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)
 * - coreexample002.c and nodeexample002.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>
#include <estnode.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

#define ATTR_DELIMITER "{{!}}"

/* prototype */
char *attr2text(ESTDOC *doc, char *attr);
char *node_attr2text(ESTRESDOC *rdoc, 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,"attributes: %s", attr);
                char *curr_attr;
                curr_attr = strtok(attr, ATTR_DELIMITER);
                while (curr_attr) {
                        elog(DEBUG1,"est_cond_add_attr(%s)", curr_attr);
                        est_cond_add_attr(cond, curr_attr);
                        curr_attr = strtok(NULL, ATTR_DELIMITER);
                }
        }

        /* 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;
        } 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;
}

/*
 * variation on theme: use node API which doesn't open index on
 * every query which is much faster for large indexes
 *
 */

PG_FUNCTION_INFO_V1(pgest_node);
Datum pgest_node(PG_FUNCTION_ARGS)
{
        ArrayType       *attr_arr = PG_GETARG_ARRAYTYPE_P(8);
        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 */
        ESTNODE *node;
        ESTCOND *cond;
        ESTNODERES *nres;
        ESTRESDOC *rdoc;
        const CBLIST *texts;
        int resnum = 0;
        int limit = 0;
        int offset = 0;

        char            *node_url;
        char            *user, *passwd;
        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 */

        /* node URL */
        if (PG_ARGISNULL(0)) {
                ereport(ERROR,
                                (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
                                 errmsg("node URL can't be null"),
                                 errdetail("Node URL must be valid URL to HyperEstraier node")));
        }
        node_url = _textout(PG_GETARG_TEXT_P(0));

        /* login and password */
        if (PG_ARGISNULL(1) || PG_ARGISNULL(2)) {
                ereport(ERROR,
                                (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
                                 errmsg("username and password can't be NULL"),
                                 errdetail("You must specify valid username and password to HyperEstraier node")));
        }
        user = _textout(PG_GETARG_TEXT_P(1));
        passwd = _textout(PG_GETARG_TEXT_P(2));

        /* query string */
        if (PG_ARGISNULL(3)) {
                query = "";
        } else {
                query = _textout(PG_GETARG_TEXT_P(3));
        }

        /* atribute filter */
        if (PG_ARGISNULL(4)) {
                attr = "";
        } else {
                attr = _textout(PG_GETARG_TEXT_P(4));
        }
        
        /* sort order */
        if (PG_ARGISNULL(5)) {
                order = "";
        } else {
                order = _textout(PG_GETARG_TEXT_P(5));
        }


        /* limit */
        if (PG_ARGISNULL(6)) {
                limit = 0;
        } else {
                limit = PG_GETARG_INT32(6);
        }

        /* offset */
        if (PG_ARGISNULL(7)) {
                offset = 0;
        } else {
                offset = PG_GETARG_INT32(7);
        }

        /* initialize the network environment */
        if(!est_init_net_env()){
                ereport(ERROR, (errcode(ERRCODE_QUERY_CANCELED),
                        errmsg("pgest_node: can't create network enviroment")));
        }

        /* create the node connection object */
        elog(DEBUG1, "pgest_node: est_node_new(%s) as %s", node_url, user);
        node = est_node_new(node_url);
        est_node_set_auth(node, user, passwd);

        elog(DEBUG1, "pgest_node: query[%s] attr[%s] limit %d offset %d", query, (PG_ARGISNULL(4) ? "NULL" : attr), limit, offset);
        
        /* create a search condition object */
        if (!(cond = est_cond_new())) {
                ereport(ERROR, (errcode(ERRCODE_QUERY_CANCELED),
                        errmsg("pgest_node: est_cond_new failed")));
        }
        
        /* set the search phrase to the search condition object */
        if (! PG_ARGISNULL(3) && strlen(query) > 0)
                est_cond_set_phrase(cond, query);

        /* minimum valid attribute length is 10: @a STREQ a */
        if (! PG_ARGISNULL(4) && strlen(attr) >= 10) {
                elog(DEBUG1,"attributes: %s", attr);
                char *curr_attr;
                curr_attr = strtok(attr, ATTR_DELIMITER);
                while (curr_attr) {
                        elog(DEBUG1,"est_cond_add_attr(%s)", curr_attr);
                        est_cond_add_attr(cond, curr_attr);
                        curr_attr = strtok(NULL, ATTR_DELIMITER);
                }
        }

