1 /*
2    Copyright (c) 2000, 2021, Oracle and/or its affiliates.
3 
4    This program is free software; you can redistribute it and/or modify
5    it under the terms of the GNU General Public License, version 2.0,
6    as published by the Free Software Foundation.
7 
8    This program is also distributed with certain software (including
9    but not limited to OpenSSL) that is licensed under separate terms,
10    as designated in a particular file or component or in included license
11    documentation.  The authors of MySQL hereby grant you an additional
12    permission to link the program and your derivative works with the
13    separately licensed software that they have included with MySQL.
14 
15    This program is distributed in the hope that it will be useful,
16    but WITHOUT ANY WARRANTY; without even the implied warranty of
17    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
18    GNU General Public License, version 2.0, for more details.
19 
20    You should have received a copy of the GNU General Public License
21    along with this program; if not, write to the Free Software
22    Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301  USA */
23 
24 #ifndef _m_string_h
25 #define _m_string_h
26 
27 #include "my_global.h"                          /* HAVE_* */
28 
29 #include <string.h>
30 
31 #include "mysql/service_my_snprintf.h"
32 
33 #if defined(__cplusplus)
34 extern "C" {
35 #endif
36 
37 /*
38   my_str_malloc(), my_str_realloc() and my_str_free() are assigned to
39   implementations in strings/alloc.c, but can be overridden in
40   the calling program.
41  */
42 extern void *(*my_str_malloc)(size_t);
43 extern void *(*my_str_realloc)(void *, size_t);
44 extern void (*my_str_free)(void *);
45 
46 /* Declared in int2str() */
47 extern char _dig_vec_upper[];
48 extern char _dig_vec_lower[];
49 
50 	/* Prototypes for string functions */
51 
52 extern	void bchange(uchar *dst,size_t old_len,const uchar *src,
53 		     size_t new_len,size_t tot_len);
54 extern	void strappend(char *s,size_t len,pchar fill);
55 extern	char *strend(const char *s);
56 extern  char *strcend(const char *, pchar);
57 extern	char *strfill(char * s,size_t len,pchar fill);
58 extern	char *strmake(char *dst,const char *src,size_t length);
59 
60 extern	char *my_stpmov(char *dst,const char *src);
61 extern	char *my_stpnmov(char *dst, const char *src, size_t n);
62 extern	char *strcont(const char *src, const char *set);
63 extern	char *strxmov(char *dst, const char *src, ...);
64 extern	char *strxnmov(char *dst, size_t len, const char *src, ...);
65 
66 /**
67    Copy a string from src to dst until (and including) terminating null byte.
68 
69    @param dst   Destination
70    @param src   Source
71 
72    @note src and dst cannot overlap.
73          Use my_stpmov() if src and dst overlaps.
74 
75    @note Unsafe, consider using my_stpnpy() instead.
76 
77    @return pointer to terminating null byte.
78 */
my_stpcpy(char * dst,const char * src)79 static inline char *my_stpcpy(char *dst, const char *src)
80 {
81 #if defined(HAVE_BUILTIN_STPCPY)
82   return __builtin_stpcpy(dst, src);
83 #elif defined(HAVE_STPCPY)
84   return stpcpy(dst, src);
85 #else
86   /* Fallback to implementation supporting overlap. */
87   return my_stpmov(dst, src);
88 #endif
89 }
90 
91 /**
92    Copy fixed-size string from src to dst.
93 
94    @param dst   Destination
95    @param src   Source
96    @param n     Maximum number of characters to copy.
97 
98    @note src and dst cannot overlap
99          Use my_stpnmov() if src and dst overlaps.
100 
101    @return pointer to terminating null byte.
102 */
my_stpncpy(char * dst,const char * src,size_t n)103 static inline char *my_stpncpy(char *dst, const char *src, size_t n)
104 {
105 #if defined(HAVE_STPNCPY)
106   return stpncpy(dst, src, n);
107 #else
108   /* Fallback to implementation supporting overlap. */
109   return my_stpnmov(dst, src, n);
110 #endif
111 }
112 
113 /**
114    Copies strlen(src) characters of source to destination.
115    If strlen(src) is equal or bigger than num then dst will be truncated.
116    The null-character is always appended at the end of destination.
117    Destination is not padded with zeros until a total of num characters.
118 
119    @param dst   Destination
120    @param src   Source
121    @param n     Maximum number of characters to copy.
122 
123    @return pointer to Destination is returned.
