Annotation of parser3/src/include/pa_string.h, revision 1.137
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:
1.137 ! paf 7: $Id: pa_string.h,v 1.136 2002/04/19 08:28:35 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:
1.9 paf 16: #ifndef NO_STRING_ORIGIN
1.33 paf 17: # define STRING_APPEND_PARAMS \
18: const char *src, size_t size, \
1.119 paf 19: uchar lang, \
1.33 paf 20: const char *file, uint line
1.54 paf 21: /// appends piece to String @see String::real_append
22: # define APPEND(src, size, lang, file, line) \
23: real_append(src, size, lang, file, line)
1.9 paf 24: #else
1.33 paf 25: # define STRING_APPEND_PARAMS \
26: const char *src, \
27: size_t size, \
1.119 paf 28: uchar lang
1.54 paf 29: /// appends piece to String @see String::real_append
30: # define APPEND(src, size, lang, file, line) \
31: real_append(src, size, lang)
32: #endif
1.83 paf 33: /// appends clean piece to String @see String::real_append
1.54 paf 34: #define APPEND_CLEAN(src, size, file, line) \
1.62 paf 35: APPEND(src, size, String::UL_CLEAN, file, line)
1.83 paf 36: /// appends piece to String as-is @see String::real_append
37: #define APPEND_AS_IS(src, size, file, line) \
38: APPEND(src, size, String::UL_AS_IS, file, line)
1.48 paf 39: /// appends tainted piece to String @see String::real_append
1.54 paf 40: #define APPEND_TAINTED(src, size, file, line) \
1.62 paf 41: APPEND(src, size, String::UL_TAINTED, file, line)
1.48 paf 42: /// handy: appends const char* piece to String @see String::real_append
1.83 paf 43: #define APPEND_CONST(src) APPEND_AS_IS(src, 0, 0, 0)
1.9 paf 44:
1.135 paf 45: class Table;
1.62 paf 46: class Array;
1.71 paf 47: class SQL_Connection;
1.101 parser 48: class Dictionary;
1.62 paf 49:
1.42 paf 50: /**
1.41 paf 51: Pooled string.
52:
1.47 paf 53: Internal structure:
54: @verbatim
55: String Chunk0
56: ====== ========
57: head--------------->[ptr, size, ...]
58: append_here-------->[ptr, size, ...]
59: .
60: .
61: [ptr, size, ...]
62: link_row----------->[link to the next chunk]
1.41 paf 63: @endverbatim
64:
65: All pieces remember
66: - the file and its line they are from [can be turned off by NO_STRING_ORIGIN]
67: - whether they are tainted or not,
68: and the language which should be used to detaint them
69: */
1.114 paf 70: #include "pa_pragma_pack_begin.h"
1.15 paf 71: class String : public Pooled {
1.1 paf 72: public:
1.48 paf 73:
1.1 paf 74: enum {
1.89 parser 75: CR_PREALLOCATED_COUNT=2, ///< default preallocated item count
1.125 paf 76: CR_GROW_COUNT=1 ///< each time the String chunk_is_full() string expanded()
1.1 paf 77: };
78:
1.106 parser 79: /** piece is tainted or not. the language to use when detaint
80: remember to change String_Untaint_lang_name @ untaint.C along
81: */
1.27 paf 82: enum Untaint_lang {
1.72 paf 83: UL_UNSPECIFIED=0, ///< zero value handy for hash lookup @see untaint_lang_name2enum
1.122 paf 84: // these two must go before others, there are checks for >UL_AS_IS
1.62 paf 85: UL_CLEAN, ///< clean
1.122 paf 86: UL_AS_IS, ///< leave all characters intact
87:
1.48 paf 88: UL_PASS_APPENDED,
1.41 paf 89: /**<
90: leave language built into string being appended.
