[1.41.17]

espeakedit: add "format *_rules file" feature.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@228 d46cf337-b52f-0410-862d-fd96e6ae7743

src/espeakedit.cpp

@@ -59,6 +59,7 @@ const char *path_data;
 extern void TestTest(int control);
 extern void CompareLexicon(int);
 extern void ConvertToUtf8();
+extern void FormatDictionary(const char *dictname);
 
 extern void init_z();
 extern void CompileInit(void);
@@ -73,6 +74,10 @@ extern void MakeWordFreqList();
 
 extern const char *dict_name;
 
+extern wxMenu *speak_menu;
+extern wxMenu *data_menu;
+
+
 MyFrame *myframe = NULL;
 SpectDisplay *currentcanvas = NULL;
 
@@ -158,21 +163,21 @@ extern void VoiceReset(int control);
 
 
   // Create the main frame window
-  myframe = new MyFrame(NULL, -1, AppName, wxPoint(0, 0), wxSize(1024, 768),
+	myframe = new MyFrame(NULL, -1, AppName, wxPoint(0, 0), wxSize(1024, 768),
                       wxDEFAULT_FRAME_STYLE |
                       wxNO_FULL_REPAINT_ON_RESIZE |
                       wxHSCROLL | wxVSCROLL);
 
 
-  // Make a menubar
-  myframe->SetMenuBar(MakeMenu(0));
+	// Make a menubar
+	myframe->SetMenuBar(MakeMenu(0));
+	myframe->CreateStatusBar();
+	myframe->SetVoiceTitle(voice_name2);
 
-  myframe->CreateStatusBar();
+	myframe->Maximize();
+	myframe->Show(TRUE);
 
-  myframe->Maximize();
-  myframe->Show(TRUE);
-
-  SetTopWindow(myframe);
+	SetTopWindow(myframe);
 	wxInitAllImageHandlers();
 //	wxImage::AddHandler(wxPNGHandler);
   return TRUE;
@@ -194,6 +199,7 @@ BEGIN_EVENT_TABLE(MyFrame, wxMDIParentFrame)
    EVT_MENU(MENU_COMPILE_PH, MyFrame::OnTools)
 	EVT_MENU(MENU_COMPILE_DICT, MyFrame::OnTools)
 	EVT_MENU(MENU_COMPILE_DICT_DEBUG, MyFrame::OnTools)
+	EVT_MENU(MENU_FORMAT_DICTIONARY, MyFrame::OnTools)
 	EVT_MENU(MENU_COMPILE_MBROLA, MyFrame::OnTools)
 	EVT_MENU(MENU_CLOSE_ALL, MyFrame::OnQuit)
 	EVT_MENU(MENU_QUIT, MyFrame::OnQuit)
@@ -284,10 +290,6 @@ wxSashLayoutWindow *win;
 			wxLogError(_T("Failed to load voice data"));
 		strcpy(dictionary_name,"en");
 	}
-	else
-	{
-		SetVoiceTitle(voice_name2);
-	}
 	WavegenSetVoice(voice);
 
 	for(param=0; param<N_SPEECH_PARAM; param++)
@@ -312,7 +314,15 @@ wxSashLayoutWindow *win;
 
 void MyFrame::SetVoiceTitle(char *voice_name)
 {//==========================================
+	char buf[100];
 	SetTitle(AppName + _T(" - ") + wxString(voice_name,wxConvLocal) + _T("  voice"));
+	if(data_menu != NULL)
+	{
+		sprintf(buf,"Compile &dictionary '%s'",dictionary_name);
+		data_menu->SetLabel(MENU_COMPILE_DICT, wxString(buf,wxConvLocal));
+		sprintf(buf,"&Format '%s_rules' file",dictionary_name);
+		data_menu->SetLabel(MENU_FORMAT_DICTIONARY, wxString(buf,wxConvLocal));
+	}
 }
 
 
@@ -605,6 +615,10 @@ void MyFrame::OnTools(wxCommandEvent& event)
 			fclose(log);
 		break;
 