        /* set the search phrase to the search condition object */
        if (! PG_ARGISNULL(5) && 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 */
        /* FIXME: allow user to specify depath of search */
        nres = est_node_search(node, cond, 0);

        if (! nres) {
                int status = est_node_status(node);
                est_cond_delete(cond);
                est_node_delete(node);
                est_free_net_env();
                ereport(ERROR, (errcode(ERRCODE_QUERY_CANCELED),
                        errmsg("pgest_node: search failed, node status %d", status)));
        }

        /* get number of results */
        resnum = est_noderes_doc_num(nres);

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

        /* total number of tuples to be returned */
        if (limit && limit < resnum) {
                nrows = limit;
        } else {
                nrows = resnum - offset;
        }


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


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

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

                /* get result from estraier */
                if (! ( rdoc = est_noderes_get_doc(nres, i + offset) )) {
                        elog(INFO, "can't find result %d", i + offset);
                } else {
                        elog(DEBUG1, "URI: %s\n Title: %s\n",
                                est_resdoc_attr(rdoc, "@uri"),
                                est_resdoc_attr(rdoc, "@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 && rdoc)
                                values[j] = DatumGetCString(
                                        node_attr2text(rdoc,
                                                (char *)DirectFunctionCall1(textout, dvalue)
                                        ));
                        else
                                values[j] = NULL;
                }
                /* construct the tuple */
                tuple = BuildTupleFromCStrings(attinmeta, values);

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

        }

        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);

        /* delete the node result object */
        est_noderes_delete(nres);

        /* destroy the search condition object */                          
        est_cond_delete(cond);

        /* destroy the node object */
        est_node_delete(node);

        /* free the networking environment */
        est_free_net_env();

        return (Datum) 0;
}

/* make text var from node attr */
char *node_attr2text(ESTRESDOC *rdoc, char *attr) {
        char *val;
        const char *attrval;
        int len;
        int attrlen;

