The R Project SVN – R-packages – Diff – /trunk/SQLiteDF/src/sqlite_vector.c


#include "sqlite_dataframe.h"
#include <math.h>
#include "Rmath.h"
 
/****************************************************************************
 * UTILITY FUNCTIONS
 ****************************************************************************/
char *_create_svector1(SEXP name, const char *type, int * _namelen, int protect) {
    int namelen, res;
    char *iname = _create_sdf_skeleton1(name, &namelen, protect);
 
    if (iname == NULL) return NULL;
 
    sprintf(g_sql_buf[2], "create table [%s].sdf_data ([row name] text, "
            "V1 %s, primary key ([row name]))", iname, type);
    res = _sqlite_exec(g_sql_buf[2]);
    _sqlite_error(res);
 
    if (_namelen != NULL) *_namelen = namelen;
    return iname;
}
 
SEXP _create_svector_sexp(const char *iname, const char *tblname,
        const char *varname, const char *type) {
    SEXP ret, value; int nprotected = 0;
    PROTECT(ret = NEW_LIST(3)); nprotected++;
 
    /* set list names */
    PROTECT(value = NEW_CHARACTER(3)); nprotected++;
    SET_STRING_ELT(value, 0, mkChar("iname"));
    SET_STRING_ELT(value, 1, mkChar("tblname"));
    SET_STRING_ELT(value, 2, mkChar("varname"));
    SET_NAMES(ret, value);
 
    /* set list values */
    SET_VECTOR_ELT(ret, 0, mkString(iname));
    SET_VECTOR_ELT(ret, 1, mkString(tblname));
    SET_VECTOR_ELT(ret, 2, mkString(varname));
 
    /* set sexp class */
    SET_CLASS(ret, mkString("sqlite.vector"));
 
    /* set sdf.vector.type */
    SET_SDFVECTORTYPE(ret, mkString(type));
 
    UNPROTECT(nprotected);
 
    return ret;
}
 
/* if ret == NULL, 3rd arg is the length of the vector to be created. otherwise
 * it is the index in the vector ret where we will put the result extracted
 * from ret */
int _get_vector_index_typed_result(sqlite3_stmt *stmt, SEXP *ret, int colidx,
        int idx_or_len, int *coltype) {
    int added = 1, ctype;
    if (*ret == NULL || *ret == R_NilValue) {
 
        if ((ctype = sqlite3_column_type(stmt, colidx)) == SQLITE_NULL) {
            added = 0;
        }
        
        if (ctype == SQLITE_TEXT) {
            PROTECT(*ret = NEW_CHARACTER(idx_or_len));
            if (added) 
                SET_STRING_ELT(*ret, 0, mkChar((char *)sqlite3_column_text(stmt, colidx)));
        } else if (ctype == SQLITE_FLOAT) {
            PROTECT(*ret = NEW_NUMERIC(idx_or_len));
            if (added) REAL(*ret)[0] = sqlite3_column_double(stmt, colidx);
        } else if (ctype == SQLITE_INTEGER) {
            /* sqlite does not differentiate b/w ints & bits, so we have to
             * do it ourselves since R distinguishes b/w ints & logicals */
            if (strcmp(sqlite3_column_decltype(stmt, colidx), "bit") == 0) {
                ctype = SQLITEDF_BIT; 
                PROTECT(*ret = NEW_LOGICAL(idx_or_len));
                if (added) LOGICAL(*ret)[0] = sqlite3_column_int(stmt, colidx);
            } else {
                PROTECT(*ret = NEW_INTEGER(idx_or_len));
                if (added) INTEGER(*ret)[0] = sqlite3_column_int(stmt, colidx);
            }
        } else added = 0;
 
        if (added) UNPROTECT(1);
        *coltype = ctype;
    } else if (stmt != NULL) {
        int ctype = *coltype;
 
        /* first row must have been NULL so that *coltype was not set */
        if (ctype == SQLITE_NULL) {
            ctype = *coltype = sqlite3_column_type(stmt, colidx);
            /* distinguish int or logical */
            if (ctype == SQLITE_INTEGER) 
                ctype = (strcmp(sqlite3_column_decltype(stmt, colidx), "bit") == 0) ?
                    SQLITEDF_BIT : SQLITE_INTEGER;
        }
 
        if (sqlite3_column_type(stmt, colidx) == SQLITE_NULL) {
            added = 0;
        } else if (ctype == SQLITE_TEXT) {
            SET_STRING_ELT(*ret, idx_or_len, 
                    mkChar((char *)sqlite3_column_text(stmt, colidx)));
        } else if (ctype == SQLITE_FLOAT) {
            REAL(*ret)[idx_or_len] = sqlite3_column_double(stmt, colidx);
        } else if (ctype == SQLITE_INTEGER) {
            INTEGER(*ret)[idx_or_len] = sqlite3_column_int(stmt, colidx);
        } else if (ctype == SQLITEDF_BIT) {
 
 
            LOGICAL(*ret)[idx_or_len] = sqlite3_column_int(stmt, colidx);
        } else added = 0;
    } else if (stmt == NULL) {
        /* used when there is no row returned (sqlite3_step(stmt)!= QLITE_ROW),
         * and we just insert NAs in the result */
        ctype = *coltype;
        if (ctype == SQLITE_NULL) { 
            /* since stmt == NULL, we can't check its column type */
            added = 0;
        } else if (ctype == SQLITE_TEXT) {
            SET_STRING_ELT(*ret, idx_or_len, NA_STRING);
        } else if (ctype == SQLITE_FLOAT) {
            REAL(*ret)[idx_or_len] = NA_REAL;
 
        } else if (ctype == SQLITE_INTEGER) {
            INTEGER(*ret)[idx_or_len] = NA_INTEGER;
        } else if (ctype == SQLITEDF_BIT) {
            LOGICAL(*ret)[idx_or_len] = NA_LOGICAL;
        } else added = 0;
    }
        
    return added;
}
 
/****************************************************************************
 * SVEC FUNCTIONS
 ****************************************************************************/
SEXP sdf_get_variable(SEXP sdf, SEXP name) {
    char *iname, *varname, *svec_type = NULL;
    const char *coltype;
    int type = -1, res, nprotected = 0;
    SEXP ret, value;
 
    if (!IS_CHARACTER(name)) {
        error("argument is not a string.\n");
    }
 
    iname = SDF_INAME(sdf);
    varname = CHAR_ELT(name, 0);
 
    if (!USE_SDF1(iname, TRUE, FALSE)) return R_NilValue;
 
    /* check if sdf & varname w/in that sdf exists */
    sqlite3_stmt *stmt;
    sprintf(g_sql_buf[0], "select [%s] from [%s].sdf_data", varname, iname);
 
    res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0);
 
    if (_sqlite_error(res)) return R_NilValue;
 
    coltype = sqlite3_column_decltype(stmt, 0);
    sqlite3_finalize(stmt);
    
    PROTECT(ret = NEW_LIST(3)); nprotected++;
 
    /* set list names */
    PROTECT(value = NEW_CHARACTER(3)); nprotected++;
    SET_STRING_ELT(value, 0, mkChar("iname"));
    SET_STRING_ELT(value, 1, mkChar("tblname"));
    SET_STRING_ELT(value, 2, mkChar("varname"));
    SET_NAMES(ret, value);
 
    /* set list values */
    SET_VECTOR_ELT(ret, 0, mkString(iname));
    SET_VECTOR_ELT(ret, 1, mkString("sdf_data"));
    SET_VECTOR_ELT(ret, 2, mkString(varname));
 
    /* set class */
    if (strcmp(coltype, "text") == 0) svec_type = "character";
    else if (strcmp(coltype, "double") == 0) svec_type = "numeric";
    else if (strcmp(coltype, "bit") == 0) svec_type = "logical";
    else if (strcmp(coltype, "integer") == 0 || strcmp(coltype, "int") == 0) {
        /* determine if int, factor or ordered */
        type = _get_factor_levels1(iname, varname, ret, FALSE);
        switch(type) {
            case VAR_INTEGER: svec_type = "integer"; break;
            case VAR_FACTOR: svec_type = "factor"; break;
            case VAR_ORDERED: svec_type = "ordered";
        }
 
    }
 
    SET_CLASS(ret, mkString("sqlite.vector"));
    SET_SDFVECTORTYPE(ret, mkString(svec_type));
 
    UNPROTECT(nprotected);
    return ret;
 
}
 
 
SEXP sdf_get_variable_length(SEXP svec) {
    char *iname = SDF_INAME(svec);
    if (!USE_SDF1(iname, TRUE, FALSE)) return R_NilValue;
 
    return ScalarInteger(_get_row_count2(iname, 1));
}
 
    
SEXP sdf_get_variable_index(SEXP svec, SEXP idx) {
    SEXP ret = R_NilValue, tmp;
    char *iname = SDF_INAME(svec), *tblname = SVEC_TBLNAME(svec),
         *varname = SVEC_VARNAME(svec);
    int index, idxlen, i, retlen=0, res, coltype;
    sqlite3_stmt *stmt;
 
    if (!USE_SDF1(iname, TRUE, FALSE)) return R_NilValue;
 
    idxlen = LENGTH(idx);
    if (idxlen < 1) return ret;
 
    sprintf(g_sql_buf[0], "select [%s] from [%s].[%s] where rowid=?",
            varname, iname, tblname);
    res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0);
    if (_sqlite_error(res)) error("cannot complete request");
 
    /* get data based on index */
    if (IS_NUMERIC(idx)) {
        /* special handling of 1st index, bec this is where we create the
         * SEXP to be used to hold the result */
        index = (int) REAL(idx)[0];
        if (index < 1 && idxlen == 1) return ret;
 
        if (index >= 1) {
            sqlite3_bind_int(stmt, 1, index);
            res = sqlite3_step(stmt);
            if (res == SQLITE_ROW) { 
                retlen = _get_vector_index_typed_result(stmt, &ret, 0, idxlen, &coltype);
            } 
        } 
        
        if (index < 1 || res != SQLITE_ROW) {
            /* something wrong w/ 1st idx, and it is quietly ignored. we make 
             * a "dummy" call to setup the SEXP */
            sqlite3_reset(stmt);
            sqlite3_bind_int(stmt, 1, 0);
            res = sqlite3_step(stmt);
            if (res == SQLITE_ROW) {
                /* populate with 1st row. this should be overwritten by 
                 * succeeding calls. if there are no valid result, then 
                 * everything will be removed by _shrink_vector().
                 */
                _get_vector_index_typed_result(stmt, &ret, 0, idxlen - 1, &coltype);
            }
            retlen = 0;
        }
 
 
        if (idxlen > 1) {
            for (i = 1; i < idxlen; i++) {
                index = (int) REAL(idx)[i];
                if (index < 1) continue;
                sqlite3_reset(stmt);
                sqlite3_bind_int(stmt, 1, index);
                res = sqlite3_step(stmt);
                if (res == SQLITE_ROW) {
                    retlen += _get_vector_index_typed_result(stmt, &ret, 0, retlen, &coltype); 
                } else {
                    retlen += _get_vector_index_typed_result(NULL, &ret, 0, retlen, &coltype); 
                }
            }
        }
    } else if (IS_INTEGER(idx)) {
        /* similar to REAL (IS_NUMERIC) above, except that we don't have
         * to cast idx to int. can't refactor this out, sucks. */
        index = INTEGER(idx)[0];
        if (index < 1 && idxlen == 1) return ret;
 
        if (index >= 1) {
            sqlite3_bind_int(stmt, 1, index);
            res = sqlite3_step(stmt);
            if (res == SQLITE_ROW) { 
                retlen = _get_vector_index_typed_result(stmt, &ret, 0, idxlen, &coltype);
            } 
        } 
        
        if (index < 1 || res != SQLITE_ROW) {
            /* something wrong w/ 1st idx, and it is quietly ignored. we make 
             * a "dummy" call to setup the SEXP */
            sqlite3_reset(stmt);
            sqlite3_bind_int(stmt, 1, 0);
            res = sqlite3_step(stmt);
            if (res == SQLITE_ROW) {
                _get_vector_index_typed_result(stmt, &ret, 0, idxlen - 1, &coltype);
            }
            retlen = 0;
        }
 
        if (idxlen > 1) {
            for (i = 1; i < idxlen; i++) {
                index = INTEGER(idx)[i];
                if (index < 1) continue;
                sqlite3_reset(stmt);
                sqlite3_bind_int(stmt, 1, index);
                res = sqlite3_step(stmt);
                if (res == SQLITE_ROW) {
                    retlen += _get_vector_index_typed_result(stmt, &ret, 0, retlen, &coltype); 
                } else {
                    retlen += _get_vector_index_typed_result(NULL, &ret, 0, retlen, &coltype);
                }
            }
        }
 
    } else if (IS_LOGICAL(idx)) {
        /* have to deal with recycling */
        int veclen = _get_row_count2(iname, 1);
 
        /* in this indexing, idx[i] == TRUE is equiv to vector[i]. therefore
         * idxlen <= veclen. if idxlen < veclen, then idx is recycled. */
 
        /* find if there is any TRUE element in the vector */
        for (i = 1; i < idxlen && i < veclen; i++) {
            if (LOGICAL(idx)[i]) {
                sqlite3_bind_int(stmt, 1, i+1);
                sqlite3_step(stmt);
                /* there are at least (idxlen-i) TRUE per cycle of the LOGICAL
                 * index. there are at least (veclen/idxlen) cycles (int div).
                 * at the last cycle, if (veclen%idxlen > 0), there will be
                 * at least (veclen%idxlen - i) if veclen%idxlen > i */
                retlen = (idxlen-i) * (veclen/idxlen);
                if (veclen%idxlen > i) retlen += (veclen%idxlen - i);
 
