]> arthur.barton.de Git - netatalk.git/blob - etc/spotlight/slmod_rdf_parser.y
Merge remote-tracking branch 'origin/develop' into spotlight
[netatalk.git] / etc / spotlight / slmod_rdf_parser.y
1 %{
2   #include <atalk/standards.h>
3
4   #include <stdbool.h>
5   #include <stdio.h>
6   #include <string.h>
7   #include <time.h>
8
9   #include <gio/gio.h>
10   #include <tracker.h>
11
12   #include <atalk/talloc.h>
13   #include <atalk/logger.h>
14   #include <atalk/errchk.h>
15   #include <atalk/spotlight.h>
16
17   #include "slmod_rdf_map.h"
18
19   struct yy_buffer_state;
20   typedef struct yy_buffer_state *YY_BUFFER_STATE;
21   extern int yylex (void);
22   extern void yyerror (char const *);
23   extern void *yyterminate(void);
24   extern YY_BUFFER_STATE yy_scan_string( const char *str);
25   extern void yy_delete_buffer ( YY_BUFFER_STATE buffer );
26
27   /* forward declarations */
28   static const char *map_expr(const char *attr, char op, const char *val);
29   static const char *map_daterange(const char *dateattr, time_t date1, time_t date2);
30   static time_t isodate2unix(const char *s);
31  
32  /* global vars, eg needed by the lexer */
33   slq_t *srp_slq;
34
35   /* local vars */
36   static gchar *srp_result;
37   static gchar *srp_fts;
38 %}
39
40 %code provides {
41   #define SPRAW_TIME_OFFSET 978307200
42   extern int map_spotlight_to_rdf_query(slq_t *slq);
43   extern slq_t *srp_slq;
44 }
45
46 %union {
47     int ival;
48     const char *sval;
49     bool bval;
50     time_t tval;
51 }
52
53 %expect 4
54 %error-verbose
55
56 %type <sval> match expr line function
57 %type <tval> date
58
59 %token <sval> WORD
60 %token <bval> BOOL
61 %token FUNC_INRANGE
62 %token DATE_ISO
63 %token OBRACE CBRACE EQUAL UNEQUAL GT LT COMMA QUOTE
64 %left AND
65 %left OR
66 %%
67
68 input:
69 /* empty */
70 | input line
71 ;
72      
73 line:
74 expr                           {
75     srp_result = talloc_asprintf(srp_slq,
76                                  "<rdfq:Condition>\n"
77                                  "  <rdfq:and>\n"
78                                  "    <rdfq:startsWith>\n"
79                                  "      <rdfq:Property name=\"File:Path\" />\n"
80                                  "      <rdf:String>%s</rdf:String>\n"
81                                  "    </rdfq:startsWith>\n"
82                                  "    %s\n"
83                                  "  </rdfq:and>\n"
84                                  "</rdfq:Condition>\n",
85                                  srp_slq->slq_vol->v_path, $1);
86     $$ = srp_result;
87 }
88 ;
89
90 expr:
91 BOOL                             {
92     if ($1 == false)
93         YYACCEPT;
94     else
95         YYABORT;
96 }
97 | match                        {$$ = $1; if ($$ == NULL) YYABORT;}
98 | function                     {$$ = $1;}
99 | OBRACE expr CBRACE           {$$ = talloc_asprintf(srp_slq, "%s\n", $2);}
100 | expr AND expr                {$$ = talloc_asprintf(srp_slq, "<rdfq:and>\n%s\n%s\n</rdfq:and>\n", $1, $3);}
101 | expr OR expr                 {
102     if (strcmp($1, "") == 0 || strcmp($3, "") == 0) {
103         /*
104          * The default Spotlight search term issued by the Finder (10.8) is:
105          * '* == "searchterm" || kMDItemTextContent == "searchterm"'
106          * As it isn't mappable to a single Tracker RDF query, we silently
107          * map ANY FTS query expression being part of an OR compound
108          * expression to a simple filename search.
109          * FTS queries are thus only possible by explicitly requesting
110          * file content FTS search in the Finder on the client (resulting
111          * in a 'kMDItemTextContent == "searchterm"' query).