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

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

        len = strlen(attrval);
        elog(DEBUG1, "node_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	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			#include "miscadmin.h"
29			#include <estraier.h>
30			#include <cabin.h>
31	dpavlin	41	#include <estnode.h>
32	dpavlin	1
33			#define _textin(str) DirectFunctionCall1(textin, CStringGetDatum(str))
34			#define _textout(str) DatumGetPointer(DirectFunctionCall1(textout, PointerGetDatum(str)))
35			#define GET_STR(textp) DatumGetCString(DirectFunctionCall1(textout, PointerGetDatum(textp)))
36			#define GET_TEXT(cstrp) DatumGetTextP(DirectFunctionCall1(textin, CStringGetDatum(cstrp)))
37
38	dpavlin	27	/* SortMem got renamed in PostgreSQL 8.0 */
39			#ifndef SortMem
40			#define SortMem 16 * 1024
41			#endif
42
43	dpavlin	40	#define ATTR_DELIMITER "{{!}}"
44
45	dpavlin	5	/* prototype */
46			char attr2text(ESTDOC doc, char *attr);
47	dpavlin	41	char node_attr2text(ESTRESDOC rdoc, char *attr);
48	dpavlin	1
49
50			/* work in progress */
51	dpavlin	19	PG_FUNCTION_INFO_V1(pgest_attr);
52			Datum pgest_attr(PG_FUNCTION_ARGS)
53	dpavlin	1	{
54	dpavlin	31	ArrayType *attr_arr = PG_GETARG_ARRAYTYPE_P(6);
55	dpavlin	25	Oid attr_element_type = ARR_ELEMTYPE(attr_arr);
56			int attr_ndims = ARR_NDIM(attr_arr);
57			int *attr_dim_counts = ARR_DIMS(attr_arr);
58			int *attr_dim_lower_bounds = ARR_LBOUND(attr_arr);
59	dpavlin	19	int ncols = 0;
60			int nrows = 0;
61			int indx[MAXDIM];
62	dpavlin	25	int16 attr_len;
63			bool attr_byval;
64			char attr_align;
65	dpavlin	1	ReturnSetInfo rsinfo = (ReturnSetInfo ) fcinfo->resultinfo;
66			AttInMetadata *attinmeta;
67			TupleDesc tupdesc;
68	dpavlin	19	Tuplestorestate *tupstore = NULL;
69	dpavlin	1	HeapTuple tuple;
70			MemoryContext per_query_ctx;
71			MemoryContext oldcontext;
72			Datum dvalue;
73			char **values;
74	dpavlin	19	int rsinfo_ncols;
75	dpavlin	1	int i, j;
76	dpavlin	19	/* estvars */
77	dpavlin	25	ESTDB *db;
78			ESTCOND *cond;
79			ESTDOC *doc;
80			const CBLIST *texts;
81			int ecode, *est_result, resnum;
82			int limit = 0;
83			int offset = 0;
84
85	dpavlin	19	char *index_path;
86			char *query;
87			char *attr;
88	dpavlin	31	char *order;
89	dpavlin	1
90	dpavlin	19
91			/* only allow 1D input array */
92	dpavlin	25	if (attr_ndims == 1)
93	dpavlin	19	{
94	dpavlin	25	ncols = attr_dim_counts[0];
95	dpavlin	19	}
96			else
97			ereport(ERROR,
98			(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
99			errmsg("invalid input array"),
100			errdetail("Input array must have 1 dimension")));
101
102	dpavlin	1	/* check to see if caller supports us returning a tuplestore */
103			if (!rsinfo \|\| !(rsinfo->allowedModes & SFRM_Materialize))
104			ereport(ERROR,
105			(errcode(ERRCODE_SYNTAX_ERROR),
106			errmsg("materialize mode required, but it is not " \
107			"allowed in this context")));
108
109	dpavlin	19	/* get info about element type needed to construct the array */
110	dpavlin	25	get_typlenbyvalalign(attr_element_type, &attr_len, &attr_byval, &attr_align);
111	dpavlin	19
112	dpavlin	1	/* get the requested return tuple description */
113			tupdesc = rsinfo->expectedDesc;
114	dpavlin	19	rsinfo_ncols = tupdesc->natts;
115	dpavlin	1
116			/*
117			* The requested tuple description better match up with the array
118			* we were given.
119			*/
120	dpavlin	19	if (rsinfo_ncols != ncols)
121			ereport(ERROR,
122			(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
123			errmsg("invalid input array"),
124			errdetail("Number of elements in array must match number of query specified columns.")));
125
126	dpavlin	1	/* OK, use it */
127			attinmeta = TupleDescGetAttInMetadata(tupdesc);
128
129			/* Now go to work */
130			rsinfo->returnMode = SFRM_Materialize;
131
132			per_query_ctx = fcinfo->flinfo->fn_mcxt;
133			oldcontext = MemoryContextSwitchTo(per_query_ctx);
134
135			/* initialize our tuplestore */