                /* create the vector */
                retlen = _get_vector_index_typed_result(stmt, &ret, 0, retlen, &coltype);
                break;
            }
        }
 
        if (i < idxlen && i < veclen) {
            for (i++; i < veclen; i++) {
                if (LOGICAL(idx)[i%idxlen]) {
                    sqlite3_reset(stmt);
                    sqlite3_bind_int(stmt, 1, i+1);
                    res = sqlite3_step(stmt);
                    if (res == SQLITE_ROW) {
                        retlen += _get_vector_index_typed_result(stmt, &ret, 0, retlen, &coltype);
                    } else {
                        retlen += _get_vector_index_typed_result(NULL, &ret, 0, retlen, &coltype);
                    }
                }
            }
        }
    }
 
    sqlite3_finalize(stmt);
 
    if (ret != R_NilValue) {
        ret = _shrink_vector(ret, retlen);
        tmp = GET_LEVELS(svec);
        if (tmp != R_NilValue) {
            SET_LEVELS(ret, duplicate(tmp));
            if (TEST_SDFVECTORTYPE(svec, "factor")) {
                SET_CLASS(ret, mkString("factor"));
            } else {
                PROTECT(tmp = NEW_CHARACTER(2));
                SET_STRING_ELT(tmp, 0, mkChar("ordered"));
                SET_STRING_ELT(tmp, 1, mkChar("factor"));
                UNPROTECT(1);
            }
        }
    }
 
    return ret;
}
 
/* sqlite.vector.[<- */
SEXP sdf_set_variable_index(SEXP svec, SEXP idx, SEXP value) {
    int idx_len, val_len, svec_len, i, res;
    char *iname = SDF_INAME(svec), *tblname = SVEC_TBLNAME(svec),
         *varname = SVEC_VARNAME(svec);
    sqlite3_stmt *stmt;
 
    /* match levels if factor or ordered */
    if (inherits(value, "ordered")) {
    } else if (inherits(value, "factor")) {
    }
 
    idx_len = LENGTH(idx);
    val_len = LENGTH(value);
    svec_len = _get_row_count2(iname, TRUE);
 
    if (idx_len % val_len != 0) 
        warning("number of items to replace is not a multiple of replacement length");
 
    sprintf(g_sql_buf[0], "update [%s].[%s] set [%s]=? where rowid=?",
            iname, tblname, varname);
    _sqlite_begin;
    res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0);
    if (_sqlite_error(res)) {
        error("cannot complete request");
        _sqlite_commit;
    }
 
    if (IS_NUMERIC(idx)) {
        for (i = 0; i < idx_len; i++) {
            /* determine type of svec, cast value to type of it */
            /* value can only be of a single type, because it is a vector! */
            /* if svec is a factor, value is a string, must check with
             * factor levels */
        }
    } else if (IS_INTEGER(idx)) {
    } else if (IS_LOGICAL(idx)) {
    }
 
    _sqlite_commit;
    return R_NilValue;
}
 
SEXP sdf_variable_summary(SEXP svec, SEXP maxsum) {
    char *iname, *tblname, *varname, *type;
    sqlite3_stmt *stmt;
    int nprotected = 0;
    SEXP ret, names;
 
    iname = SDF_INAME(svec);
    tblname = SVEC_TBLNAME(svec);
    varname = SVEC_VARNAME(svec);
    USE_SDF1(iname, TRUE, FALSE);
 
    if ((TEST_SDFVECTORTYPE(svec, "ordered") && ((type = "ordered"))) ||
            (TEST_SDFVECTORTYPE(svec, "factor") && ((type = "factor")))) {
        int nrows, i, max_rows = INTEGER(maxsum)[0];
 
 
        sprintf(g_sql_buf[0], "[%s].[%s %s]", iname, type, varname);
        nrows = _get_row_count2(g_sql_buf[0], FALSE);
        if (nrows <= max_rows) max_rows = nrows;
        else { nrows = max_rows; max_rows--; }
 
        PROTECT(ret = NEW_INTEGER(nrows)); nprotected = 1;
        PROTECT(names = NEW_CHARACTER(nrows)); nprotected++;
 
        sprintf(g_sql_buf[0], "select [%s].[%s %s].label, count(*) from "
                "[%s].[%s] join [%s].[%s %s] on [%s].[%s].[%s]=[%s].[%s %s].level "
                "group by [%s].[%s].[%s], [%s].[%s %s].level order by count(*) desc",
                iname, type, varname, iname, tblname, iname, type, varname, iname, tblname, varname,
                iname, type, varname, iname, tblname, varname, iname, type, varname);
        sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0);
 
        for (i = 0; i < max_rows; i++) {
            sqlite3_step(stmt);
            SET_STRING_ELT(names, i, mkChar((char *)sqlite3_column_text(stmt, 0)));
            INTEGER(ret)[i] = sqlite3_column_int(stmt, 1);
        }
 
        if (nrows > max_rows) {
            int others_sum = 0;
            SET_STRING_ELT(names, nrows-1, mkChar("(Others)"));
            while (sqlite3_step(stmt) == SQLITE_ROW) {
                others_sum += sqlite3_column_int(stmt, 1);
            }
            INTEGER(ret)[nrows-1] = others_sum;
        }
    } else if (TEST_SDFVECTORTYPE(svec, "logical")) {
        sprintf(g_sql_buf[0], "select count(*) from "
                "[%s].[%s] group by [%s] order by [%s]", iname, tblname, varname, varname);
 
        PROTECT(names = NEW_CHARACTER(3)); nprotected = 1;
        PROTECT(ret = NEW_CHARACTER(3)); nprotected++;
 
        SET_STRING_ELT(names, 0, mkChar("Mode"));
        SET_STRING_ELT(names, 1, mkChar("FALSE"));
        SET_STRING_ELT(names, 2, mkChar("TRUE"));
 
        sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0);
        SET_STRING_ELT(ret, 0, mkChar("logical"));
        sqlite3_step(stmt);
        SET_STRING_ELT(ret, 1, mkChar((const char *)sqlite3_column_text(stmt, 0)));
        sqlite3_step(stmt);
        SET_STRING_ELT(ret, 2, mkChar((const char *)sqlite3_column_text(stmt, 0)));
 
    } else return R_NilValue;
 
    sqlite3_finalize(stmt);
    SET_NAMES(ret, names);
    SET_CLASS(ret, mkString("table"));
    UNPROTECT(nprotected);
    return ret;
}
            
 
/* the global accumulator should be safe if we only do 1 cummulative or
 * aggregate at any time. it won't work for stuffs like "select max(col)-min(col)
 * from sdf_data". */
static long double g_accumulator = 0.0; /* accumulator var for cumsum, cumprod, etc. */
static int g_start = 0;            /* flag for start of cummulation */
static int g_narm = 0;             /* for Summary group */
 
void _init_sqlite_function_accumulator() {
    g_accumulator = 0.0;   /* initialize accumulator */
    g_start = 1;           /* flag that we are at start of accumulating */
}
 
SEXP sdf_do_variable_math(SEXP func, SEXP vector, SEXP other_args) {
    char *iname, *iname_src, *varname_src, *funcname;
    int namelen, res;
    sqlite3_stmt *stmt;
 
    /* get data from arguments (function name and sqlite.vector stuffs) */
    funcname = CHAR_ELT(func, 0);
    iname_src = SDF_INAME(vector);
    varname_src = SVEC_VARNAME(vector);
 
    if (!USE_SDF1(iname_src, TRUE, TRUE)) return R_NilValue;
 
    /* create a new sdf, with 1 column named V1 */
    iname = _create_svector1(R_NilValue, "double", &namelen, TRUE);
 
    /* insert into <newsdf>.col, row.names select func(col), rownames */
    if (strcmp(funcname, "round") == 0 || strcmp(funcname, "signif") == 0) {
        double digits = REAL(_getListElement(other_args, "digits"))[0];
        sprintf(g_sql_buf[0], "insert into [%s].sdf_data([row name], V1) "
                "select [row name], %s([%s],?) from [%s].sdf_data", iname, funcname,
                varname_src, iname_src);
        res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0); 
        if (_sqlite_error(res)) goto vecmath_prepare_error;
        res = sqlite3_bind_double(stmt, 1, digits);
    } else if (strcmp(funcname, "log") == 0) {
        double base = REAL(_getListElement(other_args, "base"))[0];
        sprintf(g_sql_buf[0], "insert into [%s].sdf_data([row name], V1) "
                "select [row name], %s([%s],?) from [%s].sdf_data", iname, funcname,
                varname_src, iname_src);
        res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0); 
        if (_sqlite_error(res)) goto vecmath_prepare_error;
        res = sqlite3_bind_double(stmt, 1, base);
    } else {
        sprintf(g_sql_buf[0], "insert into [%s].sdf_data([row name], V1) "
                "select [row name], %s([%s]) from [%s].sdf_data", iname, funcname,
                varname_src, iname_src);
        res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0); 
    }
 
    if (_sqlite_error(res)) {
vecmath_prepare_error:
        sprintf(g_sql_buf[0], "detach %s", iname);
        _sqlite_exec(g_sql_buf[0]);
 
        /* we will return a string with the file name, and do file.remove
         * at R */
        iname[namelen] = '.';
        return mkString(iname);
    }
 
    _init_sqlite_function_accumulator();
    sqlite3_step(stmt);
    sqlite3_finalize(stmt);
 
    UNUSE_SDF2(iname);
    UNUSE_SDF2(iname_src);
 
    return _create_svector_sexp(iname, "sdf_data", "V1", "numeric");
}
 
SEXP sdf_do_variable_op(SEXP func, SEXP vector, SEXP op2, SEXP arg_reversed) {
    char *iname = NULL, *iname_src, *varname_src, *funcname;
    int res, functype = -1, op2_len, svec_len, i, reversed;
    sqlite3_stmt *stmt, *stmt2;
 
    /* get data from arguments (function name and sqlite.vector stuffs) */
    funcname = CHAR_ELT(func, 0);
    iname_src = SDF_INAME(vector);
    varname_src = SVEC_VARNAME(vector);
    reversed = LOGICAL(arg_reversed)[0];
 
    if (!USE_SDF1(iname_src, TRUE, TRUE)) return R_NilValue;
    switch(funcname[0]) {
        case '+' :
        case '-' :
        case '*' :
        case '/' :
        case '^' :
        case '%' : /* %% and %/% */ 
            functype = 0; break;  /* output is REAL */
        case '&' :
        case '|' :
        case '!' :  /* ! and != */
            if (funcname[1] == 0) { functype = 1; break; }
        case '=' :  /* == */
        case '<' :  /* < and <= */
        case '>' :  /* > and >= */
            functype = 2; 
    }
 
    svec_len = _get_row_count2(iname_src, 1);
    if (functype == 0 || functype == 2 || (functype == 1 && funcname[0] != '!')) {
        char *insert_fmt_string1c, *insert_fmt_string1s;
        char *insert_fmt_string2c, *insert_fmt_string2s;
        int vec_idx, sdf_idx;
 
        iname = _create_svector1(R_NilValue, (functype == 0) ? "double" : "bit", NULL, TRUE);
 
        /* insert_fmt_string prefixes:
         * 1 - op2 is an ordinary vector, op2_len = 1, straightforward insert-select
         * 2 - op2 is an ordinary vector, op2_len > 1, may need recycling, loop both sides
         * s - operator will be printed as string
         * c - operator is either INTDIV or RMOD (int division, R modulo). initially, I
         *     thought I could cast both op to int then use / and % of sqlite, which is just
         *     the 2nd character of the funcname. however, R's INTDIV and casts to int 
         *     after division (e.g. 3.5 %/% 1.5 == 2). RMOD operates on doubles too,
         *     which is the remainder of the largest int multiple of "divisor"
         *     (e.g. 3.5 %% 1.5 = 0.5)
         */ 
        if (functype == 1) { /* boolean binary operators */
            if (!reversed) {
                insert_fmt_string1s = "insert into [%s].sdf_data "
                            "select [row name], ([%s] != 0) %s (? != 0) from [%s].sdf_data";
            } else {
                insert_fmt_string1s = "insert into [%s].sdf_data "
                            "select [row name], (? != 0) %s ([%s] != 0) from [%s].sdf_data";
            }
            insert_fmt_string2s = "insert into [%s].sdf_data values(?, (? != 0) %s (? != 0))";
            /* no need for char version of fmt_string2, since %% only occurs for functype==0 */
            insert_fmt_string1c = insert_fmt_string1s;
            insert_fmt_string2c = insert_fmt_string2s;
        } else {
            if (!reversed) {
                insert_fmt_string1s = "insert into [%s].sdf_data "
                            "select [row name], [%s] %s ? from [%s].sdf_data";
                insert_fmt_string1c = "insert into [%s].sdf_data "
                            "select [row name], %s([%s],?) from [%s].sdf_data";
            } else {
                insert_fmt_string1s = "insert into [%s].sdf_data "
                            "select [row name], ? %s [%s] from [%s].sdf_data";
                insert_fmt_string1c = "insert into [%s].sdf_data "
                            "select [row name], %s(?,[%s]) from [%s].sdf_data";
            }
            /* fmt_string2c format operator from a char, used for %% and %/% */
            insert_fmt_string2c = "insert into [%s].sdf_data values(?, %s(?,?))";
            insert_fmt_string2s = "insert into [%s].sdf_data values(?, ? %s ?)";
        }
 
