Annotation of parser3/src/include/pa_string.h, revision 1.74
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:
1.74 ! paf 8: $Id: pa_string.h,v 1.73 2001/04/05 16:30:41 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:
1.73 paf 16: #include <string.h>
1.1 paf 17: #include <stddef.h>
18:
1.15 paf 19: #include "pa_pool.h"
1.4 paf 20: #include "pa_types.h"
21:
1.63 paf 22: class Table;
23:
1.50 paf 24: /**
25: $MAIN:html-typo table elements must enlarge string not more that that
26: that's a tradeoff - otherwise we'd have to scan string twice:
27: - first for buffer length
28: - second for replacements themselves
29: */
1.31 paf 30: #define UNTAINT_TIMES_BIGGER 10
31:
1.9 paf 32: #ifndef NO_STRING_ORIGIN
1.33 paf 33: # define STRING_APPEND_PARAMS \
34: const char *src, size_t size, \
1.52 paf 35: String::Untaint_lang lang, \
1.33 paf 36: const char *file, uint line
1.54 paf 37: /// appends piece to String @see String::real_append
38: # define APPEND(src, size, lang, file, line) \
39: real_append(src, size, lang, file, line)
1.9 paf 40: #else
1.33 paf 41: # define STRING_APPEND_PARAMS \
42: const char *src, \
43: size_t size, \
1.52 paf 44: String::Untaint_lang lang
1.54 paf 45: /// appends piece to String @see String::real_append
46: # define APPEND(src, size, lang, file, line) \
47: real_append(src, size, lang)
48: #endif
1.48 paf 49: /// appends clean piece to String @see String::real_append
1.54 paf 50: #define APPEND_CLEAN(src, size, file, line) \
1.62 paf 51: APPEND(src, size, String::UL_CLEAN, file, line)
1.48 paf 52: /// appends tainted piece to String @see String::real_append
1.54 paf 53: #define APPEND_TAINTED(src, size, file, line) \
1.62 paf 54: APPEND(src, size, String::UL_TAINTED, file, line)
1.48 paf 55: /// handy: appends const char* piece to String @see String::real_append
1.54 paf 56: #define APPEND_CONST(src) APPEND_CLEAN(src, 0, 0, 0)
1.9 paf 57:
1.62 paf 58: class Array;
1.71 paf 59: class SQL_Connection;
1.62 paf 60:
1.42 paf 61: /**
1.41 paf 62: Pooled string.
63:
1.47 paf 64: Internal structure:
65: @verbatim
66: String Chunk0
67: ====== ========
68: head--------------->[ptr, size, ...]
69: append_here-------->[ptr, size, ...]
70: .
71: .
72: [ptr, size, ...]
73: link_row----------->[link to the next chunk]
1.41 paf 74: @endverbatim
75:
76: All pieces remember
77: - the file and its line they are from [can be turned off by NO_STRING_ORIGIN]
78: - whether they are tainted or not,
79: and the language which should be used to detaint them
80: */
1.15 paf 81: class String : public Pooled {
1.1 paf 82: public:
1.48 paf 83:
1.1 paf 84: enum {
1.41 paf 85: CR_PREALLOCATED_COUNT=5, ///< default preallocated item count
86: CR_GROW_PERCENT=60 ///< each time the Array chunk_is_full() array expanded()
1.1 paf 87: };
88:
1.41 paf 89: /// piece is tainted or not. the language to use when detaint
1.27 paf 90: enum Untaint_lang {
1.72 paf 91: UL_UNSPECIFIED=0, ///< zero value handy for hash lookup @see untaint_lang_name2enum
1.62 paf 92: UL_CLEAN, ///< clean
93: UL_TAINTED, ///< tainted, untaint language as assigned later
1.27 paf 94: // untaint languages. assigned by ^untaint[lang]{...}
1.48 paf 95: UL_PASS_APPENDED,
1.41 paf 96: /**<
97: leave language built into string being appended.
98: just a flag, that value not stored
99: */
1.48 paf 100: UL_AS_IS, ///< leave all characters intact
101: UL_FILE_NAME, ///< filename
1.74 ! paf 102: UL_HTTP_HEADER, ///< text in HTTP response header
! 103: UL_MAIL_HEADER, ///< text in mail header
1.48 paf 104: UL_URI, ///< text in uri
105: UL_TABLE, ///< ^table:set body
106: UL_SQL, ///< ^table:sql body
107: UL_JS, ///< JavaScript code
108: UL_HTML, ///< HTML code (for editing)
109: UL_HTML_TYPO ///< HTML code with TYPOgraphic replacements (for showing)
1.27 paf 110: };
111:
1.8 paf 112: public:
113:
1.37 paf 114: String(Pool& apool, const char *src=0, bool tainted=false);
1.14 paf 115: String(const String& src);
116: size_t size() const { return fsize; }
1.69 paf 117: /// convert to C string. if 'lang' known, forcing 'lang' to it
1.72 paf 118: char *cstr(Untaint_lang lang=UL_UNSPECIFIED, SQL_Connection *connection=0) const {
1.50 paf 119: char *result=(char *)malloc(size()*UNTAINT_TIMES_BIGGER+1);
1.71 paf 120: char *eol=store_to(result, lang, connection);
1.50 paf 121: *eol=0;
122: return result;
123: }
1.46 paf 124: /** append fragment
1.54 paf 125: @see APPEND_CLEAN, APPEND_TAINTED, APPEND_CONST
1.46 paf 126: */
1.9 paf 127: String& real_append(STRING_APPEND_PARAMS);
1.44 paf 128: /// @return <0 ==0 or >0 depending on comparison result
1.62 paf 129: int cmp (int& partial, const String& src,
1.72 paf 130: size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.56 paf 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; }
133: bool operator <= (const String& src) const { int p; return cmp(p, src)<=0; }
134: bool operator >= (const String& src) const { int p; return cmp(p, src)>=0; }
1.26 paf 135: bool operator == (const String& src) const {
136: if(size()!=src.size()) // can speed up in trivial case
137: return false;
1.56 paf 138: int p; return cmp(p, src)==0;
1.26 paf 139: }
1.56 paf 140: bool operator != (const String& src) const { int p; return cmp(p, src)!=0; }
1.26 paf 141:
1.50 paf 142: /**
143: @param partial
144: returns partial match status.
