Merge pull request #19 from varnamproject/improve-search-symbol-table

Improve search symbol table, add all config vars to varnam_config(), deprecated previous config functions

c-shared.go

@@ -271,22 +271,25 @@ func varnam_set_indic_digits(varnamHandleID C.int, val C.int) {
 	varnam_config(varnamHandleID, C.VARNAM_CONFIG_USE_INDIC_DIGITS, val)
 }
 
-// TODO move all config to varnam_config()
+// Deprecated. Use varnam_config()
 //export varnam_set_dictionary_suggestions_limit
 func varnam_set_dictionary_suggestions_limit(varnamHandleID C.int, val C.int) {
 	getVarnamHandle(varnamHandleID).varnam.DictionarySuggestionsLimit = int(val)
 }
 
+// Deprecated. Use varnam_config()
 //export varnam_set_pattern_dictionary_suggestions_limit
 func varnam_set_pattern_dictionary_suggestions_limit(varnamHandleID C.int, val C.int) {
 	getVarnamHandle(varnamHandleID).varnam.PatternDictionarySuggestionsLimit = int(val)
 }
 
+// Deprecated. Use varnam_config()
 //export varnam_set_tokenizer_suggestions_limit
 func varnam_set_tokenizer_suggestions_limit(varnamHandleID C.int, val C.int) {
 	getVarnamHandle(varnamHandleID).varnam.TokenizerSuggestionsLimit = int(val)
 }
 
+// Deprecated. Use varnam_config()
 //export varnam_set_dictionary_match_exact
 func varnam_set_dictionary_match_exact(varnamHandleID C.int, val C.int) {
 	if val == 0 {
@@ -404,6 +407,27 @@ func varnam_get_vst_path(varnamHandleID C.int) *C.char {
 	return C.CString(handle.varnam.VSTPath)
 }
 
+//export varnam_new_search_symbol
+func varnam_new_search_symbol(resultPointer **C.struct_Symbol_t) C.int {
+	symbol := govarnam.NewSearchSymbol()
+	*resultPointer = C.makeSymbol(
+		C.int(symbol.Identifier),
+		C.int(symbol.Type),
+		C.int(symbol.MatchType),
+		C.CString(symbol.Pattern),
+		C.CString(symbol.Value1),
+		C.CString(symbol.Value2),
+		C.CString(symbol.Value3),
+		C.CString(symbol.Tag),
+		C.int(symbol.Weight),
+		C.int(symbol.Priority),
+		C.int(symbol.AcceptCondition),
+		C.int(symbol.Flags),
+	)
+
+	return C.VARNAM_SUCCESS
+}
+
 //export varnam_search_symbol_table
 func varnam_search_symbol_table(varnamHandleID C.int, id C.int, searchCriteria C.struct_Symbol_t, resultPointer **C.varray) C.int {
 	ctx, cancel := makeContext(id)
@@ -633,6 +657,18 @@ func varnam_config(varnamHandleID C.int, key C.int, value C.int) C.int {
 	case C.VARNAM_CONFIG_IGNORE_DUPLICATE_TOKEN:
 		handle.varnam.VSTMakerConfig.IgnoreDuplicateTokens = cintToBool(value)
 		break
+	case C.VARNAM_CONFIG_SET_DICTIONARY_SUGGESTIONS_LIMIT:
+		handle.varnam.DictionarySuggestionsLimit = int(value)
+		break
+	case C.VARNAM_CONFIG_SET_PATTERN_DICTIONARY_SUGGESTIONS_LIMIT:
+		handle.varnam.PatternDictionarySuggestionsLimit = int(value)
+		break
+	case C.VARNAM_CONFIG_SET_TOKENIZER_SUGGESTIONS_LIMIT:
+		handle.varnam.TokenizerSuggestionsLimit = int(value)
+		break
+	case C.VARNAM_CONFIG_SET_DICTIONARY_MATCH_EXACT:
+		handle.varnam.DictionaryMatchExact = cintToBool(value)
+		break
 	}
 
