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)))

/* 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(5);
        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;


        /* 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(0)) {
                query = "";
        } else {
                query = _textout(PG_GETARG_TEXT_P(1));
        }

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

        /* limit */
        if (PG_ARGISNULL(3)) {
                limit = 0;
        } else {
                limit = PG_GETARG_INT32(3);
        }

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


        /* 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);
        }

        /* 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 - offset;
        } 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);

        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	5	/* prototype */
38			char attr2text(ESTDOC doc, char *attr);
39	dpavlin	1
40
41			/* work in progress */
42	dpavlin	19	PG_FUNCTION_INFO_V1(pgest_attr);
43			Datum pgest_attr(PG_FUNCTION_ARGS)
44	dpavlin	1	{
45	dpavlin	19	ArrayType *attr_arr = PG_GETARG_ARRAYTYPE_P(5);
46	dpavlin	25	Oid attr_element_type = ARR_ELEMTYPE(attr_arr);
47			int attr_ndims = ARR_NDIM(attr_arr);
48			int *attr_dim_counts = ARR_DIMS(attr_arr);
49			int *attr_dim_lower_bounds = ARR_LBOUND(attr_arr);
50	dpavlin	19	int ncols = 0;
51			int nrows = 0;
52			int indx[MAXDIM];
53	dpavlin	25	int16 attr_len;
54			bool attr_byval;
55			char attr_align;
56	dpavlin	1	ReturnSetInfo rsinfo = (ReturnSetInfo ) fcinfo->resultinfo;
57			AttInMetadata *attinmeta;
58			TupleDesc tupdesc;
59	dpavlin	19	Tuplestorestate *tupstore = NULL;
60	dpavlin	1	HeapTuple tuple;
61			MemoryContext per_query_ctx;
62			MemoryContext oldcontext;
63			Datum dvalue;
64			char **values;
65	dpavlin	19	int rsinfo_ncols;
66	dpavlin	1	int i, j;
67	dpavlin	19	/* estvars */
68	dpavlin	25	ESTDB *db;
69			ESTCOND *cond;
70			ESTDOC *doc;
71			const CBLIST *texts;
72			int ecode, *est_result, resnum;
73			int limit = 0;
74			int offset = 0;
75
76	dpavlin	19	char *index_path;
77			char *query;
78			char *attr;
79	dpavlin	1
80	dpavlin	19
81			/* only allow 1D input array */
82	dpavlin	25	if (attr_ndims == 1)
83	dpavlin	19	{
84	dpavlin	25	ncols = attr_dim_counts[0];
85	dpavlin	19	}
86			else
87			ereport(ERROR,
88			(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
89			errmsg("invalid input array"),
90			errdetail("Input array must have 1 dimension")));
91
92	dpavlin	1	/* check to see if caller supports us returning a tuplestore */
93			if (!rsinfo \|\| !(rsinfo->allowedModes & SFRM_Materialize))
94			ereport(ERROR,
95			(errcode(ERRCODE_SYNTAX_ERROR),
96			errmsg("materialize mode required, but it is not " \
97			"allowed in this context")));
98
99	dpavlin	19	/* get info about element type needed to construct the array */
100	dpavlin	25	get_typlenbyvalalign(attr_element_type, &attr_len, &attr_byval, &attr_align);
101	dpavlin	19
102	dpavlin	1	/* get the requested return tuple description */
103			tupdesc = rsinfo->expectedDesc;
104	dpavlin	19	rsinfo_ncols = tupdesc->natts;
105	dpavlin	1
106			/*
107			* The requested tuple description better match up with the array
108			* we were given.
109			*/
110	dpavlin	19	if (rsinfo_ncols != ncols)
111			ereport(ERROR,
112			(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
113			errmsg("invalid input array"),
114			errdetail("Number of elements in array must match number of query specified columns.")));
115
116	dpavlin	1	/* OK, use it */
117			attinmeta = TupleDescGetAttInMetadata(tupdesc);
118
119			/* Now go to work */
120			rsinfo->returnMode = SFRM_Materialize;
121
122			per_query_ctx = fcinfo->flinfo->fn_mcxt;
123			oldcontext = MemoryContextSwitchTo(per_query_ctx);
124
125			/* initialize our tuplestore */
126			tupstore = tuplestore_begin_heap(true, false, SortMem);
127
128	dpavlin	19
129			/* take rest of arguments from function */
130
131			/* index path */
132			if (PG_ARGISNULL(0)) {
133			ereport(ERROR,
134			(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
135			errmsg("index path can't be null"),
136			errdetail("Index path must be valid full path to HyperEstraier index")));
137			}
138			index_path = _textout(PG_GETARG_TEXT_P(0));
139
140			/* query string */
141			if (PG_ARGISNULL(0)) {
142			query = "";
143			} else {
144			query = _textout(PG_GETARG_TEXT_P(1));
145			}
146
147			/* atribute filter */
148			if (PG_ARGISNULL(2)) {
149			attr = "";
150			} else {
151			attr = _textout(PG_GETARG_TEXT_P(2));
152			}
153
154			/* limit */
155			if (PG_ARGISNULL(3)) {
