Annotation of parser3/src/include/pa_string.h, revision 1.134
1.41 paf 1: /** @file
1.43 paf 2: Parser: string class decl.
3:
1.123 paf 4: Copyright (c) 2001, 2002 ArtLebedev Group (http://www.artlebedev.com)
1.124 paf 5: Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
1.29 paf 6:
1.134 ! paf 7: $Id: pa_string.h,v 1.133 2002/04/01 09:37:50 paf Exp $
1.1 paf 8: */
9:
10: #ifndef PA_STRING_H
11: #define PA_STRING_H
12:
1.15 paf 13: #include "pa_pool.h"
1.4 paf 14: #include "pa_types.h"
15:
1.63 paf 16: class Table;
17:
1.9 paf 18: #ifndef NO_STRING_ORIGIN
1.33 paf 19: # define STRING_APPEND_PARAMS \
20: const char *src, size_t size, \
1.119 paf 21: uchar lang, \
1.33 paf 22: const char *file, uint line
1.54 paf 23: /// appends piece to String @see String::real_append
24: # define APPEND(src, size, lang, file, line) \
25: real_append(src, size, lang, file, line)
1.9 paf 26: #else
1.33 paf 27: # define STRING_APPEND_PARAMS \
28: const char *src, \
29: size_t size, \
1.119 paf 30: uchar lang
1.54 paf 31: /// appends piece to String @see String::real_append
32: # define APPEND(src, size, lang, file, line) \
33: real_append(src, size, lang)
34: #endif
1.83 paf 35: /// appends clean piece to String @see String::real_append
1.54 paf 36: #define APPEND_CLEAN(src, size, file, line) \
1.62 paf 37: APPEND(src, size, String::UL_CLEAN, file, line)
1.83 paf 38: /// appends piece to String as-is @see String::real_append
39: #define APPEND_AS_IS(src, size, file, line) \
40: APPEND(src, size, String::UL_AS_IS, file, line)
1.48 paf 41: /// appends tainted piece to String @see String::real_append
1.54 paf 42: #define APPEND_TAINTED(src, size, file, line) \
1.62 paf 43: APPEND(src, size, String::UL_TAINTED, file, line)
1.48 paf 44: /// handy: appends const char* piece to String @see String::real_append
1.83 paf 45: #define APPEND_CONST(src) APPEND_AS_IS(src, 0, 0, 0)
1.9 paf 46:
1.62 paf 47: class Array;
1.71 paf 48: class SQL_Connection;
1.101 parser 49: class Dictionary;
1.62 paf 50:
1.42 paf 51: /**
1.41 paf 52: Pooled string.
53:
1.47 paf 54: Internal structure:
55: @verbatim
56: String Chunk0
57: ====== ========
58: head--------------->[ptr, size, ...]
59: append_here-------->[ptr, size, ...]
60: .
61: .
62: [ptr, size, ...]
63: link_row----------->[link to the next chunk]
1.41 paf 64: @endverbatim
65:
66: All pieces remember
67: - the file and its line they are from [can be turned off by NO_STRING_ORIGIN]
68: - whether they are tainted or not,
69: and the language which should be used to detaint them
70: */
1.114 paf 71: #include "pa_pragma_pack_begin.h"
1.15 paf 72: class String : public Pooled {
1.1 paf 73: public:
1.48 paf 74:
1.1 paf 75: enum {
1.89 parser 76: CR_PREALLOCATED_COUNT=2, ///< default preallocated item count
1.125 paf 77: CR_GROW_COUNT=1 ///< each time the String chunk_is_full() string expanded()
1.1 paf 78: };
79:
1.106 parser 80: /** piece is tainted or not. the language to use when detaint
81: remember to change String_Untaint_lang_name @ untaint.C along
82: */
1.27 paf 83: enum Untaint_lang {
1.72 paf 84: UL_UNSPECIFIED=0, ///< zero value handy for hash lookup @see untaint_lang_name2enum
1.122 paf 85: // these two must go before others, there are checks for >UL_AS_IS
1.62 paf 86: UL_CLEAN, ///< clean
1.122 paf 87: UL_AS_IS, ///< leave all characters intact
88:
1.48 paf 89: UL_PASS_APPENDED,
1.41 paf 90: /**<
91: leave language built into string being appended.
