char *
to_utf8(const char *src, char *dest)
{
unsigned char *p = (unsigned char *) dest;
int len = 0, n;
while ((n=get_unichar(&src)) > 0) {
if (n >= 0x2000) {
*p++ = 0xe0 | (n >> 12);
*p++ = 0x80 | ((n >> 6) & 0x3f);
*p++ = 0x80 | (n & 0x3f);
} else if (n >= 0x80) {
int ivona_get_msgpart(struct dumbtts_conf *conf, SPDMessageType type,
char **msg, char *icon, char **buf, int *len,
int cap_mode, char *delimeters, int punct_mode,
char *punct_some)
{
int rc;
int isicon;
int n, pos, bytes;
wchar_t wc;
char xbuf[1024];
if (!*msg)
return 1;
if (!**msg)
return 1;
isicon = 0;
icon[0] = 0;
if (*buf)
**buf = 0;
log_msg(OTTS_LOG_INFO, "Ivona message %s type %d\n", *msg, type);
switch (type) {
case SPD_MSGTYPE_SOUND_ICON:
if (strlen(*msg) < 63) {
strcpy(icon, *msg);
rc = 0;
} else {
rc = 1;
}
*msg = NULL;
return rc;
case SPD_MSGTYPE_SPELL:
wc = get_unichar(msg);
if (!wc) {
*msg = NULL;
return 1;
}
n = dumbtts_WCharString(conf, wc, *buf, *len, cap_mode,
&isicon);
if (n > 0) {
*len = n + 128;
*buf = g_realloc(*buf, *len);
n = dumbtts_WCharString(conf, wc, *buf, *len, cap_mode,
&isicon);
}
if (n) {
*msg = NULL;
return 1;
}
if (isicon)
strcpy(icon, "capital");
return 0;
case SPD_MSGTYPE_KEY:
case SPD_MSGTYPE_CHAR:
if (type == SPD_MSGTYPE_KEY) {
n = dumbtts_KeyString(conf, *msg, *buf, *len, cap_mode,
&isicon);
} else {
n = dumbtts_CharString(conf, *msg, *buf, *len, cap_mode,
&isicon);
}
log_msg(OTTS_LOG_INFO, "Got n=%d", n);
if (n > 0) {
*len = n + 128;
*buf = g_realloc(*buf, *len);
if (type == SPD_MSGTYPE_KEY) {
n = dumbtts_KeyString(conf, *msg, *buf, *len,
cap_mode, &isicon);
} else {
n = dumbtts_CharString(conf, *msg, *buf, *len,
cap_mode, &isicon);
}
}
*msg = NULL;
if (!n && isicon)
strcpy(icon, "capital");
return n;
case SPD_MSGTYPE_TEXT:
pos = 0;
bytes =
module_get_message_part(*msg, xbuf, &pos, 1023, delimeters);
log_msg(OTTS_LOG_DEBUG, "Got bytes %d, %s", bytes, xbuf);
if (bytes <= 0) {
*msg = NULL;
return 1;
}
*msg += pos;
xbuf[bytes] = 0;
n = dumbtts_GetString(conf, xbuf, *buf, *len, punct_mode,
punct_some, ",.;:!?");
if (n > 0) {
*len = n + 128;
*buf = g_realloc(*buf, *len);
n = dumbtts_GetString(conf, xbuf, *buf, *len,
punct_mode, punct_some, ",.;:!?");
}
if (n) {
*msg = NULL;
return 1;
}
log_msg(OTTS_LOG_INFO, "Returning to Ivona |%s|", *buf);
return 0;
default:
*msg = NULL;
log_msg(OTTS_LOG_WARN, "Unknown message type\n");
return 1;
}
}
int mbrpipe_get_msgpart(void *conf, SPDMessageType type, char **msg,
char *icon, char **buf, int *len, int cap_mode,
char *delimiters, int punct_mode, char *punct_some)
{
int rc;
int isicon;
int n, bytes;
unsigned int pos;
wchar_t wc;
char xbuf[1024];
if (!*msg)
return 1;
if (!**msg)
return 1;
isicon = 0;
icon[0] = 0;
if (*buf)
**buf = 0;
log_msg(OTTS_LOG_DEBUG, MODULE_NAME": message %s type %d\n", *msg, type);
if (!dumb_conf) { /* No dumbtts library? */
pos=0;
bytes =
module_get_message_part(*msg, xbuf, &pos, 1023, delimiters);
log_msg(OTTS_LOG_DEBUG, MODULE_NAME": Got bytes %d, %s",
bytes, xbuf);
if (bytes <= 0) {
*msg = NULL;
return 1;
}
*msg += pos;
xbuf[bytes] = 0;
g_free(*buf);
