[1.41.15]

Fixes to: announce punctuation ( --punct option)


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@227 d46cf337-b52f-0410-862d-fd96e6ae7743

src/compiledict.cpp

@@ -223,8 +223,6 @@ static int compile_line(char *linebuf, char *dict_line, int *hash)
 	unsigned char bad_phoneme[4];
 static char nullstring[] = {0};
 
-	WORD_TAB winfo;
-
 	comment = NULL;
 	text_not_phonemes = 0;
 	phonetic = word = nullstring;
@@ -437,12 +435,11 @@ step=1;  // TEST
 		if(word[0] == '_')
 		{
 			// This is a special word, used by eSpeak.  Translate this into phonemes now
-			memset(&winfo,0,sizeof(winfo));
 			strcat(phonetic, " ");     // need a space to indicate word-boundary
 
 	// PROBLEM  vowel reductions are not applied to the translated phonemes
 	// condition rules are not applied
-			TranslateWord(translator,phonetic,0,&winfo);
+			TranslateWord(translator,phonetic,0,NULL);
 			text_not_phonemes = 0;
 			strncpy0(encoded_ph, word_phonemes, N_WORD_BYTES-4);
 

src/dictionary.cpp

@@ -247,6 +247,12 @@ int LoadDictionary(Translator *tr, const char *name, int no_error)
 	sprintf(fname,"%s%c%s_dict",path_home,PATHSEP,name);
 	size = GetFileLength(fname);
 
+	if(tr->data_dictlist != NULL)
+	{
+		Free(tr->data_dictlist);
+		tr->data_dictlist = NULL;
+	}
+
 	f = fopen(fname,"rb");
 	if((f == NULL) || (size <= 0))
 	{
@@ -257,9 +263,6 @@ int LoadDictionary(Translator *tr, const char *name, int no_error)
 		return(1);
 	}
 
-	if(tr->data_dictlist != NULL)
-		Free(tr->data_dictlist);
-
 	tr->data_dictlist = Alloc(size);
 	fread(tr->data_dictlist,size,1,f);
 	fclose(f);

src/extras.cpp

@@ -317,7 +317,6 @@ void Lexicon_It(int pass)
 	char phonemes[80];
 	char phonemes2[80];
 	char buf_out[120];
-	WORD_TAB winfo;
 	static const char *vowels1 = "aeiou";
 	static const char *vowels2 = "aeou";
 
@@ -438,8 +437,7 @@ void Lexicon_It(int pass)
 			}
 		}
 		// translate
-		memset(&winfo,0,sizeof(winfo));
-		TranslateWord(translator,&word1[1],0,&winfo);
+		TranslateWord(translator,&word1[1],0, NULL);
 		DecodePhonemes(word_phonemes,phonemes);
 
 		stress_posn1 = 0;
@@ -462,8 +460,7 @@ void Lexicon_It(int pass)
 				vowel_ix++;
 		}
 
-		memset(&winfo,0,sizeof(winfo));
-		TranslateWord(translator,&word2[1],0,&winfo);
+		TranslateWord(translator,&word2[1],0, NULL);
 		DecodePhonemes(word_phonemes,phonemes2);
 
 		vowel_ix = 1;
@@ -619,7 +616,6 @@ void Lexicon_De()
 	char pronounce2[80];
 	char phonemes[80];
 	char phonemes2[80];
-	WORD_TAB winfo;
 
 	static const char *vowels = "aeiouyAEIOUY29@";
 
@@ -728,8 +724,7 @@ void Lexicon_De()
 		}
 
 		// translate
-		memset(&winfo,0,sizeof(winfo));
-		TranslateWord(translator,&word2[1],0,&winfo);
+		TranslateWord(translator,&word2[1],0, NULL);
 
 		DecodePhonemes2(word_phonemes,phonemes);  // also need to change some phoneme names
 
@@ -794,7 +789,6 @@ void Lexicon_Ru()
 	int wlen;
 	int len;
 	int check_root;
-	WORD_TAB winfo;
 
 	char word[80];
 	char word2[80];
@@ -933,8 +927,7 @@ p_unicode = unicode;
 		}
 
 		// translate
-		memset(&winfo,0,sizeof(winfo));
-		TranslateWord(translator, &word2[1],0,&winfo);
+		TranslateWord(translator, &word2[1],0, NULL);
 		DecodePhonemes(word_phonemes,phonemes);
 
