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


#include <stdio.h>
#include <string.h>
#include "sqlite_dataframe.h"
 
/****************************************************************************
 * UTILITY FUNCTIONS
 ****************************************************************************/
 
/* if user supplied a name, return that name. otherwise, use the 
 * default "data" */
static int _check_sdf_name(SEXP name, char **rname, char **iname, int *file_idx) {
    int namelen = 0;
 
    /* check if arg name is supplied */
    if (IS_CHARACTER(name)) {
        *rname = CHAR_ELT(name,0);
        if (!_is_r_sym(*rname)) { 
            Rprintf("Error: supplied name \"%s\"is not a valid R symbol.\n", 
                    *rname); 
            return FALSE; 
        }
        namelen = strlen(*rname);
        *iname = (char*)R_alloc(namelen + 9, sizeof(char)); /* <name>10000.db\0 */
        sprintf(*iname, "%s.db", *rname);
    } else if (name == R_NilValue) {
        *rname = "data";
        namelen = 5;
        *iname = (char*)R_alloc(13, sizeof(char)); /* data10000.db\0 */
        *file_idx = 1;
        sprintf(*iname, "%s%d.db", *rname, *file_idx);
    } else {
        Rprintf("Error: the supplied value for arg name is not a string.\n");
    }
    return namelen;
}
 
static int _find_free_filename(char *rname, char **iname, int *namelen, int *file_idx) {
    do {
        if (!_file_exists(*iname)) break;
        *namelen = sprintf(*iname, "%s%d.db", rname, ++(*file_idx)) - 3;
    } while (*file_idx < 10000);
 
    return *file_idx;
}
 
/* creates an sdf attribute table */
static int _create_sdf_attribute2(const char *iname) {
    sprintf(g_sql_buf[2], "create table [%s].sdf_attributes(attr text, "
            "value text, primary key (attr))", iname);
    int res;
    res = _sqlite_exec(g_sql_buf[2]);
    if (res == SQLITE_OK) {
        sprintf(g_sql_buf[2], "insert into [%s].sdf_attributes values ('name',"
                "'%s');", iname, iname);
        res = _sqlite_exec(g_sql_buf[2]);
    }
    return res;
}
 
char *_create_sdf_skeleton1(SEXP name, int *onamelen) {
    char *iname, *rname;
    int namelen, file_idx = 0, res;
 
    namelen = _check_sdf_name(name, &rname, &iname, &file_idx);
 
    if (!namelen) return NULL;
 
    _find_free_filename(rname, &iname, &namelen, &file_idx);
 
    if (file_idx >= 10000) { 
        Rprintf("Error: cannot find free SDF name.\n");
        return NULL;
    }
 
    /* add to workspace */
    strcpy(g_sql_buf[3], iname);
    iname[namelen] = 0; /* remove ".db" */
    res = _add_sdf1(g_sql_buf[3], iname);
    if (_sqlite_error(res)) return NULL;
 
    /* detach SDF's if necessary to attach this one. if file does not
     * exist, then a sqlite db will be created after we make our 1st table */
    if (!USE_SDF1(iname, FALSE)) { /* _delete_sdf2(iname); */ return NULL; }
 
 
    /* create attributes table */
    res = _create_sdf_attribute2(iname);
    if (_sqlite_error(res)) {
        sprintf(g_sql_buf[2], "detach '%s'", iname);
        _delete_sdf2(iname);
        return NULL; 
    }
 
    *onamelen = namelen;
    return iname;
}
/* checks if a column has a corresponding factor|ordered table */
static int _is_factor2(const char *iname, const char *factor_type, const char *colname) {
    sqlite3_stmt *stmt;
    sprintf(g_sql_buf[2], "select * from [%s].[%s %s]", iname, factor_type,
            colname);
    int res;
    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
    sqlite3_finalize(stmt);
    return res == SQLITE_OK;
}
 
/* create a factor|ordered table */
static int _create_factor_table2(const char *iname, const char *factor_type, 
        const char *colname) {
    sqlite3_stmt *stmt;
    sprintf(g_sql_buf[2], "create table [%s].[%s %s] (level int, label text, "
            "primary key(level), unique(label));", iname, factor_type, colname);
    int res;
    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
    if (res == SQLITE_OK) sqlite3_step(stmt);
    sqlite3_finalize(stmt);
    return res; /* error on dup name? */
}
 
