Fixed <say-as interpret-as="tts:key> to speak all single characters, and ignore _ and - after an alphabetic character.

git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@123 d46cf337-b52f-0410-862d-fd96e6ae7743

src/dictionary.cpp

 	const char *word2;
 	unsigned char c;
 	int  nbytes;
-	int  c2;
 	int  len;
 	char word[N_WORD_BYTES];
 	static char word_replacement[N_WORD_BYTES];
 	length = 0;
 	word2 = word1 = *wordptr;
 
-	while((word2[nbytes = utf8_in(&c2,word2,0)]==' ') && (word2[nbytes+1]=='.'))
+	while((word2[nbytes = utf8_nbytes(word2)]==' ') && (word2[nbytes+1]=='.'))
 	{
 		// look for an abbreviation of the form a.b.c
 		// try removing the spaces between the dots and looking for a match
 	if(length > 0)
 	{
 		// found an abbreviation containing dots
-		nbytes = utf8_in(&c2,word2,0);
+		nbytes = utf8_nbytes(word2);
 		memcpy(&word[length],word2,nbytes);
 		word[length+nbytes] = 0;
 		found =  LookupDict2(word, word2, ph_out, flags, end_flags, wtab);
 		if(found)
 		{
-			*flags = *flags & ~(7 << 5) | (length << 5);
+			// set the skip words flag
+			flags[0] |= FLAG_SKIPWORDS;
+			dictionary_skipwords = length;
 			return(1);
 		}
 	}

src/speak_lib.cpp

 	event_list[0].unique_identifier = unique_identifier;
 	event_list[0].user_data = user_data;
 	event_list[1].type = espeakEVENT_LIST_TERMINATED;
+	event_list[1].unique_identifier = unique_identifier;
+	event_list[1].user_data = user_data;
 
 	if (my_mode==AUDIO_OUTPUT_PLAYBACK)
 	{

src/translate.cpp

 }  // end of utf8_out
 
 
+int utf8_nbytes(const char *buf)
+{//=============================
+// Returns the number of bytes for the first UTF-8 character in buf
+	unsigned char c = (unsigned char)buf[0];
+	if(c < 0x80)
+		return(1);
+	if(c < 0xe0)
+		return(2);
+	if(c < 0xf0)
+		return(3);
+	return(4);
+}
+
+
 int utf8_in(int *c, const char *buf, int backwards)
 {//================================================
+// Read a unicode characater from a UTF8 string 
+// Returns the number of UTF8 bytes used.
+// backwards: set if we are moving backwards through the UTF8 string
 	int c1;
 	int n_bytes;
 	int ix;
 
 	// try an initial lookup in the dictionary list, we may find a pronunciation specified, or
 	// we may just find some flags
-	if((option_sayas & 0xf0) == 0x10)
+	spell_word = 0;
+	if(option_sayas == SAYAS_KEY)
+	{
+		if(word_length == 1)
+			spell_word = 4;
+	}
+
+	if(option_sayas & 0x10)
 	{
 		// SAYAS_CHAR, SAYAS_GYLPH, or SAYAS_SINGLE_CHAR
 		spell_word = option_sayas & 0xf;    // 2,3,4
 	}
 	else
 	{
-		spell_word = 0;
 		found = LookupDictList(&word1, phonemes, dictionary_flags, FLAG_ALLOW_TEXTMODE, wtab);   // the original word
 
 		// if textmode, LookupDictList() replaces word1 by the new text and returns found=0
 			}
 		}
 
+		if(option_sayas2 == SAYAS_KEY)
+		{
+			if(((c == '_') || (c == '-')) && IsAlpha(prev_in))
+			{
+				c = ' ';
+			}
+			c = towlower(c);
+		}
+
 		if(phoneme_mode)
 		{
 			all_upper_case = FLAG_PHONEMES;
 			}
 		}
 		else
-		if((option_sayas2 & 0xf0) != 0x10)
+		if((option_sayas2 & 0x30) == 0)
 		{
 			// speak as words
 

src/translate.h

 #define SAYAS_CHARS     0x12
 #define SAYAS_GLYPHS    0x13
 #define SAYAS_SINGLE_CHARS 0x14
-#define SAYAS_KEY       0x20
-#define SAYAS_DIGITS1   0x21
-#define SAYAS_DIGITS    0x30  // + number of digits
+#define SAYAS_KEY       0x24
+#define SAYAS_DIGITS    0x40  // + number of digits
+#define SAYAS_DIGITS1   0xc1
 
 // Rule:
 // [4] [match] [1 pre] [2 post] [3 phonemes] 0
 int TransposeAlphabet(char *text, int offset, int min, int max);
 int utf8_in(int *c, const char *buf, int backwards);
 int utf8_out(unsigned int c, char *buf);
+int utf8_nbytes(const char *buf);
 int lookupwchar(const unsigned short *list,int c);
 int Eof(void);
 char *strchr_w(const char *s, int c);