124 */
my_strncpy_trunc(char * dst,const char * src,size_t num)125 static inline char *my_strncpy_trunc(char *dst, const char *src, size_t num)
126 {
127   size_t len= strlen(src);
128   if (unlikely(len >= num)) {
129     len= num - 1;
130     memcpy(dst, src, len);
131     dst[len]= '\0';
132   } else {
133     memcpy(dst, src, len + 1);
134   }
135   return dst;
136 }
137 
my_strtoll(const char * nptr,char ** endptr,int base)138 static inline longlong my_strtoll(const char *nptr, char **endptr, int base)
139 {
140 #if defined _WIN32
141   return _strtoi64(nptr, endptr, base);
142 #else
143   return strtoll(nptr, endptr, base);
144 #endif
145 }
146 
my_strtoull(const char * nptr,char ** endptr,int base)147 static inline ulonglong my_strtoull(const char *nptr, char **endptr, int base)
148 {
149 #if defined _WIN32
150   return _strtoui64(nptr, endptr, base);
151 #else
152   return strtoull(nptr, endptr, base);
153 #endif
154 }
155 
my_strtok_r(char * str,const char * delim,char ** saveptr)156 static inline char *my_strtok_r(char *str, const char *delim, char **saveptr)
157 {
158 #if defined _WIN32
159   return strtok_s(str, delim, saveptr);
160 #else
161   return strtok_r(str, delim, saveptr);
162 #endif
163 }
164 
165 /* native_ rather than my_ since my_strcasecmp already exists */
native_strcasecmp(const char * s1,const char * s2)166 static inline int native_strcasecmp(const char *s1, const char *s2)
167 {
168 #if defined _WIN32
169   return _stricmp(s1, s2);
170 #else
171   return strcasecmp(s1, s2);
172 #endif
173 }
174 
175 /* native_ rather than my_ for consistency with native_strcasecmp */
native_strncasecmp(const char * s1,const char * s2,size_t n)176 static inline int native_strncasecmp(const char *s1, const char *s2, size_t n)
177 {
178 #if defined _WIN32
179   return _strnicmp(s1, s2, n);
180 #else
181   return strncasecmp(s1, s2, n);
182 #endif
183 }
184 
185 /* Prototypes of normal stringfunctions (with may ours) */
186 #ifndef HAVE_STRNLEN
187 extern size_t strnlen(const char *s, size_t n);
188 #endif
189 
190 extern int is_prefix(const char *, const char *);
191 
192 /* Conversion routines */
193 typedef enum {
194   MY_GCVT_ARG_FLOAT,
195   MY_GCVT_ARG_DOUBLE
196 } my_gcvt_arg_type;
197 
198 double my_strtod(const char *str, char **end, int *error);
199 double my_atof(const char *nptr);
200 size_t my_fcvt(double x, int precision, char *to, my_bool *error);
201 size_t my_gcvt(double x, my_gcvt_arg_type type, int width, char *to,
202                my_bool *error);
203 
204 #define NOT_FIXED_DEC 31
205 
206 /*
207   The longest string my_fcvt can return is 311 + "precision" bytes.
208   Here we assume that we never cal my_fcvt() with precision >= NOT_FIXED_DEC
209   (+ 1 byte for the terminating '\0').
210 */
211 #define FLOATING_POINT_BUFFER (311 + NOT_FIXED_DEC)
212 
213 /*
214   We want to use the 'e' format in some cases even if we have enough space
215   for the 'f' one just to mimic sprintf("%.15g") behavior for large integers,
216   and to improve it for numbers < 10^(-4).
217   That is, for |x| < 1 we require |x| >= 10^(-15), and for |x| > 1 we require
218   it to be integer and be <= 10^DBL_DIG for the 'f' format to be used.
219   We don't lose precision, but make cases like "1e200" or "0.00001" look nicer.
220 */
221 #define MAX_DECPT_FOR_F_FORMAT DBL_DIG
222 
223 /*
224   The maximum possible field width for my_gcvt() conversion.
225   (DBL_DIG + 2) significant digits + sign + "." + ("e-NNN" or
226   MAX_DECPT_FOR_F_FORMAT zeros for cases when |x|<1 and the 'f' format is used).
227 */
228 #define MY_GCVT_MAX_FIELD_WIDTH (DBL_DIG + 4 + MY_MAX(5, MAX_DECPT_FOR_F_FORMAT)) \
229 
230 extern char *llstr(longlong value,char *buff);
231 extern char *ullstr(longlong value,char *buff);
232 
233 extern char *int2str(long val, char *dst, int radix, int upcase);
234 extern char *int10_to_str(long val,char *dst,int radix);
235 extern char *str2int(const char *src,int radix,long lower,long upper,
236 			 long *val);
237 longlong my_strtoll10(const char *nptr, char **endptr, int *error);
238 #if SIZEOF_LONG == SIZEOF_LONG_LONG
239 #define ll2str(A,B,C,D) int2str((A),(B),(C),(D))
240 #define longlong10_to_str(A,B,C) int10_to_str((A),(B),(C))
241 #undef strtoll
242 #define strtoll(A,B,C) strtol((A),(B),(C))
243 #define strtoull(A,B,C) strtoul((A),(B),(C))
244 #else
245 extern char *ll2str(longlong val,char *dst,int radix, int upcase);
246 extern char *longlong10_to_str(longlong val,char *dst,int radix);
247 #endif
248 #define longlong2str(A,B,C) ll2str((A),(B),(C),1)
249 
250 #if defined(__cplusplus)
