1 | /* |
2 | * psql - the PostgreSQL interactive terminal |
3 | * |
4 | * Copyright (c) 2000-2019, PostgreSQL Global Development Group |
5 | * |
6 | * src/bin/psql/stringutils.c |
7 | */ |
8 | #include "postgres_fe.h" |
9 | |
10 | #include <ctype.h> |
11 | |
12 | #include "common.h" |
13 | #include "stringutils.h" |
14 | |
15 | |
16 | /* |
17 | * Replacement for strtok() (a.k.a. poor man's flex) |
18 | * |
19 | * Splits a string into tokens, returning one token per call, then NULL |
20 | * when no more tokens exist in the given string. |
21 | * |
22 | * The calling convention is similar to that of strtok, but with more |
23 | * frammishes. |
24 | * |
25 | * s - string to parse, if NULL continue parsing the last string |
26 | * whitespace - set of whitespace characters that separate tokens |
27 | * delim - set of non-whitespace separator characters (or NULL) |
28 | * quote - set of characters that can quote a token (NULL if none) |
29 | * escape - character that can quote quotes (0 if none) |
30 | * e_strings - if true, treat E'...' syntax as a valid token |
31 | * del_quotes - if true, strip quotes from the returned token, else return |
32 | * it exactly as found in the string |
33 | * encoding - the active character-set encoding |
34 | * |
35 | * Characters in 'delim', if any, will be returned as single-character |
36 | * tokens unless part of a quoted token. |
37 | * |
38 | * Double occurrences of the quoting character are always taken to represent |
39 | * a single quote character in the data. If escape isn't 0, then escape |
40 | * followed by anything (except \0) is a data character too. |
41 | * |
42 | * The combination of e_strings and del_quotes both true is not currently |
43 | * handled. This could be fixed but it's not needed anywhere at the moment. |
44 | * |
45 | * Note that the string s is _not_ overwritten in this implementation. |
46 | * |
47 | * NB: it's okay to vary delim, quote, and escape from one call to the |
48 | * next on a single source string, but changing whitespace is a bad idea |
49 | * since you might lose data. |
50 | */ |
51 | char * |
52 | strtokx(const char *s, |
53 | const char *whitespace, |
54 | const char *delim, |
55 | const char *quote, |
56 | char escape, |
57 | bool e_strings, |
58 | bool del_quotes, |
59 | int encoding) |
60 | { |
61 | static char *storage = NULL; /* store the local copy of the users |
62 | * string here */ |
63 | static char *string = NULL; /* pointer into storage where to continue on |
64 | * next call */ |
65 | |
66 | /* variously abused variables: */ |
67 | unsigned int offset; |
68 | char *start; |
69 | char *p; |
70 | |
71 | if (s) |
72 | { |
73 | free(storage); |
74 | |
75 | /* |
76 | * We may need extra space to insert delimiter nulls for adjacent |
77 | * tokens. 2X the space is a gross overestimate, but it's unlikely |
78 | * that this code will be used on huge strings anyway. |
79 | */ |
80 | storage = pg_malloc(2 * strlen(s) + 1); |
81 | strcpy(storage, s); |
82 | string = storage; |
83 | } |
84 | |
85 | if (!storage) |
86 | return NULL; |
87 | |
88 | /* skip leading whitespace */ |
89 | offset = strspn(string, whitespace); |
90 | start = &string[offset]; |
91 | |
92 | /* end of string reached? */ |
93 | if (*start == '\0') |
94 | { |
95 | /* technically we don't need to free here, but we're nice */ |
96 | free(storage); |
97 | storage = NULL; |
98 | string = NULL; |
99 | return NULL; |
100 | } |
101 | |
102 | /* test if delimiter character */ |
103 | if (delim && strchr(delim, *start)) |
104 | { |
105 | /* |
106 | * If not at end of string, we need to insert a null to terminate the |
107 | * returned token. We can just overwrite the next character if it |
108 | * happens to be in the whitespace set ... otherwise move over the |
109 | * rest of the string to make room. (This is why we allocated extra |
110 | * space above). |
111 | */ |
112 | p = start + 1; |
113 | if (*p != '\0') |
114 | { |
115 | if (!strchr(whitespace, *p)) |
116 | memmove(p + 1, p, strlen(p) + 1); |
117 | *p = '\0'; |
118 | string = p + 1; |
119 | } |
120 | else |
121 | { |
122 | /* at end of string, so no extra work */ |
123 | string = p; |
124 | } |
125 | |
126 | return start; |
127 | } |
128 | |
129 | /* check for E string */ |
130 | p = start; |
131 | if (e_strings && |
132 | (*p == 'E' || *p == 'e') && |
133 | p[1] == '\'') |
134 | { |
135 | quote = "'" ; |
136 | escape = '\\'; /* if std strings before, not any more */ |
137 | p++; |
138 | } |
139 | |
140 | /* test if quoting character */ |
141 | if (quote && strchr(quote, *p)) |
142 | { |
143 | /* okay, we have a quoted token, now scan for the closer */ |
144 | char thisquote = *p++; |
145 | |
146 | for (; *p; p += PQmblen(p, encoding)) |
147 | { |
148 | if (*p == escape && p[1] != '\0') |
149 | p++; /* process escaped anything */ |
150 | else if (*p == thisquote && p[1] == thisquote) |
151 | p++; /* process doubled quote */ |
