encoding.c: Support determining the string length for length < 0.

src/libespeak-ng/encoding.c

@@ -17,8 +17,10 @@
 
 #include "config.h"
 
+#include <string.h>
 #include <stdint.h>
 #include <stdlib.h>
+#include <wchar.h>
 
 #include <espeak-ng/espeak_ng.h>
 
@@ -684,6 +686,8 @@ text_decoder_decode_string(espeak_ng_TEXT_DECODER *decoder,
 	if (enc->get == NULL)
 		return ENS_UNKNOWN_TEXT_ENCODING;
 
+	if (length < 0) length = string ? strlen(string) + 1 : 0;
+
 	decoder->get = string ? enc->get : null_decoder_getc;
 	decoder->codepage = enc->codepage;
 	decoder->current = (const uint8_t *)string;
@@ -704,6 +708,8 @@ text_decoder_decode_string_auto(espeak_ng_TEXT_DECODER *decoder,
 	if (enc->get == NULL)
 		return ENS_UNKNOWN_TEXT_ENCODING;
 
+	if (length < 0) length = string ? strlen(string) + 1 : 0;
+
 	decoder->get = string ? string_decoder_getc_auto : null_decoder_getc;
 	decoder->codepage = enc->codepage;
 	decoder->current = (const uint8_t *)string;
@@ -716,6 +722,8 @@ text_decoder_decode_wstring(espeak_ng_TEXT_DECODER *decoder,
                             const wchar_t *string,
                             int length)
 {
+	if (length < 0) length = string ? wcslen(string) + 1 : 0;
+
 	decoder->get = string ? string_decoder_getc_wchar : null_decoder_getc;
 	decoder->codepage = NULL;
 	decoder->current = (const uint8_t *)string;

src/libespeak-ng/readclause.c

@@ -1,7 +1,7 @@
 /*
  * Copyright (C) 2005 to 2015 by Jonathan Duddington
  * email: [email protected]
- * Copyright (C) 2015-2016 Reece H. Dunn
+ * Copyright (C) 2015-2017 Reece H. Dunn
  *
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by

src/libespeak-ng/translate.c

@@ -1,7 +1,7 @@
 /*
  * Copyright (C) 2005 to 2014 by Jonathan Duddington
  * email: [email protected]
- * Copyright (C) 2015-2016 Reece H. Dunn
+ * Copyright (C) 2015-2017 Reece H. Dunn
  *
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
@@ -1916,21 +1916,6 @@ int UpperCaseInWord(Translator *tr, char *word, int c)
 	return 0;
 }
 
-static inline espeak_ng_STATUS init_wstring_decoder(const wchar_t *text)
-{
-	return text_decoder_decode_wstring(p_decoder, text, wcslen(text) + 1);
-}
-
-static inline espeak_ng_STATUS init_string_decoder(const char *text, espeak_ng_ENCODING encoding)
-{
-	return text_decoder_decode_string(p_decoder, text, strlen(text) + 1, encoding);
-}
-
-static inline espeak_ng_STATUS init_string_decoder_auto(const char *text, espeak_ng_ENCODING encoding)
-{
-	return text_decoder_decode_string_auto(p_decoder, text, strlen(text) + 1, encoding);
-}
-
 const void *TranslateClause(Translator *tr, const void *vp_input, int *tone_out, char **voice_change)
 {
 	int ix;
@@ -1988,19 +1973,19 @@ const void *TranslateClause(Translator *tr, const void *vp_input, int *tone_out,
 	switch (option_multibyte)
 	{
 	case espeakCHARS_WCHAR:
-		init_wstring_decoder((const wchar_t *)vp_input);
+		text_decoder_decode_wstring(p_decoder, (const wchar_t *)vp_input, -1);
 		break;
 	case espeakCHARS_AUTO:
-		init_string_decoder_auto((const char *)vp_input, tr->encoding);
+		text_decoder_decode_string_auto(p_decoder, (const char *)vp_input, -1, tr->encoding);
 		break;
 	case espeakCHARS_UTF8:
-		init_string_decoder((const char *)vp_input, ESPEAKNG_ENCODING_UTF_8);
+		text_decoder_decode_string(p_decoder, (const char *)vp_input, -1, ESPEAKNG_ENCODING_UTF_8);
 		break;
 	case espeakCHARS_8BIT:
-		init_string_decoder((const char *)vp_input, tr->encoding);
+		text_decoder_decode_string(p_decoder, (const char *)vp_input, -1, tr->encoding);
 		break;
 	case espeakCHARS_16BIT:
-		init_string_decoder((const char *)vp_input, ESPEAKNG_ENCODING_ISO_10646_UCS_2);
+		text_decoder_decode_string(p_decoder, (const char *)vp_input, -1, ESPEAKNG_ENCODING_ISO_10646_UCS_2);
 		break;
 	default:
 		return NULL; // unknown multibyte option value

tests/encoding.c

@@ -736,11 +736,21 @@ test_char_decoder()
 	espeak_ng_TEXT_DECODER *decoder = create_text_decoder();
 