136			tupstore = tuplestore_begin_heap(true, false, SortMem);
137
138	dpavlin	19
139			/* take rest of arguments from function */
140
141			/* index path */
142			if (PG_ARGISNULL(0)) {
143			ereport(ERROR,
144			(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
145			errmsg("index path can't be null"),
146			errdetail("Index path must be valid full path to HyperEstraier index")));
147			}
148			index_path = _textout(PG_GETARG_TEXT_P(0));
149
150			/* query string */
151	dpavlin	31	if (PG_ARGISNULL(1)) {
152	dpavlin	19	query = "";
153			} else {
154			query = _textout(PG_GETARG_TEXT_P(1));
155			}
156
157			/* atribute filter */
158			if (PG_ARGISNULL(2)) {
159			attr = "";
160			} else {
161			attr = _textout(PG_GETARG_TEXT_P(2));
162			}
163	dpavlin	31
164			/* sort order */
165			if (PG_ARGISNULL(3)) {
166			order = "";
167			} else {
168			order = _textout(PG_GETARG_TEXT_P(3));
169			}
170	dpavlin	19
171	dpavlin	31
172	dpavlin	19	/* limit */
173	dpavlin	31	if (PG_ARGISNULL(4)) {
174	dpavlin	19	limit = 0;
175			} else {
176	dpavlin	31	limit = PG_GETARG_INT32(4);
177	dpavlin	19	}
178
179			/* offset */
180	dpavlin	31	if (PG_ARGISNULL(5)) {
181	dpavlin	19	offset = 0;
182			} else {
183	dpavlin	31	offset = PG_GETARG_INT32(5);
184	dpavlin	19	}
185
186
187			/* open the database */
188			elog(DEBUG1, "pgest_attr: est_db_open(%s)", index_path);
189
190			if(!(db = est_db_open(index_path, ESTDBREADER, &ecode))){
191			ereport(ERROR, (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
192			errmsg("est_db_open: can't open %s: %d", index_path, ecode),
193			errdetail(est_err_msg(ecode))));
194			}
195
196	dpavlin	20	elog(DEBUG1, "pgest_attr: query[%s] attr[%s] limit %d offset %d", query, (PG_ARGISNULL(2) ? "NULL" : attr), limit, offset);
197	dpavlin	19
198			/* create a search condition object */
199			if (!(cond = est_cond_new())) {
200			ereport(ERROR, (errcode(ERRCODE_QUERY_CANCELED),
201			errmsg("pgest_attr: est_cond_new failed")));
202			}
203
204			/* set the search phrase to the search condition object */
205			if (! PG_ARGISNULL(1) && strlen(query) > 0)
206			est_cond_set_phrase(cond, query);
207
208			/* minimum valid attribute length is 10: @a STREQ a */
209			if (! PG_ARGISNULL(2) && strlen(attr) >= 10) {
210	dpavlin	40	elog(DEBUG1,"attributes: %s", attr);
211			char *curr_attr;
212			curr_attr = strtok(attr, ATTR_DELIMITER);
213			while (curr_attr) {
214			elog(DEBUG1,"est_cond_add_attr(%s)", curr_attr);
215			est_cond_add_attr(cond, curr_attr);
216			curr_attr = strtok(NULL, ATTR_DELIMITER);
217			}
218	dpavlin	19	}
219
220	dpavlin	31	/* set the search phrase to the search condition object */
221			if (! PG_ARGISNULL(3) && strlen(order) > 0) {
222			elog(DEBUG1,"est_cond_set_order(%s)", order);
223			est_cond_set_order(cond, order);
224			}
225
226	dpavlin	38	if (limit) {
227			elog(DEBUG1,"est_cond_set_max(%d)", limit + offset);
228			est_cond_set_max(cond, limit + offset);
229			}
230
231	dpavlin	19	/* get the result of search */
232			est_result = est_db_search(db, cond, &resnum, NULL);
233
234			/* check if results exists */
235			if ( 0 == resnum ) {
236			elog(INFO, "pgest_attr: no results for: %s", query );
237			}
238
239			/* total number of tuples to be returned */
240			if (limit && limit < resnum) {
241	dpavlin	31	nrows = limit;
242	dpavlin	19	} else {
243			nrows = resnum - offset;
244			}
245
246
247			elog(DEBUG1, "pgest_attr: found %d hits for %s", resnum, query);
248
249
250	dpavlin	1	values = (char *) palloc(ncols sizeof(char *));
251
252			for (i = 0; i < nrows; i++)
253			{
254	dpavlin	19
255			/* get result from estraier */
256			if (! ( doc = est_db_get_doc(db, est_result[i + offset], 0)) ) {
257			elog(INFO, "can't find result %d", i + offset);
258			} else {
259			elog(DEBUG1, "URI: %s\n Title: %s\n",
260			est_doc_attr(doc, "@uri"),
261			est_doc_attr(doc, "@title")
262			);
263			}
264
265			/* iterate over results */
266	dpavlin	1	for (j = 0; j < ncols; j++)
267			{
268	dpavlin	19	bool isnull;
269
