cleanup

Author: tobz

lib/saluki-components/src/encoders/datadog/metrics/v2/constants.rs

@@ -5,7 +5,7 @@ pub const SERIES_V2_UNCOMPRESSED_SIZE_LIMIT: usize = 5_242_880; // 5 MiB
 
 // Protocol Buffers field numbers for series and sketch payload messages in the V2 format.
 //
-// These field numbers come from the Protocol Buffers definitions in `lib/datadog-protos/proto/agent_payload.proto`.
+// These field numbers come from the Protocol Buffers definitions in `lib/protos/datadog/proto/agent-payload/agent_payload.proto`.
 pub const RESOURCES_TYPE_FIELD_NUMBER: u32 = 1;
 pub const RESOURCES_NAME_FIELD_NUMBER: u32 = 2;
 

lib/saluki-components/src/encoders/datadog/metrics/v3/interner.rs

@@ -1,14 +1,16 @@
 //! Generic interning for dictionary deduplication.
 
-use std::{borrow::Borrow, collections::HashMap, hash::Hash};
+use std::{borrow::Borrow, hash::Hash};
+
+use saluki_common::collections::FastHashMap;
 
 /// Generic interning structure for dictionary deduplication.
 ///
 /// Assigns unique 1-based IDs to values, returning the same ID for duplicate values.
 /// ID 0 is reserved for "empty/none" in the V3 format.
 #[derive(Debug)]
 pub struct Interner<K: Eq + Hash> {
-    index: HashMap<K, i64>,
+    index: FastHashMap<K, i64>,
     last_id: i64,
 }
 
@@ -22,7 +24,7 @@ impl<K: Eq + Hash> Interner<K> {
     /// Creates a new empty interner.
     pub fn new() -> Self {
         Self {
-            index: HashMap::new(),
+            index: FastHashMap::default(),
             last_id: 0,
         }
     }
@@ -46,16 +48,10 @@ impl<K: Eq + Hash> Interner<K> {
     }
 
     /// Returns the number of interned values.
-    #[allow(dead_code)]
+    #[cfg(test)]
     pub fn len(&self) -> usize {
         self.index.len()
     }
-
-    /// Returns true if no values have been interned.
-    #[allow(dead_code)]
-    pub fn is_empty(&self) -> bool {
-        self.index.is_empty()
-    }
 }
 
 #[cfg(test)]

lib/saluki-components/src/encoders/datadog/metrics/v3/mod.rs

@@ -10,6 +10,7 @@
 //! - Batch encoding - all metrics must be collected before serialization
 //! - Separate value columns for different numeric types (sint64, float32, float64)
 
+mod constants;
 mod interner;
 mod types;
 mod writer;

lib/saluki-components/src/encoders/datadog/metrics/v3/types.rs

@@ -1,47 +1,8 @@
 //! V3 payload type definitions and protocol buffer field numbers.
 
-/// Protocol buffer field numbers for MetricData message.
-///
-/// These correspond to the field numbers in `payload_v3.proto`.
-pub mod field_numbers {
-    // Dictionary fields
-    pub const DICT_NAME_STR: u32 = 1;
-    pub const DICT_TAGS_STR: u32 = 2;
-    pub const DICT_TAGSETS: u32 = 3;
-    pub const DICT_RESOURCE_STR: u32 = 4;
-    pub const DICT_RESOURCE_LEN: u32 = 5;
-    pub const DICT_RESOURCE_TYPE: u32 = 6;
-    pub const DICT_RESOURCE_NAME: u32 = 7;
-    pub const DICT_SOURCE_TYPE_NAME: u32 = 8;
-    pub const DICT_ORIGIN_INFO: u32 = 9;
-
-    // Per-metric columns
-    pub const TYPES: u32 = 10;
-    pub const NAMES: u32 = 11;
-    pub const TAGS: u32 = 12;
-    pub const RESOURCES: u32 = 13;
-    pub const INTERVALS: u32 = 14;
-    pub const NUM_POINTS: u32 = 15;
-
-    // Point data
-    pub const TIMESTAMPS: u32 = 16;
-    pub const VALS_SINT64: u32 = 17;
-    pub const VALS_FLOAT32: u32 = 18;
-    pub const VALS_FLOAT64: u32 = 19;
-
-    // Sketch data
-    pub const SKETCH_NUM_BINS: u32 = 20;
-    pub const SKETCH_BIN_KEYS: u32 = 21;
-    pub const SKETCH_BIN_CNTS: u32 = 22;
-
-    // Additional per-metric columns
-    pub const SOURCE_TYPE_NAME: u32 = 23;
-    pub const ORIGIN_INFO: u32 = 24;
-}
-
 /// V3 metric type values.
 ///
-/// These match the `metricType` enum in `payload_v3.proto`.
+/// These match the `metricType` enum in `intake_v3.proto`.
 #[derive(Debug, Clone, Copy, PartialEq, Eq)]
 #[repr(u8)]
 pub enum V3MetricType {
@@ -53,7 +14,7 @@ pub enum V3MetricType {
 
