mjohnson/espeak-ng
1
0
Fork 0
mirror of https://github.com/rhasspy/espeak-ng.git synced 2026-10-31 13:38:47 +00:00
Code Issues Packages Projects Releases Wiki Activity

encoding.c: Support the ISO 8859-1 encoding.

Browse Source
This commit is contained in:
Reece H. Dunn
2017-03-27 23:31:30 +01:00
parent 0590da5da7
commit 26bec1eedf
3 changed files with 67 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/include/espeak-ng/espeak_ng.h
+1
View File
@@ -190,6 +190,7 @@ typedef enum
{
ESPEAKNG_ENCODING_UNKNOWN,
ESPEAKNG_ENCODING_US_ASCII,
ESPEAKNG_ENCODING_ISO_8859_1,
} espeak_ng_ENCODING;
ESPEAK_NG_API espeak_ng_ENCODING
src/libespeak-ng/encoding.c
+32 -2
View File
@@ -30,12 +30,21 @@ MNEM_TAB mnem_encoding[] = {
{ "ANSI_X3.4-1968", ESPEAKNG_ENCODING_US_ASCII },
{ "ANSI_X3.4-1986", ESPEAKNG_ENCODING_US_ASCII },
{ "IBM367", ESPEAKNG_ENCODING_US_ASCII },
{ "IBM819", ESPEAKNG_ENCODING_ISO_8859_1 },
{ "ISO_646.irv:1991", ESPEAKNG_ENCODING_US_ASCII },
{ "ISO_8859-1", ESPEAKNG_ENCODING_ISO_8859_1 },
{ "ISO646-US", ESPEAKNG_ENCODING_US_ASCII },
{ "ISO-8859-1", ESPEAKNG_ENCODING_ISO_8859_1 },
{ "ISO-8859-1:1987", ESPEAKNG_ENCODING_ISO_8859_1 },
{ "US-ASCII", ESPEAKNG_ENCODING_US_ASCII },
{ "cp367", ESPEAKNG_ENCODING_US_ASCII },
{ "cp819", ESPEAKNG_ENCODING_ISO_8859_1 },
{ "csASCII", ESPEAKNG_ENCODING_US_ASCII },
{ "csISOLatin1", ESPEAKNG_ENCODING_ISO_8859_1 },
{ "iso-ir-6", ESPEAKNG_ENCODING_US_ASCII },
{ "iso-ir-100", ESPEAKNG_ENCODING_ISO_8859_1 },
{ "latin1", ESPEAKNG_ENCODING_ISO_8859_1 },
{ "l1", ESPEAKNG_ENCODING_ISO_8859_1 },
{ "us", ESPEAKNG_ENCODING_US_ASCII },
{ NULL, ESPEAKNG_ENCODING_UNKNOWN }
};
@@ -54,8 +63,25 @@ struct espeak_ng_TEXT_DECODER_
{
const char *current;
const char *end;
uint32_t (*get)(espeak_ng_TEXT_DECODER *decoder);
};
static uint32_t
text_decoder_getc_us_ascii(espeak_ng_TEXT_DECODER *decoder)
{
uint8_t c = *decoder->current++ & 0xFF;
return (c >= 0x80) ? 0xFFFD : c;
}
// Reference: http://www.iana.org/go/rfc1345
// Reference: http://www.unicode.org/Public/MAPPINGS/ISO8859/8859-1.TXT
static uint32_t
text_decoder_getc_iso_8859_1(espeak_ng_TEXT_DECODER *decoder)
{
return *decoder->current++ & 0xFF;
}
espeak_ng_TEXT_DECODER *
create_text_decoder(void)
{
@@ -64,6 +90,7 @@ create_text_decoder(void)
decoder->current = NULL;
decoder->end = NULL;
decoder->get = NULL;
return decoder;
}
@@ -80,6 +107,10 @@ initialize_encoding(espeak_ng_TEXT_DECODER *decoder,
switch (encoding)
{
case ESPEAKNG_ENCODING_US_ASCII:
decoder->get = text_decoder_getc_us_ascii;
break;
case ESPEAKNG_ENCODING_ISO_8859_1:
decoder->get = text_decoder_getc_iso_8859_1;
break;
default:
return 0;
@@ -110,6 +141,5 @@ text_decoder_eof(espeak_ng_TEXT_DECODER *decoder)
uint32_t
text_decoder_getc(espeak_ng_TEXT_DECODER *decoder)
{
uint8_t c = *decoder->current++ & 0xFF;
return (c >= 0x80) ? 0xFFFD : c;
return decoder->get(decoder);
}
tests/encoding.c
+34
View File
@@ -91,12 +91,46 @@ test_us_ascii_encoding()
destroy_text_decoder(decoder);
}
void
test_iso_8859_1_encoding()
{
printf("testing ISO-8859-1 encoding\n");
assert(espeak_ng_EncodingFromName("ISO-8859-1:1987") == ESPEAKNG_ENCODING_ISO_8859_1);
assert(espeak_ng_EncodingFromName("ISO-8859-1") == ESPEAKNG_ENCODING_ISO_8859_1);
assert(espeak_ng_EncodingFromName("ISO_8859-1") == ESPEAKNG_ENCODING_ISO_8859_1);
assert(espeak_ng_EncodingFromName("iso-ir-100") == ESPEAKNG_ENCODING_ISO_8859_1);
assert(espeak_ng_EncodingFromName("latin1") == ESPEAKNG_ENCODING_ISO_8859_1);
assert(espeak_ng_EncodingFromName("l1") == ESPEAKNG_ENCODING_ISO_8859_1);
assert(espeak_ng_EncodingFromName("IBM819") == ESPEAKNG_ENCODING_ISO_8859_1);
assert(espeak_ng_EncodingFromName("cp819") == ESPEAKNG_ENCODING_ISO_8859_1);
assert(espeak_ng_EncodingFromName("csISOLatin1") == ESPEAKNG_ENCODING_ISO_8859_1);
espeak_ng_TEXT_DECODER *decoder = create_text_decoder();
assert(text_decoder_decode_string(decoder, "aG\x92\xA0\xDE", 5, 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) == 0x92);
assert(text_decoder_eof(decoder) == 0);
assert(text_decoder_getc(decoder) == 0xA0);
assert(text_decoder_eof(decoder) == 0);
assert(text_decoder_getc(decoder) == 0xDE);
assert(text_decoder_eof(decoder) == 1);
destroy_text_decoder(decoder);
}
int
main(int argc, char **argv)
{
test_unbound_text_decoder();
test_unknown_encoding();
test_us_ascii_encoding();
test_iso_8859_1_encoding();
printf("done\n");
return EXIT_SUCCESS;