        /* for 2* insert statements, the values below specifies the index of bind().
         * if !reversed, then e1 is svec, e2 is anything. otherwise, e2 is the svec */
        if (reversed) { sdf_idx = 3; vec_idx = 2; }
        else { sdf_idx = 2; vec_idx = 3; }
 
        if (IS_NUMERIC(op2)) {
            op2_len = LENGTH(op2);
 
            if (op2_len == 1) {
                if (funcname[0] == '%') {
                    sprintf(g_sql_buf[2], insert_fmt_string1c, iname, 
                            (funcname[1] == '%') ? "r_mod" : "r_intdiv", varname_src, iname_src);
                } else {
                    sprintf(g_sql_buf[2], insert_fmt_string1s, iname,
                            varname_src, funcname, iname_src);
                }
                res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
                _sqlite_error(res);
                sqlite3_bind_double(stmt, 1, REAL(op2)[0]);
                sqlite3_step(stmt);
                sqlite3_finalize(stmt);
            } else if (op2_len <= svec_len) {  /* recycle op2 */
                sprintf(g_sql_buf[2], "select [row name], [%s] from [%s].sdf_data",
                        varname_src, iname_src);
                res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt2, 0);
                _sqlite_error(res);
 
                if (funcname[0] == '%') {
                    sprintf(g_sql_buf[2], insert_fmt_string2c, iname, 
                            (funcname[1] == '%') ? "r_mod" : "r_intdiv");  
                    _sqlite_begin;
                    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
                    _sqlite_error(res);
 
                    for (i = 0; i < svec_len; i++) {
                        sqlite3_step(stmt2);
 
                        sqlite3_reset(stmt);
                        sqlite3_bind_text(stmt, 1, (char *)sqlite3_column_text(stmt2, 0), -1, SQLITE_STATIC);
                        sqlite3_bind_int(stmt, sdf_idx, (int)sqlite3_column_double(stmt2, 1));
                        sqlite3_bind_int(stmt, vec_idx, (int)REAL(op2)[i % op2_len]);
                        sqlite3_step(stmt);
                    }
                    sqlite3_finalize(stmt);
                    sqlite3_finalize(stmt2);
                    _sqlite_commit;
                } else { /* non-integer binary operation */
                    sprintf(g_sql_buf[2], insert_fmt_string2s, iname, funcname);
                    _sqlite_begin;
                    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
                    _sqlite_error(res);
 
                    for (i = 0; i < svec_len; i++) {
                        sqlite3_step(stmt2);
 
                        sqlite3_reset(stmt);
                        sqlite3_bind_text(stmt, 1, (char *)sqlite3_column_text(stmt2, 0), -1, SQLITE_STATIC);
                        sqlite3_bind_double(stmt, sdf_idx, sqlite3_column_double(stmt2, 1));
                        sqlite3_bind_double(stmt, vec_idx, REAL(op2)[i % op2_len]);
                        sqlite3_step(stmt);
                    }
                    sqlite3_finalize(stmt);
                    sqlite3_finalize(stmt2);
                    _sqlite_commit;
                }
            } else { /* op2_len > svec_len, recycle svec */
                sprintf(g_sql_buf[1], "select [%s] from [%s].sdf_data",
                        varname_src, iname_src);
 
                if (funcname[0] == '%') {
                    sprintf(g_sql_buf[2], insert_fmt_string2c, iname, 
                            (funcname[1] == '%') ? "r_mod" : "r_intdiv");
                    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
                    _sqlite_error(res);
                } else {
                    sprintf(g_sql_buf[2], insert_fmt_string2s, iname, funcname);
                    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
                    _sqlite_error(res);
                }
 
                i = 0; _sqlite_begin;
                while (i < op2_len) {
                    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt2, 0);
                    _sqlite_error(res);
 
                    if (funcname[0] == '%') {
                        for ( ; i < op2_len && sqlite3_step(stmt) == SQLITE_ROW ; i++) {
                            sqlite3_step(stmt2);
 
                            sqlite3_reset(stmt);
                            /* simplify my life, just use ints for row names so that we
                             * don't have to worry about duplicates */
                            sqlite3_bind_int(stmt, 1, i); 
                            sqlite3_bind_int(stmt, sdf_idx, sqlite3_column_int(stmt2, 1));
                            sqlite3_bind_int(stmt, vec_idx, (int)REAL(op2)[i % op2_len]);
                            sqlite3_step(stmt);
                        }
                    } else {
                        for ( ; i < op2_len && sqlite3_step(stmt) == SQLITE_ROW ; i++) {
                            sqlite3_step(stmt2);
 
                            sqlite3_reset(stmt);
                            sqlite3_bind_int(stmt, 1, i);
                            sqlite3_bind_double(stmt, sdf_idx, sqlite3_column_double(stmt2, 1));
                            sqlite3_bind_double(stmt, vec_idx, REAL(op2)[i % op2_len]);
                            sqlite3_step(stmt);
                        }
                    }
 
                    /* recycle on svec if we loop again */
                    sqlite3_finalize(stmt2);
                }
 
                sqlite3_finalize(stmt);
                _sqlite_commit;
            }
 
        } else if (IS_INTEGER(op2) || IS_LOGICAL(op2)) { /* I N T E G E R */
            /* we are taking advantage of the fact that logicals are stored as int.
             * if this becomes untrue in the future, then this is a bug */
            /* we are binding double because ___ (?) */
            op2_len = LENGTH(op2);
 
            if (op2_len == 1) {
                if (funcname[0] == '%') {
                    sprintf(g_sql_buf[2], insert_fmt_string1c, iname, 
                            varname_src, (funcname[1] == '%') ? "r_mod" : "r_intdiv", iname_src);
                } else {
                    sprintf(g_sql_buf[2], insert_fmt_string1s, iname,
                            varname_src, funcname, iname_src);
                }
                res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
                _sqlite_error(res);
                sqlite3_bind_double(stmt, 1, (double)INTEGER(op2)[0]);
                sqlite3_step(stmt);
            } else if (op2_len <= svec_len) {
                sprintf(g_sql_buf[2], "select [row name], [%s] from [%s].sdf_data",
                        varname_src, iname_src);
                res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt2, 0);
                _sqlite_error(res);
 
                if (funcname[0] == '%') {
                    sprintf(g_sql_buf[2], insert_fmt_string2c, iname, 
                            (funcname[1] == '%') ? "r_mod" : "r_intdiv");
                    _sqlite_begin;
                    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
                    _sqlite_error(res);
 
                    for (i = 0; i < svec_len; i++) {
                        sqlite3_step(stmt2);
 
                        sqlite3_reset(stmt);
                        sqlite3_bind_text(stmt, 1, (char *)sqlite3_column_text(stmt2, 0), -1, SQLITE_STATIC);
                        sqlite3_bind_int(stmt, sdf_idx, sqlite3_column_int(stmt2, 1));
                        sqlite3_bind_int(stmt, vec_idx, INTEGER(op2)[i % op2_len]);
                        sqlite3_step(stmt);
                    }
                    sqlite3_finalize(stmt);
                    sqlite3_finalize(stmt2);
                    _sqlite_commit;
                } else {
                    sprintf(g_sql_buf[2], insert_fmt_string2s, iname, funcname);
                    _sqlite_begin;
                    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
                    _sqlite_error(res);
 
                    for (i = 0; i < svec_len; i++) {
                        sqlite3_step(stmt2);
 
                        sqlite3_reset(stmt);
                        sqlite3_bind_text(stmt, 1, (char *)sqlite3_column_text(stmt2, 0), -1, SQLITE_STATIC);
                        sqlite3_bind_double(stmt, sdf_idx, sqlite3_column_double(stmt2, 1));
                        sqlite3_bind_double(stmt, vec_idx, (double)INTEGER(op2)[i % op2_len]);
                        sqlite3_step(stmt);
                    }
                    sqlite3_finalize(stmt);
                    sqlite3_finalize(stmt2);
                    _sqlite_commit;
                }
            } else {
                sprintf(g_sql_buf[1], "select [%s] from [%s].sdf_data",
                        varname_src, iname_src);
 
                if (funcname[0] == '%') {
                    sprintf(g_sql_buf[2], insert_fmt_string2c, iname,
                            (funcname[1] == '%') ? "r_mod" : "r_intdiv");
                    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
                    _sqlite_error(res);
                } else {
                    sprintf(g_sql_buf[2], insert_fmt_string2s, iname, funcname);
                    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
                    _sqlite_error(res);
                }
               
                i = 0; _sqlite_begin; 
                while (i < op2_len) {
                    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt2, 0);
                    _sqlite_error(res);
 
                    if (funcname[0] == '%') {
                        for ( ; i < op2_len && sqlite3_step(stmt) == SQLITE_ROW ; i++) {
                            sqlite3_step(stmt2);
 
                            sqlite3_reset(stmt);
                            /* simplify my life, just use ints for row names so that we
                             * don't have to worry about duplicates */
                            sqlite3_bind_int(stmt, 1, i); 
                            sqlite3_bind_int(stmt, sdf_idx, sqlite3_column_int(stmt2, 1));
                            sqlite3_bind_int(stmt, vec_idx, INTEGER(op2)[i % op2_len]);
                            sqlite3_step(stmt);
                        }
                    } else {
                        for ( ; i < op2_len && sqlite3_step(stmt) == SQLITE_ROW ; i++) {
                            sqlite3_step(stmt2);
 
                            sqlite3_reset(stmt);
                            sqlite3_bind_int(stmt, 1, i);
                            sqlite3_bind_double(stmt, sdf_idx, sqlite3_column_double(stmt2, 1));
                            sqlite3_bind_double(stmt, vec_idx, (double)INTEGER(op2)[i % op2_len]);
                            sqlite3_step(stmt);
                        }
                    }
 
                    /* recycle on svec if we loop again */
                    sqlite3_finalize(stmt2);
                }
                sqlite3_finalize(stmt);
                _sqlite_commit;
            }
 
        } else if (IS_CHARACTER(op2)) {
            if (functype != 2) error("not supported");
 
            op2_len = LENGTH(op2);
 
            if (op2_len == 1) {
                sprintf(g_sql_buf[2], insert_fmt_string1s, iname,
                            varname_src, funcname, iname_src);
                res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
                _sqlite_error(res);
                sqlite3_bind_text(stmt, 1, CHAR_ELT(op2, 0), -1, SQLITE_STATIC);
                sqlite3_step(stmt);
            } else if (op2_len <= svec_len) {
                sprintf(g_sql_buf[2], "select [row name], [%s] from [%s].sdf_data",
                        varname_src, iname_src);
                res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt2, 0);
                _sqlite_error(res);
 
                sprintf(g_sql_buf[2], insert_fmt_string2s, iname, funcname);
                _sqlite_begin;
                res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
                _sqlite_error(res);
 
                for (i = 0; i < svec_len; i++) {
                    sqlite3_step(stmt2);
 
                    sqlite3_reset(stmt);
                    sqlite3_bind_text(stmt, 1, (char *)sqlite3_column_text(stmt2, 0), -1, SQLITE_STATIC);
                    sqlite3_bind_text(stmt, sdf_idx, (char *)sqlite3_column_text(stmt2, 1), -1, SQLITE_STATIC);
                    sqlite3_bind_text(stmt, vec_idx, CHAR_ELT(op2, i % op2_len), -1 , SQLITE_STATIC);
                    sqlite3_step(stmt);
                }
                sqlite3_finalize(stmt2);
                _sqlite_commit;
            } else {
                sprintf(g_sql_buf[1], "select [%s] from [%s].sdf_data",
                        varname_src, iname_src);
 
                sprintf(g_sql_buf[2], insert_fmt_string2s, iname, funcname);
                res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
                _sqlite_error(res);
               
                i = 0; _sqlite_begin; 
                while (i < op2_len) {
                    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt2, 0);
                    _sqlite_error(res);
 
                    for ( ; i < op2_len && sqlite3_step(stmt) == SQLITE_ROW ; i++) {
                        sqlite3_step(stmt2);
 
                        sqlite3_reset(stmt);
                        sqlite3_bind_int(stmt, 1, i);
                        sqlite3_bind_text(stmt, sdf_idx, (char *)sqlite3_column_text(stmt2, 1), -1, SQLITE_STATIC);
                        sqlite3_bind_text(stmt, vec_idx, CHAR_ELT(op2, i % op2_len), -1 , SQLITE_STATIC);
                        sqlite3_step(stmt);
                    }
 
                    /* recycle on svec if we loop again */
                    sqlite3_finalize(stmt2);
                }
                sqlite3_finalize(stmt);
                _sqlite_commit;
            }
        
        } else if (inherits(op2, "sqlite.vector")) { 
            /* op2 is surely not a factor, as handled by the R wrapper */
            /* even though it is impossible for reversed to be FALSE, still use
             * sdf_idx and vec_idx so that code would be less confusing */
            char *iname_op2, *varname_op2;
            sqlite3_stmt *stmt3;
            iname_op2 = SDF_INAME(op2);
            varname_op2 = SVEC_VARNAME(op2);
 
            if (!USE_SDF1(iname_op2, TRUE, TRUE)) {
                /* delete created sqlite.vector */
                warning("detaching created result SDF %s\n", iname);
                sdf_detach_sdf(mkString(iname));
                return R_NilValue;
            }
            _sqlite_begin;
 
            op2_len = _get_row_count2(iname_op2, 1);
            
            sprintf(g_sql_buf[2], "select [%s] from [%s].sdf_data", varname_src, iname_src);
            res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt2, 0);
            _sqlite_error(res);
 
            sprintf(g_sql_buf[2], "select [%s] from [%s].sdf_data", varname_op2, iname_op2);
            res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt3, 0);
            _sqlite_error(res);
 
            if (funcname[0] == '%') {
                sprintf(g_sql_buf[2], insert_fmt_string2c, iname,
                            (funcname[1] == '%') ? "r_mod" : "r_intdiv");
                res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
                _sqlite_error(res);
 
                if (svec_len == op2_len) {
                    for (i = 0; i < svec_len; i++) {
                        sqlite3_step(stmt2); sqlite3_step(stmt3);
 
