1 |
dpavlin |
8 |
/* |
2 |
|
|
* integrate swish-e into PostgreSQL |
3 |
|
|
* |
4 |
|
|
* Dobrica Pavlinusic <dpavlin@rot13.org> 2005-02-18 |
5 |
|
|
* |
6 |
|
|
* TODO: |
7 |
|
|
* - check null input using PG_ARGISNULL before using PG_GETARG_xxxx |
8 |
|
|
* - support composite type arguments |
9 |
|
|
* |
10 |
|
|
* NOTES: |
11 |
|
|
* - clear structures with memset to support hash indexes (who whould like |
12 |
|
|
* to create hash index on table returned from function?) |
13 |
dpavlin |
10 |
* - number of returned rows is set by PostgreSQL evaluator, see: |
14 |
|
|
* http://archives.postgresql.org/pgsql-hackers/2005-02/msg00546.php |
15 |
dpavlin |
8 |
* |
16 |
dpavlin |
9 |
* Based on: |
17 |
|
|
* - C example from PostgreSQL documentation (BSD licence) |
18 |
|
|
* - swish-e example src/libtest.c (GPL) |
19 |
|
|
* - _textin/_textout from pgcurl.c (LGPL) |
20 |
|
|
* |
21 |
|
|
* This code is licenced under GPL |
22 |
dpavlin |
8 |
*/ |
23 |
|
|
|
24 |
|
|
#include "postgres.h" |
25 |
|
|
#include "fmgr.h" |
26 |
|
|
#include "funcapi.h" |
27 |
dpavlin |
9 |
#include "utils/builtins.h" |
28 |
dpavlin |
16 |
#include "utils/array.h" |
29 |
|
|
#include "miscadmin.h" |
30 |
dpavlin |
8 |
#include <swish-e.h> |
31 |
|
|
|
32 |
dpavlin |
9 |
#define _textin(str) DirectFunctionCall1(textin, CStringGetDatum(str)) |
33 |
|
|
#define _textout(str) DatumGetPointer(DirectFunctionCall1(textout, PointerGetDatum(str))) |
34 |
dpavlin |
11 |
#define GET_STR(textp) DatumGetCString(DirectFunctionCall1(textout, PointerGetDatum(textp))) |
35 |
|
|
#define GET_TEXT(cstrp) DatumGetTextP(DirectFunctionCall1(textin, CStringGetDatum(cstrp))) |
36 |
dpavlin |
8 |
|
37 |
dpavlin |
9 |
|
38 |
dpavlin |
8 |
SW_HANDLE swish_handle = NULL;/* Database handle */ |
39 |
|
|
SW_SEARCH search = NULL; /* search handle -- holds search parameters */ |
40 |
dpavlin |
11 |
SW_RESULTS swish_results = NULL; /* results handle -- holds list of results */ |
41 |
dpavlin |
14 |
SW_RESULT *sw_res = NULL; /* one row from swish-e results */ |
42 |
dpavlin |
8 |
|
43 |
|
|
/* define PostgreSQL v1 function */ |
44 |
|
|
PG_FUNCTION_INFO_V1(pgswish); |
45 |
|
|
Datum pgswish(PG_FUNCTION_ARGS) { |
46 |
|
|
|
47 |
|
|
FuncCallContext *funcctx; |
48 |
|
|
int call_cntr; |
49 |
|
|
int max_calls; |
50 |
|
|
TupleDesc tupdesc; |
51 |
|
|
TupleTableSlot *slot; |
52 |
|
|
AttInMetadata *attinmeta; |
53 |
dpavlin |
9 |
char *index_path; |
54 |
|
|
char *query; |
55 |
dpavlin |
8 |
|
56 |
|
|
/* stuff done only on the first call of the function */ |
57 |
|
|
if (SRF_IS_FIRSTCALL()) { |
58 |
|
|
MemoryContext oldcontext; |
59 |
|
|
|
60 |
dpavlin |
9 |
/* take arguments from function */ |
61 |
|
|
//index_path = _textout(PG_GETARG_TEXT_P(0)); |
62 |
|
|
index_path = _textout(PG_GETARG_TEXT_P(0)); |
63 |
|
|
query = _textout(PG_GETARG_TEXT_P(1)); |
64 |
|
|
|
65 |
dpavlin |
8 |
/* create a function context for cross-call persistence */ |
66 |
|
|
funcctx = SRF_FIRSTCALL_INIT(); |
67 |
|
|
|
68 |
|
|
/* switch to memory context appropriate for multiple function calls */ |
69 |
|
|
oldcontext = MemoryContextSwitchTo(funcctx->multi_call_memory_ctx); |
70 |
|
|
|
71 |
dpavlin |
9 |
|
72 |
dpavlin |
8 |
/* Send any errors or warnings to stderr (default is stdout) */ |
73 |
|
|
SwishErrorsToStderr(); |
74 |
|
|
|
75 |
dpavlin |
9 |
elog(INFO, "pgswish: SwishInit(%s)", index_path); |
76 |
dpavlin |
8 |
|
77 |
dpavlin |
9 |
swish_handle = SwishInit( index_path ); |
78 |
dpavlin |
8 |
|
79 |
|
|
if (! swish_handle) { |
80 |
dpavlin |
9 |
elog(ERROR, "pgswish: can't open %s", index_path); |
81 |
dpavlin |
8 |
SRF_RETURN_DONE(funcctx); |
82 |
|
|
} |
83 |
|
|
|
84 |
dpavlin |
12 |
error_or_abort( swish_handle ); |
85 |
dpavlin |
8 |
/* set ranking scheme. default is 0 */ |
86 |
dpavlin |
10 |
SwishRankScheme( swish_handle, 0 ); |
87 |
dpavlin |
12 |
error_or_abort( swish_handle ); |
88 |
dpavlin |
8 |
|
89 |
dpavlin |
9 |
elog(INFO, "pgswish: SwishQuery(%s)", query); |
90 |
dpavlin |
8 |
/* Here's a short-cut to searching that creates a search object and searches at the same time */ |
91 |
dpavlin |
11 |
swish_results = SwishQuery( swish_handle, query); |
92 |
dpavlin |
12 |
error_or_abort( swish_handle ); |
93 |
dpavlin |
8 |
|
94 |
|
|
/* total number of tuples to be returned */ |
95 |
dpavlin |
11 |
funcctx->max_calls = SwishHits( swish_results ); |
96 |
dpavlin |
8 |
|
97 |
|
|
/* check if results exists */ |
98 |
|
|
if ( 0 == funcctx->max_calls ) |
99 |
dpavlin |
9 |
elog(INFO, "no results for: %s", query ); |
100 |
dpavlin |
8 |
|
101 |
dpavlin |
9 |
elog(INFO, "pgswish: SwishHits = %d", funcctx->max_calls); |
102 |
|
|
|
103 |
dpavlin |
8 |
/* Build a tuple description for a __pgswish tuple */ |
104 |
|
|
tupdesc = RelationNameGetTupleDesc("__pgswish"); |
105 |
|
|
|
106 |
|
|
/* allocate a slot for a tuple with this tupdesc */ |
107 |
|
|
slot = TupleDescGetSlot(tupdesc); |
108 |
|
|
|
109 |
|
|
/* assign slot to function context */ |
110 |
|
|
funcctx->slot = slot; |
111 |
|
|
|
112 |
|
|
/* |
113 |
|
|
* generate attribute metadata needed later to produce tuples from raw |
114 |
|
|
* C strings |
115 |
|
|
*/ |
116 |
|
|
attinmeta = TupleDescGetAttInMetadata(tupdesc); |
117 |
|
|
funcctx->attinmeta = attinmeta; |
118 |
|
|
|
119 |
|
|
MemoryContextSwitchTo(oldcontext); |
120 |
dpavlin |
13 |
|
121 |
|
|
elog(INFO, "SRF_IS_FIRSTCALL done"); |
122 |
dpavlin |
8 |
} |
123 |
|
|
|
124 |
|
|
/* stuff done on every call of the function */ |
125 |
|
|
funcctx = SRF_PERCALL_SETUP(); |
126 |
|
|
|
127 |
|
|
call_cntr = funcctx->call_cntr; |
128 |
|
|
max_calls = funcctx->max_calls; |
129 |
|
|
slot = funcctx->slot; |
130 |
|
|
attinmeta = funcctx->attinmeta; |
131 |
|
|
|
132 |
|
|
if (call_cntr < max_calls) { |
133 |
|
|
char **values; |
134 |
|
|
HeapTuple tuple; |
135 |
|
|
Datum result; |
136 |
|
|
|
137 |
dpavlin |
13 |
elog(INFO, "pgswish: loop count %d", call_cntr); |
138 |
dpavlin |
11 |
|
139 |
|
|
if (! swish_results) { |
140 |
|
|
elog(ERROR, "pgswish: no swish-e results"); |
141 |
|
|
SRF_RETURN_DONE(funcctx); |
142 |
|
|
} |
143 |
|
|
|
144 |
dpavlin |
14 |
elog(DEBUG1, "pgswish: check for swish-e error"); |
145 |
dpavlin |
12 |
error_or_abort( swish_handle ); |
146 |
dpavlin |
11 |
|
147 |
dpavlin |
8 |
/* |
148 |
|
|
* Prepare a values array for storage in our slot. |
149 |
|
|
* This should be an array of C strings which will |
150 |
|
|
* be processed later by the type input functions. |
151 |
|
|
*/ |
152 |
|
|
|
153 |
dpavlin |
11 |
sw_res = SwishNextResult( swish_results ); |
154 |
|
|
if (! sw_res) { |
155 |
|
|
elog(ERROR, "pgswish: swish-e sort result list: %d rows expected %d", call_cntr, max_calls - 1); |
156 |
|
|
SRF_RETURN_DONE(funcctx); |
157 |
|
|
} |
158 |
|
|
|
159 |
dpavlin |
12 |
elog(INFO, "Path: %s\n Rank: %lu\n Size: %lu\n Title: %s\n Index: %s\n Modified: %s\n Record #: %lu\n File #: %lu\n\n", |
160 |
|
|
SwishResultPropertyStr ( sw_res, "swishdocpath" ), |
161 |
|
|
SwishResultPropertyULong ( sw_res, "swishrank" ), |
162 |
|
|
SwishResultPropertyULong ( sw_res, "swishdocsize" ), |
163 |
|
|
SwishResultPropertyStr ( sw_res, "swishtitle"), |
164 |
|
|
SwishResultPropertyStr ( sw_res, "swishdbfile" ), |
165 |
|
|
SwishResultPropertyStr ( sw_res, "swishlastmodified" ), |
166 |
|
|
SwishResultPropertyULong ( sw_res, "swishreccount" ), /* can figure this out in loop, of course */ |
167 |
|
|
SwishResultPropertyULong ( sw_res, "swishfilenum" ) |
168 |
|
|
); |
169 |
dpavlin |
11 |
|
170 |
dpavlin |
13 |
values = (char **) palloc(4 * sizeof(char *)); |
171 |
|
|
|
172 |
dpavlin |
12 |
values[0] = prop2int( sw_res, "swishrank" ); |
173 |
|
|
values[1] = prop2text( sw_res, "swishdocpath" ); |
174 |
|
|
values[2] = prop2text( sw_res, "swishtitle" ); |
175 |
|
|
values[3] = prop2int( sw_res, "swishdocsize" ); |
176 |
dpavlin |
13 |
|
177 |
|
|
/* |
178 |
|
|
values[0] = (char *) palloc(16 * sizeof(char)); |
179 |
|
|
snprintf(values[0], 16, "%d", 1); |
180 |
|
|
values[1] = (char *) palloc(16 * sizeof(char)); |
181 |
|
|
snprintf(values[1], 16, "%d", 2); |
182 |
|
|
values[2] = (char *) palloc(16 * sizeof(char)); |
183 |
|
|
snprintf(values[2], 16, "%d", 3); |
184 |
|
|
values[3] = (char *) palloc(16 * sizeof(char)); |
185 |
|
|
snprintf(values[3], 16, "%d", 4); |
186 |
|
|
*/ |
187 |
dpavlin |
16 |
|
188 |
dpavlin |
8 |
/* build a tuple */ |
189 |
|
|
tuple = BuildTupleFromCStrings(attinmeta, values); |
190 |
|
|
|
191 |
|
|
/* make the tuple into a datum */ |
192 |
|
|
result = TupleGetDatum(slot, tuple); |
193 |
|
|
|
194 |
dpavlin |
12 |
/* clean up ? */ |
195 |
dpavlin |
13 |
pfree(values[0]); |
196 |
|
|
pfree(values[1]); |
197 |
|
|
pfree(values[2]); |
198 |
|
|
pfree(values[3]); |
199 |
|
|
pfree(values); |
200 |
|
|
|
201 |
dpavlin |
17 |
elog(DEBUG1, "row: %s|%s|%s|%s",values[0],values[1],values[2],values[3]); |
202 |
dpavlin |
12 |
|
203 |
dpavlin |
8 |
SRF_RETURN_NEXT(funcctx, result); |
204 |
|
|
} else { |
205 |
dpavlin |
13 |
elog(INFO, "loop over"); |
206 |
|
|
|
207 |
dpavlin |
8 |
/* free swish object and close */ |
208 |
|
|
Free_Search_Object( search ); |
209 |
|
|
SwishClose( swish_handle ); |
210 |
|
|
|
211 |
|
|
/* do when there is no more left */ |
212 |
|
|
SRF_RETURN_DONE(funcctx); |
213 |
|
|
} |
214 |
|
|
} |
215 |
|
|
|
216 |
dpavlin |
16 |
/* work in progress */ |
217 |
|
|
PG_FUNCTION_INFO_V1(pgswish2); |
218 |
|
|
Datum pgswish2(PG_FUNCTION_ARGS) |
219 |
|
|
{ |
220 |
|
|
int nrows = 3; |
221 |
|
|
int16 typlen; |
222 |
|
|
bool typbyval; |
223 |
|
|
char typalign; |
224 |
|
|
ReturnSetInfo *rsinfo = (ReturnSetInfo *) fcinfo->resultinfo; |
225 |
|
|
AttInMetadata *attinmeta; |
226 |
|
|
TupleDesc tupdesc; |
227 |
|
|
Tuplestorestate *tupstore = NULL; |
228 |
|
|
HeapTuple tuple; |
229 |
|
|
MemoryContext per_query_ctx; |
230 |
|
|
MemoryContext oldcontext; |
231 |
|
|
Datum dvalue; |
232 |
|
|
char **values; |
233 |
dpavlin |
18 |
int ncols; |
234 |
dpavlin |
16 |
int i, j; |
235 |
dpavlin |
8 |
|
236 |
dpavlin |
16 |
/* check to see if caller supports us returning a tuplestore */ |
237 |
|
|
if (!rsinfo || !(rsinfo->allowedModes & SFRM_Materialize)) |
238 |
|
|
ereport(ERROR, |
239 |
|
|
(errcode(ERRCODE_SYNTAX_ERROR), |
240 |
|
|
errmsg("materialize mode required, but it is not " \ |
241 |
|
|
"allowed in this context"))); |
242 |
dpavlin |
14 |
|
243 |
dpavlin |
16 |
/* get the requested return tuple description */ |
244 |
|
|
tupdesc = rsinfo->expectedDesc; |
245 |
dpavlin |
18 |
ncols = tupdesc->natts; |
246 |
dpavlin |
14 |
|
247 |
dpavlin |
16 |
/* |
248 |
|
|
* The requested tuple description better match up with the array |
249 |
|
|
* we were given. |
250 |
|
|
*/ |
251 |
|
|
/* OK, use it */ |
252 |
|
|
attinmeta = TupleDescGetAttInMetadata(tupdesc); |
253 |
|
|
|
254 |
|
|
/* Now go to work */ |
255 |
|
|
rsinfo->returnMode = SFRM_Materialize; |
256 |
|
|
|
257 |
|
|
per_query_ctx = fcinfo->flinfo->fn_mcxt; |
258 |
|
|
oldcontext = MemoryContextSwitchTo(per_query_ctx); |
259 |
|
|
|
260 |
|
|
/* initialize our tuplestore */ |
261 |
|
|
tupstore = tuplestore_begin_heap(true, false, SortMem); |
262 |
|
|
|
263 |
|
|
values = (char **) palloc(ncols * sizeof(char *)); |
264 |
|
|
|
265 |
|
|
for (i = 0; i < nrows; i++) |
266 |
|
|
{ |
267 |
|
|
for (j = 0; j < ncols; j++) |
268 |
|
|
{ |
269 |
|
|
values[j] = DatumGetCString( "foo" ); |
270 |
|
|
} |
271 |
|
|
/* construct the tuple */ |
272 |
|
|
tuple = BuildTupleFromCStrings(attinmeta, values); |
273 |
|
|
|
274 |
|
|
/* now store it */ |
275 |
|
|
tuplestore_puttuple(tupstore, tuple); |
276 |
|
|
} |
277 |
|
|
|
278 |
|
|
tuplestore_donestoring(tupstore); |
279 |
|
|
rsinfo->setResult = tupstore; |
280 |
|
|
|
281 |
|
|
/* |
282 |
|
|
* SFRM_Materialize mode expects us to return a NULL Datum. The actual |
283 |
|
|
* tuples are in our tuplestore and passed back through |
284 |
|
|
* rsinfo->setResult. rsinfo->setDesc is set to the tuple description |
285 |
|
|
* that we actually used to build our tuples with, so the caller can |
286 |
|
|
* verify we did what it was expecting. |
287 |
|
|
*/ |
288 |
|
|
rsinfo->setDesc = tupdesc; |
289 |
|
|
MemoryContextSwitchTo(oldcontext); |
290 |
|
|
|
291 |
|
|
return (Datum) 0; |
292 |
dpavlin |
14 |
} |
293 |
|
|
|
294 |
dpavlin |
16 |
|
295 |
dpavlin |
14 |
/* make text var prom property */ |
296 |
dpavlin |
12 |
char *prop2text(SW_RESULT sw_res, char *propname) { |
297 |
|
|
char *val; |
298 |
|
|
char *prop; |
299 |
|
|
int len; |
300 |
|
|
|
301 |
dpavlin |
17 |
elog(DEBUG2, "prop2text(%s)", propname); |
302 |
dpavlin |
12 |
|
303 |
|
|
prop = SwishResultPropertyStr( sw_res, propname ); |
304 |
|
|
error_or_abort( swish_handle ); |
305 |
|
|
|
306 |
|
|
len = strlen(prop); |
307 |
dpavlin |
17 |
elog(DEBUG1, "prop2text(%s) = '%s' %d bytes", propname, prop, len); |
308 |
dpavlin |
12 |
|
309 |
|
|
len++; |
310 |
|
|
len *= sizeof(char); |
311 |
|
|
|
312 |
dpavlin |
17 |
elog(DEBUG2, "palloc(%d)", len); |
313 |
dpavlin |
12 |
|
314 |
|
|
val = palloc(len); |
315 |
|
|
|
316 |
|
|
memset(val, 0, len); |
317 |
|
|
strncpy(val, prop, len); |
318 |
|
|
|
319 |
dpavlin |
17 |
elog(DEBUG2, "val=%s", val); |
320 |
dpavlin |
12 |
|
321 |
|
|
return val; |
322 |
|
|
} |
323 |
|
|
|
324 |
dpavlin |
14 |
/* make integer variable from property */ |
325 |
dpavlin |
12 |
char *prop2int(SW_RESULT sw_res, char *propname) { |
326 |
|
|
char *val; |
327 |
|
|
unsigned long prop; |
328 |
|
|
int len; |
329 |
|
|
|
330 |
dpavlin |
17 |
elog(DEBUG2, "prop2int(%s)", propname); |
331 |
dpavlin |
12 |
|
332 |
|
|
prop = SwishResultPropertyULong( sw_res, propname ); |
333 |
|
|
error_or_abort( swish_handle ); |
334 |
|
|
|
335 |
dpavlin |
17 |
elog(DEBUG1, "prop2int(%s) = %lu", propname, prop); |
336 |
dpavlin |
12 |
|
337 |
|
|
len = 128 * sizeof(char); |
338 |
dpavlin |
17 |
elog(DEBUG2, "palloc(%d)", len); |
339 |
dpavlin |
12 |
|
340 |
|
|
val = palloc(len); |
341 |
|
|
memset(val, 0, len); |
342 |
|
|
|
343 |
|
|
snprintf(val, len, "%lu", prop); |
344 |
|
|
|
345 |
dpavlin |
17 |
elog(DEBUG2, "val=%s", val); |
346 |
dpavlin |
12 |
|
347 |
|
|
return val; |
348 |
|
|
} |
349 |
|
|
|
350 |
|
|
|
351 |
dpavlin |
14 |
/* |
352 |
|
|
* check if swish has returned error, and elog it. |
353 |
|
|
*/ |
354 |
dpavlin |
8 |
static void error_or_abort( SW_HANDLE swish_handle ) { |
355 |
|
|
if ( !SwishError( swish_handle ) ) |
356 |
|
|
return; |
357 |
|
|
|
358 |
|
|
/* print a message */ |
359 |
|
|
elog(ERROR, |
360 |
|
|
"pgswish error: Number [%d], Type [%s], Optional Message: [%s]\n", |
361 |
|
|
SwishError( swish_handle ), |
362 |
|
|
SwishErrorString( swish_handle ), |
363 |
|
|
SwishLastErrorMsg( swish_handle ) |
364 |
|
|
); |
365 |
|
|
if ( search ) Free_Search_Object( search ); |
366 |
|
|
SwishClose( swish_handle ); |
367 |
|
|
} |
368 |
|
|
|