 impl V3MetricType {
     /// Returns the numeric value for encoding in the types column.
-    pub fn as_u64(self) -> u64 {
+    pub const fn as_u64(self) -> u64 {
         self as u64
     }
 }
@@ -67,10 +28,13 @@ impl V3MetricType {
 pub enum V3ValueType {
     /// Value is zero, not stored explicitly.
     Zero = 0x00,
+
     /// Value is stored in vals_sint64.
     Sint64 = 0x10,
+
     /// Value is stored in vals_float32.
     Float32 = 0x20,
+
     /// Value is stored in vals_float64.
     Float64 = 0x30,
 }

lib/saluki-components/src/encoders/datadog/metrics/v3/writer.rs

@@ -6,8 +6,9 @@
 use protobuf::CodedOutputStream;
 use saluki_error::GenericError;
 
+use super::constants::*;
 use super::interner::Interner;
-use super::types::{field_numbers, value_type_for_values, V3MetricType, V3ValueType};
+use super::types::{value_type_for_values, V3MetricType, V3ValueType};
 
 const METRIC_TYPE_DEFAULT: i32 = 0;
 const METRIC_TYPE_AGENT_HIDDEN: i32 = 9;
@@ -183,49 +184,49 @@ impl V3Writer {
 
         // Dictionary fields (bytes - varint-length-prefixed strings concatenated)
         if !data.dict_name_bytes.is_empty() {
-            os.write_bytes(field_numbers::DICT_NAME_STR, &data.dict_name_bytes)?;
+            os.write_bytes(DICT_NAME_STR_FIELD_NUMBER, &data.dict_name_bytes)?;
         }
         if !data.dict_tags_bytes.is_empty() {
-            os.write_bytes(field_numbers::DICT_TAGS_STR, &data.dict_tags_bytes)?;
+            os.write_bytes(DICT_TAGS_STR_FIELD_NUMBER, &data.dict_tags_bytes)?;
         }
 
         // Packed repeated fields for dictionaries
-        os.write_repeated_packed_sint64(field_numbers::DICT_TAGSETS, &data.dict_tagsets)?;
+        os.write_repeated_packed_sint64(DICT_TAGSETS_FIELD_NUMBER, &data.dict_tagsets)?;
 
         if !data.dict_resource_str_bytes.is_empty() {
-            os.write_bytes(field_numbers::DICT_RESOURCE_STR, &data.dict_resource_str_bytes)?;
+            os.write_bytes(DICT_RESOURCE_STR_FIELD_NUMBER, &data.dict_resource_str_bytes)?;
         }
 
-        os.write_repeated_packed_int64(field_numbers::DICT_RESOURCE_LEN, &data.dict_resource_len)?;
-        os.write_repeated_packed_sint64(field_numbers::DICT_RESOURCE_TYPE, &data.dict_resource_type)?;
-        os.write_repeated_packed_sint64(field_numbers::DICT_RESOURCE_NAME, &data.dict_resource_name)?;
+        os.write_repeated_packed_int64(DICT_RESOURCE_LEN_FIELD_NUMBER, &data.dict_resource_len)?;
+        os.write_repeated_packed_sint64(DICT_RESOURCE_TYPE_FIELD_NUMBER, &data.dict_resource_type)?;
+        os.write_repeated_packed_sint64(DICT_RESOURCE_NAME_FIELD_NUMBER, &data.dict_resource_name)?;
 
         if !data.dict_source_type_bytes.is_empty() {
-            os.write_bytes(field_numbers::DICT_SOURCE_TYPE_NAME, &data.dict_source_type_bytes)?;
+            os.write_bytes(DICT_SOURCE_TYPE_NAME_FIELD_NUMBER, &data.dict_source_type_bytes)?;
         }
 
