--- parser3/src/include/pa_string.h 2009/05/15 06:57:43 1.185 +++ parser3/src/include/pa_string.h 2010/09/16 23:34:26 1.199 @@ -8,7 +8,7 @@ #ifndef PA_STRING_H #define PA_STRING_H -static const char * const IDENT_STRING_H="$Date: 2009/05/15 06:57:43 $"; +static const char * const IDENT_STRING_H="$Date: 2010/09/16 23:34:26 $"; // includes #include "pa_types.h" @@ -17,6 +17,11 @@ static const char * const IDENT_STRING_H extern "C" { // cord's author forgot to do that #define CORD_NO_IO #include "cord.h" + +#ifdef CORD_CAT_OPTIMIZATION +#define CORD_cat(x, y) CORD_cat_optimized(x, y) +#define CORD_cat_char_star(x, y, leny) CORD_cat_char_star_optimized(x, y, leny) +#endif }; // defines @@ -24,6 +29,10 @@ extern "C" { // cord's author forgot to // cache hash code in String::Body for faster hash access #define HASH_CODE_CACHING +// cache String::Body.length() for char* strings only, CORDs have own length +#define STRING_LENGTH_CACHING + + // 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). */ @@ -49,11 +58,6 @@ double pa_atod(const char* str, const St /// this is result of pos functions which mean that substr were not found #define STRING_NOT_FOUND ((size_t)-1) -template -inline size_t get_length(T current) { - return current; -} - /** String which knows the lang of all it's langs. @@ -96,14 +100,16 @@ public: L_FILE_SPEC='F', ///< file specification L_HTTP_HEADER='h', ///< text in HTTP response header L_MAIL_HEADER='m', ///< text in mail header - L_URI='U', ///< text in uri - L_SQL='Q', ///< ^table:sql body - L_JS='J', ///< JavaScript code - L_XML='X', ///< ^dom:set xml - L_HTML='H', ///< HTML code - L_REGEX='R', ///< RegEx expression + L_URI='U', ///< text in uri + L_SQL='Q', ///< ^table:sql body + L_JS='J', ///< JavaScript code + L_XML='X', ///< ^xdoc:create xml + L_HTML='H', ///< HTML code + L_REGEX='R', ///< RegExp + L_JSON='S', ///< JSON code L_HTTP_COOKIE='C', ///< cookies encoded as %uXXXX for compartibility with js functions encode/decode - L_FILE_POST='f', ///temporary escaping zero-char + L_FILE_POST='f', ///< temporary escaping zero-char + L_PARSER_CODE='p', ///< ^process body // READ WARNING ABOVE BEFORE ADDING ANYTHING L_OPTIMIZE_BIT = 0x80 ///< flag, requiring cstr whitespace optimization }; @@ -114,6 +120,7 @@ public: TRIM_END }; + class Body; union Languages { @@ -130,33 +137,35 @@ public: } opt; CORD langs; - template - CORD make_langs(C current) const { - return opt.is_not_just_lang? - langs - :CORD_chars((char)opt.lang, get_length(current)); + CORD make_langs(const Body& current) const { + return opt.is_not_just_lang?langs:CORD_chars((char)opt.lang, current.length()); } CORD make_langs(size_t aoffset, size_t alength) const { return opt.is_not_just_lang? - CORD_substr(langs, aoffset, alength) + CORD_substr(langs, aoffset, alength, 0) :CORD_chars((char)opt.lang, alength); } /// appending when 'langs' already contain something [simple cases handled elsewhere] - template - void append(C current, - const CORD to_nonempty_target_langs) { + void append(size_t current, const CORD to_nonempty_target_langs) { assert(langs); + if(opt.is_not_just_lang) + langs=CORD_cat(langs, to_nonempty_target_langs); + else { + assert(current); + langs=CORD_cat(CORD_chars((char)opt.lang, current), to_nonempty_target_langs); + } + } + void append(const Body& current, const CORD to_nonempty_target_langs) { + assert(langs); if(opt.is_not_just_lang) langs=CORD_cat(langs, to_nonempty_target_langs); - else { // we were "just lang" - size_t current_size=get_length(current); + else { + size_t current_size=current.length(); assert(current_size); - langs=CORD_cat( - CORD_chars((char)opt.lang, current_size), // first piece [making from just 'lang'] - to_nonempty_target_langs); // new piece + langs=CORD_cat(CORD_chars((char)opt.lang, current_size), to_nonempty_target_langs); } } @@ -172,10 +181,9 @@ public: } /// MUST be called exactly prior to modification of current [uses it's length] - template - void append(C current, Language alang, size_t asize) { + void append(size_t current, Language alang, size_t length) { assert(alang); - assert(asize); + assert(length); if(!opt.is_not_just_lang) if(opt.lang) { @@ -186,12 +194,12 @@ public: return; } - append(current, CORD_chars((char)alang, asize)); + append(current, CORD_chars((char)alang, length)); } - template - void appendHelper(C current, Language alang, C asize_helper) { + void append(const Body ¤t, Language alang, size_t length) { assert(alang); + assert(length); if(!opt.is_not_just_lang) if(opt.lang) { @@ -202,30 +210,45 @@ public: return; } - append(current, CORD_chars((char)alang, asize_helper.length())); + append(current, CORD_chars((char)alang, length)); } - template - void appendHelper(C current, C length_helper, const Languages src) { + void appendHelper(const Body& current, Language alang, const Body &length_helper) { + assert(alang); - if(!langs) + if(!opt.is_not_just_lang) + if(opt.lang) { + if(opt.lang==alang) // same language? ignoring + return; + } else { + opt.lang=alang; // to uninitialized + return; + } + + append(current, CORD_chars((char)alang, length_helper.length())); + } + + void appendHelper(const Body& current, const Languages &src, const Body& length_helper) { + if(!langs){ langs=src.langs; // to uninitialized +#ifdef CORD_CAT_OPTIMIZATION + if(opt.is_not_just_lang && !CORD_IS_STRING(langs)) + CORD_concatenation_protect(langs); +#endif + } else if(!src.opt.is_not_just_lang) appendHelper(current, src.opt.lang, length_helper); // simplifying when simple source else append(current, src.make_langs(length_helper)); } - /// MUST be called exactly prior to modification of current [uses it's length] - template - void append(C current, - const Languages src, size_t aoffset, size_t alength) { + void append(const Body& current, const Languages src, size_t aoffset, size_t alength) { assert(alength); if(!langs) // to uninitialized? if(src.opt.is_not_just_lang) - langs=CORD_substr(src.langs, aoffset, alength); // to uninitialized complex + langs=CORD_substr(src.langs, aoffset, alength, 0); // to uninitialized complex else opt.lang=src.opt.lang; // to uninitialized simple else @@ -256,14 +279,20 @@ public: : 0; }; - template - void for_each(C current, - int callback(char, size_t, I), I info) const { - + template + void for_each(size_t current, int callback(char, size_t, I), I info) const { if(opt.is_not_just_lang) CORD_block_iter(langs, 0, (CORD_block_iter_fn)callback, info); else - callback(opt.lang, get_length(current), info); + callback(opt.lang, current, info); + } + + template + void for_each(const Body& current, int callback(char, size_t, I), I info) const { + if(opt.is_not_just_lang) + CORD_block_iter(langs, 0, (CORD_block_iter_fn)callback, info); + else + callback(opt.lang, current.length(), info); } bool invariant(size_t current_length) const { @@ -285,53 +314,96 @@ public: mutable uint hash_code; #endif +#ifdef STRING_LENGTH_CACHING + // cached length is reseted on modification, used only for char*, not CORD + mutable size_t string_length; +#define INIT_LENGTH ,string_length(0) +#define ZERO_LENGTH string_length=0; +#else +#define INIT_LENGTH +#define ZERO_LENGTH +#endif + public: const char* v() const; void dump() const; #ifdef HASH_CODE_CACHING - Body(): body(CORD_EMPTY), hash_code(0) {} - Body(CORD abody, uint ahash_code): body(abody), hash_code(ahash_code) {} - Body(CORD