min character-ops

.gitignore

@@ -4,4 +4,4 @@
 
 *.json
 internal/tests/vocab.txt
-internal/tests/vocab_bigram.txt
+internal/tests/vocab_bigram*

internal/lookup_compound.go

@@ -38,7 +38,7 @@ func (s *SymSpell) LookupCompound(phrase string, maxEditDistance int) *items.Sug
 	}
 	for i := range terms1 {
 		cp.terms1 = terms1[i]
-		cp.suggestions, _ = s.Lookup(cp.terms1, verbositypkg.Top, maxEditDistance)
+		s.getSuggestion(&cp, maxEditDistance)
 		// Combine adjacent terms
 		if i > 0 && !cp.isLastCombi {
 			cp.terms2 = terms1[i-1]
@@ -109,6 +109,18 @@ func (s *SymSpell) LookupCompound(phrase string, maxEditDistance int) *items.Sug
 	return s.finalizeAnswer(phrase, cp.suggestionParts)
 }
 
+func (s *SymSpell) getSuggestion(cp *compoundProcessor, maxEditDistance int) {
+	if len(cp.terms1) > s.MinimumCharToChange {
+		cp.suggestions, _ = s.Lookup(cp.terms1, verbositypkg.Top, maxEditDistance)
+	} else {
+		cp.suggestions = []items.SuggestItem{{
+			Term:     cp.terms1,
+			Distance: 0,
+			Count:    math.MaxInt,
+		}}
+	}
+}
+
 func (s *SymSpell) getBestSuggestion2(cp compoundProcessor, maxEditDistance int) items.SuggestItem {
 	var best2 items.SuggestItem
 	if len(cp.suggestions) > 0 {

internal/symspell.go

@@ -22,6 +22,7 @@ type SymSpell struct {
 	SplitThreshold            int
 	PreserveCase              bool
 	SplitWordBySpace          bool
+	MinimumCharToChange       int
 	Words                     map[string]int
 	BelowThresholdWords       map[string]int
 	Deletes                   map[string][]string
@@ -59,6 +60,7 @@ func NewSymSpell(opt ...options.Options) (*SymSpell, error) {
 		SplitThreshold:            opts.SplitItemThreshold,
 		PreserveCase:              opts.PreserveCase,
 		SplitWordBySpace:          opts.SplitWordBySpace,
+		MinimumCharToChange:       opts.MinimumCharacterToChange,
 		Words:                     make(map[string]int),
 		BelowThresholdWords:       make(map[string]int),
 		Deletes:                   make(map[string][]string),

pkg/options/options.go

@@ -7,6 +7,7 @@ var DefaultOptions = SymspellOptions{
 	SplitItemThreshold:        1,
 	PreserveCase:              false,
 	SplitWordBySpace:          false,
+	MinimumCharacterToChange:  1,
 }
 
 type SymspellOptions struct {
@@ -16,6 +17,7 @@ type SymspellOptions struct {
 	SplitItemThreshold        int
 	PreserveCase              bool
 	SplitWordBySpace          bool
+	MinimumCharacterToChange  int
 }
 
 type Options interface {
@@ -69,3 +71,9 @@ func WithSplitWordBySpace() Options {
 		options.SplitWordBySpace = true
 	})
 }
+
+func WithMinimumCharacterToChange(charLength int) Options {
+	return NewFuncWireOption(func(options *SymspellOptions) {
+		options.MinimumCharacterToChange = charLength
+	})
+}

symspell_test.go

@@ -115,6 +115,7 @@ func TestLookupCompound(t *testing.T) {
 		options.WithCountThreshold(1),
 		options.WithMaxDictionaryEditDistance(3),
 		options.WithPrefixLength(7),
+		options.WithSplitWordBySpace(),
 	)
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {
@@ -184,15 +185,25 @@ func TestSymspellLookupCompoundUnigram(t *testing.T) {
 			},
 			want: "ازمایشگاه",
 		},
+		{
+			name: "Test Min Character",
+			args: args{
+				a:               "بیمارستان ا",
+				maxEditDistance: 3,
+			},
+			want: "بیمارستان ا",
+		},
 	}
 	symSpell := NewSymSpellWithLoadBigramDictionary("internal/tests/vocab_fa.txt",
-		"",
+		"internal/tests/vocab_bigram_fa.txt",
 		0,
 		1,
-		options.WithCountThreshold(1),
+		options.WithCountThreshold(0),
 		options.WithPrefixLength(5),
 		options.WithMaxDictionaryEditDistance(3),
 		options.WithSplitItemThreshold(100),
+		options.WithSplitWordBySpace(),
+		options.WithMinimumCharacterToChange(2),
 	)
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {