[1.43.02 release]

Language improvements include Danish rules.

Fixed phoneme [n^] for Klatt synthesizer, was producing noise burst.

espeakedit: Display error details after Compile Dictionary.
espeakedit: Fix: After speaking a word which changes language, Compile
    Dictionary may compile the wrong dictionary.

Don't consider multiple spaces as thousands separator (eg. 2  000 )

Lang=Hungarian, don't allow dot as thousands separator.
Lang-Hungarian, Fix suffixes -a, -e, -az, -ez, -azt, -ezt after numbers.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@239 d46cf337-b52f-0410-862d-fd96e6ae7743

src/compiledata.cpp

@@ -3156,7 +3156,7 @@ fprintf(f_errors,"\nRefs %d,  Reused %d\n",count_references,duplicate_references
 	report.Printf(_T("Compiled phonemes: %d errors."),error_count);
 	if(error_count > 0)
 	{
-		report += _T(" See file: 'error_log'.");
+		report += _T(" See file: 'phsource/error_log'.");
 		wxLogError(report);
 	}
 	wxLogStatus(report + report_dict);

src/compiledict.cpp

@@ -1007,6 +1007,11 @@ static char *compile_rule(char *input)
 			copy_rule_string(buf,state);
 			state = 3;
 			p = buf;
+			if(input[ix+1] == ' ')
+			{
+				fprintf(f_log,"%5d: Syntax error. Space after (\n",linenum);
+				error_count++;
+			}
 			break;
 			
 		case '\n':		// end of line

src/espeakedit.cpp

@@ -590,7 +590,36 @@ void MyFrame::OnTools(wxCommandEvent& event)
 		wxLogStatus(_T("Compiled '")+wxString(dictionary_name,wxConvLocal)+_T("', %d errors"),err);
 
 		if(log != NULL)
+		{
 			fclose(log);
+
+			if(err > 0)
+			{
+				// display the error messages
+				int len;
+				FILE *f;
+				char *msg;
+				wxString msg_string;
+
+				len = GetFileLength(fname_log);
+				if(len > 0)
+				{
+					if(len > 2000)
+						len = 2000;   // restrict length to pppppevent crash in wxLogMessage()
+					msg = (char *)malloc(len+1);
+					if(msg != NULL)
+					{
+						f = fopen(fname_log,"r");
+						fread(msg,len,1,f);
+						fclose(f);
+						msg[len] = 0;
+						msg_string = wxString(msg,wxConvUTF8);
+						wxLogMessage(msg_string);
+						free(msg);
+					}
+				}
+			}
+		}
 		break;
 
 	case MENU_FORMAT_DICTIONARY:

src/numbers.cpp

@@ -758,26 +758,26 @@ void SetSpellingStress(Translator *tr, char *phonemes, int control, int n_chars)
 static char ph_ordinal2[12];
 
 
-static int CheckDotOrdinal(Translator *tr, char *word, WORD_TAB *wtab, int roman)
-{//==============================================================================
+static int CheckDotOrdinal(Translator *tr, char *word, char *word_end, WORD_TAB *wtab, int roman)
+{//==============================================================================================
 
 	int ordinal = 0;
 	int c2;
 	int nextflags;
 
-	if((tr->langopts.numbers & NUM_ORDINAL_DOT) && ((word[0] == '.') || (wtab[0].flags & FLAG_HAS_DOT)) && !(wtab[1].flags & FLAG_NOSPACE))
+	if((tr->langopts.numbers & NUM_ORDINAL_DOT) && ((word_end[0] == '.') || (wtab[0].flags & FLAG_HAS_DOT)) && !(wtab[1].flags & FLAG_NOSPACE))
 	{
 		if(roman || !(wtab[1].flags & FLAG_FIRST_UPPER))
 		{
-			utf8_in(&c2, &word[2]);
-			if((word[1] != 0) && ((c2 == 0) || (wtab[0].flags & FLAG_COMMA_AFTER) || IsAlpha(c2)))
+			utf8_in(&c2, &word_end[2]);
+			if((word_end[1] != 0) && ((c2 == 0) || (wtab[0].flags & FLAG_COMMA_AFTER) || IsAlpha(c2)))
 			{
 				// ordinal number is indicated by dot after the number
 				// but not if the next word starts with an upper-case letter
             // (c2 == 0) is for cases such as, "2.,"
 				ordinal = 2;
-				if(word[0] == '.')
-					word[0] = ' ';
+				if(word_end[0] == '.')
+					word_end[0] = ' ';
 
 				if((roman==0) && (tr->translator_name == L('h','u')))
 				{
@@ -785,7 +785,7 @@ static int CheckDotOrdinal(Translator *tr, char *word, WORD_TAB *wtab, int roman
 					nextflags = 0;
 					if(IsAlpha(c2))
 					{
-						nextflags = TranslateWord(tr, &word[2], 0, NULL);
+						nextflags = TranslateWord(tr, &word_end[2], 0, NULL);
 					}
 
 if((tr->prev_dict_flags & FLAG_ALT_TRANS) && ((c2 == 0) || (wtab[0].flags & FLAG_COMMA_AFTER) || iswdigit(c2)))
@@ -793,8 +793,15 @@ if((tr->prev_dict_flags & FLAG_ALT_TRANS) && ((c2 == 0) || (wtab[0].flags & FLAG
 
 					if(nextflags & FLAG_ALT_TRANS)
 						ordinal = 0;
-					if((tr->prev_dict_flags & (FLAG_ALT_TRANS | FLAG_ALT3_TRANS)) && (nextflags & FLAG_ALT3_TRANS))
-						ordinal = 0x22;
+
+					if(nextflags & FLAG_ALT3_TRANS)
+					{
+						if(word[-2] == '-')
+							ordinal = 0;   // eg. december 2-5. között
+
+						if(tr->prev_dict_flags & (FLAG_ALT_TRANS | FLAG_ALT3_TRANS))
+							ordinal = 0x22;
+					}
 				}
 			}
 		}
@@ -816,6 +823,7 @@ int TranslateRoman(Translator *tr, char *word, char *ph_out, WORD_TAB *wtab)
 	int subtract;
 	int repeat = 0;
 	int n_digits = 0;
+	char *word_start;
 	unsigned int flags[2];
 	char ph_roman[30];
 	char number_chars[N_WORD_BYTES];
@@ -833,6 +841,7 @@ int TranslateRoman(Translator *tr, char *word, char *ph_out, WORD_TAB *wtab)
 	if((tr->langopts.numbers & NUM_ROMAN_CAPITALS) && !(wtab[0].flags & FLAG_ALL_UPPER))
 		return(0);
 
+	word_start = word;
 	while((c = *word++) != ' ')
 	{
 		if((p2 = strchr(roman_numbers,c)) == NULL)
@@ -886,7 +895,7 @@ int TranslateRoman(Translator *tr, char *word, char *ph_out, WORD_TAB *wtab)
 
 	sprintf(number_chars,"  %d    ",acc);
 
-	if(CheckDotOrdinal(tr, word, wtab, 1))
+	if(CheckDotOrdinal(tr, word_start, word, wtab, 1))
 		wtab[0].flags |= FLAG_ORDINAL;
 
 	if(tr->langopts.numbers & NUM_ROMAN_ORDINAL)
@@ -947,31 +956,34 @@ static int LookupThousands(Translator *tr, int value, int thousandplex, int thou
 	ph_of[0] = 0;
 