270			/* array value of this position */
271	dpavlin	25	indx[0] = j + attr_dim_lower_bounds[0];
272	dpavlin	19
273	dpavlin	25	dvalue = array_ref(attr_arr, attr_ndims, indx, -1, attr_len, attr_byval, attr_align, &isnull);
274	dpavlin	19
275			if (!isnull && doc)
276			values[j] = DatumGetCString(
277			attr2text(doc,
278	dpavlin	20	(char *)DirectFunctionCall1(textout, dvalue)
279	dpavlin	19	));
280			else
281			values[j] = NULL;
282	dpavlin	1	}
283			/* construct the tuple */
284			tuple = BuildTupleFromCStrings(attinmeta, values);
285
286			/* now store it */
287			tuplestore_puttuple(tupstore, tuple);
288	dpavlin	19
289
290			/* delete estraier document object */
291			est_doc_delete(doc);
292	dpavlin	1	}
293
294			tuplestore_donestoring(tupstore);
295			rsinfo->setResult = tupstore;
296
297			/*
298			* SFRM_Materialize mode expects us to return a NULL Datum. The actual
299			* tuples are in our tuplestore and passed back through
300			* rsinfo->setResult. rsinfo->setDesc is set to the tuple description
301			* that we actually used to build our tuples with, so the caller can
302			* verify we did what it was expecting.
303			*/
304			rsinfo->setDesc = tupdesc;
305			MemoryContextSwitchTo(oldcontext);
306
307	dpavlin	31	est_cond_delete(cond);
308
309	dpavlin	19	if(!est_db_close(db, &ecode)){
310			ereport(ERROR, (errcode(ERRCODE_IO_ERROR),
311			errmsg("est_db_close: %d", ecode),
312			errdetail(est_err_msg(ecode))));
313			}
314
315	dpavlin	1	return (Datum) 0;
316			}
317
318
319			/* make text var from attr */
320			char attr2text(ESTDOC doc, char *attr) {
321			char *val;
322			const char *attrval;
323			int len;
324	dpavlin	4	int attrlen;
325	dpavlin	1
326	dpavlin	2	elog(DEBUG1, "doc: %08x, attr: %s", doc, attr);
327	dpavlin	1
328	dpavlin	4	if ( (attrval = est_doc_attr(doc, attr)) && (attrlen = strlen(attrval)) ) {
329			val = (char ) palloc(attrlen sizeof(char));
330	dpavlin	1	} else {
331			return (Datum) NULL;
332			}
333
334			len = strlen(attrval);
335	dpavlin	2	elog(DEBUG1, "attr2text(%s) = '%s' %d bytes", attr, attrval, len);
336	dpavlin	1
337			len++;
338			len *= sizeof(char);
339
340			elog(DEBUG2, "palloc(%d)", len);
341
342			val = palloc(len);
343
344			memset(val, 0, len);
345			strncpy(val, attrval, len);
346
347			elog(DEBUG2, "val=%s", val);
348
349			return val;
350			}
351
352	dpavlin	41	/*
353			* variation on theme: use node API which doesn't open index on
354			* every query which is much faster for large indexes
355			*
356			*/
357
358			PG_FUNCTION_INFO_V1(pgest_node);
359			Datum pgest_node(PG_FUNCTION_ARGS)
360			{
361			ArrayType *attr_arr = PG_GETARG_ARRAYTYPE_P(8);
362			Oid attr_element_type = ARR_ELEMTYPE(attr_arr);
363			int attr_ndims = ARR_NDIM(attr_arr);
364			int *attr_dim_counts = ARR_DIMS(attr_arr);
365			int *attr_dim_lower_bounds = ARR_LBOUND(attr_arr);
366			int ncols = 0;
367			int nrows = 0;
368			int indx[MAXDIM];
369			int16 attr_len;
370			bool attr_byval;
371			char attr_align;
372			ReturnSetInfo rsinfo = (ReturnSetInfo ) fcinfo->resultinfo;
373			AttInMetadata *attinmeta;
374			TupleDesc tupdesc;
375			Tuplestorestate *tupstore = NULL;
376			HeapTuple tuple;
377			MemoryContext per_query_ctx;
378			MemoryContext oldcontext;
379			Datum dvalue;
380			char **values;
381			int rsinfo_ncols;
382			int i, j;
383			/* estvars */
384			ESTNODE *node;
385			ESTCOND *cond;
386			ESTNODERES *nres;
387			ESTRESDOC *rdoc;
388			const CBLIST *texts;
389			int resnum = 0;
390			int limit = 0;
391			int offset = 0;
392
393			char *node_url;
394			char user, passwd;
395			char *query;
396			char *attr;
397			char *order;
398
399
400			/* only allow 1D input array */
401			if (attr_ndims == 1)
402			{
403			ncols = attr_dim_counts[0];
404			}
405			else
406			ereport(ERROR,
407			(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
408			errmsg("invalid input array"),
409			errdetail("Input array must have 1 dimension")));
410
411			/* check to see if caller supports us returning a tuplestore */