91: just a flag, that value not stored
92: */
1.122 paf 93: UL_TAINTED, ///< tainted, untaint language as assigned later
94: // untaint languages. assigned by ^untaint[lang]{...}
1.103 parser 95: UL_FILE_SPEC, ///< file specification
1.74 paf 96: UL_HTTP_HEADER, ///< text in HTTP response header
97: UL_MAIL_HEADER, ///< text in mail header
1.48 paf 98: UL_URI, ///< text in uri
99: UL_TABLE, ///< ^table:set body
100: UL_SQL, ///< ^table:sql body
101: UL_JS, ///< JavaScript code
1.102 parser 102: UL_XML, ///< ^dom:set xml
1.48 paf 103: UL_HTML, ///< HTML code (for editing)
1.119 paf 104: UL_OPTIMIZE_BIT = 0x80 ///< flag, requiring cstr whitespace optimization
1.27 paf 105: };
106:
1.8 paf 107: public:
108:
1.77 paf 109: String(Pool& apool, const char *src=0, size_t src_size=0, bool tainted=false);
1.14 paf 110: String(const String& src);
1.135 paf 111: bool is_empty() const { return append_here==head.chunk.rows; }
1.117 paf 112: size_t size() const;
1.69 paf 113: /// convert to C string. if 'lang' known, forcing 'lang' to it
1.109 paf 114: char *cstr(Untaint_lang lang=UL_AS_IS,
1.82 paf 115: SQL_Connection *connection=0,
1.133 paf 116: Charset *cstr_charset=0, const char *cstr_charset_name=0) const {
1.88 parser 117:
1.121 paf 118: char *result=(char *)malloc(cstr_bufsize(lang, connection, cstr_charset));
1.133 paf 119: char *eol=store_to(result, lang, connection, cstr_charset, cstr_charset_name);
1.50 paf 120: *eol=0;
121: return result;
122: }
1.117 paf 123: char *cstr_debug_origins() const;
1.108 parser 124: /// puts pieces to buf
125: void serialize(size_t prolog_size, void *& buf, size_t& buf_size) const;
126: /// appends pieces from buf to self
1.132 paf 127: bool deserialize(size_t prolog_size, void *buf, size_t buf_size, const char *file);
1.46 paf 128: /** append fragment
1.83 paf 129: @see APPEND_AS_IS, APPEND_CLEAN, APPEND_TAINTED, APPEND_CONST
1.46 paf 130: */
1.9 paf 131: String& real_append(STRING_APPEND_PARAMS);
1.44 paf 132: /// @return <0 ==0 or >0 depending on comparison result
1.62 paf 133: int cmp (int& partial, const String& src,
1.72 paf 134: size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.56 paf 135: bool operator < (const String& src) const { int p; return cmp(p, src)<0; }
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; }
1.26 paf 139: bool operator == (const String& src) const {
140: if(size()!=src.size()) // can speed up in trivial case
141: return false;
1.56 paf 142: int p; return cmp(p, src)==0;
1.26 paf 143: }
1.56 paf 144: bool operator != (const String& src) const { int p; return cmp(p, src)!=0; }
1.26 paf 145:
1.50 paf 146: /**
147: @param partial
148: returns partial match status.
1.51 paf 149: - -1: strings too different
150: - 0: full match
151: - 1: means @c this starts @c src
152: - 2: means @src starts @this
1.50 paf 153: */
1.61 paf 154: int cmp(int& partial, const char* src_ptr, size_t src_size=0,
1.72 paf 155: size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.50 paf 156: bool operator == (const char* src_ptr) const {
157: size_t src_size=src_ptr?strlen(src_ptr):0;
158: if(size() != src_size)
159: return false;
160: int partial; // unused
1.56 paf 161: return cmp(partial, src_ptr, src_size)==0;
1.50 paf 162: }
1.80 paf 163: bool operator != (const char* src_ptr) const {
164: int partial; // unused
165: return cmp(partial, src_ptr, 0)!=0;
166: }
167:
1.119 paf 168: String& append(const String& src, uchar lang, bool forced=false);
1.76 paf 169: String& operator << (const String& src) { return append(src, UL_PASS_APPENDED); }
170: String& operator << (const char *src) { return APPEND_CONST(src); }
1.8 paf 171:
1.41 paf 172: /// simple hash code of string. used by Hash
1.14 paf 173: uint hash_code() const;
1.100 parser 174:
175: /// extracts first char of a string
176: char first_char() const;
1.54 paf 177:
178: /// extracts [start, finish) piece of string
1.70 paf 179: String& mid(size_t start, size_t finish) const;
1.55 paf 180:
1.59 paf 181: /// @return position of substr in string, -1 means "not found" [String version]
1.62 paf 182: int pos(const String& substr,
1.112 paf 183: int this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.59 paf 184: /// @return position of substr in string, -1 means "not found" [const char* version]
1.97 parser 185: int pos(const char *substr, size_t substr_size=0,
1.112 paf 186: int this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.62 paf 187:
188: void split(Array& result,
189: size_t *pos_after_ref,
190: const char *delim, size_t delim_size,
1.86 paf 191: Untaint_lang lang=UL_UNSPECIFIED, int limit=-1) const;
1.62 paf 192: void split(Array& result,
193: size_t *pos_after_ref,
194: const String& delim,
1.86 paf 195: Untaint_lang lang=UL_UNSPECIFIED, int limit=-1) const;
1.62 paf 196:
1.136 paf 197: typedef void (*Row_action)(Table& table, Array *row,
198: int prestart, int prefinish,
199: int poststart, int postfinish,
1.68 paf 200: void *info);
1.87 parser 201: /**
202: @return true if fills table.
203: table format is defined and fixed[can be used by others]:
204: @verbatim
205: prematch/match/postmatch/1/2/3/...
