encoding.c: Create a string decoding API; support US-ASCII decoding.

src/include/espeak-ng/espeak_ng.h

@@ -60,6 +60,7 @@ typedef enum {
 
 	/* eSpeak NG 1.49.2 */
 	ENS_UNKNOWN_PHONEME_FEATURE  = 0x10000FFF,
+	ENS_UNKNOWN_TEXT_ENCODING    = 0x100010FF,
 } espeak_ng_STATUS;
 
 typedef enum {

src/libespeak-ng/encoding.c

@@ -52,6 +52,8 @@ espeak_ng_EncodingFromName(const char *encoding)
 
 struct espeak_ng_TEXT_DECODER_
 {
+	const char *current;
+	const char *end;
 };
 
 espeak_ng_TEXT_DECODER *
@@ -60,6 +62,8 @@ create_text_decoder(void)
 	espeak_ng_TEXT_DECODER *decoder = malloc(sizeof(espeak_ng_TEXT_DECODER));
 	if (!decoder) return NULL;
 
+	decoder->current = NULL;
+	decoder->end = NULL;
 	return decoder;
 }
 
@@ -69,14 +73,43 @@ destroy_text_decoder(espeak_ng_TEXT_DECODER *decoder)
 	if (decoder) free(decoder);
 }
 
+static int
+initialize_encoding(espeak_ng_TEXT_DECODER *decoder,
+                    espeak_ng_ENCODING encoding)
+{
+	switch (encoding)
+	{
+	case ESPEAKNG_ENCODING_US_ASCII:
+		break;
+	default:
+		return 0;
+	}
+	return 1;
+}
+
+espeak_ng_STATUS
+text_decoder_decode_string(espeak_ng_TEXT_DECODER *decoder,
+                           const char *string,
+                           int length,
+                           espeak_ng_ENCODING encoding)
+{
+	if (!initialize_encoding(decoder, encoding))
+		return ENS_UNKNOWN_TEXT_ENCODING;
+
+	decoder->current = string;
+	decoder->end = string + length;
+	return ENS_OK;
+}
+
 int
 text_decoder_eof(espeak_ng_TEXT_DECODER *decoder)
 {
-	return 1;
+	return decoder->current == decoder->end;
 }
 
 uint32_t
 text_decoder_getc(espeak_ng_TEXT_DECODER *decoder)
 {
-	return 0;
+	uint8_t c = *decoder->current++ & 0xFF;
+	return (c >= 0x80) ? 0xFFFD : c;
 }

src/libespeak-ng/encoding.h

@@ -30,6 +30,12 @@ create_text_decoder(void);
 void
 destroy_text_decoder(espeak_ng_TEXT_DECODER *decoder);
 
+espeak_ng_STATUS
+text_decoder_decode_string(espeak_ng_TEXT_DECODER *decoder,
+                           const char *string,
+                           int length,
+                           espeak_ng_ENCODING encoding);
+
 int
 text_decoder_eof(espeak_ng_TEXT_DECODER *decoder);
 

src/libespeak-ng/error.c

@@ -135,6 +135,9 @@ espeak_ng_GetStatusCodeMessage(espeak_ng_STATUS status,
 	case ENS_UNKNOWN_PHONEME_FEATURE:
 		strncpy0(buffer, "The phoneme feature is not recognised", length);
 		break;
+	case ENS_UNKNOWN_TEXT_ENCODING:
+		strncpy0(buffer, "The text encoding is not supported", length);
+		break;
 	default:
 		if ((status & ENS_GROUP_MASK) == ENS_GROUP_ERRNO)
 			strerror_r(status, buffer, length);

tests/encoding.c

@@ -35,7 +35,6 @@ test_unbound_text_decoder()
 
 	assert(decoder != NULL);
 	assert(text_decoder_eof(decoder) == 1);
-	assert(text_decoder_getc(decoder) == 0);
 
 	destroy_text_decoder(decoder);
 }
@@ -49,6 +48,13 @@ test_unknown_encoding()
 	assert(espeak_ng_EncodingFromName("") == ESPEAKNG_ENCODING_UNKNOWN);
 	assert(espeak_ng_EncodingFromName("abcxyz") == ESPEAKNG_ENCODING_UNKNOWN);
 	assert(espeak_ng_EncodingFromName("US") == ESPEAKNG_ENCODING_UNKNOWN); // wrong case
+
+	espeak_ng_TEXT_DECODER *decoder = create_text_decoder();
+
+	assert(text_decoder_decode_string(decoder, "aG\x92\xA0\xDE", 5, ESPEAKNG_ENCODING_UNKNOWN) == ENS_UNKNOWN_TEXT_ENCODING);
+	assert(text_decoder_eof(decoder) == 1);
+
+	destroy_text_decoder(decoder);
 }
 
 void
@@ -66,6 +72,23 @@ test_us_ascii_encoding()
 	assert(espeak_ng_EncodingFromName("IBM367") == ESPEAKNG_ENCODING_US_ASCII);
 	assert(espeak_ng_EncodingFromName("cp367") == ESPEAKNG_ENCODING_US_ASCII);
 	assert(espeak_ng_EncodingFromName("csASCII") == ESPEAKNG_ENCODING_US_ASCII);
+
+	espeak_ng_TEXT_DECODER *decoder = create_text_decoder();
+
+	assert(text_decoder_decode_string(decoder, "aG\x92\xA0\xDE", 5, ESPEAKNG_ENCODING_US_ASCII) == 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) == 0xFFFD);
+	assert(text_decoder_eof(decoder) == 0);
+	assert(text_decoder_getc(decoder) == 0xFFFD);
+	assert(text_decoder_eof(decoder) == 0);
+	assert(text_decoder_getc(decoder) == 0xFFFD);
+	assert(text_decoder_eof(decoder) == 1);
+
+	destroy_text_decoder(decoder);
 }
 
 int