 		// find the stress position in the translation

src/numbers.cpp

@@ -1314,7 +1314,7 @@ static int TranslateNumber_1(Translator *tr, char *word, char *ph_out, unsigned
 	char ph_buf2[50];
 	char suffix[20];
 	char *wordptr;
-	unsigned int dictflags[2];
+	unsigned int dictflags;
 
 	static const char str_pause[2] = {phonPAUSE_NOLINK,0};
 
@@ -1331,14 +1331,12 @@ static int TranslateNumber_1(Translator *tr, char *word, char *ph_out, unsigned
 		// 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)
-			dictflags[0] = 0;
 			wordptr = &word[ix+2];
-			LookupDictList(tr, &wordptr, ph_buf, dictflags, 0, NULL);
-			if(dictflags[0] & FLAG_ALT_TRANS) // TEST
+			dictflags = TranslateWord(tr, &word[ix+2], 0, NULL);
+			if(dictflags & FLAG_ALT_TRANS)
 				ordinal = 0;
 		}
 	}
@@ -1514,23 +1512,20 @@ static int TranslateNumber_1(Translator *tr, char *word, char *ph_out, unsigned
 				}
 				break;
 
-			case NUM_DFRACTION_1:   // italian, say "hundredths" is leading zero
+			case NUM_DFRACTION_1:   // italian, say "hundredths" if leading zero
 			case NUM_DFRACTION_5:   // hungarian, always say "tenths" etc.
-				if(decimal_count <= 4)
+				LookupNum3(tr, atoi(&word[n_digits]), ph_buf, 0,0,0);
+				if((word[n_digits]=='0') || (decimal_mode == NUM_DFRACTION_5))
 				{
-					LookupNum3(tr, atoi(&word[n_digits]), ph_buf, 0,0,0);
-					if((word[n_digits]=='0') || (decimal_mode == 0xa000))
-					{
-						// decimal part has leading zeros, so add a "hundredths" or "thousandths" suffix
-						sprintf(string,"_0Z%d",decimal_count);
-						if(Lookup(tr, string, buf1) == 0)
-							break;   // revert to speaking single digits
+					// decimal part has leading zeros, so add a "hundredths" or "thousandths" suffix
+					sprintf(string,"_0Z%d",decimal_count);
+					if(Lookup(tr, string, buf1) == 0)
+						break;   // revert to speaking single digits
 
-						strcat(ph_buf,buf1);
-					}
-					strcat(ph_out,ph_buf);
-					n_digits += decimal_count;
+					strcat(ph_buf,buf1);
 				}
+				strcat(ph_out,ph_buf);
+				n_digits += decimal_count;
 				break;
 
 			case NUM_DFRACTION_3:

src/readclause.cpp

@@ -816,8 +816,8 @@ static int LoadSoundFile2(const char *fname)
 
 
 
-static int AnnouncePunctuation(Translator *tr, int c1, int c2, char *buf, int bufix)
-{//=================================================================================
+static int AnnouncePunctuation(Translator *tr, int c1, int *c2_ptr, char *output, int *bufix, int end_clause)
+{//==========================================================================================================
 	// announce punctuation names
 	// c1:  the punctuation character
 	// c2:  the following character
@@ -826,12 +826,19 @@ static int AnnouncePunctuation(Translator *tr, int c1, int c2, char *buf, int bu
 	const char *punctname;
 	int found = 0;
 	int soundicon;
-	char *p;
+	int c2;
+	int len;
+	int bufix1;
+	char buf[200];
+	char buf2[80];
+
+	c2 = *c2_ptr;
+	buf[0] = 0;
 
 	if((soundicon = LookupSoundicon(c1)) >= 0)
 	{
 		// add an embedded command to play the soundicon
-		sprintf(&buf[bufix],"\001%dI ",soundicon);
+		sprintf(buf,"\001%dI ",soundicon);
 		UngetC(c2);
 		found = 1;
 	}
@@ -839,7 +846,7 @@ static int AnnouncePunctuation(Translator *tr, int c1, int c2, char *buf, int bu
 	if((punctname = LookupCharName(tr, c1)) != NULL)
 	{
 		found = 1;
-		if(bufix==0)
+		if((*bufix==0) || (end_clause==0))
 		{
 			punct_count=1;
 			while(c2 == c1)
@@ -847,26 +854,32 @@ static int AnnouncePunctuation(Translator *tr, int c1, int c2, char *buf, int bu
 				punct_count++;
 				c2 = GetC();
 			}
-			UngetC(c2);
+			*c2_ptr = c2;
+			if(end_clause)
+			{
+				UngetC(c2);
+			}
 