412			if (!rsinfo \|\| !(rsinfo->allowedModes & SFRM_Materialize))
413			ereport(ERROR,
414			(errcode(ERRCODE_SYNTAX_ERROR),
415			errmsg("materialize mode required, but it is not " \
416			"allowed in this context")));
417
418			/* get info about element type needed to construct the array */
419			get_typlenbyvalalign(attr_element_type, &attr_len, &attr_byval, &attr_align);
420
421			/* get the requested return tuple description */
422			tupdesc = rsinfo->expectedDesc;
423			rsinfo_ncols = tupdesc->natts;
424
425			/*
426			* The requested tuple description better match up with the array
427			* we were given.
428			*/
429			if (rsinfo_ncols != ncols)
430			ereport(ERROR,
431			(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
432			errmsg("invalid input array"),
433			errdetail("Number of elements in array must match number of query specified columns.")));
434
435			/* OK, use it */
436			attinmeta = TupleDescGetAttInMetadata(tupdesc);
437
438			/* Now go to work */
439			rsinfo->returnMode = SFRM_Materialize;
440
441			per_query_ctx = fcinfo->flinfo->fn_mcxt;
442			oldcontext = MemoryContextSwitchTo(per_query_ctx);
443
444			/* initialize our tuplestore */
445			tupstore = tuplestore_begin_heap(true, false, SortMem);
446
447
448			/* take rest of arguments from function */
449
450			/* node URL */
451			if (PG_ARGISNULL(0)) {
452			ereport(ERROR,
453			(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
454			errmsg("node URL can't be null"),
455			errdetail("Node URL must be valid URL to HyperEstraier node")));
456			}
457			node_url = _textout(PG_GETARG_TEXT_P(0));
458
459			/* login and password */
460			if (PG_ARGISNULL(1) \|\| PG_ARGISNULL(2)) {
461			ereport(ERROR,
462			(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
463			errmsg("username and password can't be NULL"),
464			errdetail("You must specify valid username and password to HyperEstraier node")));
465			}
466			user = _textout(PG_GETARG_TEXT_P(1));
467			passwd = _textout(PG_GETARG_TEXT_P(2));
468
469			/* query string */
470			if (PG_ARGISNULL(3)) {
471			query = "";
472			} else {
473			query = _textout(PG_GETARG_TEXT_P(3));
474			}
475
476			/* atribute filter */
477			if (PG_ARGISNULL(4)) {
478			attr = "";
479			} else {
480			attr = _textout(PG_GETARG_TEXT_P(4));
481			}
482
483			/* sort order */
484			if (PG_ARGISNULL(5)) {
485			order = "";
486			} else {
487			order = _textout(PG_GETARG_TEXT_P(5));
488			}
489
490
491			/* limit */
492			if (PG_ARGISNULL(6)) {
493			limit = 0;
494			} else {
495			limit = PG_GETARG_INT32(6);
496			}
497
498			/* offset */
499			if (PG_ARGISNULL(7)) {
500			offset = 0;
501			} else {
502			offset = PG_GETARG_INT32(7);
503			}
504
505			/* initialize the network environment */
506			if(!est_init_net_env()){
507			ereport(ERROR, (errcode(ERRCODE_QUERY_CANCELED),
508			errmsg("pgest_node: can't create network enviroment")));
509			}
510
511			/* create the node connection object */
512			elog(DEBUG1, "pgest_node: est_node_new(%s) as %s", node_url, user);
513			node = est_node_new(node_url);
514			est_node_set_auth(node, user, passwd);
515
516			elog(DEBUG1, "pgest_node: query[%s] attr[%s] limit %d offset %d", query, (PG_ARGISNULL(4) ? "NULL" : attr), limit, offset);
517
518			/* create a search condition object */
519			if (!(cond = est_cond_new())) {
520			ereport(ERROR, (errcode(ERRCODE_QUERY_CANCELED),
521			errmsg("pgest_node: est_cond_new failed")));
522			}
523
524			/* set the search phrase to the search condition object */
525			if (! PG_ARGISNULL(3) && strlen(query) > 0)
526			est_cond_set_phrase(cond, query);
527
528			/* minimum valid attribute length is 10: @a STREQ a */
529			if (! PG_ARGISNULL(4) && strlen(attr) >= 10) {
530			elog(DEBUG1,"attributes: %s", attr);
531			char *curr_attr;
532			curr_attr = strtok(attr, ATTR_DELIMITER);
533			while (curr_attr) {
534			elog(DEBUG1,"est_cond_add_attr(%s)", curr_attr);
535			est_cond_add_attr(cond, curr_attr);
536			curr_attr = strtok(NULL, ATTR_DELIMITER);
