--- parser3/src/include/pa_string.h	2003/03/24 13:16:12	1.144.2.28.2.27
+++ parser3/src/include/pa_string.h	2003/03/27 14:51:27	1.144.2.28.2.41
@@ -8,7 +8,7 @@
 #ifndef PA_STRING_H
 #define PA_STRING_H
 
-static const char* IDENT_STRING_H="$Date: 2003/03/24 13:16:12 $";
+static const char* IDENT_STRING_H="$Date: 2003/03/27 14:51:27 $";
 
 // includes
 
@@ -16,6 +16,7 @@ static const char* IDENT_STRING_H="$Date
 #include "pa_array.h"
 
 extern "C" { // cord's author forgot to do that
+#define CORD_NO_IO
 #include "cord.h"
 };
 
@@ -29,17 +30,6 @@ class Request_charsets;
 class String;
 typedef Array<const String*> ArrayString;
 
-// helpers
-/*
-/// appends clean piece to String @see String::append
-#define APPEND_CLEAN(src, length) append(src, length, String::L_CLEAN)
-/// appends piece to String as-is @see String::append
-#define APPEND_AS_IS(src, length) append(src, length, String::L_AS_IS)
-/// appends tainted piece to String  @see String::append
-#define APPEND_TAINTED(src, length) append(src, length, String::L_TAINTED)
-/// handy: appends const char* piece to String  @see String::append
-#define	APPEND_CONST(src) APPEND_AS_IS(src, 0)
-*/
 /// this is result of pos functions which mean that substr were not found
 #define STRING_NOT_FOUND ((size_t)-1)
 
@@ -50,10 +40,20 @@ class StringBody {
 public:
 
 	StringBody(): body(CORD_EMPTY) {}
-	StringBody(CORD abody): body(abody) {}
-	explicit StringBody(const char* str, size_t length): body(CORD_EMPTY) {
-		append(str, length);
+	StringBody(CORD abody): body(abody) {
+#ifdef _DEBUG
+		if(body && !*body)
+			switch(*(char*)(body+1)) {
+			case 1:	case 4: case 6: break;
+			default: _asm int 3; break;
+			}
+#endif
+	}
+	/// WARNING: length is only HELPER length, str in ANY case should be zero-terminated
+	StringBody(const char* str, size_t helper_length): body(CORD_EMPTY) {
+		append_help_length(str, helper_length);
 	}
+	static StringBody Format(int value);
 
 	void clear() { body=CORD_EMPTY; }
 
@@ -68,19 +68,32 @@ public:
 
 	bool is_empty() const { return body==CORD_EMPTY; }
 
-	void append(const char *str, size_t length) {
-		body=CORD_cat_char_star(body, str, length);
+	void append_help_length(const char *str, size_t help_length) {
+#ifdef _DEBUG
+		if(!*str)
+			_asm int 3;
+		bool were_empty=cstr==CORD_EMPTY;
+#endif
+		body=CORD_cat_char_star(body, str, help_length);
+#ifdef _DEBUG
+		if(were_empty && !*str)
+			_asm int 3;
+#endif
+	}
+	String& append_strdup(const char* str, size_t length) {
+		append_help_length(pa_strdup(str, length), length);
 	}
 	void append(char c) { body=CORD_cat_char(body, c); }
-	void append(StringBody src) { body=CORD_cat(body, src.body); }
+	StringBody& operator << (const StringBody src) { body=CORD_cat(body, src.body); return *this; }
+	StringBody& operator << (const char* str) { body=CORD_cat_char_star(body, str, 0); return *this; }
 
 	// could not figure out why this operator is needed [should do this chain: string->simple->==]
-	bool operator < (const StringBody& src) const { return CORD_cmp(body, src.body)<0; }
-	bool operator > (const StringBody& src) const { return CORD_cmp(body, src.body)>0; }
-	bool operator <= (const StringBody& src) const { return CORD_cmp(body, src.body)<=0; }
-	bool operator >= (const StringBody& src) const { return CORD_cmp(body, src.body)>=0; }
-	bool operator != (const StringBody& src) const { return CORD_cmp(body, src.body)!=0; }
-	bool operator == (const StringBody& src) const { return CORD_cmp(body, src.body)==0; }
+	bool operator < (const StringBody src) const { return CORD_cmp(body, src.body)<0; }
+	bool operator > (const StringBody src) const { return CORD_cmp(body, src.body)>0; }
+	bool operator <= (const StringBody src) const { return CORD_cmp(body, src.body)<=0; }
+	bool operator >= (const StringBody src) const { return CORD_cmp(body, src.body)>=0; }
+	bool operator != (const StringBody src) const { return CORD_cmp(body, src.body)!=0; }
+	bool operator == (const StringBody src) const { return CORD_cmp(body, src.body)==0; }
 
