short, required = true)]
     pub index_path_prefix: String,
 
     /// Number of threads to use.
-    #[arg(long = "num_threads", short = 'T')]
+    #[arg(long = "num-threads", short = 'T')]
     pub num_threads: Option<usize>,
 
     /// Ratio of PQ training set size to data size
-    #[arg(long = "p_val", short = 'p', default_value = "0.1")]
+    #[arg(long = "p-val", short = 'p', default_value = "0.1")]
     pub p_val: f64,
 
     /// Number of PQ bytee
-    #[arg(long = "pq_bytes", short, default_value = "10")]
+    #[arg(long = "pq-bytes", short, default_value = "10")]
     pub pq_bytes: usize,
 }

diskann-tools/src/bin/generate_synthetic_labels.rs

@@ -11,7 +11,7 @@ use tracing::{error, info};
 #[derive(Debug, Parser)]
 struct GenerateSyntheticLabelsArgs {
     /// Filename for saving the label file
-    #[arg(long = "output_file", required = true)]
+    #[arg(long = "output-file", required = true)]
     pub output_file: String,
 
     /// Number of vectors

diskann-tools/src/bin/random_data_generator.rs

@@ -10,11 +10,11 @@ use diskann_tools::utils::{write_random_data, CMDResult, CMDToolError, DataType}
 #[derive(Debug, Parser)]
 struct RandomDataGeneratorArgs {
     /// data type <int8/uint8/float/fp16> (required)
-    #[arg(long = "data_type", required = true)]
+    #[arg(long = "data-type", required = true)]
     pub data_type: DataType,
 
     /// File name for saving the random vectors
-    #[arg(long = "output_file", required = true)]
+    #[arg(long = "output-file", required = true)]
     pub output_file: String,
 
     /// Dimensionality of the vector

diskann-tools/src/bin/relative_contrast.rs

@@ -93,26 +93,26 @@ fn main() -> CMDResult<()> {
 #[derive(Debug, Parser)]
 struct RelativeContrastArgs {
     /// Data type <int8/uint8/float/fp16>
-    #[arg(long = "data_type", default_value = "fp16")]
+    #[arg(long = "data-type", default_value = "fp16")]
     pub data_type: DataType,
 
     /// Vector data file path
-    #[arg(long = "data_file", short, required = true)]
+    #[arg(long = "data-file", short, required = true)]
     pub data_file: String,
 
     /// Query file in binary format
-    #[arg(long = "query_file", short, required = true)]
+    #[arg(long = "query-file", short, required = true)]
     pub query_file: String,
 
     /// Ground truth file for the queryset
-    #[arg(long = "gt_file", required = true)]
+    #[arg(long = "gt-file", required = true)]
     pub gt_file: String,
 
     /// Number of neighbors to use from ground truth
-    #[arg(long = "recall_at", short = 'K', default_value = "10")]
+    #[arg(long = "recall-at", short = 'K', default_value = "10")]
     pub recall_at: usize,
 
     /// Number of random distances to average per query
-    #[arg(long = "search_list", short = 'L', default_value = "10")]
+    #[arg(long = "search-list", short = 'L', default_value = "10")]
     pub search_list: usize,
 }

diskann-tools/src/bin/subsample_bin.rs

@@ -24,19 +24,23 @@ use diskann_utils::io::Metadata;
 #[command(name = "subsample_bin", about = "Subsample vectors from a binary file")]
 struct Args {
     /// Data type of the vectors, one of: float, int8, uint8, fp16
-    #[arg(value_enum)]
+    #[arg(long = "data-type", value_enum)]
     data_type: DataType,
 
     /// Input base binary file
+    #[arg(long = "base-bin-file")]
     base_bin_file: PathBuf,
 
     /// Output file for sampled vectors
+    #[arg(long = "sampled-output-file")]
     sampled_output_file: PathBuf,
 
     /// Sampling probability between 0 and 1, for example 0.1
+    #[arg(long = "sampling-probability")]
     sampling_probability: f64,
 
     /// Optional random seed for reproducible sampling
+    #[arg(long = "random-seed")]
     random_seed: Option<u64>,
 }