1 |
dpavlin |
1 |
/* |
2 |
|
|
* integrate Hyper Estraier into PostgreSQL |
3 |
|
|
* |
4 |
|
|
* Dobrica Pavlinusic <dpavlin@rot13.org> 2005-05-19 |
5 |
|
|
* |
6 |
|
|
* TODO: |
7 |
|
|
* - all |
8 |
|
|
* |
9 |
|
|
* NOTES: |
10 |
|
|
* - clear structures with memset to support hash indexes (who whould like |
11 |
|
|
* to create hash index on table returned from function?) |
12 |
|
|
* - number of returned rows is set by PostgreSQL evaluator, see: |
13 |
|
|
* http://archives.postgresql.org/pgsql-hackers/2005-02/msg00546.php |
14 |
|
|
* |
15 |
|
|
* Based on: |
16 |
|
|
* - C example from PostgreSQL documentation (BSD licence) |
17 |
dpavlin |
41 |
* - coreexample002.c and nodeexample002.c from Hyper Estraier (GPL) |
18 |
dpavlin |
1 |
* - _textin/_textout from pgcurl.c (LGPL) |
19 |
|
|
* |
20 |
|
|
* This code is licenced under GPL |
21 |
|
|
*/ |
22 |
|
|
|
23 |
|
|
#include "postgres.h" |
24 |
|
|
#include "fmgr.h" |
25 |
|
|
#include "funcapi.h" |
26 |
|
|
#include "utils/builtins.h" |
27 |
|
|
#include "utils/array.h" |
28 |
dpavlin |
56 |
#include "utils/lsyscache.h" |
29 |
dpavlin |
1 |
#include "miscadmin.h" |
30 |
dpavlin |
61 |
#include "commands/trigger.h" |
31 |
|
|
#include "executor/spi.h" |
32 |
|
|
|
33 |
dpavlin |
1 |
#include <estraier.h> |
34 |
|
|
#include <cabin.h> |
35 |
dpavlin |
41 |
#include <estnode.h> |
36 |
dpavlin |
1 |
|
37 |
|
|
#define _textin(str) DirectFunctionCall1(textin, CStringGetDatum(str)) |
38 |
|
|
#define _textout(str) DatumGetPointer(DirectFunctionCall1(textout, PointerGetDatum(str))) |
39 |
|
|
#define GET_STR(textp) DatumGetCString(DirectFunctionCall1(textout, PointerGetDatum(textp))) |
40 |
|
|
#define GET_TEXT(cstrp) DatumGetTextP(DirectFunctionCall1(textin, CStringGetDatum(cstrp))) |
41 |
|
|
|
42 |
dpavlin |
27 |
/* SortMem got renamed in PostgreSQL 8.0 */ |
43 |
|
|
#ifndef SortMem |
44 |
|
|
#define SortMem 16 * 1024 |
45 |
|
|
#endif |
46 |
|
|
|
47 |
dpavlin |
40 |
#define ATTR_DELIMITER "{{!}}" |
48 |
dpavlin |
57 |
#define HINTS_PREFIX "HINTS." |
49 |
dpavlin |
40 |
|
50 |
dpavlin |
5 |
/* prototype */ |
51 |
|
|
char *attr2text(ESTDOC *doc, char *attr); |
52 |
dpavlin |
41 |
char *node_attr2text(ESTRESDOC *rdoc, char *attr); |
53 |
dpavlin |
56 |
void cond_add_attr(ESTCOND *cond, char *attr); |
54 |
dpavlin |
1 |
|
55 |
|
|
|
56 |
|
|
/* work in progress */ |
57 |
dpavlin |
19 |
PG_FUNCTION_INFO_V1(pgest_attr); |
58 |
|
|
Datum pgest_attr(PG_FUNCTION_ARGS) |
59 |
dpavlin |
1 |
{ |
60 |
dpavlin |
31 |
ArrayType *attr_arr = PG_GETARG_ARRAYTYPE_P(6); |
61 |
dpavlin |
25 |
Oid attr_element_type = ARR_ELEMTYPE(attr_arr); |
62 |
|
|
int attr_ndims = ARR_NDIM(attr_arr); |
63 |
|
|
int *attr_dim_counts = ARR_DIMS(attr_arr); |
64 |
|
|
int *attr_dim_lower_bounds = ARR_LBOUND(attr_arr); |
65 |
dpavlin |
19 |
int ncols = 0; |
66 |
|
|
int nrows = 0; |
67 |
|
|
int indx[MAXDIM]; |
68 |
dpavlin |
25 |
int16 attr_len; |
69 |
|
|
bool attr_byval; |
70 |
|
|
char attr_align; |
71 |
dpavlin |
1 |
ReturnSetInfo *rsinfo = (ReturnSetInfo *) fcinfo->resultinfo; |
72 |
|
|
AttInMetadata *attinmeta; |
73 |
|
|
TupleDesc tupdesc; |
74 |
dpavlin |
19 |
Tuplestorestate *tupstore = NULL; |
75 |
dpavlin |
1 |
HeapTuple tuple; |
76 |
|
|
MemoryContext per_query_ctx; |
77 |
|
|
MemoryContext oldcontext; |
78 |
|
|
Datum dvalue; |
79 |
|
|
char **values; |
80 |
dpavlin |
19 |
int rsinfo_ncols; |
81 |
dpavlin |
1 |
int i, j; |
82 |
dpavlin |
19 |
/* estvars */ |
83 |
dpavlin |
25 |
ESTDB *db; |
84 |
|
|
ESTCOND *cond; |
85 |
|
|
ESTDOC *doc; |
86 |
|
|
int ecode, *est_result, resnum; |
87 |
|
|
int limit = 0; |
88 |
|
|
int offset = 0; |
89 |
|
|
|
90 |
dpavlin |
19 |
char *index_path; |
91 |
|
|
char *query; |
92 |
|
|
char *attr; |
93 |
dpavlin |
31 |
char *order; |
94 |
dpavlin |
1 |
|
95 |
dpavlin |
19 |
|
96 |
|
|
/* only allow 1D input array */ |
97 |
dpavlin |
25 |
if (attr_ndims == 1) |
98 |
dpavlin |
19 |
{ |
99 |
dpavlin |
25 |
ncols = attr_dim_counts[0]; |
100 |
dpavlin |
19 |
} |
101 |
|
|
else |
102 |
|
|
ereport(ERROR, |
103 |
|
|
