Staticanalysis (#1457)

Fixes for code smells from cppcheck static code analysis tool.

Also includes some function refactoring.

src/libespeak-ng/common.c

@@ -132,7 +132,6 @@ int utf8_in2(int *c, const char *buf, int backwards)
 
 	int c1;
 	int n_bytes;
-	int ix;
 	static const unsigned char mask[4] = { 0xff, 0x1f, 0x0f, 0x07 };
 
 	// find the start of the next/previous character
@@ -155,6 +154,7 @@ int utf8_in2(int *c, const char *buf, int backwards)
 			n_bytes = 3;
 
 		c1 &= mask[n_bytes];
+		int ix;
 		for (ix = 0; ix < n_bytes; ix++)
 		{
 			if (!*buf)
@@ -271,9 +271,7 @@ int IsSpace(unsigned int c)
 int isspace2(unsigned int c)
 {
 	// can't use isspace() because on Windows, isspace(0xe1) gives TRUE !
-	int c2;
-
-	if (((c2 = (c & 0xff)) == 0) || (c > ' '))
+	if ( ((c & 0xff) == 0) || (c > ' '))
 		return 0;
 	return 1;
 }
@@ -290,10 +288,10 @@ int Read4Bytes(FILE *f)
 {
 	// Read 4 bytes (least significant first) into a word
 	int ix;
-	unsigned char c;
 	int acc = 0;
 
 	for (ix = 0; ix < 4; ix++) {
+		unsigned char c;
 		c = fgetc(f) & 0xff;
 		acc += (c << (ix*8));
 	}

src/libespeak-ng/compiledata.c

@@ -1207,13 +1207,13 @@ static int LoadWavefile(FILE *f, const char *fname)
 	fseek(f, 40, SEEK_SET);
 
 	if ((sr1 != samplerate_native) || (sr2 != sr1*2)) {
-		int fd_temp;
 		char command[sizeof(path_home)+250];
 
 		failed = false;
 
 #ifdef HAVE_MKSTEMP
 		strcpy(fname_temp, "/tmp/espeakXXXXXX");
+		int fd_temp;
 		if ((fd_temp = mkstemp(fname_temp)) >= 0)
 			close(fd_temp);
 #else

src/libespeak-ng/readclause.c

@@ -186,11 +186,11 @@ static const char *LookupSpecial(Translator *tr, const char *string, char *text_
 {
 	unsigned int flags[2];
 	char phonemes[55];
-	char phonemes2[55];
 	char *string1 = (char *)string;
 
 	flags[0] = flags[1] = 0;
 	if (LookupDictList(tr, &string1, phonemes, flags, 0, NULL)) {
+		char phonemes2[55];
 		SetWordStress(tr, phonemes, flags, -1, 0);
 		DecodePhonemes(phonemes, phonemes2);
 		sprintf(text_out, "[\002%s]]", phonemes2);
@@ -208,7 +208,6 @@ static const char *LookupCharName(Translator *tr, int c, int only)
 	unsigned int flags[2];
 	char single_letter[24];
 	char phonemes[60];
-	char phonemes2[60];
 	const char *lang_name = NULL;
 	char *string;
 	static char buf[60];
@@ -255,6 +254,7 @@ static const char *LookupCharName(Translator *tr, int c, int only)
 	}
 
 	if (phonemes[0]) {
+		char phonemes2[60];
 		if (lang_name) {
 			SetWordStress(translator2, phonemes, flags, -1, 0);
 			DecodePhonemes(phonemes, phonemes2);
@@ -277,7 +277,7 @@ static int AnnouncePunctuation(Translator *tr, int c1, int *c2_ptr, char *output
 	// c1:  the punctuation character
 	// c2:  the following character
 
-	int punct_count;
+
 	const char *punctname = NULL;
 	int soundicon;
 	int attributes;
@@ -286,7 +286,6 @@ static int AnnouncePunctuation(Translator *tr, int c1, int *c2_ptr, char *output
 	int len;
 	int bufix1;
 	char buf[200];
-	char buf2[80];
 	char ph_buf[30];
 
 	c2 = *c2_ptr;
@@ -307,8 +306,9 @@ static int AnnouncePunctuation(Translator *tr, int c1, int *c2_ptr, char *output
 		if (punctname == NULL)
 			return -1;
 