+	case MENU_FORMAT_DICTIONARY:
+		FormatDictionary(dictionary_name);
+		break;
+
 	case MENU_VOWELCHART1:
 		MakeVowelLists();
 		break;
@@ -623,8 +637,6 @@ void MyFrame::OnTools(wxCommandEvent& event)
 	}
 }
 
-extern wxMenu *speak_menu;
-
 void MyFrame::OnSpeak(wxCommandEvent& event)
 {//=========================================
 	switch(event.GetId())

src/extras.cpp

@@ -1312,6 +1312,263 @@ void ConvertToUtf8()
 
 //******************************************************************************************************
 
+void FormatDictionary(const char *dictname)
+{//========================================
+// Format the *_rules file for the current voice
+
+	FILE *f_in;
+	FILE *f_out;
+	char *p;
+	char *p_start;
+	unsigned short *pw;
+	unsigned short *pw_match;
+	unsigned short *pw_post;
+	unsigned short *pw_phonemes;
+	int c;
+	int ix;
+	int n_pre;
+	int n_match;
+	int n_post;
+	int n_phonemes;
+	int n_spaces;
+	int n_out;
+	int formatting;
+	int comment;
+	char buf[200];
+	unsigned short bufw[200];
+	char conditional[80];
+	char fname_in[200];
+	char fname_out[200];
+
+	const int tab1 = 8;
+	const int tab2 = 18;
+	const int tab3 = 28;
+
+	sprintf(fname_in,"%s/%s_rules",path_dsource,dictname);
+	sprintf(fname_out,"%s_formatted",fname_in);
+
+	if((f_in = fopen(fname_in,"r")) == NULL)
+	{
+		wxLogError(_T("Can't open rules file: ") + wxString(fname_in,wxConvLocal));
+		return;
+	}
+	if((f_out = fopen(fname_out,"w")) == NULL)
+	{
+		wxLogError(_T("Can't write to file: ") + wxString(fname_out,wxConvLocal));
+		fclose(f_in);
+		return;
+	}
+
+	formatting = 0;
+	n_match = 0;
+	while(fgets(buf, sizeof(buf)-1, f_in) != NULL)
+	{
+		buf[sizeof(buf)-1] = 0;  // ensure zero byte terminator
+
+		ix = strlen(buf) - 1;
+		while((buf[ix]=='\n') || (buf[ix]==' ') || (buf[ix]=='\t')) ix--;
+		buf[ix+1] = 0;  // strip tailing spaces
+
+		p_start = buf;
+		while((*p_start==' ') || (*p_start == '\t')) p_start++;  // skip leading spaces
+
+		comment = 0;
+		if((p_start[0]=='/') && (p_start[1]=='/'))
+			comment = 1;
+
+		ix = 0;
+		if(*p_start == '?')
+		{
+			// conditional rule
+			while(!isspace(*p_start) && (*p_start != 0))
+			{
+				conditional[ix++] = *p_start++;
+			}
+			while((*p_start == ' ') || (*p_start == '\t')) p_start++;
+		}
+		conditional[ix] = 0;
+
+		if(buf[0] == '.')
+		{
+			formatting = 0;
+		}
+		if(memcmp(p_start, ".group", 6) == 0)
+		{
+			formatting = 2;
+			if(n_match > 0)
+			{
+				// previous line was not blank, so add a blank line
+				fprintf(f_out,"\n");
+			}
+		}
+
+		n_match = 0;
+
+		if((formatting == 1) && (comment==0))
+		{
+			// convert from UTF-8 to UTF-16
+			p = p_start;
+			pw = bufw;
+			do {
+				p += utf8_in(&c, p);
+				*pw++ = c;
+			} while (c != 0);
+
+			pw = bufw;
+
