parser3/src/include/pa_string.h - diff

Return to pa_string.h CVS log

Up to [parser3project] / parser3 / src / include

Diff for /parser3/src/include/pa_string.h between versions 1.117 and 1.134

version 1.117, 2001/11/16 13:51:14	version 1.134, 2002/04/04 13:42:50
Line 1	Line 1
/** @file	/** @file
Parser: string class decl.	Parser: string class decl.

Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com)	Copyright (c) 2001, 2002 ArtLebedev Group (http://www.artlebedev.com)
Author: Alexander Petrosyan <paf@design.ru> (http://paf.design.ru)	Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)

$Id$	$Id$
*/	*/
Line 10	Line 10
#ifndef PA_STRING_H	#ifndef PA_STRING_H
#define PA_STRING_H	#define PA_STRING_H

#include "pa_config_includes.h"
#include "pa_pool.h"	#include "pa_pool.h"
#include "pa_types.h"	#include "pa_types.h"

Line 19 class Table;	Line 18 class Table;
#ifndef NO_STRING_ORIGIN	#ifndef NO_STRING_ORIGIN
# define STRING_APPEND_PARAMS \	# define STRING_APPEND_PARAMS \
const char *src, size_t size, \	const char *src, size_t size, \
String::Untaint_lang lang, \	uchar lang, \
const char *file, uint line	const char *file, uint line
/// appends piece to String @see String::real_append	/// appends piece to String @see String::real_append
# define APPEND(src, size, lang, file, line) \	# define APPEND(src, size, lang, file, line) \
Line 28 class Table;	Line 27 class Table;
# define STRING_APPEND_PARAMS \	# define STRING_APPEND_PARAMS \
const char *src, \	const char *src, \
size_t size, \	size_t size, \
String::Untaint_lang lang	uchar lang
/// appends piece to String @see String::real_append	/// appends piece to String @see String::real_append
# define APPEND(src, size, lang, file, line) \	# define APPEND(src, size, lang, file, line) \
real_append(src, size, lang)	real_append(src, size, lang)
Line 75 public:	Line 74 public:

enum {	enum {
CR_PREALLOCATED_COUNT=2, ///< default preallocated item count	CR_PREALLOCATED_COUNT=2, ///< default preallocated item count
CR_GROW_COUNT=2 ///< each time the String chunk_is_full() string expanded()	CR_GROW_COUNT=1 ///< each time the String chunk_is_full() string expanded()
};	};

/** piece is tainted or not. the language to use when detaint	/** piece is tainted or not. the language to use when detaint
Line 83 public:	Line 82 public:
*/	*/
enum Untaint_lang {	enum Untaint_lang {
UL_UNSPECIFIED=0, ///< zero value handy for hash lookup @see untaint_lang_name2enum	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_CLEAN, ///< clean
UL_TAINTED, ///< tainted, untaint language as assigned later	UL_AS_IS, ///< leave all characters intact
// untaint languages. assigned by ^untaint[lang]{...}
UL_PASS_APPENDED,	UL_PASS_APPENDED,
/**<	/**<
leave language built into string being appended.	leave language built into string being appended.
just a flag, that value not stored	just a flag, that value not stored
*/	*/
UL_AS_IS, ///< leave all characters intact	UL_TAINTED, ///< tainted, untaint language as assigned later
	// untaint languages. assigned by ^untaint[lang]{...}
UL_FILE_SPEC, ///< file specification	UL_FILE_SPEC, ///< file specification
UL_HTTP_HEADER, ///< text in HTTP response header	UL_HTTP_HEADER, ///< text in HTTP response header
UL_MAIL_HEADER, ///< text in mail header	UL_MAIL_HEADER, ///< text in mail header
Line 101 public:	Line 102 public:
UL_JS, ///< JavaScript code	UL_JS, ///< JavaScript code
UL_XML, ///< ^dom:set xml	UL_XML, ///< ^dom:set xml
UL_HTML, ///< HTML code (for editing)	UL_HTML, ///< HTML code (for editing)
UL_OPTIMIZED_HTML ///< optimized HTML code, adjucent whitespaces joined	UL_OPTIMIZE_BIT = 0x80 ///< flag, requiring cstr whitespace optimization
};	};

public:	public:

String(Pool& apool, const char *src=0, size_t src_size=0, bool tainted=false);	String(Pool& apool, const char *src=0, size_t src_size=0, bool tainted=false);
String(const String& src);	String(const String& src);
	bool is_empty() const { return append_here==head.rows; }
size_t size() const;	size_t size() const;
/// convert to C string. if 'lang' known, forcing 'lang' to it	/// convert to C string. if 'lang' known, forcing 'lang' to it
char *cstr(Untaint_lang lang=UL_AS_IS,	char *cstr(Untaint_lang lang=UL_AS_IS,
SQL_Connection *connection=0,	SQL_Connection *connection=0,
const char *charset=0) const {	Charset cstr_charset=0, const char cstr_charset_name=0) const {

char result=(char )malloc(cstr_bufsize(lang, connection, charset));	char result=(char )malloc(cstr_bufsize(lang, connection, cstr_charset));
char *eol=store_to(result, lang, connection, charset);	char *eol=store_to(result, lang, connection, cstr_charset, cstr_charset_name);
*eol=0;	*eol=0;
return result;	return result;
}	}
Line 123 public:	Line 125 public:
/// puts pieces to buf	/// puts pieces to buf
void serialize(size_t prolog_size, void *& buf, size_t& buf_size) const;	void serialize(size_t prolog_size, void *& buf, size_t& buf_size) const;
/// appends pieces from buf to self	/// appends pieces from buf to self
void deserialize(size_t prolog_size, void buf, size_t buf_size, const char file);	bool deserialize(size_t prolog_size, void buf, size_t buf_size, const char file);
/** append fragment	/** append fragment
@see APPEND_AS_IS, APPEND_CLEAN, APPEND_TAINTED, APPEND_CONST	@see APPEND_AS_IS, APPEND_CLEAN, APPEND_TAINTED, APPEND_CONST
*/	*/
Line 164 public:	Line 166 public:
return cmp(partial, src_ptr, 0)!=0;	return cmp(partial, src_ptr, 0)!=0;
}	}

/**	String& append(const String& src, uchar lang, bool forced=false);
appends other String.

marking all tainted pieces of it with @a lang.
or marking ALL pieces of it with a @a lang when @a forced to.
*/
String& append(const String& src, Untaint_lang lang, bool forced=false);
String& operator << (const String& src) { return append(src, UL_PASS_APPENDED); }	String& operator << (const String& src) { return append(src, UL_PASS_APPENDED); }
String& operator << (const char *src) { return APPEND_CONST(src); }	String& operator << (const char *src) { return APPEND_CONST(src); }

Line 208 public:	Line 204 public:
prematch/match/postmatch/1/2/3/...	prematch/match/postmatch/1/2/3/...
@endverbatim	@endverbatim
*/	*/
bool match(const unsigned char *pcre_tables,	bool match(
const String *aorigin,	const String *aorigin,
const String& regexp,	const String& regexp,
const String *options,	const String *options,
Line 219 public:	Line 215 public:
CC_UPPER,	CC_UPPER,
CC_LOWER	CC_LOWER
};	};
String& change_case(Pool& pool, const unsigned char *pcre_tables,	String& change_case(Pool& pool,
Change_case_kind kind) const;	Change_case_kind kind) const;
String& replace(Pool& pool, Dictionary& dict) const;	String& replace(Pool& pool, Dictionary& dict) const;
double as_double() const;	double as_double() const;
Line 232 public:	Line 228 public:

private:	private:

/// several String fragments	/** several String fragments
struct Chunk {
uint count; ///< the number of rows in chunk	'mutable' because can write after it's end, after it was appended to somebody
	@see String::append
	*/
	mutable struct Chunk {
	typedef uchar count_type;
	count_type count; ///< the number of rows in chunk
/// string fragment or a link to next chunk union	/// string fragment or a link to next chunk union
union Row {	union Row {
	typedef uchar item_size_type;
/// fragment	/// fragment
struct {	struct {
const char *ptr; ///< pointer to the start	const char *ptr; ///< pointer to the start
size_t size; ///< length	item_size_type size; ///< length
unsigned char/Untaint_lang/ lang; ///< untaint flag, later untaint language	uchar/Untaint_lang/ lang; ///< untaint flag, later untaint language
#ifndef NO_STRING_ORIGIN	#ifndef NO_STRING_ORIGIN
Origin origin; ///< origin	Origin origin; ///< origin
#endif	#endif
} item;	} item;
Chunk *link; ///< link to the next chunk in chain	Chunk *link; ///< link to the next chunk in chain
} rows[CR_PREALLOCATED_COUNT];	} rows[CR_PREALLOCATED_COUNT+1/for head.rows[CR_PREALLOCATED_COUNT].link/];
Chunk *preallocated_link; ///< next rows are here
}	}
head; ///< the head chunk of the chunk chain	head; ///< the head chunk of the chunk chain

/// next append would write to this record	/// next append would write to this record
Chunk::Row *append_here;	Chunk::Row *append_here;

/** the address of place where lies address
of the link to the next chunk to allocate
*/
Chunk::Row *link_row;

private:	private:
/// last chunk	/// last chunk
Chunk *last_chunk;	mutable Chunk *last_chunk;

private:	private:

bool chunk_is_full() {	bool chunk_is_full() {
return append_here == link_row;	return append_here == last_chunk->rows+last_chunk->count;
}	}
uint used_rows() const;	uint used_rows() const;
void expand();	void expand();

	Untaint_lang lang_of(size_t offset) const;

size_t cstr_bufsize(Untaint_lang lang,	size_t cstr_bufsize(Untaint_lang lang,
SQL_Connection *connection,	SQL_Connection *connection,
const char *charset) const;	Charset *buf_charset) const;
/// convert to C string, store to 'dest' which must be big enough for proper untaint	/// 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,	char store_to(char dest, Untaint_lang lang=UL_UNSPECIFIED,
SQL_Connection *connection=0,	SQL_Connection *connection=0,
const char *charset=0) const;	Charset *store_to_charset=0,
	const char *store_to_charset_name=0) const;

String& reconstruct(Pool& pool) const;
void join_chain(Pool& pool,	void join_chain(Pool& pool,
uint& ai, const Chunk& achunk, const Chunk::Row& arow,	const Chunk& achunk, const Chunk::Row& arow, uint& acountdown,
Untaint_lang& joined_lang, const char *& joined_ptr, size_t& joined_size) const;	uchar& joined_lang, const char *& joined_ptr, size_t& joined_size) const;
String& replace_in_reconstructed(Pool& pool, Dictionary& dict) const;

private: //disabled	private: //disabled

Line 293 private: //disabled	Line 290 private: //disabled
};	};
#include "pa_pragma_pack_end.h"	#include "pa_pragma_pack_end.h"

	#define STRING_PREPARED_FOREACH_ROW(self, body) \
	while(row!=(self).append_here) { \
	if(countdown==0) { \
	chunk=row->link; \
	row=chunk->rows; \
	countdown=chunk->count; \
	}; \
	{ body } \
	row++; countdown--; \
	}

	#define STRING_PREFIX_FOREACH_ROW(self, body) { \
	const Chunk *chunk=&(self).head; \
	const Chunk::Row *row=chunk->rows; \
	uint countdown=chunk->count; \
	STRING_PREPARED_FOREACH_ROW(self, body) \
	}

	#define STRING_FOREACH_ROW(body) STRING_PREFIX_FOREACH_ROW(*this, body)
	#define STRING_SRC_FOREACH_ROW(body) STRING_PREFIX_FOREACH_ROW(src, body)

#endif	#endif

E-mail:

Removed from v.1.117
changed lines
	Added in v.1.134