if (strcasecmp(MbrpipeSynthEncoding,"utf-8")) {
*buf=g_convert(xbuf,-1,MbrpipeSynthEncoding,"UTF-8",NULL,NULL,NULL);
*len=strlen(*buf);
}
else {
*len=strlen(xbuf);
*buf=g_strdup(xbuf);
}
return 0;
}
switch (type) {
case SPD_MSGTYPE_SOUND_ICON:
if (strlen(*msg) < 63) {
strcpy(icon, *msg);
rc = 0;
} else {
rc = 1;
}
*msg = NULL;
return rc;
case SPD_MSGTYPE_SPELL:
wc = get_unichar(msg);
if (!wc) {
*msg = NULL;
return 1;
}
n = dumbtts_WCharString(conf, wc, *buf, *len, cap_mode,
&isicon);
if (n > 0) {
*len = n + 128;
*buf = g_realloc(*buf, *len);
n = dumbtts_WCharString(conf, wc, *buf, *len, cap_mode,
&isicon);
}
if (n) {
*msg = NULL;
return 1;
}
if (isicon)
strcpy(icon, "capital");
return 0;
case SPD_MSGTYPE_KEY:
case SPD_MSGTYPE_CHAR:
if (type == SPD_MSGTYPE_KEY) {
n = dumbtts_KeyString(conf, *msg, *buf, *len, cap_mode,
&isicon);
} else {
n = dumbtts_CharString(conf, *msg, *buf, *len, cap_mode,
&isicon);
}
log_msg(OTTS_LOG_DEBUG, MODULE_NAME": Got n=%d", n);
if (n > 0) {
*len = n + 128;
*buf = g_realloc(*buf, *len);
if (type == SPD_MSGTYPE_KEY) {
n = dumbtts_KeyString(conf, *msg, *buf, *len,
cap_mode, &isicon);
} else {
n = dumbtts_CharString(conf, *msg, *buf, *len,
cap_mode, &isicon);
}
}
*msg = NULL;
if (!n && isicon)
strcpy(icon, "capital");
return n;
case SPD_MSGTYPE_TEXT:
pos = 0;
bytes =
module_get_message_part(*msg, xbuf, &pos, 1023, delimiters);
log_msg(OTTS_LOG_DEBUG, MODULE_NAME": Got bytes %d, %s",
bytes, xbuf);
if (bytes <= 0) {
*msg = NULL;
return 1;
}
*msg += pos;
xbuf[bytes] = 0;
n = dumbtts_GetString(conf, xbuf, *buf, *len, punct_mode,
punct_some, ",.;:!?'");
if (n > 0) {
*len = n + 128;
*buf = g_realloc(*buf, *len);
n = dumbtts_GetString(conf, xbuf, *buf, *len,
punct_mode, punct_some, ",.;:!?'");
}
if (n) {
*msg = NULL;
return 1;
}
log_msg(OTTS_LOG_DEBUG, MODULE_NAME
": Returning to translator |%s|", *buf);
return 0;
default:
*msg = NULL;
log_msg(OTTS_LOG_WARN, MODULE_NAME": Unknown message type\n");
return 1;
}
}
static int milena_get_msgpart(char **msg,char *icon,char **buf,int *phmode)
{
int rc,wc,nlen,iscap,*iscp;char *c;
if (!*msg || !**msg) return 0;
iscp=NULL;
iscap=0;
log_msg(OTTS_LOG_DEBUG, MODULE_NAME": CAP %d", milena_cap_mode);
if (milena_cap_mode == 1) iscp=&iscap;
switch(milena_message_type) {
case SPD_MSGTYPE_SOUND_ICON:
if (strlen(*msg)<63) {
strcpy(icon,*msg);
rc=1;
}
else {
rc=0;
}
*msg=NULL;
return rc;
case SPD_MSGTYPE_SPELL:
wc=get_unichar(msg);
if (!wc) {
*msg=NULL;
return 0;
}
nlen=milena_wchar(milena,wc,NULL,0,iscp);
if (nlen>0) {
*buf=malloc(nlen);
milena_wchar(milena,wc,*buf,nlen,iscp);
*phmode=16;
}
else {
*buf=strdup("b\xb3\261d");
*phmode=2;
}
*phmode=(**msg)?1:16;
if (iscap) strcpy(icon,"capital");
return 1;
case SPD_MSGTYPE_KEY:
nlen=milena_key(milena,*msg,NULL,0,iscp);
if (nlen>0) {
*buf=malloc(nlen);
milena_key(milena,*msg,*buf,nlen,iscp);
*phmode=16;
}
else {
*buf=strdup("b\xb3\261d");
*phmode=2;
}
*msg=NULL;
if (iscap) strcpy(icon,"capital");
return 1;
case SPD_MSGTYPE_CHAR:
log_msg(OTTS_LOG_DEBUG, MODULE_NAME": CHAR [%s]\n",*msg);
wc=get_unichar(msg);
nlen=milena_wchar(milena,wc,NULL,0,iscp);
