pgestraier/trunk/pgest.c

/*
 * integrate Hyper Estraier into PostgreSQL
 *
 * Dobrica Pavlinusic <dpavlin@rot13.org> 2005-05-19
 *
 * TODO:
 * - all
 *
 * NOTES:
 * - clear structures with memset to support hash indexes (who whould like
 *   to create hash index on table returned from function?)
 * - number of returned rows is set by PostgreSQL evaluator, see:
 *   http://archives.postgresql.org/pgsql-hackers/2005-02/msg00546.php
 *
 * Based on:
 * - C example from PostgreSQL documentation (BSD licence)
 * - example002.c from Hyper Estraier (GPL)
 * - _textin/_textout from pgcurl.c (LGPL)
 *
 * This code is licenced under GPL
 */

#include "postgres.h"
#include "fmgr.h"
#include "funcapi.h"
#include "utils/builtins.h"
#include "utils/array.h"
#include "miscadmin.h"
#include <estraier.h>
#include <cabin.h>

#define _textin(str) DirectFunctionCall1(textin, CStringGetDatum(str))
#define _textout(str) DatumGetPointer(DirectFunctionCall1(textout, PointerGetDatum(str)))
#define GET_STR(textp) DatumGetCString(DirectFunctionCall1(textout, PointerGetDatum(textp)))
#define GET_TEXT(cstrp) DatumGetTextP(DirectFunctionCall1(textin, CStringGetDatum(cstrp)))

/* SortMem got renamed in PostgreSQL 8.0 */
#ifndef SortMem
 #define SortMem 16 * 1024
#endif

/* prototype */
char *attr2text(ESTDOC *doc, char *attr);


/* work in progress */
PG_FUNCTION_INFO_V1(pgest_attr);
Datum pgest_attr(PG_FUNCTION_ARGS)
{
        ArrayType       *attr_arr = PG_GETARG_ARRAYTYPE_P(6);
        Oid             attr_element_type = ARR_ELEMTYPE(attr_arr);
        int             attr_ndims = ARR_NDIM(attr_arr);
        int             *attr_dim_counts = ARR_DIMS(attr_arr);
        int             *attr_dim_lower_bounds = ARR_LBOUND(attr_arr);
        int             ncols = 0;
        int             nrows = 0;
        int             indx[MAXDIM];
        int16           attr_len;
        bool            attr_byval;
        char            attr_align;
        ReturnSetInfo   *rsinfo = (ReturnSetInfo *) fcinfo->resultinfo;
        AttInMetadata   *attinmeta;
        TupleDesc       tupdesc;
        Tuplestorestate *tupstore = NULL;
        HeapTuple       tuple;
        MemoryContext   per_query_ctx;
        MemoryContext   oldcontext;
        Datum           dvalue;
        char            **values;
        int             rsinfo_ncols;
        int             i, j;
        /* estvars */
        ESTDB *db;
        ESTCOND *cond;
        ESTDOC *doc;
        const CBLIST *texts;
        int ecode, *est_result, resnum;
        int limit = 0;
        int offset = 0;

        char            *index_path;
        char            *query;
        char            *attr;
        char            *order;


        /* only allow 1D input array */
        if (attr_ndims == 1)
        {
                ncols = attr_dim_counts[0];
        }
        else
                ereport(ERROR,
                                (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
                                 errmsg("invalid input array"),
                                 errdetail("Input array must have 1 dimension")));
                
        /* check to see if caller supports us returning a tuplestore */
        if (!rsinfo || !(rsinfo->allowedModes & SFRM_Materialize))
                ereport(ERROR,
                                (errcode(ERRCODE_SYNTAX_ERROR),
                                 errmsg("materialize mode required, but it is not " \
                                                "allowed in this context")));

        /* get info about element type needed to construct the array */
        get_typlenbyvalalign(attr_element_type, &attr_len, &attr_byval, &attr_align);

        /* get the requested return tuple description */
        tupdesc = rsinfo->expectedDesc;
        rsinfo_ncols = tupdesc->natts;

        /*
         * The requested tuple description better match up with the array
         * we were given.
         */
        if (rsinfo_ncols != ncols)
                ereport(ERROR,
                                (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
                                 errmsg("invalid input array"),
                                 errdetail("Number of elements in array must match number of query specified columns.")));

        /* OK, use it */
        attinmeta = TupleDescGetAttInMetadata(tupdesc);