                        sqlite3_reset(stmt);
                        sqlite3_bind_int(stmt, 1, i);
                        sqlite3_bind_int(stmt, sdf_idx, sqlite3_column_int(stmt2, 0));
                        sqlite3_bind_int(stmt, vec_idx, sqlite3_column_int(stmt3, 0));
                        sqlite3_step(stmt);
                    }
                } else if (svec_len < op2_len) {
                    i = 0;
                    while (i < op2_len) {
                        for ( ; sqlite3_step(stmt2) == SQLITE_ROW && i < op2_len; i++) {
                            sqlite3_step(stmt3);
 
                            sqlite3_reset(stmt);
                            sqlite3_bind_int(stmt, 1, i);
                            sqlite3_bind_int(stmt, sdf_idx, sqlite3_column_int(stmt2, 0));
                            sqlite3_bind_int(stmt, vec_idx, sqlite3_column_int(stmt3, 0));
                            sqlite3_step(stmt);
                        }
                        sqlite3_reset(stmt2);
                    }
                } else {
                    i = 0;
                    while (i < svec_len) {
                        for ( ; sqlite3_step(stmt3) == SQLITE_ROW && i < svec_len; i++) {
                            sqlite3_step(stmt2);
 
                            sqlite3_reset(stmt);
                            sqlite3_bind_int(stmt, 1, i);
                            sqlite3_bind_int(stmt, sdf_idx, sqlite3_column_int(stmt2, 0));
                            sqlite3_bind_int(stmt, vec_idx, sqlite3_column_int(stmt3, 0));
                            sqlite3_step(stmt);
                        }
                        sqlite3_reset(stmt3);
                    }
                }
                _sqlite_commit;
            } else { /* not an integer op %% or %/% */
                sprintf(g_sql_buf[2], insert_fmt_string2s, iname, funcname);
                res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
                _sqlite_error(res);
 
                if (svec_len == op2_len) {
                    for (i = 0; i < svec_len; i++) {
                        sqlite3_step(stmt2); sqlite3_step(stmt3);
 
                        sqlite3_reset(stmt);
                        sqlite3_bind_int(stmt, 1, i);
                        sqlite3_bind_double(stmt, sdf_idx, sqlite3_column_double(stmt2, 0));
                        sqlite3_bind_double(stmt, vec_idx, sqlite3_column_double(stmt3, 0));
                        sqlite3_step(stmt);
                    }
                } else if (svec_len < op2_len) { /* recycle svec */
                    i = 0;
                    while (i < op2_len) {
                        for ( ; sqlite3_step(stmt2) == SQLITE_ROW && i < op2_len; i++) {
                            sqlite3_step(stmt3);
 
                            sqlite3_reset(stmt);
                            sqlite3_bind_int(stmt, 1, i);
                            sqlite3_bind_double(stmt, sdf_idx, sqlite3_column_double(stmt2, 0));
                            sqlite3_bind_double(stmt, vec_idx, sqlite3_column_double(stmt3, 0));
                            sqlite3_step(stmt);
                        }
                        sqlite3_reset(stmt2);
                    }
                } else { /* svec_len > op2_len, recycle op2 */
                    i = 0;
                    while (i < svec_len) {
                        for ( ; sqlite3_step(stmt3) == SQLITE_ROW && i < svec_len; i++) {
                            sqlite3_step(stmt2);
 
                            sqlite3_reset(stmt);
                            sqlite3_bind_int(stmt, 1, i);
                            sqlite3_bind_double(stmt, sdf_idx, sqlite3_column_double(stmt2, 0));
                            sqlite3_bind_double(stmt, vec_idx, sqlite3_column_double(stmt3, 0));
                            sqlite3_step(stmt);
                        }
                        sqlite3_reset(stmt3);
                    }
                }
            }
 
            sqlite3_finalize(stmt);
            sqlite3_finalize(stmt2);
            sqlite3_finalize(stmt3);
            _sqlite_commit;
 
            UNUSE_SDF2(iname_op2);
        }
    } else if (functype == 1 && funcname[0] != '!') { /* unary not operator */
        iname = _create_svector1(R_NilValue, "bit", NULL, TRUE);
        sprintf(g_sql_buf[2], "insert into [%s].sdf_data "
                    "select [row name], [%s] == 0 from [%s].sdf_data", 
                    iname, varname_src, iname_src);
        res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
        _sqlite_error(res);
        sqlite3_step(stmt);
        sqlite3_finalize(stmt);
    }
 
    if (iname != NULL) {
        return _create_svector_sexp(iname, "sdf_data", "V1", 
                (functype == 0) ? "numeric" : "logical");
        UNUSE_SDF2(iname);
    }
 
    UNUSE_SDF2(iname_src);
 
    return R_NilValue;
 
}
 
SEXP sdf_do_variable_summary(SEXP func, SEXP vector, SEXP na_rm) {
    char *iname_src, *varname_src, *funcname;
    int res;
    sqlite3_stmt *stmt;
    double _ret = NA_REAL, _ret2; SEXP ret;
 
    /* get data from arguments (function name and sqlite.vector stuffs) */
    funcname = CHAR_ELT(func, 0);
    iname_src = SDF_INAME(vector);
    varname_src = SVEC_VARNAME(vector);
 
    if (!USE_SDF1(iname_src, TRUE, FALSE)) return R_NilValue;
 
    g_narm = LOGICAL(na_rm)[0];
    if (strcmp(funcname, "range") == 0) {
        /* special handling for range. use min then max */
        _init_sqlite_function_accumulator();
        sprintf(g_sql_buf[0], "select min_df([%s]) from [%s].sdf_data", varname_src, iname_src);
        res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, NULL);
        if (_sqlite_error(res)) return R_NilValue;
        sqlite3_step(stmt); 
        _ret = sqlite3_column_double(stmt, 0);
        sqlite3_finalize(stmt);
 
        if (R_IsNA(_ret) && !g_narm) {
            PROTECT(ret = NEW_NUMERIC(2));
            REAL(ret)[0] = REAL(ret)[1] = R_NaReal;
            goto __sdf_do_variable_summary_out;
        }
        
        _init_sqlite_function_accumulator();
        sprintf(g_sql_buf[0], "select max_df([%s]) from [%s].sdf_data", varname_src, iname_src);
        res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, NULL);
        if (_sqlite_error(res)) return R_NilValue;
        sqlite3_step(stmt); 
        _ret2 = sqlite3_column_double(stmt, 0);
        sqlite3_finalize(stmt);
 
        /* if there is NA, then the if above should have caught it already */
        PROTECT(ret = NEW_NUMERIC(2));
        REAL(ret)[0] = _ret;
        REAL(ret)[1] = _ret2;
    } else {
        _init_sqlite_function_accumulator();
        sprintf(g_sql_buf[0], "select %s_df([%s]) from [%s].sdf_data", funcname, varname_src, iname_src);
        res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, NULL);
        if (_sqlite_error(res)) return R_NilValue;
        res = sqlite3_step(stmt); 
        _ret = sqlite3_column_double(stmt, 0);
        sqlite3_finalize(stmt);
 
        if (strcmp(funcname, "all") == 0 || strcmp(funcname, "any") == 0) {
            PROTECT(ret = NEW_LOGICAL(1));
            if (R_IsNA(_ret)) LOGICAL(ret)[0] = NA_INTEGER;
            else LOGICAL(ret)[0] = (_ret != 0);
        } else {
            PROTECT(ret = NEW_NUMERIC(1));
            REAL(ret)[0] = _ret;
        }
    }
 
__sdf_do_variable_summary_out:
    UNPROTECT(1);
    return ret;
}
 
 
SEXP sdf_sort_variable(SEXP svec, SEXP decreasing) {
    char *iname, *tblname, *varname, *type, *sort_type;
    sqlite3_stmt *stmt;
    int res;
 
    iname = SDF_INAME(svec);
    tblname = SVEC_TBLNAME(svec);
    varname = SVEC_VARNAME(svec);
 
    if (!USE_SDF1(iname, TRUE, TRUE)) return R_NilValue;
 
    /* determine type of svec */
    sprintf(g_sql_buf[0], "select [%s] from [%s].[%s] limit 1", varname, iname, tblname);
    res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0);
    _sqlite_error(res);
    sqlite3_step(stmt);
    strcpy(g_sql_buf[0], sqlite3_column_decltype(stmt, 0));
    sqlite3_finalize(stmt);
 
    if (TEST_SDFVECTORTYPE(svec, "factor")) { /* copy factor table to iname */
        if (TEST_SDFVECTORTYPE(svec, "ordered")) type = "ordered";
        else type = "factor";
        _copy_factor_levels2(type, iname, varname, iname, "V1");
    } else type = CHAR_ELT(GET_SDFVECTORTYPE(svec), 0);
 
    if (strcmp(tblname, "sdf_data") == 0) {
        /* cache sorted data if column to be sorted comes from main sdf_data */
 
        /* test if there is already a sort_varname table */
        sort_type = (LOGICAL(decreasing)[0]) ? "desc" : "asc";
        sprintf(g_sql_buf[1], "select count(*) from [%s].sqlite_master "
                "where type='table' and name='sort_%s_%s'", iname, sort_type, varname);
        sqlite3_prepare(g_workspace, g_sql_buf[1], -1, &stmt, NULL);
        sqlite3_step(stmt);
        res = sqlite3_column_int(stmt, 0);
        sqlite3_finalize(stmt);
 
        if (res == 0) {
            sprintf(g_sql_buf[1], "create table [%s].[sort_%s_%s] ([%s] %s)", 
                    iname, sort_type, varname, varname, g_sql_buf[0]);
            if (_sqlite_error(_sqlite_exec(g_sql_buf[1]))) 
                    error("Can't create table: %s", g_sql_buf[1]);
 
            /* insert to new sdf ordered */
            sprintf(g_sql_buf[0], "insert into [%s].[sort_%s_%s] "
                    "select [%s] from [%s].[%s] order by [%s] %s", 
                    iname, sort_type, varname, varname, iname, tblname, varname, sort_type);
                    
            res = _sqlite_exec(g_sql_buf[0]);
            if (_sqlite_error(res)) error("Can't insert: %s", g_sql_buf[0]);
        }
 
        UNUSE_SDF2(iname);
        sprintf(g_sql_buf[0], "sort_%s_%s", sort_type, varname);
        return _create_svector_sexp(iname, g_sql_buf[0], varname, type);
    } else {
        /* create a new sdf table, copy data to that */
 
        /* create a sorted column */
 
        error("Not yet supported.");
    }
 
    return R_NilValue;
}
 
/****************************************************************************
 * VECTOR MATH/OPS/GROUP OPERATIONS
 ****************************************************************************/
struct accumulator_t {
    /* long double acts weird here, but if it's not long double,
     * it f*cks up sum() */
    long double accumulator;
    int started;
};
 
R_INLINE int __vecmath_checkarg(sqlite3_context *ctx, sqlite3_value *arg, double *value) {
    int ret = 1;
    if (sqlite3_value_type(arg) == SQLITE_NULL) { 
        sqlite3_result_null(ctx); 
        ret = 0;
    } else {
        if (sqlite3_value_type(arg) == SQLITE_INTEGER) 
            *value = sqlite3_value_int(arg); 
        else *value = sqlite3_value_double(arg); 
    }
    return ret;
}
 
#define SQLITE_MATH_FUNC1(name, func) static void __vecmath_ ## name(\
        sqlite3_context *ctx, int argc, sqlite3_value **argv) { \
    double value; \
    if (__vecmath_checkarg(ctx, argv[0], &value)) { \
        sqlite3_result_double(ctx, func(value)); \
    }  \
}
 
#define SQLITE_MATH_FUNC2(name, func) static void __vecmath_ ## name(\
        sqlite3_context *ctx, int argc, sqlite3_value **argv) { \
    double value1, value2; \
    if (__vecmath_checkarg(ctx, argv[0], &value1) && \
        __vecmath_checkarg(ctx, argv[1], &value2)) { \
        sqlite3_result_double(ctx, func((long double)value1, (long double)value2)); \
    }  \
}
 
#define SQLITE_MATH_FUNC_CUM(name, func) static void __vecmath_ ## name(\
        sqlite3_context *ctx, int argc, sqlite3_value **argv) { \
    double value; \
    if (__vecmath_checkarg(ctx, argv[0], &value)) { \
        if (g_start) { g_start = 0; g_accumulator = value; } \
        else g_accumulator = func(g_accumulator, value); \
        sqlite3_result_double(ctx, g_accumulator); \
    }  \
}
 
#define LOGBASE(a, b) log(a)/log(b)
 
