Annotation of parser3/src/include/pa_string.h, revision 1.71
1.41 paf 1: /** @file
1.43 paf 2: Parser: string class decl.
3:
1.29 paf 4: Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com)
1.43 paf 5:
1.30 paf 6: Author: Alexander Petrosyan <paf@design.ru> (http://design.ru/paf)
1.29 paf 7:
1.71 ! paf 8: $Id: pa_string.h,v 1.70 2001/04/04 10:50:34 paf Exp $
1.1 paf 9: */
10:
11: #ifndef PA_STRING_H
12: #define PA_STRING_H
13:
1.49 paf 14: #include "pa_config_includes.h"
1.9 paf 15:
1.1 paf 16: #include <stddef.h>
17:
1.15 paf 18: #include "pa_pool.h"
1.4 paf 19: #include "pa_types.h"
20:
1.63 paf 21: class Table;
22:
1.50 paf 23: /**
24: $MAIN:html-typo table elements must enlarge string not more that that
25: that's a tradeoff - otherwise we'd have to scan string twice:
26: - first for buffer length
27: - second for replacements themselves
28: */
1.31 paf 29: #define UNTAINT_TIMES_BIGGER 10
30:
1.9 paf 31: #ifndef NO_STRING_ORIGIN
1.33 paf 32: # define STRING_APPEND_PARAMS \
33: const char *src, size_t size, \
1.52 paf 34: String::Untaint_lang lang, \
1.33 paf 35: const char *file, uint line
1.54 paf 36: /// appends piece to String @see String::real_append
37: # define APPEND(src, size, lang, file, line) \
38: real_append(src, size, lang, file, line)
1.9 paf 39: #else
1.33 paf 40: # define STRING_APPEND_PARAMS \
41: const char *src, \
42: size_t size, \
1.52 paf 43: String::Untaint_lang lang
1.54 paf 44: /// appends piece to String @see String::real_append
45: # define APPEND(src, size, lang, file, line) \
46: real_append(src, size, lang)
47: #endif
1.48 paf 48: /// appends clean piece to String @see String::real_append
1.54 paf 49: #define APPEND_CLEAN(src, size, file, line) \
1.62 paf 50: APPEND(src, size, String::UL_CLEAN, file, line)
1.48 paf 51: /// appends tainted piece to String @see String::real_append
1.54 paf 52: #define APPEND_TAINTED(src, size, file, line) \
1.62 paf 53: APPEND(src, size, String::UL_TAINTED, file, line)
1.48 paf 54: /// handy: appends const char* piece to String @see String::real_append
1.54 paf 55: #define APPEND_CONST(src) APPEND_CLEAN(src, 0, 0, 0)
1.9 paf 56:
1.62 paf 57: class Array;
1.71 ! paf 58: class SQL_Connection;
1.62 paf 59:
1.42 paf 60: /**
1.41 paf 61: Pooled string.
62:
1.47 paf 63: Internal structure:
64: @verbatim
65: String Chunk0
66: ====== ========
67: head--------------->[ptr, size, ...]
68: append_here-------->[ptr, size, ...]
69: .
70: .
71: [ptr, size, ...]
72: link_row----------->[link to the next chunk]
1.41 paf 73: @endverbatim
74:
75: All pieces remember
76: - the file and its line they are from [can be turned off by NO_STRING_ORIGIN]
77: - whether they are tainted or not,
78: and the language which should be used to detaint them
79: */
1.15 paf 80: class String : public Pooled {
1.1 paf 81: public:
1.48 paf 82:
1.1 paf 83: enum {
1.41 paf 84: CR_PREALLOCATED_COUNT=5, ///< default preallocated item count
85: CR_GROW_PERCENT=60 ///< each time the Array chunk_is_full() array expanded()
1.1 paf 86: };
87:
1.41 paf 88: /// piece is tainted or not. the language to use when detaint
1.27 paf 89: enum Untaint_lang {
1.60 paf 90: UL_UNKNOWN=0, ///< zero value handy for hash lookup @see untaint_lang_name2enum
1.62 paf 91: UL_CLEAN, ///< clean
92: UL_TAINTED, ///< tainted, untaint language as assigned later
1.27 paf 93: // untaint languages. assigned by ^untaint[lang]{...}
1.48 paf 94: UL_PASS_APPENDED,
1.41 paf 95: /**<
96: leave language built into string being appended.
97: just a flag, that value not stored
98: */
1.48 paf 99: UL_AS_IS, ///< leave all characters intact
100: UL_FILE_NAME, ///< filename
101: UL_HEADER, ///< text in response header
102: UL_URI, ///< text in uri
103: UL_TABLE, ///< ^table:set body
104: UL_SQL, ///< ^table:sql body
105: UL_JS, ///< JavaScript code
106: UL_HTML, ///< HTML code (for editing)
107: UL_HTML_TYPO ///< HTML code with TYPOgraphic replacements (for showing)
1.27 paf 108: };
109:
1.8 paf 110: public:
111:
1.37 paf 112: String(Pool& apool, const char *src=0, bool tainted=false);
1.14 paf 113: String(const String& src);
114: size_t size() const { return fsize; }
1.69 paf 115: /// convert to C string. if 'lang' known, forcing 'lang' to it
1.71 ! paf 116: char *cstr(Untaint_lang lang=UL_UNKNOWN, SQL_Connection *connection=0) const {
1.50 paf 117: char *result=(char *)malloc(size()*UNTAINT_TIMES_BIGGER+1);
1.71 ! paf 118: char *eol=store_to(result, lang, connection);
1.50 paf 119: *eol=0;
120: return result;
121: }
1.46 paf 122: /** append fragment
1.54 paf 123: @see APPEND_CLEAN, APPEND_TAINTED, APPEND_CONST
1.46 paf 124: */
1.9 paf 125: String& real_append(STRING_APPEND_PARAMS);
1.44 paf 126: /// @return <0 ==0 or >0 depending on comparison result
1.62 paf 127: int cmp (int& partial, const String& src,
128: size_t this_offset=0, Untaint_lang lang=UL_UNKNOWN) const;
1.56 paf 129: bool operator < (const String& src) const { int p; return cmp(p, src)<0; }
130: bool operator > (const String& src) const { int p; return cmp(p, src)>0; }
131: bool operator <= (const String& src) const { int p; return cmp(p, src)<=0; }
132: bool operator >= (const String& src) const { int p; return cmp(p, src)>=0; }
1.26 paf 133: bool operator == (const String& src) const {
134: if(size()!=src.size()) // can speed up in trivial case
135: return false;
1.56 paf 136: int p; return cmp(p, src)==0;
1.26 paf 137: }
1.56 paf 138: bool operator != (const String& src) const { int p; return cmp(p, src)!=0; }
1.26 paf 139:
1.50 paf 140: /**
141: @param partial
142: returns partial match status.
