--- parser3/src/include/pa_string.h 2003/09/25 09:15:02 1.146 +++ parser3/src/include/pa_string.h 2003/10/21 05:11:00 1.151 @@ -8,7 +8,7 @@ #ifndef PA_STRING_H #define PA_STRING_H -static const char* IDENT_STRING_H="$Date: 2003/09/25 09:15:02 $"; +static const char* IDENT_STRING_H="$Date: 2003/10/21 05:11:00 $"; // includes @@ -24,6 +24,7 @@ extern "C" { // cord's author forgot to /* 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 @@ -42,6 +43,7 @@ template inline size_t get_length(T current) { return current; } + /** String which knows the lang of all it's langs. @@ -54,53 +56,63 @@ 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_TABLE='L', ///< ^table:set body + L_SQL='Q', ///< ^table:sql body + L_JS='J', ///< JavaScript code + L_XML='X', ///< ^dom:set xml + L_HTML='H', ///< HTML code (for editing) + // READ WARNING ABOVE BEFORE ADDING ANYTHING L_OPTIMIZE_BIT = 0x80 ///< flag, requiring cstr whitespace optimization }; - class Languages { + union Languages { - union { - struct { - Language lang:8; - int is_not_just_lang:sizeof(CORD)*8-8; - }; - CORD langs; - }; + struct { + Language lang:8; + int is_not_just_lang:sizeof(CORD)*8-8; + } opt; + CORD langs; template CORD make_langs(C current) const { - return is_not_just_lang? + return opt.is_not_just_lang? langs - :CORD_chars((char)lang, get_length(current)); + :CORD_chars((char)opt.lang, get_length(current)); } CORD make_langs(size_t aoffset, size_t alength) const { - return is_not_just_lang? + return opt.is_not_just_lang? CORD_substr(langs, aoffset, alength) - :CORD_chars((char)lang, alength); + :CORD_chars((char)opt.lang, alength); } /// appending when 'langs' already contain something [simple cases hanled elsewhere] @@ -109,13 +121,13 @@ public: const CORD to_nonempty_target_langs) { assert(langs); - if(is_not_just_lang) + 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); assert(current_size); langs=CORD_cat( - CORD_chars((char)lang, current_size), // first piece [making from just 'lang'] + CORD_chars((char)opt.lang, current_size), // first piece [making from just 'lang'] to_nonempty_target_langs); // new piece } } @@ -125,7 +137,10 @@ public: const char* v() const; Languages(): langs(0) {} - Languages(Language alang): lang(alang), is_not_just_lang(0) {} + Languages(Language alang) { + opt.lang=alang; + opt.is_not_just_lang=0; + } /// MUST be called exactly prior to modification of current [uses it's length] template @@ -133,12 +148,12 @@ public: assert(alang); assert(asize); - if(!is_not_just_lang) - if(lang) { - if(lang==alang) // same length? ignoring + if(!opt.is_not_just_lang) + if(opt.lang) { + if(opt.lang==alang) // same length? ignoring return; } else { - lang=alang; // to uninitialized + opt.lang=alang; // to uninitialized return; } @@ -153,8 +168,8 @@ public: if(!langs) langs=src.langs; // to uninitialized - else if(!src.is_not_just_lang) - append(current, src.lang, appending_length); // simplifying when simple source + else if(!src.opt.is_not_just_lang) + append(current, src.opt.lang, appending_length); // simplifying when simple source else append(current, src.make_langs(appending_length)); } @@ -166,12 +181,12 @@ public: assert(alength); if(!langs) // to uninitialized? - if(src.is_not_just_lang) + if(src.opt.is_not_just_lang) langs=CORD_substr(src.langs, aoffset, alength); // to uninitialized complex else - lang=src.lang; // to uninitialized simple + opt.lang=src.opt.lang; // to uninitialized simple else - if(!is_not_just_lang && !src.is_not_just_lang && lang==src.lang) // both simple & of same language? + if(!opt.is_not_just_lang && !src.opt.is_not_just_lang && opt.lang==src.opt.lang) // both simple & of same language? return; // ignoring else append(current, src.make_langs(aoffset, alength)); @@ -182,26 +197,36 @@ public: if(alang==L_UNSPECIFIED) // ignore lang? return true; - if(is_not_just_lang) + if(opt.is_not_just_lang) return CORD_range_contains_chr_greater_then(langs, aoffset, alength, (unsigned)alang)==0; else - return lang<=alang; + return opt.lang<=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 { - if(is_not_just_lang) + if(opt.is_not_just_lang) CORD_block_iter(langs, 0, (CORD_block_iter_fn)callback, info); else - callback(lang, get_length(current), info); + callback(opt.lang, get_length(current), info); } bool invariant(size_t current_length) { if(!langs) return current_length==0; - if(is_not_just_lang) + if(opt.is_not_just_lang) return CORD_len(langs)==current_length; return true; // uncheckable, actually } @@ -273,16 +298,24 @@ 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 void for_each( + int (*f1)(char c, I), + int (*f2)(const char* s, I), + I info) const { + 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); } @@ -291,7 +324,6 @@ public: }*/ }; - struct C { const char *str; size_t length; @@ -318,6 +350,8 @@ private: 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) { @@ -379,11 +413,7 @@ public: } 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) { - langs.append(body, L_AS_IS, src.length()); - body<