+			while((*pw != ')') && (*pw != 0) && !iswspace(*pw)) pw++;
+
+			n_pre = 0;
+			n_post = 0;
+			n_phonemes = 0;
+			n_spaces = 0;
+
+			if(*pw != 0)
+				n_spaces = tab1;
+
+			if(*pw == ')')
+			{
+				// there is a pre-condition
+				n_pre = pw - bufw + 1;
+				n_spaces = tab1 - n_pre - 1;
+				pw++;
+				while((*pw==' ') || (*pw=='\t')) pw++;
+			}
+			else
+			{
+				pw = bufw;
+			}
+
+			pw_match = pw;
+			while(((c = *pw)!= ' ') && (c != '\t') && (c != '(') && (c != 0))
+			{
+				pw++;
+			}
+			n_match = pw - pw_match;
+
+			while(((c = *pw)==' ') || (c == '\t')) pw++;
+
+			if(*pw == '(')
+			{
+				pw_post = pw;
+				while(((c = *pw)!=' ') && (c != '\t') && (c != 0)) pw++;
+				n_post = pw - pw_post;
+				while(((c = *pw)==' ') || (c == '\t')) pw++;
+			}
+
+			if((*pw != 0) && ((*pw != '/') || (pw[1] != '/')))
+			{
+				pw_phonemes = pw;
+				while(((c = *pw)!=' ') && (c != '\t') && (c != 0)) pw++;
+				n_phonemes = pw - pw_phonemes;
+				while(((c = *pw)==' ') || (c == '\t')) pw++;
+			}
+
+			// write formatted line
+			p = buf;
+
+			if(conditional[0] != 0)
+			{
+				ix = 0;
+				while(conditional[ix] != 0)
+				{
+					*p++ = conditional[ix++];
+					n_spaces--;
+				}
+				*p++ = ' ';
+				n_spaces--;
+			}
+
+			while(n_spaces-- > 0)
+			{
+				*p++ = ' ';
+			}
+			if(n_pre > 0)
+			{
+				ix = 0;
+				for(ix=0; ix<n_pre; ix++)
+				{
+					p += utf8_out(bufw[ix], p);
+				}
+				*p++ = ' ';
+			}
+
+			// write the match condition
+			if(n_match > 0)
+			{
+				ix = 0;
+				for(ix=0; ix<n_match; ix++)
+				{
+					p += utf8_out(pw_match[ix], p);
+				}
+				*p++ = ' ';
+			}
+
+			// write the post condition
+			if(n_post > 0)
+			{
+				for(ix=0; ix<n_post; ix++)
+				{
+					p += utf8_out(pw_post[ix], p);
+				}
+				*p++ = ' ';
+				n_post++;
+			}
+
+			n_out = tab1 + n_match + n_post;
+			if(n_pre >= tab1)
+				n_out += (n_pre - tab1 + 1);
+
+			if(n_phonemes > 0)
+			{
+				n_spaces = tab2 - n_out;
+
+				while(n_spaces-- > 0)
+				{
+					*p++ = ' ';
+					n_out++;
+				}
+
+			// write the phoneme codes
+				for(ix=0; ix<n_phonemes; ix++)
+				{
+					p += utf8_out(pw_phonemes[ix], p);
+				}
+			}
+
+			if(*pw != 0)
+			{
+				*p++ = ' ';
+				n_phonemes++;
+				n_spaces = tab3 - (n_out + n_phonemes);
+
+				while(n_spaces-- > 0)
+				{
+					*p++ = ' ';
+				}
+			}
+
+			// write the remainer of the line
+			while(*pw != 0)
+			{
+				p+= utf8_out(*pw++, p);
+			}
+			*p = 0;
+		}
+
+			if(formatting > 1)
+				formatting--;
+
+		fprintf(f_out, "%s\n", buf);
+	}
+
+	fclose(f_in);
+	fclose(f_out);
+	wxLogStatus(_("Written to file: ") + wxString(fname_out,wxConvLocal));
+}  // end of FormatDictionary
+
+
+//******************************************************************************************************
 
