1 /*
2 Copyright (c) 2000, 2020, Oracle and/or its affiliates. All rights reserved.
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_INCLUDED
25 #define M_STRING_INCLUDED
26
27 /**
28 @file include/m_string.h
29 */
30
31 #include <float.h>
32 #include <limits.h>
33 #include <stdbool.h> // IWYU pragma: keep
34 #include <stdint.h>
35 #include <stdio.h>
36 #include <stdlib.h>
37 #include <string.h>
38
39 #include <algorithm>
40
41 #include "decimal.h"
42 #include "lex_string.h"
43 #include "my_config.h"
44 #include "my_inttypes.h"
45 #include "my_macros.h"
46
47 /**
48 Definition of the null string (a null pointer of type char *),
49 used in some of our string handling code. New code should use
50 nullptr instead.
51 */
52 #define NullS (char *)0
53
54 /*
55 my_str_malloc(), my_str_realloc() and my_str_free() are assigned to
56 implementations in strings/alloc.cc, but can be overridden in
57 the calling program.
58 */
59 extern void *(*my_str_malloc)(size_t);
60 extern void *(*my_str_realloc)(void *, size_t);
61 extern void (*my_str_free)(void *);
62
63 /* Declared in int2str.cc. */
64 extern const char _dig_vec_upper[];
65 extern const char _dig_vec_lower[];
66
67 /* Prototypes for string functions */
68
69 extern char *strmake(char *dst, const char *src, size_t length);
70 extern char *strcont(char *src, const char *set);
71 extern char *strxmov(char *dst, const char *src, ...);
72 extern char *strxnmov(char *dst, size_t len, const char *src, ...);
73
74 /*
75 bchange(dst, old_length, src, new_length, tot_length)
76 replaces old_length characters at dst to new_length characters from
77 src in a buffer with tot_length bytes.
78 */
bchange(uchar * dst,size_t old_length,const uchar * src,size_t new_length,size_t tot_length)79 static inline void bchange(uchar *dst, size_t old_length, const uchar *src,
80 size_t new_length, size_t tot_length) {
81 memmove(dst + new_length, dst + old_length, tot_length - old_length);
82 memcpy(dst, src, new_length);
83 }
84
85 /*
86 strend(s) returns a character pointer to the NUL which ends s. That
87 is, strend(s)-s == strlen(s). This is useful for adding things at
88 the end of strings. It is redundant, because strchr(s,'\0') could
89 be used instead, but this is clearer and faster.
90 */
strend(const char * s)91 static inline const char *strend(const char *s) {
92 while (*s++)
93 ;
94 return s - 1;
95 }
96
strend(char * s)97 static inline char *strend(char *s) {
98 while (*s++)
99 ;
100 return s - 1;
101 }
102
103 /*
104 strcend(s, c) returns a pointer to the first place in s where c
105 occurs, or a pointer to the end-null of s if c does not occur in s.
106 */
strcend(const char * s,char c)107 static inline const char *strcend(const char *s, char c) {
108 for (;;) {
109 if (*s == c) return s;
110 if (!*s++) return s - 1;
111 }
112 }
113
114 /*
115 strfill(dest, len, fill) makes a string of fill-characters. The result
116 string is of length == len. The des+len character is allways set to NULL.
117 strfill() returns pointer to dest+len;
118 */
strfill(char * s,size_t len,char fill)119 static inline char *strfill(char *s, size_t len, char fill) {
120 while (len--) *s++ = fill;
121 *(s) = '\0';
122 return (s);
123 }
124
125 /*
126 my_stpmov(dst, src) moves all the characters of src (including the
127 closing NUL) to dst, and returns a pointer to the new closing NUL in
128 dst. The similar UNIX routine strcpy returns the old value of dst,
129 which I have never found useful. my_stpmov(my_stpmov(dst,a),b) moves a//b
130 into dst, which seems useful.
131 */
my_stpmov(char * dst,const char * src)132 static inline char *my_stpmov(char *dst, const char *src) {
133 while ((*dst++ = *src++))
134 ;
135 return dst - 1;
136 }
137
138 /*
139 my_stpnmov(dst,src,length) moves length characters, or until end, of src to
140 dst and appends a closing NUL to dst if src is shorter than length.
141 The result is a pointer to the first NUL in dst, or is dst+n if dst was
142 truncated.
143 */
my_stpnmov(char * dst,const char * src,size_t n)144 static inline char *my_stpnmov(char *dst, const char *src, size_t n) {
145 while (n-- != 0) {
146 if (!(*dst++ = *src++)) return (char *)dst - 1;
147 }
148 return dst;
149 }
150
151 /**
152 Copy a string from src to dst until (and including) terminating null byte.
153
154 @param dst Destination
155 @param src Source
156
157 @note src and dst cannot overlap.