/* SQLITE_MATH_FUNC1(abs, abs)   in SQLite */
SQLITE_MATH_FUNC1(sign, sign)   /* in R */
SQLITE_MATH_FUNC1(sqrt, sqrt)
SQLITE_MATH_FUNC1(floor, floor)
SQLITE_MATH_FUNC1(ceiling, ceil)
SQLITE_MATH_FUNC1(trunc, ftrunc) /* in R */
/*SQLITE_MATH_FUNC2(round, fprec)  2 arg, in SQLite, but override with R's version */
SQLITE_MATH_FUNC2(signif, fround) /* 2 arg, in R */
SQLITE_MATH_FUNC1(exp, exp)
SQLITE_MATH_FUNC2(log, LOGBASE) /* 2 arg */
SQLITE_MATH_FUNC1(cos, cos)
SQLITE_MATH_FUNC1(sin, sin)
SQLITE_MATH_FUNC1(tan, tan)
SQLITE_MATH_FUNC1(acos, acos)
SQLITE_MATH_FUNC1(asin, asin)
SQLITE_MATH_FUNC1(atan, atan)
SQLITE_MATH_FUNC1(cosh, cosh)
SQLITE_MATH_FUNC1(sinh, sinh)
SQLITE_MATH_FUNC1(tanh, tanh)
SQLITE_MATH_FUNC1(acosh, acosh)  /* nowhere in include?? */
SQLITE_MATH_FUNC1(asinh, asinh)  /* nowhere in include?? */
SQLITE_MATH_FUNC1(atanh, atanh)  /* nowhere in include?? */
SQLITE_MATH_FUNC1(lgamma, lgammafn) /* in R */
SQLITE_MATH_FUNC1(gamma, gammafn) /* in R */
/* SQLITE_MATH_FUNC1(gammaCody, gammaCody)   * in R ?? */
SQLITE_MATH_FUNC1(digamma, digamma) /* in R */    
SQLITE_MATH_FUNC1(trigamma, trigamma) /* in R */
 
#define SUM(a, b)  ((long double)(a) + (b))
#define PROD(a, b) (a) * (b)
#define MIN(a, b) ((a) <= (b)) ? (a) : (b)
#define MAX(a, b) ((a) >= (b)) ? (a) : (b)
#define ALL(a, b) (((a) == 0) || ((b) == 0)) ? 0 : 1
#define ANY(a, b) (((a) == 0) && ((b) == 0)) ? 0 : 1
 
SQLITE_MATH_FUNC_CUM(cumsum, SUM)
SQLITE_MATH_FUNC_CUM(cumprod, PROD)
SQLITE_MATH_FUNC_CUM(cummin, MIN)
SQLITE_MATH_FUNC_CUM(cummax, MAX)
 
#define SQLITE_SUMMARY_FUNC(name, func) static void __vecsummary_ ## name(\
        sqlite3_context *ctx, int argc, sqlite3_value **argv) { \
    double value; \
    struct accumulator_t *acc; \
    acc = sqlite3_aggregate_context(ctx, sizeof(struct accumulator_t)); \
    if (!g_narm && R_IsNA(acc->accumulator)) return; /* NA if na.rm=F & NA found */ \
    if (sqlite3_value_type(argv[0]) != SQLITE_NULL) {  \
        if (sqlite3_value_type(argv[0]) == SQLITE_INTEGER) {  \
            int tmp = sqlite3_value_int(argv[0]); \
            value = (tmp == NA_INTEGER) ? R_NaReal : tmp; \
        } else value = sqlite3_value_double(argv[0]);  \
        if (R_IsNA(value)) { \
            if (!g_narm) acc->accumulator = value; return; /* else ignore */ \
        } else if (!acc->started) { \
            acc->started = 1; \
            acc->accumulator = value; \
        } else acc->accumulator = func(acc->accumulator, value); \
    } \
}
 
SQLITE_SUMMARY_FUNC(all_df, ALL)
SQLITE_SUMMARY_FUNC(any_df, ANY)
SQLITE_SUMMARY_FUNC(sum_df, SUM)
SQLITE_SUMMARY_FUNC(prod_df, PROD)
SQLITE_SUMMARY_FUNC(min_df, MIN)
SQLITE_SUMMARY_FUNC(max_df, MAX)
 
static void __vecsummary_finalize(sqlite3_context *ctx) {
    /* g_accumulator already summarizes it. just return that */
    struct accumulator_t *acc = (struct accumulator_t *)
                sqlite3_aggregate_context(ctx, sizeof(struct accumulator_t)); 
    sqlite3_result_double(ctx, acc->accumulator);
}
 
static void __r_modulo(sqlite3_context *ctx, int argc, sqlite3_value **argv) {
    if (sqlite3_value_type(argv[0]) == SQLITE_NULL ||
        sqlite3_value_type(argv[1]) == SQLITE_NULL) { 
        sqlite3_result_null(ctx); 
    } else {
        double v1, v2, q, tmp;
        if (sqlite3_value_type(argv[0]) == SQLITE_INTEGER && 
            sqlite3_value_type(argv[1]) == SQLITE_INTEGER) {
            int i1, i2;
            i1 = sqlite3_value_int(argv[0]);
            i2 = sqlite3_value_int(argv[1]);
            if (i1 > 0 && i2 > 0) { 
                sqlite3_result_int(ctx, i1 % i2); return;
            }
            v1 = i1; v2 = i2;
        } else {
            v1 = sqlite3_value_double(argv[0]);
            v2 = sqlite3_value_double(argv[1]);
        }
        /* copied from myfmod() in src/main/arithmetic.c */
        q = v1 / v2; 
        if (v2 == 0) sqlite3_result_double(ctx, R_NaN);
        tmp = v1 - floor(q) * v2;
        /* checking omitted */
        q = floor(tmp/v2);
        sqlite3_result_double(ctx, tmp - q*v2);
 
    }
}
 
static void __r_intdiv(sqlite3_context *ctx, int argc, sqlite3_value **argv) {
    if (sqlite3_value_type(argv[0]) == SQLITE_NULL ||
        sqlite3_value_type(argv[1]) == SQLITE_NULL) { 
        sqlite3_result_null(ctx); 
    } else {
        double v1, v2;
        if (sqlite3_value_type(argv[0]) == SQLITE_INTEGER && 
            sqlite3_value_type(argv[1]) == SQLITE_INTEGER) {
            int i1, i2;
            i1 = sqlite3_value_int(argv[0]);
            i2 = sqlite3_value_int(argv[1]);
            if (i1 == NA_INTEGER || i2 == NA_INTEGER) {
                sqlite3_result_int(ctx, NA_INTEGER); return;
            } else if (i2 == 0) {
                sqlite3_result_int(ctx, 0); return;
            }
            v1 = i1; v2 = i2;
        } else {
            v1 = sqlite3_value_double(argv[0]);
            v2 = sqlite3_value_double(argv[1]);
        }
        /* copied from IDIVOP cases in src/main/arithmetic.c */
        sqlite3_result_double(ctx, floor(v1/v2));
    }
}
#define VMENTRY1(func)  {#func, __vecmath_ ## func}
#define VSENTRY1(func)  {#func, __vecsummary_ ## func}
void __register_vector_math() {
    int i, res;
    static const struct {
        char *name;
        void (*func)(sqlite3_context*, int, sqlite3_value**);
    } arr_func1[] = {
        VMENTRY1(sign),
        VMENTRY1(sqrt),
        VMENTRY1(floor),
        VMENTRY1(ceiling),
        VMENTRY1(trunc),
        VMENTRY1(exp),
        VMENTRY1(cos),
        VMENTRY1(sin),
        VMENTRY1(tan),
        VMENTRY1(acos),
        VMENTRY1(asin),
        VMENTRY1(atan),
        VMENTRY1(cosh),
        VMENTRY1(sinh),
        VMENTRY1(tanh),
        VMENTRY1(acosh),
        VMENTRY1(asinh),
        VMENTRY1(atanh),
        VMENTRY1(lgamma),
        VMENTRY1(gamma),
        VMENTRY1(digamma),
        VMENTRY1(trigamma),
        VMENTRY1(cumsum),
        VMENTRY1(cumprod),
        VMENTRY1(cummin),
        VMENTRY1(cummax)
    }, arr_func2[] =  {
        {"r_mod", __r_modulo},
        {"r_intdiv", __r_intdiv},
        /*VMENTRY1(round),*/
        VMENTRY1(signif),
        VMENTRY1(log)
    }, arr_sum1[] = {
        VSENTRY1(all_df),  /* can't override sum, min, max */
        VSENTRY1(any_df),
        VSENTRY1(sum_df),
        VSENTRY1(prod_df),
        VSENTRY1(min_df),
        VSENTRY1(max_df)
    };
 
    int len = sizeof(arr_func1) / sizeof(arr_func1[0]);
 
    for (i = 0; i < len; i++) {
        res = sqlite3_create_function(g_workspace, arr_func1[i].name, 1, 
                SQLITE_ANY, NULL, arr_func1[i].func, NULL, NULL);
        _sqlite_error(res);
    }
 
    len = sizeof(arr_func2) / sizeof(arr_func2[0]);
    for (i = 0; i < len; i++) {
        res = sqlite3_create_function(g_workspace, arr_func2[i].name, 2, 
                SQLITE_ANY, NULL, arr_func2[i].func, NULL, NULL);
        _sqlite_error(res);
    }
 
    len = sizeof(arr_sum1) / sizeof(arr_sum1[0]);
    for (i = 0; i < len; i++) {
        res = sqlite3_create_function(g_workspace, arr_sum1[i].name, 1, 
                SQLITE_ANY, NULL, NULL, arr_sum1[i].func, __vecsummary_finalize);
        _sqlite_error(res);
    }
}