-			p = &buf[bufix];
 			if(punct_count==1)
 			{
-				sprintf(p,"%s %s %s",tone_punct_on,punctname,tone_punct_off);
+				sprintf(buf,"%s %s %s",tone_punct_on,punctname,tone_punct_off);
 			}
 			else
 			if(punct_count < 4)
 			{
-				sprintf(p,"\001+10S%s",tone_punct_on);
+				sprintf(buf,"\001+10S%s",tone_punct_on);
 				while(punct_count-- > 0)
-					sprintf(buf,"%s %s",buf,punctname);
-				sprintf(p,"%s %s\001-10S",buf,tone_punct_off);
+				{
+					sprintf(buf2," %s",punctname);
+					strcat(buf, buf2);
+				}
+				sprintf(buf2," %s\001-10S",tone_punct_off);
+				strcat(buf, buf2);
 			}
 			else
 			{
-				sprintf(p,"%s %s %d %s %s",
+				sprintf(buf,"%s %s %d %s %s",
 						tone_punct_on,punctname,punct_count,punctname,tone_punct_off);
-				return(CLAUSE_COMMA);
 			}
 		}
 		else
@@ -879,17 +892,25 @@ static int AnnouncePunctuation(Translator *tr, int c1, int c2, char *buf, int bu
 					ssml_ignore_l_angle = c1;  // this was &lt; which was converted to <, don't pick it up again as <
 			}
 			ungot_char2 = c1;
-			buf[bufix] = ' ';
-			buf[bufix+1] = 0;
+			buf[0] = ' ';
+			buf[1] = 0;
 		}
 	}
 
 	if(found == 0)
 		return(-1);
 
+	bufix1 = *bufix;
+	len = strlen(buf);
+	strcpy(&output[*bufix],buf);
+	*bufix += len;
+
+	if(end_clause==0)
+		return(-1);
+
 	if(c1 == '-')
 		return(CLAUSE_NONE);   // no pause
-	if(bufix > 0)
+	if(bufix1 > 0)
 		return(CLAUSE_SHORTCOMMA);
 	if((strchr_w(punct_close,c1) != NULL) && !iswalnum(c2))
 		return(CLAUSE_SHORTFALL+4);
@@ -1919,8 +1940,8 @@ int ReadClause(Translator *tr, FILE *f_in, char *buf, short *charix, int *charix
 	int any_alnum = 0;
 	int self_closing;
 	int punct_data;
-	int is_punctuation;
-	int save_c2;
+	int is_end_clause;
+	int announced_punctuation;
 	int stressed_word = 0;
 	const char *p;
 	wchar_t xml_buf[N_XML_BUF+1];
@@ -2304,35 +2325,48 @@ if(option_ssml) parag=1;
 			linelength = 0;
 		}
 
-		if(option_punctuation && (phoneme_mode==0) && (sayas_mode==0) && iswpunct(c1))
-		{
-			// option is set to explicitly speak punctuation characters
-			// if a list of allowed punctuation has been set up, check whether the character is in it
-			if((option_punctuation == 1) || (wcschr(option_punctlist,c1) != NULL))
-			{
-				if((terminator = AnnouncePunctuation(tr, c1, c2, buf, ix)) >= 0)
-					return(terminator);
-			}
-		}
+		announced_punctuation = 0;
 
-		if((phoneme_mode==0) && (sayas_mode==0) && ((punct = lookupwchar(punct_chars,c1)) != 0))
+		if((phoneme_mode==0) && (sayas_mode==0))
 		{
-			punct_data = punct_attributes[punct];
+			is_end_clause = 0;
 