 
 //#define calcspeedtab

src/main.h

@@ -51,6 +51,7 @@ class MyFrame: public wxMDIParentFrame
 
     void OnNewWindow(wxCommandEvent& event);
     void OnProsody(wxCommandEvent& event);
+	void SetVoiceTitle(char *voice_name);
 
 
 protected:
@@ -58,7 +59,6 @@ protected:
     wxSashLayoutWindow* m_leftWindow1;
     wxSashLayoutWindow* m_leftWindow2;
     wxSashLayoutWindow* m_bottomWindow;
-	void SetVoiceTitle(char *voice_name);
     wxTimer m_timer;
 
 DECLARE_EVENT_TABLE()
@@ -135,6 +135,7 @@ enum {
 	MENU_COMPILE_PH,
 	MENU_COMPILE_DICT,
 	MENU_COMPILE_DICT_DEBUG,
+	MENU_FORMAT_DICTIONARY,
 	MENU_COMPILE_MBROLA,
 
 	MENU_SPEAK_TRANSLATE,

src/menus.cpp

@@ -24,6 +24,7 @@
 
 
 wxMenu *speak_menu;
+wxMenu *data_menu = NULL;
 
 wxMenuBar *MakeMenu(int type)
 {//==========================
@@ -32,7 +33,6 @@ wxMenuBar *MakeMenu(int type)
 	wxMenu *help_menu;
 	wxMenu *option_menu = NULL;
 	wxMenu *paths_menu;
-	wxMenu *data_menu = NULL;
 	wxMenu *tools_menu;
 	wxMenu *voice_menu;
 	wxMenu *vowelchart_menu;
@@ -86,8 +86,10 @@ wxMenuBar *MakeMenu(int type)
 	data_menu = new wxMenu;
 	data_menu->Append(MENU_COMPILE_PH, _("Compile &phoneme data"));
 	data_menu->Append(MENU_COMPILE_DICT, _("Compile &dictionary"));
-	data_menu->Append(MENU_COMPILE_DICT_DEBUG, _("Compile dictionary (&debug)"));
+	data_menu->Append(MENU_COMPILE_DICT_DEBUG, _("Compile dictionary (debu&g)"));
 	data_menu->Append(MENU_COMPILE_MBROLA, _("Compile &mbrola phonemes list..."));
+	data_menu->AppendSeparator();
+	data_menu->Append(MENU_FORMAT_DICTIONARY, _("&Format *_rules file"));
 	
 	// OPTIONS MENU
 	paths_menu = new wxMenu;

src/numbers.cpp

@@ -751,8 +751,40 @@ void SetSpellingStress(Translator *tr, char *phonemes, int control, int n_chars)
 
 static char ph_ordinal2[12];
 
-int TranslateRoman(Translator *tr, char *word, char *ph_out)
-{//=====================================================
+
+static int CheckDotOrdinal(Translator *tr, char *word, WORD_TAB *wtab, int lowercase)
+{//==================================================================================
+// nextupper:  Next word must start with lower-case
+
+	int ordinal = 0;
+	int c2;
+
+	if((tr->langopts.numbers & NUM_ORDINAL_DOT) && (word[0] == '.'))
+	{
+		if((lowercase==0) || !(wtab[1].flags & FLAG_FIRST_UPPER))
+		{
+			utf8_in(&c2, &word[2]);
+			if(IsAlpha(c2))
+			{
+				// ordinal number is indicated by dot after the number
+				// but not if the next word starts with an upper-case letter
+				ordinal = 2;
+				word[0] = ' ';
+				if(tr->translator_name == L('h','u'))
+				{
+					// lang=hu don't treat dot as ordinal indicator if the next word is a month name ($alt)
+					if(TranslateWord(tr, &word[2], 0, NULL) & FLAG_ALT_TRANS)
+						ordinal = 0;
+				}
+			}
+		}
+	}
+	return(ordinal);
+}  // end of CheckDotOrdinal
+
+
+int TranslateRoman(Translator *tr, char *word, char *ph_out, WORD_TAB *wtab)
+{//=========================================================================
 	int c;
 	char *p;
 	const char *p2;
@@ -761,7 +793,7 @@ int TranslateRoman(Translator *tr, char *word, char *ph_out)
 	int value;
 	int subtract;
 	int repeat = 0;
-	WORD_TAB wtab[2];
+	int n_digits = 0;
 	unsigned int flags;
 	char ph_roman[30];
 	char number_chars[N_WORD_BYTES];
@@ -772,7 +804,9 @@ int TranslateRoman(Translator *tr, char *word, char *ph_out)
 	acc = 0;
 	prev = 0;
 	subtract = 0x7fff;
-	memset(wtab, 0, sizeof(wtab));
+
+	if((tr->langopts.numbers & NUM_ROMAN_CAPITALS) && !(wtab[0].flags & FLAG_ALL_UPPER))
+		return(0);
 