+
 		if ((*bufix == 0) || (end_clause == 0) || (tr->langopts.param[LOPT_ANNOUNCE_PUNCT] & 2)) {
-			punct_count = 1;
+			int punct_count = 1;
 			while (!Eof() && (c2 == c1) && (c1 != '<')) { // don't eat extra '<', it can miss XML tags
 				punct_count++;
 				c2 = GetC();
@@ -324,6 +324,7 @@ static int AnnouncePunctuation(Translator *tr, int c1, int *c2_ptr, char *output
 				if (embedded_value[EMBED_S] < 300)
 					sprintf(buf, "\001+10S"); // Speak punctuation name faster, unless we are already speaking fast.  It would upset Sonic SpeedUp
 
+				char buf2[80];
 				while (punct_count-- > 0) {
 					sprintf(buf2, " %s", punctname);
 					strcat(buf, buf2);
@@ -381,7 +382,6 @@ int AddNameData(const char *name, int wide)
 
 	int ix;
 	int len;
-	void *vp;
 
 	if (wide) {
 		len = (wcslen((const wchar_t *)name)+1)*sizeof(wchar_t);
@@ -391,6 +391,7 @@ int AddNameData(const char *name, int wide)
 
 	if (namedata_ix+len >= n_namedata) {
 		// allocate more space for marker names
+		void *vp;
 		if ((vp = realloc(namedata, namedata_ix+len + 1000)) == NULL)
 			return -1;  // failed to allocate, original data is unchanged but ignore this new name
 		// !!! Bug?? If the allocated data shifts position, then pointers given to user application will be invalid
@@ -490,7 +491,6 @@ int ReadClause(Translator *tr, char *buf, short *charix, int *charix_top, int n_
 	int c1 = ' '; // current character
 	int c2; // next character
 	int cprev = ' '; // previous character
-	int cprev2 = ' ';
 	int c_next;
 	int parag;
 	int ix = 0;
@@ -548,8 +548,7 @@ int ReadClause(Translator *tr, char *buf, short *charix, int *charix_top, int n_
 				return CLAUSE_NONE;
 			}
 		}
-
-		cprev2 = cprev;
+		int cprev2 = cprev;
 		cprev = c1;
 		c1 = c2;
 

src/libespeak-ng/soundicon.c

@@ -72,7 +72,6 @@ static espeak_ng_STATUS LoadSoundFile(const char *fname, int index, espeak_ng_ER
 	f = NULL;
 	if ((f = fopen(fname, "rb")) != NULL) {
 		int ix;
-		int fd_temp;
 		int header[3];
 		char command[sizeof(fname2)+sizeof(fname2)+40];
 
@@ -92,6 +91,7 @@ static espeak_ng_STATUS LoadSoundFile(const char *fname, int index, espeak_ng_ER
 
 #ifdef HAVE_MKSTEMP
 			strcpy(fname_temp, "/tmp/espeakXXXXXX");
+			int fd_temp;
 			if ((fd_temp = mkstemp(fname_temp)) >= 0)
 				close(fd_temp);
 #else

src/libespeak-ng/ssml.c

@@ -132,16 +132,16 @@ static int attrnumber(const wchar_t *pw, int default_value, int type)
 static int attrcopy_utf8(char *buf, const wchar_t *pw, int len)
 {
 	// Convert attribute string into utf8, write to buf, and return its utf8 length
-	unsigned int c;
 	int ix = 0;
-	int n;
-	int prev_c = 0;
 
 	if (pw != NULL) {
+		unsigned int c;
+		int prev_c = 0;
 		while ((ix < (len-4)) && ((c = *pw++) != 0)) {
 			if ((c == '"') && (prev_c != '\\'))
 				break; // " indicates end of attribute, unless preceded by backstroke
-			n = utf8_out(c, &buf[ix]);
+
+			int n = utf8_out(c, &buf[ix]);
 			ix += n;
 			prev_c = c;
 		}
@@ -208,12 +208,10 @@ static const char *VoiceFromStack(SSML_STACK *ssml_stack, int n_ssml_stack, espe
 	const char *p;
 	SSML_STACK *sp;
 	const char *v_id;
-	int voice_name_specified;
 	int voice_found;
 	espeak_VOICE voice_select;
 	static char voice_name[40];
 	char language[40];
-	char buf[80];
 
 	MAKE_MEM_UNDEFINED(&voice_name, sizeof(voice_name));
 
