8 |
* - support composite type arguments |
* - support composite type arguments |
9 |
* - split error_or_abort |
* - split error_or_abort |
10 |
* - use getResultPropValue not SwishResultPropertyStr |
* - use getResultPropValue not SwishResultPropertyStr |
11 |
|
* - fix everything about pgswish_arr which is broken |
12 |
* |
* |
13 |
* NOTES: |
* NOTES: |
14 |
* - clear structures with memset to support hash indexes (who whould like |
* - clear structures with memset to support hash indexes (who whould like |
234 |
} |
} |
235 |
|
|
236 |
|
|
237 |
|
/* |
238 |
|
* new function with support for property selection |
239 |
|
*/ |
240 |
|
|
241 |
|
PG_FUNCTION_INFO_V1(pgswish_arr); |
242 |
|
Datum pgswish_arr(PG_FUNCTION_ARGS) |
243 |
|
{ |
244 |
|
ArrayType *prop_arr = PG_GETARG_ARRAYTYPE_P(5); |
245 |
|
Oid prop_element_type = ARR_ELEMTYPE(prop_arr); |
246 |
|
int prop_ndims = ARR_NDIM(prop_arr); |
247 |
|
int *prop_dim_counts = ARR_DIMS(prop_arr); |
248 |
|
int *prop_dim_lower_bounds = ARR_LBOUND(prop_arr); |
249 |
|
int ncols = 0; |
250 |
|
int nrows = 0; |
251 |
|
int indx[MAXDIM]; |
252 |
|
int16 prop_len; |
253 |
|
bool prop_byval; |
254 |
|
char prop_align; |
255 |
|
ReturnSetInfo *rsinfo = (ReturnSetInfo *) fcinfo->resultinfo; |
256 |
|
AttInMetadata *attinmeta; |
257 |
|
TupleDesc tupdesc; |
258 |
|
Tuplestorestate *tupstore = NULL; |
259 |
|
HeapTuple tuple; |
260 |
|
MemoryContext per_query_ctx; |
261 |
|
MemoryContext oldcontext; |
262 |
|
Datum dvalue; |
263 |
|
char **values; |
264 |
|
int rsinfo_ncols; |
265 |
|
int i, j; |
266 |
|
/* swish-e */ |
267 |
|
FILE *logfh; |
268 |
|
int resnum; |
269 |
|
int limit = 0; |
270 |
|
int offset = 0; |
271 |
|
|
272 |
|
char *index_path; |
273 |
|
char *query; |
274 |
|
char *attr; |
275 |
|
|
276 |
|
|
277 |
|
/* only allow 1D input array */ |
278 |
|
if (prop_ndims == 1) |
279 |
|
{ |
280 |
|
ncols = prop_dim_counts[0]; |
281 |
|
} |
282 |
|
else |
283 |
|
ereport(ERROR, |
284 |
|
(errcode(ERRCODE_INVALID_PARAMETER_VALUE), |
285 |
|
errmsg("invalid input array"), |
286 |
|
errdetail("Input array must have 1 dimension"))); |
287 |
|
|
288 |
|
/* check to see if caller supports us returning a tuplestore */ |
289 |
|
if (!rsinfo || !(rsinfo->allowedModes & SFRM_Materialize)) |
290 |
|
ereport(ERROR, |
291 |
|
(errcode(ERRCODE_SYNTAX_ERROR), |
292 |
|
errmsg("materialize mode required, but it is not " \ |
293 |
|
"allowed in this context"))); |
294 |
|
|
295 |
|
/* get info about element type needed to construct the array */ |
296 |
|
get_typlenbyvalalign(prop_element_type, &prop_len, &prop_byval, &prop_align); |
297 |
|
|
298 |
|
/* get the requested return tuple description */ |
299 |
|
tupdesc = rsinfo->expectedDesc; |
300 |
|
rsinfo_ncols = tupdesc->natts; |
301 |
|
|
302 |
|
/* |
303 |
|
* The requested tuple description better match up with the array |
304 |
|
* we were given. |
305 |
|
*/ |
306 |
|
if (rsinfo_ncols != ncols) |
307 |
|
ereport(ERROR, |
308 |
|
(errcode(ERRCODE_INVALID_PARAMETER_VALUE), |
309 |
|
errmsg("invalid input array"), |
310 |
|
errdetail("Number of elements in array must match number of query specified columns."))); |
311 |
|
|
312 |
|
/* OK, use it */ |
313 |
|
attinmeta = TupleDescGetAttInMetadata(tupdesc); |
314 |
|
|
315 |
|
/* Now go to work */ |
316 |
