[1.27.10] Added inverse conditional rules in *_list and *_rules. eg. ?!3 "if rule 3 is NOT set".

lang=en-r, adjusted vowels.
Moved single characters (eg. % @) from *_list to *_rules so they can be used in combinations (eg. %-).


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@65 d46cf337-b52f-0410-862d-fd96e6ae7743

src/compiledict.cpp

@@ -161,6 +161,7 @@ int compile_line(char *linebuf, char *dict_line, int *hash)
 	int  ix;
 	int  step;
 	int  n_flag_codes = 0;
+	int  flag_offset;
 	int  length;
 	int  multiple_words = 0;
 	char *multiple_string = NULL;
@@ -189,7 +190,17 @@ int compile_line(char *linebuf, char *dict_line, int *hash)
 	
 		if((c == '?') && (step==0))
 		{
+			// conditional rule, allow only if the numbered condition is set for the voice
+			flag_offset = 100;
+
 			p++;
+			if(*p == '!')
+			{
+				// allow only if the numbered condition is NOT set
+				flag_offset = 132;
+				p++;
+			}
+
 			ix = 0;
 			if(isdigit(*p))
 			{
@@ -201,7 +212,7 @@ int compile_line(char *linebuf, char *dict_line, int *hash)
 				ix = ix*10 + (*p-'0');
 				p++;
 			}
-			flag_codes[n_flag_codes++] = ix+100;
+			flag_codes[n_flag_codes++] = ix + flag_offset;
 			c = *p;
 		}
 		
@@ -859,7 +870,17 @@ char *compile_rule(char *input)
 	if(rule_cond[0] != 0)
 	{
 		ix = -1;
-		ix = atoi(rule_cond);
+		if(rule_cond[0] == '!')
+		{
+			// allow the rule only if the condition number is NOT set for the voice
+			ix = atoi(&rule_cond[1]) + 32;
+		}
+		else
+		{
+			// allow the rule only if the condition number is set for the voice
+			ix = atoi(rule_cond);
+		}
+
 		if((ix > 0) && (ix < 255))
 		{
 			output[len++] = RULE_CONDITION;

src/dictionary.cpp

@@ -666,7 +666,7 @@ int Translator::IsVowel(int letter)
 
 void SetLetterVowel(Translator *tr, int c)
 {//=======================================
-	tr->letter_bits[c] = 0x81;   // group 0 only
+	tr->letter_bits[c] = (tr->letter_bits[c] & 0x40) | 0x81;  // keep value for group 6 (front vowels e,i,y)
 }
 
 void ResetLetterBits(Translator *tr, int groups)
@@ -1773,9 +1773,21 @@ void Translator::MatchRule(char *word[], const char *group, char *rule, MatchRec
 				case RULE_CONDITION:
 					/* conditional rule, next byte gives condition number */
 					condition_num = *rule++;
-					if((dict_condition & (1L << condition_num))==0)
-						failed = 1;
+					
+					if(condition_num >= 32)
+					{
+						// allow the rule only if the condition number is NOT set
+						if((dict_condition & (1L << (condition_num-32))) != 0)
+							failed = 1;
+					}
 					else
+					{
+						// allow the rule only if the condition number is set
+						if((dict_condition & (1L << condition_num)) == 0)
+							failed = 1;
+					}
+
+					if(!failed)
 						match.points++;  // add one point for a matched conditional rule
 					break;
 				}
@@ -2677,8 +2689,18 @@ int Translator::LookupDict2(char *word, char *word2, char *phonetic, unsigned in
 			if(flag >= 100)
 			{
 				// conditional rule
-				if((dict_condition & (1 << (flag-100))) == 0)
-					condition_failed = 1;
+				if(flag >= 132)
+				{
+					// fail if this condition is set
+					if((dict_condition & (1 << (flag-132))) != 0)
+						condition_failed = 1;
+				}
+				else
+				{
+					// allow only if this condition is set
+					if((dict_condition & (1 << (flag-100))) == 0)
+						condition_failed = 1;
+				}
 			}
 			else
 			if(flag > 64)

src/setlengths.cpp

@@ -389,7 +389,7 @@ void Translator::CalcLengths()
 				if(p->type == phLIQUID)
 				{
 					p->length = speed1;
-p->pitch1 = p->pitch2 - 20;   // post vocalic [r/]
+//p->pitch1 = p->pitch2 - 20;   // post vocalic [r/]
 				}
 
 				if(next->type == phVSTOP)

src/synthdata.cpp

@@ -35,7 +35,7 @@
 #include "translate.h"
 #include "wave.h"
 
-const char *version_string = "1.27.09  12.Jul.07";
+const char *version_string = "1.27.10  15.Jul.07";
 const int version_phdata  = 0x012701;
 
 int option_device_number = -1;

src/tr_languages.cpp

@@ -317,8 +317,7 @@ Translator *SelectTranslator(const char *name)
 	case L('h','u'):   // Hungarian
 		{
 			static int stress_amps_hu[8] = {17,17, 19,19, 20,24, 24,22 };
-			static int stress_lengths_hu[8] = {185,190, 190,190, 0,0, 210,220};
-//			static int stress_lengths_hu[8] = {180,180, 200,190, 0,0, 210,225};
+			static int stress_lengths_hu[8] = {185,195, 195,190, 0,0, 210,220};
 			static const unsigned int replace_chars_hu[] = {0xd4,0xf4,0xdb,0xfb,0};
 			static const unsigned int replacement_chars_hu[] = {0x150,0x151,0x170,0x171,0};     // allow o,u-circumflex for o,u-double-acute
 

src/translate.cpp

@@ -713,6 +713,7 @@ if((wmark > 0) && (wmark < 8))
 		length = 999;
 		while(((length < 3) && (length > 0))|| (word_length > 1 && Unpronouncable(word)))
 		{
+			char *p;
 			// This word looks "unpronouncable", so speak letters individually until we
 			// find a remainder that we can pronounce.
 			word += TranslateLetter(word,phonemes,0);
@@ -722,13 +723,15 @@ if((wmark > 0) && (wmark < 8))
 				strcpy(word_phonemes,phonemes);
 				return(0);
 			}
-			if(memcmp(word,"'s ",3) == 0)
+
+			p = &word[word_length-3];
+			if(memcmp(p,"'s ",3) == 0)
 			{
 				// remove a 's suffix and pronounce this separately (not as an individual letter)
 				add_plural_suffix = 1;
-				word[0] = ' ';
-				word[1] = ' ';
-				break;
+				p[0] = ' ';
+				p[1] = ' ';
+				last_char = p[-1];
 			}
 
 			length=0;
@@ -1903,7 +1906,8 @@ if((c == '/') && (langopts.testing & 2) && isdigit(next_in) && IsAlpha(prev_out)
 			{
 				if(!IsAlpha(prev_out) || (langopts.ideographs && (c >= 0x3000)))
 				{
-					letter_count = 0;
+					if(prev_out != '\'')
+						letter_count = 0;    // don't reset count for an apostrophy within a word
 
 					if((prev_out != ' ') && (prev_out != '\''))
 					{