[1.46.28]

Ignore character U+00AD (soft hyphen) in text.
Added stress rule code for Lang=kl (Greenlandic).
Changes to languages: it, ta, si


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@294 d46cf337-b52f-0410-862d-fd96e6ae7743

src/dictionary.cpp

@@ -1129,6 +1129,7 @@ void SetWordStress(Translator *tr, char *output, unsigned int *dictionary_flags,
 	int stressflags;
 	int dflags = 0;
 	int first_primary;
+	int long_vowel;
 
 	signed char vowel_stress[N_WORD_PHONEMES/2];
 	char syllable_weight[N_WORD_PHONEMES/2];
@@ -1444,6 +1445,39 @@ void SetWordStress(Translator *tr, char *output, unsigned int *dictionary_flags,
 				vowel_stress[ix] = 4;
 		}
 		break;
+
+	case 12:  // LANG=kl (Greenlandic)
+		long_vowel = 0;
+		for(ix=1; ix < vowel_count; ix++)
+		{
+			if(vowel_stress[ix] == 4)
+					vowel_stress[ix] = 3;    // change marked stress (consonant clusters) to secondary (except the last)
+	
+			if(vowel_length[ix] > 0)
+			{
+					long_vowel = ix;
+					vowel_stress[ix] = 3;    // give secondary stress to all long vowels
+			}
+		}
+	
+		// 'stressed_syllable' gives the last marked stress
+		if(stressed_syllable == 0)
+		{
+			// no marked stress, choose the last long vowel
+			if(long_vowel > 0)
+					stressed_syllable = long_vowel;
+			else
+			{
+					// no long vowels or consonant clusters
+					if(vowel_count > 4)
+						stressed_syllable = vowel_count - 3;
+					else
+						stressed_syllable = vowel_count - 1;
+			}
+		}
+		vowel_stress[stressed_syllable] = 4;
+		max_stress = 4;
+		break;
 	}
 
 	/* now guess the complete stress pattern */

src/numbers.cpp

@@ -1146,6 +1146,7 @@ static int LookupNum2(Translator *tr, int value, const int control, char *ph_out
 // control bit 3: use feminine form of '2' (for thousands
 // control bit 4: speak zero tens
 // control bit 5: variant of ordinal number (lang=hu)
+//         bit 8   followed by decimal fraction
 
 	int found;
 	int ix;
@@ -1385,6 +1386,22 @@ static int LookupNum2(Translator *tr, int value, const int control, char *ph_out
 		}
 	}
 
+	if(tr->langopts.numbers & NUM_SINGLE_STRESS_L)
+	{
+		// only one primary stress, on the first part (tens)
+		found = 0;
+		for(ix=0; ix < (signed)strlen(ph_out); ix++)
+		{
+			if(ph_out[ix] == phonSTRESS_P)
+			{
+				if(found)
+					ph_out[ix] = phonSTRESS_3;
+				else
+					found = 1;
+			}
+		}
+	}
+	else
 	if(tr->langopts.numbers & NUM_SINGLE_STRESS)
 	{
 		// only one primary stress
@@ -1410,6 +1427,7 @@ static int LookupNum3(Translator *tr, int value, char *ph_out, int suppress_null
 //  control  bit 0,  previous thousands
 //           bit 1,  ordinal number
 //           bit 5   variant form of ordinal number
+//           bit 8   followed by decimal fraction
 	int found;
 	int hundreds;
 	int tensunits;
@@ -1610,7 +1628,7 @@ static int LookupNum3(Translator *tr, int value, char *ph_out, int suppress_null
 				x = 8;   // use variant (feminine) for before thousands and millions
 		}
 
-		if(LookupNum2(tr, tensunits, x, buf2) != 0)
+		if(LookupNum2(tr, tensunits, x | control & 0x100, buf2) != 0)
 		{
 			if(tr->langopts.numbers & NUM_SINGLE_AND)
 				ph_hundred_and[0] = 0;  // don't put 'and' after 'hundred' if there's 'and' between tens and units
@@ -1858,7 +1876,7 @@ static int TranslateNumber_1(Translator *tr, char *word, char *ph_out, unsigned
 	{
 		// this "word" ends with a decimal point
 		Lookup(tr, "_dpt", ph_append);
-		decimal_point = 1;
+		decimal_point = 0x100;
 	}
 	else
 	if(suppress_null == 0)
@@ -1913,6 +1931,7 @@ static int TranslateNumber_1(Translator *tr, char *word, char *ph_out, unsigned
 //		if((buf_digit_lookup[0] == 0) && (*p != '0') && (dot_ordinal==0))
 		if((buf_digit_lookup[0] == 0) && (*p != '0'))
 		{
+			// LANG=hu ?
 			// not found, lookup only the last digit (?? but not if dot-ordinal has been found)
 			if(LookupDictList(tr, &p, buf_digit_lookup, flags, FLAG_SUFX, wtab))  // don't match '0', or entries with $only
 			{
@@ -1920,24 +1939,38 @@ static int TranslateNumber_1(Translator *tr, char *word, char *ph_out, unsigned
 			}
 		}
 