/* copy a factor|ordered table from a sdf(db) to another sdf(db) */
int _copy_factor_levels2(const char *factor_type, const char *iname_src,
        const char *colname_src, const char *iname_dst, const char *colname_dst) {
    sqlite3_stmt *stmt;
    int res;
    res = _create_factor_table2(iname_dst, factor_type, colname_dst);
    if (res == SQLITE_OK) {
        sprintf(g_sql_buf[2], "insert into [%s].[%s %s] select * from [%s].[%s %s]",
                iname_src, factor_type, colname_src, iname_dst, factor_type,
                colname_dst);
        res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
        if (res == SQLITE_OK) sqlite3_step(stmt);
        sqlite3_finalize(stmt);
    }
    return res; /* error on dup name? */
}
 
 
/* assumes stmt has col_cnt + 1 columns, col 0 is [row name]. returned SEXP is 
 * not UNPROTECT-ed, user will have to do that. will create the names &
 * attach factor infos */
static SEXP _setup_df_sexp1(sqlite3_stmt *stmt, const char *iname, 
        int col_cnt, int row_cnt, int *dup_indices) {
    SEXP ret, names, value, class = R_NilValue;
    int i, type;
    const char *colname, *coltype;
 
    PROTECT(ret = NEW_LIST(col_cnt));
 
    /* set up names. */
    PROTECT(names = NEW_CHARACTER(col_cnt));
    for (i = 0; i < col_cnt; i++) {
        colname = sqlite3_column_name(stmt, i+1);  /* +1 bec [row name is 0] */
        coltype = sqlite3_column_decltype(stmt, i+1);
 
        if (dup_indices[i] == 0) {
            strcpy(g_sql_buf[1], colname);
        } else {
            sprintf(g_sql_buf[1], "%s.%d", colname, dup_indices[i]);
        }
 
        SET_STRING_ELT(names, i, mkChar(g_sql_buf[1]));
 
        if (strcmp(coltype, "text") == 0) {
            PROTECT(value = NEW_CHARACTER(row_cnt));
        } else if (strcmp(coltype, "double") == 0) {
            PROTECT(value = NEW_NUMERIC(row_cnt));
        } else if (strcmp(coltype, "bit") == 0) {
            PROTECT(value = NEW_LOGICAL(row_cnt));
        } else if (strcmp(coltype, "integer") == 0 || 
                   strcmp(coltype, "int") == 0) {
            PROTECT(value = NEW_INTEGER(row_cnt));
            type = _get_factor_levels1(iname, colname, value);
            if (type == VAR_FACTOR) {
                PROTECT(class = mkString("factor"));
                SET_CLASS(value, class);
                UNPROTECT(1);
            } else if (type == VAR_ORDERED) {
                PROTECT(class = NEW_CHARACTER(2));
                SET_STRING_ELT(class, 0, mkChar("ordered"));
                SET_STRING_ELT(class, 1, mkChar("factor"));
                SET_CLASS(value, class);
                UNPROTECT(1);
            }
        } else {
            Rprintf("Error: not supported type %s for %s\n", coltype, colname);
            UNPROTECT(2); /* unprotect ret, names */
            return R_NilValue;
        }
 
        SET_VECTOR_ELT(ret, i, value);
        UNPROTECT(1); /* unprotect value */
                    
    }
    SET_NAMES(ret, names);
    UNPROTECT(1); /* unprotect names only */
    return ret;
}
 