 	// null string
-	assert(text_decoder_decode_string(decoder, NULL, 5, ESPEAKNG_ENCODING_ISO_8859_1) == ENS_OK);
+	assert(text_decoder_decode_string(decoder, NULL, -1, ESPEAKNG_ENCODING_ISO_8859_1) == ENS_OK);
 	assert(text_decoder_eof(decoder) == 1);
 	assert(text_decoder_getc(decoder) == 0);
 	assert(text_decoder_eof(decoder) == 1);
 
+	// string length
+	assert(text_decoder_decode_string(decoder, "aG", -1, ESPEAKNG_ENCODING_ISO_8859_1) == ENS_OK);
+	assert(text_decoder_eof(decoder) == 0);
+	assert(text_decoder_getc(decoder) == 'a');
+	assert(text_decoder_eof(decoder) == 0);
+	assert(text_decoder_getc(decoder) == 'G');
+	assert(text_decoder_eof(decoder) == 0);
+	assert(text_decoder_getc(decoder) == 0);
+	assert(text_decoder_eof(decoder) == 1);
+
 	destroy_text_decoder(decoder);
 }
 
@@ -752,11 +762,12 @@ test_wchar_decoder()
 	espeak_ng_TEXT_DECODER *decoder = create_text_decoder();
 
 	// null string
-	assert(text_decoder_decode_wstring(decoder, NULL, 5) == ENS_OK);
+	assert(text_decoder_decode_wstring(decoder, NULL, -1) == ENS_OK);
 	assert(text_decoder_eof(decoder) == 1);
 	assert(text_decoder_getc(decoder) == 0);
 	assert(text_decoder_eof(decoder) == 1);
 
+	// wide-character string
 	assert(text_decoder_decode_wstring(decoder, L"aG\xA0\x2045", 4) == ENS_OK);
 	assert(text_decoder_eof(decoder) == 0);
 	assert(text_decoder_getc(decoder) == 'a');
@@ -768,6 +779,20 @@ test_wchar_decoder()
 	assert(text_decoder_getc(decoder) == 0x2045);
 	assert(text_decoder_eof(decoder) == 1);
 
+	// string length
+	assert(text_decoder_decode_wstring(decoder, L"aG\xA0\x2045", -1) == ENS_OK);
+	assert(text_decoder_eof(decoder) == 0);
+	assert(text_decoder_getc(decoder) == 'a');
+	assert(text_decoder_eof(decoder) == 0);
+	assert(text_decoder_getc(decoder) == 'G');
+	assert(text_decoder_eof(decoder) == 0);
+	assert(text_decoder_getc(decoder) == 0xA0);
+	assert(text_decoder_eof(decoder) == 0);
+	assert(text_decoder_getc(decoder) == 0x2045);
+	assert(text_decoder_eof(decoder) == 0);
+	assert(text_decoder_getc(decoder) == 0);
+	assert(text_decoder_eof(decoder) == 1);
+
 	destroy_text_decoder(decoder);
 }
 
@@ -779,7 +804,7 @@ test_auto_decoder()
 	espeak_ng_TEXT_DECODER *decoder = create_text_decoder();
 
 	// null string
-	assert(text_decoder_decode_string_auto(decoder, NULL, 5, ESPEAKNG_ENCODING_ISO_8859_1) == ENS_OK);
+	assert(text_decoder_decode_string_auto(decoder, NULL, -1, ESPEAKNG_ENCODING_ISO_8859_1) == ENS_OK);
 	assert(text_decoder_eof(decoder) == 1);
 	assert(text_decoder_getc(decoder) == 0);
 	assert(text_decoder_eof(decoder) == 1);
@@ -808,6 +833,16 @@ test_auto_decoder()
 	assert(text_decoder_getc(decoder) == 'f');
 	assert(text_decoder_eof(decoder) == 1);
 
+	// string length
+	assert(text_decoder_decode_string_auto(decoder, "aG", -1, ESPEAKNG_ENCODING_ISO_8859_1) == ENS_OK);
+	assert(text_decoder_eof(decoder) == 0);
+	assert(text_decoder_getc(decoder) == 'a');
+	assert(text_decoder_eof(decoder) == 0);
+	assert(text_decoder_getc(decoder) == 'G');
+	assert(text_decoder_eof(decoder) == 0);
+	assert(text_decoder_getc(decoder) == 0);
+	assert(text_decoder_eof(decoder) == 1);
+
 	destroy_text_decoder(decoder);
 }