|
|
| version 1.84, 2001/04/26 14:55:25 | version 1.144.2.28.2.16, 2003/03/20 16:57:43 |
|---|---|
| Line 1 | Line 1 |
| /** @file | /** @file |
| Parser: string class decl. | Parser: string class decl. |
| Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com) | Copyright (c) 2001-2003 ArtLebedev Group (http://www.artlebedev.com) |
| Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru) | |
| Author: Alexander Petrosyan <paf@design.ru> (http://design.ru/paf) | |
| $Id$ | |
| */ | */ |
| #ifndef PA_STRING_H | #ifndef PA_STRING_H |
| #define PA_STRING_H | #define PA_STRING_H |
| #include "pa_config_includes.h" | static const char* IDENT_STRING_H="$Date$"; |
| #include <string.h> | // includes |
| #include <stddef.h> | |
| #include "pa_pool.h" | |
| #include "pa_types.h" | #include "pa_types.h" |
| #include "pa_array.h" | |
| class Table; | extern "C" { // cord's author forgot to do that |
| #include "cord.h" | |
| /** | }; |
| $MAIN:html-typo table elements must enlarge string not more that that | |
| that's a tradeoff - otherwise we'd have to scan string twice: | |
| - first for buffer length | |
| - second for replacements themselves | |
| */ | |
| #define UNTAINT_TIMES_BIGGER 10 | |
| #ifndef NO_STRING_ORIGIN | // forwards |
| # define STRING_APPEND_PARAMS \ | |
| const char *src, size_t size, \ | |
| String::Untaint_lang lang, \ | |
| const char *file, uint line | |
| /// appends piece to String @see String::real_append | |
| # define APPEND(src, size, lang, file, line) \ | |
| real_append(src, size, lang, file, line) | |
| #else | |
| # define STRING_APPEND_PARAMS \ | |
| const char *src, \ | |
| size_t size, \ | |
| String::Untaint_lang lang | |
| /// appends piece to String @see String::real_append | |
| # define APPEND(src, size, lang, file, line) \ | |
| real_append(src, size, lang) | |
| #endif | |
| /// appends clean piece to String @see String::real_append | |
| #define APPEND_CLEAN(src, size, file, line) \ | |
| APPEND(src, size, String::UL_CLEAN, file, line) | |
| /// appends piece to String as-is @see String::real_append | |
| #define APPEND_AS_IS(src, size, file, line) \ | |
| APPEND(src, size, String::UL_AS_IS, file, line) | |
| /// appends tainted piece to String @see String::real_append | |
| #define APPEND_TAINTED(src, size, file, line) \ | |
| APPEND(src, size, String::UL_TAINTED, file, line) | |
| /// handy: appends const char* piece to String @see String::real_append | |
| #define APPEND_CONST(src) APPEND_AS_IS(src, 0, 0, 0) | |
| class Array; | class Charset; |
| class Table; | |
| class SQL_Connection; | class SQL_Connection; |
| class Dictionary; | |
| class Request_charsets; | |
| class String; | |
| typedef Array<const String* > ArrayString; | |
| // helpers | |
| /* | |
| /// appends clean piece to String @see String::append | |
| #define APPEND_CLEAN(src, length) append(src, length, String::UL_CLEAN) | |
| /// appends piece to String as-is @see String::append | |
| #define APPEND_AS_IS(src, length) append(src, length, String::UL_AS_IS) | |
| /// appends tainted piece to String @see String::append | |
| #define APPEND_TAINTED(src, length) append(src, length, String::UL_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) | |
| /** | |
| Pooled string. | |
| Internal structure: | /** |
| @verbatim | String which knows the language of all it's fragments. |
| String Chunk0 | |
| ====== ======== | |
| head--------------->[ptr, size, ...] | |
| append_here-------->[ptr, size, ...] | |
| . | |
| . | |
| [ptr, size, ...] | |
| link_row----------->[link to the next chunk] | |
| @endverbatim | |
| All pieces remember | All pieces remember |
| - the file and its line they are from [can be turned off by NO_STRING_ORIGIN] | |
| - whether they are tainted or not, | - whether they are tainted or not, |
| and the language which should be used to detaint them | and the language which should be used to detaint them |
| */ | */ |
| class String : public Pooled { | class String: public PA_Object { |
| public: | |
| enum { | public: |
| CR_PREALLOCATED_COUNT=5, ///< default preallocated item count | |