1.51 paf 145: - -1: strings too different
146: - 0: full match
147: - 1: means @c this starts @c src
148: - 2: means @src starts @this
1.50 paf 149: */
1.61 paf 150: int cmp(int& partial, const char* src_ptr, size_t src_size=0,
1.72 paf 151: size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.50 paf 152: bool operator == (const char* src_ptr) const {
153: size_t src_size=src_ptr?strlen(src_ptr):0;
154: if(size() != src_size)
155: return false;
156: int partial; // unused
1.56 paf 157: return cmp(partial, src_ptr, src_size)==0;
1.50 paf 158: }
159:
1.42 paf 160: /**
161: appends other String.
1.41 paf 162:
1.47 paf 163: marking all tainted pieces of it with @a lang.
164: or marking ALL pieces of it with a @a lang when @a forced to.
1.41 paf 165: */
1.39 paf 166: String& append(const String& src, Untaint_lang lang, bool forced=false);
1.8 paf 167:
1.41 paf 168: /// simple hash code of string. used by Hash
1.14 paf 169: uint hash_code() const;
1.54 paf 170:
171: /// extracts [start, finish) piece of string
1.70 paf 172: String& mid(size_t start, size_t finish) const;
1.55 paf 173:
1.59 paf 174: /// @return position of substr in string, -1 means "not found" [String version]
1.62 paf 175: int pos(const String& substr,
1.72 paf 176: size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.59 paf 177: /// @return position of substr in string, -1 means "not found" [const char* version]
1.62 paf 178: int pos(const char *substr, size_t substr_size,
1.72 paf 179: size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.62 paf 180:
181: void split(Array& result,
182: size_t *pos_after_ref,
183: const char *delim, size_t delim_size,
184: Untaint_lang lang, int limit=-1) const;
185: void split(Array& result,
186: size_t *pos_after_ref,
187: const String& delim,
188: Untaint_lang lang, int limit=-1) const;
189:
1.68 paf 190: typedef void (*Row_action)(Table& table, Array *row, int start, int finish,
191: void *info);
1.64 paf 192: bool match(const String *aorigin,
193: const String& regexp,
1.65 paf 194: const String *options,
1.66 paf 195: Table **table,
196: Row_action row_action, void *info) const;
1.13 paf 197:
1.41 paf 198: #ifndef NO_STRING_ORIGIN
199: /// origin of string. calculated by first row
1.43 paf 200: const Origin& origin() const;
1.41 paf 201: #endif
1.8 paf 202:
1.10 paf 203: private:
204:
1.1 paf 205: struct Chunk {
1.6 paf 206: // the number of rows in chunk
1.56 paf 207: size_t count;
1.1 paf 208: union Row {
1.27 paf 209: // fragment
210: struct {
211: const char *ptr; // pointer to the start
212: size_t size; // length
213: Untaint_lang lang; // untaint flag, later untaint language
214: #ifndef NO_STRING_ORIGIN
215: Origin origin; // origin
216: #endif
1.1 paf 217: } item;
218: Chunk *link; // link to the next chunk in chain
1.2 paf 219: } rows[CR_PREALLOCATED_COUNT];
1.1 paf 220: // next rows are here
221: Chunk *preallocated_link;
222: }
223: head; // the head chunk of the chunk chain
224:
225: // next append would write to this record
226: Chunk::Row *append_here;
227:
228: // the address of place where lies address
229: // of the link to the next chunk to allocate
230: Chunk::Row *link_row;
231:
1.5 paf 232: private:
1.25 paf 233: // last chunk
234: Chunk *last_chunk;
1.5 paf 235:
236: // string size
237: size_t fsize;
238:
239: // used rows in all chunks
240: int fused_rows;
241:
242: private:
1.1 paf 243:
244: bool chunk_is_full() {
245: return append_here == link_row;
246: }
247: void expand();
1.71 paf 248: char *String::store_to(char *dest,
1.72 paf 249: Untaint_lang lang=UL_UNSPECIFIED, SQL_Connection *connection=0) const;
1.7 paf 250:
251: private: //disabled
252:
1.12 paf 253: String& operator = (const String&) { return *this; }
1.7 paf 254:
1.1 paf 255: };
256:
257: #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