| 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 |
* - example002.c from Hyper Estraier (GPL) |
| 18 |
* - _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 |
#include "miscadmin.h" |
| 29 |
#include <estraier.h> |
| 30 |
#include <cabin.h> |
| 31 |
|
| 32 |
#define _textin(str) DirectFunctionCall1(textin, CStringGetDatum(str)) |
| 33 |
#define _textout(str) DatumGetPointer(DirectFunctionCall1(textout, PointerGetDatum(str))) |
| 34 |
#define GET_STR(textp) DatumGetCString(DirectFunctionCall1(textout, PointerGetDatum(textp))) |
| 35 |
#define GET_TEXT(cstrp) DatumGetTextP(DirectFunctionCall1(textin, CStringGetDatum(cstrp))) |
| 36 |
|
| 37 |
/* SortMem got renamed in PostgreSQL 8.0 */ |
| 38 |
#ifndef SortMem |
| 39 |
#define SortMem 16 * 1024 |
| 40 |
#endif |
| 41 |
|
| 42 |
/* prototype */ |
| 43 |
char *attr2text(ESTDOC *doc, char *attr); |
| 44 |
|
| 45 |
|
| 46 |
/* work in progress */ |
| 47 |
PG_FUNCTION_INFO_V1(pgest_attr); |
| 48 |
Datum pgest_attr(PG_FUNCTION_ARGS) |
| 49 |
{ |
| 50 |
ArrayType *attr_arr = PG_GETARG_ARRAYTYPE_P(5); |
| 51 |
Oid attr_element_type = ARR_ELEMTYPE(attr_arr); |
| 52 |
int attr_ndims = ARR_NDIM(attr_arr); |
| 53 |
int *attr_dim_counts = ARR_DIMS(attr_arr); |
| 54 |
int *attr_dim_lower_bounds = ARR_LBOUND(attr_arr); |
| 55 |
int ncols = 0; |
| 56 |
int nrows = 0; |
| 57 |
int indx[MAXDIM]; |
| 58 |
int16 attr_len; |
| 59 |
bool attr_byval; |
| 60 |
char attr_align; |
| 61 |
ReturnSetInfo *rsinfo = (ReturnSetInfo *) fcinfo->resultinfo; |
| 62 |
AttInMetadata *attinmeta; |
| 63 |
TupleDesc tupdesc; |
| 64 |
Tuplestorestate *tupstore = NULL; |
| 65 |
HeapTuple tuple; |
| 66 |
MemoryContext per_query_ctx; |
| 67 |
MemoryContext oldcontext; |
| 68 |
Datum dvalue; |
| 69 |
char **values; |
| 70 |
int rsinfo_ncols; |
| 71 |
int i, j; |
| 72 |
/* estvars */ |
| 73 |
ESTDB *db; |
| 74 |
ESTCOND *cond; |
| 75 |
ESTDOC *doc; |
| 76 |
const CBLIST *texts; |
| 77 |
int ecode, *est_result, resnum; |
| 78 |
int limit = 0; |
| 79 |
int offset = 0; |
| 80 |
|
| 81 |
char *index_path; |
| 82 |
char *query; |
| 83 |
char *attr; |
| 84 |
|
| 85 |
|
| 86 |
/* only allow 1D input array */ |
| 87 |
if (attr_ndims == 1) |
| 88 |
{ |
| 89 |
ncols = attr_dim_counts[0]; |
| 90 |
} |
| 91 |
else |
| 92 |
ereport(ERROR, |
| 93 |
(errcode(ERRCODE_INVALID_PARAMETER_VALUE), |
| 94 |
errmsg("invalid input array"), |
| 95 |
errdetail("Input array must have 1 dimension"))); |
| 96 |
|
| 97 |
/* check to see if caller supports us returning a tuplestore */ |
| 98 |
if (!rsinfo || !(rsinfo->allowedModes & SFRM_Materialize)) |
| 99 |
ereport(ERROR, |
| 100 |
(errcode(ERRCODE_SYNTAX_ERROR), |
| 101 |
errmsg("materialize mode required, but it is not " \ |
| 102 |
"allowed in this context"))); |
| 103 |
|
| 104 |
/* get info about element type needed to construct the array */ |
| 105 |
get_typlenbyvalalign(attr_element_type, &attr_len, &attr_byval, &attr_align); |
| 106 |
|
| 107 |
/* get the requested return tuple description */ |
| 108 |
tupdesc = rsinfo->expectedDesc; |
| 109 |
rsinfo_ncols = tupdesc->natts; |
| 110 |
|
| 111 |
/* |
| 112 |
* The requested tuple description better match up with the array |
| 113 |
* we were given. |
| 114 |
*/ |
| 115 |
if (rsinfo_ncols != ncols) |
| 116 |
ereport(ERROR, |
| 117 |
