Return to pa_string.h CVS log

Up to [parser3project] / parser3 / src / include

vrequest,vdouble,vint compiled

/** @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: 2003/01/29 11:06:43 $";

// includes

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

// forwards

class String; DECLARE_OBJECT_PTR(String);
typedef object_ptr<const String> ConstStringPtr; const ConstStringPtr ConstStringPtrZero(0);

// forwards

class Table;
typedef object_ptr<Table> TablePtr;
class SQL_Connection;
class Dictionary;
struct Request_charsets;

// helpers

#ifndef NO_STRING_ORIGIN
#	define STRING_APPEND_PARAMS \
		const char *src, size_t size,  \
		String_UL lang, \
		const char *file, uint line
/// appends piece to String  @see String::real_append
#	define APPEND(src, size, lang, file, line) \
		real_append(src, size, lang, file, line)
#else
#	define STRING_APPEND_PARAMS \
		const char *src, \
		size_t size, \
		String_UL lang
/// appends piece to String  @see String::real_append
#	define APPEND(src, size, lang, file, line) \
		real_append(src, size, lang)
#endif
/// appends clean piece to String @see String::real_append
#define APPEND_CLEAN(src, size, file, line) \
	APPEND(src, size, String::UL_CLEAN, file, line)
/// appends piece to String as-is @see String::real_append
#define APPEND_AS_IS(src, size, file, line) \
	APPEND(src, size, String::UL_AS_IS, file, line)
/// appends tainted piece to String  @see String::real_append
#define APPEND_TAINTED(src, size, file, line) \
	APPEND(src, size, String::UL_TAINTED, file, line)
/// handy: appends const char* piece to String  @see String::real_append
#define	APPEND_CONST(src) APPEND_AS_IS(src, 0, 0, 0)

typedef uint String_UL;

struct String_fragment { 
	String_UL/*Untaint_lang*/ lang; ///< untaint flag, later untaint language
	const char *ptr;  ///< pointer to the start
	size_t size;  ///< length

#ifndef NO_STRING_ORIGIN
	/// all String pieces hold information of where they come from
	struct Origin {
		const char *file;  ///< macros file name | load file name | sql query text
		uint line; ///< file line no | record no
	} origin;  ///< origin
#endif
};

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

	All pieces remember 
	- the file and its line they are from [can be turned off by NO_STRING_ORIGIN]
	- whether they are tainted or not, 
	  and the language which should be used to detaint them
*/

class String: public Array<String_fragment> {
	size_t fsize;
	int count() const; // hiding fro accidental use instead of size()
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
	};

public:

	explicit String(const char *src=0, size_t src_size=0, bool tainted=false);
	String(const String& src);
	bool is_empty() const { return !size(); }
	size_t size() const { return fsize; }
	/// convert to C string. if 'lang' known, forcing 'lang' to it
	CharPtr cstr(Untaint_lang lang=UL_AS_IS, 
		SQL_Connection *connection=0,
		const Request_charsets *charsets=0) const {

		CharPtr result(new char[cstr_bufsize(lang, connection, charsets)]);
		char *eol=store_to(result.get(), lang, connection, charsets);
		*eol=0;
		return result;
	}
	char *cstr(Pool& pool, Untaint_lang lang=UL_AS_IS, 
		SQL_Connection *connection=0,
		const Request_charsets *charsets=0) const {

		char *result=new(pool) char[cstr_bufsize(lang, connection, charsets)];
		char *eol=store_to(result, lang, connection, charsets);
		*eol=0;
		return result;
	}
	char *cstr_debug_origins(Pool& pool) const;
	/// puts pieces to buf
	void serialize(Pool& pool, size_t prolog_size,  char *& buf, size_t& buf_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
		@see APPEND_AS_IS, APPEND_CLEAN, APPEND_TAINTED, APPEND_CONST
	*/
	String& real_append(STRING_APPEND_PARAMS);
	/// @return <0 ==0 or >0 depending on comparison result
	int cmp (int& partial, const String& src, 
		size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
	bool operator < (const String& src) const {	int p; return cmp(p, src)<0; }
	bool operator > (const String& src) const {	int p; return cmp(p, src)>0; }
	bool operator <= (const String& src) const { int p; return cmp(p, src)<=0; }
	bool operator >= (const String& src) const { int p; return cmp(p, src)>=0; }
	bool operator == (const String& src) const {
		if(size()!=src.size()) // can speed up in trivial case
			return false;
		int p; return cmp(p, src)==0;
	}
	bool operator != (const String& src) const { int p; return cmp(p, src)!=0; }