 	return C.VARNAM_SUCCESS

c-shared.h

@@ -8,10 +8,15 @@
 #define VARNAM_ERROR   2
 #define VARNAM_CANCELLED  3
 
-#define VARNAM_CONFIG_USE_DEAD_CONSONANTS			 100
-#define VARNAM_CONFIG_IGNORE_DUPLICATE_TOKEN	 101
-#define VARNAM_CONFIG_ENABLE_SUGGESTIONS			 102
-#define VARNAM_CONFIG_USE_INDIC_DIGITS				 103
+#define VARNAM_CONFIG_USE_DEAD_CONSONANTS 100
+#define VARNAM_CONFIG_IGNORE_DUPLICATE_TOKEN 101
+// VARNAM_CONFIG_ENABLE_SUGGESTIONS hasn't been implemented yet 
+#define VARNAM_CONFIG_ENABLE_SUGGESTIONS 102
+#define VARNAM_CONFIG_USE_INDIC_DIGITS 103
+#define VARNAM_CONFIG_SET_DICTIONARY_SUGGESTIONS_LIMIT 104
+#define VARNAM_CONFIG_SET_PATTERN_DICTIONARY_SUGGESTIONS_LIMIT 105
+#define VARNAM_CONFIG_SET_TOKENIZER_SUGGESTIONS_LIMIT 106
+#define VARNAM_CONFIG_SET_DICTIONARY_MATCH_EXACT 107
 
 typedef struct Suggestion_t {
   char* Word;

govarnam/constants.go

@@ -12,6 +12,12 @@ var (
 	VersionString string
 )
 
+// Go's struct int has default value 0.
+// For SearchSymbolTable usecase this is a problem.
+// Hence we use a constructor with default value setting.
+// https://stackoverflow.com/q/37135193/1372424
+const STRUCT_INT_DEFAULT_VALUE = -1
+
 /* General */
 const ZWNJ = "\u200c"
 const ZWJ = "\u200d"

govarnam/govarnam.go

@@ -469,6 +469,10 @@ func (varnam *Varnam) ReverseTransliterate(word string) ([]Suggestion, error) {
 
 	tokens := varnam.splitTextByConjunct(ctx, word)
 
+	if varnam.Debug {
+		fmt.Println(tokens)
+	}
+
 	for i, token := range tokens {
 		for j, symbol := range token.symbols {
 			tokens[i].symbols[j].Value1 = symbol.Pattern

govarnam/govarnam_ml_test.go

@@ -206,19 +206,25 @@ func TestMLAtomicChil(t *testing.T) {
 
 func TestMLReverseTransliteration(t *testing.T) {
 	varnam := getVarnamInstance("ml")
+	oldLimit := varnam.TokenizerSuggestionsLimit
+	varnam.TokenizerSuggestionsLimit = 30
 
 	sugs, err := varnam.ReverseTransliterate("മലയാളം")
 	checkError(err)
 
 	// The order of this will fail if VST weights change
-	expected := []string{"malayaaLam", "malayALam", "malayaalam", "malayAlam", "malayaLam", "malayalam"}
-	for i, sug := range sugs {
-		assertEqual(t, sug.Word, expected[i])
+	expected := []string{"malayaaLam", "malayaaLam_", "malayALam", "malayALam_", "malayaalam", "malayaalam_", "malayAlam", "malayAlam_", "malayaLam", "malayaLam_", "malayalam", "malayalam_"}
+
+	assertEqual(t, len(sugs), len(expected))
+	for i, expectedWord := range expected {
+		assertEqual(t, sugs[i].Word, expectedWord)
 	}
 
 	sugs, err = varnam.ReverseTransliterate("2019 ഏപ്രിൽ 17-ന് മലയാളം വിക്കിപീഡിയയിലെ ലേഖനങ്ങളുടെ എണ്ണം 63,000 പിന്നിട്ടു.")
 
