tokenizer.c: Support semicolon tokens.

src/libespeak-ng/tokenizer.c

@@ -97,9 +97,10 @@ typedef enum {
 	ESPEAKNG_CTYPE_EXCLAMATION_MARK,
 	ESPEAKNG_CTYPE_COMMA,
 	ESPEAKNG_CTYPE_COLON,
+	ESPEAKNG_CTYPE_SEMICOLON,
 } espeakng_CTYPE;
 
-#define ESPEAKNG_CTYPE_PROPERTY_MASK 0xF80000000000C001ull
+#define ESPEAKNG_CTYPE_PROPERTY_MASK 0xFC0000000000C001ull
 
 // Reference: http://www.unicode.org/reports/tr14/tr14-32.html -- Unicode Line Breaking Algorithm
 static espeakng_CTYPE codepoint_type(uint32_t c)
@@ -148,6 +149,8 @@ static espeakng_CTYPE codepoint_type(uint32_t c)
 		return ESPEAKNG_CTYPE_COMMA;
 	case ESPEAKNG_PROPERTY_COLON:
 		return ESPEAKNG_CTYPE_COLON;
+	case ESPEAKNG_PROPERTY_SEMI_COLON:
+		return ESPEAKNG_CTYPE_SEMICOLON;
 	}
 
 	// 4. Classify the remaining codepoints.
@@ -299,6 +302,10 @@ tokenizer_state_default(espeak_ng_TOKENIZER *tokenizer)
 		current += utf8_out(c, current);
 		*current = '\0';
 		return ESPEAKNG_TOKEN_COLON;
+	case ESPEAKNG_CTYPE_SEMICOLON:
+		current += utf8_out(c, current);
+		*current = '\0';
+		return ESPEAKNG_TOKEN_SEMICOLON;
 	default:
 		current += utf8_out(c, current);
 		*current = '\0';

src/libespeak-ng/tokenizer.h

@@ -56,6 +56,7 @@ typedef enum
 	ESPEAKNG_TOKEN_EXCLAMATION_MARK,
 	ESPEAKNG_TOKEN_COMMA,
 	ESPEAKNG_TOKEN_COLON,
+	ESPEAKNG_TOKEN_SEMICOLON,
 } espeak_ng_TOKEN_TYPE;
 
 espeak_ng_TOKEN_TYPE

tests/tokenizer.c

@@ -465,7 +465,7 @@ test_Latn_punctuation_tokens()
 	espeak_ng_TOKENIZER *tokenizer = create_tokenizer();
 	espeak_ng_TEXT_DECODER *decoder = create_text_decoder();
 
-	assert(text_decoder_decode_string(decoder, ". ? ! , :", -1, ESPEAKNG_ENCODING_US_ASCII) == ENS_OK);
+	assert(text_decoder_decode_string(decoder, ". ? ! , : ;", -1, ESPEAKNG_ENCODING_US_ASCII) == ENS_OK);
 	assert(tokenizer_reset(tokenizer, decoder, ESPEAKNG_TOKENIZER_OPTION_TEXT) == 1);
 
 	assert(tokenizer_read_next_token(tokenizer) == ESPEAKNG_TOKEN_FULL_STOP);
@@ -504,6 +504,14 @@ test_Latn_punctuation_tokens()
 	assert(tokenizer_get_token_text(tokenizer) != NULL);
 	assert(strcmp(tokenizer_get_token_text(tokenizer), ":") == 0);
 
+	assert(tokenizer_read_next_token(tokenizer) == ESPEAKNG_TOKEN_WHITESPACE);
+	assert(tokenizer_get_token_text(tokenizer) != NULL);
+	assert(strcmp(tokenizer_get_token_text(tokenizer), " ") == 0);
+
+	assert(tokenizer_read_next_token(tokenizer) == ESPEAKNG_TOKEN_SEMICOLON);
+	assert(tokenizer_get_token_text(tokenizer) != NULL);
+	assert(strcmp(tokenizer_get_token_text(tokenizer), ";") == 0);
+
 	assert(tokenizer_read_next_token(tokenizer) == ESPEAKNG_TOKEN_END_OF_BUFFER);
 	assert(tokenizer_get_token_text(tokenizer) != NULL);
 	assert(*tokenizer_get_token_text(tokenizer) == '\0');
@@ -608,6 +616,9 @@ print_tokens(espeak_ng_TEXT_DECODER *decoder)
 	case ESPEAKNG_TOKEN_COLON:
 		printf("colon              : %s\n", tokenizer_get_token_text(tokenizer));
 		break;
+	case ESPEAKNG_TOKEN_SEMICOLON:
+		printf("semicolon          : %s\n", tokenizer_get_token_text(tokenizer));
+		break;
 	}
 }