156			limit = 0;
157			} else {
158			limit = PG_GETARG_INT32(3);
159			}
160
161			/* offset */
162			if (PG_ARGISNULL(4)) {
163			offset = 0;
164			} else {
165			offset = PG_GETARG_INT32(4);
166			}
167
168
169			/* open the database */
170			elog(DEBUG1, "pgest_attr: est_db_open(%s)", index_path);
171
172			if(!(db = est_db_open(index_path, ESTDBREADER, &ecode))){
173			ereport(ERROR, (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
174			errmsg("est_db_open: can't open %s: %d", index_path, ecode),
175			errdetail(est_err_msg(ecode))));
176			}
177
178	dpavlin	20	elog(DEBUG1, "pgest_attr: query[%s] attr[%s] limit %d offset %d", query, (PG_ARGISNULL(2) ? "NULL" : attr), limit, offset);
179	dpavlin	19
180			/* create a search condition object */
181			if (!(cond = est_cond_new())) {
182			ereport(ERROR, (errcode(ERRCODE_QUERY_CANCELED),
183			errmsg("pgest_attr: est_cond_new failed")));
184			}
185
186			/* set the search phrase to the search condition object */
187			if (! PG_ARGISNULL(1) && strlen(query) > 0)
188			est_cond_set_phrase(cond, query);
189
190			/* minimum valid attribute length is 10: @a STREQ a */
191			if (! PG_ARGISNULL(2) && strlen(attr) >= 10) {
192	dpavlin	20	elog(DEBUG1,"est_cond_add_attr(%s)", attr);
193	dpavlin	19	est_cond_add_attr(cond, attr);
194			}
195
196			/* get the result of search */
197			est_result = est_db_search(db, cond, &resnum, NULL);
198
199			/* check if results exists */
200			if ( 0 == resnum ) {
201			elog(INFO, "pgest_attr: no results for: %s", query );
202			}
203
204			/* total number of tuples to be returned */
205			if (limit && limit < resnum) {
206			nrows = limit - offset;
207			} else {
208			nrows = resnum - offset;
209			}
210
211
212			elog(DEBUG1, "pgest_attr: found %d hits for %s", resnum, query);
213
214
215	dpavlin	1	values = (char *) palloc(ncols sizeof(char *));
216
217			for (i = 0; i < nrows; i++)
218			{
219	dpavlin	19
220			/* get result from estraier */
221			if (! ( doc = est_db_get_doc(db, est_result[i + offset], 0)) ) {
222			elog(INFO, "can't find result %d", i + offset);
223			} else {
224			elog(DEBUG1, "URI: %s\n Title: %s\n",
225			est_doc_attr(doc, "@uri"),
226			est_doc_attr(doc, "@title")
227			);
228			}
229
230			/* iterate over results */
231	dpavlin	1	for (j = 0; j < ncols; j++)
232			{
233	dpavlin	19	bool isnull;
234
235			/* array value of this position */
236	dpavlin	25	indx[0] = j + attr_dim_lower_bounds[0];
237	dpavlin	19
238	dpavlin	25	dvalue = array_ref(attr_arr, attr_ndims, indx, -1, attr_len, attr_byval, attr_align, &isnull);
239	dpavlin	19
240			if (!isnull && doc)
241			values[j] = DatumGetCString(
242			attr2text(doc,
243	dpavlin	20	(char *)DirectFunctionCall1(textout, dvalue)
244	dpavlin	19	));
245			else
246			values[j] = NULL;
247	dpavlin	1	}
248			/* construct the tuple */
249			tuple = BuildTupleFromCStrings(attinmeta, values);
250
251			/* now store it */
252			tuplestore_puttuple(tupstore, tuple);
253	dpavlin	19
254
255			/* delete estraier document object */
256			est_doc_delete(doc);
257	dpavlin	1	}
258
259			tuplestore_donestoring(tupstore);
260			rsinfo->setResult = tupstore;
261
262			/*
263			* SFRM_Materialize mode expects us to return a NULL Datum. The actual
264			* tuples are in our tuplestore and passed back through
265			* rsinfo->setResult. rsinfo->setDesc is set to the tuple description
266			* that we actually used to build our tuples with, so the caller can
267			* verify we did what it was expecting.
268			*/
269			rsinfo->setDesc = tupdesc;
270			MemoryContextSwitchTo(oldcontext);
271
272	dpavlin	19	if(!est_db_close(db, &ecode)){
273			ereport(ERROR, (errcode(ERRCODE_IO_ERROR),
274			errmsg("est_db_close: %d", ecode),
275			errdetail(est_err_msg(ecode))));
276			}
277
278	dpavlin	1	return (Datum) 0;
279			}
280
281
282			/* make text var from attr */
283			char attr2text(ESTDOC doc, char *attr) {
284			char *val;
285			const char *attrval;
286			int len;
287	dpavlin	4	int attrlen;
288	dpavlin	1
289	dpavlin	2	elog(DEBUG1, "doc: %08x, attr: %s", doc, attr);
290	dpavlin	1
291	dpavlin	4	if ( (attrval = est_doc_attr(doc, attr)) && (attrlen = strlen(attrval)) ) {
292			val = (char ) palloc(attrlen sizeof(char));
293	dpavlin	1	} else {
294			return (Datum) NULL;
295			}
296
297			len = strlen(attrval);
298	dpavlin	2	elog(DEBUG1, "attr2text(%s) = '%s' %d bytes", attr, attrval, len);
299	dpavlin	1
300			len++;
301			len *= sizeof(char);
302
303			elog(DEBUG2, "palloc(%d)", len);
304
305			val = palloc(len);
306
307			memset(val, 0, len);
308			strncpy(val, attrval, len);
309
310			elog(DEBUG2, "val=%s", val);
311
312			return val;
313			}
314