|
rsinfo->returnMode = SFRM_Materialize; |
317 |
|
|
318 |
|
per_query_ctx = fcinfo->flinfo->fn_mcxt; |
319 |
|
oldcontext = MemoryContextSwitchTo(per_query_ctx); |
320 |
|
|
321 |
|
/* initialize our tuplestore */ |
322 |
|
tupstore = tuplestore_begin_heap(true, false, SortMem); |
323 |
|
|
324 |
|
|
325 |
|
/* take rest of arguments from function */ |
326 |
|
|
327 |
|
/* index path */ |
328 |
|
if (PG_ARGISNULL(0)) { |
329 |
|
ereport(ERROR, |
330 |
|
(errcode(ERRCODE_INVALID_PARAMETER_VALUE), |
331 |
|
errmsg("index path can't be null"), |
332 |
|
errdetail("Index path must be valid full path to swish-e index"))); |
333 |
|
} |
334 |
|
index_path = _textout(PG_GETARG_TEXT_P(0)); |
335 |
|
|
336 |
|
/* query string */ |
337 |
|
if (PG_ARGISNULL(0)) { |
338 |
|
query = ""; |
339 |
|
} else { |
340 |
|
query = _textout(PG_GETARG_TEXT_P(1)); |
341 |
|
} |
342 |
|
|
343 |
|
/* atribute filter */ |
344 |
|
if (PG_ARGISNULL(2)) { |
345 |
|
attr = ""; |
346 |
|
} else { |
347 |
|
attr = _textout(PG_GETARG_TEXT_P(2)); |
348 |
|
} |
349 |
|
|
350 |
|
/* limit */ |
351 |
|
if (PG_ARGISNULL(3)) { |
352 |
|
limit = 0; |
353 |
|
} else { |
354 |
|
limit = PG_GETARG_INT32(3); |
355 |
|
} |
356 |
|
|
357 |
|
/* offset */ |
358 |
|
if (PG_ARGISNULL(4)) { |
359 |
|
offset = 0; |
360 |
|
} else { |
361 |
|
offset = PG_GETARG_INT32(4); |
362 |
|
} |
363 |
|
|
364 |
|
|
365 |
|
/* Send any errors or warnings to log, as well as |
366 |
|
* STDOUT and STDERR (just to be sure) */ |
367 |
|
if ( logfh = fopen("/tmp/pgswish.log", "a") ) { |
368 |
|
set_error_handle( logfh ); |
369 |
|
elog(INFO, "loggin swish-e errors to /tmp/pgswish.log"); |
370 |
|
/* redirect STDOUT and STDERR to log */ |
371 |
|
dup2(1, logfh); |
372 |
|
dup2(2, logfh); |
373 |
|
} else { |
374 |
|
elog(INFO, "can't open /tmp/pgswish.log -- errors from swish-e won't be cought and may result in back-end crashes!"); |
375 |
|
} |
376 |
|
|
377 |
|
elog(INFO, "pgswish: SwishInit(%s)", index_path); |
378 |
|
|
379 |
|
swish_handle = SwishInit( index_path ); |
380 |
|
|
381 |
|
if ( SwishError( swish_handle ) || ! swish_handle ) |
382 |
|
ereport(ERROR, (errcode(ERRCODE_INVALID_PARAMETER_VALUE), |
383 |
|
errmsg("pgswish: SwishInit(%s) failed", index_path ), |
384 |
|
errdetail( SwishErrorString( swish_handle ) ) |
385 |
|
)); |
386 |
|
|
387 |
|
elog(DEBUG1, "pgswish: query[%s] attr[%s] limit %d offset %d", query, (PG_ARGISNULL(2) ? "NULL" : attr), limit, offset); |
388 |
|
|
389 |
|
|
390 |
|
/* set ranking scheme. default is 0 */ |
391 |
|
SwishRankScheme( swish_handle, 0 ); |
392 |
|
error_or_abort( swish_handle ); |
393 |
|
|
394 |
|
elog(INFO, "pgswish: SwishQuery(%s)", query); |
395 |
|
/* Here's a short-cut to searching that creates a search object |
396 |
|
* and searches at the same time */ |
397 |
|
|
398 |
|
/* set the search phrase to the search condition object */ |
399 |
|
if (! PG_ARGISNULL(1) && strlen(query) > 0) |
400 |
|
swish_results = SwishQuery( swish_handle, query); |
401 |
|
error_or_abort( swish_handle ); |
402 |
|
|
403 |
|
/* total number of tuples to be returned */ |
404 |
|
resnum = SwishHits( swish_results ); |
405 |
|
|
406 |
|
/* FIXME */ |
407 |
|
if (! PG_ARGISNULL(2) && strlen(attr) >= 10) { |
408 |
|
elog(DEBUG1,"ignored: %s", attr); |
409 |
|
} |
410 |
|
|
411 |
|
/* check if results exists */ |
412 |
|
if ( 0 == resnum ) { |
413 |
|
elog(INFO, "pgswish: no results for: %s", query ); |
414 |
|
} |
415 |
|
|
416 |
|
/* total number of tuples to be returned */ |
417 |
|
if (limit && limit < resnum) { |
418 |
|
nrows = limit - offset; |
419 |
|
} else { |
420 |
|
nrows = resnum - offset; |
421 |
|
} |
422 |
|
|
423 |
|
|
424 |
|
elog(DEBUG1, "pgswish: found %d hits for %s", resnum, query); |
425 |
|
|
426 |
|
|
427 |
|
values = (char **) palloc(ncols * sizeof(char *)); |
428 |
|
|
429 |
|
for (i = 0; i < nrows; i++) |
430 |
|
{ |
431 |
|
SwishSeekResult( swish_results, i + offset ); |
432 |
|
sw_res = SwishNextResult( swish_results ); |
433 |
|
|
434 |
|
/* get result from swish-e */ |
435 |
|
if (! ( SwishErrorString( swish_handle ) ) ) { |
436 |
|
elog(INFO, "can't find result %d", i + offset); |
437 |
|
} else { |
438 |
|
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", |
439 |
|
SwishResultPropertyStr ( sw_res, "swishdocpath" ), |
440 |
|
SwishResultPropertyULong ( sw_res, "swishrank" ), |
441 |
|
SwishResultPropertyULong ( sw_res, "swishdocsize" ), |
442 |
|
SwishResultPropertyStr ( sw_res, "swishtitle"), |
443 |
|
SwishResultPropertyStr ( sw_res, "swishdbfile" ), |
444 |
|
SwishResultPropertyStr ( sw_res, "swishlastmodified" ), |
445 |
|
SwishResultPropertyULong ( sw_res, "swishreccount" ), /* can figure this out in loop, of course */ |
446 |
|
SwishResultPropertyULong ( sw_res, "swishfilenum" ) |
447 |
|
); |
448 |
|
} |
449 |
|
|
450 |
|
/* iterate over results */ |
451 |
|
for (j = 0; j < ncols; j++) |
452 |
|
{ |
453 |
|
bool isnull; |
454 |
|
|
455 |
|
/* array value of this position */ |
456 |
|
indx[0] = j + prop_dim_lower_bounds[0]; |
457 |
|
|
458 |
|
dvalue = array_ref(prop_arr, prop_ndims, indx, -1, prop_len, prop_byval, prop_align, &isnull); |
459 |
|
|
460 |
|
if (!isnull && sw_res) |
461 |
|
values[j] = DatumGetCString( |
462 |
|
prop2text( sw_res, |
463 |
|
(char *)DirectFunctionCall1(textout, dvalue) |
464 |
|
)); |
465 |
|
else |
466 |
|
values[j] = NULL; |
467 |
|
} |
468 |
|
/* construct the tuple */ |
469 |
|
tuple = BuildTupleFromCStrings(attinmeta, values); |
470 |
|
|
471 |
|
/* now store it */ |
472 |
|
tuplestore_puttuple(tupstore, tuple); |
473 |
|
|
474 |
|
} |
475 |
|
|
476 |
|
tuplestore_donestoring(tupstore); |
477 |
|
rsinfo->setResult = tupstore; |
478 |
|
|
479 |
|
/* |
480 |
|
* SFRM_Materialize mode expects us to return a NULL Datum. The actual |
481 |
|
* tuples are in our tuplestore and passed back through |
482 |
|
* rsinfo->setResult. rsinfo->setDesc is set to the tuple description |
483 |
|
* that we actually used to build our tuples with, so the caller can |
484 |
|
* verify we did what it was expecting. |
485 |
|
*/ |
486 |
|
rsinfo->setDesc = tupdesc; |
487 |
|
MemoryContextSwitchTo(oldcontext); |
488 |
|
|
489 |
|
/* free swish object and close */ |
490 |
|
Free_Search_Object( search ); |
491 |
|
SwishClose( swish_handle ); |
492 |
|
|
493 |
|
return (Datum) 0; |
494 |
|
} |
495 |
|
|
496 |
|
|
497 |
|
|
498 |
|
|
499 |
/* make text var from property */ |
/* make text var from property */ |
500 |
char *prop2text(SW_RESULT sw_res, char *propname) { |
char *prop2text(SW_RESULT sw_res, char *propname) { |
501 |
char *val; |
char *val; |