| CR_GROW_PERCENT=60 ///< each time the Array chunk_is_full() array expanded() | |
| }; | |
| /// piece is tainted or not. the language to use when detaint | /** piece is tainted or not. the language to use when detaint |
| remember to change String_Untaint_lang_name @ untaint.C along | |
| */ | |
| enum Untaint_lang { | enum Untaint_lang { |
| UL_UNSPECIFIED=0, ///< zero value handy for hash lookup @see untaint_lang_name2enum | 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_CLEAN, ///< clean |
| UL_TAINTED, ///< tainted, untaint language as assigned later | UL_AS_IS, ///< leave all characters intact |
| // untaint languages. assigned by ^untaint[lang]{...} | |
| UL_PASS_APPENDED, | UL_PASS_APPENDED, |
| /**< | /**< |
| leave language built into string being appended. | leave language built into string being appended. |
| just a flag, that value not stored | just a flag, that value not stored |
| */ | */ |
| UL_AS_IS, ///< leave all characters intact | UL_TAINTED, ///< tainted, untaint language as assigned later |
| UL_FILE_NAME, ///< filename | // untaint languages. assigned by ^untaint[lang]{...} |
| UL_FILE_SPEC, ///< file specification | |
| UL_HTTP_HEADER, ///< text in HTTP response header | UL_HTTP_HEADER, ///< text in HTTP response header |
| UL_MAIL_HEADER, ///< text in mail header | UL_MAIL_HEADER, ///< text in mail header |
| UL_URI, ///< text in uri | UL_URI, ///< text in uri |
| UL_TABLE, ///< ^table:set body | UL_TABLE, ///< ^table:set body |
| UL_SQL, ///< ^table:sql body | UL_SQL, ///< ^table:sql body |
| UL_JS, ///< JavaScript code | UL_JS, ///< JavaScript code |
| UL_XML, ///< ^dom:set xml | |
| UL_HTML, ///< HTML code (for editing) | UL_HTML, ///< HTML code (for editing) |
| UL_HTML_TYPO ///< HTML code with TYPOgraphic replacements (for showing) | UL_OPTIMIZE_BIT = 0x8000 ///< flag, requiring cstr whitespace optimization |
| }; | |
| struct Fragment { | |
| Untaint_lang lang; ///< untaint flag, later untaint language | |
| size_t length; ///< length | |
| Fragment(Untaint_lang alang, size_t asize): lang(alang), length(asize) {} | |
| }; | |
| class ArrayFragment: public Array<Fragment> { | |
| void append(element_type src) { | |
| *static_cast<Array<Fragment> *>(this)+=src; | |
| } | |
| /// hiding from accidental USE, use append_positions | |
| void append(const ArrayFragment& src, int offset, int limit) { | |
| static_cast<Array<Fragment> *>(this)->append(src, offset, limit); | |
| } | |
| public: | |
| ArrayFragment& operator += (element_type src) { | |
| if(count()) { // not empty? | |
| // try to join with last | |
| Fragment& last=get(count()-1); | |
| if(last.lang==src.lang) { | |
| last.length+=src.length; | |
| return *this; | |
| } | |
| } | |
| append(src); | |
| return *this; | |
| } | |
| 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 { | |
| char *str; | |
| size_t length; | |
| operator char *() { return str; } | |
| C(char *astr, size_t asize): str(astr), length(asize) {} | |
| }; | }; |
| private: | |
| CORD body; ///< all characters of string | |
| ArrayFragment fragments; ///< fragment language+length info | |
| public: | public: |
| String(Pool& apool, const char *src=0, size_t src_size=0, bool tainted=false); | /// @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); | String(const String& src); |
| size_t size() const { return fsize; } | operator bool() const { return body!=CORD_EMPTY; } |
| /// convert to C string. if 'lang' known, forcing 'lang' to it | size_t length() const { return CORD_len(body); } |
| char *cstr(Untaint_lang lang=UL_UNSPECIFIED, | |
| /// convert to CORD. if 'lang' known, forcing 'lang' to it | |
| CORD cstr_to_cord(Untaint_lang lang=UL_AS_IS, | |
| SQL_Connection *connection=0, | SQL_Connection *connection=0, |
| const char *charset=0) const { | const Request_charsets *charsets=0) const; |
| char *result=(char *)malloc(size()*UNTAINT_TIMES_BIGGER+1); | |
| char *eol=store_to(result, lang, connection, charset); | /// convert to constant C string. if 'lang' known, forcing 'lang' to it |
| *eol=0; | const char* cstr(Untaint_lang lang=UL_AS_IS, |
| return result; | SQL_Connection *connection=0, |