/* expected that 1st col of stmt is [row name], so we start with index 1 */
static int _add_row_to_df(SEXP df, sqlite3_stmt *stmt, int row, int ncols) {
    SEXP vec;
    int type = -1, is_null, i;
    for (i = 1; i <= ncols; i++) {
        vec = VECTOR_ELT(df, i-1);
        type = TYPEOF(vec);
 
        is_null = (sqlite3_column_type(stmt, row) == SQLITE_NULL);
 
        if (type == CHARSXP) {
            SET_STRING_ELT(vec, row, mkChar((char *)sqlite3_column_text(stmt, i)));
        } else if (type == INTSXP) {
            INTEGER(vec)[row] = sqlite3_column_int(stmt, i);
        } else if (type == REALSXP) {
            REAL(vec)[row] = sqlite3_column_double(stmt, i);
        } else if (type == LGLSXP) {
            LOGICAL(vec)[row] = sqlite3_column_int(stmt, i);
        }
    }
    return type;
}
 
static void _set_rownames2(SEXP df) {
    SEXP value = VECTOR_ELT(df, 0);
    int len = LENGTH(value), i;
    PROTECT(value = NEW_CHARACTER(len));
    for (i = 0; i < len; i++) {
        sprintf(g_sql_buf[2], "%d", i+1);
        SET_STRING_ELT(value, i, mkChar(g_sql_buf[2]));
    }
 
    SET_ROWNAMES(df, value);
    UNPROTECT(1);
}
 
 
/****************************************************************************
 * SDF FUNCTIONS
 ****************************************************************************/
 
SEXP sdf_create_sdf(SEXP df, SEXP name) {
    SEXP ret;
    char *iname; 
    int namelen, res, i, j;
 
    /* find free name, attach sdf, create sdf_attributes */
    iname = _create_sdf_skeleton1(name, &namelen);
    
    if (iname != NULL) {
        int sql_len, sql_len2;
        sqlite3_stmt *stmt;
 
        /* create sdf_data table */
        SEXP names = GET_NAMES(df), variable, levels;
        int ncols = GET_LENGTH(names), type, *types;
        char *col_name, *class, *factor;
 
        /* variables for adding rownames */
        SEXP rownames;
        int nrows;
        char *row_name;
 
        /* TODO: put constraints on table after inserting everything? */
 
 
        /* 
         * create the create table and insert sql scripts by looping through
         * the columns of df
         */
        types = (int *)R_alloc(ncols, sizeof(int));
        sql_len = sprintf(g_sql_buf[0], "create table [%s].sdf_data ([row name] text", iname);
        sql_len2 = sprintf(g_sql_buf[1], "insert into [%s].sdf_data values(?", iname);
 
        for (i = 0; i < ncols; i++) {
            col_name = CHAR(STRING_ELT(names, i));
 
            /* add column definition to the create table sql */
            _expand_buf(0, sql_len+strlen(col_name)+10);
 
            variable = _getListElement(df, col_name);
            class = CHAR(STRING_ELT(GET_CLASS(variable), 0));
            type = TYPEOF(variable);
            types[i] = type;
 
            sql_len += sprintf(g_sql_buf[0]+sql_len, ", [%s] %s", col_name, 
                    _get_column_type(class, type));
 
            /* add handler to insert table sql */
            _expand_buf(1, sql_len+5);
            strcpy(g_sql_buf[1]+sql_len2, ",?");
            sql_len2 += 2; 
 
            /* create separate table for factors decode */
            if (strcmp(class, "factor") == 0 || strcmp(class, "ordered") == 0){
                if (_create_factor_table2(iname, class, col_name)) 
                    return R_NilValue; /* dup tbl name? */
 
                levels = GET_LEVELS(variable);
                sprintf(g_sql_buf[2], "insert into [%s].[%s %s] values(?, ?);",
                        iname, class, col_name);
                res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, NULL);
                if (_sqlite_error(res)) return R_NilValue; /* dup tbl name? */
 
                for (j = 0; j < GET_LENGTH(levels); j++) {
                    sqlite3_reset(stmt);
                    factor = CHAR(STRING_ELT(levels, j));
                    sqlite3_bind_int(stmt, 1, j+1);
                    sqlite3_bind_text(stmt, 2, factor, -1, SQLITE_STATIC);
                    sqlite3_step(stmt);
                }
                sqlite3_finalize(stmt);
            }
        }
        
        _expand_buf(0,sql_len+35);
        sql_len += sprintf(g_sql_buf[0]+sql_len, ", primary key([row name]));");
        /* sql_len += sprintf(g_sql_buf[0]+sql_len, ");"); */
        res = _sqlite_exec(g_sql_buf[0]);
        if (_sqlite_error(res)) return R_NilValue; /* why? */
 