112          */
113         if (strcmp($1, "") == 0)
114             $$ = talloc_asprintf(srp_slq, $3);
115         else
116             $$ = talloc_asprintf(srp_slq, $1);
117         talloc_free(srp_fts);
118         srp_fts = NULL;
119     } else {
120         $$ = talloc_asprintf(srp_slq, "<rdfq:or>\n%s\n%s\n</rdfq:or>\n", $1, $3);
121     }
122 }
123 ;
124
125 match:
126 WORD EQUAL QUOTE WORD QUOTE     {$$ = map_expr($1, '=', $4);}
127 | WORD UNEQUAL QUOTE WORD QUOTE {$$ = map_expr($1, '!', $4);}
128 | WORD LT QUOTE WORD QUOTE      {$$ = map_expr($1, '<', $4);}
129 | WORD GT QUOTE WORD QUOTE      {$$ = map_expr($1, '>', $4);}
130 | WORD EQUAL QUOTE WORD QUOTE WORD    {$$ = map_expr($1, '=', $4);}
131 | WORD UNEQUAL QUOTE WORD QUOTE WORD {$$ = map_expr($1, '!', $4);}
132 | WORD LT QUOTE WORD QUOTE WORD     {$$ = map_expr($1, '<', $4);}
133 | WORD GT QUOTE WORD QUOTE WORD     {$$ = map_expr($1, '>', $4);}
134 ;
135
136 function:
137 FUNC_INRANGE OBRACE WORD COMMA date COMMA date CBRACE {$$ = map_daterange($3, $5, $7);}
138 ;
139
140 date:
141 DATE_ISO OBRACE WORD CBRACE    {$$ = isodate2unix($3);}
142 | WORD                         {$$ = atoi($1) + SPRAW_TIME_OFFSET;}
143 ;
144
145 %%
146
147 static time_t isodate2unix(const char *s)
148 {
149     struct tm tm;
150
151     if (strptime(s, "%Y-%m-%dT%H:%M:%SZ", &tm) == NULL)
152         return (time_t)-1;
153     return mktime(&tm);
154 }
155
156 static const char *map_daterange(const char *dateattr, time_t date1, time_t date2)
157 {
158     EC_INIT;
159     char *result = NULL;
160     struct spotlight_rdf_map *p;
161     struct tm *tmp;
162     char buf1[64], buf2[64];
163
164     EC_NULL_LOG( tmp = localtime(&date1) );
165     strftime(buf1, sizeof(buf1), "%Y-%m-%dT%H:%M:%SZ", tmp);
166     EC_NULL_LOG( tmp = localtime(&date2) );
167     strftime(buf2, sizeof(buf2), "%Y-%m-%dT%H:%M:%SZ", tmp);
168
169     for (p = spotlight_rdf_map; p->srm_spotlight_attr; p++) {
170         if (strcmp(dateattr, p->srm_spotlight_attr) == 0) {
171                 result = talloc_asprintf(srp_slq,
172                                          "<rdfq:and>\n"
173                                          "  <rdfq:greaterThan>\n"
174                                          "    <rdfq:Property name=\"%s\" />\n"
175                                          "    <rdf:Date>%s</rdf:Date>\n"
176                                          "  </rdfq:greaterThan>\n"
177                                          "  <rdfq:lessThan>\n"
178                                          "    <rdfq:Property name=\"%s\" />\n"
179                                          "    <rdf:Date>%s</rdf:Date>\n"
180                                          "  </rdfq:lessThan>\n"
181                                          "</rdfq:and>\n",
182                                          p->srm_rdf_attr, buf1,
183                                          p->srm_rdf_attr, buf2);
184             break;
185         }
186     }
187
188 EC_CLEANUP:
189     if (ret != 0)
190         return NULL;
191     return result;
192 }
193
194 static char *map_type_search(const char *attr, char op, const char *val)
195 {
196     char *result = NULL;
197
198     for (struct MDTypeMap *p = MDTypeMap; p->mdtm_value; p++) {
199         if (strcmp(p->mdtm_value, val) == 0) {
200             if (!p->mdtm_type)
201                 return NULL;
202             if (val[0] == '9') {
203                 srp_slq->slq_service = SERVICE_FOLDERS;
204                 return "";
205             }
206             result = talloc_asprintf(srp_slq,
207                                      "<rdfq:%s>\n"
208                                      "  <rdfq:Property name=\"File:Mime\" />\n"
209                                      "  <rdf:String>%s</rdf:String>\n"
210                                      "</rdfq:%s>\n",
211                                      p->mdtm_rdfop,
212                                      p->mdtm_type,
213                                      p->mdtm_rdfop);
214             break;
215         }
216     }
217     return result;
218 }
219
220 static const char *map_expr(const char *attr, char op, const char *val)
221 {
222     EC_INIT;
223     char *result = NULL;
224     struct spotlight_rdf_map *p;
225     time_t t;
226     struct tm *tmp;
227     char buf1[64];
228     bstring q = NULL, search = NULL, replace = NULL;
229     char *rdfop;
230
231     for (p = spotlight_rdf_map; p->srm_spotlight_attr; p++) {
232         if (p->srm_rdf_attr && strcmp(p->srm_spotlight_attr, attr) == 0) {
233             switch (p->srm_type) {
234             case srmt_num:
235                 q = bformat("^%s$", val);
236                 search = bfromcstr("*");
237                 replace = bfromcstr(".*");
238                 bfindreplace(q, search, replace, 0);
239                 result = talloc_asprintf(srp_slq,
240                                          "<rdfq:regex>\n"