158 Use my_stpmov() if src and dst overlaps.
159
160 @note Unsafe, consider using my_stpnpy() instead.
161
162 @return pointer to terminating null byte.
163 */
my_stpcpy(char * dst,const char * src)164 static inline char *my_stpcpy(char *dst, const char *src) {
165 #if defined(HAVE_BUILTIN_STPCPY)
166 /*
167 If __builtin_stpcpy() is available, use it instead of stpcpy(), since GCC in
168 some situations is able to transform __builtin_stpcpy() into more efficient
169 strcpy() or memcpy() calls. It does not perform these transformations for a
170 plain call to stpcpy() when the compiler runs in strict mode. See GCC bug
171 82429.
172 */
173 return __builtin_stpcpy(dst, src);
174 #elif defined(HAVE_STPCPY)
175 return stpcpy(dst, src);
176 #else
177 /* Fallback to implementation supporting overlap. */
178 return my_stpmov(dst, src);
179 #endif
180 }
181
182 /**
183 Copy fixed-size string from src to dst.
184
185 @param dst Destination
186 @param src Source
187 @param n Maximum number of characters to copy.
188
189 @note src and dst cannot overlap
190 Use my_stpnmov() if src and dst overlaps.
191
192 @return pointer to terminating null byte.
193 */
my_stpncpy(char * dst,const char * src,size_t n)194 static inline char *my_stpncpy(char *dst, const char *src, size_t n) {
195 #if defined(HAVE_STPNCPY)
196 return stpncpy(dst, src, n);
197 #else
198 /* Fallback to implementation supporting overlap. */
199 return my_stpnmov(dst, src, n);
200 #endif
201 }
202
my_strtoll(const char * nptr,char ** endptr,int base)203 static inline longlong my_strtoll(const char *nptr, char **endptr, int base) {
204 #if defined _WIN32
205 return _strtoi64(nptr, endptr, base);
206 #else
207 return strtoll(nptr, endptr, base);
208 #endif
209 }
210
my_strtoull(const char * nptr,char ** endptr,int base)211 static inline ulonglong my_strtoull(const char *nptr, char **endptr, int base) {
212 #if defined _WIN32
213 return _strtoui64(nptr, endptr, base);
214 #else
215 return strtoull(nptr, endptr, base);
216 #endif
217 }
218
my_strtok_r(char * str,const char * delim,char ** saveptr)219 static inline char *my_strtok_r(char *str, const char *delim, char **saveptr) {
220 #if defined _WIN32
221 return strtok_s(str, delim, saveptr);
222 #else
223 return strtok_r(str, delim, saveptr);
224 #endif
225 }
226
227 /* native_ rather than my_ since my_strcasecmp already exists */
native_strcasecmp(const char * s1,const char * s2)228 static inline int native_strcasecmp(const char *s1, const char *s2) {
229 #if defined _WIN32
230 return _stricmp(s1, s2);
231 #else
232 return strcasecmp(s1, s2);
233 #endif
234 }
235
236 /* native_ rather than my_ for consistency with native_strcasecmp */
native_strncasecmp(const char * s1,const char * s2,size_t n)237 static inline int native_strncasecmp(const char *s1, const char *s2, size_t n) {
238 #if defined _WIN32
239 return _strnicmp(s1, s2, n);
240 #else
241 return strncasecmp(s1, s2, n);
242 #endif
243 }
244
245 /*
246 is_prefix(s, t) returns 1 if s starts with t.
247 A empty t is always a prefix.
248 */
is_prefix(const char * s,const char * t)249 static inline int is_prefix(const char *s, const char *t) {
250 while (*t)
251 if (*s++ != *t++) return 0;
252 return 1; /* WRONG */
253 }
254
255 /* Conversion routines */
256 typedef enum { MY_GCVT_ARG_FLOAT, MY_GCVT_ARG_DOUBLE } my_gcvt_arg_type;
257
258 double my_strtod(const char *str, const char **end, int *error);
259 double my_atof(const char *nptr);
260 size_t my_fcvt(double x, int precision, char *to, bool *error);
261 size_t my_fcvt_compact(double x, char *to, bool *error);
262 size_t my_gcvt(double x, my_gcvt_arg_type type, int width, char *to,
263 bool *error);
264
265 /*
266 The longest string my_fcvt can return is 311 + "precision" bytes.
267 Here we assume that we never call my_fcvt() with precision >=
268 DECIMAL_NOT_SPECIFIED
269 (+ 1 byte for the terminating '\0').