@@ -226,7 +224,7 @@ static const char *VoiceFromStack(SSML_STACK *ssml_stack, int n_ssml_stack, espe
 
 	for (ix = 0; ix < n_ssml_stack; ix++) {
 		sp = &ssml_stack[ix];
-		voice_name_specified = 0;
+		int voice_name_specified = 0;
 
 		if ((sp->voice_name[0] != 0) && (SelectVoiceByName(NULL, sp->voice_name) != NULL)) {
 			voice_name_specified = 1;
@@ -270,6 +268,7 @@ static const char *VoiceFromStack(SSML_STACK *ssml_stack, int n_ssml_stack, espe
 
 	if ((strchr(v_id, '+') == NULL) && ((voice_select.gender == ENGENDER_UNKNOWN) || (voice_select.gender == base_voice->gender)) && (base_voice_variant_name[0] != 0)) {
 		// a voice variant has not been selected, use the original voice variant
+		char buf[80];
 		sprintf(buf, "%s+%s", v_id, base_voice_variant_name);
 		strncpy0(voice_name, buf, sizeof(voice_name));
 		return voice_name;
@@ -318,12 +317,6 @@ static int GetVoiceAttributes(wchar_t *pw, int tag_type, SSML_STACK *ssml_sp, SS
 	// a voice change.
 	// Returns  CLAUSE_TYPE_VOICE_CHANGE if there is a voice change
 
-	const wchar_t *lang;
-	const wchar_t *gender;
-	const wchar_t *name;
-	const wchar_t *age;
-	const wchar_t *variant;
-	int value;
 	const char *new_voice_id;
 
 	static const MNEM_TAB mnem_gender[] = {
@@ -338,6 +331,12 @@ static int GetVoiceAttributes(wchar_t *pw, int tag_type, SSML_STACK *ssml_sp, SS
 		if (n_ssml_stack > 1)
 			n_ssml_stack--;
 	} else {
+		const wchar_t *lang;
+    	const wchar_t *gender;
+    	const wchar_t *name;
+    	const wchar_t *age;
+    	const wchar_t *variant;
+
 		// add a stack frame if any voice details are specified
 		lang = GetSsmlAttribute(pw, "xml:lang");
 
@@ -359,6 +358,8 @@ static int GetVoiceAttributes(wchar_t *pw, int tag_type, SSML_STACK *ssml_sp, SS
 
 		ssml_sp = &ssml_stack[n_ssml_stack++];
 
+		int value;
+
 		attrcopy_utf8(ssml_sp->language, lang, sizeof(ssml_sp->language));
 		attrcopy_utf8(ssml_sp->voice_name, name, sizeof(ssml_sp->voice_name));
 		if ((value = attrnumber(variant, 1, 0)) > 0)
@@ -384,7 +385,6 @@ static void ProcessParamStack(char *outbuf, int *outix, int n_param_stack, PARAM
 	// Set the speech parameters from the parameter stack
 	int param;
 	int ix;
-	int value;
 	char buf[20];
 	int new_parameters[N_SPEECH_PARAM];
 	static const char cmd_letter[N_SPEECH_PARAM] = { 0, 'S', 'A', 'P', 'R', 0, 'C', 0, 0, 0, 0, 0, 'F' }; // embedded command letters
@@ -400,6 +400,7 @@ static void ProcessParamStack(char *outbuf, int *outix, int n_param_stack, PARAM
 	}
 
 	for (param = 0; param < N_SPEECH_PARAM; param++) {
+		int value;
 		if ((value = new_parameters[param]) != speech_parameters[param]) {
 			buf[0] = 0;
 
@@ -471,14 +472,14 @@ static int ReplaceKeyName(char *outbuf, int index, int *outix)
 		{ NULL,            0 }
 	};
 
-	int ix;
 	int letter;
 	char *p;
 
 	p = &outbuf[index];
 
 	if ((letter = LookupMnem(keynames, p)) != 0) {
-		ix = utf8_out(letter, p);
+		int ix;
+		 ix = utf8_out(letter, p);
 		*outix = index + ix;
 		return letter;
 	}
@@ -488,7 +489,7 @@ static int ReplaceKeyName(char *outbuf, int index, int *outix)
 static void SetProsodyParameter(int param_type, const wchar_t *attr1, PARAM_STACK *sp, PARAM_STACK *param_stack, int *speech_parameters)
 {
 	int value;
-	int sign;
+
 
 	static const MNEM_TAB mnem_volume[] = {
 		{ "default", 100 },
@@ -539,7 +540,7 @@ static void SetProsodyParameter(int param_type, const wchar_t *attr1, PARAM_STAC
 		// mnemonic specifies a value as a percentage of the base pitch/range/rate/volume
 		sp->parameter[param_type] = (param_stack[0].parameter[param_type] * value)/100;
 	} else {
-		sign = attr_prosody_value(param_type, attr1, &value);
+		int sign = attr_prosody_value(param_type, attr1, &value);
 