(errcode(ERRCODE_INVALID_PARAMETER_VALUE), |
104 |
|
|
errmsg("invalid input array"), |
105 |
|
|
errdetail("Input array must have 1 dimension"))); |
106 |
|
|
|
107 |
dpavlin |
1 |
/* check to see if caller supports us returning a tuplestore */ |
108 |
|
|
if (!rsinfo || !(rsinfo->allowedModes & SFRM_Materialize)) |
109 |
|
|
ereport(ERROR, |
110 |
|
|
(errcode(ERRCODE_SYNTAX_ERROR), |
111 |
|
|
errmsg("materialize mode required, but it is not " \ |
112 |
|
|
"allowed in this context"))); |
113 |
|
|
|
114 |
dpavlin |
19 |
/* get info about element type needed to construct the array */ |
115 |
dpavlin |
25 |
get_typlenbyvalalign(attr_element_type, &attr_len, &attr_byval, &attr_align); |
116 |
dpavlin |
19 |
|
117 |
dpavlin |
1 |
/* get the requested return tuple description */ |
118 |
|
|
tupdesc = rsinfo->expectedDesc; |
119 |
dpavlin |
19 |
rsinfo_ncols = tupdesc->natts; |
120 |
dpavlin |
1 |
|
121 |
|
|
/* |
122 |
|
|
* The requested tuple description better match up with the array |
123 |
|
|
* we were given. |
124 |
|
|
*/ |
125 |
dpavlin |
19 |
if (rsinfo_ncols != ncols) |
126 |
|
|
ereport(ERROR, |
127 |
|
|
(errcode(ERRCODE_INVALID_PARAMETER_VALUE), |
128 |
|
|
errmsg("invalid input array"), |
129 |
|
|
errdetail("Number of elements in array must match number of query specified columns."))); |
130 |
|
|
|
131 |
dpavlin |
1 |
/* OK, use it */ |
132 |
|
|
attinmeta = TupleDescGetAttInMetadata(tupdesc); |
133 |
|
|
|
134 |
|
|
/* Now go to work */ |
135 |
|
|
rsinfo->returnMode = SFRM_Materialize; |
136 |
|
|
|
137 |
|
|
per_query_ctx = fcinfo->flinfo->fn_mcxt; |
138 |
|
|
oldcontext = MemoryContextSwitchTo(per_query_ctx); |
139 |
|
|
|
140 |
|
|
/* initialize our tuplestore */ |
141 |
|
|
tupstore = tuplestore_begin_heap(true, false, SortMem); |
142 |
|
|
|
143 |
dpavlin |
19 |
|
144 |
|
|
/* take rest of arguments from function */ |
145 |
|
|
|
146 |
|
|
/* index path */ |
147 |
|
|
if (PG_ARGISNULL(0)) { |
148 |
|
|
ereport(ERROR, |
149 |
|
|
(errcode(ERRCODE_INVALID_PARAMETER_VALUE), |
150 |
|
|
errmsg("index path can't be null"), |
151 |
|
|
errdetail("Index path must be valid full path to HyperEstraier index"))); |
152 |
|
|
} |
153 |
|
|
index_path = _textout(PG_GETARG_TEXT_P(0)); |
154 |
|
|
|
155 |
|
|
/* query string */ |
156 |
dpavlin |
31 |
if (PG_ARGISNULL(1)) { |
157 |
dpavlin |
19 |
query = ""; |
158 |
|
|
} else { |
159 |
|
|
query = _textout(PG_GETARG_TEXT_P(1)); |
160 |
|
|
} |
161 |
|
|
|
162 |
|
|
/* atribute filter */ |
163 |
|
|
if (PG_ARGISNULL(2)) { |
164 |
|
|
attr = ""; |
165 |
|
|
} else { |
166 |
|
|
attr = _textout(PG_GETARG_TEXT_P(2)); |
167 |
|
|
} |
168 |
dpavlin |
31 |
|
169 |
|
|
/* sort order */ |
170 |
|
|
if (PG_ARGISNULL(3)) { |
171 |
|
|
order = ""; |
172 |
|
|
} else { |
173 |
|
|
order = _textout(PG_GETARG_TEXT_P(3)); |
174 |
|
|
} |
175 |
dpavlin |
19 |
|
176 |
dpavlin |
31 |
|
177 |
dpavlin |
19 |
/* limit */ |
178 |
dpavlin |
31 |
if (PG_ARGISNULL(4)) { |
179 |
dpavlin |
19 |
limit = 0; |
180 |
|
|
} else { |
181 |
dpavlin |
31 |
limit = PG_GETARG_INT32(4); |
182 |
dpavlin |
19 |
} |
183 |
|
|
|
184 |
|
|
/* offset */ |
185 |
dpavlin |
31 |
if (PG_ARGISNULL(5)) { |
186 |
dpavlin |
19 |
offset = 0; |
187 |
|
|
} else { |
188 |
dpavlin |
31 |
offset = PG_GETARG_INT32(5); |
189 |
dpavlin |
19 |
} |
190 |
|
|
|
191 |
|
|
|
192 |
|
|
/* open the database */ |
193 |
|
|
elog(DEBUG1, "pgest_attr: est_db_open(%s)", index_path); |
194 |
|
|
|
195 |
|
|
if(!(db = est_db_open(index_path, ESTDBREADER, &ecode))){ |
196 |
|
|
ereport(ERROR, (errcode(ERRCODE_INVALID_PARAMETER_VALUE), |
197 |
|
|
errmsg("est_db_open: can't open %s: %d", index_path, ecode), |
198 |
|
|
errdetail(est_err_msg(ecode)))); |
199 |
|
|
} |
200 |
|
|
|
201 |
dpavlin |
20 |
elog(DEBUG1, "pgest_attr: query[%s] attr[%s] limit %d offset %d", query, (PG_ARGISNULL(2) ? "NULL" : attr), limit, offset); |
202 |
dpavlin |
19 |
|
203 |
|
|
/* create a search condition object */ |
204 |
|
|
if (!(cond = est_cond_new())) { |
205 |
|
|
ereport(ERROR, (errcode(ERRCODE_QUERY_CANCELED), |
206 |