 	// first look for a match with the exact value of thousands
-	if(thousands_exact & 1)
+	if(value > 0)
 	{
-		if(thousands_exact & 2)
+		if(thousands_exact & 1)
 		{
-			// ordinal number
-			sprintf(string,"_%dM%do",value,thousandplex);
+			if(thousands_exact & 2)
+			{
+				// ordinal number
+				sprintf(string,"_%dM%do",value,thousandplex);
+				found_value = Lookup(tr, string, ph_thousands);
+			}
+			if(!found_value & (number_control & 1))
+			{
+				// look for the 'e' variant
+				sprintf(string,"_%dM%de",value,thousandplex);
+				found_value = Lookup(tr, string, ph_thousands);
+			}
+			if(!found_value)
+			{
+				// is there a different pronunciation if there are no hundreds,tens,or units ? (LANG=ta)
+				sprintf(string,"_%dM%dx",value,thousandplex);
+				found_value = Lookup(tr, string, ph_thousands);
+			}
+		}
+		if(found_value == 0)
+		{
+			sprintf(string,"_%dM%d",value,thousandplex);
 			found_value = Lookup(tr, string, ph_thousands);
 		}
-		if(!found_value & (number_control & 1))
-		{
-			// look for the 'e' variant
-			sprintf(string,"_%dM%de",value,thousandplex);
-			found_value = Lookup(tr, string, ph_thousands);
-		}
-		if(!found_value)
-		{
-			// is there a different pronunciation if there are no hundreds,tens,or units ? (LANG=ta)
-			sprintf(string,"_%dM%dx",value,thousandplex);
-			found_value = Lookup(tr, string, ph_thousands);
-		}
-	}
-	if(found_value == 0)
-	{
-		sprintf(string,"_%dM%d",value,thousandplex);
-		found_value = Lookup(tr, string, ph_thousands);
 	}
 
 	if(found_value == 0)
@@ -1504,7 +1516,7 @@ static int TranslateNumber_1(Translator *tr, char *word, char *ph_out, unsigned
 	if((tr->langopts.thousands_sep == ' ') || (tr->langopts.numbers & NUM_ALLOW_SPACE))
 	{
 		// thousands groups can be separated by spaces
-		if((n_digits == 3) && isdigit(word[-2]))
+		if((n_digits == 3) && !(wtab->flags & FLAG_MULTIPLE_SPACES) && isdigit(word[-2]))
 		{
 			prev_thousands = 1;
 		}
@@ -1520,7 +1532,7 @@ static int TranslateNumber_1(Translator *tr, char *word, char *ph_out, unsigned
 	if(prev_thousands || (word[0] != '0'))
 	{
 		// don't check for ordinal if the number has a leading zero
-		ordinal = CheckDotOrdinal(tr, &word[ix], wtab, 0);
+		ordinal = CheckDotOrdinal(tr, word, &word[ix], wtab, 0);
 	}
 
 	if((word[ix] == '.') && !isdigit(word[ix+1]) && !isdigit(word[ix+2]) && !(wtab[1].flags & FLAG_NOSPACE))
@@ -1627,8 +1639,9 @@ static int TranslateNumber_1(Translator *tr, char *word, char *ph_out, unsigned
 
 	if(tr->translator_name == L('h','u'))
 	{
-		// variant form of numbers when followed by hyphen and a suffix starting with 'a' or 'e'
-		if((wtab[thousandplex].flags & FLAG_HYPHEN_AFTER) && (thousands_exact==1) && ((word[suffix_ix] == 'a') || (word[suffix_ix] == 'e')))
+		// variant form of numbers when followed by hyphen and a suffix starting with 'a' or 'e' (buit not a, e, az, ez, azt, ezt
+		if((wtab[thousandplex].flags & FLAG_HYPHEN_AFTER) && (thousands_exact==1)
+			&& ((word[suffix_ix] == 'a') || (word[suffix_ix] == 'e')) && ((c = word[suffix_ix+1]) != ' ') && (c != 'z'))
 		{
 			number_control |= 1;  // use _1e variant of number
 		}

src/speak_lib.cpp

@@ -47,7 +47,6 @@
 #include "wave.h"
 
 unsigned char *outbuf=NULL;
-extern FILE *f_logespeak;
 extern espeak_VOICE voice_selected;
 
 espeak_EVENT *event_list=NULL;

src/synthdata.cpp

@@ -35,8 +35,8 @@
 #include "translate.h"
 #include "wave.h"
 
-const char *version_string = "1.43.03  24.Feb.10";
-const int version_phdata  = 0x014220;
+const char *version_string = "1.43.02  08.Mar.10";
+const int version_phdata  = 0x014300;
 
 int option_device_number = -1;
 FILE *f_logespeak = NULL;

src/tr_languages.cpp

@@ -598,7 +598,9 @@ Translator *SelectTranslator(const char *name)
 			tr->langopts.param[LOPT_ANNOUNCE_PUNCT] = 2;  // don't break clause before announcing . ? !
 			tr->langopts.param[LOPT_MIN_LONG_VOWEL] = 65;
 
