tokenizer.c: Tokenize line separator codepoints as newline tokens.

src/libespeak-ng/tokenizer.c

 
 #define ESPEAKNG_CTYPE_PROPERTY_MASK 0x0000000000000001ull
 
+// Reference: http://www.unicode.org/reports/tr14/tr14-32.html -- Unicode Line Breaking Algorithm
 static espeakng_CTYPE codepoint_type(uint32_t c)
 {
 	// 1. Detect and classify specific codepoints.
 	ucd_category cat = ucd_lookup_category(c);
 	switch (cat)
 	{
-	case UCD_CATEGORY_Zl: return ESPEAKNG_CTYPE_WHITESPACE;
+	case UCD_CATEGORY_Zl: return ESPEAKNG_CTYPE_NEWLINE;
 	case UCD_CATEGORY_Zp: return ESPEAKNG_CTYPE_WHITESPACE;
 	case UCD_CATEGORY_Zs: return ESPEAKNG_CTYPE_WHITESPACE;
 	}

tests/tokenizer.c

 	espeak_ng_TOKENIZER *tokenizer = create_tokenizer();
 	espeak_ng_TEXT_DECODER *decoder = create_text_decoder();
 
-	assert(text_decoder_decode_string(decoder, "\xC2\x85\xC2\x85", -1, ESPEAKNG_ENCODING_UTF_8) == ENS_OK);
+	assert(text_decoder_decode_string(decoder, "\xC2\x85\xC2\x85\xE2\x80\xA8\xE2\x80\xA8", -1, ESPEAKNG_ENCODING_UTF_8) == ENS_OK);
 	assert(tokenizer_reset(tokenizer, decoder) == 1);
 
 	// U+0085 : NEXT LINE (NEL) -- Used in EBCDIC systems as a combined CR+LF character.
 	assert(tokenizer_get_token_text(tokenizer) != NULL);
 	assert(strcmp(tokenizer_get_token_text(tokenizer), "\xC2\x85") == 0);
 
+	// General Category: Zl -- LINE SEPARATOR
+	assert(tokenizer_read_next_token(tokenizer) == ESPEAKNG_TOKEN_NEWLINE);
+	assert(tokenizer_get_token_text(tokenizer) != NULL);
+	assert(strcmp(tokenizer_get_token_text(tokenizer), "\xE2\x80\xA8") == 0);
+
+	assert(tokenizer_read_next_token(tokenizer) == ESPEAKNG_TOKEN_NEWLINE);
+	assert(tokenizer_get_token_text(tokenizer) != NULL);
+	assert(strcmp(tokenizer_get_token_text(tokenizer), "\xE2\x80\xA8") == 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');
 	espeak_ng_TOKENIZER *tokenizer = create_tokenizer();
 	espeak_ng_TEXT_DECODER *decoder = create_text_decoder();
 
-	assert(text_decoder_decode_string(decoder, "\t\t\n \xE3\x80\x80 \n\xC2\xA0\xC2\xA0\n\xE2\x80\xA8\n\xE2\x80\xA9", -1, ESPEAKNG_ENCODING_UTF_8) == ENS_OK);
+	assert(text_decoder_decode_string(decoder, "\t\t\n \xE3\x80\x80 \n\xC2\xA0\xC2\xA0\n\xE2\x80\xA9", -1, ESPEAKNG_ENCODING_UTF_8) == ENS_OK);
 	assert(tokenizer_reset(tokenizer, decoder) == 1);
 
 	// General Category: Cc, Property: White_Space
 	assert(tokenizer_get_token_text(tokenizer) != NULL);
 	assert(strcmp(tokenizer_get_token_text(tokenizer), "\n") == 0);
 
-	// General Category: Zl -- LINE SEPARATOR
-	assert(tokenizer_read_next_token(tokenizer) == ESPEAKNG_TOKEN_WHITESPACE);
-	assert(tokenizer_get_token_text(tokenizer) != NULL);
-	assert(strcmp(tokenizer_get_token_text(tokenizer), "\xE2\x80\xA8") == 0);
-
-	assert(tokenizer_read_next_token(tokenizer) == ESPEAKNG_TOKEN_NEWLINE);
-	assert(tokenizer_get_token_text(tokenizer) != NULL);
-	assert(strcmp(tokenizer_get_token_text(tokenizer), "\n") == 0);
-
 	// General Category: Zp -- PARAGRAPH SEPARATOR
 	assert(tokenizer_read_next_token(tokenizer) == ESPEAKNG_TOKEN_WHITESPACE);
 	assert(tokenizer_get_token_text(tokenizer) != NULL);