Rev 3855	Rev 4160
1	`#include "sqlite_dataframe.h"`	1	`#include "sqlite_dataframe.h"`
2	`#include <math.h>`	2	`#include <math.h>`
3	`#include "Rmath.h"`	3	`#include "Rmath.h"`
4		4
5	`/****************************************************************************`	5	`/****************************************************************************`
6	`* UTILITY FUNCTIONS`	6	`* UTILITY FUNCTIONS`
7	`****************************************************************************/`	7	`****************************************************************************/`
8	`char _create_svector1(SEXP name, const char type, int * _namelen, int protect) {`	8	`char _create_svector1(SEXP name, const char type, int * _namelen, int protect) {`
9	`int namelen, res;`	9	`int namelen, res;`
10	`char *iname = _create_sdf_skeleton1(name, &namelen, protect);`	10	`char *iname = _create_sdf_skeleton1(name, &namelen, protect);`
11		11
12	`if (iname == NULL) return NULL;`	12	`if (iname == NULL) return NULL;`
13		13
14	`sprintf(g_sql_buf[2], "create table [%s].sdf_data ([row name] text, "`	14	`sprintf(g_sql_buf[2], "create table [%s].sdf_data ([row name] text, "`
15	`"V1 %s, primary key ([row name]))", iname, type);`	15	`"V1 %s, primary key ([row name]))", iname, type);`
16	`res = _sqlite_exec(g_sql_buf[2]);`	16	`res = _sqlite_exec(g_sql_buf[2]);`
17	`_sqlite_error(res);`	17	`_sqlite_error(res);`
18		18
19	`if (_namelen != NULL) *_namelen = namelen;`	19	`if (_namelen != NULL) *_namelen = namelen;`
20	`return iname;`	20	`return iname;`
21	`}`	21	`}`
22		22
23	`SEXP _create_svector_sexp(const char iname, const char tblname,`	23	`SEXP _create_svector_sexp(const char iname, const char tblname,`
24	`const char varname, const char type) {`	24	`const char varname, const char type) {`
25	`SEXP ret, value; int nprotected = 0;`	25	`SEXP ret, value; int nprotected = 0;`
26	`PROTECT(ret = NEW_LIST(3)); nprotected++;`	26	`PROTECT(ret = NEW_LIST(3)); nprotected++;`
27		27
28	`/* set list names */`	28	`/* set list names */`
29	`PROTECT(value = NEW_CHARACTER(3)); nprotected++;`	29	`PROTECT(value = NEW_CHARACTER(3)); nprotected++;`
30	`SET_STRING_ELT(value, 0, mkChar("iname"));`	30	`SET_STRING_ELT(value, 0, mkChar("iname"));`
31	`SET_STRING_ELT(value, 1, mkChar("tblname"));`	31	`SET_STRING_ELT(value, 1, mkChar("tblname"));`
32	`SET_STRING_ELT(value, 2, mkChar("varname"));`	32	`SET_STRING_ELT(value, 2, mkChar("varname"));`
33	`SET_NAMES(ret, value);`	33	`SET_NAMES(ret, value);`
34		34
35	`/* set list values */`	35	`/* set list values */`
36	`SET_VECTOR_ELT(ret, 0, mkString(iname));`	36	`SET_VECTOR_ELT(ret, 0, mkString(iname));`
37	`SET_VECTOR_ELT(ret, 1, mkString(tblname));`	37	`SET_VECTOR_ELT(ret, 1, mkString(tblname));`
38	`SET_VECTOR_ELT(ret, 2, mkString(varname));`	38	`SET_VECTOR_ELT(ret, 2, mkString(varname));`
39		39
40	`/* set sexp class */`	40	`/* set sexp class */`
41	`SET_CLASS(ret, mkString("sqlite.vector"));`	41	`SET_CLASS(ret, mkString("sqlite.vector"));`
42		42
43	`/* set sdf.vector.type */`	43	`/* set sdf.vector.type */`
44	`SET_SDFVECTORTYPE(ret, mkString(type));`	44	`SET_SDFVECTORTYPE(ret, mkString(type));`
45		45
46	`UNPROTECT(nprotected);`	46	`UNPROTECT(nprotected);`
47		47
48	`return ret;`	48	`return ret;`
49	`}`	49	`}`
50		50
51	`/* if ret == NULL, 3rd arg is the length of the vector created. otherwise`	51	`/* if ret == NULL, 3rd arg is the length of the vector to be created. otherwise`
52	`* it is the index in the vector ret where we will put the result extracted`	52	`* it is the index in the vector ret where we will put the result extracted`
53	`* from ret */`	53	`* from ret */`
54	`int _get_vector_index_typed_result(sqlite3_stmt stmt, SEXP ret, int idx_or_len) {`	54	`int _get_vector_index_typed_result(sqlite3_stmt stmt, SEXP ret, int colidx,`
-		55	`int idx_or_len, int *coltype) {`
55	`int added = 1;`	56	`int added = 1, ctype;`
56	`if (ret == NULL \|\| ret == R_NilValue) {`	57	`if (ret == NULL \|\| ret == R_NilValue) {`
57	`const char *coltype = sqlite3_column_decltype(stmt, 0);`	-
58	`if (sqlite3_column_type(stmt, 0) == SQLITE_NULL) {`	58	`if ((ctype = sqlite3_column_type(stmt, colidx)) == SQLITE_NULL) {`
59	`added = 0;`	59	`added = 0;`
60	`}`	60	`}`
61		61
62	`if (strcmp(coltype, "text") == 0) {`	62	`if (ctype == SQLITE_TEXT) {`
63	`PROTECT(*ret = NEW_CHARACTER(idx_or_len));`	63	`PROTECT(*ret = NEW_CHARACTER(idx_or_len));`
64	`if (added)`	64	`if (added)`
65	`SET_STRING_ELT(ret, 0, mkChar((char )sqlite3_column_text(stmt, 0)));`	65	`SET_STRING_ELT(ret, 0, mkChar((char )sqlite3_column_text(stmt, colidx)));`
66	`} else if (strcmp(coltype, "double") == 0) {`	66	`} else if (ctype == SQLITE_FLOAT) {`
67	`PROTECT(*ret = NEW_NUMERIC(idx_or_len));`	67	`PROTECT(*ret = NEW_NUMERIC(idx_or_len));`
68	`if (added) REAL(*ret)[0] = sqlite3_column_double(stmt, 0);`	68	`if (added) REAL(*ret)[0] = sqlite3_column_double(stmt, colidx);`
69	`} else if (strcmp(coltype, "bit") == 0) {`	69	`} else if (ctype == SQLITE_INTEGER) {`
70	`PROTECT(*ret = NEW_LOGICAL(idx_or_len));`	70	`/* sqlite does not differentiate b/w ints & bits, so we have to`
71	`if (added) INTEGER(*ret)[0] = sqlite3_column_int(stmt, 0);`	71	`* do it ourselves since R distinguishes b/w ints & logicals */`
72	`} else if (strcmp(coltype, "integer") == 0 \|\|`	72	`if (strcmp(sqlite3_column_decltype(stmt, colidx), "bit") == 0) {`
73	`strcmp(coltype, "int") == 0) {`	73	`ctype = SQLITEDF_BIT;`
-		74	`PROTECT(*ret = NEW_LOGICAL(idx_or_len));`
74	`/* caller should just copy off the vars level attr for factors */`	75	`if (added) LOGICAL(*ret)[0] = sqlite3_column_int(stmt, colidx);`
-		76	`} else {`
75	`PROTECT(*ret = NEW_INTEGER(idx_or_len));`	77	`PROTECT(*ret = NEW_INTEGER(idx_or_len));`
76	`if (added) INTEGER(*ret)[0] = sqlite3_column_int(stmt, 0);`	78	`if (added) INTEGER(*ret)[0] = sqlite3_column_int(stmt, colidx);`
-		79	`}`
77	`} else added = 0;`	80	`} else added = 0;`
78		81
79	`UNPROTECT(1);`	82	`if (added) UNPROTECT(1);`
-		83	`*coltype = ctype;`
80	`} else if (stmt != NULL) {`	84	`} else if (stmt != NULL) {`
-		85	`int ctype = *coltype;`
-		86
-		87	`/* first row must have been NULL so that coltype was not set /`
-		88	`if (ctype == SQLITE_NULL) {`
81	`const char *coltype = sqlite3_column_decltype(stmt, 0);`	89	`ctype = *coltype = sqlite3_column_type(stmt, colidx);`
-		90	`/* distinguish int or logical */`
-		91	`if (ctype == SQLITE_INTEGER)`
-		92	`ctype = (strcmp(sqlite3_column_decltype(stmt, colidx), "bit") == 0) ?`
-		93	`SQLITEDF_BIT : SQLITE_INTEGER;`
-		94	`}`
-		95
82	`if (sqlite3_column_type(stmt, 0) == SQLITE_NULL) {`	96	`if (sqlite3_column_type(stmt, colidx) == SQLITE_NULL) {`
83	`added = 0;`	97	`added = 0;`
84	`} else if (strcmp(coltype, "text") == 0) {`	98	`} else if (ctype == SQLITE_TEXT) {`
85	`SET_STRING_ELT(*ret, idx_or_len,`	99	`SET_STRING_ELT(*ret, idx_or_len,`
86	`mkChar((char *)sqlite3_column_text(stmt, 0)));`	100	`mkChar((char *)sqlite3_column_text(stmt, colidx)));`
87	`} else if (strcmp(coltype, "double") == 0) {`	101	`} else if (ctype == SQLITE_FLOAT) {`
88	`REAL(*ret)[idx_or_len] = sqlite3_column_double(stmt, 0);`	102	`REAL(*ret)[idx_or_len] = sqlite3_column_double(stmt, colidx);`
89	`} else if (strcmp(coltype, "bit") == 0) {`	103	`} else if (ctype == SQLITE_INTEGER) {`
90	`INTEGER(*ret)[idx_or_len] = sqlite3_column_int(stmt, 0);`	104	`INTEGER(*ret)[idx_or_len] = sqlite3_column_int(stmt, colidx);`
91	`} else if (strcmp(coltype, "integer") == 0 \|\|`	105	`} else if (ctype == SQLITEDF_BIT) {`
92	`strcmp(coltype, "int") == 0) {`	-
93	`/* caller should just copy off the vars level attr for factors */`	-
94	`INTEGER(*ret)[idx_or_len] = sqlite3_column_int(stmt, 0);`	106	`LOGICAL(*ret)[idx_or_len] = sqlite3_column_int(stmt, colidx);`
95	`} else added = 0;`	107	`} else added = 0;`
96	`} else if (stmt == NULL) {`	108	`} else if (stmt == NULL) {`
97	`const char *coltype = sqlite3_column_decltype(stmt, 0);`	109	`/* used when there is no row returned (sqlite3_step(stmt)!= QLITE_ROW),`
-		110	`* and we just insert NAs in the result */`
-		111	`ctype = *coltype;`
98	`if (sqlite3_column_type(stmt, 0) == SQLITE_NULL) {`	112	`if (ctype == SQLITE_NULL) {`
-		113	`/* since stmt == NULL, we can't check its column type */`
99	`added = 0;`	114	`added = 0;`
100	`} else if (strcmp(coltype, "text") == 0) {`	115	`} else if (ctype == SQLITE_TEXT) {`
101	`SET_STRING_ELT(*ret, idx_or_len, NA_STRING);`	116	`SET_STRING_ELT(*ret, idx_or_len, NA_STRING);`
102	`} else if (strcmp(coltype, "double") == 0) {`	117	`} else if (ctype == SQLITE_FLOAT) {`
103	`REAL(*ret)[idx_or_len] = NA_REAL;`	118	`REAL(*ret)[idx_or_len] = NA_REAL;`
104	`} else if (strcmp(coltype, "bit") == 0 \|\| strcmp(coltype, "integer") == 0 \|\|`	-
105	`strcmp(coltype, "int") == 0) {`	119	`} else if (ctype == SQLITE_INTEGER) {`
106	`INTEGER(*ret)[idx_or_len] = NA_INTEGER;`	120	`INTEGER(*ret)[idx_or_len] = NA_INTEGER;`
-		121	`} else if (ctype == SQLITEDF_BIT) {`
-		122	`LOGICAL(*ret)[idx_or_len] = NA_LOGICAL;`
107	`} else added = 0;`	123	`} else added = 0;`
108	`}`	124	`}`
109		125
110	`return added;`	126	`return added;`
111	`}`	127	`}`
112		128
113	`/****************************************************************************`	129	`/****************************************************************************`
114	`* SVEC FUNCTIONS`	130	`* SVEC FUNCTIONS`
115	`****************************************************************************/`	131	`****************************************************************************/`
116	`SEXP sdf_get_variable(SEXP sdf, SEXP name) {`	132	`SEXP sdf_get_variable(SEXP sdf, SEXP name) {`
117	`char iname, varname, *svec_type = NULL;`	133	`char iname, varname, *svec_type = NULL;`
118	`const char *coltype;`	134	`const char *coltype;`
119	`int type = -1, res, nprotected = 0;`	135	`int type = -1, res, nprotected = 0;`
120	`SEXP ret, value;`	136	`SEXP ret, value;`
121		137
122	`if (!IS_CHARACTER(name)) {`	138	`if (!IS_CHARACTER(name)) {`
123	`error("argument is not a string.\n");`	139	`error("argument is not a string.\n");`
124	`}`	140	`}`
125		141
126	`iname = SDF_INAME(sdf);`	142	`iname = SDF_INAME(sdf);`
127	`varname = CHAR_ELT(name, 0);`	143	`varname = CHAR_ELT(name, 0);`
128		144
129	`if (!USE_SDF1(iname, TRUE, FALSE)) return R_NilValue;`	145	`if (!USE_SDF1(iname, TRUE, FALSE)) return R_NilValue;`
130		146
131	`/* check if sdf & varname w/in that sdf exists */`	147	`/* check if sdf & varname w/in that sdf exists */`
132	`sqlite3_stmt *stmt;`	148	`sqlite3_stmt *stmt;`
133	`sprintf(g_sql_buf[0], "select [%s] from [%s].sdf_data", varname, iname);`	149	`sprintf(g_sql_buf[0], "select [%s] from [%s].sdf_data", varname, iname);`
134		150
135	`res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0);`	151	`res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0);`
136		152
137	`if (_sqlite_error(res)) return R_NilValue;`	153	`if (_sqlite_error(res)) return R_NilValue;`
138		154
139	`coltype = sqlite3_column_decltype(stmt, 0);`	155	`coltype = sqlite3_column_decltype(stmt, 0);`
140	`sqlite3_finalize(stmt);`	156	`sqlite3_finalize(stmt);`
141		157
142	`PROTECT(ret = NEW_LIST(3)); nprotected++;`	158	`PROTECT(ret = NEW_LIST(3)); nprotected++;`
