diff options
author | Alan Pearce | 2025-03-22 16:51:43 +0100 |
---|---|---|
committer | Alan Pearce | 2025-03-22 16:51:43 +0100 |
commit | 1328df9c5ca44e903a052c9bfc9a67dd34e25704 (patch) | |
tree | a18ca6edf71d019a0895367c0bca53786d2f892e /internal | |
parent | d39204eadf673b6c7ae940203fd75c0805245a96 (diff) | |
download | searchix-1328df9c5ca44e903a052c9bfc9a67dd34e25704.tar.lz searchix-1328df9c5ca44e903a052c9bfc9a67dd34e25704.tar.zst searchix-1328df9c5ca44e903a052c9bfc9a67dd34e25704.zip |
feat: promote prefix/exact name/attr matches
Fixes: https://todo.sr.ht/~alanpearce/searchix/18
Diffstat (limited to 'internal')
-rw-r--r-- | internal/index/indexer.go | 15 | ||||
-rw-r--r-- | internal/index/search.go | 27 | ||||
-rw-r--r-- | internal/index/search_test.go | 150 |
3 files changed, 181 insertions, 11 deletions
diff --git a/internal/index/indexer.go b/internal/index/indexer.go index 8cbc8e2..c4032e8 100644 --- a/internal/index/indexer.go +++ b/internal/index/indexer.go @@ -8,6 +8,7 @@ import ( "math" "os" "path" + "path/filepath" "slices" "go.alanpearce.eu/searchix/internal/file" @@ -93,13 +94,6 @@ func createIndexMapping() (mapping.IndexMapping, errors.E) { if err != nil { return nil, errors.WithMessage(err, "could not add custom analyser") } - err = indexMapping.AddCustomAnalyzer("keyword_single", map[string]any{ - "type": keyword.Name, - "tokenizer": letter.Name, - }) - if err != nil { - return nil, errors.WithMessage(err, "could not add custom analyser") - } identityFieldMapping := bleve.NewKeywordFieldMapping() @@ -222,6 +216,13 @@ func OpenOrCreate( ) (*ReadIndex, *WriteIndex, bool, errors.E) { var err errors.E bleve.SetLog(zap.NewStdLog(options.Logger.Named("bleve").GetLogger())) + if !filepath.IsAbs(dataRoot) { + wd, err := os.Getwd() + if err != nil { + return nil, nil, false, errors.WithMessagef(err, "could not get working directory") + } + dataRoot = filepath.Join(wd, dataRoot) + } indexPath := path.Join(dataRoot, indexBaseName) metaPath := path.Join(dataRoot, metaBaseName) diff --git a/internal/index/search.go b/internal/index/search.go index 292661e..a8124c7 100644 --- a/internal/index/search.go +++ b/internal/index/search.go @@ -128,8 +128,8 @@ func (index *ReadIndex) Search( // match the user's query in any field ... query.AddMust(bleve.NewDisjunctionQuery( - bleve.NewTermQuery(keyword), - bleve.NewPrefixQuery(keyword), + setBoost(bleve.NewTermQuery(keyword), 50), + setBoost(bleve.NewPrefixQuery(keyword), 25), bleve.NewMatchPhraseQuery(keyword), bleve.NewMatchQuery(keyword), )) @@ -140,13 +140,23 @@ func (index *ReadIndex) Search( ) } else { q := bleve.NewDisjunctionQuery( - setBoost(setField(bleve.NewTermQuery("nixpkgs"), "Source"), -150), - setBoost(setField(bleve.NewTermQuery("nur"), "Source"), -200), + setBoost(setField(bleve.NewTermQuery("nixpkgs"), "Source"), -1000), + setBoost(setField(bleve.NewTermQuery("nur"), "Source"), -5000), ) query.AddShould(q) } + mainProgramQuery := bleve.NewMatchQuery(keyword) + mainProgramQuery.SetField("MainProgram") + mainProgramQuery.SetBoost(50) + query.AddShould(mainProgramQuery) + + mainProgramLiteralQuery := bleve.NewTermQuery(keyword) + mainProgramLiteralQuery.SetField("MainProgram") + mainProgramLiteralQuery.SetBoost(100) + query.AddShould(mainProgramLiteralQuery) + programsQuery := bleve.NewMatchQuery(keyword) programsQuery.SetField("Programs") programsQuery.SetBoost(2) @@ -215,3 +225,12 @@ func (index *ReadIndex) GetDocument( return nil, err } + +func (index *ReadIndex) Close() error { + err := index.index.Close() + if err != nil { + return errors.WithStack(err) + } + + return nil +} diff --git a/internal/index/search_test.go b/internal/index/search_test.go new file mode 100644 index 0000000..339a0de --- /dev/null +++ b/internal/index/search_test.go @@ -0,0 +1,150 @@ +package index_test + +import ( + "context" + "maps" + "math" + "slices" + "testing" + "time" + + "go.alanpearce.eu/searchix/internal/config" + "go.alanpearce.eu/searchix/internal/index" + "go.alanpearce.eu/searchix/internal/nix" + "go.alanpearce.eu/x/log" +) + +const dataRoot = "../../data" + +func TestSearchGitPackagesFirst(t *testing.T) { + log := log.Configure(false) + cfg := config.DefaultConfig + + read, _, exists, err := index.OpenOrCreate(dataRoot, false, &index.Options{ + Logger: log.Named("index"), + LowMemory: false, + }) + defer read.Close() + if err != nil { + t.Fatal(err) + } + if !exists { + t.Fatal("expected index to exist") + } + + ctx, cancel := context.WithTimeout(context.Background(), 1*time.Second) + defer cancel() + + source := cfg.Importer.Sources["nixpkgs"] + if source == nil || !source.Enable { + t.Fatal("expected source to exist and be enabled") + } + + result, err := read.Search( + ctx, + source, + "git", + 0, + 100, + ) + if err != nil { + t.Fatal(err) + } + + if result.Total < 4 { + t.Errorf("Expected at least 4 results, got %d", result.Total) + } + important := map[string]int{ + "git": 0, + "git-doc": 0, + "gitFull": 0, + "gitMinimal": 0, + "gitSVN": 0, + } + var i int + for hit := range result.Hits { + data := hit.Data.(nix.Package) + if _, found := important[data.Attribute]; found { + important[data.Attribute] = i + } + i++ + } + if slices.Max(slices.Collect(maps.Values(important))) > len(important) { + t.Errorf( + "Expected all of %s to be the first %d matches, got %v", + slices.Collect(maps.Keys(important)), + len(important), + important, + ) + } +} + +func TestSearchJujutsuPackagesFirst(t *testing.T) { + log := log.Configure(false) + cfg := config.DefaultConfig + + read, _, exists, err := index.OpenOrCreate(dataRoot, false, &index.Options{ + Logger: log.Named("index"), + LowMemory: false, + }) + defer read.Close() + if err != nil { + t.Fatal(err) + } + if !exists { + t.Fatal("expected index to exist") + } + + ctx, cancel := context.WithTimeout(context.Background(), 1*time.Second) + defer cancel() + + source := cfg.Importer.Sources["nixpkgs"] + if source == nil || !source.Enable { + t.Fatal("expected source to exist and be enabled") + } + + result, err := read.Search( + ctx, + source, + "jj", + 0, + 100, + ) + if err != nil { + t.Fatal(err) + } + + if result.Total < 4 { + t.Errorf("Expected at least 4 results, got %d", result.Total) + } + important := map[string]int{ + "jj": 0, + "jujutsu": 0, + "lazyjj": 0, + "jjui": 0, + "jj-fzf": 0, + } + matches := []string{} + unwanted := "javacc" + unwantedIndex := math.MaxInt + var i int + for hit := range result.Hits { + data := hit.Data.(nix.Package) + if _, found := important[data.Attribute]; found { + matches = append(matches, data.Attribute) + } else if data.Attribute == unwanted { + unwantedIndex = i + matches = append(matches, data.Attribute) + } + i++ + } + if slices.Max(slices.Collect(maps.Values(important))) > unwantedIndex { + t.Errorf( + "Expected all of %s to be above unwanted result %s at index %d. Results: %v", + slices.Collect(maps.Keys(important)), + unwanted, + unwantedIndex, + matches, + ) + } +} |