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 1.104 and 1.125

version 1.104, 2001/09/21 15:35:45	version 1.125, 2002/02/20 09:13:07
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: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
Author: Alexander Petrosyan <paf@design.ru> (http://design.ru/paf)

$Id$	$Id$
*/	*/
Line 11	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"

class Table;	class Table;

/**
$MAIN:html-typo table elements must enlarge string not more that that.

that's a tradeoff - otherwise we'd have to scan string twice:
- first for buffer length
- second for replacements themselves
*/
#define UNTAINT_TIMES_BIGGER 10

#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 38 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 79 class Dictionary;	Line 68 class Dictionary;
- whether they are tainted or not,	- whether they are tainted or not,
and the language which should be used to detaint them	and the language which should be used to detaint them
*/	*/
	#include "pa_pragma_pack_begin.h"
class String : public Pooled {	class String : public Pooled {
public:	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
	remember to change String_Untaint_lang_name @ untaint.C along
	*/
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 108 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_USER_HTML ///< HTML code with USER chars	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);
size_t size() const { return fsize; }	size_t size() const;
/// makes store_to produce origin:text\n output
void set_origins_mode(bool aorigins_mode) const { forigins_mode=aorigins_mode; }
/// 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_UNSPECIFIED,	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 result=(char )malloc(cstr_bufsize(lang));	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);
*eol=0;	*eol=0;
return result;	return result;
}	}
	char *cstr_debug_origins() const;
	/// puts pieces to buf
	void serialize(size_t prolog_size, void *& buf, size_t& buf_size) const;
	/// appends pieces from buf to self
	void 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 168 public:	Line 165 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 190 public:	Line 180 public:

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

void split(Array& result,	void split(Array& result,
size_t *pos_after_ref,	size_t *pos_after_ref,
Line 213 public:	Line 203 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 224 public:	Line 214 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 235 public:	Line 225 public:
const Origin& origin() const;	const Origin& origin() const;
#endif	#endif

private:	public: // debug private:

/// several String fragments	/// several String fragments
struct Chunk {	struct Chunk {
size_t count; ///< the number of rows in 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
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];
Chunk *preallocated_link; ///< next rows are here
}	}
head; ///< the head chunk of the chunk chain	head; ///< the head chunk of the chunk chain
	Chunk *head_link; ///< next rows are here must go right after 'head'

/// next append would write to this record	/// next append would write to this record
Chunk::Row *append_here;	Chunk::Row *append_here;
Line 266 private:	Line 258 private:
Chunk::Row *link_row;	Chunk::Row *link_row;

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

// string size
size_t fsize;

// used rows in all chunks
int fused_rows;

/// origins_mode changes behaviour of store_to[cstr()]
mutable bool forigins_mode;

	public: //debug
	uint used_rows() const;
private:	private:

bool chunk_is_full() {	bool chunk_is_full() {
return append_here == link_row;	return append_here == link_row;
}	}
	//uint used_rows() const;
void expand();	void expand();

	Untaint_lang lang_of(size_t offset) const;

size_t cstr_bufsize(Untaint_lang lang) const;	size_t cstr_bufsize(Untaint_lang lang,
	SQL_Connection *connection,
	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;

	String& reconstruct(Pool& pool) const;
	void join_chain(Pool& pool,
	uint& ai, const Chunk& achunk, const Chunk::Row& arow,
	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

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

};	};
	#include "pa_pragma_pack_end.h"

	#define STRING_PREFIX_FOREACH_ROW(src, body) { \
	const Chunk *chunk=&(src).head; \
	do { \
	const Chunk::Row *row=chunk->rows; \
	for(uint i=0; i<chunk->count; i++, row++) { \
	if(row==(src).append_here) \
	goto break2; \
	\
	body \
	} \
	chunk=row->link; \
	} while(chunk); \
	}

	#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.1.104
changed lines
	Added in v.1.125