[1.24.12] Produce dictsource/dict_phonemes, a list of which phonemes are used by each language's *_rules and *_list files.

Fix crash with -X option when spelling words (acronyms).
Rules files: added special character X meaning "no vowel until the word boundary". Used for lang=no.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@14 d46cf337-b52f-0410-862d-fd96e6ae7743

src/compiledata.cpp

@@ -75,6 +75,7 @@
 
 extern void Write4Bytes(FILE *f, int value);
 extern void MakeVowelLists(void);
+extern void FindPhonemesUsed(void);
 extern int CompileDictionary(const char *dsource, const char *dict_name, FILE *log, char *fname);
 extern char voice_name[];
 
@@ -1824,6 +1825,76 @@ void Compile::Report(void)
 }
 
 
+
+static int ph_sorter(char **a, char **b)
+{//======================================
+	int ix;
+	int t1, t2;
+	char mnem1[6];
+
+	PHONEME_TAB *p1 = (PHONEME_TAB *)(*a);
+	PHONEME_TAB *p2 = (PHONEME_TAB *)(*b);
+
+	t1 = p1->type;
+	if(t1 > phVOWEL) t1 = phVOWEL+1;
+
+	t2 = p2->type;
+	if(t2 > phVOWEL) t2 = phVOWEL+1;
+
+	if((ix = t1 - t2) != 0)
+		return(ix);
+
+	strcpy(mnem1,WordToString(p1->mnemonic));
+  return(strcasecmp(mnem1,WordToString(p2->mnemonic)));
+}   /* end of ph_sorter */
+
+
+
+void PrintPhonemesUsed(FILE *f, const char *dictname)
+{//==================================================
+	int ix;
+	PHONEME_TAB *ph;
+	PHONEME_TAB *ph_tab[N_PHONEME_TAB];
+	int count = 0;
+	int n_ph = 0;
+	int section = 0;
+
+	fprintf(f,"\n\nDictionary %s_dict\n",dictname);
+	fflush(f);
+
+	for(ix=0; (ix<N_PHONEME_TAB) && (phoneme_tab[ix] != NULL); ix++)
+	{
+		if(phoneme_tab_flags[ix] & 2)
+		{
+			ph_tab[n_ph++] = phoneme_tab[ix];
+		}
+	}
+
+	qsort((void *)ph_tab,n_ph,sizeof(PHONEME_TAB *),(int (*)(const void *,const void *))ph_sorter);
+
+	for(ix=0; ix<n_ph; ix++)
+	{
+		ph = ph_tab[ix];
+
+		if(ph->type > 1)
+		{
+			if((ph->type > phVOWEL) && (section == 0))
+			{
+				section = 1;
+				count = 0;
+				fputc('\n',f);
+			}
+			if((count & 0x7) == 0)
+				fputc('\n',f);
+			fprintf(f,"%-4s ",WordToString(ph->mnemonic));
+			count++;
+		}
+	}
+	fputc('\n',f);
+}  // end of  PrintPhonemesUsed
+
+
+
 wxString CompileAllDictionaries()
 {//==============================
 	wxString filename;
@@ -1834,6 +1905,7 @@ wxString CompileAllDictionaries()
 	int errors = 0;
 	int dict_count = 0;
 	FILE *log;
+	FILE *f_phused;
 	char dictname[80];
 	char fname_log[80];
 	char save_voice_name[80];
@@ -1866,6 +1938,13 @@ wxString CompileAllDictionaries()
 
 	sprintf(fname_log,"%s%s",path_dsource,"dict_log");
 	log = fopen(fname_log,"w");
+	sprintf(fname_log,"%s%s",path_dsource,"dict_phonemes");
+	f_phused = fopen(fname_log,"w");
+
+	if(f_phused)
+	{
+		fprintf(f_phused,"Phonemes which are used in the *_rules and *_list files\n");
+	}
 
 	bool cont = dir.GetFirst(&filename, _T("*_rules"), wxDIR_FILES);
 	while ( cont )
@@ -1883,10 +1962,19 @@ wxString CompileAllDictionaries()
 			errors += err;
 		}
 