 	while((c = *word++) != ' ')
 	{
@@ -807,9 +841,10 @@ int TranslateRoman(Translator *tr, char *word, char *ph_out)
 		else
 			acc += prev;
 		prev = value;
+		n_digits++;
 	}
 	acc += prev;
-	if(acc < 2)
+	if(acc < tr->langopts.min_roman)
 		return(0);
 
 	if(acc > tr->langopts.max_roman)
@@ -826,12 +861,22 @@ int TranslateRoman(Translator *tr, char *word, char *ph_out)
 
 	sprintf(number_chars," %d ",acc);
 
-	if(tr->langopts.numbers & NUM_ROMAN_ORDINAL)
+	if(CheckDotOrdinal(tr, word, wtab, 0))
 		wtab[0].flags |= FLAG_ORDINAL;
+
+	if(tr->langopts.numbers & NUM_ROMAN_ORDINAL)
+	{
+		if((n_digits <= 1) && !(wtab[0].flags & FLAG_ORDINAL))
+			return(0);
+		wtab[0].flags |= FLAG_ORDINAL;
+	}
+
+
 	TranslateNumber(tr, &number_chars[1], p, &flags, wtab);
 
 	if(tr->langopts.numbers & NUM_ROMAN_AFTER)
 		strcat(ph_out,ph_roman);
+
 	return(1);
 }  // end of TranslateRoman
 
@@ -1313,8 +1358,6 @@ static int TranslateNumber_1(Translator *tr, char *word, char *ph_out, unsigned
 	char ph_buf[200];
 	char ph_buf2[50];
 	char suffix[20];
-	char *wordptr;
-	unsigned int dictflags;
 
 	static const char str_pause[2] = {phonPAUSE_NOLINK,0};
 
@@ -1325,22 +1368,10 @@ static int TranslateNumber_1(Translator *tr, char *word, char *ph_out, unsigned
 	value = this_value = atoi(word);
 
 	ph_ordinal2[0] = 0;
-	if((tr->langopts.numbers & NUM_ORDINAL_DOT) && (word[ix] == '.') && IsAlpha(word[ix+2]) && !(wtab[1].flags & FLAG_FIRST_UPPER))
-	{
-		// ordinal number is indicated by dot after the number
-		// but not if the next word starts with an upper-case letter
-		ordinal = 2;
-		word[ix] = ' ';
-		if(tr->translator_name == L('h','u'))
-		{
-			// lang=hu don't treat dot as ordinal indicator if the next word is a month name ($alt)
-			wordptr = &word[ix+2];
-			dictflags = TranslateWord(tr, &word[ix+2], 0, NULL);
-			if(dictflags & FLAG_ALT_TRANS)
-				ordinal = 0;
-		}
-	}
-	else
+
+	ordinal = CheckDotOrdinal(tr, &word[ix], wtab, 1);
+
+	if(ordinal == 0)
 	{
 		// look for an ordinal number suffix after the number
 		ix++;

src/readclause.cpp

@@ -357,6 +357,15 @@ int towlower2(unsigned int c)
 	return(towlower(c));
 }
 
+
+static int IsRomanU(unsigned int c)
+{//================================
+	if((c=='I') || (c=='V') || (c=='X') || (c=='L'))
+		return(1);
+	return(0);
+}
+
+
 static void GetC_unget(int c)
 {//==========================
 // This is only called with UTF8 input, not wchar input
@@ -2388,10 +2397,11 @@ if(option_ssml) parag=1;
 
 				if((nl_count==0) && (c1 == '.'))
 				{
-					if(iswdigit(cprev) && (tr->langopts.numbers & NUM_ORDINAL_DOT))
+					if((tr->langopts.numbers & NUM_ORDINAL_DOT) && 
+						(iswdigit(cprev) || (IsRomanU(cprev) && (IsRomanU(cprev2) || iswspace(cprev2)))))  // lang=hu
 					{
 						// dot after a number indicates an ordinal number
-						if(islower(c_next) || (c_next == '<'))
+						if(!iswdigit(cprev) || iswlower(c_next) || (c_next == '<'))
 							is_end_clause = 0;      // only if followed by lower-case, (or if there is a XML tag)
 					}
 					else

src/synthdata.cpp

@@ -35,7 +35,7 @@
 #include "translate.h"
 #include "wave.h"
 
