]>
Commit | Line | Data |
---|---|---|
ef416fc2 | 1 | /* |
7e86f2f6 | 2 | * Search routines for CUPS. |
ef416fc2 | 3 | * |
3416fe95 | 4 | * Copyright 2007-2018 by Apple Inc. |
7e86f2f6 | 5 | * Copyright 1997-2006 by Easy Software Products. |
ef416fc2 | 6 | * |
3416fe95 MS |
7 | * Licensed under Apache License v2.0. See the file "LICENSE" for more |
8 | * information. | |
ef416fc2 | 9 | */ |
10 | ||
11 | /* | |
12 | * Include necessary headers... | |
13 | */ | |
14 | ||
15 | #include "cgi-private.h" | |
16 | #include <regex.h> | |
17 | ||
18 | ||
19 | /* | |
20 | * 'cgiCompileSearch()' - Compile a search string. | |
21 | */ | |
22 | ||
23 | void * /* O - Search context */ | |
24 | cgiCompileSearch(const char *query) /* I - Query string */ | |
25 | { | |
26 | regex_t *re; /* Regular expression */ | |
27 | char *s, /* Regular expression string */ | |
28 | *sptr, /* Pointer into RE string */ | |
29 | *sword; /* Pointer to start of word */ | |
7e86f2f6 | 30 | size_t slen; /* Allocated size of RE string */ |
ef416fc2 | 31 | const char *qptr, /* Pointer into query string */ |
32 | *qend; /* End of current word */ | |
33 | const char *prefix; /* Prefix to add to next word */ | |
34 | int quoted; /* Word is quoted */ | |
7e86f2f6 | 35 | size_t wlen; /* Word length */ |
ef416fc2 | 36 | char *lword; /* Last word in query */ |
37 | ||
38 | ||
1f0275e3 MS |
39 | DEBUG_printf(("cgiCompileSearch(query=\"%s\")\n", query)); |
40 | ||
41 | /* | |
42 | * Range check input... | |
43 | */ | |
44 | ||
45 | if (!query) | |
46 | return (NULL); | |
ef416fc2 | 47 | |
48 | /* | |
49 | * Allocate a regular expression storage structure... | |
50 | */ | |
51 | ||
91c84a35 MS |
52 | if ((re = (regex_t *)calloc(1, sizeof(regex_t))) == NULL) |
53 | return (NULL); | |
ef416fc2 | 54 | |
55 | /* | |
56 | * Allocate a buffer to hold the regular expression string, starting | |
57 | * at 1024 bytes or 3 times the length of the query string, whichever | |
58 | * is greater. We'll expand the string as needed... | |
59 | */ | |
60 | ||
61 | slen = strlen(query) * 3; | |
62 | if (slen < 1024) | |
63 | slen = 1024; | |
64 | ||
91c84a35 MS |
65 | if ((s = (char *)malloc(slen)) == NULL) |
66 | { | |
67 | free(re); | |
68 | return (NULL); | |
69 | } | |
ef416fc2 | 70 | |
71 | /* | |
72 | * Copy the query string to the regular expression, handling basic | |
73 | * AND and OR logic... | |
74 | */ | |
75 | ||
76 | prefix = ".*"; | |
77 | qptr = query; | |
78 | sptr = s; | |
79 | lword = NULL; | |
80 | ||
81 | while (*qptr) | |
82 | { | |
83 | /* | |
84 | * Skip leading whitespace... | |
85 | */ | |
86 | ||
87 | while (isspace(*qptr & 255)) | |
88 | qptr ++; | |
89 | ||
90 | if (!*qptr) | |
91 | break; | |
92 | ||
93 | /* | |
94 | * Find the end of the current word... | |
95 | */ | |
96 | ||
97 | if (*qptr == '\"' || *qptr == '\'') | |
98 | { | |
99 | /* | |
100 | * Scan quoted string... | |
101 | */ | |
102 | ||
103 | quoted = *qptr ++; | |
104 | for (qend = qptr; *qend && *qend != quoted; qend ++); | |