if (nlen>0) {
*buf=malloc(nlen);
milena_wchar(milena,wc,*buf,nlen,iscp);
*phmode=16;
}
else {
*buf=strdup("b\xb3\261d");
*phmode=2;
}
*msg=NULL;
if (iscap) strcpy(icon,"capital");
return 1;
case SPD_MSGTYPE_TEXT:
c=*msg;
nlen=milena_GetPhraseWithPunct(milena,msg,NULL,0,phmode,milena_punct,MilenaPunctuationSome);
if (nlen<=0) {
*msg=NULL;
return 0;
}
*msg=c;
*buf=malloc(nlen+1);
milena_GetPhraseWithPunct(milena,msg,*buf,nlen+1,phmode,milena_punct,MilenaPunctuationSome);
return 1;
default:
*msg=NULL;
log_msg(OTTS_LOG_WARN, MODULE_NAME": Unknown message type\n");
return 0;
}
return 0;
}
/*
* The scanner
*
*/
int mcy_lex(void)
{
static const WCHAR ustr_dot1[] = { '.', '\n', 0 };
static const WCHAR ustr_dot2[] = { '.', '\r', '\n', 0 };
static int isinit = 0;
int ch;
if(!isinit)
{
isinit++;
set_codepage(WMC_DEFAULT_CODEPAGE);
add_token(tok_keyword, ustr_codepages, tCODEPAGE, 0, NULL, 0);
add_token(tok_keyword, ustr_facility, tFACILITY, 0, NULL, 1);
add_token(tok_keyword, ustr_facilitynames, tFACNAMES, 0, NULL, 1);
add_token(tok_keyword, ustr_language, tLANGUAGE, 0, NULL, 1);
add_token(tok_keyword, ustr_languagenames, tLANNAMES, 0, NULL, 1);
add_token(tok_keyword, ustr_messageid, tMSGID, 0, NULL, 1);
add_token(tok_keyword, ustr_messageidtypedef, tTYPEDEF, 0, NULL, 1);
add_token(tok_keyword, ustr_outputbase, tBASE, 0, NULL, 1);
add_token(tok_keyword, ustr_severity, tSEVERITY, 0, NULL, 1);
add_token(tok_keyword, ustr_severitynames, tSEVNAMES, 0, NULL, 1);
add_token(tok_keyword, ustr_symbolicname, tSYMNAME, 0, NULL, 1);
add_token(tok_severity, ustr_error, 0x03, 0, NULL, 0);
add_token(tok_severity, ustr_warning, 0x02, 0, NULL, 0);
add_token(tok_severity, ustr_informational, 0x01, 0, NULL, 0);
add_token(tok_severity, ustr_success, 0x00, 0, NULL, 0);
add_token(tok_facility, ustr_application, 0xFFF, 0, NULL, 0);
add_token(tok_facility, ustr_system, 0x0FF, 0, NULL, 0);
add_token(tok_language, ustr_english, 0x409, 437, ustr_msg00001, 0);
}
empty_unichar_stack();
while(1)
{
if(want_line)
{
while((ch = get_unichar()) != '\n')
{
if(ch == EOF)
xyyerror("Unexpected EOF\n");
push_unichar(ch);
}
newline();
push_unichar(ch);
push_unichar(0);
if(!unistrcmp(ustr_dot1, get_unichar_stack()) || !unistrcmp(ustr_dot2, get_unichar_stack()))
{
want_line = 0;
/* Reset the codepage to our default after each message */
set_codepage(WMC_DEFAULT_CODEPAGE);
return tMSGEND;
}
mcy_lval.str = xunistrdup(get_unichar_stack());
return tLINE;
}
ch = get_unichar();
if(ch == EOF)
return EOF;
if(ch == '\n')
{
newline();
if(want_nl)
{
want_nl = 0;
return tNL;
}
continue;
}
if(isisochar(ch))
{
if(want_file)
{
int n = 0;
while(n < 8 && isisochar(ch))
{
int t = char_table[ch];
if((t & CH_PUNCT) || !(t & CH_SHORTNAME))
break;
push_unichar(ch);
n++;
ch = get_unichar();
}
unget_unichar(ch);
push_unichar(0);
want_file = 0;
mcy_lval.str = xunistrdup(get_unichar_stack());
return tFILE;
}
if(char_table[ch] & CH_IDENT)
{
token_t *tok;
while(isisochar(ch) && (char_table[ch] & (CH_IDENT|CH_NUMBER)))
{
push_unichar(ch);
ch = get_unichar();
}
unget_unichar(ch);
push_unichar(0);