        /* Now go to work */
        rsinfo->returnMode = SFRM_Materialize;

        per_query_ctx = fcinfo->flinfo->fn_mcxt;
        oldcontext = MemoryContextSwitchTo(per_query_ctx);

        /* initialize our tuplestore */
        tupstore = tuplestore_begin_heap(true, false, SortMem);


        /* take rest of arguments from function */

        /* index path */
        if (PG_ARGISNULL(0)) {
                ereport(ERROR,
                                (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
                                 errmsg("index path can't be null"),
                                 errdetail("Index path must be valid full path to HyperEstraier index")));
        }
        index_path = _textout(PG_GETARG_TEXT_P(0));

        /* query string */
        if (PG_ARGISNULL(1)) {
                query = "";
        } else {
                query = _textout(PG_GETARG_TEXT_P(1));
        }

        /* atribute filter */
        if (PG_ARGISNULL(2)) {
                attr = "";
        } else {
                attr = _textout(PG_GETARG_TEXT_P(2));
        }
        
        /* sort order */
        if (PG_ARGISNULL(3)) {
                order = "";
        } else {
                order = _textout(PG_GETARG_TEXT_P(3));
        }


        /* limit */
        if (PG_ARGISNULL(4)) {
                limit = 0;
        } else {
                limit = PG_GETARG_INT32(4);
        }

        /* offset */
        if (PG_ARGISNULL(5)) {
                offset = 0;
        } else {
                offset = PG_GETARG_INT32(5);
        }


        /* open the database */
        elog(DEBUG1, "pgest_attr: est_db_open(%s)", index_path);
                
        if(!(db = est_db_open(index_path, ESTDBREADER, &ecode))){
                ereport(ERROR, (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
                        errmsg("est_db_open: can't open %s: %d", index_path, ecode),
                        errdetail(est_err_msg(ecode))));
        }
                
        elog(DEBUG1, "pgest_attr: query[%s] attr[%s] limit %d offset %d", query, (PG_ARGISNULL(2) ? "NULL" : attr), limit, offset);
        
        /* create a search condition object */
        if (!(cond = est_cond_new())) {
                ereport(ERROR, (errcode(ERRCODE_QUERY_CANCELED),
                        errmsg("pgest_attr: est_cond_new failed")));
        }
        
        /* set the search phrase to the search condition object */
        if (! PG_ARGISNULL(1) && strlen(query) > 0)
                est_cond_set_phrase(cond, query);

        /* minimum valid attribute length is 10: @a STREQ a */
        if (! PG_ARGISNULL(2) && strlen(attr) >= 10) {
                elog(DEBUG1,"est_cond_add_attr(%s)", attr);
                est_cond_add_attr(cond, attr);
        }

        /* set the search phrase to the search condition object */
        if (! PG_ARGISNULL(3) && strlen(order) > 0) {
                elog(DEBUG1,"est_cond_set_order(%s)", order);
                est_cond_set_order(cond, order);
        }

        if (limit) {
                elog(DEBUG1,"est_cond_set_max(%d)", limit + offset);
                est_cond_set_max(cond, limit + offset);
        }

        /* get the result of search */
        est_result = est_db_search(db, cond, &resnum, NULL);

        /* check if results exists */
        if ( 0 == resnum ) {
                elog(INFO, "pgest_attr: no results for: %s", query );
        }

        /* total number of tuples to be returned */
        if (limit && limit < resnum) {
                nrows = limit;
                elog(INFO, "This is probably a bug in limit implementation. Please report it to dpavlin@rot13.org");
        } else {
                nrows = resnum - offset;
        }


        elog(DEBUG1, "pgest_attr: found %d hits for %s", resnum, query);


        values = (char **) palloc(ncols * sizeof(char *));

        for (i = 0; i < nrows; i++)
        {

                /* get result from estraier */
                if (! ( doc = est_db_get_doc(db, est_result[i + offset], 0)) ) {
                        elog(INFO, "can't find result %d", i + offset);
                } else {
                        elog(DEBUG1, "URI: %s\n Title: %s\n",
                                est_doc_attr(doc, "@uri"),
                                est_doc_attr(doc, "@title")
                        );
                }

                /* iterate over results */
                for (j = 0; j < ncols; j++)
                {
                        bool    isnull;

                        /* array value of this position */
                        indx[0] = j + attr_dim_lower_bounds[0];

                        dvalue = array_ref(attr_arr, attr_ndims, indx, -1, attr_len, attr_byval, attr_align, &isnull);