        /*
         * add the data in df to the sdf
         */
        rownames = getAttrib(df, R_RowNamesSymbol);
        nrows = GET_LENGTH(rownames);
 
        sprintf(g_sql_buf[1]+sql_len2, ")");
        res = sqlite3_prepare(g_workspace, g_sql_buf[1], -1, &stmt, NULL);
 
        for (i = 0; i < nrows; i++) {
            sqlite3_reset(stmt);
 
            /* since this is coming from a real dataframe, we're assured that
             * the rownames are unique */
            if (IS_CHARACTER(rownames)) {
                row_name = CHAR(STRING_ELT(rownames, i));
                if (*row_name) /* if not empty string */
                    sqlite3_bind_text(stmt, 1, row_name, strlen(row_name), SQLITE_STATIC);
                else sqlite3_bind_int(stmt, 1, i);
            } else if (IS_INTEGER(rownames)) {
                sqlite3_bind_int(stmt, 1, INTEGER(rownames)[i]);
            } else sqlite3_bind_int(stmt, 1, i);
 
            for (j = 0; j < ncols; j++) {
                variable = VECTOR_ELT(df, j);
                switch(types[j]) {
                    case INTSXP : 
                        sqlite3_bind_int(stmt, j+2, INTEGER(variable)[i]);
                        break;
                    case REALSXP:
                        sqlite3_bind_double(stmt, j+2, REAL(variable)[i]);
                        break;
                    case CHARSXP:
                        col_name = CHAR(STRING_ELT(variable,i));
                        sqlite3_bind_text(stmt, j+2, col_name, strlen(col_name), SQLITE_STATIC);
                }
                /* TODO: handle NA's & NULL's */
            }
 
            res = sqlite3_step(stmt);
            if (res != SQLITE_DONE) { 
                sqlite3_finalize(stmt);
                Rprintf("ERROR: %s\n", sqlite3_errmsg(g_workspace));
                return R_NilValue; /* why? */
            }
        }
                
        sqlite3_finalize(stmt);
 
        /* create a new object representing the sdf */
        ret = _create_sdf_sexp(iname);
 
    } else {
        Rprintf("ERROR: unable to create a sqlite data frame.\n");
        ret = R_NilValue;
    }
        
    return ret;
}
 
SEXP sdf_get_names(SEXP sdf) {
    char *iname;
    iname = SDF_INAME(sdf);
    if (!USE_SDF1(iname, TRUE)) return R_NilValue;
 
    int len;
    len = sprintf(g_sql_buf[0], "select * from [%s].sdf_data;", iname);
 
    sqlite3_stmt *stmt;
    int res, i;
   
    res = sqlite3_prepare(g_workspace, g_sql_buf[0], len, &stmt, NULL);
    if (_sqlite_error(res)) return R_NilValue;
 
    SEXP ret;
 
    len = sqlite3_column_count(stmt)-1;
    PROTECT(ret = NEW_CHARACTER(len));
 
    for (i = 0; i < len; i++) {
        SET_STRING_ELT(ret, i, mkChar(sqlite3_column_name(stmt, i+1)));
    }
 
    sqlite3_finalize(stmt);
    UNPROTECT(1);
    return ret;
}
 
SEXP sdf_get_length(SEXP sdf) {
    char *iname;
    iname  = CHAR(STRING_ELT(_getListElement(sdf, "iname"),0));
    if (!USE_SDF1(iname, TRUE)) return R_NilValue;
 
    int len = sprintf(g_sql_buf[0], "select * from [%s].sdf_data;", iname);
 
    sqlite3_stmt *stmt;
    int res;
   
    res = sqlite3_prepare(g_workspace, g_sql_buf[0], len, &stmt, NULL);
    if (_sqlite_error(res)) return R_NilValue;
 
    SEXP ret;
 
    len = sqlite3_column_count(stmt)-1;
    PROTECT(ret = NEW_INTEGER(1));
    INTEGER(ret)[0] = len;
 
    sqlite3_finalize(stmt);
    UNPROTECT(1);
    return ret;
}
 
/* get row count */
SEXP sdf_get_row_count(SEXP sdf) {
    char *iname = CHAR(STRING_ELT(_getListElement(sdf, "iname"),0));
    if (!USE_SDF1(iname, TRUE)) return R_NilValue;
 
    int nrow;
 