|
|
errmsg("pgest_attr: est_cond_new failed"))); |
207 |
|
|
} |
208 |
|
|
|
209 |
|
|
/* set the search phrase to the search condition object */ |
210 |
|
|
if (! PG_ARGISNULL(1) && strlen(query) > 0) |
211 |
|
|
est_cond_set_phrase(cond, query); |
212 |
|
|
|
213 |
|
|
/* minimum valid attribute length is 10: @a STREQ a */ |
214 |
|
|
if (! PG_ARGISNULL(2) && strlen(attr) >= 10) { |
215 |
dpavlin |
40 |
elog(DEBUG1,"attributes: %s", attr); |
216 |
dpavlin |
56 |
cond_add_attr(cond, attr); |
217 |
dpavlin |
19 |
} |
218 |
|
|
|
219 |
dpavlin |
31 |
/* set the search phrase to the search condition object */ |
220 |
|
|
if (! PG_ARGISNULL(3) && strlen(order) > 0) { |
221 |
|
|
elog(DEBUG1,"est_cond_set_order(%s)", order); |
222 |
|
|
est_cond_set_order(cond, order); |
223 |
|
|
} |
224 |
|
|
|
225 |
dpavlin |
38 |
if (limit) { |
226 |
|
|
elog(DEBUG1,"est_cond_set_max(%d)", limit + offset); |
227 |
|
|
est_cond_set_max(cond, limit + offset); |
228 |
|
|
} |
229 |
|
|
|
230 |
dpavlin |
19 |
/* get the result of search */ |
231 |
|
|
est_result = est_db_search(db, cond, &resnum, NULL); |
232 |
|
|
|
233 |
|
|
/* check if results exists */ |
234 |
|
|
if ( 0 == resnum ) { |
235 |
|
|
elog(INFO, "pgest_attr: no results for: %s", query ); |
236 |
|
|
} |
237 |
|
|
|
238 |
|
|
/* total number of tuples to be returned */ |
239 |
|
|
if (limit && limit < resnum) { |
240 |
dpavlin |
31 |
nrows = limit; |
241 |
dpavlin |
19 |
} else { |
242 |
|
|
nrows = resnum - offset; |
243 |
|
|
} |
244 |
|
|
|
245 |
|
|
|
246 |
|
|
elog(DEBUG1, "pgest_attr: found %d hits for %s", resnum, query); |
247 |
|
|
|
248 |
dpavlin |
1 |
values = (char **) palloc(ncols * sizeof(char *)); |
249 |
|
|
|
250 |
|
|
for (i = 0; i < nrows; i++) |
251 |
|
|
{ |
252 |
dpavlin |
19 |
|
253 |
|
|
/* get result from estraier */ |
254 |
|
|
if (! ( doc = est_db_get_doc(db, est_result[i + offset], 0)) ) { |
255 |
dpavlin |
47 |
elog(INFO, "pgest_attr: can't find result %d", i + offset); |
256 |
dpavlin |
19 |
} else { |
257 |
|
|
elog(DEBUG1, "URI: %s\n Title: %s\n", |
258 |
|
|
est_doc_attr(doc, "@uri"), |
259 |
|
|
est_doc_attr(doc, "@title") |
260 |
|
|
); |
261 |
|
|
} |
262 |
|
|
|
263 |
|
|
/* iterate over results */ |
264 |
dpavlin |
1 |
for (j = 0; j < ncols; j++) |
265 |
|
|
{ |
266 |
dpavlin |
19 |
bool isnull; |
267 |
|
|
|
268 |
|
|
/* array value of this position */ |
269 |
dpavlin |
25 |
indx[0] = j + attr_dim_lower_bounds[0]; |
270 |
dpavlin |
19 |
|
271 |
dpavlin |
25 |
dvalue = array_ref(attr_arr, attr_ndims, indx, -1, attr_len, attr_byval, attr_align, &isnull); |
272 |
dpavlin |
19 |
|
273 |
|
|
if (!isnull && doc) |
274 |
|
|
values[j] = DatumGetCString( |
275 |
|
|
attr2text(doc, |
276 |
dpavlin |
20 |
(char *)DirectFunctionCall1(textout, dvalue) |
277 |
dpavlin |
19 |
)); |
278 |
|
|
else |
279 |
|
|
values[j] = NULL; |
280 |
dpavlin |
1 |
} |
281 |
|
|
/* construct the tuple */ |
282 |
|
|
tuple = BuildTupleFromCStrings(attinmeta, values); |
283 |
|
|
|
284 |
|
|
/* now store it */ |
285 |
|
|
tuplestore_puttuple(tupstore, tuple); |
286 |
dpavlin |
19 |
|
287 |
|
|
/* delete estraier document object */ |
288 |
dpavlin |
47 |
if (doc) est_doc_delete(doc); |
289 |
dpavlin |
1 |
} |
290 |
|
|
|
291 |
|
|
tuplestore_donestoring(tupstore); |
292 |
|
|
rsinfo->setResult = tupstore; |
293 |
|
|
|
294 |
|
|
/* |
295 |
|
|
* SFRM_Materialize mode expects us to return a NULL Datum. The actual |
296 |
|
|
* tuples are in our tuplestore and passed back through |
297 |
|
|
* rsinfo->setResult. rsinfo->setDesc is set to the tuple description |
298 |
|
|
* that we actually used to build our tuples with, so the caller can |
299 |
|
|
* verify we did what it was expecting. |
300 |
|
|
*/ |
301 |
|
|
rsinfo->setDesc = tupdesc; |
302 |
|
|
MemoryContextSwitchTo(oldcontext); |
303 |
|
|
|
304 |
dpavlin |
31 |
est_cond_delete(cond); |
305 |
|
|
|
306 |
dpavlin |
19 |
if(!est_db_close(db, &ecode)){ |
307 |
|
|
ereport(ERROR, (errcode(ERRCODE_IO_ERROR), |
308 |
|
|
errmsg("est_db_close: %d", ecode), |
309 |
|
|