                        if (!isnull && doc)
                                values[j] = DatumGetCString(
                                        attr2text(doc,
                                                (char *)DirectFunctionCall1(textout, dvalue)
                                        ));
                        else
                                values[j] = NULL;
                }
                /* construct the tuple */
                tuple = BuildTupleFromCStrings(attinmeta, values);

                /* now store it */
                tuplestore_puttuple(tupstore, tuple);


                /* delete estraier document object */
                est_doc_delete(doc);
        }

        tuplestore_donestoring(tupstore);
        rsinfo->setResult = tupstore;

        /*
         * SFRM_Materialize mode expects us to return a NULL Datum. The actual
         * tuples are in our tuplestore and passed back through
         * rsinfo->setResult. rsinfo->setDesc is set to the tuple description
         * that we actually used to build our tuples with, so the caller can
         * verify we did what it was expecting.
         */
        rsinfo->setDesc = tupdesc;
        MemoryContextSwitchTo(oldcontext);

        est_cond_delete(cond);

        if(!est_db_close(db, &ecode)){
                ereport(ERROR, (errcode(ERRCODE_IO_ERROR),
                        errmsg("est_db_close: %d", ecode),
                        errdetail(est_err_msg(ecode))));
        }

        return (Datum) 0;
}


/* make text var from attr */
char *attr2text(ESTDOC *doc, char *attr) {
        char *val;
        const char *attrval;
        int len;
        int attrlen;

        elog(DEBUG1, "doc: %08x, attr: %s", doc, attr);

        if ( (attrval = est_doc_attr(doc, attr)) && (attrlen = strlen(attrval)) ) {
                val = (char *) palloc(attrlen * sizeof(char));
        } else {
                return (Datum) NULL;
        }

        len = strlen(attrval);
        elog(DEBUG1, "attr2text(%s) = '%s' %d bytes", attr, attrval, len);

        len++;
        len *= sizeof(char);

        elog(DEBUG2, "palloc(%d)", len);

        val = palloc(len);

        memset(val, 0, len);
        strncpy(val, attrval, len);

        elog(DEBUG2, "val=%s", val);

