tokenizer.c: Support full stop tokens.

src/libespeak-ng/tokenizer.c

@@ -92,9 +92,10 @@ typedef enum {
 	ESPEAKNG_CTYPE_WHITESPACE,
 	ESPEAKNG_CTYPE_LOWERCASE,
 	ESPEAKNG_CTYPE_UPPERCASE,
+	ESPEAKNG_CTYPE_FULL_STOP,
 } espeakng_CTYPE;
 
-#define ESPEAKNG_CTYPE_PROPERTY_MASK 0x000000000000C001ull
+#define ESPEAKNG_CTYPE_PROPERTY_MASK 0x800000000000C001ull
 
 // Reference: http://www.unicode.org/reports/tr14/tr14-32.html -- Unicode Line Breaking Algorithm
 static espeakng_CTYPE codepoint_type(uint32_t c)
@@ -133,6 +134,8 @@ static espeakng_CTYPE codepoint_type(uint32_t c)
 		return ESPEAKNG_CTYPE_LOWERCASE;
 	case UCD_PROPERTY_OTHER_UPPERCASE:
 		return ESPEAKNG_CTYPE_UPPERCASE;
+	case ESPEAKNG_PROPERTY_FULL_STOP:
+		return ESPEAKNG_CTYPE_FULL_STOP;
 	}
 
 	// 4. Classify the remaining codepoints.
@@ -264,6 +267,10 @@ tokenizer_state_default(espeak_ng_TOKENIZER *tokenizer)
 	case ESPEAKNG_CTYPE_UPPERCASE:
 		current += utf8_out(c, current);
 		return tokenizer_read_word_token(tokenizer, current, ESPEAKNG_TOKEN_WORD_UPPERCASE);
+	case ESPEAKNG_CTYPE_FULL_STOP:
+		current += utf8_out(c, current);
+		*current = '\0';
+		return ESPEAKNG_TOKEN_FULL_STOP;
 	default:
 		current += utf8_out(c, current);
 		*current = '\0';

src/libespeak-ng/tokenizer.h

@@ -51,6 +51,7 @@ typedef enum
 	ESPEAKNG_TOKEN_WORD_LOWERCASE,
 	ESPEAKNG_TOKEN_WORD_MIXEDCASE,
 	ESPEAKNG_TOKEN_WORD_CAPITALIZED,
+	ESPEAKNG_TOKEN_FULL_STOP,
 } espeak_ng_TOKEN_TYPE;
 
 espeak_ng_TOKEN_TYPE

tests/tokenizer.c

@@ -457,6 +457,29 @@ test_Latn_word_tokens()
 	destroy_tokenizer(tokenizer);
 }
 
+void
+test_Latn_punctuation_tokens()
+{
+	printf("testing Latin (Latn) script punctuation tokens\n");
+
+	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(tokenizer_reset(tokenizer, decoder, ESPEAKNG_TOKENIZER_OPTION_TEXT) == 1);
+
+	assert(tokenizer_read_next_token(tokenizer) == ESPEAKNG_TOKEN_FULL_STOP);
+	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');
+
+	destroy_text_decoder(decoder);
+	destroy_tokenizer(tokenizer);
+}
+
 void
 run_tests()
 {
@@ -482,6 +505,7 @@ run_tests()
 	test_whitespace_tokens();
 
 	test_Latn_word_tokens();
+	test_Latn_punctuation_tokens();
 
 	printf("done\n");
 }
@@ -537,6 +561,9 @@ print_tokens(espeak_ng_TEXT_DECODER *decoder)
 	case ESPEAKNG_TOKEN_WORD_CAPITALIZED:
 		printf("word (capitalized) : %s\n", tokenizer_get_token_text(tokenizer));
 		break;
+	case ESPEAKNG_TOKEN_FULL_STOP:
+		printf("full stop          : %s\n", tokenizer_get_token_text(tokenizer));
+		break;
 	}
 }