(errcode(ERRCODE_INVALID_PARAMETER_VALUE), |
| 118 |
errmsg("invalid input array"), |
| 119 |
errdetail("Number of elements in array must match number of query specified columns."))); |
| 120 |
|
| 121 |
/* OK, use it */ |
| 122 |
attinmeta = TupleDescGetAttInMetadata(tupdesc); |
| 123 |
|
| 124 |
/* Now go to work */ |
| 125 |
rsinfo->returnMode = SFRM_Materialize; |
| 126 |
|
| 127 |
per_query_ctx = fcinfo->flinfo->fn_mcxt; |
| 128 |
oldcontext = MemoryContextSwitchTo(per_query_ctx); |
| 129 |
|
| 130 |
/* initialize our tuplestore */ |
| 131 |
tupstore = tuplestore_begin_heap(true, false, SortMem); |
| 132 |
|
| 133 |
|
| 134 |
/* take rest of arguments from function */ |
| 135 |
|
| 136 |
/* index path */ |
| 137 |
if (PG_ARGISNULL(0)) { |
| 138 |
ereport(ERROR, |
| 139 |
(errcode(ERRCODE_INVALID_PARAMETER_VALUE), |
| 140 |
errmsg("index path can't be null"), |
| 141 |
errdetail("Index path must be valid full path to HyperEstraier index"))); |
| 142 |
} |
| 143 |
index_path = _textout(PG_GETARG_TEXT_P(0)); |
| 144 |
|
| 145 |
/* query string */ |
| 146 |
if (PG_ARGISNULL(0)) { |
| 147 |
query = ""; |
| 148 |
} else { |
| 149 |
query = _textout(PG_GETARG_TEXT_P(1)); |
| 150 |
} |
| 151 |
|
| 152 |
/* atribute filter */ |
| 153 |
if (PG_ARGISNULL(2)) { |
| 154 |
attr = ""; |
| 155 |
} else { |
| 156 |
attr = _textout(PG_GETARG_TEXT_P(2)); |
| 157 |
} |
| 158 |
|
| 159 |
/* limit */ |
| 160 |
if (PG_ARGISNULL(3)) { |
| 161 |
limit = 0; |
| 162 |
} else { |
| 163 |
limit = PG_GETARG_INT32(3); |
| 164 |
} |
| 165 |
|
| 166 |
/* offset */ |
| 167 |
if (PG_ARGISNULL(4)) { |
| 168 |
offset = 0; |
| 169 |
} else { |
| 170 |
offset = PG_GETARG_INT32(4); |
| 171 |
} |
| 172 |
|
| 173 |
|
| 174 |
/* open the database */ |
| 175 |
elog(DEBUG1, "pgest_attr: est_db_open(%s)", index_path); |
| 176 |
|
| 177 |
if(!(db = est_db_open(index_path, ESTDBREADER, &ecode))){ |
| 178 |
ereport(ERROR, (errcode(ERRCODE_INVALID_PARAMETER_VALUE), |
| 179 |
errmsg("est_db_open: can't open %s: %d", index_path, ecode), |
| 180 |
errdetail(est_err_msg(ecode)))); |
| 181 |
} |
| 182 |
|
| 183 |
elog(DEBUG1, "pgest_attr: query[%s] attr[%s] limit %d offset %d", query, (PG_ARGISNULL(2) ? "NULL" : attr), limit, offset); |
| 184 |
|
| 185 |
/* create a search condition object */ |
| 186 |
if (!(cond = est_cond_new())) { |
| 187 |
ereport(ERROR, (errcode(ERRCODE_QUERY_CANCELED), |
| 188 |
errmsg("pgest_attr: est_cond_new failed"))); |
| 189 |
} |
| 190 |
|
| 191 |
/* set the search phrase to the search condition object */ |
| 192 |
if (! PG_ARGISNULL(1) && strlen(query) > 0) |
| 193 |
est_cond_set_phrase(cond, query); |
| 194 |
|
| 195 |
/* minimum valid attribute length is 10: @a STREQ a */ |
| 196 |
if (! PG_ARGISNULL(2) && strlen(attr) >= 10) { |
| 197 |
elog(DEBUG1,"est_cond_add_attr(%s)", attr); |
| 198 |
est_cond_add_attr(cond, attr); |
| 199 |
} |
| 200 |
|
| 201 |
/* get the result of search */ |
| 202 |
est_result = est_db_search(db, cond, &resnum, NULL); |
| 203 |
|
| 204 |
/* check if results exists */ |
| 205 |
if ( 0 == resnum ) { |
| 206 |
elog(INFO, "pgest_attr: no results for: %s", query ); |
| 207 |
} |
| 208 |
|
| 209 |
/* total number of tuples to be returned */ |
| 210 |
if (limit && limit < resnum) { |
| 211 |
nrows = limit - offset; |
| 212 |
} else { |
