--- parser3/src/main/pa_string.C	2003/09/30 05:20:16	1.183
+++ parser3/src/main/pa_string.C	2003/12/10 14:17:45	1.192
@@ -5,7 +5,7 @@
 	Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
 */
 
-static const char* IDENT_STRING_C="$Date: 2003/09/30 05:20:16 $";
+static const char * const IDENT_STRING_C="$Date: 2003/12/10 14:17:45 $";
 
 #include "pcre.h"
 
@@ -15,6 +15,8 @@ static const char* IDENT_STRING_C="$Date
 #include "pa_dictionary.h"
 #include "pa_charset.h"
 
+const String String::Empty;
+
 // cord lib extension
 
 #ifndef DOXYGEN
@@ -41,7 +43,7 @@ int CORD_range_contains_chr_greater_then
     return(CORD_block_iter(x, i, CORD_range_contains_chr_greater_then_proc, &d) == 1/*alternatives: 0 normally ended, 2=struck 'n'*/);
 }
 
-static int CORD_block_count_proc(char c, size_t size, void* client_data)
+static int CORD_block_count_proc(char /*c*/, size_t /*size*/, void* client_data)
 {
     int* result=(int*)client_data;
     (*result)++;
@@ -192,7 +194,7 @@ void String::split(ArrayString& result,
 		   Language lang, int limit) const {
 	size_t self_length=length();
 	if(size_t delim_length=strlen(delim)) {
-		int pos_before;
+		size_t pos_before;
 		// while we have 'delim'...
 		for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) {
 			result+=&mid(pos_after, pos_before);
@@ -214,7 +216,7 @@ void String::split(ArrayString& result,
 		   const String& delim, Language lang, 
 		   int limit) const {
 	if(!delim.is_empty()) {
-		int pos_before;
+		size_t pos_before;
 		// while we have 'delim'...
 		for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) {
 			result+=&mid(pos_after, pos_before);
@@ -239,13 +241,13 @@ static void regex_options(const String*
 		int *result;
 		bool *flag;
     } regex_option[]={
-		{"i", "I", 0, PCRE_CASELESS, result}, // a=A
-		{"s", "S", 0, PCRE_DOTALL, result}, // \n\n$ [default]
-		{"x", "U", 0, PCRE_EXTENDED, result}, // whitespace in regex ignored
-		{"m", "M", PCRE_DOTALL, PCRE_MULTILINE, result}, // ^aaa\n$^bbb\n$
-		{"g", "G", 0, true, result+1}, // many rows
+		{"i", "I", 0, PCRE_CASELESS, result, 0}, // a=A
+		{"s", "S", 0, PCRE_DOTALL, result, 0}, // \n\n$ [default]
+		{"x", "U", 0, PCRE_EXTENDED, result, 0}, // whitespace in regex ignored
+		{"m", "M", PCRE_DOTALL, PCRE_MULTILINE, result, 0}, // ^aaa\n$^bbb\n$
+		{"g", "G", 0, 1, result+1, 0}, // many rows
 		{"'", 0, 0, 0, 0, &need_pre_post_match},
-		{0}
+		{0, 0, 0, 0, 0, 0}
     };
 	result[0]=PCRE_EXTRA | PCRE_DOTALL | PCRE_DOLLAR_ENDONLY;
 	result[1]=0;
@@ -341,9 +343,9 @@ Table* String::match(Charset& source_cha
 			*row+=&mid(prefinish, poststart); // .match
 			*row+=&mid(poststart, postfinish); // .postmatch
 		} else {
-			*row+=0; // .prematch column value
-			*row+=0; // .match
-			*row+=0; // .postmatch
+			*row+=&Empty; // .prematch column value
+			*row+=&Empty; // .match
+			*row+=&Empty; // .postmatch
 		}
 		
 		for(int i=1; i<exec_substrings; i++) {
@@ -373,14 +375,14 @@ String& String::change_case(Charset& sou
 		return result;
 
 	char* new_cstr=cstrm();
-	char *dest=new_cstr;
+	size_t new_cstr_len=length();
 	if(source_charset.isUTF8()) {
 		switch(kind) {
 		case CC_UPPER:
-			change_case_UTF8((const XMLByte*)new_cstr, (XMLByte*)new_cstr, UTF8CaseToUpper);
+			change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToUpper);
 			break;
 		case CC_LOWER:
-			change_case_UTF8((const XMLByte*)new_cstr, (XMLByte*)new_cstr, UTF8CaseToLower);
+			change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToLower);
 			break;
 		default:
 			assert(!"unknown change case kind");
@@ -407,8 +409,9 @@ String& String::change_case(Charset& sou
 			break; // never
 		}	
 