 		if (sign == 0)
 			sp->parameter[param_type] = value; // absolute value in Hz
@@ -560,7 +561,6 @@ int ProcessSsmlTag(wchar_t *xml_buf, char *outbuf, int *outix, int n_outbuf, con
 
 	unsigned int ix;
 	int index;
-	int c;
 	int tag_type;
 	int value;
 	int value2;
@@ -571,7 +571,6 @@ int ProcessSsmlTag(wchar_t *xml_buf, char *outbuf, int *outix, int n_outbuf, con
 	const wchar_t *attr2;
 	const wchar_t *attr3;
 	int terminator;
-	char *uri;
 	int param_type;
 	char tag_name[40];
 	char buf[160];
@@ -653,6 +652,7 @@ int ProcessSsmlTag(wchar_t *xml_buf, char *outbuf, int *outix, int n_outbuf, con
 	};
 
 	for (ix = 0; ix < (sizeof(tag_name)-1); ix++) {
+		int c;
 		if (((c = xml_buf[ix]) == 0) || iswspace(c))
 			break;
 		tag_name[ix] = tolower((char)c);
@@ -815,11 +815,11 @@ int ProcessSsmlTag(wchar_t *xml_buf, char *outbuf, int *outix, int n_outbuf, con
 		sp = PushParamStack(tag_type, n_param_stack, (PARAM_STACK *)param_stack);
 
 		if ((attr1 = GetSsmlAttribute(px, "src")) != NULL) {
-			char fname[256];
 			attrcopy_utf8(buf, attr1, sizeof(buf));
 
 			if (uri_callback == NULL) {
 				if ((xmlbase != NULL) && (buf[0] != '/')) {
+					char fname[256];
 					sprintf(fname, "%s/%s", xmlbase, buf);
 					index = LoadSoundFile2(fname);
 				} else
@@ -832,6 +832,7 @@ int ProcessSsmlTag(wchar_t *xml_buf, char *outbuf, int *outix, int n_outbuf, con
 				}
 			} else {
 				if ((index = AddNameData(buf, 0)) >= 0) {
+					char *uri;
 					uri = &namedata[index];
 					if (uri_callback(1, uri, xmlbase) == 0) {
 						sprintf(buf, "%c%dU", CTRL_EMBEDDED, index);

src/libespeak-ng/translate.c

@@ -45,7 +45,9 @@
 #include "speech.h"               // for MAKE_MEM_UNDEFINED
 #include "translateword.h"
 
+static int CalcWordLength(int source_index, int charix_top, short int *charix, WORD_TAB *words, int word_count);
 static void CombineFlag(Translator *tr, WORD_TAB *wtab, char *word, int *flags, unsigned char *p, char *word_phonemes);
+static void SwitchLanguage(char *word, char *word_phonemes);
 
 Translator *translator = NULL; // the main translator
 Translator *translator2 = NULL; // secondary translator for certain words
@@ -143,8 +145,7 @@ int TranslateWord(Translator *tr, char *word_start, WORD_TAB *wtab, char *word_o
 {
 	char words_phonemes[N_WORD_PHONEMES]; // a word translated into phoneme codes
 	char *phonemes = words_phonemes;
-	int available = N_WORD_PHONEMES;
-	bool first_word = true;
+
 
 	int flags = TranslateWord3(tr, word_start, wtab, word_out, &any_stressed_words, current_alphabet, word_phonemes, sizeof(word_phonemes));
 	if (flags & FLAG_TEXTMODE && word_out) {
@@ -156,6 +157,8 @@ int TranslateWord(Translator *tr, char *word_start, WORD_TAB *wtab, char *word_o
 		strcpy(word+2, word_out);
 		word_out = word+2;
 
