--- parser3/src/include/pa_string.h 2003/03/20 15:05:57 1.144.2.28.2.14 +++ parser3/src/include/pa_string.h 2003/03/24 08:27:45 1.144.2.28.2.21 @@ -8,7 +8,7 @@ #ifndef PA_STRING_H #define PA_STRING_H -static const char* IDENT_STRING_H="$Date: 2003/03/20 15:05:57 $"; +static const char* IDENT_STRING_H="$Date: 2003/03/24 08:27:45 $"; // includes @@ -27,22 +27,72 @@ class SQL_Connection; class Dictionary; class Request_charsets; class String; -typedef Array ArrayString; +typedef Array ArrayString; // helpers /* /// appends clean piece to String @see String::append -#define APPEND_CLEAN(src, length) append(src, length, String::UL_CLEAN) +#define APPEND_CLEAN(src, length) append(src, length, String::L_CLEAN) /// appends piece to String as-is @see String::append -#define APPEND_AS_IS(src, length) append(src, length, String::UL_AS_IS) +#define APPEND_AS_IS(src, length) append(src, length, String::L_AS_IS) /// appends tainted piece to String @see String::append -#define APPEND_TAINTED(src, length) append(src, length, String::UL_TAINTED) +#define APPEND_TAINTED(src, length) append(src, length, String::L_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) +class StringBody { + + CORD body; + +public: + + StringBody(): body(CORD_EMPTY) {} + StringBody(CORD abody): body(abody) {} + + void clear() { body=CORD_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(const char *str, size_t length) { + CORD_cat_char_star(body, str, length); + } + void append(char c) { CORD_cat_char(body, c); } + void append(StringBody src) { CORD_cat(body, src.body); } + + // 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(size_t offset, const char* substr) const { return CORD_str(body, offset, substr); } + size_t pos(size_t offset, const StringBody substr) const { return CORD_str(body, offset, substr.body); } + + template 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); } + +}; /** String which knows the language of all it's fragments. @@ -50,54 +100,55 @@ typedef Array ArrayStrin 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 { + friend class StringBody; + 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 + enum Language { + L_UNSPECIFIED=0, ///< zero value handy for hash lookup @see untaint_lang_name2enum + // these two must go before others, there are checks for >L_AS_IS + L_CLEAN, ///< clean + L_AS_IS, ///< leave all characters intact - UL_PASS_APPENDED, + L_PASS_APPENDED, /**< leave language built into string being appended. just a flag, that value not stored */ - UL_TAINTED, ///< tainted, untaint language as assigned later + L_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 + 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 = 0x8000 ///< flag, requiring cstr whitespace optimization }; struct Fragment { - Untaint_lang lang; ///< untaint flag, later untaint language + Language lang; ///< untaint flag, later untaint language size_t length; ///< length - Fragment(Untaint_lang alang, size_t asize): lang(alang), length(asize) {} + Fragment(Language alang, size_t asize): lang(alang), length(asize) {} }; class ArrayFragment: public Array { - void simple_append(element_type src) { + void append(element_type src) { *static_cast *>(this)+=src; } + /// hiding from accidental USE, use append_positions + void append(const ArrayFragment& src, int offset, int limit) { + static_cast *>(this)->append(src, offset, limit); + } public: ArrayFragment& operator += (element_type src) { if(count()) { // not empty? @@ -108,10 +159,11 @@ public: return *this; } } - simple_append(src); + append(src); return *this; } - ArrayFragment& append(const ArrayFragment& src, size_t substr_begin, size_t substr_end); + void append(const ArrayFragment& src) { append(src, 0, 0); } + void append_positions(const ArrayFragment& src, size_t substr_begin, size_t substr_end); }; struct C { @@ -123,7 +175,7 @@ public: private: - CORD body; ///< all characters of string + StringBody body; ///< all characters of string ArrayFragment fragments; ///< fragment language+length info public: @@ -131,95 +183,101 @@ 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); } + String(StringBody abody, Language alang): body(abody) { + fragments+=Fragment(alang, abody.length()); + } + + /// for convinient hash lookup + operator 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 - CORD cstr_to_cord(Untaint_lang lang=UL_AS_IS, - SQL_Connection *connection=0, + 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(Untaint_lang lang=UL_AS_IS, - SQL_Connection *connection=0, + const char* cstr(Language lang=L_AS_IS, + SQL_Connection* connection=0, const Request_charsets *charsets=0) const { - return CORD_to_const_char_star(cstr_to_cord(lang, connection, charsets)); + return cstr_to_string_body(lang, connection, charsets).cstr(); } /// convert to Modifiable C string. if 'lang' known, forcing 'lang' to it - char *cstrm(Untaint_lang lang=UL_AS_IS, - SQL_Connection *connection=0, + char *cstrm(Language lang=L_AS_IS, + SQL_Connection* connection=0, const Request_charsets *charsets=0) const { - return CORD_to_char_star(cstr_to_cord(lang, connection, charsets)); + return cstr_to_string_body(lang, connection, charsets).cstrm(); } /// 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 + /** append fragment, + @c length can be zero, it will be autocalced + when length (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 char* y) const { return body==StringBody(y); } + bool operator != (const char* y) const { return body!=StringBody(y); } /// this starts with y - bool starts_with(CORD y) const { - return CORD_ncmp(body, 0/*x_begin*/, y, 0/*y_begin*/, CORD_len(y))==0; + bool starts_with(const char* y) const { + return body.ncmp(0/*x_begin*/, StringBody(y), 0/*y_begin*/, strlen(y))==0; } /// y starts with this - bool this_starts(CORD y) const { - return CORD_ncmp(body, 0/*x_begin*/, y, 0/*y_begin*/, CORD_len(body))==0; + bool this_starts(const char* y) const { + return body.ncmp(0/*x_begin*/, StringBody(y), 0/*y_begin*/, strlen(y))==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) { + 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, 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; + String& operator << (const String& src) { return append(src, L_PASS_APPENDED); } + String& operator << (const char* src) { return append(src, 0, L_AS_IS); } + String& operator << (const StringBody& src) { + body.append(src); + fragments+=Fragment(L_AS_IS, src.length()); + return *this; + } /// extracts first char of a string, if any char first_char() const { - return operator bool()?CORD_fetch(body, 0):0; + return is_empty()?0:body.fetch(0); } + bool operator < (const String& src) const { return body (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 const String& mid(size_t substr_begin, size_t substr_end) const; /** - ignore lang if it's UL_UNSPECIFIED + 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(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 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, Untaint_lang lang=UL_UNSPECIFIED) const; + size_t this_offset=0, Language lang=L_UNSPECIFIED) const; void split(ArrayString& result, - size_t& pos_after_ref, + size_t& pos_after, const char* delim, - Untaint_lang lang=UL_UNSPECIFIED, int limit=-1); + Language lang=L_UNSPECIFIED, int limit=-1); void split(ArrayString& result, - size_t& pos_after_ref, + size_t& pos_after, const String& delim, - Untaint_lang lang=UL_UNSPECIFIED, int limit=-1) const; + Language lang=L_UNSPECIFIED, int limit=-1) const; typedef void (*Row_action)(Table* table, ArrayString* row, int prestart, int prefinish, @@ -254,7 +312,7 @@ private: //disabled }; /// simple hash code of string. used by Hash -inline uint hash_code(const String& self) { +inline uint hash_code(const StringBody& self) { return self.hash_code(); }