152 | else if (*p == thisquote) |
153 | { |
154 | p++; /* skip trailing quote */ |
155 | break; |
156 | } |
157 | } |
158 | |
159 | /* |
160 | * If not at end of string, we need to insert a null to terminate the |
161 | * returned token. See notes above. |
162 | */ |
163 | if (*p != '\0') |
164 | { |
165 | if (!strchr(whitespace, *p)) |
166 | memmove(p + 1, p, strlen(p) + 1); |
167 | *p = '\0'; |
168 | string = p + 1; |
169 | } |
170 | else |
171 | { |
172 | /* at end of string, so no extra work */ |
173 | string = p; |
174 | } |
175 | |
176 | /* Clean up the token if caller wants that */ |
177 | if (del_quotes) |
178 | strip_quotes(start, thisquote, escape, encoding); |
179 | |
180 | return start; |
181 | } |
182 | |
183 | /* |
184 | * Otherwise no quoting character. Scan till next whitespace, delimiter |
185 | * or quote. NB: at this point, *start is known not to be '\0', |
186 | * whitespace, delim, or quote, so we will consume at least one character. |
187 | */ |
188 | offset = strcspn(start, whitespace); |
189 | |
190 | if (delim) |
191 | { |
192 | unsigned int offset2 = strcspn(start, delim); |
193 | |
194 | if (offset > offset2) |
195 | offset = offset2; |
196 | } |
197 | |
198 | if (quote) |
199 | { |
200 | unsigned int offset2 = strcspn(start, quote); |
201 | |
202 | if (offset > offset2) |
203 | offset = offset2; |
204 | } |
205 | |
206 | p = start + offset; |
207 | |
208 | /* |
209 | * If not at end of string, we need to insert a null to terminate the |
210 | * returned token. See notes above. |
211 | */ |
212 | if (*p != '\0') |
213 | { |
214 | if (!strchr(whitespace, *p)) |
215 | memmove(p + 1, p, strlen(p) + 1); |
216 | *p = '\0'; |
217 | string = p + 1; |
218 | } |
219 | else |
220 | { |
221 | /* at end of string, so no extra work */ |
222 | string = p; |
223 | } |
224 | |
225 | return start; |
226 | } |
227 | |
228 | |
229 | /* |
230 | * strip_quotes |
231 | * |
232 | * Remove quotes from the string at *source. Leading and trailing occurrences |
233 | * of 'quote' are removed; embedded double occurrences of 'quote' are reduced |
234 | * to single occurrences; if 'escape' is not 0 then 'escape' removes special |
235 | * significance of next character. |
236 | * |
237 | * Note that the source string is overwritten in-place. |
238 | */ |
239 | void |
240 | strip_quotes(char *source, char quote, char escape, int encoding) |
241 | { |
242 | char *src; |
243 | char *dst; |
244 | |
245 | Assert(source != NULL); |
246 | Assert(quote != '\0'); |
247 | |
248 | src = dst = source; |
249 | |
250 | if (*src && *src == quote) |
251 | src++; /* skip leading quote */ |
252 | |
253 | while (*src) |
254 | { |
255 | char c = *src; |
256 | int i; |
257 | |
258 | if (c == quote && src[1] == '\0') |
259 | break; /* skip trailing quote */ |
260 | else if (c == quote && src[1] == quote) |
261 | src++; /* process doubled quote */ |
262 | else if (c == escape && src[1] != '\0') |
263 | src++; /* process escaped character */ |
264 | |
265 | i = PQmblen(src, encoding); |
266 | while (i--) |
267 | *dst++ = *src++; |
268 | } |
269 | |
270 | *dst = '\0'; |
271 | } |
272 | |
273 | |
274 | /* |
275 | * quote_if_needed |
276 | * |
277 | * Opposite of strip_quotes(). If "source" denotes itself literally without |
278 | * quoting or escaping, returns NULL. Otherwise, returns a malloc'd copy with |
279 | * quoting and escaping applied: |
280 | * |
281 | * source - string to parse |
282 | * entails_quote - any of these present? need outer quotes |
283 | * quote - doubled within string, affixed to both ends |
284 | * escape - doubled within string |
285 | * encoding - the active character-set encoding |
286 | * |
287 | * Do not use this as a substitute for PQescapeStringConn(). Use it for |
288 | * strings to be parsed by strtokx() or psql_scan_slash_option(). |
289 | */ |
290 | char * |
291 | quote_if_needed(const char *source, const char *entails_quote, |
292 | char quote, char escape, int encoding) |
293 | { |
294 | const char *src; |
295 | char *ret; |
296 | char *dst; |
297 | bool need_quotes = false; |
298 | |
299 | Assert(source != NULL); |
300 | Assert(quote != '\0'); |
301 | |
302 | src = source; |
303 | dst = ret = pg_malloc(2 * strlen(src) + 3); /* excess */ |
304 | |
305 | *dst++ = quote; |
306 | |
307 | while (*src) |
308 | { |
309 | char c = *src; |
310 | int i; |
311 | |
312 | if (c == quote) |
313 | { |
314 | need_quotes = true; |
315 | *dst++ = quote; |
316 | } |
317 | else if (c == escape) |
318 | { |
319 | need_quotes = true; |
320 | *dst++ = escape; |
321 | } |
322 | else if (strchr(entails_quote, c)) |
323 | need_quotes = true; |
324 | |
325 | i = PQmblen(src, encoding); |
326 | while (i--) |
327 | *dst++ = *src++; |
328 | } |
329 | |
330 | *dst++ = quote; |
331 | *dst = '\0'; |
332 | |
333 | if (!need_quotes) |
334 | { |
335 | free(ret); |
336 | ret = NULL; |
337 | } |
338 | |
339 | return ret; |
340 | } |
341 | |