92: just a flag, that value not stored
93: */
1.122 paf 94: UL_TAINTED, ///< tainted, untaint language as assigned later
95: // untaint languages. assigned by ^untaint[lang]{...}
1.103 parser 96: UL_FILE_SPEC, ///< file specification
1.74 paf 97: UL_HTTP_HEADER, ///< text in HTTP response header
98: UL_MAIL_HEADER, ///< text in mail header
1.48 paf 99: UL_URI, ///< text in uri
100: UL_TABLE, ///< ^table:set body
101: UL_SQL, ///< ^table:sql body
102: UL_JS, ///< JavaScript code
1.102 parser 103: UL_XML, ///< ^dom:set xml
1.48 paf 104: UL_HTML, ///< HTML code (for editing)
1.119 paf 105: UL_OPTIMIZE_BIT = 0x80 ///< flag, requiring cstr whitespace optimization
1.27 paf 106: };
107:
1.8 paf 108: public:
109:
1.77 paf 110: String(Pool& apool, const char *src=0, size_t src_size=0, bool tainted=false);
1.14 paf 111: String(const String& src);
1.126 paf 112: bool is_empty() const { return append_here==head.rows; }
1.117 paf 113: size_t size() const;
1.69 paf 114: /// convert to C string. if 'lang' known, forcing 'lang' to it
1.109 paf 115: char *cstr(Untaint_lang lang=UL_AS_IS,
1.82 paf 116: SQL_Connection *connection=0,
1.133 paf 117: Charset *cstr_charset=0, const char *cstr_charset_name=0) const {
1.88 parser 118:
1.121 paf 119: char *result=(char *)malloc(cstr_bufsize(lang, connection, cstr_charset));
1.133 paf 120: char *eol=store_to(result, lang, connection, cstr_charset, cstr_charset_name);
1.50 paf 121: *eol=0;
122: return result;
123: }
1.117 paf 124: char *cstr_debug_origins() const;
1.108 parser 125: /// puts pieces to buf
126: void serialize(size_t prolog_size, void *& buf, size_t& buf_size) const;
127: /// appends pieces from buf to self
1.132 paf 128: bool deserialize(size_t prolog_size, void *buf, size_t buf_size, const char *file);
1.46 paf 129: /** append fragment
1.83 paf 130: @see APPEND_AS_IS, APPEND_CLEAN, APPEND_TAINTED, APPEND_CONST
1.46 paf 131: */
1.9 paf 132: String& real_append(STRING_APPEND_PARAMS);
1.44 paf 133: /// @return <0 ==0 or >0 depending on comparison result
1.62 paf 134: int cmp (int& partial, const String& src,
1.72 paf 135: size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.56 paf 136: bool operator < (const String& src) const { int p; return cmp(p, src)<0; }
137: bool operator > (const String& src) const { int p; return cmp(p, src)>0; }
138: bool operator <= (const String& src) const { int p; return cmp(p, src)<=0; }
139: bool operator >= (const String& src) const { int p; return cmp(p, src)>=0; }
1.26 paf 140: bool operator == (const String& src) const {
141: if(size()!=src.size()) // can speed up in trivial case
142: return false;
1.56 paf 143: int p; return cmp(p, src)==0;
1.26 paf 144: }
1.56 paf 145: bool operator != (const String& src) const { int p; return cmp(p, src)!=0; }
1.26 paf 146:
1.50 paf 147: /**
148: @param partial
149: returns partial match status.
1.51 paf 150: - -1: strings too different
151: - 0: full match
152: - 1: means @c this starts @c src
153: - 2: means @src starts @this
1.50 paf 154: */
1.61 paf 155: int cmp(int& partial, const char* src_ptr, size_t src_size=0,
1.72 paf 156: size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.50 paf 157: bool operator == (const char* src_ptr) const {
158: size_t src_size=src_ptr?strlen(src_ptr):0;
159: if(size() != src_size)
160: return false;
161: int partial; // unused
1.56 paf 162: return cmp(partial, src_ptr, src_size)==0;
1.50 paf 163: }
1.80 paf 164: bool operator != (const char* src_ptr) const {
165: int partial; // unused
166: return cmp(partial, src_ptr, 0)!=0;
167: }
168:
1.119 paf 169: String& append(const String& src, uchar lang, bool forced=false);
1.76 paf 170: String& operator << (const String& src) { return append(src, UL_PASS_APPENDED); }
171: String& operator << (const char *src) { return APPEND_CONST(src); }
1.8 paf 172:
1.41 paf 173: /// simple hash code of string. used by Hash
1.14 paf 174: uint hash_code() const;
1.100 parser 175:
176: /// extracts first char of a string
177: char first_char() const;
1.54 paf 178:
179: /// extracts [start, finish) piece of string
1.70 paf 180: String& mid(size_t start, size_t finish) const;
1.55 paf 181:
1.59 paf 182: /// @return position of substr in string, -1 means "not found" [String version]
1.62 paf 183: int pos(const String& substr,
1.112 paf 184: int this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.59 paf 185: /// @return position of substr in string, -1 means "not found" [const char* version]
1.97 parser 186: int pos(const char *substr, size_t substr_size=0,
1.112 paf 187: int this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.62 paf 188:
189: void split(Array& result,
190: size_t *pos_after_ref,
191: const char *delim, size_t delim_size,
1.86 paf 192: Untaint_lang lang=UL_UNSPECIFIED, int limit=-1) const;
1.62 paf 193: void split(Array& result,
194: size_t *pos_after_ref,
195: const String& delim,
1.86 paf 196: Untaint_lang lang=UL_UNSPECIFIED, int limit=-1) const;
1.62 paf 197:
1.68 paf 198: typedef void (*Row_action)(Table& table, Array *row, int start, int finish,
199: void *info);
1.87 parser 200: /**
201: @return true if fills table.
