read from path or zip; unify errors

src/fastmultigather.rs

@@ -14,7 +14,7 @@ use std::sync::atomic::AtomicUsize;
 use std::collections::BinaryHeap;
 
 use crate::utils::{prepare_query, write_prefetch, PrefetchResult,
-    load_sketchlist_filenames, load_sketches, consume_query_by_gather};
+    consume_query_by_gather, load_sigpaths_from_zip_or_pathlist, load_sketches_from_zip_or_pathlist, ReportType};
 
 pub fn fastmultigather<P: AsRef<Path> + std::fmt::Debug + Clone>(
     query_filenames: P,
@@ -32,14 +32,9 @@ pub fn fastmultigather<P: AsRef<Path> + std::fmt::Debug + Clone>(
     let template = Sketch::MinHash(template_mh);
 
     // load the list of query paths
-    let querylist_paths = load_sketchlist_filenames(&query_filenames)?;
+    let (querylist_paths, temp_dir) = load_sigpaths_from_zip_or_pathlist(&query_filenames)?;
     println!("Loaded {} sig paths in querylist", querylist_paths.len());
 
-    // build the list of paths to match against.
-    println!("Loading matchlist");
-    let matchlist_paths = load_sketchlist_filenames(&matchlist_filename)?;
-    println!("Loaded {} sig paths in matchlist", matchlist_paths.len());
-
     let threshold_hashes : u64 = {
         let x = threshold_bp / scaled;
         if x > 0 {
@@ -52,22 +47,7 @@ pub fn fastmultigather<P: AsRef<Path> + std::fmt::Debug + Clone>(
     println!("threshold overlap: {} {}", threshold_hashes, threshold_bp);
 
     // Load all the against sketches
-    let result = load_sketches(matchlist_paths, &template)?;
-    let (sketchlist, skipped_paths, failed_paths) = result;
-
-    eprintln!("Loaded {} sketches to search against.", sketchlist.len());
-    if failed_paths > 0 {
-        eprintln!("WARNING: {} search paths failed to load. See error messages above.",
-                  failed_paths);
-    }
-    if skipped_paths > 0 {
-        eprintln!("WARNING: skipped {} search paths - no compatible signatures.",
-                  skipped_paths);
-    }
-
-    if sketchlist.is_empty() {
-        bail!("No sketches loaded to search against!?")
-    }
+    let sketchlist = load_sketches_from_zip_or_pathlist(&matchlist_filename, &template, ReportType::Against)?;
 
     // Iterate over all queries => do prefetch and gather!
     let processed_queries = AtomicUsize::new(0);

src/index.rs

@@ -1,8 +1,8 @@
-use std::path::{Path, PathBuf};
+use std::path::Path;
 use sourmash::sketch::Sketch;
 use sourmash::index::revindex::RevIndex;
 
-use crate::utils::load_paths_from_zip_or_pathlist;
+use crate::utils::load_sigpaths_from_zip_or_pathlist;
 
 pub fn index<P: AsRef<Path>>(
     siglist: P,
@@ -13,7 +13,7 @@ pub fn index<P: AsRef<Path>>(
 ) -> Result<(), Box<dyn std::error::Error>> {
     println!("Loading siglist");
 
-    let (index_sigs, temp_dir) = load_paths_from_zip_or_pathlist(&siglist)?;
+    let (index_sigs, temp_dir) = load_sigpaths_from_zip_or_pathlist(&siglist)?;
 
     // if index_sigs pathlist is empty, bail
     if index_sigs.is_empty() {

src/python/tests/test_multigather.py

@@ -263,8 +263,8 @@ def test_empty_against(runtmp, capfd):
     captured = capfd.readouterr()
     print(captured.err)
 
-    assert "Loaded 0 sketches to search against." in captured.err
-    assert "Error: No sketches loaded to search against!?" in captured.err
+    assert "Loaded 0 search signature(s)" in captured.err
+    assert "Error: No search signatures loaded, exiting." in captured.err
 
 
 def test_nomatch_in_against(runtmp, capfd):

src/python/tests/test_multisearch.py

@@ -193,7 +193,7 @@ def test_bad_query_2(runtmp, capfd):
     print(captured.err)
 
     assert "WARNING: could not load sketches from path 'no-exist'" in captured.err
-    assert "WARNING: 1 query signature paths failed to load. See error messages above." in captured.err
+    assert "WARNING: 1 query paths failed to load. See error messages above." in captured.err
 
 
 def test_bad_query_3(runtmp, capfd):
@@ -297,7 +297,7 @@ def test_bad_against_2(runtmp, capfd):
     print(captured.err)
 
     assert "WARNING: could not load sketches from path 'no-exist'" in captured.err
-    assert "WARNING: 1 against signature paths failed to load. See error messages above." in captured.err
+    assert "WARNING: 1 search paths failed to load. See error messages above." in captured.err
 
 
 def test_empty_query(runtmp):

src/utils.rs

@@ -439,7 +439,7 @@ pub fn load_sketches_from_zip<P: AsRef<Path>>(
 }
 
 
-pub fn load_paths_from_zip_or_pathlist<P: AsRef<Path>>(
+pub fn load_sigpaths_from_zip_or_pathlist<P: AsRef<Path>>(
     sketchlist_path: P,
 ) -> Result<(Vec<PathBuf>, Option<tempfile::TempDir>)> {
     eprintln!("Reading list of filepaths from: '{}'", sketchlist_path.as_ref().display());
@@ -462,19 +462,13 @@ pub enum ReportType {
     Against,
 }
 
-impl ReportType {
-    fn as_str(&self, plural: bool) -> &'static str {
-        match (self, plural) {
-            (ReportType::Query, true) => "queries",
-            (ReportType::Query, false) => "query",
-            (ReportType::Against, _) => "against",
-        }
-    }
-}
-
 impl std::fmt::Display for ReportType {
     fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
-        write!(f, "{}", self.as_str(false)) // assume not plural?
+        let description = match self {
+            ReportType::Query => "query",
+            ReportType::Against => "search",
+        };
+        write!(f, "{}", description)
     }
 }
 
@@ -483,7 +477,7 @@ pub fn load_sketches_from_zip_or_pathlist<P: AsRef<Path>>(
     template: &Sketch,
     report_type: ReportType,
 ) -> Result<Vec<SmallSignature>> {
-    eprintln!("Reading list of {} from: '{}'", report_type.as_str(true), sketchlist_path.as_ref().display());
+    eprintln!("Reading list of {} paths from: '{}'", report_type, sketchlist_path.as_ref().display());
 
     let (sketchlist, skipped_paths, failed_paths) =
         if sketchlist_path.as_ref().extension().map(|ext| ext == "zip").unwrap_or(false) {
@@ -507,7 +501,7 @@ pub fn report_on_sketch_loading(
 
     if failed_paths > 0 {
         eprintln!(
-            "WARNING: {} {} signature paths failed to load. See error messages above.",
+            "WARNING: {} {} paths failed to load. See error messages above.",
             failed_paths,
             report_type
         );
@@ -521,7 +515,7 @@ pub fn report_on_sketch_loading(
     }
 
     // Validate sketches
-    eprintln!("Loaded {} {} signatures", sketchlist.len(), report_type.as_str(false));
+    eprintln!("Loaded {} {} signature(s)", sketchlist.len(), report_type);
     if sketchlist.is_empty() {
         bail!("No {} signatures loaded, exiting.", report_type);
     }