206: @endverbatim
207: */
1.137 ! paf 208: bool match(const char *acstr,
1.81 paf 209: const String *aorigin,
1.64 paf 210: const String& regexp,
1.65 paf 211: const String *options,
1.66 paf 212: Table **table,
1.99 parser 213: Row_action row_action, void *info,
214: bool *was_global=0) const;
1.87 parser 215: enum Change_case_kind {
216: CC_UPPER,
217: CC_LOWER
218: };
1.120 paf 219: String& change_case(Pool& pool,
1.87 parser 220: Change_case_kind kind) const;
1.101 parser 221: String& replace(Pool& pool, Dictionary& dict) const;
1.96 parser 222: double as_double() const;
223: int as_int() const;
1.137 ! paf 224:
! 225: bool is_join_chains_profitable() const;
! 226: String& join_chains(Pool& pool, char** cstr) const;
1.13 paf 227:
1.41 paf 228: #ifndef NO_STRING_ORIGIN
229: /// origin of string. calculated by first row
1.43 paf 230: const Origin& origin() const;
1.41 paf 231: #endif
1.8 paf 232:
1.129 paf 233: private:
1.10 paf 234:
1.128 paf 235: /** several String fragments
236: */
1.135 paf 237: struct Chunk {
1.125 paf 238: typedef uchar count_type;
239: count_type count; ///< the number of rows in chunk
1.135 paf 240: // here could be some padding bytes
1.84 paf 241: /// string fragment or a link to next chunk union
1.135 paf 242: typedef union Row {
1.125 paf 243: typedef uchar item_size_type;
1.84 paf 244: /// fragment
1.27 paf 245: struct {
1.84 paf 246: const char *ptr; ///< pointer to the start
1.125 paf 247: item_size_type size; ///< length
248: uchar/*Untaint_lang*/ lang; ///< untaint flag, later untaint language
1.27 paf 249: #ifndef NO_STRING_ORIGIN
1.84 paf 250: Origin origin; ///< origin
1.27 paf 251: #endif
1.1 paf 252: } item;
1.135 paf 253: // we are using the fact that there's no padding before this field!
1.84 paf 254: Chunk *link; ///< link to the next chunk in chain
1.135 paf 255: } rows_type[CR_PREALLOCATED_COUNT];
256: rows_type rows;
257: };
258: /**
259: 'mutable' because can write after it's end, after it was appended to somebody
260: @see String::append
261: */
262: mutable struct {
263: Chunk chunk;
264: Chunk *link_storage;
265: } head; ///< the head chunk of the chunk chain
1.1 paf 266:
1.84 paf 267: /// next append would write to this record
1.1 paf 268: Chunk::Row *append_here;
269:
1.5 paf 270: private:
1.110 paf 271: /// last chunk
1.125 paf 272: mutable Chunk *last_chunk;
1.5 paf 273:
274: private:
1.1 paf 275:
276: bool chunk_is_full() {
1.127 paf 277: return append_here == last_chunk->rows+last_chunk->count;
1.1 paf 278: }
1.129 paf 279: uint used_rows() const;
1.1 paf 280: void expand();
1.125 paf 281:
282: Untaint_lang lang_of(size_t offset) const;
1.79 paf 283:
1.116 paf 284: size_t cstr_bufsize(Untaint_lang lang,
285: SQL_Connection *connection,
1.121 paf 286: Charset *buf_charset) const;
1.79 paf 287: /// convert to C string, store to 'dest' which must be big enough for proper untaint
1.82 paf 288: char *store_to(char *dest, Untaint_lang lang=UL_UNSPECIFIED,
289: SQL_Connection *connection=0,
1.133 paf 290: Charset *store_to_charset=0,
291: const char *store_to_charset_name=0) const;
1.107 parser 292:
293: void join_chain(Pool& pool,
1.130 paf 294: const Chunk*& achunk, const Chunk::Row*& arow, uint& acountdown,
1.119 paf 295: uchar& joined_lang, const char *& joined_ptr, size_t& joined_size) const;
1.7 paf 296:
297: private: //disabled
298:
1.12 paf 299: String& operator = (const String&) { return *this; }
1.7 paf 300:
1.1 paf 301: };
1.114 paf 302: #include "pa_pragma_pack_end.h"
1.119 paf 303:
1.130 paf 304: #define STRING_PREPARED_FOREACH_ROW(self, body) \
305: while(row!=(self).append_here) { \
306: if(countdown==0) { \
307: chunk=row->link; \
308: row=chunk->rows; \
309: countdown=chunk->count; \
310: }; \
311: { body } \
1.131 paf 312: row++; countdown--; \
1.130 paf 313: }
314:
315: #define STRING_PREFIX_FOREACH_ROW(self, body) { \
1.135 paf 316: const Chunk *chunk=&(self).head.chunk; \
1.130 paf 317: const Chunk::Row *row=chunk->rows; \
318: uint countdown=chunk->count; \
319: STRING_PREPARED_FOREACH_ROW(self, body) \
1.119 paf 320: }
321:
322: #define STRING_FOREACH_ROW(body) STRING_PREFIX_FOREACH_ROW(*this, body)
323: #define STRING_SRC_FOREACH_ROW(body) STRING_PREFIX_FOREACH_ROW(src, body)
1.1 paf 324:
325: #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