abody): body(abody), hash_code(0) { + Body(): body(CORD_EMPTY), hash_code(0) INIT_LENGTH {} + Body(CORD abody, uint ahash_code): body(abody), hash_code(ahash_code) INIT_LENGTH {} + Body(CORD abody): body(abody), hash_code(0) INIT_LENGTH { #else - Body(): body(CORD_EMPTY) {} - Body(CORD abody): body(abody) { + Body(): body(CORD_EMPTY) INIT_LENGTH {} + Body(CORD abody): body(abody) INIT_LENGTH { #endif +#ifdef CORD_CAT_OPTIMIZATION assert(!body // no body || *body // ordinary string || body[1]==1 // CONCAT_HDR + || body[1]==3 // CONCAT_HDR_READ_ONLY || body[1]==4 // FN_HDR || body[1]==6 // SUBSTR_HDR ); +#else + assert(!body // no body + || *body // ordinary string + || body[1]==1 // CONCAT_HDR + || body[1]==4 // FN_HDR + || body[1]==6 // SUBSTR_HDR + ); +#endif } static Body Format(int value); - void clear() { body=CORD_EMPTY; } + void clear() { ZERO_LENGTH body=CORD_EMPTY; } bool operator! () const { return is_empty(); } CORD get_cord() const { return body; } uint get_hash_code() const; - const char* cstr() const { return CORD_to_const_char_star(body); } - char* cstrm() const { return CORD_to_char_star(body); } + const char* cstr() const { +#ifdef STRING_LENGTH_CACHING + string_length = length(); + if(string_length) + return const_cast(this)->body=CORD_to_const_char_star(body, string_length); +#endif + return CORD_to_const_char_star(body, length()); + } + + char* cstrm() const { return CORD_to_char_star(body, length()); } +#ifdef STRING_LENGTH_CACHING + void set_length(size_t alength){ string_length = alength; } + size_t length() const { return body ? CORD_IS_STRING(body) ? string_length ? string_length : (string_length=strlen(body)) : CORD_len(body) : 0; } +#else size_t length() const { return CORD_len(body); } +#endif 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); + if(known_length){ + if(body){ + body = CORD_cat_char_star(body, str, known_length); + ZERO_LENGTH + } else { + body=str; +#ifdef STRING_LENGTH_CACHING + string_length=known_length; +#endif + } + } } 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); } - Body& operator << (const Body src) { body=CORD_cat(body, src.body); return *this; } + void append(char c) { ZERO_LENGTH body=CORD_cat_char(body, c); } + Body& operator << (const Body src) { ZERO_LENGTH body=CORD_cat(body, src.body); return *this; } + Body& 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->==] @@ -347,14 +419,14 @@ public: } char fetch(size_t index) const { return CORD_fetch(body, index); } - Body 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); } + Body mid(size_t aindex, size_t alength) const { return CORD_substr(body, aindex, alength, length()); } + size_t pos(const char* substr, size_t offset=0) const { return CORD_str(body, offset, substr, length()); } size_t pos(const Body substr, size_t offset=0) const { if(substr.is_empty()) return STRING_NOT_FOUND; // in this case CORD_str returns 0 [parser users got used to -1] // CORD_str checks for bad offset [CORD_chr does not] - return CORD_str(body, offset, substr.body); + return CORD_str(body, offset, substr.body, length()); } size_t pos(char c, size_t offset=0) const { @@ -364,16 +436,13 @@ public: return CORD_chr(body, offset, c); } - template int for_each( - int (*f)(char c, I), - I info) const { + template + int for_each(int (*f)(char c, I), I info) const { return CORD_iter(body, (CORD_iter_fn)f, (void*)info); } - template int for_each( - int (*f1)(char c, I), - int (*f2)(const char* s, I), - I info) const { + template + int for_each(int (*f1)(char c, I), int (*f2)(const char* s, I), I info) const { return CORD_iter5(body, 0, (CORD_iter_fn)f1, (CORD_batched_iter_fn)f2, info); } @@ -385,7 +454,7 @@ public: /// @returns this or 0 or mid. if returns this or 0 out_* are not filled Body trim(Trim_kind kind=TRIM_BOTH, const char* chars=0, - size_t* out_start=0, size_t* out_length=0) const; + size_t* out_start=0, size_t* out_length=0, Charset* source_charset=0) const; }; struct C { @@ -425,12 +494,16 @@ public: langs=alang; } } - explicit String(const String::C cstr, Language alang=L_CLEAN){ - if(cstr.length){ - body=cstr.str; + explicit String(const char* cstr, Language alang, size_t alength){ + if(cstr && *cstr){ + body=cstr; +#ifdef STRING_LENGTH_CACHING + body.set_length(alength); +#endif langs=alang; } } + String(int value, char *format); String(Body abody, Language alang): body(abody), langs(alang) { ASSERT_STRING_INVARIANT(*this); @@ -450,23 +523,38 @@ public: size_t length() const { return body.length(); } size_t length(Charset& charset) const; - /// convert to CORD. if 'lang' known, forcing 'lang' to it - Body 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(); + /// convert to CORD forcing lang tainting + Body cstr_to_string_body_taint(Language lang, SQL_Connection* connection=0, const Request_charsets *charsets=0) const; + /// convert to CORD with tainting dirty to lang + Body cstr_to_string_body_untaint(Language lang, SQL_Connection* connection=0, const Request_charsets *charsets=0) const; + + /// + const char* cstr() const { + return body.cstr(); + } + /// + char* cstrm() const { + return body.cstrm(); + } + + /// convert to constant C string forcing lang tainting + const char* taint_cstr(Language lang, SQL_Connection* connection=0, const Request_charsets *charsets=0) const { + return cstr_to_string_body_taint(lang, connection, charsets).cstr(); } + char *taint_cstrm(Language lang, SQL_Connection* connection=0, const Request_charsets *charsets=0) const { + return cstr_to_string_body_taint(lang, connection, charsets).cstrm(); + } + + /// convert to constant C string with tainting dirty to lang + const char* untaint_cstr(Language lang, SQL_Connection* connection=0, const Request_charsets *charsets=0) const { + return cstr_to_string_body_untaint(lang, connection, charsets).cstr(); + } + char *untaint_cstrm(Language lang, SQL_Connection* connection=0, const Request_charsets *charsets=0) const { + return cstr_to_string_body_untaint(lang, connection, charsets).cstrm(); + } + + const char* untaint_and_transcode_cstr(Language lang, const Request_charsets *charsets) const; + /// puts pieces to buf Cm serialize(size_t prolog_size) const; /// appends pieces from buf to self @@ -489,13 +577,25 @@ public: return Body(x).ncmp(0/*x_begin*/, body, 0/*y_begin*/, length())==0; } - String& append_to(String& dest, Language lang, bool forced) const; + String& append_to(String& dest, Language lang, bool forced=false) const; String& append(const String& src, Language lang, bool forced=false) { return src.append_to(*this, lang, forced); } + String& append_quoted(const String* src, Language lang=L_JSON){ + *this << "\""; + if(src) + this->append(*src, lang, true/*forced lang*/); + *this << "\""; + return *this; + } + 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 Body src); + String& operator << (const Body& src){ + langs.appendHelper(body, L_AS_IS, src); + body< -inline size_t get_length(String::Body body) { - return body.length(); -} - #ifndef HASH_CODE_CACHING /// simple hash code of string. used by Hash inline uint hash_code(const String::Body self) { @@ -586,12 +681,4 @@ inline uint hash_code(const String::Body #endif -/// now that we've declared specialization we can use it -inline String& String::operator << (const String::Body src) { - langs.append(body, L_AS_IS, src.length()); - body<