+			bool first_word = true;
+			int available = N_WORD_PHONEMES;
 		while (*word_out && available > 1) {
 			int c;
 			utf8_in(&c, word_out);
@@ -232,11 +235,9 @@ static void Word_EmbeddedCmd()
 {
 	// Process embedded commands for emphasis, sayas, and break
 	int embedded_cmd;
-	int value;
-
 	do {
 		embedded_cmd = embedded_list[embedded_read++];
-		value = embedded_cmd >> 8;
+		int value = embedded_cmd >> 8;
 
 		switch (embedded_cmd & 0x1f)
 		{
@@ -318,11 +319,8 @@ static int TranslateWord2(Translator *tr, char *word, WORD_TAB *wtab, int pre_pa
 	bool first_phoneme = true;
 	int source_ix;
 	int len;
-	int ix;
-	const char *new_language;
 	int bad_phoneme;
 	int word_flags;
-	int word_copy_len;
 	char word_copy[N_WORD_BYTES+1];
 	char word_replaced[N_WORD_BYTES+1];
 	char old_dictionary_name[40];
@@ -347,12 +345,8 @@ static int TranslateWord2(Translator *tr, char *word, WORD_TAB *wtab, int pre_pa
 	if ((word[0] == 0) || (word_flags & FLAG_DELETE_WORD)) {
 		// nothing to translate.  Add a dummy phoneme to carry any embedded commands
 		if (embedded_flag) {
-			ph_list2[n_ph_list2].phcode = phonEND_WORD;
-			ph_list2[n_ph_list2].stresslevel = 0;
+			SetPlist2(&ph_list2[n_ph_list2], phonEND_WORD);
 			ph_list2[n_ph_list2].wordstress = 0;
-			ph_list2[n_ph_list2].tone_ph = 0;
-			ph_list2[n_ph_list2].synthflags = embedded_flag;
-			ph_list2[n_ph_list2].sourceix = 0;
 			n_ph_list2++;
 			embedded_flag = 0;
 		}
@@ -391,32 +385,18 @@ static int TranslateWord2(Translator *tr, char *word, WORD_TAB *wtab, int pre_pa
 	p = (unsigned char *)word_phonemes;
 	if (word_flags & FLAG_PHONEMES) {
 		// The input is in phoneme mnemonics, not language text
-		int c1;
-		char lang_name[12];
 
 		if (memcmp(word, "_^_", 3) == 0) {
-			// switch languages
-			word += 3;
-			for (ix = 0;;) {
-				c1 = *word++;
-				if ((c1 == ' ') || (c1 == 0))
-					break;
-				lang_name[ix++] = tolower(c1);
-			}
-			lang_name[ix] = 0;
-
-			if ((ix = LookupPhonemeTable(lang_name)) > 0) {
-				SelectPhonemeTable(ix);
-				word_phonemes[0] = phonSWITCH;
-				word_phonemes[1] = ix;
-				word_phonemes[2] = 0;
-			}
-		} else
+			SwitchLanguage(word, word_phonemes);
+		} else {
 			EncodePhonemes(word, word_phonemes, &bad_phoneme);
+		}
+
 		flags = FLAG_FOUND;
 	} else {
 		int c2;
-		ix = 0;
+		int ix = 0;
+		int word_copy_len;
 		while (((c2 = word_copy[ix] = word[ix]) != ' ') && (c2 != 0) && (ix < N_WORD_BYTES)) ix++;
 		word_copy_len = ix;
 
@@ -440,6 +420,7 @@ static int TranslateWord2(Translator *tr, char *word, WORD_TAB *wtab, int pre_pa
 				// this word uses a different language
 				memcpy(word, word_copy, word_copy_len);
 
+				const char *new_language;
 				new_language = (char *)(&p[1]);
 				if (new_language[0] == 0)
 					new_language = ESPEAKNG_DEFAULT_VOICE;
@@ -853,9 +834,6 @@ static int TranslateChar(Translator *tr, char *ptr, int prev_in, unsigned int c,
 	// To allow language specific examination and replacement of characters
 
 	int code;
-	int initial;
-	int medial;
-	int final;
 	int next2;
 
 	static const unsigned char hangul_compatibility[0x34] = {
@@ -871,9 +849,9 @@ static int TranslateChar(Translator *tr, char *ptr, int prev_in, unsigned int c,
 	// check for Korean Hangul letters
 	if (((code = c - 0xac00) >= 0) && (c <= 0xd7af)) {
 		// break a syllable hangul into 2 or 3 individual jamo
-		initial = (code/28)/21;
-		medial = (code/28) % 21;
-		final = code % 28;
+		int initial = (code/28)/21;
+		int medial = (code/28) % 21;
+		int final = code % 28;
 
 		if (initial == 11) {
 			// null initial
@@ -922,13 +900,12 @@ static const char *UCase_ga[] = { "bp", "bhf", "dt", "gc", "hA", "mb", "nd", "ng
 
 static int UpperCaseInWord(Translator *tr, char *word, int c)
 {
-	int ix;
-	int len;
-	const char *p;
-
 	if (tr->translator_name == L('g', 'a')) {
-		// Irish
+		int ix;
+		const char *p;
+
 		for (ix = 0;; ix++) {
+			int len;
 			if ((p = UCase_ga[ix]) == NULL)
 				break;
 