errdetail(est_err_msg(ecode)))); |
310 |
|
|
} |
311 |
|
|
|
312 |
dpavlin |
1 |
return (Datum) 0; |
313 |
|
|
} |
314 |
|
|
|
315 |
|
|
|
316 |
|
|
/* make text var from attr */ |
317 |
|
|
char *attr2text(ESTDOC *doc, char *attr) { |
318 |
|
|
char *val; |
319 |
|
|
const char *attrval; |
320 |
|
|
int len; |
321 |
dpavlin |
4 |
int attrlen; |
322 |
dpavlin |
1 |
|
323 |
dpavlin |
47 |
if (! doc) return (Datum) NULL; |
324 |
|
|
|
325 |
dpavlin |
56 |
elog(DEBUG1, "doc: %p, attr: %s", doc, attr); |
326 |
dpavlin |
1 |
|
327 |
dpavlin |
4 |
if ( (attrval = est_doc_attr(doc, attr)) && (attrlen = strlen(attrval)) ) { |
328 |
|
|
val = (char *) palloc(attrlen * sizeof(char)); |
329 |
dpavlin |
1 |
} else { |
330 |
|
|
return (Datum) NULL; |
331 |
|
|
} |
332 |
|
|
|
333 |
|
|
len = strlen(attrval); |
334 |
dpavlin |
2 |
elog(DEBUG1, "attr2text(%s) = '%s' %d bytes", attr, attrval, len); |
335 |
dpavlin |
1 |
|
336 |
|
|
len++; |
337 |
|
|
len *= sizeof(char); |
338 |
|
|
|
339 |
|
|
elog(DEBUG2, "palloc(%d)", len); |
340 |
|
|
|
341 |
|
|
val = palloc(len); |
342 |
|
|
|
343 |
|
|
memset(val, 0, len); |
344 |
|
|
strncpy(val, attrval, len); |
345 |
|
|
|
346 |
|
|
elog(DEBUG2, "val=%s", val); |
347 |
|
|
|
348 |
|
|
return val; |
349 |
|
|
} |
350 |
|
|
|
351 |
dpavlin |
41 |
/* |
352 |
|
|
* variation on theme: use node API which doesn't open index on |
353 |
|
|
* every query which is much faster for large indexes |
354 |
|
|
* |
355 |
|
|
*/ |
356 |
|
|
|
357 |
dpavlin |
48 |
/* select * from pgest( */ |
358 |
|
|
#define _arg_node_uri 0 |
359 |
|
|
#define _arg_login 1 |
360 |
|
|
#define _arg_passwd 2 |
361 |
dpavlin |
49 |
#define _arg_depth 3 |
362 |
|
|
#define _arg_query 4 |
363 |
|
|
#define _arg_attr 5 |
364 |
|
|
#define _arg_order 6 |
365 |
|
|
#define _arg_limit 7 |
366 |
|
|
#define _arg_offset 8 |
367 |
|
|
#define _arg_attr_array 9 |
368 |
dpavlin |
48 |
/* as (foo text, ... ); */ |
369 |
|
|
|
370 |
|
|
|
371 |
dpavlin |
41 |
PG_FUNCTION_INFO_V1(pgest_node); |
372 |
|
|
Datum pgest_node(PG_FUNCTION_ARGS) |
373 |
|
|
{ |
374 |
dpavlin |
48 |
ArrayType *attr_arr = PG_GETARG_ARRAYTYPE_P(_arg_attr_array); |
375 |
dpavlin |
41 |
Oid attr_element_type = ARR_ELEMTYPE(attr_arr); |
376 |
|
|
int attr_ndims = ARR_NDIM(attr_arr); |
377 |
|
|
int *attr_dim_counts = ARR_DIMS(attr_arr); |
378 |
|
|
int *attr_dim_lower_bounds = ARR_LBOUND(attr_arr); |
379 |
|
|
int ncols = 0; |
380 |
|
|
int nrows = 0; |
381 |
|
|
int indx[MAXDIM]; |
382 |
|
|
int16 attr_len; |
383 |
|
|
bool attr_byval; |
384 |
|
|
char attr_align; |
385 |
|
|
ReturnSetInfo *rsinfo = (ReturnSetInfo *) fcinfo->resultinfo; |
386 |
|
|
AttInMetadata *attinmeta; |
387 |
|
|
TupleDesc tupdesc; |
388 |
|
|
Tuplestorestate *tupstore = NULL; |
389 |
|
|
HeapTuple tuple; |
390 |
|
|
MemoryContext per_query_ctx; |
391 |
|
|
MemoryContext oldcontext; |
392 |
|
|
Datum dvalue; |
393 |
|
|
char **values; |
394 |
|
|
int rsinfo_ncols; |
395 |
|
|
int i, j; |
396 |
|
|
/* estvars */ |
397 |
|
|
ESTNODE *node; |
398 |
|
|
ESTCOND *cond; |
399 |
|
|
ESTNODERES *nres; |
400 |
|
|
ESTRESDOC *rdoc; |
401 |
dpavlin |
57 |
CBMAP *hints; |
402 |
dpavlin |
41 |
int resnum = 0; |
403 |
|
|
int limit = 0; |
404 |
|
|
int offset = 0; |
405 |
dpavlin |
49 |
int depth = 0; |
406 |
dpavlin |
41 |
|
407 |
|
|
char *node_url; |
408 |
|
|
char *user, *passwd; |
409 |
|
|
char *query; |
410 |
|
|
char *attr; |
411 |
|
|
char *order; |
412 |
|
|
|
413 |
|
|
|
414 |
|
|
/* only allow 1D input array */ |
415 |
|
|
if (attr_ndims == 1) |
416 |
|
|
{ |
417 |
|
|
ncols = attr_dim_counts[0]; |
418 |
|
|
} |
419 |
|
|
else |
420 |
|
|
ereport(ERROR, |
421 |
|
|
(errcode(ERRCODE_INVALID_PARAMETER_VALUE), |
422 |
|
|
errmsg("invalid input array"), |
423 |
|
|
errdetail("Input array must have 1 dimension"))); |
424 |
|
|
|
425 |
|
|
/* check to see if caller supports us returning a tuplestore */ |
426 |
|
|
if (!rsinfo || !(rsinfo->allowedModes & SFRM_Materialize)) |
427 |
|
|
ereport(ERROR, |
428 |
|
|
(errcode(ERRCODE_SYNTAX_ERROR), |
429 |
|
|