-const char *version_string = "1.41.15  12.Oct.09";
+const char *version_string = "1.41.17  15.Oct.09";
 const int version_phdata  = 0x014100;
 
 int option_device_number = -1;

src/tr_languages.cpp

@@ -164,6 +164,7 @@ static Translator* NewTranslator(void)
 	tr->langopts.long_stop = 100;
 
 	tr->langopts.max_roman = 49;
+	tr->langopts.min_roman = 2;
 	tr->langopts.thousands_sep = ',';
 	tr->langopts.decimal_sep = '.';
 	tr->langopts.break_numbers = BREAK_THOUSANDS;   // 1000, 1000,000  1,000,000 etc
@@ -588,7 +589,8 @@ Translator *SelectTranslator(const char *name)
 			tr->langopts.param[LOPT_IT_DOUBLING] = 1;
 			tr->langopts.param[LOPT_COMBINE_WORDS] = 99;  // combine some prepositions with the following word
 
-			tr->langopts.numbers = NUM_DECIMAL_COMMA | NUM_ALLOW_SPACE | NUM_DFRACTION_5 | NUM_ROMAN | NUM_ROMAN_ORDINAL | NUM_ORDINAL_DOT | NUM_OMIT_1_HUNDRED;
+			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.min_roman = 1;
 			SetLetterVowel(tr,'y');
 			tr->langopts.spelling_stress = 1;
 SetLengthMods(tr,3);  // all equal

src/translate.cpp

@@ -740,7 +740,7 @@ if((wmark > 0) && (wmark < 8))
 
 			if((tr->langopts.numbers & NUM_ROMAN) || ((tr->langopts.numbers & NUM_ROMAN_UC) && (wflags & FLAG_ALL_UPPER)))
 			{
-				if((found = TranslateRoman(tr, word1, phonemes)) != 0)
+				if((found = TranslateRoman(tr, word1, phonemes, wtab)) != 0)
 					dictionary_flags[0] |= FLAG_ABBREV;   // prevent emphasis if capitals
 			}
 		}

src/translate.h

@@ -352,6 +352,7 @@ typedef struct {
 #define NUM_ROMAN_AFTER  0x200000
 #define NUM_VIGESIMAL    0x400000
 #define NUM_ROMAN_ORDINAL 0x800000
+#define NUM_ROMAN_CAPITALS 0x1000000
 
 	// bits0-1=which numbers routine to use.
 	// bit2=  thousands separator must be space
@@ -374,6 +375,7 @@ typedef struct {
 	// bit21= say "roman" after the number, not before
 	// bit22= vigesimal number, if tens are not found
 	// bit23=Roman numbers are ordinal numbers
+	// bit24=Roman numbers must be capital letters
 	int numbers;
 
 	// bits 1-4  use variant form of numbers before thousands,millions,etc.
@@ -386,6 +388,7 @@ typedef struct {
 #define BREAK_THOUSANDS   0x49249248
 	int break_numbers;  // which digits to break the number into thousands, millions, etc (Hindi has 100,000 not 1,000,000)
 	int max_roman;
+	int min_roman;
 	int thousands_sep;
 	int decimal_sep;
 
@@ -570,7 +573,7 @@ void DeleteTranslator(Translator *tr);
 int Lookup(Translator *tr, const char *word, char *ph_out);
 
 int TranslateNumber(Translator *tr, char *word1, char *ph_out, unsigned int *flags, WORD_TAB *wtab);
-int TranslateRoman(Translator *tr, char *word, char *ph_out);
+int TranslateRoman(Translator *tr, char *word, char *ph_out, WORD_TAB *wtab);
 
 void ChangeWordStress(Translator *tr, char *word, int new_stress);
 void SetSpellingStress(Translator *tr, char *phonemes, int control, int n_chars);