@@ -948,11 +925,9 @@ void TranslateClause(Translator *tr, int *tone_out, char **voice_change)
 	int c;
 	int cc = 0;
 	unsigned int source_index = 0;
-	unsigned int prev_source_index = 0;
 	int source_index_word = 0;
 	int prev_in;
 	int prev_out = ' ';
-	int prev_out2;
 	int prev_in_save = 0;
 	int next_in;
 	int next_in_nbytes;
@@ -1082,16 +1057,9 @@ void TranslateClause(Translator *tr, int *tone_out, char **voice_change)
 	words[0].start = ix;
 	words[0].flags = 0;
 
-	for (j = 0; charix[j] <= 0; j++) ;
-	words[0].sourceix = charix[j];
-	k = 0;
-	while (charix[j] != 0) {
-		// count the number of characters (excluding multibyte continuation bytes)
-		if (charix[j++] != -1)
-			k++;
-	}
-	words[0].length = k;
+	words[0].length = CalcWordLength(source_index, charix_top, charix, words, 0);
 
+	int prev_out2;
 	while (!finished && (ix < (int)sizeof(sbuf) - 1)) {
 		prev_out2 = prev_out;
 		utf8_in2(&prev_out, &sbuf[ix-1], 1);
@@ -1107,7 +1075,7 @@ void TranslateClause(Translator *tr, int *tone_out, char **voice_change)
 		} else if (source_index > 0)
 			utf8_in2(&prev_in, &source[source_index-1], 1);
 
-		prev_source_index = source_index;
+		unsigned int prev_source_index = source_index;
 
 		if (char_inserted) {
 			c = char_inserted;
@@ -1467,16 +1435,7 @@ void TranslateClause(Translator *tr, int *tone_out, char **voice_change)
 				words[word_count].start = ix;
 				words[word_count].flags = 0;
 
-				for (j = source_index; j < charix_top && charix[j] <= 0; j++) // skip blanks
-					;
-				words[word_count].sourceix = charix[j];
-				k = 0;
-				while (charix[j] != 0) {
-					// count the number of characters (excluding multibyte continuation bytes)
-					if (charix[j++] != -1)
-						k++;
-				}
-				words[word_count].length = k;
+				words[word_count].length = CalcWordLength(source_index, charix_top, charix, words, word_count);
 
 				word_flags = next_word_flags;
 				next_word_flags = 0;
@@ -1530,7 +1489,6 @@ void TranslateClause(Translator *tr, int *tone_out, char **voice_change)
 		int c_temp;
 		char *pn;
 		char *pw;
-		int nw;
 		char number_buf[150];
 		WORD_TAB num_wtab[50]; // copy of 'words', when splitting numbers into parts
 
@@ -1575,12 +1533,13 @@ void TranslateClause(Translator *tr, int *tone_out, char **voice_change)
 
 		if (n_digits > 4 && n_digits <= 32) {
 			// word is entirely digits, insert commas and break into 3 digit "words"
+			int nw = 0;
+
 			number_buf[0] = ' ';
 			number_buf[1] = ' ';
 			number_buf[2] = ' ';
 			pn = &number_buf[3];
 			nx = n_digits;
-			nw = 0;
 
 			if ((n_digits > tr->langopts.max_digits) || (word[0] == '0'))
 				words[ix].flags |= FLAG_INDIVIDUAL_DIGITS;
@@ -1711,6 +1670,20 @@ void TranslateClause(Translator *tr, int *tone_out, char **voice_change)
 	}
 }
 
+static int CalcWordLength(int source_index, int charix_top, short int *charix, WORD_TAB *words, int word_count) {
+	int j;
+	int k;
+
+	for (j = source_index; j < charix_top && charix[j] <= 0; j++); // skip blanks
+	words[word_count].sourceix = charix[j];
+	k = 0;
+	while (charix[j] != 0) {
+		// count the number of characters (excluding multibyte continuation bytes)
+		if (charix[j++] != -1)
+			k++;
+	}
+	return k;
+	}
 
 static void CombineFlag(Translator *tr, WORD_TAB *wtab, char *word, int *flags, unsigned char *p, char *word_phonemes) {
 	// combine a preposition with the following word
@@ -1775,6 +1748,29 @@ static void CombineFlag(Translator *tr, WORD_TAB *wtab, char *word, int *flags,
 	}
 }
 
