Return to pa_string.h CVS log

Up to [parser3project] / parser3 / src / include

warning

/** @file
	Parser: string class decl.

	Copyright (c) 2001-2003 ArtLebedev Group (http://www.artlebedev.com)
	Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
*/

#ifndef PA_STRING_H
#define PA_STRING_H

static const char* IDENT_STRING_H="$Date: 2003/03/20 14:45:16 $";

// includes

#include "pa_types.h"
#include "pa_array.h"

extern "C" { // cord's author forgot to do that
#include "cord.h"
};

// forwards

class Charset;
class Table;
class SQL_Connection;
class Dictionary;
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::UL_CLEAN)
/// appends piece to String as-is @see String::append
#define APPEND_AS_IS(src, length) append(src, length, String::UL_AS_IS)
/// appends tainted piece to String  @see String::append
#define APPEND_TAINTED(src, length) append(src, length, String::UL_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)


/** 
	String which knows the language of all it's fragments.

	All pieces remember 
	- whether they are tainted or not, 
	  and the language which should be used to detaint them

	GENERAL ASSUMPTION: they append only cstr's to body, not PART of cstr's

	HENCE WARNING: DO NOT PASS HELPER_LENGTH TO REDUCE SIZE OF CONSTRUCTURED STRING
	USE append_copy INSTEAD
*/
class String: public PA_Object {

public:

	/** piece is tainted or not. the language to use when detaint
		remember to change String_Untaint_lang_name @ untaint.C along
	*/
	enum Untaint_lang {
		UL_UNSPECIFIED=0, ///< zero value handy for hash lookup @see untaint_lang_name2enum
		// these two must go before others, there are checks for >UL_AS_IS
		UL_CLEAN, ///< clean
		UL_AS_IS,     ///< leave all characters intact

		UL_PASS_APPENDED,
			/**<
				leave language built into string being appended.
				just a flag, that value not stored
			*/
		UL_TAINTED,  ///< tainted, untaint language as assigned later 
		// untaint languages. assigned by ^untaint[lang]{...}
		UL_FILE_SPEC, ///< file specification
		UL_HTTP_HEADER,    ///< text in HTTP response header
		UL_MAIL_HEADER,    ///< text in mail header
		UL_URI,       ///< text in uri
		UL_TABLE,     ///< ^table:set body
		UL_SQL,       ///< ^table:sql body
		UL_JS,        ///< JavaScript code
		UL_XML,		///< ^dom:set xml
		UL_HTML,      ///< HTML code (for editing)
		UL_OPTIMIZE_BIT = 0x8000  ///< flag, requiring cstr whitespace optimization
	};

	struct Fragment { 
		Untaint_lang lang; ///< untaint flag, later untaint language
		size_t length;  ///< length
		Fragment(Untaint_lang alang, size_t asize): lang(alang), length(asize) {}
	};

	class ArrayFragment: public Array<Fragment> {
		void simple_append(element_type src) {
			*static_cast<Array<Fragment> *>(this)+=src;
		}
	public:
		ArrayFragment& operator += (element_type src) {
			if(count()) { // not empty?
				// try to join with last
				Fragment& last=get(count()-1);
				if(last.lang==src.lang) {
					last.length+=src.length;
					return *this;
				}
			}
			simple_append(src);
			return *this;
		}
		ArrayFragment& append(const ArrayFragment& src, size_t substr_start, size_t substr_end);
	};

	struct C {
		char *str;
		size_t length;
		operator char *() { return str; }
		C(char *astr, size_t asize): str(astr), length(asize) {}
	};

private:

	CORD body; ///< all characters of string
	ArrayFragment fragments; ///< fragment language+length info

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);
	String(const String& src);
	operator bool() const { return body!=CORD_EMPTY; }
	size_t length() const { return CORD_len(body); }

	/// convert to CORD. if 'lang' known, forcing 'lang' to it
	CORD cstr_to_cord(Untaint_lang lang=UL_AS_IS, 
		SQL_Connection *connection=0,
		const Request_charsets *charsets=0) const;

	/// convert to constant C string. if 'lang' known, forcing 'lang' to it
	const char* cstr(Untaint_lang lang=UL_AS_IS, 
		SQL_Connection *connection=0,
		const Request_charsets *charsets=0) const {
		return CORD_to_const_char_star(cstr_to_cord(lang, connection, charsets));
	}
	/// convert to Modifiable C string. if 'lang' known, forcing 'lang' to it
	char *cstrm(Untaint_lang lang=UL_AS_IS, 
		SQL_Connection *connection=0,
		const Request_charsets *charsets=0) const {
		return CORD_to_char_star(cstr_to_cord(lang, connection, charsets));
	}
	/// puts pieces to buf
	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, const char* file);
	/** append fragment,
		@c heap_variable considered as cstr, that is zeroterminated
	*/
	String& append_cstr(const char* heap_variable, size_t helper_length, Untaint_lang lang);
	/** append fragment, making heap copy of @c auto_variable
		@c auto_variable considered MAY BE NOT zeroterminated
		if @c length=0 @c auto_variable considered zeroterminated
	*/
	String& append_copy(const char* auto_variable, size_t helper_length, Untaint_lang lang);

	bool operator < (const String& src) const { return CORD_cmp(body, src.body)<0; }
	bool operator > (const String& src) const { return CORD_cmp(body, src.body)>0; }
	bool operator <= (const String& src) const { return CORD_cmp(body, src.body)<=0; }
	bool operator >= (const String& src) const { return CORD_cmp(body, src.body)>=0; }
	bool operator == (const String& src) const { return CORD_cmp(body, src.body)==0; }
	bool operator != (const String& src) const { return CORD_cmp(body, src.body)!=0; }

	/// this starts with y
	bool starts_with(CORD y) const {
		return CORD_ncmp(body, 0/*x_start*/, y, 0/*y_start*/, CORD_len(y))==0;
	}
	/// y starts with this
	bool this_starts(CORD y) const {
		return CORD_ncmp(body, 0/*x_start*/, y, 0/*y_start*/, CORD_len(body))==0;
	}
	bool operator == (CORD y) const { return CORD_cmp(body, y)==0; }
	bool operator != (CORD y) const { return CORD_cmp(body, y)!=0; }

	String& append_to(String& dest, Untaint_lang lang, bool forced) const;
	String& append(const String& src, Untaint_lang lang, bool forced=false) { 
		return src.append_to(*this, lang, forced);
	}
	String& operator << (const String& src) { return append(src, UL_PASS_APPENDED); }
	String& operator << (const char* src) { return append_cstr(src, 0, UL_AS_IS); }

	/// simple hash code of string
	uint hash_code() const;

	/// extracts first char of a string, if any
	char first_char() const {
		return operator bool()?CORD_fetch(body, 0):0;
	}

	/// extracts [start, finish) piece of string
	const String& mid(size_t substr_start, size_t substr_end) const;

	/** 
		ignore lang if it's UL_UNSPECIFIED
		but when specified: look for substring that lies in ONE fragment in THAT lang
		@return position of substr in string, -1 means "not found" [const char* version]
	*/
	size_t pos(CORD substr, 
		size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
	/// String version of @see pos(const char*, int, Untaint_lang)
	size_t pos(const String& substr, 
		size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;

	void split(ArrayString& result, 
		size_t& pos_after_ref, 
		const char* delim, 
		Untaint_lang lang=UL_UNSPECIFIED, int limit=-1);
	void split(ArrayString& result, 
		size_t& pos_after_ref, 
		const String& delim, 
		Untaint_lang lang=UL_UNSPECIFIED, int limit=-1) const;

	typedef void (*Row_action)(Table* table, ArrayString* row, 
		int prestart, int prefinish, 
		int poststart, int postfinish,
		void *info);
	/**
		@return table of found items, if any.
		table format is defined and fixed[can be used by others]: 
		@verbatim
			prematch/match/postmatch/1/2/3/...
		@endverbatim
	*/
	Table* match(Charset& source_charset,
		const String& regexp, 
		const String& options,
		Row_action row_action, void *info,
		bool& just_matched) const;
	enum Change_case_kind {
		CC_UPPER,
		CC_LOWER
	};
	String& change_case(Charset& source_charset,
		Change_case_kind kind);
	const String& replace(const Dictionary& dict) const;
	double as_double() const;
	int as_int() const;

private: //disabled

	String& operator = (const String&) { return *this; }

};

/// simple hash code of string. used by Hash
inline uint hash_code(const String& self) {
	return self.hash_code();
}

#endif