    SEXP ret;
   
 
 
    nrow = _get_row_count2(iname, TRUE);
 
 
    if (nrow < 1) ret = R_NilValue;
    else {
        PROTECT(ret = NEW_INTEGER(1));
        INTEGER(ret)[0] = nrow;
        UNPROTECT(1);
    }
 
 
    return ret;
}
 
    
SEXP sdf_import_table(SEXP _filename, SEXP _name, SEXP _sep, SEXP _quote, 
        SEXP _rownames, SEXP _colnames) {
    char *filename = CHAR_ELT(_filename, 0);
    FILE *f = fopen(filename, "r");
 
    if (f == NULL) {
        Rprintf("Error: File %s does not exist.", filename);
        return R_NilValue;
    }
 
    /*char *sep = CHAR_ELT(_sep, 0), *quote = CHAR_ELT(_quote,0);
    char *name = CHAR_ELT(_name, 0);*/
 
    /* create the table */
    /* insert the stuffs */
    /* register to workspace */
 
    return R_NilValue;
}
 
SEXP sdf_get_index(SEXP sdf, SEXP row, SEXP col) {
    SEXP ret = R_NilValue;
    char *iname = SDF_INAME(sdf);
    if (!USE_SDF1(iname, TRUE)) return R_NilValue;
 
    sqlite3_stmt *stmt;
    int buflen = 0, idxlen, col_cnt, row_cnt, index;
    int i, j,res;
    int *col_indices, col_index_len, *dup_indices;
    int row_index_len = 0;
 
    sprintf(g_sql_buf[0], "select * from [%s].sdf_data ", iname);
    res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0);
    if (_sqlite_error(res)) return R_NilValue;
 
    buflen = sprintf(g_sql_buf[0], "select [row name]");
    idxlen = LENGTH(col);
    col_cnt = sqlite3_column_count(stmt) - 1;
 
    /*
     * PROCESS COLUMN INDEX
     */
    if (col == R_NilValue) {
        for (i = 1; i < col_cnt+1; i++) {
            buflen += sprintf(g_sql_buf[0]+buflen, ",[%s]", 
                    sqlite3_column_name(stmt, i)); 
        }
        _expand_buf(0, buflen+20+strlen(iname));
        buflen += sprintf(g_sql_buf[0]+buflen, " from [%s].sdf_data", iname);
        col_index_len = col_cnt;
        col_indices = (int *)R_alloc(idxlen, sizeof(int));
        dup_indices = (int *)R_alloc(idxlen, sizeof(int));
        for (i = 0; i < col_cnt; i++) { col_indices[i] = i; dup_indices[i] = 0; }
    } else if (col == R_NilValue || idxlen < 1) {
        sqlite3_finalize(stmt);
        return R_NilValue;
    } else if (IS_NUMERIC(col)) {
        col_indices = (int *)R_alloc(idxlen, sizeof(int));
        dup_indices = (int *)R_alloc(idxlen, sizeof(int));
        col_index_len = 0;
 
        for (i = 0; i < idxlen; i++) {
            /* no need to correct for 0 base because col 0 is [row name] */
            index = ((int) REAL(col)[i]); 
            if (index > col_cnt) {
                sqlite3_finalize(stmt);
                Rprintf("Error: undefined columns selected\n");
                return R_NilValue;
            } else if (index > 0) {
                buflen += sprintf(g_sql_buf[0]+buflen, ",[%s]", 
                        sqlite3_column_name(stmt,index));
                dup_indices[col_index_len] = 0;
                for (j = 0; j < col_index_len; j++) {
                    if (col_indices[j] == index) dup_indices[col_index_len]++;
                }
                col_indices[col_index_len++] = index;
            } else if (index < 0) {
                sqlite3_finalize(stmt);
                Rprintf("Error: negative indices not supported.\n");
                return R_NilValue;
            }
        }
 