537			}
538			}
539
540			/* set the search phrase to the search condition object */
541			if (! PG_ARGISNULL(5) && strlen(order) > 0) {
542			elog(DEBUG1,"est_cond_set_order(%s)", order);
543			est_cond_set_order(cond, order);
544			}
545
546			if (limit) {
547			elog(DEBUG1,"est_cond_set_max(%d)", limit + offset);
548			est_cond_set_max(cond, limit + offset);
549			}
550
551			/* get the result of search */
552			/* FIXME: allow user to specify depath of search */
553			nres = est_node_search(node, cond, 0);
554
555	dpavlin	44	if (! nres) {
556			int status = est_node_status(node);
557			est_cond_delete(cond);
558			est_node_delete(node);
559			est_free_net_env();
560			ereport(ERROR, (errcode(ERRCODE_QUERY_CANCELED),
561			errmsg("pgest_node: search failed, node status %d", status)));
562			}
563
564	dpavlin	41	/* get number of results */
565			resnum = est_noderes_doc_num(nres);
566
567			/* check if results exists */
568			if ( 0 == resnum ) {
569			elog(INFO, "pgest_node: no results for: %s", query );
570			}
571
572			/* total number of tuples to be returned */
573			if (limit && limit < resnum) {
574			nrows = limit;
575			} else {
576			nrows = resnum - offset;
577			}
578
579
580			elog(DEBUG1, "pgest_node: found %d hits for %s", resnum, query);
581
582
583			values = (char *) palloc(ncols sizeof(char *));
584
585			for (i = 0; i < nrows; i++)
586			{
587
588			/* get result from estraier */
589			if (! ( rdoc = est_noderes_get_doc(nres, i + offset) )) {
590			elog(INFO, "can't find result %d", i + offset);
591			} else {
592			elog(DEBUG1, "URI: %s\n Title: %s\n",
593			est_resdoc_attr(rdoc, "@uri"),
594			est_resdoc_attr(rdoc, "@title")
595			);
596			}
597
598			/* iterate over results */
599			for (j = 0; j < ncols; j++)
600			{
601			bool isnull;
602
603			/* array value of this position */
604			indx[0] = j + attr_dim_lower_bounds[0];
605
606			dvalue = array_ref(attr_arr, attr_ndims, indx, -1, attr_len, attr_byval, attr_align, &isnull);
607
608			if (!isnull && rdoc)
609			values[j] = DatumGetCString(
610			node_attr2text(rdoc,
611			(char *)DirectFunctionCall1(textout, dvalue)
612			));
613			else
614			values[j] = NULL;
615			}
616			/* construct the tuple */
617			tuple = BuildTupleFromCStrings(attinmeta, values);
618
619			/* now store it */
620			tuplestore_puttuple(tupstore, tuple);
621
622			}
623
624			tuplestore_donestoring(tupstore);
625			rsinfo->setResult = tupstore;
626
627			/*
628			* SFRM_Materialize mode expects us to return a NULL Datum. The actual
629			* tuples are in our tuplestore and passed back through
630			* rsinfo->setResult. rsinfo->setDesc is set to the tuple description
631			* that we actually used to build our tuples with, so the caller can
632			* verify we did what it was expecting.
633			*/
634			rsinfo->setDesc = tupdesc;
635			MemoryContextSwitchTo(oldcontext);
636
637			/* delete the node result object */
638			est_noderes_delete(nres);
639
640			/* destroy the search condition object */
641			est_cond_delete(cond);
642
643			/* destroy the node object */
644			est_node_delete(node);
645
646			/* free the networking environment */
647			est_free_net_env();
648
649			return (Datum) 0;
650			}
651
652			/* make text var from node attr */
653			char node_attr2text(ESTRESDOC rdoc, char *attr) {
654			char *val;
655			const char *attrval;
656			int len;
657			int attrlen;
658
659			elog(DEBUG1, "doc: %08x, attr: %s", rdoc, attr);
660
661			if ( (attrval = est_resdoc_attr(rdoc, attr)) && (attrlen = strlen(attrval)) ) {
662			val = (char ) palloc(attrlen sizeof(char));
663			} else {
664			return (Datum) NULL;
665			}
666
667			len = strlen(attrval);
668			elog(DEBUG1, "node_attr2text(%s) = '%s' %d bytes", attr, attrval, len);
669
670			len++;
671			len *= sizeof(char);
672
673			elog(DEBUG2, "palloc(%d)", len);
674
675			val = palloc(len);
676
677			memset(val, 0, len);
678			strncpy(val, attrval, len);
679
680			elog(DEBUG2, "val=%s", val);
681
682			return val;
683			}
684