105 | ||
106 | if (!*qend) | |
107 | { | |
108 | /* | |
109 | * No closing quote, error out! | |
110 | */ | |
111 | ||
112 | free(s); | |
113 | free(re); | |
114 | ||
115 | if (lword) | |
116 | free(lword); | |
117 | ||
118 | return (NULL); | |
119 | } | |
120 | } | |
121 | else | |
122 | { | |
123 | /* | |
124 | * Scan whitespace-delimited string... | |
125 | */ | |
126 | ||
127 | quoted = 0; | |
128 | for (qend = qptr + 1; *qend && !isspace(*qend); qend ++); | |
129 | } | |
130 | ||
7e86f2f6 | 131 | wlen = (size_t)(qend - qptr); |
ef416fc2 | 132 | |
133 | /* | |
134 | * Look for logic words: AND, OR | |
135 | */ | |
136 | ||
88f9aafc | 137 | if (wlen == 3 && !_cups_strncasecmp(qptr, "AND", 3)) |
ef416fc2 | 138 | { |
139 | /* | |
140 | * Logical AND with the following text... | |
141 | */ | |
142 | ||
143 | if (sptr > s) | |
144 | prefix = ".*"; | |
145 | ||
146 | qptr = qend; | |
147 | } | |
88f9aafc | 148 | else if (wlen == 2 && !_cups_strncasecmp(qptr, "OR", 2)) |
ef416fc2 | 149 | { |
150 | /* | |
151 | * Logical OR with the following text... | |
152 | */ | |
153 | ||
154 | if (sptr > s) | |
155 | prefix = ".*|.*"; | |
156 | ||
157 | qptr = qend; | |
158 | } | |
159 | else | |
160 | { | |
161 | /* | |
162 | * Add a search word, making sure we have enough room for the | |
163 | * string + RE overhead... | |
164 | */ | |
165 | ||
7e86f2f6 | 166 | wlen = (size_t)(sptr - s) + 2 * 4 * wlen + 2 * strlen(prefix) + 11; |
839a51c8 MS |
167 | if (lword) |
168 | wlen += strlen(lword); | |
ef416fc2 | 169 | |
170 | if (wlen > slen) | |
171 | { | |
172 | /* | |
173 | * Expand the RE string buffer... | |
174 | */ | |
175 | ||
176 | char *temp; /* Temporary string pointer */ | |
177 | ||
178 | ||
179 | slen = wlen + 128; | |
180 | temp = (char *)realloc(s, slen); | |
181 | if (!temp) | |
182 | { | |
183 | free(s); | |
184 | free(re); | |
185 | ||
186 | if (lword) | |
187 | free(lword); | |
188 | ||
189 | return (NULL); | |
190 | } | |
191 | ||
192 | sptr = temp + (sptr - s); | |
193 | s = temp; | |
194 | } | |
195 | ||
196 | /* | |
197 | * Add the prefix string... | |
198 | */ | |
199 | ||
5a9febac | 200 | memcpy(sptr, prefix, strlen(prefix) + 1); |
ef416fc2 | 201 | sptr += strlen(sptr); |
202 | ||
203 | /* | |
204 | * Then quote the remaining word characters as needed for the | |
205 | * RE... | |
206 | */ | |
207 | ||
208 | sword = sptr; | |
209 | ||
210 | while (qptr < qend) | |
211 | { | |
212 | /* | |
213 | * Quote: ^ . [ $ ( ) | * + ? { \ | |
214 | */ | |
215 | ||
216 | if (strchr("^.[$()|*+?{\\", *qptr)) | |
217 | *sptr++ = '\\'; | |
218 | ||
219 | *sptr++ = *qptr++; | |
220 | } | |
221 | ||
07725fee | 222 | *sptr = '\0'; |
223 | ||
ef416fc2 | 224 | /* |
225 | * For "word1 AND word2", add reciprocal "word2 AND word1"... | |
226 | */ | |
227 | ||
228 | if (!strcmp(prefix, ".*") && lword) | |
229 | { | |
230 | char *lword2; /* New "last word" */ | |
231 | ||
232 | ||
91c84a35 MS |
233 | if ((lword2 = strdup(sword)) == NULL) |
234 | { | |