 	assertEqual(t, sugs[0].Word, "2019 Epril 17-n~ malayaaLam vikkipeeDiyayile lEkhanangaLuTe eNNam 63,000 pinnittu.")
+
+	varnam.TokenizerSuggestionsLimit = oldLimit
 }
 
 func TestDictionaryLimit(t *testing.T) {
@@ -389,7 +395,7 @@ func TestMLExportAndImport(t *testing.T) {
 func TestMLSearchSymbolTable(t *testing.T) {
 	varnam := getVarnamInstance("ml")
 
-	var search Symbol
+	search := NewSearchSymbol()
 	search.Value1 = "ക"
 	results, err := varnam.SearchSymbolTable(context.Background(), search)
 	checkError(err)

govarnam/govarnam_test.go

@@ -17,12 +17,12 @@ var testTempDir string
 
 // AssertEqual checks if values are equal
 // Thanks https://gist.github.com/samalba/6059502#gistcomment-2710184
-func assertEqual(t *testing.T, a interface{}, b interface{}) {
-	if a == b {
+func assertEqual(t *testing.T, value interface{}, expected interface{}) {
+	if value == expected {
 		return
 	}
 	debug.PrintStack()
-	t.Errorf("Received %v (type %v), expected %v (type %v)", a, reflect.TypeOf(a), b, reflect.TypeOf(b))
+	t.Errorf("Received %v (type %v), expected %v (type %v)", value, reflect.TypeOf(value), expected, reflect.TypeOf(expected))
 }
 
 func checkError(err error) {
@@ -65,6 +65,9 @@ func getVarnamInstance(schemeID string) *Varnam {
 	if ok {
 		return instance
 	}
+
+	log.Fatalf("Varnam instance for %s not found", schemeID)
+
 	return nil
 }
 

govarnam/symbol.go

@@ -512,6 +512,23 @@ func removeNonExactTokens(tokens []Token) []Token {
 	return tokens
 }
 
+// NewSearchSymbol a constructor for making Symbol.
+// We're doing this because default int value in
+// go structs is 0. This won't work with searching
+// because fields can have 0 value.
+// https://stackoverflow.com/q/37135193/137242
+func NewSearchSymbol() Symbol {
+	symbol := Symbol{}
+	symbol.Identifier = STRUCT_INT_DEFAULT_VALUE
+	symbol.Type = STRUCT_INT_DEFAULT_VALUE
+	symbol.MatchType = STRUCT_INT_DEFAULT_VALUE
+	symbol.Weight = STRUCT_INT_DEFAULT_VALUE
+	symbol.Priority = STRUCT_INT_DEFAULT_VALUE
+	symbol.AcceptCondition = STRUCT_INT_DEFAULT_VALUE
+	symbol.Flags = STRUCT_INT_DEFAULT_VALUE
+	return symbol
+}
+
 // SearchSymbolTable For searching symbol table
 func (varnam *Varnam) SearchSymbolTable(ctx context.Context, searchCriteria Symbol) ([]Symbol, error) {
 	var (
@@ -534,7 +551,7 @@ func (varnam *Varnam) SearchSymbolTable(ctx context.Context, searchCriteria Symb
 				return
 			}
 		} else {
-			if val.(int) == 0 {
+			if valInt, ok := val.(int); !ok || valInt == STRUCT_INT_DEFAULT_VALUE {
 				return
 			}
 		}
@@ -587,7 +604,7 @@ func (varnam *Varnam) SearchSymbolTable(ctx context.Context, searchCriteria Symb
 }
 
 func (varnam *Varnam) getVirama() (string, error) {
-	var viramaSymbol Symbol
+	viramaSymbol := NewSearchSymbol()
 	viramaSymbol.Pattern = "~"
 	results, _ := varnam.SearchSymbolTable(context.Background(), viramaSymbol)
 

govarnam/vst_maker.go

@@ -225,7 +225,7 @@ func (varnam *Varnam) vmPersistToken(pattern string, value1 string, value2 strin
 		return fmt.Errorf("arguments invalid")
 	}
 