        if (col_index_len == 0) {
            sqlite3_finalize(stmt);
            Rprintf("Error: no indices detected??\n");
            return R_NilValue;
        } else { 
            _expand_buf(0, buflen+20+strlen(iname));
            buflen += sprintf(g_sql_buf[0]+buflen, " from [%s].sdf_data", iname);
        }
    } else if (IS_INTEGER(col)) {
        /* identical logic with IS_NUMERIC, except that we don't have to cast
         * stuffs from SEXP col. the alternative is doing if idxlen times. */
        col_indices = (int *)R_alloc(idxlen, sizeof(int));
        dup_indices = (int *)R_alloc(idxlen, sizeof(int));
        col_index_len = 0;
 
        for (i = 0; i < idxlen; i++) {
            /* no need to correct for 0 base because col 0 is [row name] */
            index = INTEGER(col)[i];
            if (index > col_cnt) {
                sqlite3_finalize(stmt);
                Rprintf("Error: undefined columns selected\n");
                return R_NilValue;
            } else if (index > 0) {
                buflen += sprintf(g_sql_buf[0]+buflen, ",[%s]", 
                        sqlite3_column_name(stmt,index));
                dup_indices[col_index_len] = 0;
                for (j = 0; j < col_index_len; j++) {
                    if (col_indices[j] == index) dup_indices[col_index_len]++;
                }
                col_indices[col_index_len++] = index;
            } else if (index < 0) {
                sqlite3_finalize(stmt);
                Rprintf("Error: negative indices not supported.\n");
                return R_NilValue;
            }
        }
 
        if (col_index_len == 0) {
            sqlite3_finalize(stmt);
            Rprintf("Error: no indices detected??\n");
            return R_NilValue;
        } else { 
            _expand_buf(0, buflen+20+strlen(iname));
            buflen += sprintf(g_sql_buf[0]+buflen, " from [%s].sdf_data", iname);
        }
    } else if (IS_LOGICAL(col)) {
        /* recycling stuff, so max column output is the # of cols in the df */
        col_indices = (int *)R_alloc(col_cnt, sizeof(int));
        dup_indices = (int *)R_alloc(col_cnt, sizeof(int));
        col_index_len = 0;
        for (i = 0; i < col_cnt; i++) {
            if (LOGICAL(col)[i%idxlen]) {
                buflen += sprintf(g_sql_buf[0]+buflen, ",[%s]", 
                        sqlite3_column_name(stmt,i+1));
                dup_indices[col_index_len] = 0;
                col_indices[col_index_len++] = i;
            }
        }
 
        if (col_index_len == 0) {
            sqlite3_finalize(stmt);
            Rprintf("Warning: no column selected.\n");
            return R_NilValue;
        } else { 
            _expand_buf(0, buflen+20+strlen(iname));
            buflen += sprintf(g_sql_buf[0]+buflen, " from [%s].sdf_data", iname);
        }
    } else {
        sqlite3_finalize(stmt);
        Rprintf("Error: don't know how to handle column index.\n");
        return R_NilValue;
    }
 
    /* 
     * PROCESS ROW INDEX
     */
    idxlen = LENGTH(row);
    if (row == R_NilValue) {
        if (col_index_len == 1) {
            ret = sdf_get_variable(sdf, mkString(sqlite3_column_name(stmt,col_indices[0])));
            sqlite3_finalize(stmt);
            return ret;
        } if (col_index_len > 1) {
            /* create a new SDF, logic similar to sdf_create_sdf */
 
            /* find a new name. data<n> ? */
            char *iname2;
            const char *colname;
            int namelen, sql_len, sql_len2;
 
            iname2 = _create_sdf_skeleton1(R_NilValue, &namelen);
 
            /* create sdf_data table */
            sql_len = sprintf(g_sql_buf[1], "create table [%s].sdf_data ([row name] text", iname2);
            sql_len2 = 0;
 
            for (i = 0; i < col_index_len; i++) {
                colname = sqlite3_column_name(stmt, col_indices[i]);
                if (dup_indices[i] == 0) {
                    sql_len += sprintf(g_sql_buf[1]+sql_len, ", [%s] %s", colname,
                            sqlite3_column_decltype(stmt, col_indices[i]));
                } else {
                    /* un-duplicate col names by appending num, just like R */
                    sql_len += sprintf(g_sql_buf[1]+sql_len, ", [%s.%d] %s", colname,
                            dup_indices[i], 
                            sqlite3_column_decltype(stmt, col_indices[i]));
                }
 