 	int ncmp(size_t x_begin, const StringBody y, size_t y_begin, size_t size) const {
 		return CORD_ncmp(body, x_begin, y.body, y_begin, size);
@@ -89,7 +102,15 @@ public:
 	char fetch(size_t index) const { return CORD_fetch(body, index); }
 	StringBody mid(size_t index, size_t length) const { return CORD_substr(body, index, length); }
 	size_t pos(const char* substr, size_t offset=0) const { return CORD_str(body, offset, substr); }
-	size_t pos(const StringBody substr, size_t offset=0) const { return CORD_str(body, offset, substr.body); }
+	size_t pos(const StringBody substr, size_t offset=0) const { 
+		if(!substr.length())
+			return STRING_NOT_FOUND; // in this case CORD_str returns 0 [parser users got used to -1]
+		return CORD_str(body, offset, substr.body); 
+	}
+	size_t pos(char c, 
+		size_t offset=0) const {
+		return CORD_chr(body, offset, c);
+	}
 
 	template<typename I> void for_each(int (*callback)(const char* s, I), I info) const {
 		CORD_iter5(body, 0, 0, (CORD_batched_iter_fn)callback, info);
@@ -143,7 +164,14 @@ public:
 	struct Fragment { 
 		Language lang; ///< untaint flag, later untaint language
 		size_t length;  ///< length
-		Fragment(Language alang, size_t asize): lang(alang), length(asize) {}
+		Fragment(Language alang, size_t asize): lang(alang), length(asize) {
+#ifdef _DEBUG
+			if(alang==L_UNSPECIFIED)
+				_asm int 3; // MUST NOT
+			if(asize==(size_t)-1)
+				_asm int 3; // MUST NOT
+#endif
+		}
 	};
 
 	class ArrayFragment: public Array<Fragment> {
@@ -156,9 +184,9 @@ public:
 		}
 	public:
 		ArrayFragment& operator += (element_type src) {
-			if(count()) { // not empty?
+			if(size_t lcount=count()) { // not empty?
 				// try to join with last
-				Fragment& last=get(count()-1);
+				Fragment& last=unchecked_get_ref(lcount-1);
 				if(last.lang==src.lang) {
 					last.length+=src.length;
 					return *this;
@@ -185,8 +213,7 @@ private:
 
 public:
 
-	/// @todo check all places for length to be only HELPER length, and never be !=strlen(cstr)
-	explicit String(const char* cstr=0, size_t helper_length=0, bool tainted=false);
+	explicit String(const char* cstr=0, size_t help_length=0, bool tainted=false);
 	String(const String& src);
 	String(StringBody abody, Language alang): body(abody) {
 		fragments+=Fragment(alang, abody.length());
@@ -219,11 +246,9 @@ public:
 	C serialize(size_t prolog_size) const;
 	/// appends pieces from buf to self
 	bool deserialize(size_t prolog_size, void *buf, size_t buf_size);
-	/** append fragment, 
-		@c length can be zero, it will be autocalced
-		when length<strlen(str) makes heap copy of @c str piece
-	*/
-	String& append(const char* str, size_t length, Language lang);
+	/// @see StringBody::append_help_length
+	String& append_help_length(const char* str, size_t helper_length, Language lang);
+	String& append_strdup(const char* str, size_t length, Language lang);
 
 	bool operator == (const char* y) const { return body==StringBody(y); }
 	bool operator != (const char* y) const { return body!=StringBody(y); }
@@ -232,9 +257,9 @@ public:
 	bool starts_with(const char* y) const {
 		return body.ncmp(0/*x_begin*/, StringBody(y), 0/*y_begin*/, strlen(y))==0;
 	}
-	/// y starts with this
-	bool this_starts(const char* y) const {
-		return body.ncmp(0/*x_begin*/, StringBody(y), 0/*y_begin*/, strlen(y))==0;
+	/// x starts with this
+	bool this_starts(const char* x) const {
+		return StringBody(x).ncmp(0/*x_begin*/, body, 0/*y_begin*/, length())==0;
 	}
 
 	String& append_to(String& dest, Language lang, bool forced) const;
@@ -242,9 +267,9 @@ public:
 		return src.append_to(*this, lang, forced);
 	}
 	String& operator << (const String& src) { return append(src, L_PASS_APPENDED); }
-	String& operator << (const char* src) { return append(src, 0, L_AS_IS); }
-	String& operator << (const StringBody& src) { 
-		body.append(src);
+	String& operator << (const char* src) { return append_help_length(src, 0, L_AS_IS); }
+	String& operator << (const StringBody src) { 
+		body<<src;
 		fragments+=Fragment(L_AS_IS, src.length());
 		return *this;
 	}
@@ -274,6 +299,10 @@ public:
 	/// String version of @see pos(const char*, int, Language)
 	size_t pos(const String& substr, 
 		size_t this_offset=0, Language lang=L_UNSPECIFIED) const;
+	size_t pos(char c, 
+		size_t this_offset=0) const {
+		return body.pos(c, this_offset);
+	}
 
 	void split(ArrayString& result, 
 		size_t& pos_after,
@@ -317,7 +346,7 @@ private: //disabled
 };
 
 /// simple hash code of string. used by Hash
-inline uint hash_code(const StringBody& self) {
+inline uint hash_code(const StringBody self) {
 	return self.hash_code();
 }