errmsg("materialize mode required, but it is not " \ |
430 |
|
|
"allowed in this context"))); |
431 |
|
|
|
432 |
|
|
/* get info about element type needed to construct the array */ |
433 |
|
|
get_typlenbyvalalign(attr_element_type, &attr_len, &attr_byval, &attr_align); |
434 |
|
|
|
435 |
|
|
/* get the requested return tuple description */ |
436 |
|
|
tupdesc = rsinfo->expectedDesc; |
437 |
|
|
rsinfo_ncols = tupdesc->natts; |
438 |
|
|
|
439 |
|
|
/* |
440 |
|
|
* The requested tuple description better match up with the array |
441 |
|
|
* we were given. |
442 |
|
|
*/ |
443 |
|
|
if (rsinfo_ncols != ncols) |
444 |
|
|
ereport(ERROR, |
445 |
|
|
(errcode(ERRCODE_INVALID_PARAMETER_VALUE), |
446 |
|
|
errmsg("invalid input array"), |
447 |
|
|
errdetail("Number of elements in array must match number of query specified columns."))); |
448 |
|
|
|
449 |
|
|
/* OK, use it */ |
450 |
|
|
attinmeta = TupleDescGetAttInMetadata(tupdesc); |
451 |
|
|
|
452 |
|
|
/* Now go to work */ |
453 |
|
|
rsinfo->returnMode = SFRM_Materialize; |
454 |
|
|
|
455 |
|
|
per_query_ctx = fcinfo->flinfo->fn_mcxt; |
456 |
|
|
oldcontext = MemoryContextSwitchTo(per_query_ctx); |
457 |
|
|
|
458 |
|
|
/* initialize our tuplestore */ |
459 |
|
|
tupstore = tuplestore_begin_heap(true, false, SortMem); |
460 |
|
|
|
461 |
|
|
|
462 |
|
|
/* take rest of arguments from function */ |
463 |
|
|
|
464 |
|
|
/* node URL */ |
465 |
dpavlin |
48 |
if (PG_ARGISNULL(_arg_node_uri)) { |
466 |
dpavlin |
41 |
ereport(ERROR, |
467 |
|
|
(errcode(ERRCODE_INVALID_PARAMETER_VALUE), |
468 |
|
|
errmsg("node URL can't be null"), |
469 |
|
|
errdetail("Node URL must be valid URL to HyperEstraier node"))); |
470 |
|
|
} |
471 |
dpavlin |
48 |
node_url = _textout(PG_GETARG_TEXT_P(_arg_node_uri)); |
472 |
dpavlin |
41 |
|
473 |
|
|
/* login and password */ |
474 |
dpavlin |
48 |
if (PG_ARGISNULL(_arg_login) || PG_ARGISNULL(_arg_passwd)) { |
475 |
dpavlin |
41 |
ereport(ERROR, |
476 |
|
|
(errcode(ERRCODE_INVALID_PARAMETER_VALUE), |
477 |
|
|
errmsg("username and password can't be NULL"), |
478 |
|
|
errdetail("You must specify valid username and password to HyperEstraier node"))); |
479 |
|
|
} |
480 |
dpavlin |
48 |
user = _textout(PG_GETARG_TEXT_P(_arg_login)); |
481 |
|
|
passwd = _textout(PG_GETARG_TEXT_P(_arg_passwd)); |
482 |
dpavlin |
41 |
|
483 |
dpavlin |
49 |
/* depth of search */ |
484 |
|
|
if (PG_ARGISNULL(_arg_depth)) { |
485 |
|
|
depth = 0; |
486 |
|
|
} else { |
487 |
|
|
depth = PG_GETARG_INT32(_arg_depth); |
488 |
|
|
} |
489 |
|
|
|
490 |
dpavlin |
41 |
/* query string */ |
491 |
dpavlin |
48 |
if (PG_ARGISNULL(_arg_query)) { |
492 |
dpavlin |
41 |
query = ""; |
493 |
|
|
} else { |
494 |
dpavlin |
48 |
query = _textout(PG_GETARG_TEXT_P(_arg_query)); |
495 |
dpavlin |
41 |
} |
496 |
|
|
|
497 |
|
|
/* atribute filter */ |
498 |
dpavlin |
48 |
if (PG_ARGISNULL(_arg_attr)) { |
499 |
dpavlin |
41 |
attr = ""; |
500 |
|
|
} else { |
501 |
dpavlin |
48 |
attr = _textout(PG_GETARG_TEXT_P(_arg_attr)); |
502 |
dpavlin |
41 |
} |
503 |
|
|
|
504 |
|
|
/* sort order */ |
505 |
dpavlin |
48 |
if (PG_ARGISNULL(_arg_order)) { |
506 |
dpavlin |
41 |
order = ""; |
507 |
|
|
} else { |
508 |
dpavlin |
48 |
order = _textout(PG_GETARG_TEXT_P(_arg_order)); |
509 |
dpavlin |
41 |
} |
510 |
|
|
|
511 |
|
|
|
512 |
|
|
/* limit */ |
513 |
dpavlin |
48 |
if (PG_ARGISNULL(_arg_limit)) { |
514 |
dpavlin |
41 |
limit = 0; |
515 |
|
|
} else { |
516 |
dpavlin |
48 |
limit = PG_GETARG_INT32(_arg_limit); |
517 |
dpavlin |
41 |
} |
518 |
|
|
|
519 |
|
|
/* offset */ |
520 |
dpavlin |
48 |
if (PG_ARGISNULL(_arg_offset)) { |
521 |
dpavlin |
41 |
offset = 0; |
522 |
|
|
} else { |
523 |
dpavlin |
48 |
offset = PG_GETARG_INT32(_arg_offset); |
524 |
dpavlin |
41 |
} |
525 |
|
|
|
526 |
|
|
/* initialize the network environment */ |
527 |
|
|
if(!est_init_net_env()){ |
528 |
|
|
ereport(ERROR, (errcode(ERRCODE_QUERY_CANCELED), |
529 |
|
|
errmsg("pgest_node: can't create network enviroment"))); |
530 |
|
|
} |
531 |
|
|
|
532 |
|
|
/* create the node connection object */ |