270 */
271 static constexpr int FLOATING_POINT_BUFFER{311 + DECIMAL_NOT_SPECIFIED};
272
273 /*
274 We want to use the 'e' format in some cases even if we have enough space
275 for the 'f' one just to mimic sprintf("%.15g") behavior for large integers,
276 and to improve it for numbers < 10^(-4).
277 That is, for |x| < 1 we require |x| >= 10^(-15), and for |x| > 1 we require
278 it to be integer and be <= 10^DBL_DIG for the 'f' format to be used.
279 We don't lose precision, but make cases like "1e200" or "0.00001" look nicer.
280 */
281 #define MAX_DECPT_FOR_F_FORMAT DBL_DIG
282
283 /*
284 The maximum possible field width for my_gcvt() conversion.
285 (DBL_DIG + 2) significant digits + sign + "." + ("e-NNN" or
286 MAX_DECPT_FOR_F_FORMAT zeros for cases when |x|<1 and the 'f' format is used).
287 */
288 #define MY_GCVT_MAX_FIELD_WIDTH \
289 (DBL_DIG + 4 + std::max(5, MAX_DECPT_FOR_F_FORMAT))
290
291 const char *str2int(const char *src, int radix, long lower, long upper,
292 long *val);
293 longlong my_strtoll10(const char *nptr, const char **endptr, int *error);
294 char *ll2str(int64_t val, char *dst, int radix, bool upcase);
295 char *longlong10_to_str(int64_t val, char *dst, int radix);
296
longlong2str(int64_t val,char * dst,int radix)297 inline char *longlong2str(int64_t val, char *dst, int radix) {
298 return ll2str(val, dst, radix, true);
299 }
300
301 /*
302 This function saves a longlong value in a buffer and returns the pointer to
303 the buffer.
304 */
llstr(longlong value,char * buff)305 static inline char *llstr(longlong value, char *buff) {
306 longlong10_to_str(value, buff, -10);
307 return buff;
308 }
309
ullstr(longlong value,char * buff)310 static inline char *ullstr(longlong value, char *buff) {
311 longlong10_to_str(value, buff, 10);
312 return buff;
313 }
314
315 #define STRING_WITH_LEN(X) (X), ((sizeof(X) - 1))
316
317 /**
318 Skip trailing space (ASCII spaces only).
319
320 @return New end of the string.
321 */
skip_trailing_space(const uchar * ptr,size_t len)322 static inline const uchar *skip_trailing_space(const uchar *ptr, size_t len) {
323 const uchar *end = ptr + len;
324 while (end - ptr >= 8) {
325 uint64_t chunk;
326 memcpy(&chunk, end - 8, sizeof(chunk));
327 if (chunk != 0x2020202020202020ULL) break;
328 end -= 8;
329 }
330 while (end > ptr && end[-1] == 0x20) end--;
331 return (end);
332 }
333
334 /*
335 Format a double (representing number of bytes) into a human-readable string.
336
337 @param buf Buffer used for printing
338 @param buf_len Length of buffer
339 @param dbl_val Value to be formatted
340
341 @note
342 Sample output format: 42 1K 234M 2G
343 If we exceed ULLONG_MAX YiB we give up, and convert to "+INF".
344
345 @todo Consider writing KiB GiB etc, since we use 1024 rather than 1000
346 */
human_readable_num_bytes(char * buf,int buf_len,double dbl_val)347 static inline void human_readable_num_bytes(char *buf, int buf_len,
348 double dbl_val) {
349 const char size[] = {'\0', 'K', 'M', 'G', 'T', 'P', 'E', 'Z', 'Y'};
350 unsigned int i;
351 for (i = 0; dbl_val > 1024 && i < sizeof(size) - 1; i++) dbl_val /= 1024;
352 const char mult = size[i];
353 // 18446744073709551615 Yottabytes should be enough for most ...
354 // ULLONG_MAX is not exactly representable as a double. This is the largest
355 // double that is still below ULLONG_MAX.
356 if (dbl_val > 18446744073709549568.0)
357 snprintf(buf, buf_len, "+INF");
358 else
359 snprintf(buf, buf_len, "%llu%c", (unsigned long long)dbl_val, mult);
360 }
361
lex_string_set(LEX_STRING * lex_str,char * c_str)362 static inline void lex_string_set(LEX_STRING *lex_str, char *c_str) {
363 lex_str->str = c_str;
364 lex_str->length = strlen(c_str);
365 }
366
lex_cstring_set(LEX_CSTRING * lex_str,const char * c_str)367 static inline void lex_cstring_set(LEX_CSTRING *lex_str, const char *c_str) {
368 lex_str->str = c_str;
369 lex_str->length = strlen(c_str);
370 }
371
372 #endif // M_STRING_INCLUDED
373