| const Request_charsets *charsets=0) const { | |
| return CORD_to_const_char_star(cstr_to_cord(lang, connection, charsets)); | |
| } | } |
| /** append fragment | /// convert to Modifiable C string. if 'lang' known, forcing 'lang' to it |
| @see APPEND_AS_IS, APPEND_CLEAN, APPEND_TAINTED, APPEND_CONST | char *cstrm(Untaint_lang lang=UL_AS_IS, |
| */ | SQL_Connection *connection=0, |
| String& real_append(STRING_APPEND_PARAMS); | const Request_charsets *charsets=0) const { |
| /// @return <0 ==0 or >0 depending on comparison result | return CORD_to_char_star(cstr_to_cord(lang, connection, charsets)); |
| int cmp (int& partial, const String& src, | |
| size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const; | |
| bool operator < (const String& src) const { int p; return cmp(p, src)<0; } | |
| bool operator > (const String& src) const { int p; return cmp(p, src)>0; } | |
| bool operator <= (const String& src) const { int p; return cmp(p, src)<=0; } | |
| bool operator >= (const String& src) const { int p; return cmp(p, src)>=0; } | |
| bool operator == (const String& src) const { | |
| if(size()!=src.size()) // can speed up in trivial case | |
| return false; | |
| int p; return cmp(p, src)==0; | |
| } | } |
| bool operator != (const String& src) const { int p; return cmp(p, src)!=0; } | /// puts pieces to buf |
| C serialize(size_t prolog_size) const; | |
| /** | /// appends pieces from buf to self |
| @param partial | bool deserialize(size_t prolog_size, void *buf, size_t buf_size, const char* file); |
| returns partial match status. | /** append fragment, |
| - -1: strings too different | @c helper_length can be zero, it will be autocalced |
| - 0: full match | when length<strlen(str) makes heap copy of @c str piece |
| - 1: means @c this starts @c src | |
| - 2: means @src starts @this | |
| */ | */ |
| int cmp(int& partial, const char* src_ptr, size_t src_size=0, | String& append(const char* str, size_t helper_length, Untaint_lang lang); |
| size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const; | |
| bool operator == (const char* src_ptr) const { | bool operator < (const String& src) const { return CORD_cmp(body, src.body)<0; } |
| size_t src_size=src_ptr?strlen(src_ptr):0; | bool operator > (const String& src) const { return CORD_cmp(body, src.body)>0; } |
| if(size() != src_size) | bool operator <= (const String& src) const { return CORD_cmp(body, src.body)<=0; } |
| return false; | bool operator >= (const String& src) const { return CORD_cmp(body, src.body)>=0; } |
| int partial; // unused | bool operator == (const String& src) const { return CORD_cmp(body, src.body)==0; } |
| return cmp(partial, src_ptr, src_size)==0; | bool operator != (const String& src) const { return CORD_cmp(body, src.body)!=0; } |
| /// 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 operator != (const char* src_ptr) const { | /// y starts with this |
| int partial; // unused | bool this_starts(CORD y) const { |
| return cmp(partial, src_ptr, 0)!=0; | return CORD_ncmp(body, 0/*x_begin*/, y, 0/*y_begin*/, CORD_len(body))==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) { |
| appends other String. | return src.append_to(*this, lang, forced); |
| } | |
| marking all tainted pieces of it with @a lang. | |
| or marking ALL pieces of it with a @a lang when @a forced to. | |
| */ | |
| String& append(const String& src, Untaint_lang lang, bool forced=false); | |
| String& operator << (const String& src) { return append(src, UL_PASS_APPENDED); } | String& operator << (const String& src) { return append(src, UL_PASS_APPENDED); } |
| String& operator << (const char *src) { return APPEND_CONST(src); } | String& operator << (const char* src) { return append(src, 0, UL_AS_IS); } |
| /// simple hash code of string. used by Hash | /// simple hash code of string |
| uint hash_code() const; | uint hash_code() const; |
| /// extracts first char of a string, if any | |
| char first_char() const { | |
| return operator bool()?CORD_fetch(body, 0):0; | |
| } | |
| /// extracts [start, finish) piece of string | /// extracts [start, finish) piece of string |