143		159
144	`/* set list names */`	160	`/* set list names */`
145	`PROTECT(value = NEW_CHARACTER(3)); nprotected++;`	161	`PROTECT(value = NEW_CHARACTER(3)); nprotected++;`
146	`SET_STRING_ELT(value, 0, mkChar("iname"));`	162	`SET_STRING_ELT(value, 0, mkChar("iname"));`
147	`SET_STRING_ELT(value, 1, mkChar("tblname"));`	163	`SET_STRING_ELT(value, 1, mkChar("tblname"));`
148	`SET_STRING_ELT(value, 2, mkChar("varname"));`	164	`SET_STRING_ELT(value, 2, mkChar("varname"));`
149	`SET_NAMES(ret, value);`	165	`SET_NAMES(ret, value);`
150		166
151	`/* set list values */`	167	`/* set list values */`
152	`SET_VECTOR_ELT(ret, 0, mkString(iname));`	168	`SET_VECTOR_ELT(ret, 0, mkString(iname));`
153	`SET_VECTOR_ELT(ret, 1, mkString("sdf_data"));`	169	`SET_VECTOR_ELT(ret, 1, mkString("sdf_data"));`
154	`SET_VECTOR_ELT(ret, 2, mkString(varname));`	170	`SET_VECTOR_ELT(ret, 2, mkString(varname));`
155		171
156	`/* set class */`	172	`/* set class */`
157	`if (strcmp(coltype, "text") == 0) svec_type = "character";`	173	`if (strcmp(coltype, "text") == 0) svec_type = "character";`
158	`else if (strcmp(coltype, "double") == 0) svec_type = "numeric";`	174	`else if (strcmp(coltype, "double") == 0) svec_type = "numeric";`
159	`else if (strcmp(coltype, "bit") == 0) svec_type = "logical";`	175	`else if (strcmp(coltype, "bit") == 0) svec_type = "logical";`
160	`else if (strcmp(coltype, "integer") == 0 \|\| strcmp(coltype, "int") == 0) {`	176	`else if (strcmp(coltype, "integer") == 0 \|\| strcmp(coltype, "int") == 0) {`
161	`/* determine if int, factor or ordered */`	177	`/* determine if int, factor or ordered */`
162	`type = _get_factor_levels1(iname, varname, ret);`	178	`type = _get_factor_levels1(iname, varname, ret, FALSE);`
163	`switch(type) {`	179	`switch(type) {`
164	`case VAR_INTEGER: svec_type = "integer"; break;`	180	`case VAR_INTEGER: svec_type = "integer"; break;`
165	`case VAR_FACTOR: svec_type = "factor"; break;`	181	`case VAR_FACTOR: svec_type = "factor"; break;`
166	`case VAR_ORDERED: svec_type = "ordered";`	182	`case VAR_ORDERED: svec_type = "ordered";`
167	`}`	183	`}`
168		-
169	`}`	184	`}`
170		185
171	`SET_CLASS(ret, mkString("sqlite.vector"));`	186	`SET_CLASS(ret, mkString("sqlite.vector"));`
172	`SET_SDFVECTORTYPE(ret, mkString(svec_type));`	187	`SET_SDFVECTORTYPE(ret, mkString(svec_type));`
173		188
174	`UNPROTECT(nprotected);`	189	`UNPROTECT(nprotected);`
175	`return ret;`	190	`return ret;`
176		191
177	`}`	192	`}`
178		193
179		194
180	`SEXP sdf_get_variable_length(SEXP svec) {`	195	`SEXP sdf_get_variable_length(SEXP svec) {`
181	`char *iname = SDF_INAME(svec);`	196	`char *iname = SDF_INAME(svec);`
182	`if (!USE_SDF1(iname, TRUE, FALSE)) return R_NilValue;`	197	`if (!USE_SDF1(iname, TRUE, FALSE)) return R_NilValue;`
183		198
184	`return ScalarInteger(_get_row_count2(iname, 1));`	199	`return ScalarInteger(_get_row_count2(iname, 1));`
185	`}`	200	`}`
186		201
187		202
188	`SEXP sdf_get_variable_index(SEXP svec, SEXP idx) {`	203	`SEXP sdf_get_variable_index(SEXP svec, SEXP idx) {`
189	`SEXP ret = R_NilValue, tmp;`	204	`SEXP ret = R_NilValue, tmp;`
190	`char iname = SDF_INAME(svec), tblname = SVEC_TBLNAME(svec),`	205	`char iname = SDF_INAME(svec), tblname = SVEC_TBLNAME(svec),`
191	`*varname = SVEC_VARNAME(svec);`	206	`*varname = SVEC_VARNAME(svec);`
192	`int index, idxlen, i, retlen=0, res;`	207	`int index, idxlen, i, retlen=0, res, coltype;`
193	`sqlite3_stmt *stmt;`	208	`sqlite3_stmt *stmt;`
194		209
195	`if (!USE_SDF1(iname, TRUE, FALSE)) return R_NilValue;`	210	`if (!USE_SDF1(iname, TRUE, FALSE)) return R_NilValue;`
196		211
197	`idxlen = LENGTH(idx);`	212	`idxlen = LENGTH(idx);`
198	`if (idxlen < 1) return ret;`	213	`if (idxlen < 1) return ret;`
199		214
200	`sprintf(g_sql_buf[0], "select [%s] from [%s].[%s] limit ?,1",`	215	`sprintf(g_sql_buf[0], "select [%s] from [%s].[%s] where rowid=?",`
201	`varname, iname, tblname);`	216	`varname, iname, tblname);`
202	`res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0);`	217	`res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0);`
203	`if (_sqlite_error(res)) error("cannot complete request");`	218	`if (_sqlite_error(res)) error("cannot complete request");`
204		219
205	`/* get data based on index */`	220	`/* get data based on index */`
206	`if (IS_NUMERIC(idx)) {`	221	`if (IS_NUMERIC(idx)) {`
-		222	`/* special handling of 1st index, bec this is where we create the`
-		223	`* SEXP to be used to hold the result */`
207	`index = ((int) REAL(idx)[0]) - 1;`	224	`index = (int) REAL(idx)[0];`
208	`if (index < 0 && idxlen == 1) return ret;`	225	`if (index < 1 && idxlen == 1) return ret;`
209		226
210	`if (index >= 0) {`	227	`if (index >= 1) {`
211	`sqlite3_bind_int(stmt, 1, index);`	228	`sqlite3_bind_int(stmt, 1, index);`
212	`res = sqlite3_step(stmt);`	229	`res = sqlite3_step(stmt);`
213	`if (res == SQLITE_ROW) {`	230	`if (res == SQLITE_ROW) {`
214	`retlen = _get_vector_index_typed_result(stmt, &ret, idxlen);`	231	`retlen = _get_vector_index_typed_result(stmt, &ret, 0, idxlen, &coltype);`
215	`}`	232	`}`
216	`}`	233	`}`
217		234
218	`if (index < 0 \|\| res != SQLITE_ROW) {`	235	`if (index < 1 \|\| res != SQLITE_ROW) {`
219	`/* something wrong w/ 1st idx, and it is quietly ignored. we make`	236	`/* something wrong w/ 1st idx, and it is quietly ignored. we make`
220	`* a "dummy" call to setup the SEXP */`	237	`* a "dummy" call to setup the SEXP */`
221	`sqlite3_reset(stmt);`	238	`sqlite3_reset(stmt);`
222	`sqlite3_bind_int(stmt, 1, 0);`	239	`sqlite3_bind_int(stmt, 1, 0);`
223	`res = sqlite3_step(stmt);`	240	`res = sqlite3_step(stmt);`
224	`if (res == SQLITE_ROW) {`	241	`if (res == SQLITE_ROW) {`
-		242	`/* populate with 1st row. this should be overwritten by`
-		243	`* succeeding calls. if there are no valid result, then`
-		244	`* everything will be removed by _shrink_vector().`
-		245	`*/`
225	`_get_vector_index_typed_result(stmt, &ret, idxlen - 1);`	246	`_get_vector_index_typed_result(stmt, &ret, 0, idxlen - 1, &coltype);`
226	`}`	247	`}`
227	`retlen = 0;`	248	`retlen = 0;`
228	`}`	249	`}`
229		250
230		251
231	`if (idxlen > 1) {`	252	`if (idxlen > 1) {`
232	`for (i = 1; i < idxlen; i++) {`	253	`for (i = 1; i < idxlen; i++) {`
233	`index = ((int) REAL(idx)[i]) - 1;`	254	`index = (int) REAL(idx)[i];`
234	`if (index < 0) continue;`	255	`if (index < 1) continue;`
235	`sqlite3_reset(stmt);`	256	`sqlite3_reset(stmt);`
236	`sqlite3_bind_int(stmt, 1, index);`	257	`sqlite3_bind_int(stmt, 1, index);`
237	`res = sqlite3_step(stmt);`	258	`res = sqlite3_step(stmt);`
238	`if (res == SQLITE_ROW) {`	259	`if (res == SQLITE_ROW) {`
239	`retlen += _get_vector_index_typed_result(stmt, &ret, retlen);`	260	`retlen += _get_vector_index_typed_result(stmt, &ret, 0, retlen, &coltype);`
240	`} else {`	261	`} else {`
241	`retlen += _get_vector_index_typed_result(NULL, &ret, retlen);`	262	`retlen += _get_vector_index_typed_result(NULL, &ret, 0, retlen, &coltype);`
242	`}`	263	`}`
243	`}`	264	`}`
244	`}`	265	`}`
245	`} else if (IS_INTEGER(idx)) {`	266	`} else if (IS_INTEGER(idx)) {`
246	`/* similar to REAL (IS_NUMERIC) above, except that we don't have`	267	`/* similar to REAL (IS_NUMERIC) above, except that we don't have`
247	`* to cast idx to int. can't refactor this out, sucks. */`	268	`* to cast idx to int. can't refactor this out, sucks. */`
248	`index = INTEGER(idx)[0] - 1;`	269	`index = INTEGER(idx)[0];`
249	`if (index < 0 && idxlen == 1) return ret;`	270	`if (index < 1 && idxlen == 1) return ret;`
250		271
251	`if (index >= 0) {`	272	`if (index >= 1) {`
252	`sqlite3_bind_int(stmt, 1, index);`	273	`sqlite3_bind_int(stmt, 1, index);`
253	`res = sqlite3_step(stmt);`	274	`res = sqlite3_step(stmt);`
254	`if (res == SQLITE_ROW) {`	275	`if (res == SQLITE_ROW) {`
255	`retlen = _get_vector_index_typed_result(stmt, &ret, idxlen);`	276	`retlen = _get_vector_index_typed_result(stmt, &ret, 0, idxlen, &coltype);`
256	`}`	277	`}`
257	`}`	278	`}`
258		279
259	`if (index < 0 \|\| res != SQLITE_ROW) {`	280	`if (index < 1 \|\| res != SQLITE_ROW) {`
260	`/* something wrong w/ 1st idx, and it is quietly ignored. we make`	281	`/* something wrong w/ 1st idx, and it is quietly ignored. we make`
261	`* a "dummy" call to setup the SEXP */`	282	`* a "dummy" call to setup the SEXP */`
262	`sqlite3_reset(stmt);`	283	`sqlite3_reset(stmt);`
263	`sqlite3_bind_int(stmt, 1, 0);`	284	`sqlite3_bind_int(stmt, 1, 0);`
264	`res = sqlite3_step(stmt);`	285	`res = sqlite3_step(stmt);`
265	`if (res == SQLITE_ROW) {`	286	`if (res == SQLITE_ROW) {`
266	`_get_vector_index_typed_result(stmt, &ret, idxlen - 1);`	287	`_get_vector_index_typed_result(stmt, &ret, 0, idxlen - 1, &coltype);`
267	`}`	288	`}`
268	`retlen = 0;`	289	`retlen = 0;`
269	`}`	290	`}`
270		291
271	`if (idxlen > 1) {`	292	`if (idxlen > 1) {`
272	`for (i = 1; i < idxlen; i++) {`	293	`for (i = 1; i < idxlen; i++) {`
273	`index = INTEGER(idx)[i] - 1;`	294	`index = INTEGER(idx)[i];`
274	`if (index < 0) continue;`	295	`if (index < 1) continue;`
275	`sqlite3_reset(stmt);`	296	`sqlite3_reset(stmt);`
276	`sqlite3_bind_int(stmt, 1, index);`	297	`sqlite3_bind_int(stmt, 1, index);`
277	`res = sqlite3_step(stmt);`	298	`res = sqlite3_step(stmt);`
278	`if (res == SQLITE_ROW) {`	299	`if (res == SQLITE_ROW) {`
279	`retlen += _get_vector_index_typed_result(stmt, &ret, retlen);`	300	`retlen += _get_vector_index_typed_result(stmt, &ret, 0, retlen, &coltype);`
280	`} else {`	301	`} else {`
281	`retlen += _get_vector_index_typed_result(NULL, &ret, retlen);`	302	`retlen += _get_vector_index_typed_result(NULL, &ret, 0, retlen, &coltype);`
282	`}`	303	`}`
283	`}`	304	`}`
284	`}`	305	`}`
285		306
286	`} else if (IS_LOGICAL(idx)) {`	307	`} else if (IS_LOGICAL(idx)) {`
287	`/* have to deal with recycling */`	308	`/* have to deal with recycling */`
288	`int veclen = _get_row_count2(iname, 1);`	309	`int veclen = _get_row_count2(iname, 1);`
289		310
-		311	`/* in this indexing, idx[i] == TRUE is equiv to vector[i]. therefore`
-		312	`* idxlen <= veclen. if idxlen < veclen, then idx is recycled. */`
-		313
290	`/* find if there is any TRUE element in the vector */`	314	`/* find if there is any TRUE element in the vector */`
291	`for (i = 0; i < idxlen && i < veclen; i++) {`	315	`for (i = 1; i < idxlen && i < veclen; i++) {`
292	`if (LOGICAL(idx)[i]) {`	316	`if (LOGICAL(idx)[i]) {`
293	`sqlite3_bind_int(stmt, 1, i);`	317	`sqlite3_bind_int(stmt, 1, i+1);`
294	`sqlite3_step(stmt);`	318	`sqlite3_step(stmt);`
295	`/* there are at least (idxlen-i) TRUE per cycle of the LOGICAL`	319	`/* there are at least (idxlen-i) TRUE per cycle of the LOGICAL`
296	`* index. there are at least (veclen/idxlen) cycles (int div).`	320	`* index. there are at least (veclen/idxlen) cycles (int div).`
297	`* at the last cycle, if (veclen%idxlen > 0), there will be`	321	`* at the last cycle, if (veclen%idxlen > 0), there will be`
298	`* at least (veclen%idxlen - i) if veclen%idxlen > i */`	322	`* at least (veclen%idxlen - i) if veclen%idxlen > i */`
299	`retlen = (idxlen-i) * (veclen/idxlen);`	323	`retlen = (idxlen-i) * (veclen/idxlen);`