241                                          "  <rdfq:Property name=\"%s\" />\n"
242                                          "  <rdf:String>%s</rdf:String>\n"
243                                          "</rdfq:regex>\n",
244                                          p->srm_rdf_attr,
245                                          bdata(q));
246                 bdestroy(q);
247                 break;
248
249             case srmt_str:
250                 q = bformat("^%s$", val);
251                 search = bfromcstr("*");
252                 replace = bfromcstr(".*");
253                 bfindreplace(q, search, replace, 0);
254                 result = talloc_asprintf(srp_slq,
255                                          "<rdfq:regex>\n"
256                                          "  <rdfq:Property name=\"%s\" />\n"
257                                          "  <rdf:String>%s</rdf:String>\n"
258                                          "</rdfq:regex>\n",
259                                          p->srm_rdf_attr,
260                                          bdata(q));
261                 bdestroy(q);
262                 break;
263
264             case srmt_fts:
265                 if (srp_fts) {
266                     yyerror("only single fts query allowed");
267                     EC_FAIL;
268                 }
269                 q = bfromcstr(val);
270                 search = bfromcstr("*");
271                 replace = bfromcstr("");
272                 bfindreplace(q, search, replace, 0);
273                 srp_fts = talloc_strdup(srp_slq, bdata(q));
274                 result = "";
275                 break;
276
277             case srmt_date:
278                 t = atoi(val) + SPRAW_TIME_OFFSET;
279                 EC_NULL( tmp = localtime(&t) );
280                 strftime(buf1, sizeof(buf1), "%Y-%m-%dT%H:%M:%SZ", tmp);
281
282                 switch (op) {
283                 case '=':
284                     rdfop = "equals";
285                 case '<':
286                     rdfop = "lessThan";
287                 case '>':
288                     rdfop = "greaterThan";
289                 default:
290                     yyerror("unknown date comparison");
291                     EC_FAIL;
292                 }
293                 result = talloc_asprintf(srp_slq,
294                                          "<rdfq:%s>\n"
295                                          "  <rdfq:Property name=\"%s\" />\n"
296                                          "  <rdf:Date>%s</rdf:Date>\n"
297                                          "</rdfq:%s>\n",
298                                          rdfop,
299                                          p->srm_rdf_attr,
300                                          buf1,
301                                          rdfop);
302
303                 break;
304
305             case srmt_type:
306                 result = map_type_search(attr, op, val);
307                 break;
308
309             default:
310                 yyerror("unknown Spotlight attribute type");
311                 EC_FAIL;
312             }
313             break;
314         }
315     }
316
317 EC_CLEANUP:
318     if (q)
319         bdestroy(q);
320     if (search)
321         bdestroy(search);
322     if (replace)
323         bdestroy(replace);
324     return result;
325 }
326
327 void yyerror(const char *str)
328 {
329 #ifdef MAIN
330     printf("yyerror: %s\n", str);
331 #else
332     LOG(log_error, logtype_sl, "yyerror: %s", str);
333 #endif
334 }
335  
336 int yywrap()
337 {
338     return 1;
339
340
341 /**
342  * Map a Spotlight RAW query string to a RDF query
343  *
344  * @param[in]     slq            Spotlight query handle
345  * @return        0 on success, -1 on error
346  **/
347 int map_spotlight_to_rdf_query(slq_t *slq)
348 {
349     EC_INIT;
350     YY_BUFFER_STATE s = NULL;
351     srp_result = NULL;
352     srp_fts = NULL;
353     slq->slq_service = SERVICE_FILES;
354     srp_slq = slq;
355     s = yy_scan_string(slq->slq_qstring);
356
357     EC_ZERO( yyparse() );
358
359 EC_CLEANUP:
360     if (s)
361         yy_delete_buffer(s);
362     if (ret == 0) {
363         slq->slq_trackerquery = srp_result;
364         slq->slq_fts = srp_fts;
365     }
366     EC_EXIT;
367 }
368
369 #ifdef MAIN
370 int main(int argc, char **argv)
371 {
372     int ret;
373     YY_BUFFER_STATE s;
374
375     if (argc != 2) {
376         printf("usage: %s QUERY\n", argv[0]);
377         return 1;
378     }
379
380     srp_slq = talloc_zero(NULL, slq_t);
381     struct vol *vol = talloc_zero(srp_slq, struct vol);
382     vol->v_path = "/Volumes/test";
383     srp_slq->slq_vol = vol;
384
385     s = yy_scan_string(argv[1]);
386
387     ret = yyparse();
388
389     yy_delete_buffer(s);
390
391     if (ret == 0)
392         printf("RDF:\n%s\nFTS: %s\n",
393                srp_result ? srp_result : "(empty)",
394                srp_fts ? srp_fts : "(none)");
395     return 0;
396
397 #endif