--- parser3/src/include/pa_string.h 2003/09/25 09:41:24 1.147 +++ parser3/src/include/pa_string.h 2009/04/22 04:37:52 1.178 @@ -1,17 +1,16 @@ /** @file Parser: string class decl. - Copyright (c) 2001-2003 ArtLebedev Group (http://www.artlebedev.com) + Copyright (c) 2001-2005 ArtLebedev Group (http://www.artlebedev.com) Author: Alexandr Petrosian (http://paf.design.ru) */ #ifndef PA_STRING_H #define PA_STRING_H -static const char* IDENT_STRING_H="$Date: 2003/09/25 09:41:24 $"; +static const char * const IDENT_STRING_H="$Date: 2009/04/22 04:37:52 $"; // includes - #include "pa_types.h" #include "pa_array.h" @@ -20,10 +19,13 @@ extern "C" { // cord's author forgot to #include "cord.h" }; +// defines + // 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_range_contains_chr_greater_then(CORD x, size_t i, size_t n, int c); +size_t CORD_block_count(CORD x); // forwards @@ -34,6 +36,12 @@ class Dictionary; class Request_charsets; class String; typedef Array ArrayString; +class VRegex; + +// generally useful + +int pa_atoi(const char* str, const String* problem_source=0); +double pa_atod(const char* str, const String* problem_source=0); /// this is result of pos functions which mean that substr were not found #define STRING_NOT_FOUND ((size_t)-1) @@ -50,42 +58,72 @@ inline size_t get_length(T current) { - 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 + + WARNING WARNING WARNING WARNING WARNING WARNING + + pos function compares(<=) languages, that is used in searching + for table column separator being L_CLEAN or L_AS_IS. + they search for AS_IS, meaning AS_IS|CLEAN [doing <=L_AS_IS check]. + + letters assigned for debugging, but it's important for no language-letter + come before L_AS_IS other then L_CLEAN + + WARNING WARNING WARNING WARNING WARNING WARNING */ 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='1', ///< clean - L_AS_IS, ///< leave all characters intact + L_CLEAN='0', ///< clean WARNING: read above warning before changing + L_AS_IS='A', ///< leave all characters intact WARNING: read above warning before changing - L_PASS_APPENDED, + L_PASS_APPENDED='P', /**< leave lang built into string being appended. just a flag, that value not stored */ - L_TAINTED, ///< tainted, untaint lang as assigned later + L_TAINTED='T', ///< 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_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_HTTP_COOKIE='C', ///< cookies encoded as %uXXXX for compartibility with js functions encode/decode + L_FILE_POST='f', ///temporary escaping zero-char + // READ WARNING ABOVE BEFORE ADDING ANYTHING L_OPTIMIZE_BIT = 0x80 ///< flag, requiring cstr whitespace optimization }; + enum Trim_kind { + TRIM_BOTH, + TRIM_START, + TRIM_END + }; + + union Languages { struct { +#ifdef PA_LITTLE_ENDIAN Language lang:8; int is_not_just_lang:sizeof(CORD)*8-8; +#elif defined(PA_BIG_ENDIAN) + int is_not_just_lang:sizeof(CORD)*8-8; + Language lang:8; +#else +# error word endianness not determined for some obscure reason +#endif } opt; CORD langs; @@ -102,7 +140,7 @@ public: :CORD_chars((char)opt.lang, alength); } - /// appending when 'langs' already contain something [simple cases hanled elsewhere] + /// appending when 'langs' already contain something [simple cases handled elsewhere] template void append(C current, const CORD to_nonempty_target_langs) { @@ -122,6 +160,7 @@ public: public: const char* v() const; + void dump() const; Languages(): langs(0) {} Languages(Language alang) { @@ -147,20 +186,34 @@ public: append(current, CORD_chars((char)alang, asize)); } - /// MUST be called exactly prior to modification of current [uses it's length] template - void append(C current, size_t appending_length, - const Languages src) { - assert(appending_length); + void appendHelper(C current, Language alang, C asize_helper) { + assert(alang); + + if(!opt.is_not_just_lang) + if(opt.lang) { + if(opt.lang==alang) // same length? ignoring + return; + } else { + opt.lang=alang; // to uninitialized + return; + } + + append(current, CORD_chars((char)alang, asize_helper.length())); + } + + template + void appendHelper(C current, C length_helper, const Languages src) { if(!langs) langs=src.langs; // to uninitialized else if(!src.opt.is_not_just_lang) - append(current, src.opt.lang, appending_length); // simplifying when simple source + appendHelper(current, src.opt.lang, length_helper); // simplifying when simple source else - append(current, src.make_langs(appending_length)); + 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, @@ -187,9 +240,19 @@ public: if(opt.is_not_just_lang) return CORD_range_contains_chr_greater_then(langs, aoffset, alength, (unsigned)alang)==0; else - return opt.lang<=alang; + return (unsigned)opt.lang<=(unsigned)alang; } + /// @returns count of blocks + /// @todo currently there can be adjucent blocks of same language. someday merge them + size_t count() const { + return opt.is_not_just_lang? + CORD_block_count(langs) + : opt.lang? + 1 + : 0; + }; + template void for_each(C current, int callback(char, size_t, I), I info) const { @@ -200,7 +263,7 @@ public: callback(opt.lang, get_length(current), info); } - bool invariant(size_t current_length) { + bool invariant(size_t current_length) const { if(!langs) return current_length==0; if(opt.is_not_just_lang) @@ -216,6 +279,7 @@ public: public: const char* v() const; + void dump() const; Body(): body(CORD_EMPTY) {} Body(CORD abody): body(abody) { @@ -275,28 +339,47 @@ public: size_t pos(const Body 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] + + // CORD_str checks for bad offset [CORD_chr does not] return CORD_str(body, offset, substr.body); } size_t pos(char c, size_t offset=0) const { + if(offset>=length()) // CORD_chr does not check that [and ABORT's in that case] + return STRING_NOT_FOUND; + return CORD_chr(body, offset, c); } - /* template void for_each(int (*callback)(const char* s, I), I info) const { - CORD_iter5(body, 0, 0, (CORD_batched_iter_fn)callback, info); - }*/ + 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 { + return CORD_iter5(body, 0, (CORD_iter_fn)f1, (CORD_batched_iter_fn)f2, info); + } void set_pos(CORD_pos& pos, size_t index) const { CORD_set_pos(pos, body, index); } /*Body normalize() const { return Body(CORD_balance(body)); }*/ + + /// @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; }; struct C { const char *str; size_t length; operator const char *() { return str; } + C(): str(0), length(0) {} C(const char *astr, size_t asize): str(astr), length(asize) {} }; @@ -304,25 +387,27 @@ public: char *str; size_t length; //operator char *() { return str; } + Cm(): str(0), length(0) {} Cm(char *astr, size_t asize): str(astr), length(asize) {} }; private: - Languages langs; ///< string characters lang Body body; ///< all characters of string + Languages langs; ///< string characters lang const char* v() const; - -#define ASSERT_STRING_INVARIANT(string) \ - assert((string).langs.invariant((string).body.length())) + void dump() const; + #define ASSERT_STRING_INVARIANT(string) \ + assert((string).langs.invariant((string).body.length())) public: + static const String Empty; + explicit String(const char* cstr=0, size_t helper_length=0, bool tainted=false); explicit String(const C cstr, bool tainted=false); String(Body abody, Language alang): body(abody), langs(alang) { - assert(!body.is_empty()); ASSERT_STRING_INVARIANT(*this); } String(const String& src): body(src.body), langs(src.langs) { @@ -334,6 +419,7 @@ public: bool is_empty() const { return body.is_empty(); } 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, @@ -396,6 +482,7 @@ public: /// extracts [start, finish) piece of string String& mid(size_t substr_begin, size_t substr_end) const; + String& mid(Charset& charset, size_t from, size_t to, size_t helper_length=0) const; /** ignore lang if it's L_UNSPECIFIED @@ -411,6 +498,9 @@ public: size_t this_offset=0) const { return body.pos(c, this_offset); } + size_t pos(Charset& charset, + const String& substr, + size_t this_offset=0, Language lang=L_UNSPECIFIED) const; void split(ArrayString& result, size_t& pos_after, @@ -432,11 +522,9 @@ public: prematch/match/postmatch/1/2/3/... @endverbatim */ - Table* match(Charset& source_charset, - const String& regexp, - const String* options, + Table* match(VRegex* vregex, Row_action row_action, void *info, - bool& just_matched) const; + int& matches_count) const; enum Change_case_kind { CC_UPPER, CC_LOWER @@ -444,8 +532,11 @@ public: 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; + const String& trim(Trim_kind kind=TRIM_BOTH, const char* chars=0) const; + double as_double() const { return pa_atod(cstr(), this); } + int as_int() const { return pa_atoi(cstr(), this); } + bool as_bool() const { return as_int()!=0; } + const String& escape(Charset& source_charset) const; private: //disabled