if(!(tok = lookup_token(get_unichar_stack())))
{
mcy_lval.str = xunistrdup(get_unichar_stack());
return tIDENT;
}
switch(tok->type)
{
case tok_keyword:
return tok->token;
case tok_language:
codepage = tok->codepage;
/* Fall through */
case tok_severity:
case tok_facility:
mcy_lval.tok = tok;
return tTOKEN;
default:
internal_error(__FILE__, __LINE__, "Invalid token type encountered\n");
}
}
if(isspace(ch)) /* Ignore space */
continue;
if(isdigit(ch))
return scan_number(ch);
}
switch(ch)
{
case ':':
case '=':
case '+':
case '(':
case ')':
return ch;
case ';':
while(ch != '\n' && ch != EOF)
{
push_unichar(ch);
ch = get_unichar();
}
newline();
push_unichar(ch); /* Include the newline */
push_unichar(0);
mcy_lval.str = xunistrdup(get_unichar_stack());
return tCOMMENT;
default:
xyyerror("Invalid character '%c' (0x%04x)\n", isisochar(ch) && isprint(ch) ? ch : '.', ch);
}
}
}
/*
* Number scanner
*
* state | ch | next state
* ------+-----------------+--------------------------
* 0 | [0] | 1
* 0 | [1-9] | 4
* 0 | . | error (should never occur)
* 1 | [xX] | 2
* 1 | [0-7] | 3
* 1 | [89a-wyzA-WYZ_] | error invalid digit
* 1 | . | return 0
* 2 | [0-9a-fA-F] | 2
* 2 | [g-zG-Z_] | error invalid hex digit
* 2 | . | return (hex-number) if TOS != [xX] else error
* 3 | [0-7] | 3
* 3 | [89a-zA-Z_] | error invalid octal digit
* 3 | . | return (octal-number)
* 4 | [0-9] | 4
* 4 | [a-zA-Z_] | error invalid decimal digit
* 4 | . | return (decimal-number)
*
* All non-identifier characters [^a-zA-Z_0-9] terminate the scan
* and return the value. This is not entirely correct, but close
* enough (should check punctuators as trailing context, but the
* char_table is not adapted to that and it is questionable whether
* it is worth the trouble).
* All non-iso-8859-1 characters are an error.
*/
static int scan_number(int ch)
{
int state = 0;
int base = 10;
empty_char_stack();
while(1)
{
if(!isisochar(ch))
xyyerror("Invalid digit\n");
switch(state)
{
case 0:
if(isdigit(ch))
{
push_char(ch);
if(ch == '0')
state = 1;
else
state = 4;
}
else
internal_error(__FILE__, __LINE__, "Non-digit in first number-scanner state\n");
break;
case 1:
if(ch == 'x' || ch == 'X')
{
push_char(ch);
state = 2;
}
else if(ch >= '0' && ch <= '7')
{
push_char(ch);
state = 3;
}
else if(isalpha(ch) || ch == '_')
xyyerror("Invalid number digit\n");
else
{
unget_unichar(ch);
mcy_lval.num = 0;
return tNUMBER;
}
break;
case 2:
if(isxdigit(ch))
push_char(ch);
else if(isalpha(ch) || ch == '_' || !isxdigit(tos_char_stack()))
xyyerror("Invalid hex digit\n");
else
{
base = 16;
goto finish;
}
break;
case 3:
if(ch >= '0' && ch <= '7')
push_char(ch);
else if(isalnum(ch) || ch == '_')
xyyerror("Invalid octal digit\n");
else
{
base = 8;
goto finish;
}
break;
case 4:
if(isdigit(ch))
push_char(ch);
else if(isalnum(ch) || ch == '_')
xyyerror("Invalid decimal digit\n");
else
{
base = 10;
goto finish;
}
break;
default:
internal_error(__FILE__, __LINE__, "Invalid state in number-scanner\n");
}
ch = get_unichar();
}
finish:
unget_unichar(ch);
push_char(0);
mcy_lval.num = strtoul(get_char_stack(), NULL, base);
return tNUMBER;
}