| 213 |
nrows = resnum - offset; |
| 214 |
} |
| 215 |
|
| 216 |
|
| 217 |
elog(DEBUG1, "pgest_attr: found %d hits for %s", resnum, query); |
| 218 |
|
| 219 |
|
| 220 |
values = (char **) palloc(ncols * sizeof(char *)); |
| 221 |
|
| 222 |
for (i = 0; i < nrows; i++) |
| 223 |
{ |
| 224 |
|
| 225 |
/* get result from estraier */ |
| 226 |
if (! ( doc = est_db_get_doc(db, est_result[i + offset], 0)) ) { |
| 227 |
elog(INFO, "can't find result %d", i + offset); |
| 228 |
} else { |
| 229 |
elog(DEBUG1, "URI: %s\n Title: %s\n", |
| 230 |
est_doc_attr(doc, "@uri"), |
| 231 |
est_doc_attr(doc, "@title") |
| 232 |
); |
| 233 |
} |
| 234 |
|
| 235 |
/* iterate over results */ |
| 236 |
for (j = 0; j < ncols; j++) |
| 237 |
{ |
| 238 |
bool isnull; |
| 239 |
|
| 240 |
/* array value of this position */ |
| 241 |
indx[0] = j + attr_dim_lower_bounds[0]; |
| 242 |
|
| 243 |
dvalue = array_ref(attr_arr, attr_ndims, indx, -1, attr_len, attr_byval, attr_align, &isnull); |
| 244 |
|
| 245 |
if (!isnull && doc) |
| 246 |
values[j] = DatumGetCString( |
| 247 |
attr2text(doc, |
| 248 |
(char *)DirectFunctionCall1(textout, dvalue) |
| 249 |
)); |
| 250 |
else |
| 251 |
values[j] = NULL; |
| 252 |
} |
| 253 |
/* construct the tuple */ |
| 254 |
tuple = BuildTupleFromCStrings(attinmeta, values); |
| 255 |
|
| 256 |
/* now store it */ |
| 257 |
tuplestore_puttuple(tupstore, tuple); |
| 258 |
|
| 259 |
|
| 260 |
/* delete estraier document object */ |
| 261 |
est_doc_delete(doc); |
| 262 |
} |
| 263 |
|
| 264 |
tuplestore_donestoring(tupstore); |
| 265 |
rsinfo->setResult = tupstore; |
| 266 |
|
| 267 |
/* |
| 268 |
* SFRM_Materialize mode expects us to return a NULL Datum. The actual |
| 269 |
* tuples are in our tuplestore and passed back through |
| 270 |
* rsinfo->setResult. rsinfo->setDesc is set to the tuple description |
| 271 |
* that we actually used to build our tuples with, so the caller can |
| 272 |
* verify we did what it was expecting. |
| 273 |
*/ |
| 274 |
rsinfo->setDesc = tupdesc; |
| 275 |
MemoryContextSwitchTo(oldcontext); |
| 276 |
|
| 277 |
if(!est_db_close(db, &ecode)){ |
| 278 |
ereport(ERROR, (errcode(ERRCODE_IO_ERROR), |
| 279 |
errmsg("est_db_close: %d", ecode), |
| 280 |
errdetail(est_err_msg(ecode)))); |
| 281 |
} |
| 282 |
|
| 283 |
return (Datum) 0; |
| 284 |
} |
| 285 |
|
| 286 |
|
| 287 |
/* make text var from attr */ |
| 288 |
char *attr2text(ESTDOC *doc, char *attr) { |
| 289 |
char *val; |
| 290 |
const char *attrval; |
| 291 |
int len; |
| 292 |
int attrlen; |
| 293 |
|
| 294 |
elog(DEBUG1, "doc: %08x, attr: %s", doc, attr); |
| 295 |
|
| 296 |
if ( (attrval = est_doc_attr(doc, attr)) && (attrlen = strlen(attrval)) ) { |
| 297 |
val = (char *) palloc(attrlen * sizeof(char)); |
| 298 |
} else { |
| 299 |
return (Datum) NULL; |
| 300 |
} |
| 301 |
|
| 302 |
len = strlen(attrval); |
| 303 |
elog(DEBUG1, "attr2text(%s) = '%s' %d bytes", attr, attrval, len); |
| 304 |
|
| 305 |
len++; |
| 306 |
len *= sizeof(char); |
| 307 |
|
| 308 |
elog(DEBUG2, "palloc(%d)", len); |
| 309 |
|
| 310 |
val = palloc(len); |
| 311 |
|
| 312 |
memset(val, 0, len); |
| 313 |
strncpy(val, attrval, len); |
| 314 |
|
| 315 |
elog(DEBUG2, "val=%s", val); |
| 316 |
|
| 317 |
return val; |
| 318 |
} |
| 319 |
|
Parent Directory
| 
Revision Log