MRG: rename internal file & functions to use 'rocksdb' (#494)

* rename internal file & functions to use 'rocksdb'

* fix fmt and clippy

src/fastmultigather.rs

@@ -25,6 +25,7 @@ use crate::utils::{
     consume_query_by_gather, load_collection, write_prefetch, PrefetchResult, ReportType,
 };
 
+#[allow(clippy::too_many_arguments)]
 pub fn fastmultigather(
     query_filepath: String,
     against_filepath: String,
@@ -48,7 +49,7 @@ pub fn fastmultigather(
     let scaled = match scaled {
         Some(s) => s,
         None => {
-            let scaled = query_collection.max_scaled().expect("no records!?").clone() as usize;
+            let scaled = *query_collection.max_scaled().expect("no records!?") as usize;
             eprintln!(
                 "Setting scaled={} based on max scaled in query collection",
                 scaled

src/mastiff_manygather.rs → src/fastmultigather_rocksdb.rs

@@ -1,4 +1,4 @@
-/// mastiff_manygather: mastiff-indexed version of fastmultigather.
+/// fastmultigather_rocksdb: rocksdb-indexed version of fastmultigather.
 use anyhow::Result;
 use camino::Utf8PathBuf as PathBuf;
 use rayon::prelude::*;
@@ -12,7 +12,7 @@ use crate::utils::{
     csvwriter_thread, is_revindex_database, load_collection, BranchwaterGatherResult, ReportType,
 };
 
-pub fn mastiff_manygather(
+pub fn fastmultigather_rocksdb(
     queries_file: String,
     index: PathBuf,
     selection: Selection,

src/lib.rs

@@ -15,11 +15,11 @@ mod check;
 mod cluster;
 mod fastgather;
 mod fastmultigather;
+mod fastmultigather_rocksdb;
 mod index;
 mod manysearch;
+mod manysearch_rocksdb;
 mod manysketch;
-mod mastiff_manygather;
-mod mastiff_manysearch;
 mod multisearch;
 mod pairwise;
 mod singlesketch;
@@ -28,6 +28,7 @@ use camino::Utf8PathBuf as PathBuf;
 
 #[pyfunction]
 #[pyo3(signature = (querylist_path, siglist_path, threshold, ksize, scaled, moltype, output_path=None, ignore_abundance=false))]
+#[allow(clippy::too_many_arguments)]
 fn do_manysearch(
     querylist_path: String,
     siglist_path: String,
@@ -45,10 +46,10 @@ fn do_manysearch(
 
     let ignore_abundance = ignore_abundance.unwrap_or(false);
 
-    // if siglist_path is revindex, run mastiff_manysearch; otherwise run manysearch
+    // if siglist_path is revindex, run rocksdb manysearch; otherwise run manysearch
     if is_revindex_database(&againstfile_path) {
-        // note: mastiff_manysearch ignores abundance automatically.
-        match mastiff_manysearch::mastiff_manysearch(
+        // note: manysearch_rocksdb ignores abundance automatically.
+        match manysearch_rocksdb::manysearch_rocksdb(
             querylist_path,
             againstfile_path,
             selection,
@@ -133,9 +134,9 @@ fn do_fastmultigather(
     let selection = build_selection(ksize, scaled, &moltype);
     let allow_failed_sigpaths = true;
 
-    // if a siglist path is a revindex, run mastiff_manygather. If not, run multigather
+    // if a siglist path is a revindex, run rocksdb fastmultigather. If not, run multigather
     if is_revindex_database(&againstfile_path) {
-        match mastiff_manygather::mastiff_manygather(
+        match fastmultigather_rocksdb::fastmultigather_rocksdb(
             query_filenames,
             againstfile_path,
             selection.clone(),

src/mastiff_manysearch.rs → src/manysearch_rocksdb.rs

@@ -1,4 +1,4 @@
-/// mastiff_manysearch: mastiff-indexed version of manysearch.
+/// manysearch_rocksdb: rocksdb-indexed version of manysearch.
 use anyhow::Result;
 use camino::Utf8PathBuf as PathBuf;
 use log::debug;
@@ -15,7 +15,7 @@ use crate::utils::{
     csvwriter_thread, is_revindex_database, load_collection, ReportType, SearchResult,
 };
 
-pub fn mastiff_manysearch(
+pub fn manysearch_rocksdb(
     queries_path: String,
     index: PathBuf,
     selection: Selection,

src/multisearch.rs

@@ -34,7 +34,7 @@ pub fn multisearch(
     let scaled = match selection.scaled() {
         Some(s) => s,
         None => {
-            let scaled = query_collection.max_scaled().expect("no records!?").clone() as u32;
+            let scaled = *query_collection.max_scaled().expect("no records!?") as u32;
             eprintln!(
                 "Setting scaled={} based on max scaled in query collection",
                 scaled

src/singlesketch.rs

@@ -34,7 +34,7 @@ pub fn singlesketch(
     };
 
     // Build signature templates based on parsed parameters and detected moltype
-    let mut sigs = crate::manysketch::build_siginfo(&params_vec, &moltype);
+    let mut sigs = crate::manysketch::build_siginfo(&params_vec, moltype);
 
     if sigs.is_empty() {
         bail!("No signatures to build for the given parameters.");

src/utils/mod.rs

@@ -737,7 +737,7 @@ pub fn branchwater_calculate_gather_stats(
     // If abundance, calculate abund-related metrics (vs current query)
     if calc_abund_stats {
         // take abunds from subtracted query
-        let (abunds, unique_weighted_found) = match match_mh.inflated_abundances(&query) {
+        let (abunds, unique_weighted_found) = match match_mh.inflated_abundances(query) {
             Ok((abunds, unique_weighted_found)) => (abunds, unique_weighted_found),
             Err(e) => return Err(e.into()),
         };