300	`if (veclen%idxlen > i) retlen += (veclen%idxlen - i);`	324	`if (veclen%idxlen > i) retlen += (veclen%idxlen - i);`
301		325
302	`/* create the vector */`	326	`/* create the vector */`
303	`retlen = _get_vector_index_typed_result(stmt, &ret, retlen);`	327	`retlen = _get_vector_index_typed_result(stmt, &ret, 0, retlen, &coltype);`
304	`break;`	328	`break;`
305	`}`	329	`}`
306	`}`	330	`}`
307		331
308	`if (i < idxlen && i < veclen) {`	332	`if (i < idxlen && i < veclen) {`
309	`for (i++; i < veclen; i++) {`	333	`for (i++; i < veclen; i++) {`
310	`if (LOGICAL(idx)[i%idxlen]) {`	334	`if (LOGICAL(idx)[i%idxlen]) {`
311	`sqlite3_reset(stmt);`	335	`sqlite3_reset(stmt);`
312	`sqlite3_bind_int(stmt, 1, i);`	336	`sqlite3_bind_int(stmt, 1, i+1);`
313	`res = sqlite3_step(stmt);`	337	`res = sqlite3_step(stmt);`
314	`if (res == SQLITE_ROW) {`	338	`if (res == SQLITE_ROW) {`
315	`retlen += _get_vector_index_typed_result(stmt, &ret, retlen);`	339	`retlen += _get_vector_index_typed_result(stmt, &ret, 0, retlen, &coltype);`
316	`} else {`	340	`} else {`
317	`retlen += _get_vector_index_typed_result(NULL, &ret, retlen);`	341	`retlen += _get_vector_index_typed_result(NULL, &ret, 0, retlen, &coltype);`
318	`}`	342	`}`
319	`}`	343	`}`
320	`}`	344	`}`
321	`}`	345	`}`
322	`}`	346	`}`
323		347
324	`sqlite3_finalize(stmt);`	348	`sqlite3_finalize(stmt);`
325		349
326	`if (ret != R_NilValue) {`	350	`if (ret != R_NilValue) {`
327	`ret = _shrink_vector(ret, retlen);`	351	`ret = _shrink_vector(ret, retlen);`
328	`tmp = GET_LEVELS(svec);`	352	`tmp = GET_LEVELS(svec);`
329	`if (tmp != R_NilValue) {`	353	`if (tmp != R_NilValue) {`
330	`SET_LEVELS(ret, duplicate(tmp));`	354	`SET_LEVELS(ret, duplicate(tmp));`
331	`if (TEST_SDFVECTORTYPE(svec, "factor")) {`	355	`if (TEST_SDFVECTORTYPE(svec, "factor")) {`
332	`SET_CLASS(ret, mkString("factor"));`	356	`SET_CLASS(ret, mkString("factor"));`
333	`} else {`	357	`} else {`
334	`PROTECT(tmp = NEW_CHARACTER(2));`	358	`PROTECT(tmp = NEW_CHARACTER(2));`
335	`SET_STRING_ELT(tmp, 0, mkChar("ordered"));`	359	`SET_STRING_ELT(tmp, 0, mkChar("ordered"));`
336	`SET_STRING_ELT(tmp, 1, mkChar("factor"));`	360	`SET_STRING_ELT(tmp, 1, mkChar("factor"));`
337	`UNPROTECT(1);`	361	`UNPROTECT(1);`
338	`}`	362	`}`
339	`}`	363	`}`
340	`}`	364	`}`
341		365
342	`return ret;`	366	`return ret;`
343	`}`	367	`}`
344		368
345	`/* sqlite.vector.[<- */`	369	`/* sqlite.vector.[<- */`
346	`SEXP sdf_set_variable_index(SEXP svec, SEXP idx, SEXP value) {`	370	`SEXP sdf_set_variable_index(SEXP svec, SEXP idx, SEXP value) {`
347	`int idx_len, val_len;`	371	`int idx_len, val_len, svec_len, i, res;`
-		372	`char iname = SDF_INAME(svec), tblname = SVEC_TBLNAME(svec),`
-		373	`*varname = SVEC_VARNAME(svec);`
-		374	`sqlite3_stmt *stmt;`
348		375
349	`/* match levels if factor or ordered */`	376	`/* match levels if factor or ordered */`
350	`if (inherits(value, "ordered")) {`	377	`if (inherits(value, "ordered")) {`
351	`} else if (inherits(value, "factor")) {`	378	`} else if (inherits(value, "factor")) {`
352	`}`	379	`}`
353		380
354	`idx_len = LENGTH(idx);`	381	`idx_len = LENGTH(idx);`
355	`val_len = LENGTH(value);`	382	`val_len = LENGTH(value);`
-		383	`svec_len = _get_row_count2(iname, TRUE);`
-		384
-		385	`if (idx_len % val_len != 0)`
-		386	`warning("number of items to replace is not a multiple of replacement length");`
-		387
-		388	`sprintf(g_sql_buf[0], "update [%s].[%s] set [%s]=? where rowid=?",`
-		389	`iname, tblname, varname);`
-		390	`_sqlite_begin;`
-		391	`res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0);`
-		392	`if (_sqlite_error(res)) {`
-		393	`error("cannot complete request");`
-		394	`_sqlite_commit;`
-		395	`}`
356		396
357	`if (IS_NUMERIC(idx)) {`	397	`if (IS_NUMERIC(idx)) {`
-		398	`for (i = 0; i < idx_len; i++) {`
-		399	`/* determine type of svec, cast value to type of it */`
-		400	`/* value can only be of a single type, because it is a vector! */`
-		401	`/* if svec is a factor, value is a string, must check with`
-		402	`* factor levels */`
-		403	`}`
358	`} else if (IS_INTEGER(idx)) {`	404	`} else if (IS_INTEGER(idx)) {`
359	`} else if (IS_LOGICAL(idx)) {`	405	`} else if (IS_LOGICAL(idx)) {`
360	`}`	406	`}`
361		407
-		408	`_sqlite_commit;`
362	`return R_NilValue;`	409	`return R_NilValue;`
363	`}`	410	`}`
364		411
365	`SEXP sdf_variable_summary(SEXP svec, SEXP maxsum) {`	412	`SEXP sdf_variable_summary(SEXP svec, SEXP maxsum) {`
366	`char iname, tblname, varname, type;`	413	`char iname, tblname, varname, type;`
367	`sqlite3_stmt *stmt;`	414	`sqlite3_stmt *stmt;`
368	`int nprotected = 0;`	415	`int nprotected = 0;`
369	`SEXP ret, names;`	416	`SEXP ret, names;`
370		417
371	`iname = SDF_INAME(svec);`	418	`iname = SDF_INAME(svec);`
372	`tblname = SVEC_TBLNAME(svec);`	419	`tblname = SVEC_TBLNAME(svec);`
373	`varname = SVEC_VARNAME(svec);`	420	`varname = SVEC_VARNAME(svec);`
374	`USE_SDF1(iname, TRUE, FALSE);`	421	`USE_SDF1(iname, TRUE, FALSE);`
375		422
376	`if ((TEST_SDFVECTORTYPE(svec, "ordered") && ((type = "ordered"))) \|\|`	423	`if ((TEST_SDFVECTORTYPE(svec, "ordered") && ((type = "ordered"))) \|\|`
377	`(TEST_SDFVECTORTYPE(svec, "factor") && ((type = "factor")))) {`	424	`(TEST_SDFVECTORTYPE(svec, "factor") && ((type = "factor")))) {`
378	`int nrows, i, max_rows = INTEGER(maxsum)[0];`	425	`int nrows, i, max_rows = INTEGER(maxsum)[0];`
379		426
380		427
381	`sprintf(g_sql_buf[0], "[%s].[%s %s]", iname, type, varname);`	428	`sprintf(g_sql_buf[0], "[%s].[%s %s]", iname, type, varname);`
382	`nrows = _get_row_count2(g_sql_buf[0], FALSE);`	429	`nrows = _get_row_count2(g_sql_buf[0], FALSE);`
383	`if (nrows <= max_rows) max_rows = nrows;`	430	`if (nrows <= max_rows) max_rows = nrows;`
384	`else { nrows = max_rows; max_rows--; }`	431	`else { nrows = max_rows; max_rows--; }`
385		432
386	`PROTECT(ret = NEW_INTEGER(nrows)); nprotected = 1;`	433	`PROTECT(ret = NEW_INTEGER(nrows)); nprotected = 1;`
387	`PROTECT(names = NEW_CHARACTER(nrows)); nprotected++;`	434	`PROTECT(names = NEW_CHARACTER(nrows)); nprotected++;`
388		435
389	`sprintf(g_sql_buf[0], "select [%s].[%s %s].label, count(*) from "`	436	`sprintf(g_sql_buf[0], "select [%s].[%s %s].label, count(*) from "`
390	`"[%s].[%s] join [%s].[%s %s] on [%s].[%s].[%s]=[%s].[%s %s].level "`	437	`"[%s].[%s] join [%s].[%s %s] on [%s].[%s].[%s]=[%s].[%s %s].level "`
391	`"group by [%s].[%s].[%s], [%s].[%s %s].level order by count(*) desc",`	438	`"group by [%s].[%s].[%s], [%s].[%s %s].level order by count(*) desc",`
392	`iname, type, varname, iname, tblname, iname, type, varname, iname, tblname, varname,`	439	`iname, type, varname, iname, tblname, iname, type, varname, iname, tblname, varname,`
393	`iname, type, varname, iname, tblname, varname, iname, type, varname);`	440	`iname, type, varname, iname, tblname, varname, iname, type, varname);`
394	`sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0);`	441	`sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0);`
395		442
396	`for (i = 0; i < max_rows; i++) {`	443	`for (i = 0; i < max_rows; i++) {`
397	`sqlite3_step(stmt);`	444	`sqlite3_step(stmt);`
398	`SET_STRING_ELT(names, i, mkChar((char *)sqlite3_column_text(stmt, 0)));`	445	`SET_STRING_ELT(names, i, mkChar((char *)sqlite3_column_text(stmt, 0)));`
399	`INTEGER(ret)[i] = sqlite3_column_int(stmt, 1);`	446	`INTEGER(ret)[i] = sqlite3_column_int(stmt, 1);`
400	`}`	447	`}`
401		448
402	`if (nrows > max_rows) {`	449	`if (nrows > max_rows) {`
403	`int others_sum = 0;`	450	`int others_sum = 0;`
404	`SET_STRING_ELT(names, nrows-1, mkChar("(Others)"));`	451	`SET_STRING_ELT(names, nrows-1, mkChar("(Others)"));`
405	`while (sqlite3_step(stmt) == SQLITE_ROW) {`	452	`while (sqlite3_step(stmt) == SQLITE_ROW) {`
406	`others_sum += sqlite3_column_int(stmt, 1);`	453	`others_sum += sqlite3_column_int(stmt, 1);`
407	`}`	454	`}`
408	`INTEGER(ret)[nrows-1] = others_sum;`	455	`INTEGER(ret)[nrows-1] = others_sum;`
409	`}`	456	`}`
410	`} else if (TEST_SDFVECTORTYPE(svec, "logical")) {`	457	`} else if (TEST_SDFVECTORTYPE(svec, "logical")) {`
411	`sprintf(g_sql_buf[0], "select count(*) from "`	458	`sprintf(g_sql_buf[0], "select count(*) from "`
412	`"[%s].[%s] group by [%s] order by [%s]", iname, tblname, varname, varname);`	459	`"[%s].[%s] group by [%s] order by [%s]", iname, tblname, varname, varname);`
413		460
414	`PROTECT(names = NEW_CHARACTER(3)); nprotected = 1;`	461	`PROTECT(names = NEW_CHARACTER(3)); nprotected = 1;`
415	`PROTECT(ret = NEW_CHARACTER(3)); nprotected++;`	462	`PROTECT(ret = NEW_CHARACTER(3)); nprotected++;`
416		463
417	`SET_STRING_ELT(names, 0, mkChar("Mode"));`	464	`SET_STRING_ELT(names, 0, mkChar("Mode"));`
418	`SET_STRING_ELT(names, 1, mkChar("FALSE"));`	465	`SET_STRING_ELT(names, 1, mkChar("FALSE"));`
419	`SET_STRING_ELT(names, 2, mkChar("TRUE"));`	466	`SET_STRING_ELT(names, 2, mkChar("TRUE"));`
420		467
421	`sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0);`	468	`sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0);`
422	`SET_STRING_ELT(ret, 0, mkChar("logical"));`	469	`SET_STRING_ELT(ret, 0, mkChar("logical"));`
423	`sqlite3_step(stmt);`	470	`sqlite3_step(stmt);`
424	`SET_STRING_ELT(ret, 1, mkChar((const char *)sqlite3_column_text(stmt, 0)));`	471	`SET_STRING_ELT(ret, 1, mkChar((const char *)sqlite3_column_text(stmt, 0)));`
425	`sqlite3_step(stmt);`	472	`sqlite3_step(stmt);`
426	`SET_STRING_ELT(ret, 2, mkChar((const char *)sqlite3_column_text(stmt, 0)));`	473	`SET_STRING_ELT(ret, 2, mkChar((const char *)sqlite3_column_text(stmt, 0)));`
427		474
428	`} else return R_NilValue;`	475	`} else return R_NilValue;`
429		476
430	`sqlite3_finalize(stmt);`	477	`sqlite3_finalize(stmt);`
431	`SET_NAMES(ret, names);`	478	`SET_NAMES(ret, names);`
432	`SET_CLASS(ret, mkString("table"));`	479	`SET_CLASS(ret, mkString("table"));`
433	`UNPROTECT(nprotected);`	480	`UNPROTECT(nprotected);`
434	`return ret;`	481	`return ret;`
435	`}`	482	`}`
436		483
437		484
438	`/* the global accumulator should be safe if we only do 1 cummulative or`	485	`/* the global accumulator should be safe if we only do 1 cummulative or`
439	`* aggregate at any time. it won't work for stuffs like "select max(col)-min(col)`	486	`* aggregate at any time. it won't work for stuffs like "select max(col)-min(col)`
440	`* from sdf_data". */`	487	`* from sdf_data". */`
441	`static long double g_accumulator = 0.0; /* accumulator var for cumsum, cumprod, etc. */`	488	`static long double g_accumulator = 0.0; /* accumulator var for cumsum, cumprod, etc. */`
442	`static int g_start = 0; /* flag for start of cummulation */`	489	`static int g_start = 0; /* flag for start of cummulation */`
443	`static int g_narm = 0; /* for Summary group */`	490	`static int g_narm = 0; /* for Summary group */`
444		491
445	`void _init_sqlite_function_accumulator() {`	492	`void _init_sqlite_function_accumulator() {`
446	`g_accumulator = 0.0; /* initialize accumulator */`	493	`g_accumulator = 0.0; /* initialize accumulator */`
447	`g_start = 1; /* flag that we are at start of accumulating */`	494	`g_start = 1; /* flag that we are at start of accumulating */`
448	`}`	495	`}`

The R Project SVN R-packages

(root)/trunk/SQLiteDF/src/sqlite_vector.c – Rev 3855 → 4160