-			if(punct_data & PUNCT_IN_WORD)
+			if((punct = lookupwchar(punct_chars,c1)) != 0)
 			{
-				// Armenian punctuation inside a word
-				stressed_word = 1;
-				*tone_type = punct_data >> 12 & 0xf;   // override the end-of-sentence type
-				continue;
+				punct_data = punct_attributes[punct];
+	
+				if(punct_data & PUNCT_IN_WORD)
+				{
+					// Armenian punctuation inside a word
+					stressed_word = 1;
+					*tone_type = punct_data >> 12 & 0xf;   // override the end-of-sentence type
+					continue;
+				}
+
+				if((iswspace(c2) || (punct_data & 0x8000) || IsBracket(c2) || (c2=='?') || (c2=='-') || Eof()))
+				{
+					// note: (c2='?') is for when a smart-quote has been replaced by '?'
+					is_end_clause = 1;
+				}
 			}
 
-			if((iswspace(c2) || (punct_data & 0x8000) || IsBracket(c2) || (c2=='?') || (c2=='-') || Eof()))
+			if(option_punctuation && iswpunct(c1))
 			{
-				// note: (c2='?') is for when a smart-quote has been replaced by '?'
-
-				is_punctuation = 1;
+				// option is set to explicitly speak punctuation characters
+				// if a list of allowed punctuation has been set up, check whether the character is in it
+				if((option_punctuation == 1) || (wcschr(option_punctlist,c1) != NULL))
+				{
+					if((terminator = AnnouncePunctuation(tr, c1, &c2, buf, &ix, is_end_clause)) >= 0)
+						return(terminator);
+					announced_punctuation = c1;
+				}
+			}
+	
 
+			if(is_end_clause)
+			{
+				int c_next;
+	
 				if((c1 == '.') && (cprev == '.'))
 				{
 					c1 = 0x2026;
@@ -2340,47 +2374,43 @@ if(option_ssml) parag=1;
 				}
 	
 				nl_count = 0;
-				save_c2 = c2;
+				c_next = c2;
 
-				if(iswspace(c2))
+				if(iswspace(c_next))
 				{
-					while(!Eof() && iswspace(c2))
+					while(!Eof() && iswspace(c_next))
 					{
-						if(c2 == '\n')
+						if(c_next == '\n')
 							nl_count++;
-						c2 = GetC();   // skip past space(s)
-					}
-					if(!Eof())
-					{
-						UngetC(c2);
+						c_next = GetC();   // skip past space(s)
 					}
 				}
 
 				if((nl_count==0) && (c1 == '.'))
 				{
-//					if(iswdigit(cprev) && (tr->langopts.numbers & NUM_ORDINAL_DOT) && islower(c2))
 					if(iswdigit(cprev) && (tr->langopts.numbers & NUM_ORDINAL_DOT))
 					{
 						// dot after a number indicates an ordinal number
-						is_punctuation = 0;
+						if(islower(c_next) || (c_next == '<'))
+							is_end_clause = 0;      // only if followed by lower-case, (or if there is a XML tag)
 					}
 					else
-					if(iswlower(c2))
+					if(iswlower(c_next))
 					{
 						// next word has no capital letter, this dot is probably from an abbreviation
 						c1 = ' ';
-						is_punctuation = 0;
+						is_end_clause = 0;
 					}
 					if(any_alnum==0)
 					{
 						c1 = ' ';   // no letters or digits yet, so probably not a sentence terminator
-						is_punctuation = 0;
+						is_end_clause = 0;
 					}
 				}
-				c2 = save_c2;
 
-				if(is_punctuation)
+				if(is_end_clause)
 				{
+					UngetC(c_next);
 					buf[ix] = ' ';
 					buf[ix+1] = 0;
 		
@@ -2393,12 +2423,33 @@ if(option_ssml) parag=1;
 					}
 					return(punct_data);   // only recognise punctuation if followed by a blank or bracket/quote
 				}
+				else
+				{
+					if(!Eof())
+					{
+						if(iswspace(c2))
+							UngetC(c_next);
+					}
+				}
 			}
 		}
 
 		if(speech_parameters[espeakSILENCE]==1)
 			continue;
 
+		if(c1 == announced_punctuation)
+		{
+//			const unsigned short keep_punctuation[] = {
+//			'\'', '-', 0x92, 0xb4, 0x2019, 0x2032, 0 };
+
+			// This character has already been announced, so delete it so that it isn't spoken a second time.
+			// Unless it's a hyphen or apostrophe (which is used by TranslateClause() )
+			if(!IsBracket(c1))
+			{
+				c1 = ' ';
+			}
+		}
+
 		j = ix+1;
 		ix += utf8_out(c1,&buf[ix]);    //	buf[ix++] = c1;
 		if(!iswspace(c1) && !IsBracket(c1))

src/synthdata.cpp

@@ -35,7 +35,7 @@
 #include "translate.h"
 #include "wave.h"
 