533 |
|
|
elog(DEBUG1, "pgest_node: est_node_new(%s) as %s", node_url, user); |
534 |
|
|
node = est_node_new(node_url); |
535 |
|
|
est_node_set_auth(node, user, passwd); |
536 |
|
|
|
537 |
dpavlin |
49 |
elog(DEBUG1, "pgest_node: node: %s (d:%d) query[%s] attr[%s] limit %d offset %d", node_url, depth, query, (PG_ARGISNULL(_arg_attr) ? "NULL" : attr), limit, offset); |
538 |
dpavlin |
41 |
|
539 |
|
|
/* create a search condition object */ |
540 |
|
|
if (!(cond = est_cond_new())) { |
541 |
|
|
ereport(ERROR, (errcode(ERRCODE_QUERY_CANCELED), |
542 |
|
|
errmsg("pgest_node: est_cond_new failed"))); |
543 |
|
|
} |
544 |
|
|
|
545 |
|
|
/* set the search phrase to the search condition object */ |
546 |
dpavlin |
48 |
if (! PG_ARGISNULL(_arg_query) && strlen(query) > 0) |
547 |
dpavlin |
41 |
est_cond_set_phrase(cond, query); |
548 |
|
|
|
549 |
|
|
/* minimum valid attribute length is 10: @a STREQ a */ |
550 |
dpavlin |
48 |
if (! PG_ARGISNULL(_arg_attr) && strlen(attr) >= 10) { |
551 |
dpavlin |
41 |
elog(DEBUG1,"attributes: %s", attr); |
552 |
dpavlin |
56 |
cond_add_attr(cond, attr); |
553 |
dpavlin |
41 |
} |
554 |
|
|
|
555 |
|
|
/* set the search phrase to the search condition object */ |
556 |
dpavlin |
48 |
if (! PG_ARGISNULL(_arg_order) && strlen(order) > 0) { |
557 |
dpavlin |
41 |
elog(DEBUG1,"est_cond_set_order(%s)", order); |
558 |
|
|
est_cond_set_order(cond, order); |
559 |
|
|
} |
560 |
|
|
|
561 |
|
|
if (limit) { |
562 |
|
|
elog(DEBUG1,"est_cond_set_max(%d)", limit + offset); |
563 |
|
|
est_cond_set_max(cond, limit + offset); |
564 |
|
|
} |
565 |
|
|
|
566 |
dpavlin |
51 |
if (offset) { |
567 |
|
|
elog(DEBUG1,"est_cond_set_skip(%d)", offset); |
568 |
|
|
est_cond_set_skip(cond, offset); |
569 |
|
|
} |
570 |
|
|
|
571 |
dpavlin |
41 |
/* get the result of search */ |
572 |
dpavlin |
49 |
nres = est_node_search(node, cond, depth); |
573 |
dpavlin |
41 |
|
574 |
dpavlin |
44 |
if (! nres) { |
575 |
|
|
int status = est_node_status(node); |
576 |
|
|
est_cond_delete(cond); |
577 |
|
|
est_node_delete(node); |
578 |
|
|
est_free_net_env(); |
579 |
|
|
ereport(ERROR, (errcode(ERRCODE_QUERY_CANCELED), |
580 |
|
|
errmsg("pgest_node: search failed, node status %d", status))); |
581 |
|
|
} |
582 |
|
|
|
583 |
dpavlin |
41 |
/* get number of results */ |
584 |
|
|
resnum = est_noderes_doc_num(nres); |
585 |
|
|
|
586 |
|
|
/* check if results exists */ |
587 |
|
|
if ( 0 == resnum ) { |
588 |
|
|
elog(INFO, "pgest_node: no results for: %s", query ); |
589 |
|
|
} |
590 |
|
|
|
591 |
|
|
/* total number of tuples to be returned */ |
592 |
|
|
if (limit && limit < resnum) { |
593 |
|
|
nrows = limit; |
594 |
|
|
} else { |
595 |
|
|
nrows = resnum - offset; |
596 |
|
|
} |
597 |
|
|
|
598 |
dpavlin |
57 |
/* get hints */ |
599 |
|
|
hints = est_noderes_hints(nres); |
600 |
|
|
|
601 |
dpavlin |
41 |
elog(DEBUG1, "pgest_node: found %d hits for %s", resnum, query); |
602 |
|
|
|
603 |
|
|
|
604 |
|
|
values = (char **) palloc(ncols * sizeof(char *)); |
605 |
|
|
|
606 |
|
|
for (i = 0; i < nrows; i++) |
607 |
|
|
{ |
608 |
|
|
|
609 |
|
|
/* get result from estraier */ |
610 |
dpavlin |
51 |
if (! ( rdoc = est_noderes_get_doc(nres, i) )) { |
611 |
dpavlin |
47 |
elog(INFO, "pgest_node: can't find result %d", i + offset); |
612 |
dpavlin |
41 |
} else { |
613 |
|
|
elog(DEBUG1, "URI: %s\n Title: %s\n", |
614 |
|
|
est_resdoc_attr(rdoc, "@uri"), |
615 |
|
|
est_resdoc_attr(rdoc, "@title") |
616 |
|
|
); |
617 |
|
|
} |
618 |
|
|
|
619 |
|
|
/* iterate over results */ |
620 |
|
|
for (j = 0; j < ncols; j++) |
621 |
|
|
{ |
622 |
|
|
bool isnull; |
623 |
dpavlin |
57 |
char *attr; /* current attribute name */ |
624 |
|
|
char *hint; /* position of current hint in attribute */ |
625 |
|
|
char *hint_val; |
626 |
dpavlin |
41 |
|
627 |
|
|
/* array value of this position */ |
628 |
|
|
indx[0] = j + attr_dim_lower_bounds[0]; |
629 |
|
|
|
630 |
|
|
dvalue = array_ref(attr_arr, attr_ndims, indx, -1, attr_len, attr_byval, attr_align, &isnull); |
631 |
dpavlin |
57 |