-	persisted, err := varnam.vmAlreadyPersisted(pattern, value1, matchType)
+	persisted, err := varnam.vmAlreadyPersisted(pattern, value1, matchType, acceptCondition)
 	if err != nil {
 		return err
 	}
@@ -260,9 +260,10 @@ func (varnam *Varnam) vmPersistToken(pattern string, value1 string, value2 strin
 	return nil
 }
 
-func (varnam *Varnam) vmAlreadyPersisted(pattern string, value1 string, matchType int) (bool, error) {
-	var searchCriteria Symbol
+func (varnam *Varnam) vmAlreadyPersisted(pattern string, value1 string, matchType int, acceptCondition int) (bool, error) {
+	searchCriteria := NewSearchSymbol()
 	searchCriteria.Pattern = pattern
+	searchCriteria.AcceptCondition = acceptCondition
 
 	if matchType == VARNAM_MATCH_EXACT {
 		searchCriteria.MatchType = matchType

govarnamgo/govarnamgo.go

@@ -633,6 +633,29 @@ func (handle *VarnamHandle) GetVSTPath() string {
 	return C.GoString(cStr)
 }
 
+func makeGoSymbol(cSymbol *C.Symbol) Symbol {
+	var goSymbol Symbol
+	goSymbol.Identifier = int(cSymbol.Identifier)
+	goSymbol.Type = int(cSymbol.Type)
+	goSymbol.MatchType = int(cSymbol.MatchType)
+	goSymbol.Pattern = C.GoString(cSymbol.Pattern)
+	goSymbol.Value1 = C.GoString(cSymbol.Value1)
+	goSymbol.Value2 = C.GoString(cSymbol.Value2)
+	goSymbol.Value3 = C.GoString(cSymbol.Value3)
+	goSymbol.Tag = C.GoString(cSymbol.Tag)
+	goSymbol.Weight = int(cSymbol.Weight)
+	goSymbol.Priority = int(cSymbol.Priority)
+	goSymbol.AcceptCondition = int(cSymbol.AcceptCondition)
+	goSymbol.Flags = int(cSymbol.Flags)
+	return goSymbol
+}
+
+func NewSearchSymbol() Symbol {
+	var resultPointer *C.Symbol
+	C.varnam_new_search_symbol(&resultPointer)
+	return makeGoSymbol(resultPointer)
+}
+
 // SearchSymbolTable search VST
 func (handle *VarnamHandle) SearchSymbolTable(ctx context.Context, searchCriteria Symbol) []Symbol {
 	var goResults []Symbol
@@ -671,21 +694,7 @@ func (handle *VarnamHandle) SearchSymbolTable(ctx context.Context, searchCriteri
 		for i < int(C.varray_length(resultPointer)) {
 			result := (*C.Symbol)(C.varray_get(resultPointer, C.int(i)))
 
-			var goResult Symbol
-			goResult.Identifier = int(result.Identifier)
-			goResult.Type = int(result.Type)
-			goResult.MatchType = int(result.MatchType)
-			goResult.Pattern = C.GoString(result.Pattern)
-			goResult.Value1 = C.GoString(result.Value1)
-			goResult.Value2 = C.GoString(result.Value2)
-			goResult.Value3 = C.GoString(result.Value3)
-			goResult.Tag = C.GoString(result.Tag)
-			goResult.Weight = int(result.Weight)
-			goResult.Priority = int(result.Priority)
-			goResult.AcceptCondition = int(result.AcceptCondition)
-			goResult.Flags = int(result.Flags)
-
-			goResults = append(goResults, goResult)
+			goResults = append(goResults, makeGoSymbol(result))
 			i++
 		}
 

govarnamgo/govarnamgo_ml_test.go

@@ -91,7 +91,7 @@ func TestRecentlyLearnedWords(t *testing.T) {
 func TestSearchSymbolTable(t *testing.T) {
 	varnam := getVarnamInstance("ml")
 
-	var symbol Symbol
+	symbol := NewSearchSymbol()
 	symbol.Pattern = "la"
 	result := varnam.SearchSymbolTable(context.Background(), symbol)