Return to pa_string.h CVS log

Up to [parser3project] / parser3 / src / include

just compiled
todo: serialize&deserialize

/** @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: 2003/09/23 10:37:05 $";

// includes

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

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

CORD CORD_chars_gen(char c, size_t i);

// cord extension
/* Returns true if x does contain                                       */
/* char not_c at positions i..i+n. Value i,i+n must be < CORD_len(x).	*/
int CORD_nnchr(CORD x, size_t i, size_t n, int not_c);

// forwards

class Charset;
class Table;
class SQL_Connection;
class Dictionary;
class Request_charsets;
class String;
typedef Array<const String*> ArrayString;

/// this is result of pos functions which mean that substr were not found
#define STRING_NOT_FOUND ((size_t)-1)

class StringBody {

	CORD body;

public:

	StringBody(): body(CORD_EMPTY) {}
	StringBody(CORD abody): body(abody) {
		assert(!body // no body
			|| *body // ordinary string
			|| body[1]==1 // CONCAT_HDR
			|| body[1]==4 // FN_HDR 
			|| body[1]==6 // SUBSTR_HDR 
			);
	}
	/// 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_know_length(str, helper_length?helper_length:strlen(str));
	}
	static StringBody Format(int value);

	void clear() { body=CORD_EMPTY; }

	bool operator! () const { return is_empty(); }

	uint hash_code() const;

	const char* cstr() const { return CORD_to_const_char_star(body); }
	char* cstrm() const { return CORD_to_char_star(body); }

	size_t length() const { return CORD_len(body); }

	bool is_empty() const { return body==CORD_EMPTY; }

	void append_know_length(const char *str, size_t known_length) {
		if(known_length)
			body=CORD_cat_char_star(body, str, known_length);
	}
	void append_strdup_know_length(const char* str, size_t known_length) {
		if(known_length)
			append_know_length(pa_strdup(str, known_length), known_length);
	}
	void append(char c) { body=CORD_cat_char(body, c); }
	StringBody& operator << (const StringBody src) { body=CORD_cat(body, src.body); return *this; }
	StringBody& operator << (const char* str) { append_know_length(str, strlen(str)); 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; }

	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);
	}

	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 { 
		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);
	}*/

	void set_pos(CORD_pos& pos, size_t index) const { CORD_set_pos(pos, body, index); }

	/*StringBody normalize() const {
		return StringBody(CORD_balance(body));
	}*/

	void dump() const {
		CORD_dump(body);
	}
};

/** 
	String which knows the lang of all it's langs.

	All pieces remember 
	- whether they are tainted or not, 
	  and the lang which should be used to detaint them
*/
class String: public PA_Object {
public:

	/** piece is tainted or not. the lang to use when detaint
		remember to change String_Untaint_lang_name @ untaint.C along
	*/
	enum Language {
		L_UNSPECIFIED=0, ///< no real string has parts of this lange: it's just convinient to check when string's empty
		// these two must go before others, there are checks for >L_AS_IS
		L_CLEAN, ///< clean
		L_AS_IS,     ///< leave all characters intact

		L_PASS_APPENDED,
			/**<
				leave lang built into string being appended.
				just a flag, that value not stored
			*/
		L_TAINTED,  ///< tainted, untaint lang as assigned later 
		// untaint langs. assigned by ^untaint[lang]{...}
		L_FILE_SPEC, ///< file specification
		L_HTTP_HEADER,    ///< text in HTTP response header
		L_MAIL_HEADER,    ///< text in mail header
		L_URI,       ///< text in uri
		L_TABLE,     ///< ^table:set body
		L_SQL,       ///< ^table:sql body
		L_JS,        ///< JavaScript code
		L_XML,		///< ^dom:set xml
		L_HTML,      ///< HTML code (for editing)
		L_OPTIMIZE_BIT = 0x80  ///< flag, requiring cstr whitespace optimization
	};

	class Languages {

		union {
			struct {
				Language lang:8;
				int is_not_just_lang:16-8;
			};
			CORD langs;
		};

		CORD make_langs(const StringBody body) const {
			return is_not_just_lang?
				langs
				:CORD_chars_gen((char)lang, body.length());
		}

		CORD make_langs(size_t aoffset, size_t alength)  const {
			return is_not_just_lang?
				CORD_substr(langs, aoffset, alength)
				:CORD_chars_gen((char)lang, alength);
		}

		/// @returns true if appended, false if not: 
		/// when appending can be simplified by just assigning lang/langs
		bool append(const StringBody current_body, 
			const CORD to_nonempty_target_langs) {
			if(!lang)
				return false;

			if(is_not_just_lang)
				langs=CORD_cat(langs, to_nonempty_target_langs);
			else { // we were "just lang"
				size_t current_size=current_body.length();
				assert(current_size);
				langs=CORD_cat(
					CORD_chars_gen((char)lang, current_size),  // first piece [making from just 'lang']
					to_nonempty_target_langs); // new piece
			}

			return true;
		}

	public:

		Languages(): langs(0) {}
		Languages(Language alang): lang(alang), is_not_just_lang(0) {}

		/// MUST be called prior to modification of current_body [uses it's original length]
		void append(const StringBody current_body, Language alang, size_t asize) {
			assert(alang);
			assert(asize);

			if(!is_not_just_lang && lang==alang)
				return;

			if(!append(current_body, CORD_chars((char)alang, asize)))
				lang=alang;
		}

		/// MUST be called prior to modification of current_body [uses it's original length]
		void append(const StringBody current_body, const StringBody appending_body, 
			const Languages src) {
			assert(appending_body.length());

			if(!is_not_just_lang && !src.is_not_just_lang && lang==src.lang)
				return;

			if(!append(current_body, src.make_langs(appending_body)))
				langs=src.langs;
		}
		
		/// MUST be called prior to modification of current_body [uses it's original length]
		void append(const StringBody current_body, 
			const Languages src, size_t aoffset, size_t alength) {
			assert(alength);
			if(!is_not_just_lang && !src.is_not_just_lang && lang==src.lang)
				return;

			if(!append(current_body, src.make_langs(aoffset, alength)))
				if(src.is_not_just_lang)
					langs=CORD_substr(src.langs, aoffset, alength);
				else
					lang=src.lang;
		}

		/// checks if we have alang all from aoffset to aoffset+alength
		bool check_lang(Language alang, size_t aoffset, size_t alength) const {
			if(alang==L_UNSPECIFIED) // ignore lang?
				return true;

			if(is_not_just_lang)
				return CORD_nnchr(langs, aoffset, alength, (unsigned)alang)!=0;
			else
				return lang==alang;
		}

		template<typename I> void for_each(const StringBody current_body, 
			int callback(char, size_t, I), I info) const {
			
			if(is_not_just_lang)
				CORD_block_iter(langs, 0, (CORD_block_iter_fn)callback, info);
			else
				callback(lang, current_body.length(), info);
		}
	};

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

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

private:

	Languages langs; ///< string characters lang
	StringBody body; ///< all characters of string

public:

	explicit String(const char* cstr=0, size_t helper_length=0, bool tainted=false);
	explicit String(const C cstr, bool tainted=false);
	String(StringBody abody, Language alang): body(abody), langs(alang) {}
	String(const String& src): body(src.body), langs(src.langs) {}

	/// for convinient hash lookup
	operator const StringBody() const { return body; }

	bool is_empty() const { return body.is_empty(); }
	size_t length() const { return body.length(); }

	/// convert to CORD. if 'lang' known, forcing 'lang' to it
	StringBody cstr_to_string_body(Language lang=L_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(Language lang=L_AS_IS, 
		SQL_Connection* connection=0,
		const Request_charsets *charsets=0) const {
		return cstr_to_string_body(lang, connection, charsets).cstr();
	}
	/// convert to Modifiable C string. if 'lang' known, forcing 'lang' to it
	char *cstrm(Language lang=L_AS_IS, 
		SQL_Connection* connection=0,
		const Request_charsets *charsets=0) const {
		return cstr_to_string_body(lang, connection, charsets).cstrm();
	}
	/// puts pieces to buf
	Cm serialize(size_t prolog_size) const;
	/// appends pieces from buf to self
	bool deserialize(size_t prolog_size, void *buf, size_t buf_size);
	/// @see StringBody::append_know_length
	String& append_know_length(const char* str, size_t known_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 helper_length, Language lang);

	bool operator == (const char* y) const { return body==StringBody(y); }
	bool operator != (const char* y) const { return body!=StringBody(y); }

	/// this starts with y
	bool starts_with(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;
	String& append(const String& src, Language lang, bool forced=false) { 
		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_help_length(src, 0, L_AS_IS); }
	String& operator << (const StringBody src) { 
		langs.append(body, L_AS_IS, src.length());
		body<<src;
		return *this;
	}

	/// extracts first char of a string, if any
	char first_char() const {
		return is_empty()?0:body.fetch(0);
	}

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

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

	/** 
		ignore lang if it's L_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(const StringBody substr, 
		size_t this_offset=0, Language lang=L_UNSPECIFIED) const;
	/// 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,
		const char* delim, 
		Language lang=L_UNSPECIFIED, int limit=-1) const;
	void split(ArrayString& result, 
		size_t& pos_after, 
		const String& delim, 
		Language lang=L_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;
	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 StringBody self) {
	return self.hash_code();
}

#endif