-		if((tr->langopts.numbers2 & NUM2_PERCENT_BEFORE) && (prev_thousands == 0))
+		if(prev_thousands == 0)
 		{
-			// LANG=si, say "percent" before the number
-			p2 = word;
-			while((*p2 != ' ') && (*p2 != 0))
+			if((decimal_point == 0) && (ordinal == 0))
 			{
-				p2++;
+				// Look for special pronunciation for this number in isolation (LANG=kl)
+				sprintf(string, "_%dn", value);
+				if(Lookup(tr, string, ph_out))
+				{
+					return(1);
+				}
 			}
-			if(p2[1] == '%')
+
+			if(tr->langopts.numbers2 & NUM2_PERCENT_BEFORE)
 			{
-				Lookup(tr, "%", ph_out);
-				ph_out += strlen(ph_out);
-				p2[1] = ' ';
+				// LANG=si, say "percent" before the number
+				p2 = word;
+				while((*p2 != ' ') && (*p2 != 0))
+				{
+					p2++;
+				}
+				if(p2[1] == '%')
+				{
+					Lookup(tr, "%", ph_out);
+					ph_out += strlen(ph_out);
+					p2[1] = ' ';
+				}
 			}
 		}
+
 	}
 
-	LookupNum3(tr, value, ph_buf, suppress_null, thousandplex, prev_thousands | ordinal);
+	LookupNum3(tr, value, ph_buf, suppress_null, thousandplex, prev_thousands | ordinal | decimal_point);
 	if((thousandplex > 0) && (tr->langopts.numbers2 & 0x200))
 		sprintf(ph_out,"%s%s%s%s",ph_zeros,ph_append,ph_buf2,ph_buf);  // say "thousands" before its number
 	else
@@ -2005,6 +2038,17 @@ static int TranslateNumber_1(Translator *tr, char *word, char *ph_out, unsigned
 						n_digits += decimal_count;
 				}
 				break;
+
+			case NUM_DFRACTION_7:
+				// alternative form of decimal fraction digits, except the final digit
+				while(decimal_count-- > 1)
+				{
+					sprintf(string,"_%cd", word[n_digits]);
+					if(Lookup(tr, string, buf1) == 0)
+						break;
+					n_digits++;
+					strcat(ph_out, buf1);
+				} 
 			}
 		}
 

src/synthdata.cpp

@@ -35,7 +35,7 @@
 #include "translate.h"
 #include "wave.h"
 
-const char *version_string = "1.46.27  21.Oct.12";
+const char *version_string = "1.46.28  16.Nov.12";
 const int version_phdata  = 0x014624;
 
 int option_device_number = -1;

src/tr_languages.cpp

@@ -113,12 +113,14 @@ static void SetLetterBitsRange(Translator *tr, int group, int first, int last)
 
 // ignore these characters
 static const unsigned short chars_ignore_default[] = {
+	0xad,    1, // soft hyphtn
 	0x200c,  1, // zero width non-joiner
 	0x200d,  1, // zero width joiner
 	0, 0 };
 
 // alternatively, ignore characters but allow zero-width-non-joiner (lang-fa)
 static const unsigned short chars_ignore_zwnj_hyphen[] = {
+	0xad,    1, // soft hyphen
 	0x200c,  '-', // zero width non-joiner, replace with hyphen
 	0x200d,  1, // zero width joiner
 	0, 0 };
@@ -943,6 +945,14 @@ SetLengthMods(tr,3);  // all equal
 		}
 		break;
 
+	case L('k','l'):   // Greenlandic
+		{
+			tr->langopts.stress_rule = 12;
+			tr->langopts.stress_flags = S_NO_AUTO_2;
+			tr->langopts.numbers = NUM_DECIMAL_COMMA | NUM_SWAP_TENS | NUM_HUNDRED_AND | NUM_OMIT_1_HUNDRED | NUM_ORDINAL_DOT | NUM_1900 | NUM_ROMAN | NUM_ROMAN_CAPITALS | NUM_ROMAN_ORDINAL;
+		}
+		break;
+
 	case L('k','o'):   // Korean, TEST
 		{
 			static const char ko_ivowels[] = {0x63,0x64,0x67,0x68,0x6d,0x72,0x74,0x75,0};  // y and i vowels
@@ -1207,7 +1217,7 @@ SetLengthMods(tr,3);  // all equal
 		
 			tr->langopts.param[LOPT_UNPRONOUNCABLE] = 1;   // disable check for unpronouncable words
 			tr->langopts.suffix_add_e = tr->letter_bits_offset + 0x4a;   //virama
-			tr->langopts.numbers =  NUM_OMIT_1_THOUSAND;
+			tr->langopts.numbers =  NUM_OMIT_1_THOUSAND | NUM_SINGLE_STRESS_L | NUM_DFRACTION_7;
 			tr->langopts.numbers2 =  NUM2_PERCENT_BEFORE;
 			tr->langopts.break_numbers = 0x14aa8;  // for languages which have numbers for 100,000 and 100,00,000, eg Hindi
 		}

src/translate.h

@@ -402,6 +402,7 @@ typedef struct {
 #define NUM_DFRACTION_4  0x8000
 #define NUM_DFRACTION_5  0xa000
 #define NUM_DFRACTION_6  0xc000
+#define NUM_DFRACTION_7  0xe000    // lang=si, alternative form of number for decimal fraction digits (except the last)
 #define NUM_ORDINAL_DOT   0x10000
 #define NUM_NOPAUSE       0x20000
 #define NUM_AND_HUNDRED   0x40000
@@ -414,6 +415,7 @@ typedef struct {
 #define NUM_ROMAN_CAPITALS 0x2000000
 #define NUM_ROMAN_AFTER    0x4000000
 #define NUM_ROMAN_ORDINAL  0x8000000
+#define NUM_SINGLE_STRESS_L  0x10000000
 
 	// bits0-1=which numbers routine to use.
 	// bit2=  thousands separator must be space
@@ -442,6 +444,7 @@ typedef struct {
 	// bit25= Roman numbers only if upper case
 	// bit26= say "roman" after the number, not before
 	// bit27= Roman numbers are ordinal numbers
+   // bit28= only one primary stress in tens+units (on the tens)
 	int numbers;
 
 #define NUM2_THOUSANDS_VAR1     0x40