attr = (char *)DirectFunctionCall1(textout, dvalue); |
632 |
dpavlin |
41 |
|
633 |
dpavlin |
57 |
if (!isnull && (hint = strstr(attr, HINTS_PREFIX)) != NULL) { |
634 |
|
|
/* skip HINTS. prefix */ |
635 |
|
|
hint += strlen(HINTS_PREFIX); |
636 |
|
|
|
637 |
dpavlin |
72 |
hint_val = (char *)cbmapget(hints, hint, -1, NULL); |
638 |
dpavlin |
57 |
elog(DEBUG2, "hint %s = %s", hint, hint_val); |
639 |
|
|
|
640 |
|
|
if (hint_val != NULL) { |
641 |
|
|
values[j] = DatumGetCString( hint_val ); |
642 |
|
|
} else { |
643 |
|
|
elog(INFO, "can't get hint in results: %s", hint); |
644 |
|
|
values[j] = NULL; |
645 |
|
|
} |
646 |
|
|
} else if (!isnull && rdoc) |
647 |
|
|
values[j] = DatumGetCString( node_attr2text(rdoc, attr) ); |
648 |
dpavlin |
41 |
else |
649 |
|
|
values[j] = NULL; |
650 |
|
|
} |
651 |
|
|
/* construct the tuple */ |
652 |
|
|
tuple = BuildTupleFromCStrings(attinmeta, values); |
653 |
|
|
|
654 |
|
|
/* now store it */ |
655 |
|
|
tuplestore_puttuple(tupstore, tuple); |
656 |
|
|
|
657 |
|
|
} |
658 |
|
|
|
659 |
|
|
tuplestore_donestoring(tupstore); |
660 |
|
|
rsinfo->setResult = tupstore; |
661 |
|
|
|
662 |
|
|
/* |
663 |
|
|
* SFRM_Materialize mode expects us to return a NULL Datum. The actual |
664 |
|
|
* tuples are in our tuplestore and passed back through |
665 |
|
|
* rsinfo->setResult. rsinfo->setDesc is set to the tuple description |
666 |
|
|
* that we actually used to build our tuples with, so the caller can |
667 |
|
|
* verify we did what it was expecting. |
668 |
|
|
*/ |
669 |
|
|
rsinfo->setDesc = tupdesc; |
670 |
|
|
MemoryContextSwitchTo(oldcontext); |
671 |
|
|
|
672 |
|
|
/* delete the node result object */ |
673 |
|
|
est_noderes_delete(nres); |
674 |
|
|
|
675 |
|
|
/* destroy the search condition object */ |
676 |
|
|
est_cond_delete(cond); |
677 |
|
|
|
678 |
|
|
/* destroy the node object */ |
679 |
|
|
est_node_delete(node); |
680 |
|
|
|
681 |
|
|
/* free the networking environment */ |
682 |
|
|
est_free_net_env(); |
683 |
|
|
|
684 |
|
|
return (Datum) 0; |
685 |
|
|
} |
686 |
|
|
|
687 |
|
|
/* make text var from node attr */ |
688 |
|
|
char *node_attr2text(ESTRESDOC *rdoc, char *attr) { |
689 |
|
|
char *val; |
690 |
|
|
const char *attrval; |
691 |
|
|
int len; |
692 |
|
|
int attrlen; |
693 |
|
|
|
694 |
dpavlin |
47 |
if (! rdoc) return (Datum) NULL; |
695 |
|
|
|
696 |
dpavlin |
56 |
elog(DEBUG1, "doc: %p, attr: %s", rdoc, attr); |
697 |
dpavlin |
41 |
|
698 |
|
|
if ( (attrval = est_resdoc_attr(rdoc, attr)) && (attrlen = strlen(attrval)) ) { |
699 |
|
|
val = (char *) palloc(attrlen * sizeof(char)); |
700 |
|
|
} else { |
701 |
|
|
return (Datum) NULL; |
702 |
|
|
} |
703 |
|
|
|
704 |
|
|
len = strlen(attrval); |
705 |
|
|
elog(DEBUG1, "node_attr2text(%s) = '%s' %d bytes", attr, attrval, len); |
706 |
|
|
|
707 |
|
|
len++; |
708 |
|
|
len *= sizeof(char); |
709 |
|
|
|
710 |
|
|
elog(DEBUG2, "palloc(%d)", len); |
711 |
|
|
|
712 |
|
|
val = palloc(len); |
713 |
|
|
|
714 |
|
|
memset(val, 0, len); |
715 |
|
|
strncpy(val, attrval, len); |
716 |
|
|
|
717 |
|
|
elog(DEBUG2, "val=%s", val); |
718 |
|
|
|
719 |
|
|
return val; |
720 |
|
|
} |
721 |
|
|
|
722 |
dpavlin |
56 |
/* parse attributes and add them to confition */ |
723 |
|
|
void cond_add_attr(ESTCOND *cond, char *attr) { |
724 |
|
|
char *next; |
725 |
|
|
char *curr_attr; |
726 |
|
|
while ( strlen(attr) > 0 ) { |
727 |
|
|
printf("len [%s] = %d\n", attr, strlen(attr)); |
728 |
|
|
if ((next = strstr(attr, ATTR_DELIMITER)) != NULL) { |
729 |
|
|
curr_attr = palloc( next - attr + 1 ); |
730 |
|
|
memcpy(curr_attr, attr, next-attr); |
731 |
|
|
curr_attr[next-attr] = '\0'; |
732 |
|
|
next += strlen(ATTR_DELIMITER); |
733 |
|
|
} else { |
734 |
|
|
next = ""; |
735 |
|
|
curr_attr = attr; |
736 |
|
|
} |
737 |
|
|
elog(DEBUG1, "est_cond_add_attr(%s)", curr_attr); |
738 |
|
|
est_cond_add_attr(cond, curr_attr); |
739 |
|
|
attr = next; |
740 |
|
|
} |
741 |
|
|
} |
742 |
dpavlin |
61 |
|
743 |
|
|
/* trigger to keep data in Hyper Estraier index up-to-date */ |
744 |
|
|
/* CREATE FUNCTION pgest_trigger() RETURNS TRIGGER AS ... */ |
745 |
|
|
|
746 |
|
|
/* |
747 |
|
|
* UPDATE, INSERT and DELETE triggers are like this: |
748 |
|
|
|
749 |
|
|
CREATE TRIGGER pgest_trigger_update AFTER UPDATE |
750 |
|
|
ON table_name FOR EACH ROW |
751 |
|
|
EXECUTE PROCEDURE pgest_trigger('http://localhost:1978/node/trivia','admin','admin', |
752 |
|
|
'name_of_pk', 'column', 'another_column', 'and_so_on' |
753 |
|
|
) |
754 |
|
|
|
755 |
|
|
*/ |
756 |
|
|
|
757 |
|
|
PG_FUNCTION_INFO_V1(pgest_trigger); |
758 |
|
|
Datum pgest_trigger(PG_FUNCTION_ARGS) { |
759 |
|
|
|
760 |
|
|
TriggerData *data; |
761 |
|
|
TupleDesc tupdesc; |
762 |
|
|
HeapTuple ret; |
763 |
|
|
|
764 |
|
|
char **args; |
765 |
|
|
char *keycol = NULL; |
766 |
|
|
char *key = NULL; |
767 |
|
|
char *col_data = NULL; |
768 |
|
|
int knumber; |
769 |
|
|
int i; |
770 |
|
|
int create_doc = 0; |
771 |
|
|
|
772 |
|
|
ESTNODE *node; |
773 |
|
|
ESTDOC *doc; |
774 |
|
|
|
775 |
|
|
|
776 |
|
|
|
777 |
|
|
if (! CALLED_AS_TRIGGER(fcinfo)) { |
778 |
|
|
elog(ERROR, "pgest_trigger() must be called as a trigger"); |
779 |
|
|
} |
780 |
|
|
|
781 |
|
|
data = (TriggerData *) fcinfo->context; |
782 |
|
|
|
783 |
|
|
if (data->tg_trigger->tgnargs < 5) |
784 |
|
|
elog(ERROR, "pgest_trigger() requires at least 5 parameters ('http://localhost:1978/node/trivia', 'user', 'passwd', 'pk_column', 'column', ... )"); |
785 |
|
|
|
786 |
|
|
args = data->tg_trigger->tgargs; |
787 |
|
|
keycol = args[3]; |
788 |
|
|
|
789 |
|
|
tupdesc = data->tg_relation->rd_att; |
790 |
|
|
|
791 |
|
|
knumber = SPI_fnumber(tupdesc, keycol); |
792 |
|
|
key = SPI_getvalue(data->tg_trigtuple, tupdesc, knumber); |
793 |
|
|
|
794 |
|
|
|
795 |
|
|
/* initialize the network environment */ |
796 |
|
|
if( ! est_init_net_env() ) |
797 |
|
|
elog(ERROR, "pgest_trigger: network is unavailable\n"); |
798 |
|
|
|
799 |
|
|
/* create and configure the node connection object */ |
800 |
|
|
node = est_node_new( args[0] ); |
801 |
|
|
est_node_set_auth(node, args[1], args[2]); |
802 |
|
|
|
803 |
|
|
|
804 |
|
|
if (TRIGGER_FIRED_BY_INSERT(data->tg_event)) { |
805 |
|
|
/* There is no old data */ |
806 |
|
|
ret = data->tg_trigtuple; |
807 |
|
|
|
808 |
|
|
create_doc++; |
809 |
|
|
|
810 |
|
|
} else if (TRIGGER_FIRED_BY_UPDATE(data->tg_event)) { |
811 |
|
|
ret = data->tg_newtuple; |
812 |
|
|
|
813 |
|
|
create_doc++; |
814 |
|
|
|
815 |
|
|
} else if (TRIGGER_FIRED_BY_DELETE(data->tg_event)) { |
816 |
|
|
/* There is no new data */ |
817 |
|
|
ret = data->tg_trigtuple; |
818 |
|
|
|
819 |
|
|
if (! est_node_out_doc_by_uri(node, key) ) |
820 |
|
|
elog(ERROR, "est_node_doc_by_uri(%s): %d\n", key, est_node_status(node)); |
821 |
|
|
|
822 |
|
|
} else { |
823 |
|
|
elog(ERROR, "pgest_trigger() not called from INSERT/UPDATE/DELETE"); |
824 |
|
|
} |
825 |
|
|
|
826 |
|
|
if ( create_doc ) { |
827 |
|
|
|
828 |
|
|
/* create a document object */ |
829 |
|
|
doc = est_doc_new(); |
830 |
|
|
est_doc_add_attr(doc, "@uri", key); |
831 |
|
|
|
832 |
|
|
elog(DEBUG1, "est_doc_new @uri=%s", key); |
833 |
|
|
|
834 |
|
|
for( i = 4; i < data->tg_trigger->tgnargs; i++ ) { |
835 |
|
|
|
836 |
|
|
col_data = SPI_getvalue(ret, tupdesc, SPI_fnumber(tupdesc, args[i])); |
837 |
|
|
|
838 |
|
|
if (data) { |
839 |
|
|
elog(DEBUG1, " + %s = %s", args[i], col_data); |
840 |
|
|
est_doc_add_attr(doc, args[i], col_data); |
841 |
|
|
est_doc_add_text(doc, col_data); |
842 |
|
|
} |
843 |
|
|
|
844 |
|
|
} |
845 |
|
|
|
846 |
|
|
/* register the document object to the node */ |
847 |
|
|
if( ! est_node_put_doc(node, doc) ) |
848 |
|
|
elog(ERROR, "est_node_put_doc: %d\n", est_node_status(node)); |
849 |
|
|
|
850 |
|
|
/* destroy the document object */ |
851 |
|
|
est_doc_delete(doc); |
852 |
|
|
|
853 |
|
|
} |
854 |
|
|
|
855 |
|
|
/* destroy the node object */ |
856 |
|
|
est_node_delete(node); |
857 |
|
|
/* free the networking environment */ |
858 |
|
|
est_free_net_env(); |
859 |
|
|
|
860 |
|
|
|
861 |
|
|
return PointerGetDatum(ret); |
862 |
|
|
} |
863 |
|
|
|