251 }
252 #endif
253 
254 /*
255   LEX_STRING -- a pair of a C-string and its length.
256   (it's part of the plugin API as a MYSQL_LEX_STRING)
257   Ditto LEX_CSTRING/MYSQL_LEX_CSTRING.
258 */
259 
260 #include <mysql/mysql_lex_string.h>
261 typedef struct st_mysql_lex_string LEX_STRING;
262 typedef struct st_mysql_const_lex_string LEX_CSTRING;
263 
264 #define STRING_WITH_LEN(X) (X), ((sizeof(X) - 1))
265 #define USTRING_WITH_LEN(X) ((uchar*) X), ((sizeof(X) - 1))
266 #define C_STRING_WITH_LEN(X) ((char *) (X)), ((sizeof(X) - 1))
267 
268 
269 /**
270   Skip trailing space.
271 
272   On most systems reading memory in larger chunks (ideally equal to the size of
273   the chunks that the machine physically reads from memory) causes fewer memory
274   access loops and hence increased performance.
275   This is why the 'int' type is used : it's closest to that (according to how
276   it's defined in C).
277   So when we determine the amount of whitespace at the end of a string we do
278   the following :
279     1. We divide the string into 3 zones :
280       a) from the start of the string (__start) to the first multiple
281         of sizeof(int)  (__start_words)
282       b) from the end of the string (__end) to the last multiple of sizeof(int)
283         (__end_words)
284       c) a zone that is aligned to sizeof(int) and can be safely accessed
285         through an int *
286     2. We start comparing backwards from (c) char-by-char. If all we find is
287        space then we continue
288     3. If there are elements in zone (b) we compare them as unsigned ints to a
289        int mask (SPACE_INT) consisting of all spaces
290     4. Finally we compare the remaining part (a) of the string char by char.
291        This covers for the last non-space unsigned int from 3. (if any)
292 
293    This algorithm works well for relatively larger strings, but it will slow
294    the things down for smaller strings (because of the additional calculations
295    and checks compared to the naive method). Thus the barrier of length 20
296    is added.
297 
298    @param     ptr   pointer to the input string
299    @param     len   the length of the string
300    @return          the last non-space character
301 */
302 #if defined(__sparc) || defined(__sparcv9)
skip_trailing_space(const uchar * ptr,size_t len)303 static inline const uchar *skip_trailing_space(const uchar *ptr,size_t len)
304 {
305   /* SPACE_INT is a word that contains only spaces */
306 #if SIZEOF_INT == 4
307   const unsigned SPACE_INT= 0x20202020U;
308 #elif SIZEOF_INT == 8
309   const unsigned SPACE_INT= 0x2020202020202020ULL;
310 #else
311 #error define the appropriate constant for a word full of spaces
312 #endif
313 
314   const uchar *end= ptr + len;
315 
316   if (len > 20)
317   {
318     const uchar *end_words= (const uchar *)(intptr)
319       (((ulonglong)(intptr)end) / SIZEOF_INT * SIZEOF_INT);
320     const uchar *start_words= (const uchar *)(intptr)
321        ((((ulonglong)(intptr)ptr) + SIZEOF_INT - 1) / SIZEOF_INT * SIZEOF_INT);
322 
323     assert(end_words > ptr);
324     while (end > end_words && end[-1] == 0x20)
325       end--;
326     if (end[-1] == 0x20 && start_words < end_words)
327       while (end > start_words && ((unsigned *)end)[-1] == SPACE_INT)
328         end -= SIZEOF_INT;
329   }
330   while (end > ptr && end[-1] == 0x20)
331     end--;
332   return (end);
333 }
334 #else
335 /*
336   Reads 8 bytes at a time, ignoring alignment.
337   We use uint8korr, which is fast (it simply reads a *ulonglong)
338   on all platforms, except sparc.
339 */
skip_trailing_space(const uchar * ptr,size_t len)340 static inline const uchar *skip_trailing_space(const uchar *ptr, size_t len)
341 {
342   const uchar *end= ptr + len;
343   while (end - ptr >= 8)
344   {
345     if (uint8korr(end-8) != 0x2020202020202020ULL)
346       break;
347     end-= 8;
348   }
349   while (end > ptr && end[-1] == 0x20)
350     end--;
351   return (end);
352 }
353 #endif
354 
lex_string_set(LEX_STRING * lex_str,const char * c_str)355 static inline void lex_string_set(LEX_STRING *lex_str, const char *c_str)
356 {
357   lex_str->str= (char *) c_str;
358   lex_str->length= strlen(c_str);
359 }
360 
lex_cstring_set(LEX_CSTRING * lex_str,const char * c_str)361 static inline void lex_cstring_set(LEX_CSTRING *lex_str, const char *c_str)
362 {
363   lex_str->str= c_str;
364   lex_str->length= strlen(c_str);
365 }
366 
367 #endif
368