	/**
		 @param partial 
			returns partial match status. 
			- -1: strings too different
			-  0: full match
			-  1: means @c this starts @c src
			-  2: means @src starts @this
	*/
	int cmp(int& partial, const char* src_ptr, size_t src_size=0, 
		size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
	/// this starts with src
	bool starts_with(const char* src_ptr, size_t src_size=0) const {
		int p; cmp(p, src_ptr, src_size);
		return p==0 || p==2;
	}
	bool operator == (const char* src_ptr) const { 
		size_t src_size=src_ptr?strlen(src_ptr):0;
		if(size() != src_size)
			return false;
		int partial; // unused
		return cmp(partial, src_ptr, src_size)==0; 
	}
	bool operator != (const char* src_ptr) const { 
		int partial; // unused
		return cmp(partial, src_ptr, 0)!=0; 
	}

	String& append_to(String& dest, uchar lang, bool forced) const;
	String& append(const String& src, uchar 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 StringPtr src) { return append(*src, UL_PASS_APPENDED); }
	String& operator << (const char *src) { return APPEND_CONST(src); }

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

	/// extracts first char of a string, if any
	char first_char() const {
		return is_empty()?0:*felements[0].ptr;
	}

	/// extracts [start, finish) piece of string
	StringPtr mid(size_t start, size_t finish) const;

	/// @return position of substr in string, -1 means "not found" [String version]
	int pos(const String& substr, 
		int this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
	/// @return position of substr in string, -1 means "not found" [const char* version]
	int pos(const char *substr, size_t substr_size=0, 
		int this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;

	void split(Array<ConstStringPtr>& result, 
		size_t *pos_after_ref, 
		const char *delim, size_t delim_size, 
		Untaint_lang lang=UL_UNSPECIFIED, int limit=-1) const;
	void split(Array<ConstStringPtr>& result, 
		size_t *pos_after_ref, 
		const String& delim, 
		Untaint_lang lang=UL_UNSPECIFIED, int limit=-1) const;

	typedef void (*Row_action)(Table& table, Array<ConstStringPtr> *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
	*/
	TablePtr match(Charset& source_charset,
		ConstStringPtr aorigin,		
		const String& regexp, 
		ConstStringPtr options,
		Row_action row_action, void *info,
		bool *was_global=0) const;
	enum Change_case_kind {
		CC_UPPER,
		CC_LOWER
	};
	StringPtr change_case(Pool& pool, Charset& source_charset,
		Change_case_kind kind) const;
	StringPtr replace(Pool& pool, Dictionary& dict) const;
	double as_double() const;
	int as_int() const;

	StringPtr join_chains(Pool& pool) const;

#ifndef NO_STRING_ORIGIN
	/// origin of string. calculated by first row
	const String_fragment::Origin& origin() const;
#endif

	size_t cstr_bufsize(Untaint_lang lang=UL_UNSPECIFIED,
		SQL_Connection *connection=0,
		const Request_charsets *charsets=0) const;
	/// convert to C string, store to 'dest' which must be big enough for proper untaint
	char *store_to(char *dest, Untaint_lang lang=UL_UNSPECIFIED, 
		SQL_Connection *connection=0,
		const Request_charsets *charsets=0) const;
/*
	void join_chain(
					   const Chunk*& achunk, const Chunk::Row*& arow, uint& acountdown, 
					   uchar& joined_lang, const char *& joined_ptr, size_t& joined_size) 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();
}
/// simple hash code of string. used by Hash
inline uint hash_code(ConstStringPtr self) {
	return self->hash_code();
}

// helpers

#define STRING_PREPARED_FOREACH_FRAGMENT(code) \
	for(; fragment<last; fragment++) { \
		code; \
	}

#define STRING_FOREACH_FRAGMENT(code) \
	{ \
		element_type *last = felements+fused; \
		element_type *fragment = felements; \
		STRING_PREPARED_FOREACH_FRAGMENT(code) \
	}

#endif