                /* deal with possibly factor columns */
                if (_is_factor2(iname, "factor", colname)) {
                    /* found a factor column */
 
                    if (dup_indices[i] == 0)  {
                        _copy_factor_levels2("factor", iname, colname, iname2, colname);
                    } else {
                        sprintf(g_sql_buf[3], "%s.%d", colname, dup_indices[i]);
                        _copy_factor_levels2("factor", iname, colname, iname2, 
                                g_sql_buf[3]);
                    }
                }
 
                /* and deal with ordered factors too... life is hard, then you die */
                if (_is_factor2(iname, "ordered", colname)) {
                    
                    if (dup_indices[i] == 0)  {
                        _copy_factor_levels2("ordered", iname, colname, iname2, 
                                colname);
                    } else {
                        sprintf(g_sql_buf[3], "%s.%d", colname, dup_indices[i]);
                        _copy_factor_levels2("ordered", iname, colname, iname2, 
                                g_sql_buf[3]);
                    }
                }
                                
            }
 
            /* don't need it anymore */
            sqlite3_finalize(stmt);
 
            /* no table index created, that's for v2 I hope*/
            sql_len += sprintf(g_sql_buf[1]+sql_len, ");");
            res = _sqlite_exec(g_sql_buf[1]);
            if (_sqlite_error(res)) { Rprintf("here? %s\n", g_sql_buf[1]); return R_NilValue; }
 
            /* insert data (with row names). buf[0] contains a select */
            sprintf(g_sql_buf[1], "insert into [%s].sdf_data %s", iname2,
                    g_sql_buf[0]);
            res = _sqlite_exec(g_sql_buf[1]);
            if (_sqlite_error(res)) return R_NilValue;
 
            /* add new sdf to workspace */
            sprintf(g_sql_buf[0], "%s.db", iname2);
            res = _add_sdf1(g_sql_buf[0], iname2);
            if (_sqlite_error(res)) return R_NilValue;
 
            /* create SEXP for the SDF */
            ret = _create_sdf_sexp(iname2);
            return ret;
        } else { sqlite3_finalize(stmt); return R_NilValue; }
    } else if (row == R_NilValue || idxlen < 1) {
        sqlite3_finalize(stmt);
        return R_NilValue;
    } else if (IS_NUMERIC(row)) {
        sqlite3_finalize(stmt);
 
        sprintf(g_sql_buf[1], "[%s].sdf_data", iname);
        row_cnt = _get_row_count2(g_sql_buf[1], 0);
 
        /* append " limit ?,1" to the formed select statement */
        _expand_buf(0, buflen+10);
        buflen += sprintf(g_sql_buf[0]+buflen, " limit ?,1");
        /*Rprintf("sql [%d]: %s\n", buflen, g_sql_buf[0]); */
 
        index = ((int) REAL(row)[0]) - 1;
        if (idxlen < 0 && idxlen == 1) return R_NilValue;
 
        res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0);
        if (_sqlite_error(res)) return R_NilValue;
 
        sqlite3_bind_int(stmt, 1, index);
        res = sqlite3_step(stmt);
 
        /* create data frame */
        ret = _setup_df_sexp1(stmt, iname, col_index_len, idxlen, dup_indices);
        if (ret == R_NilValue) { sqlite3_finalize(stmt); return R_NilValue; }
 
        /* put data in it */
        if (index >= 0 && index < row_cnt) {
            _add_row_to_df(ret, stmt, row_index_len++, col_index_len);
        }
 
        for (i = 1; i < idxlen; i++) {
            sqlite3_reset(stmt);
            index = ((int) REAL(row)[i]) - 1;
            if (index >= 0 && index < row_cnt) {
                sqlite3_bind_int(stmt, 1, index);
                res = sqlite3_step(stmt);
                _add_row_to_df(ret, stmt, row_index_len++, col_index_len);
            }
        }
        sqlite3_finalize(stmt);
 