+		char *dest=new_cstr;
 		unsigned char index;
-		for(const char* current=new_cstr; index=(unsigned char)*current; current++) {
+		for(const char* current=new_cstr; (index=(unsigned char)*current); current++) {
 			unsigned char c=a[index];
 			if(b)
 				c=b[c];
@@ -429,7 +432,7 @@ const String& String::replace(const Dict
 
 	const char* current=old_cstr;
 	while(*current) {
-		if(Table::element_type row=dict.first_that_begins(current)) {
+		if(Dictionary::Subst subst=dict.first_that_begins(current)) {
 			// prematch
 			if(size_t prematch_length=current-prematch_begin) {
 				result.langs.append(result.body, langs, prematch_begin-old_cstr, prematch_length);
@@ -437,15 +440,11 @@ const String& String::replace(const Dict
 			}
 
 			// match
-
-			const String* a=row->get(0);
 			// skip 'a' in 'current'; move prematch_begin
-			current+=a->length(); prematch_begin=current;
+			current+=subst.from_length; prematch_begin=current;
 
-			if(row->count()>1) { // are there any b?
-				const String* b=row->get(1);
+			if(const String* b=subst.to) // are there any b?
 				result<<*b;
-			}
 		} else // simply advance
 			current++; 
 	}
@@ -526,8 +525,8 @@ static int serialize_body_piece(const ch
 };
 static int serialize_lang_piece(char alang, size_t asize, char** cur) {
 	// lang
-	memcpy(*cur, &alang, sizeof(alang));  *cur+=sizeof(alang);
-	// length
+	**cur=alang; (*cur)++;
+	// length [WARNING: not cast, addresses must be %4=0 on sparc]
 	memcpy(*cur, &asize, sizeof(asize));  *cur+=sizeof(asize);
 
 	return 0; // 0=continue
@@ -544,7 +543,7 @@ String::Cm String::serialize(size_t prol
 
 	// 1: prolog
 	char *cur=result.str+prolog_length;
-	// 2: langs.count
+	// 2: langs.count [WARNING: not cast, addresses must be %4=0 on sparc]
 	memcpy(cur, &fragments_count, sizeof(fragments_count));  cur+=sizeof(fragments_count);
 	// 3: lang info
 	langs.for_each(body, serialize_lang_piece, &cur);
@@ -565,22 +564,30 @@ bool String::deserialize(size_t prolog_l
 	const char* cur=(const char* )buf+prolog_length;
 
 	// 2: langs.count
-	if(buf_length<sizeof(size_t)) // langs.count don't fit?
+	size_t fragments_count;
+	if(buf_length<sizeof(fragments_count)) // langs.count don't fit?
 		return false;
-	size_t fragments_count=*reinterpret_cast<const size_t*>(cur);  cur+=sizeof(size_t);
-	buf_length-=sizeof(size_t);
+	// [WARNING: not cast, addresses must be %4=0 on sparc]
+	memcpy(&fragments_count, cur, sizeof(fragments_count));  cur+=sizeof(fragments_count);
+	buf_length-=sizeof(fragments_count);
 	
 	if(fragments_count) {
 		// 3: lang info
 		size_t total_length=0;
 		for(size_t f=0; f<fragments_count; f++) {
-			size_t piece_length=sizeof(char)+sizeof(size_t);
+			char lang;
+			size_t fragment_length;
+			size_t piece_length=sizeof(lang)+sizeof(fragment_length);
 			if(buf_length<piece_length) // lang+length
 				return false;
 
-			Language lang=*reinterpret_cast<const Language *>(cur);  cur+=sizeof(char);
-			size_t fragment_length=*reinterpret_cast<const size_t*>(cur);  cur+=sizeof(size_t);
-			langs.append(total_length, lang, fragment_length);
+			// lang
+			lang=*cur++;
+			// length [WARNING: not cast, addresses must be %4=0 on sparc]
+			memcpy(&fragment_length, cur, sizeof(fragment_length));  cur+=sizeof(fragment_length);
+
+			// uchar needed to prevent propagating 0x80 bit to upper bytes
+			langs.append(total_length, (String::Language)(uchar)lang, fragment_length);
 			total_length+=fragment_length;
 
 			buf_length-=piece_length;