fix(search): remove bm25 config file dependency (#48)

Author: garrettladley

Cargo.lock

@@ -23,7 +23,6 @@ workspace = true
 ignore = "^0.4"
 rust-stemmers = "^1.2"
 stop-words = "^0.10"
-config = { version = "^0.15", default-features = false, features = ["yaml"] }
 serde = { version = "^1.0", features = ["derive"] }
 serde_json = "^1.0"
 rayon = "^1.11"

configuration/bm25.yml

@@ -1,11 +1,3 @@
-#[derive(Debug, thiserror::Error)]
-pub enum RankingError {
-    #[error("failed to load BM25 configuration")]
-    ConfigLoad(#[from] config::ConfigError),
-    #[error("failed to determine current directory")]
-    CurrentDir(#[source] std::io::Error),
-}
-
 #[derive(Debug, thiserror::Error)]
 pub enum EvalError {
     #[error("failed to parse evaluation data")]

crates/core/Cargo.toml

@@ -2,27 +2,21 @@ use dashmap::DashMap;
 use rayon::iter::{ParallelBridge, ParallelIterator};
 
 use crate::{
-    error::RankingError,
     index::{DocId, PostingList, RankedIndexReader, Term},
     query::{AnalyzedQuery, QueryTerm},
     ranking::{scorer::Scorer, utils::idf},
 };
 
-#[derive(serde::Deserialize, Debug, Clone)]
+#[derive(Debug, Clone)]
 pub struct BM25HyperParams {
     pub k1: f64,
     pub b: f64,
 }
 
-pub fn get_configuration() -> Result<BM25HyperParams, RankingError> {
-    let base_path = std::env::current_dir().map_err(RankingError::CurrentDir)?;
-    let configuration_directory = base_path.join("configuration");
-
-    let settings = config::Config::builder()
-        .add_source(config::File::from(configuration_directory.join("bm25.yml")))
-        .build()?;
-
-    Ok(settings.try_deserialize::<BM25HyperParams>()?)
+impl Default for BM25HyperParams {
+    fn default() -> Self {
+        Self { k1: 1.2, b: 0.75 }
+    }
 }
 
 pub struct BM25 {

crates/core/src/error.rs

@@ -4,7 +4,6 @@ use dashmap::DashMap;
 use rayon::iter::{ParallelBridge, ParallelIterator};
 
 use crate::{
-    error::RankingError,
     index::{DocId, DocumentField, PostingList, RankedIndexReader, Term, TermDocument},
     query::{AnalyzedQuery, QueryIntent, QueryTerm},
     ranking::{scorer::Scorer, utils::idf},
@@ -106,10 +105,6 @@ impl BM25FHyperParams {
     }
 }
 
-pub fn get_configuration() -> Result<BM25FHyperParams, RankingError> {
-    Ok(BM25FHyperParams::code_search_defaults())
-}
-
 pub struct BM25F {
     pub hyper_params: BM25FHyperParams,
 }

crates/core/src/ranking/bm25.rs

@@ -10,7 +10,7 @@ pub mod scorer;
 pub mod tf_idf;
 mod utils;
 
-pub use bm25::{BM25, BM25HyperParams, get_configuration};
+pub use bm25::{BM25, BM25HyperParams};
 pub use bm25f::{BM25F, BM25FHyperParams};
 pub use cosine_similarity::CosineSimilarity;
 pub use explanation::{

crates/core/src/ranking/bm25f.rs

@@ -10,7 +10,7 @@ use crate::{
         BM25, BM25F, BM25FHyperParams, BM25HyperParams, CosineSimilarity, FieldContribution,
         ProximityConfig, QueryLikelihood, QueryLikelihoodParams, ScoreExplanation,
         ScoreWithExplanation, ScoredWithExplanations, StaticQualityContribution, TFIDF,
-        TermExplanation, get_configuration, idf,
+        TermExplanation, idf,
     },
 };
 
@@ -518,22 +518,12 @@ impl FromStr for RankingAlgo {
     fn from_str(s: &str) -> Result<Self, Self::Err> {
         match s {
             "cosim" => Ok(RankingAlgo::CosineSimilarity),
-            "bm25" => {
-                let hyper_params =
-                    get_configuration().map_err(|e| format!("failed to load BM25 config: {e}"))?;
-
-                Ok(RankingAlgo::BM25(hyper_params))
-            }
+            "bm25" => Ok(RankingAlgo::BM25(BM25HyperParams::default())),
             "bm25f" => Ok(RankingAlgo::BM25F(BM25FHyperParams::code_search_defaults())),
-            "bm25-proximity" | "proximity" => {
-                let hyper_params =
-                    get_configuration().map_err(|e| format!("failed to load BM25 config: {e}"))?;
-
-                Ok(RankingAlgo::BM25Proximity(
-                    hyper_params,
-                    ProximityConfig::default(),
-                ))
-            }
+            "bm25-proximity" | "proximity" => Ok(RankingAlgo::BM25Proximity(
+                BM25HyperParams::default(),
+                ProximityConfig::default(),
+            )),
             "ql" | "ql-dirichlet" | "query-likelihood" => Ok(RankingAlgo::QueryLikelihood(
                 QueryLikelihoodParams::dirichlet_defaults(),
             )),

crates/core/src/ranking/mod.rs

@@ -197,20 +197,20 @@
       ]
     },
     {
-      "query": "bm25 hyperparameters load configuration bm25 yml",
-      "narrative": "Find where BM25 k1 and b are loaded from configuration.",
+      "query": "bm25 hyperparameters default k1 b",
+      "narrative": "Find where BM25 k1 and b defaults are defined.",
       "query_shape": "configuration",
       "results": [
         {
           "path": "core/src/ranking/bm25.rs",
-          "content": "get_configuration reads configuration/bm25.yml and deserializes BM25HyperParams.",
+          "content": "BM25HyperParams defines the built-in k1 and b defaults.",
           "relevant": true,
           "rank": 1,
           "evidence": [
             {
               "start_line": 13,
-              "end_line": 28,
-              "snippet": "pub struct BM25HyperParams"
+              "end_line": 21,
+              "snippet": "impl Default for BM25HyperParams"
             }
           ]
         }