| String& mid(size_t start, size_t finish) const; | const String& mid(size_t substr_begin, size_t substr_end) const; |
| /// @return position of substr in string, -1 means "not found" [String version] | /** |
| int pos(const String& substr, | ignore lang if it's UL_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; | size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const; |
| /// @return position of substr in string, -1 means "not found" [const char* version] | /// String version of @see pos(const char*, int, Untaint_lang) |
| int pos(const char *substr, size_t substr_size, | size_t pos(const String& substr, |
| size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const; | size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const; |
| void split(Array& result, | void split(ArrayString& result, |
| size_t *pos_after_ref, | size_t& pos_after, |
| const char *delim, size_t delim_size, | const char* delim, |
| Untaint_lang lang, int limit=-1) const; | Untaint_lang lang=UL_UNSPECIFIED, int limit=-1); |
| void split(Array& result, | void split(ArrayString& result, |
| size_t *pos_after_ref, | size_t& pos_after, |
| const String& delim, | const String& delim, |
| Untaint_lang lang, int limit=-1) const; | Untaint_lang lang=UL_UNSPECIFIED, int limit=-1) const; |
| typedef void (*Row_action)(Table& table, Array *row, int start, int finish, | typedef void (*Row_action)(Table* table, ArrayString* row, |
| int prestart, int prefinish, | |
| int poststart, int postfinish, | |
| void *info); | void *info); |
| bool match(const unsigned char *pcre_tables, | /** |
| const String *aorigin, | @return table of found items, if any. |
| const String& regexp, | table format is defined and fixed[can be used by others]: |
| const String *options, | @verbatim |
| Table **table, | prematch/match/postmatch/1/2/3/... |
| Row_action row_action, void *info) const; | @endverbatim |
| #ifndef NO_STRING_ORIGIN | |
| /// origin of string. calculated by first row | |
| const Origin& origin() const; | |
| #endif | |
| private: | |
| /// several String fragments | |
| struct Chunk { | |
| size_t count; ///< the number of rows in chunk | |
| /// string fragment or a link to next chunk union | |
| union Row { | |
| /// fragment | |
| struct { | |
| const char *ptr; ///< pointer to the start | |
| size_t size; ///< length | |
| Untaint_lang lang; ///< untaint flag, later untaint language | |
| #ifndef NO_STRING_ORIGIN | |
| Origin origin; ///< origin | |
| #endif | |
| } item; | |
| Chunk *link; ///< link to the next chunk in chain | |
| } rows[CR_PREALLOCATED_COUNT]; | |
| Chunk *preallocated_link; ///< next rows are here | |
| } | |
| head; ///< the head chunk of the chunk chain | |
| /// next append would write to this record | |
| Chunk::Row *append_here; | |
| /** the address of place where lies address | |
| of the link to the next chunk to allocate | |
| */ | */ |
| Chunk::Row *link_row; | Table* match(Charset& source_charset, |
| const String& regexp, | |
| private: | const String& options, |
| // last chunk | Row_action row_action, void *info, |
| Chunk *last_chunk; | bool& just_matched) const; |
| enum Change_case_kind { | |
| // string size | CC_UPPER, |
| size_t fsize; | CC_LOWER |
| }; | |
| // used rows in all chunks | String& change_case(Charset& source_charset, |
| int fused_rows; | Change_case_kind kind); |
| const String& replace(const Dictionary& dict) const; | |
| private: | double as_double() const; |
| int as_int() const; | |
| bool chunk_is_full() { | |
| return append_here == link_row; | |
| } | |
| void expand(); | |
| /// convert to C string, store to 'dest' which must be big enough for proper untaint | |
| char *store_to(char *dest, Untaint_lang lang=UL_UNSPECIFIED, | |
| SQL_Connection *connection=0, | |
| const char *charset=0) const; | |
| private: //disabled | private: //disabled |
| Line 271 private: //disabled | Line 249 private: //disabled |
| }; | }; |
| /// simple hash code of string. used by Hash | |
| inline uint hash_code(const String& self) { | |
| return self.hash_code(); | |
| } | |
| #endif | #endif |