+static void SwitchLanguage(char *word, char *word_phonemes) {
+	char lang_name[12];
+	int ix;
+
+	word += 3;
+
+	for (ix = 0;;) {
+		int  c1;
+		c1 = *word++;
+		if ((c1 == ' ') || (c1 == 0))
+			break;
+		lang_name[ix++] = tolower(c1);
+	}
+	lang_name[ix] = 0;
+
+	if ((ix = LookupPhonemeTable(lang_name)) > 0) {
+		SelectPhonemeTable(ix);
+		word_phonemes[0] = phonSWITCH;
+		word_phonemes[1] = ix;
+		word_phonemes[2] = 0;
+	}
+}
+
 void InitText(int control)
 {
 	count_sentences = 0;
@@ -1797,4 +1793,4 @@ void InitText(int control)
 
 	if ((control & espeakKEEP_NAMEDATA) == 0)
 		InitNamedata();
-}
+}

src/libespeak-ng/translateword.c

@@ -674,16 +674,15 @@ int TranslateWord3(Translator *tr, char *word_start, WORD_TAB *wtab, char *word_
 void ApplySpecialAttribute2(Translator *tr, char *phonemes, int dict_flags)
 {
 	// apply after the translation is complete
-
-	int ix;
 	int len;
-	char *p;
+
 
 	len = strlen(phonemes);
 
 	if (tr->langopts.param[LOPT_ALT] & 2) {
-		for (ix = 0; ix < (len-1); ix++) {
+		for (int ix = 0; ix < (len-1); ix++) {
 			if (phonemes[ix] == phonSTRESS_P) {
+				char *p;
 				p = &phonemes[ix+1];
 				if ((dict_flags & FLAG_ALT2_TRANS) != 0) {
 					if (*p == PhonemeCode('E'))
@@ -795,18 +794,11 @@ static int TranslateLetter(Translator *tr, char *word, char *phonemes, int contr
 	int n_bytes;
 	int letter;
 	int len;
-	int ix;
-	int c;
-	char *p2;
-	char *pbuf;
-	const char *modifier;
 	ALPHABET *alphabet;
 	int al_offset;
 	int al_flags;
-	int language;
 	int number;
 	int phontab_1;
-	int speak_letter_number;
 	char capital[30];
 	char ph_buf[80];
 	char ph_buf2[80];
@@ -834,8 +826,11 @@ static int TranslateLetter(Translator *tr, char *word, char *phonemes, int contr
 
 	if (ph_buf[0] == 0) {
 		// is this a subscript or superscript letter ?
+		int c;
 		if ((c = IsSuperscript(letter)) != 0) {
 			letter = c & 0x3fff;
+
+			const char *modifier;
 			if ((control & 4 ) && ((modifier = modifiers[c >> 14]) != NULL)) {
 				// don't say "superscript" during normal text reading
 				Lookup(tr, modifier, capital);
@@ -909,6 +904,7 @@ static int TranslateLetter(Translator *tr, char *word, char *phonemes, int contr
 	// caution: SetWordStress() etc don't expect phonSWITCH + phoneme table number
 
 	if (ph_buf[0] == 0) {
+		int language;
 		if ((al_offset != 0) && (al_offset == translator->langopts.alt_alphabet))
 			language = translator->langopts.alt_alphabet_lang;
 		else if ((alphabet != NULL) && (alphabet->language != 0) && !(al_flags & AL_NOT_LETTERS))
@@ -918,19 +914,21 @@ static int TranslateLetter(Translator *tr, char *word, char *phonemes, int contr
 
 		if ((language != tr->translator_name) || (language == L('k', 'o'))) {
 			char *p3;
-			int initial, code;
+			//int initial, code;
 			char hangul_buf[12];
 
 			// speak in the language for this alphabet (or English)
 			ph_buf[2] = SetTranslator3(WordToString2(language));
 
 			if (translator3 != NULL) {
+				int code;
 				if (((code = letter - 0xac00) >= 0) && (letter <= 0xd7af)) {
 					// Special case for Korean letters.
 					// break a syllable hangul into 2 or 3 individual jamo
 
 					hangul_buf[0] = ' ';
 					p3 = &hangul_buf[1];
+					int initial;
 					if ((initial = (code/28)/21) != 11) {
 						p3 += utf8_out(initial + 0x1100, p3);
 					}
@@ -966,51 +964,50 @@ static int TranslateLetter(Translator *tr, char *word, char *phonemes, int contr
 
 	if (ph_buf[0] == 0) {
 		// character name not found
+		int speak_letter_number = 1;
+		if (!(al_flags & AL_NO_SYMBOL)) {
+			if (iswalpha(letter))
+				Lookup(translator, "_?A", ph_buf);
 