+		if(f_phused != NULL)
+		{
+			memset(phoneme_tab_flags,0,sizeof(phoneme_tab_flags));
+			FindPhonemesUsed();
+			PrintPhonemesUsed(f_phused,dictname);
+		}
+
 		cont = dir.GetNext(&filename);
 	}
 	if(log != NULL)
 		fclose(log);
+	if(f_phused != NULL)
+		fclose(f_phused);
 
 	LoadVoice(save_voice_name,1);
 

src/compiledict.cpp

@@ -657,6 +657,9 @@ void copy_rule_string(char *string, int &state)
 				case 'W':
 					c = RULE_SPELLING;
 					break;
+				case 'X':
+					c = RULE_NOVOWELS;
+					break;
 				case 'L':
 					// expect two digits
 					c = *p++ - '0';

src/dictionary.cpp

@@ -1261,7 +1261,11 @@ void Translator::SetWordStress(char *output, unsigned int dictionary_flags, int
 				else
 				{
 					// unstressed syllable within a word
-					v_stress = 1;      /* change from 0 (unstressed) to 1 (diminished stress) */
+					if((vowel_stress[v-1] != 1) || ((langopts.stress_flags & 0x10000) == 0))
+					{
+						v_stress = 1;      /* change from 0 (unstressed) to 1 (diminished stress) */
+						vowel_stress[v] = v_stress;
+					}
 				}
 			}
 
@@ -1383,7 +1387,7 @@ char *Translator::DecodeRule(const char *group, char *rule)
 	static char output[60];
 
 	static char symbols[] = {' ',' ',' ',' ',' ',' ',' ',' ',' ',
-			'@','&','%','+','#','S','D','Z','A','B','C','H','F','G','Y','N','K','V','L','T'};
+			'@','&','%','+','#','S','D','Z','A','B','C','H','F','G','Y','N','K','V','L','T','X','?','W'};
 
 
 	match_type = 0;
@@ -1805,6 +1809,23 @@ void Translator::MatchRule(char *word[], const char *group, char *rule, MatchRec
 					}
 					break;
 
+				case RULE_NOVOWELS:
+					{
+						char *p = post_ptr + letter_xbytes;
+						while(letter_w != RULE_SPACE)
+						{
+							if(IsLetter(letter_w,LETTERGP_VOWEL2))
+							{
+								failed = 1;
+								break;
+							}
+							p += utf8_in(&letter_w,p,0);
+						}
+						if(!failed)
+							match.points += (19-distance_right);
+					}
+					break;
+
 				case RULE_INC_SCORE:
 					match.points += 20;      // force an increase in points
 					break;
@@ -1932,6 +1953,13 @@ void Translator::MatchRule(char *word[], const char *group, char *rule, MatchRec
 						failed = 1;
 					break;
 
+				case RULE_NOVOWELS:
+					if(word_vowel_count== 0)
+						match.points += 19;
+					else
+						failed =1;
+					break;
+
 				case RULE_IFVERB:
 					if(expect_verb)
 						match.points += 1;
@@ -2710,7 +2738,8 @@ int Translator::LookupDictList(char *word1, char *ph_out, unsigned int *flags, i
 
 int Translator::Lookup(char *word, char *ph_out)
 {//=============================================
-	return(LookupDictList(word,ph_out,NULL,0));
+	unsigned int flags;
+	return(LookupDictList(word,ph_out,&flags,0));
 }
 
 

src/synthdata.cpp

@@ -35,7 +35,7 @@
 #include "translate.h"
 #include "wave.h"
 