        return val;
}

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			* - 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	dpavlin	27	/* SortMem got renamed in PostgreSQL 8.0 */
38			#ifndef SortMem
39			#define SortMem 16 * 1024
40			#endif
41
42	dpavlin	5	/* prototype */
43			char attr2text(ESTDOC doc, char *attr);
44	dpavlin	1
45
46			/* work in progress */
47	dpavlin	19	PG_FUNCTION_INFO_V1(pgest_attr);
48			Datum pgest_attr(PG_FUNCTION_ARGS)
49	dpavlin	1	{
50	dpavlin	31	ArrayType *attr_arr = PG_GETARG_ARRAYTYPE_P(6);
51	dpavlin	25	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	dpavlin	19	int ncols = 0;
56			int nrows = 0;
57			int indx[MAXDIM];
58	dpavlin	25	int16 attr_len;
59			bool attr_byval;
60			char attr_align;
61	dpavlin	1	ReturnSetInfo rsinfo = (ReturnSetInfo ) fcinfo->resultinfo;
62			AttInMetadata *attinmeta;
63			TupleDesc tupdesc;
64	dpavlin	19	Tuplestorestate *tupstore = NULL;
65	dpavlin	1	HeapTuple tuple;
66			MemoryContext per_query_ctx;
67			MemoryContext oldcontext;
68			Datum dvalue;
69			char **values;
70	dpavlin	19	int rsinfo_ncols;
71	dpavlin	1	int i, j;
72	dpavlin	19	/* estvars */
73	dpavlin	25	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	dpavlin	19	char *index_path;
82			char *query;
83			char *attr;
84	dpavlin	31	char *order;
85	dpavlin	1
86	dpavlin	19
87			/* only allow 1D input array */
88	dpavlin	25	if (attr_ndims == 1)
89	dpavlin	19	{
90	dpavlin	25	ncols = attr_dim_counts[0];
91	dpavlin	19	}
92			else
93			ereport(ERROR,
94			(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
95			errmsg("invalid input array"),
96			errdetail("Input array must have 1 dimension")));
97
98	dpavlin	1	/* check to see if caller supports us returning a tuplestore */
99			if (!rsinfo \|\| !(rsinfo->allowedModes & SFRM_Materialize))
100			ereport(ERROR,
101			(errcode(ERRCODE_SYNTAX_ERROR),
102			errmsg("materialize mode required, but it is not " \
103			"allowed in this context")));
104
105	dpavlin	19	/* get info about element type needed to construct the array */
106	dpavlin	25	get_typlenbyvalalign(attr_element_type, &attr_len, &attr_byval, &attr_align);
107	dpavlin	19
108	dpavlin	1	/* get the requested return tuple description */
109			tupdesc = rsinfo->expectedDesc;
110	dpavlin	19	rsinfo_ncols = tupdesc->natts;
111	dpavlin	1
112			/*
113			* The requested tuple description better match up with the array
114			* we were given.
115			*/
116	dpavlin	19	if (rsinfo_ncols != ncols)
117			ereport(ERROR,
118			(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
119			errmsg("invalid input array"),
120			errdetail("Number of elements in array must match number of query specified columns.")));
121
122	dpavlin	1	/* OK, use it */
123			attinmeta = TupleDescGetAttInMetadata(tupdesc);
124
125			/* Now go to work */
126			rsinfo->returnMode = SFRM_Materialize;
127
128			per_query_ctx = fcinfo->flinfo->fn_mcxt;
129			oldcontext = MemoryContextSwitchTo(per_query_ctx);
130
131			/* initialize our tuplestore */
132			tupstore = tuplestore_begin_heap(true, false, SortMem);
133
134	dpavlin	19
135			/* take rest of arguments from function */
136
137			/* index path */
138			if (PG_ARGISNULL(0)) {
139			ereport(ERROR,
140			(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
141			errmsg("index path can't be null"),
142			errdetail("Index path must be valid full path to HyperEstraier index")));
143			}
144			index_path = _textout(PG_GETARG_TEXT_P(0));
145
146			/* query string */
147	dpavlin	31	if (PG_ARGISNULL(1)) {
148	dpavlin	19	query = "";
149			} else {
150			query = _textout(PG_GETARG_TEXT_P(1));
151			}
152
153			/* atribute filter */
154			if (PG_ARGISNULL(2)) {
155			attr = "";
156			} else {
157			attr = _textout(PG_GETARG_TEXT_P(2));
158			}
159	dpavlin	31
160			/* sort order */
161			if (PG_ARGISNULL(3)) {
162			order = "";
163			} else {
164			order = _textout(PG_GETARG_TEXT_P(3));
165			}
166	dpavlin	19
167	dpavlin	31
168	dpavlin	19	/* limit */
169	dpavlin	31	if (PG_ARGISNULL(4)) {
170	dpavlin	19	limit = 0;
171			} else {
172	dpavlin	31	limit = PG_GETARG_INT32(4);
173	dpavlin	19	}
174
175			/* offset */
176	dpavlin	31	if (PG_ARGISNULL(5)) {
177	dpavlin	19	offset = 0;
178			} else {
179	dpavlin	31	offset = PG_GETARG_INT32(5);
180	dpavlin	19	}
181
182
183			/* open the database */
184			elog(DEBUG1, "pgest_attr: est_db_open(%s)", index_path);
185
186			if(!(db = est_db_open(index_path, ESTDBREADER, &ecode))){
187			ereport(ERROR, (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
188			errmsg("est_db_open: can't open %s: %d", index_path, ecode),
189			errdetail(est_err_msg(ecode))));
190			}
191