-		if (ph_buf[0] == 0) {
-			speak_letter_number = 1;
-			if (!(al_flags & AL_NO_SYMBOL)) {
-				if (iswalpha(letter))
-					Lookup(translator, "_?A", ph_buf);
+			if ((ph_buf[0] == 0) && !iswspace(letter))
+				Lookup(translator, "_??", ph_buf);
 
-				if ((ph_buf[0] == 0) && !iswspace(letter))
-					Lookup(translator, "_??", ph_buf);
-
-				if (ph_buf[0] == 0)
-					EncodePhonemes("l'et@", ph_buf, NULL);
-			}
+			if (ph_buf[0] == 0)
+				EncodePhonemes("l'et@", ph_buf, NULL);
+		}
 
-			if (!(control & 4) && (al_flags & AL_NOT_CODE)) {
-				// don't speak the character code number, unless we want full details of this character
-				speak_letter_number = 0;
-			}
+		if (!(control & 4) && (al_flags & AL_NOT_CODE)) {
+			// don't speak the character code number, unless we want full details of this character
+			speak_letter_number = 0;
+		}
 
-			if (speak_letter_number) {
-				if (al_offset == 0x2800) {
-					// braille dots symbol, list the numbered dots
-					p2 = hexbuf;
-					for (ix = 0; ix < 8; ix++) {
-						if (letter & (1 << ix))
-							*p2++ = '1'+ix;
-					}
-					*p2 = 0;
-				} else {
-					// speak the hexadecimal number of the character code
-					sprintf(hexbuf, "%x", letter);
+		if (speak_letter_number) {
+			char *p2;
+			if (al_offset == 0x2800) {
+				// braille dots symbol, list the numbered dots
+				p2 = hexbuf;
+				for (int ix = 0; ix < 8; ix++) {
+					if (letter & (1 << ix))
+						*p2++ = '1'+ix;
 				}
+				*p2 = 0;
+			} else {
+				// speak the hexadecimal number of the character code
+				sprintf(hexbuf, "%x", letter);
+			}
 
-				pbuf = ph_buf;
-				for (p2 = hexbuf; *p2 != 0; p2++) {
-					pbuf += strlen(pbuf);
-					*pbuf++ = phonPAUSE_VSHORT;
-					LookupLetter(translator, *p2, 0, pbuf, 1);
-					if (((pbuf[0] == 0) || (pbuf[0] == phonSWITCH)) && (*p2 >= 'a')) {
-						// This language has no translation for 'a' to 'f', speak English names using base phonemes
-						EncodePhonemes(hex_letters[*p2 - 'a'], pbuf, NULL);
-					}
+			char *pbuf;
+			pbuf = ph_buf;
+			for (p2 = hexbuf; *p2 != 0; p2++) {
+				pbuf += strlen(pbuf);
+				*pbuf++ = phonPAUSE_VSHORT;
+				LookupLetter(translator, *p2, 0, pbuf, 1);
+				if (((pbuf[0] == 0) || (pbuf[0] == phonSWITCH)) && (*p2 >= 'a')) {
+					// This language has no translation for 'a' to 'f', speak English names using base phonemes
+					EncodePhonemes(hex_letters[*p2 - 'a'], pbuf, NULL);
 				}
-				strcat(pbuf, pause_string);
 			}
+			strcat(pbuf, pause_string);
 		}
 	}
 
@@ -1049,8 +1046,6 @@ static int CheckDottedAbbrev(char *word1)
 {
 	int wc;
 	int count = 0;
-	int nbytes;
-	int ok;
 	int ix;
 	char *word;
 	char *wbuf;
@@ -1060,8 +1055,8 @@ static int CheckDottedAbbrev(char *word1)
 	wbuf = word_buf;
 
 	for (;;) {
-		ok = 0;
-		nbytes = utf8_in(&wc, word);
+		int ok = 0;
+		int nbytes = utf8_in(&wc, word);
 		if ((word[nbytes] == ' ') && IsAlpha(wc)) {
 			if (word[nbytes+1] == '.') {
 				if (word[nbytes+2] == ' ')