-const char *version_string = "1.24.11  17.May.07";
+const char *version_string = "1.24.12  18.May.07";
 const int version_phdata  = 0x012201;
 
 int option_device_number = -1;

src/tr_languages.cpp

@@ -244,7 +244,7 @@ Translator *SelectTranslator(const char *name)
 			static const wchar_t replace_chars_hi[11] = {0x966,0x967,0x968,0x969,0x96a,0x96b,0x96c,0x96d,0x96e,0x96f,0};  // digits 0-9
 			static const unsigned int replacement_chars_hi[11] = {0x30,0x31,0x32,0x33,0x34,0x35,0x36,0x37,0x38,0x39,0};
 			static int stress_lengths_hi[8] = {190, 190,  210, 210,  0, 0,  230, 250};
-			static int stress_amps_hi[8] = {17,14, 20,20, 20,24, 24,22 };
+			static int stress_amps_hi[8] = {17,14, 20,19, 20,24, 24,22 };
 
 			tr = new Translator();
 			SetupTranslator(tr,stress_lengths_hi,stress_amps_hi);
@@ -252,7 +252,7 @@ Translator *SelectTranslator(const char *name)
 			tr->langopts.length_mods0 = tr->langopts.length_mods;  // don't lengthen vowels in the last syllable
 
 			tr->langopts.stress_rule = 6;      // stress on last heaviest syllable
-			tr->langopts.stress_flags =  0x4;   // use 'diminished' for unstressed final syllable
+			tr->langopts.stress_flags =  0x10004;   // use 'diminished' for unstressed final syllable
 			tr->langopts.numbers = 0x811 + 0x40000;
 			tr->letter_bits_offset = OFFSET_DEVANAGARI;
 			tr->langopts.replace_chars = replace_chars_hi;
@@ -392,7 +392,7 @@ Translator *SelectTranslator(const char *name)
 	case L('n','o'):  // Norwegian
 		{
 			static int stress_amps_no[] = {16,16, 20,20, 20,24, 24,22 };
-			static int stress_lengths_no[8] = {160,140, 200,200, 0,0, 250,270};
+			static int stress_lengths_no[8] = {160,140, 200,190, 0,0, 220,240};
 			tr = new Translator();
 			SetupTranslator(tr,stress_lengths_no,stress_amps_no);
 

src/translate.h

@@ -122,9 +122,10 @@
 #define RULE_NOTVOWEL   25   // K
 #define RULE_IFVERB     26   // V
 #define RULE_LETTERGP   27   // L + letter group number
-#define RULE_ALT1       28   // word has $alt attribute
-#define RULE_SPELLING   31   // while spelling letter-by-letter
-#define RULE_LAST_RULE   28
+#define RULE_ALT1       28   // T word has $alt attribute
+#define RULE_NOVOWELS   29   // X no vowels up to word boundary
+#define RULE_SPELLING   31   // W while spelling letter-by-letter
+#define RULE_LAST_RULE   31
 
 #define LETTERGP_A	0
 #define LETTERGP_B	1
@@ -258,6 +259,7 @@ typedef struct {
 // bit9=stress last syllable if it doesn't end in vowel or "s" or "n"  LANG=Spanish
 // bit12= In a 2-syllable word, if one has primary stress then give the other secondary stress
 // bit13= If there is only one syllable before the primary stress, give it a secondary stress
+// bit16= Don't diminish consecutive syllables within a word.
 
 	int stress_flags; 
 	int unstressed_wd1; // stress for $u word of 1 syllable

src/vowelchart.cpp

@@ -373,10 +373,12 @@ void VowelChart(int control, char *fname)
 
 
 
-static void FindPhonemesUsed(void)
-{//===============================
+void FindPhonemesUsed(void)
+{//========================
 	int hash;
 	char *p;
+	char *start;
+	char *group;
 	char *next;
 	unsigned char c;
 	int count = 0;
@@ -386,15 +388,24 @@ static void FindPhonemesUsed(void)
 	p = translator->data_dictrules;
 	while(*p != 0)
 	{
+		if(*p == RULE_CONDITION)
+			p+=2;
 		if(*p == RULE_GROUP_END)
+		{
 			p++;
+			if(*p == 0) break;
+		}
 		if(*p == RULE_GROUP_START)
+		{
+			group = p;
 			p += (strlen(p)+1);
+		}
 
 		while((((c = *p) != RULE_PHONEMES)) && (c != 0)) p++;
 		count++;
 		if(c == RULE_PHONEMES)
 		{
+			start = p;
 			p++;
 			while(*p != 0)
 			{