-const char *version_string = "1.41.13  10.Oct.09";
+const char *version_string = "1.41.15  12.Oct.09";
 const int version_phdata  = 0x014100;
 
 int option_device_number = -1;

src/synthesize.cpp

@@ -535,7 +535,7 @@ static short vcolouring[N_VCOLOUR][5] = {
 //	fprintf(stderr,"FMT%d %3s  %3d-%3d f1=%d  f2=%4d %4d %4d  f3=%4d %3d\n",
 //		which,WordToString(other_ph->mnemonic),len,rms,f1,f2,f2_min,f2_max,f3_adj,f3_amp);
 
-	if(other_ph->mnemonic == '?')
+	if((other_ph != NULL) && (other_ph->mnemonic == '?'))
 		flags |= 8;
 
 	if(which == 1)

src/tr_languages.cpp

@@ -464,6 +464,13 @@ Translator *SelectTranslator(const char *name)
 		break;
 
 
+	case L('f','a'):   // Farsi
+		{
+			tr->langopts.param[LOPT_UNPRONOUNCABLE] = 1;   // disable check for unpronouncable words
+		}
+		break;
+
+
 	case L('f','i'):   // Finnish
 		{
 			static const unsigned char stress_amps_fi[8] = {18,16, 22,22, 20,22, 22,22 };

src/translate.cpp

@@ -601,8 +601,9 @@ int TranslateWord(Translator *tr, char *word1, int next_pause, WORD_TAB *wtab)
 	int spell_word;
 	int stress_bits;
 	int emphasize_allcaps = 0;
-	int wflags = wtab->flags;
-	int wmark = wtab->wmark;
+	int wflags;
+	int wmark;
+	WORD_TAB wtab_null[8];
 
 	// translate these to get pronunciations of plural 's' suffix (different forms depending on
 	// the preceding letter
@@ -610,6 +611,14 @@ int TranslateWord(Translator *tr, char *word1, int next_pause, WORD_TAB *wtab)
 	static char word_iz[4] = {0,'i','z',0};
 	static char word_ss[4] = {0,'s','s',0};
 
+	if(wtab == NULL)
+	{
+		memset(wtab_null, 0, sizeof(wtab_null));
+		wtab = wtab_null;
+	}
+	wflags = wtab->flags;
+	wmark = wtab->wmark;
+
 	dictionary_flags[0] = 0;
 	dictionary_flags[1] = 0;
 	dictionary_flags2[0] = 0;
@@ -620,6 +629,13 @@ int TranslateWord(Translator *tr, char *word1, int next_pause, WORD_TAB *wtab)
 	end_phonemes[0] = 0;
 	ph_limit = &phonemes[N_WORD_PHONEMES];
 
+	if(tr->data_dictlist == NULL)
+	{
+		// dictionary is not loaded
+		word_phonemes[0] = 0;
+		return(0);
+	}
+
 	// count the length of the word
 	if(*word1 == ' ') word1++;   // possibly a dot was replaced by space:  $dot
 	wordx = word1;
@@ -1053,6 +1069,7 @@ strcpy(phonemes2,phonemes);
 					AppendPhonemes(tr,phonemes, N_WORD_PHONEMES, end_phonemes);
 					end_phonemes[0] = 0;
 				}
+				memcpy(wordx,word_copy,strlen(word_copy));
 			}
 			wordx[-1] = c_temp;
 		}

src/translate.h

@@ -98,10 +98,10 @@
 #define FLAG_DONT_SWITCH_TRANSLATOR  0x1000
 #define FLAG_SUFFIX_REMOVED  0x2000
 #define FLAG_HYPHEN_AFTER    0x4000
+#define FLAG_ORDINAL       0x8000   // passed to TranslateNumber() to indicate an ordinal number
 
 #define FLAG_NO_TRACE      0x10000   // passed to TranslateRules() to suppress dictionary lookup printout
 #define FLAG_NO_PREFIX     0x20000
-#define FLAG_ORDINAL       0x40000   // passed to TranslateNumber() to indicate an ordinal number
 
 // prefix/suffix flags (bits 8 to 14, bits 16 to 22) don't use 0x8000, 0x800000
 #define SUFX_E        0x0100   // e may have been added