1.51 paf 143: - -1: strings too different
144: - 0: full match
145: - 1: means @c this starts @c src
146: - 2: means @src starts @this
1.50 paf 147: */
1.61 paf 148: int cmp(int& partial, const char* src_ptr, size_t src_size=0,
1.62 paf 149: size_t this_offset=0, Untaint_lang lang=UL_UNKNOWN) const;
1.50 paf 150: bool operator == (const char* src_ptr) const {
151: size_t src_size=src_ptr?strlen(src_ptr):0;
152: if(size() != src_size)
153: return false;
154: int partial; // unused
1.56 paf 155: return cmp(partial, src_ptr, src_size)==0;
1.50 paf 156: }
157:
1.42 paf 158: /**
159: appends other String.
1.41 paf 160:
1.47 paf 161: marking all tainted pieces of it with @a lang.
162: or marking ALL pieces of it with a @a lang when @a forced to.
1.41 paf 163: */
1.39 paf 164: String& append(const String& src, Untaint_lang lang, bool forced=false);
1.8 paf 165:
1.41 paf 166: /// simple hash code of string. used by Hash
1.14 paf 167: uint hash_code() const;
1.54 paf 168:
169: /// extracts [start, finish) piece of string
1.70 paf 170: String& mid(size_t start, size_t finish) const;
1.55 paf 171:
1.59 paf 172: /// @return position of substr in string, -1 means "not found" [String version]
1.62 paf 173: int pos(const String& substr,
174: size_t this_offset=0, Untaint_lang lang=UL_UNKNOWN) const;
1.59 paf 175: /// @return position of substr in string, -1 means "not found" [const char* version]
1.62 paf 176: int pos(const char *substr, size_t substr_size,
177: size_t this_offset=0, Untaint_lang lang=UL_UNKNOWN) const;
178:
179: void split(Array& result,
180: size_t *pos_after_ref,
181: const char *delim, size_t delim_size,
182: Untaint_lang lang, int limit=-1) const;
183: void split(Array& result,
184: size_t *pos_after_ref,
185: const String& delim,
186: Untaint_lang lang, int limit=-1) const;
187:
1.68 paf 188: typedef void (*Row_action)(Table& table, Array *row, int start, int finish,
189: void *info);
1.64 paf 190: bool match(const String *aorigin,
191: const String& regexp,
1.65 paf 192: const String *options,
1.66 paf 193: Table **table,
194: Row_action row_action, void *info) const;
1.13 paf 195:
1.41 paf 196: #ifndef NO_STRING_ORIGIN
197: /// origin of string. calculated by first row
1.43 paf 198: const Origin& origin() const;
1.41 paf 199: #endif
1.8 paf 200:
1.10 paf 201: private:
202:
1.1 paf 203: struct Chunk {
1.6 paf 204: // the number of rows in chunk
1.56 paf 205: size_t count;
1.1 paf 206: union Row {
1.27 paf 207: // fragment
208: struct {
209: const char *ptr; // pointer to the start
210: size_t size; // length
211: Untaint_lang lang; // untaint flag, later untaint language
212: #ifndef NO_STRING_ORIGIN
213: Origin origin; // origin
214: #endif
1.1 paf 215: } item;
216: Chunk *link; // link to the next chunk in chain
1.2 paf 217: } rows[CR_PREALLOCATED_COUNT];
1.1 paf 218: // next rows are here
219: Chunk *preallocated_link;
220: }
221: head; // the head chunk of the chunk chain
222:
223: // next append would write to this record
224: Chunk::Row *append_here;
225:
226: // the address of place where lies address
227: // of the link to the next chunk to allocate
228: Chunk::Row *link_row;
229:
1.5 paf 230: private:
1.25 paf 231: // last chunk
232: Chunk *last_chunk;
1.5 paf 233:
234: // string size
235: size_t fsize;
236:
237: // used rows in all chunks
238: int fused_rows;
239:
240: private:
1.1 paf 241:
242: bool chunk_is_full() {
243: return append_here == link_row;
244: }
245: void expand();
1.71 ! paf 246: char *String::store_to(char *dest,
! 247: Untaint_lang lang=UL_UNKNOWN, SQL_Connection *connection=0) const;
1.7 paf 248:
249: private: //disabled
250:
1.12 paf 251: String& operator = (const String&) { return *this; }
1.7 paf 252:
1.1 paf 253: };
254:
255: #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