-        os.write_repeated_packed_int32(field_numbers::DICT_ORIGIN_INFO, &data.dict_origin_info)?;
+        os.write_repeated_packed_int32(DICT_ORIGIN_INFO_FIELD_NUMBER, &data.dict_origin_info)?;
 
         // Per-metric columns
-        os.write_repeated_packed_uint64(field_numbers::TYPES, &data.types)?;
-        os.write_repeated_packed_sint64(field_numbers::NAMES, &data.names)?;
-        os.write_repeated_packed_sint64(field_numbers::TAGS, &data.tags)?;
-        os.write_repeated_packed_sint64(field_numbers::RESOURCES, &data.resources)?;
-        os.write_repeated_packed_uint64(field_numbers::INTERVALS, &data.intervals)?;
-        os.write_repeated_packed_uint64(field_numbers::NUM_POINTS, &data.num_points)?;
-        os.write_repeated_packed_sint64(field_numbers::SOURCE_TYPE_NAME, &data.source_type_names)?;
-        os.write_repeated_packed_sint64(field_numbers::ORIGIN_INFO, &data.origin_infos)?;
+        os.write_repeated_packed_uint64(TYPES_FIELD_NUMBER, &data.types)?;
+        os.write_repeated_packed_sint64(NAMES_FIELD_NUMBER, &data.names)?;
+        os.write_repeated_packed_sint64(TAGS_FIELD_NUMBER, &data.tags)?;
+        os.write_repeated_packed_sint64(RESOURCES_FIELD_NUMBER, &data.resources)?;
+        os.write_repeated_packed_uint64(INTERVALS_FIELD_NUMBER, &data.intervals)?;
+        os.write_repeated_packed_uint64(NUM_POINTS_FIELD_NUMBER, &data.num_points)?;
+        os.write_repeated_packed_sint64(SOURCE_TYPE_NAME_FIELD_NUMBER, &data.source_type_names)?;
+        os.write_repeated_packed_sint64(ORIGIN_INFO_FIELD_NUMBER, &data.origin_infos)?;
 
         // Point data
-        os.write_repeated_packed_sint64(field_numbers::TIMESTAMPS, &data.timestamps)?;
-        os.write_repeated_packed_sint64(field_numbers::VALS_SINT64, &data.vals_sint64)?;
-        os.write_repeated_packed_float(field_numbers::VALS_FLOAT32, &data.vals_float32)?;
-        os.write_repeated_packed_double(field_numbers::VALS_FLOAT64, &data.vals_float64)?;
+        os.write_repeated_packed_sint64(TIMESTAMPS_FIELD_NUMBER, &data.timestamps)?;
+        os.write_repeated_packed_sint64(VALS_SINT64_FIELD_NUMBER, &data.vals_sint64)?;
+        os.write_repeated_packed_float(VALS_FLOAT32_FIELD_NUMBER, &data.vals_float32)?;
+        os.write_repeated_packed_double(VALS_FLOAT64_FIELD_NUMBER, &data.vals_float64)?;
 
         // Sketch data
-        os.write_repeated_packed_uint64(field_numbers::SKETCH_NUM_BINS, &data.sketch_num_bins)?;
-        os.write_repeated_packed_sint32(field_numbers::SKETCH_BIN_KEYS, &data.sketch_bin_keys)?;
-        os.write_repeated_packed_uint32(field_numbers::SKETCH_BIN_CNTS, &data.sketch_bin_cnts)?;
+        os.write_repeated_packed_uint64(SKETCH_NUM_BINS_FIELD_NUMBER, &data.sketch_num_bins)?;
+        os.write_repeated_packed_sint32(SKETCH_BIN_KEYS_FIELD_NUMBER, &data.sketch_bin_keys)?;
+        os.write_repeated_packed_uint32(SKETCH_BIN_CNTS_FIELD_NUMBER, &data.sketch_bin_cnts)?;
 
         os.flush()?;
         Ok(())