235 | free(lword); | |
236 | free(s); | |
237 | free(re); | |
238 | return (NULL); | |
239 | } | |
ef416fc2 | 240 | |
5a9febac | 241 | memcpy(sptr, ".*|.*", 6); |
ef416fc2 | 242 | sptr += 5; |
243 | ||
5a9febac | 244 | memcpy(sptr, lword2, strlen(lword2) + 1); |
ef416fc2 | 245 | sptr += strlen(sptr); |
246 | ||
5a9febac | 247 | memcpy(sptr, ".*", 3); |
ef416fc2 | 248 | sptr += 2; |
249 | ||
5a9febac | 250 | memcpy(sptr, lword, strlen(lword) + 1); |
ef416fc2 | 251 | sptr += strlen(sptr); |
252 | ||
253 | free(lword); | |
254 | lword = lword2; | |
255 | } | |
256 | else | |
257 | { | |
258 | if (lword) | |
259 | free(lword); | |
260 | ||
261 | lword = strdup(sword); | |
262 | } | |
263 | ||
264 | prefix = ".*|.*"; | |
265 | } | |
266 | ||
267 | /* | |
268 | * Advance to the next string... | |
269 | */ | |
270 | ||
271 | if (quoted) | |
272 | qptr ++; | |
273 | } | |
274 | ||
275 | if (lword) | |
276 | free(lword); | |
277 | ||
278 | if (sptr > s) | |
5a9febac | 279 | memcpy(sptr, ".*", 3); |
ef416fc2 | 280 | else |
281 | { | |
282 | /* | |
283 | * No query data, return NULL... | |
284 | */ | |
285 | ||
286 | free(s); | |
287 | free(re); | |
288 | ||
289 | return (NULL); | |
290 | } | |
291 | ||
292 | /* | |
293 | * Compile the regular expression... | |
294 | */ | |
295 | ||
296 | DEBUG_printf((" s=\"%s\"\n", s)); | |
297 | ||
298 | if (regcomp(re, s, REG_EXTENDED | REG_ICASE)) | |
299 | { | |
300 | free(re); | |
301 | free(s); | |
302 | ||
303 | return (NULL); | |
304 | } | |
305 | ||
306 | /* | |
307 | * Free the RE string and return the new regular expression we compiled... | |
308 | */ | |
309 | ||
310 | free(s); | |
311 | ||
312 | return ((void *)re); | |
313 | } | |
314 | ||
315 | ||
316 | /* | |
317 | * 'cgiDoSearch()' - Do a search of some text. | |
318 | */ | |
319 | ||
320 | int /* O - Number of matches */ | |
321 | cgiDoSearch(void *search, /* I - Search context */ | |
322 | const char *text) /* I - Text to search */ | |
323 | { | |
324 | int i; /* Looping var */ | |
325 | regmatch_t matches[100]; /* RE matches */ | |
326 | ||
327 | ||
328 | /* | |
329 | * Range check... | |
330 | */ | |
331 | ||
332 | if (!search || !text) | |
333 | return (0); | |
334 | ||
335 | /* | |
336 | * Do a lookup... | |
337 | */ | |
338 | ||
339 | if (!regexec((regex_t *)search, text, sizeof(matches) / sizeof(matches[0]), | |
340 | matches, 0)) | |
341 | { | |
342 | /* | |
343 | * Figure out the number of matches in the string... | |
344 | */ | |
345 | ||
346 | for (i = 0; i < (int)(sizeof(matches) / sizeof(matches[0])); i ++) | |
347 | if (matches[i].rm_so < 0) | |
348 | break; | |
349 | ||
350 | return (i); | |
351 | } | |
352 | else | |
353 | return (0); | |
354 | } | |
355 | ||
356 | ||
357 | /* | |
358 | * 'cgiFreeSearch()' - Free a compiled search context. | |
359 | */ | |
360 | ||
361 | void | |
362 | cgiFreeSearch(void *search) /* I - Search context */ | |
363 | { | |
364 | regfree((regex_t *)search); | |
3416fe95 | 365 | free(search); |
ef416fc2 | 366 | } |