202: table format is defined and fixed[can be used by others]:
203: @verbatim
204: prematch/match/postmatch/1/2/3/...
205: @endverbatim
206: */
1.120 paf 207: bool match(
1.81 paf 208: const String *aorigin,
1.64 paf 209: const String& regexp,
1.65 paf 210: const String *options,
1.66 paf 211: Table **table,
1.99 parser 212: Row_action row_action, void *info,
213: bool *was_global=0) const;
1.87 parser 214: enum Change_case_kind {
215: CC_UPPER,
216: CC_LOWER
217: };
1.120 paf 218: String& change_case(Pool& pool,
1.87 parser 219: Change_case_kind kind) const;
1.101 parser 220: String& replace(Pool& pool, Dictionary& dict) const;
1.96 parser 221: double as_double() const;
222: int as_int() const;
1.13 paf 223:
1.41 paf 224: #ifndef NO_STRING_ORIGIN
225: /// origin of string. calculated by first row
1.43 paf 226: const Origin& origin() const;
1.41 paf 227: #endif
1.8 paf 228:
1.129 paf 229: private:
1.10 paf 230:
1.128 paf 231: /** several String fragments
232:
233: 'mutable' because can write after it's end, after it was appended to somebody
234: @see String::append
235: */
236: mutable struct Chunk {
1.125 paf 237: typedef uchar count_type;
238: count_type count; ///< the number of rows in chunk
1.84 paf 239: /// string fragment or a link to next chunk union
1.1 paf 240: union Row {
1.125 paf 241: typedef uchar item_size_type;
1.84 paf 242: /// fragment
1.27 paf 243: struct {
1.84 paf 244: const char *ptr; ///< pointer to the start
1.125 paf 245: item_size_type size; ///< length
246: uchar/*Untaint_lang*/ lang; ///< untaint flag, later untaint language
1.27 paf 247: #ifndef NO_STRING_ORIGIN
1.84 paf 248: Origin origin; ///< origin
1.27 paf 249: #endif
1.1 paf 250: } item;
1.84 paf 251: Chunk *link; ///< link to the next chunk in chain
1.130 paf 252: } rows[CR_PREALLOCATED_COUNT+1/*for head.rows[CR_PREALLOCATED_COUNT].link*/];
1.1 paf 253: }
1.84 paf 254: head; ///< the head chunk of the chunk chain
1.1 paf 255:
1.84 paf 256: /// next append would write to this record
1.1 paf 257: Chunk::Row *append_here;
258:
1.5 paf 259: private:
1.110 paf 260: /// last chunk
1.125 paf 261: mutable Chunk *last_chunk;
1.5 paf 262:
263: private:
1.1 paf 264:
265: bool chunk_is_full() {
1.127 paf 266: return append_here == last_chunk->rows+last_chunk->count;
1.1 paf 267: }
1.129 paf 268: uint used_rows() const;
1.1 paf 269: void expand();
1.125 paf 270:
271: Untaint_lang lang_of(size_t offset) const;
1.79 paf 272:
1.116 paf 273: size_t cstr_bufsize(Untaint_lang lang,
274: SQL_Connection *connection,
1.121 paf 275: Charset *buf_charset) const;
1.79 paf 276: /// convert to C string, store to 'dest' which must be big enough for proper untaint
1.82 paf 277: char *store_to(char *dest, Untaint_lang lang=UL_UNSPECIFIED,
278: SQL_Connection *connection=0,
1.133 paf 279: Charset *store_to_charset=0,
280: const char *store_to_charset_name=0) const;
1.107 parser 281:
282: void join_chain(Pool& pool,
1.130 paf 283: const Chunk*& achunk, const Chunk::Row*& arow, uint& acountdown,
1.119 paf 284: uchar& joined_lang, const char *& joined_ptr, size_t& joined_size) const;
1.7 paf 285:
286: private: //disabled
287:
1.12 paf 288: String& operator = (const String&) { return *this; }
1.7 paf 289:
1.1 paf 290: };
1.114 paf 291: #include "pa_pragma_pack_end.h"
1.119 paf 292:
1.130 paf 293: #define STRING_PREPARED_FOREACH_ROW(self, body) \
294: while(row!=(self).append_here) { \
295: if(countdown==0) { \
296: chunk=row->link; \
297: row=chunk->rows; \
298: countdown=chunk->count; \
299: }; \
300: { body } \
1.131 paf 301: row++; countdown--; \
1.130 paf 302: }
303:
304: #define STRING_PREFIX_FOREACH_ROW(self, body) { \
305: const Chunk *chunk=&(self).head; \
306: const Chunk::Row *row=chunk->rows; \
307: uint countdown=chunk->count; \
308: STRING_PREPARED_FOREACH_ROW(self, body) \
1.119 paf 309: }
310:
311: #define STRING_FOREACH_ROW(body) STRING_PREFIX_FOREACH_ROW(*this, body)
312: #define STRING_SRC_FOREACH_ROW(body) STRING_PREFIX_FOREACH_ROW(src, body)
1.1 paf 313:
314: #endif
E-mail:
![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to pa_string.h CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [parser3project] / parser3 / src / include