192	dpavlin	20	elog(DEBUG1, "pgest_attr: query[%s] attr[%s] limit %d offset %d", query, (PG_ARGISNULL(2) ? "NULL" : attr), limit, offset);
193	dpavlin	19
194			/* create a search condition object */
195			if (!(cond = est_cond_new())) {
196			ereport(ERROR, (errcode(ERRCODE_QUERY_CANCELED),
197			errmsg("pgest_attr: est_cond_new failed")));
198			}
199
200			/* set the search phrase to the search condition object */
201			if (! PG_ARGISNULL(1) && strlen(query) > 0)
202			est_cond_set_phrase(cond, query);
203
204			/* minimum valid attribute length is 10: @a STREQ a */
205			if (! PG_ARGISNULL(2) && strlen(attr) >= 10) {
206	dpavlin	20	elog(DEBUG1,"est_cond_add_attr(%s)", attr);
207	dpavlin	19	est_cond_add_attr(cond, attr);
208			}
209
210	dpavlin	31	/* set the search phrase to the search condition object */
211			if (! PG_ARGISNULL(3) && strlen(order) > 0) {
212			elog(DEBUG1,"est_cond_set_order(%s)", order);
213			est_cond_set_order(cond, order);
214			}
215
216	dpavlin	38	if (limit) {
217			elog(DEBUG1,"est_cond_set_max(%d)", limit + offset);
218			est_cond_set_max(cond, limit + offset);
219			}
220
221	dpavlin	19	/* get the result of search */
222			est_result = est_db_search(db, cond, &resnum, NULL);
223
224			/* check if results exists */
225			if ( 0 == resnum ) {
226			elog(INFO, "pgest_attr: no results for: %s", query );
227			}
228
229			/* total number of tuples to be returned */
230			if (limit && limit < resnum) {
231	dpavlin	31	nrows = limit;
232	dpavlin	38	elog(INFO, "This is probably a bug in limit implementation. Please report it to dpavlin@rot13.org");
233	dpavlin	19	} else {
234			nrows = resnum - offset;
235			}
236
237
238			elog(DEBUG1, "pgest_attr: found %d hits for %s", resnum, query);
239
240
241	dpavlin	1	values = (char *) palloc(ncols sizeof(char *));
242
243			for (i = 0; i < nrows; i++)
244			{
245	dpavlin	19
246			/* get result from estraier */
247			if (! ( doc = est_db_get_doc(db, est_result[i + offset], 0)) ) {
248			elog(INFO, "can't find result %d", i + offset);
249			} else {
250			elog(DEBUG1, "URI: %s\n Title: %s\n",
251			est_doc_attr(doc, "@uri"),
252			est_doc_attr(doc, "@title")
253			);
254			}
255
256			/* iterate over results */
257	dpavlin	1	for (j = 0; j < ncols; j++)
258			{
259	dpavlin	19	bool isnull;
260
261			/* array value of this position */
262	dpavlin	25	indx[0] = j + attr_dim_lower_bounds[0];
263	dpavlin	19
264	dpavlin	25	dvalue = array_ref(attr_arr, attr_ndims, indx, -1, attr_len, attr_byval, attr_align, &isnull);
265	dpavlin	19
266			if (!isnull && doc)
267			values[j] = DatumGetCString(
268			attr2text(doc,
269	dpavlin	20	(char *)DirectFunctionCall1(textout, dvalue)
270	dpavlin	19	));
271			else
272			values[j] = NULL;
273	dpavlin	1	}
274			/* construct the tuple */
275			tuple = BuildTupleFromCStrings(attinmeta, values);
276
277			/* now store it */
278			tuplestore_puttuple(tupstore, tuple);
279	dpavlin	19
280
281			/* delete estraier document object */
282			est_doc_delete(doc);
283	dpavlin	1	}
284
285			tuplestore_donestoring(tupstore);
286			rsinfo->setResult = tupstore;
287
288			/*
289			* SFRM_Materialize mode expects us to return a NULL Datum. The actual
290			* tuples are in our tuplestore and passed back through
291			* rsinfo->setResult. rsinfo->setDesc is set to the tuple description
292			* that we actually used to build our tuples with, so the caller can
293			* verify we did what it was expecting.
294			*/
295			rsinfo->setDesc = tupdesc;
296			MemoryContextSwitchTo(oldcontext);
297
298	dpavlin	31	est_cond_delete(cond);
299
300	dpavlin	19	if(!est_db_close(db, &ecode)){
301			ereport(ERROR, (errcode(ERRCODE_IO_ERROR),
302			errmsg("est_db_close: %d", ecode),
303			errdetail(est_err_msg(ecode))));
304			}
305
306	dpavlin	1	return (Datum) 0;
307			}
308
309
310			/* make text var from attr */
311			char attr2text(ESTDOC doc, char *attr) {
312			char *val;
313			const char *attrval;
314			int len;
315	dpavlin	4	int attrlen;
316	dpavlin	1
317	dpavlin	2	elog(DEBUG1, "doc: %08x, attr: %s", doc, attr);
318	dpavlin	1
319	dpavlin	4	if ( (attrval = est_doc_attr(doc, attr)) && (attrlen = strlen(attrval)) ) {
320			val = (char ) palloc(attrlen sizeof(char));
321	dpavlin	1	} else {
322			return (Datum) NULL;
323			}
324
325			len = strlen(attrval);
326	dpavlin	2	elog(DEBUG1, "attr2text(%s) = '%s' %d bytes", attr, attrval, len);
327	dpavlin	1
328			len++;
329			len *= sizeof(char);
330
331			elog(DEBUG2, "palloc(%d)", len);
332
333			val = palloc(len);
334
335			memset(val, 0, len);
336			strncpy(val, attrval, len);
337
338			elog(DEBUG2, "val=%s", val);
339
340			return val;
341			}
342