-			tr->langopts.numbers = NUM_DECIMAL_COMMA | NUM_ALLOW_SPACE | NUM_DFRACTION_5 | NUM_ROMAN | NUM_ROMAN_ORDINAL | NUM_ROMAN_CAPITALS | NUM_ORDINAL_DOT | NUM_OMIT_1_HUNDRED;
+			tr->langopts.numbers = NUM_DFRACTION_5 | NUM_ALLOW_SPACE | NUM_ROMAN | NUM_ROMAN_ORDINAL | NUM_ROMAN_CAPITALS | NUM_ORDINAL_DOT | NUM_OMIT_1_HUNDRED;
+			tr->langopts.thousands_sep = ' ';   // don't allow dot as thousands separator
+			tr->langopts.decimal_sep = ',';
 			tr->langopts.min_roman = 1;
 			SetLetterVowel(tr,'y');
 			tr->langopts.spelling_stress = 1;

src/translate.cpp

@@ -1361,6 +1361,18 @@ static int TranslateWord2(Translator *tr, char *word, WORD_TAB *wtab, int pre_pa
 	int word_flags;
 	int word_copy_len;
 	char word_copy[N_WORD_BYTES+1];
+	char old_dictionary_name[40];
+
+	if((f_logespeak != NULL) && (logging_type & 8))
+	{
+		fprintf(f_logespeak,"WORD: flg=%.5x len=%d  '",wtab->flags,wtab->length);
+		for(ix=0; ix<40; ix++)
+		{
+			if(word[ix]==0) break;
+			fputc(word[ix], f_logespeak);
+		}
+		fprintf(f_logespeak,"'\n");
+	}
 
 	len = wtab->length;
 	if(len > 31) len = 31;
@@ -1550,6 +1562,7 @@ static int TranslateWord2(Translator *tr, char *word, WORD_TAB *wtab, int pre_pa
 			if(new_language[0]==0)
 				new_language = "en";
 
+			strcpy(old_dictionary_name, dictionary_name);
 			switch_phonemes = SetTranslator2(new_language);
 
 			if(switch_phonemes >= 0)
@@ -1784,6 +1797,7 @@ static int TranslateWord2(Translator *tr, char *word, WORD_TAB *wtab, int pre_pa
 	if(switch_phonemes >= 0)
 	{
 		// this word uses a different phoneme table, now switch back
+		strcpy(dictionary_name, old_dictionary_name);
 		SelectPhonemeTable(voice->phoneme_tab_ix);
 		SetPlist2(&ph_list2[n_ph_list2],phonSWITCH);
 		ph_list2[n_ph_list2++].tone_ph = voice->phoneme_tab_ix;  // original phoneme table number
@@ -2076,6 +2090,14 @@ void *TranslateClause(Translator *tr, FILE *f_text, const void *vp_input, int *t
 		charix[ix] = 0;
 	terminator = ReadClause(tr, f_text, source, charix, &charix_top, N_TR_SOURCE, &tone2);
 
+	if((f_logespeak != NULL) && (logging_type & 4))
+	{
+		fprintf(f_logespeak,"CLAUSE %x:\n",terminator);
+		for(p=source; *p != 0; p++)
+			fputc(*p, f_logespeak);
+		fprintf(f_logespeak,"ENDCLAUSE\n");
+	}
+
 	charix[charix_top+1] = 0;
 	charix[charix_top+2] = 0x7fff;
 	charix[charix_top+3] = 0;

src/translate.h

@@ -642,3 +642,5 @@ void InterpretPhoneme(Translator *tr, int control, PHONEME_LIST *plist, PHONEME_
 void InterpretPhoneme2(int phcode, PHONEME_DATA *phdata);
 
 extern FILE *f_trans;		// for logging
+extern FILE *f_logespeak;
+extern int logging_type;  // from config file