        /* shrink vectors */
        if (row_index_len < idxlen) {
            for (i = 0; i < col_index_len; i++) {
                SET_VECTOR_ELT(ret, i, 
                        _shrink_vector(VECTOR_ELT(ret, i), row_index_len));
            }
        }
 
        UNPROTECT(1); /* for the ret from _setup_df_sexp1 */
    } else if (IS_INTEGER(row)) {
        /* same stuff as with IS_INTEGER, except for setting of var index*/
        sqlite3_finalize(stmt);
 
        sprintf(g_sql_buf[1], "[%s].sdf_data", iname);
        row_cnt = _get_row_count2(g_sql_buf[1], 0);
 
        /* append " limit ?,1" to the formed select statement */
        _expand_buf(0, buflen+10);
        buflen += sprintf(g_sql_buf[0]+buflen, " limit ?,1");
 
        index = INTEGER(row)[0] - 1;
        if (idxlen < 0 && idxlen == 1) return R_NilValue;
 
        res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0);
        if (_sqlite_error(res)) return R_NilValue;
 
        sqlite3_bind_int(stmt, 1, index);
        res = sqlite3_step(stmt);
 
        /* create data frame */
        ret = _setup_df_sexp1(stmt, iname, col_index_len, idxlen, dup_indices);
        if (ret == R_NilValue) { sqlite3_finalize(stmt); return R_NilValue; }
 
        /* put data in it */
        if (index >= 0 && index < row_cnt) {
            _add_row_to_df(ret, stmt, row_index_len++, col_index_len);
        }
 
        for (i = 1; i < idxlen; i++) {
            sqlite3_reset(stmt);
            index = INTEGER(row)[i] - 1;
            if (index >= 0 && index < row_cnt) {
                sqlite3_bind_int(stmt, 1, index);
                res = sqlite3_step(stmt); 
                _add_row_to_df(ret, stmt, row_index_len++, col_index_len);
            }
        }
        sqlite3_finalize(stmt);
 
        /* shrink vectors */
        if (row_index_len < idxlen) {
            for (i = 0; i < col_index_len; i++) {
                SET_VECTOR_ELT(ret, i, 
                        _shrink_vector(VECTOR_ELT(ret, i), row_index_len));
            }
        }
        UNPROTECT(1); /* for the ret from _setup_df_sexp1 */
    } else if (IS_LOGICAL(row)) {
        /* */
        sqlite3_finalize(stmt);
        int est_row_cnt = 0;
 
        sprintf(g_sql_buf[1], "[%s].sdf_data", iname);
        row_cnt = _get_row_count2(g_sql_buf[1], 0);
 
        /* append " limit ?,1" to the formed select statement */
        _expand_buf(0, buflen+10);
        buflen += sprintf(g_sql_buf[0]+buflen, " limit ?,1");
 
        /* find if there is any TRUE element in the vector */
        for (i = 0; i < idxlen && i < row_cnt; i++) {
            if (LOGICAL(row)[i]) break;
        }
 
        if (i < idxlen && i < row_cnt) {
            est_row_cnt = (idxlen-i) * (row_cnt/idxlen);
            if (row_cnt%idxlen > i) est_row_cnt += (row_cnt%idxlen - i);
 
            res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0);
            if (_sqlite_error(res)) return R_NilValue;
 
            sqlite3_bind_int(stmt, 1, i);
            sqlite3_step(stmt);
            ret = _setup_df_sexp1(stmt, iname, col_index_len, est_row_cnt, dup_indices);
            _add_row_to_df(ret, stmt, row_index_len++, col_index_len);
            
            for (i++ ; i < row_cnt; i++) {
                if (LOGICAL(row)[i%idxlen]) {
                    sqlite3_reset(stmt);
                    sqlite3_bind_int(stmt, 1, i);
                    sqlite3_step(stmt);
                    _add_row_to_df(ret, stmt, row_index_len++, col_index_len);
                }
            }
        }
 
        sqlite3_finalize(stmt);
 
        /* shrink vectors */
        if (est_row_cnt > 0 && row_index_len < est_row_cnt) {
            for (i = 0; i < col_index_len; i++) {
                SET_VECTOR_ELT(ret, i, 
                        _shrink_vector(VECTOR_ELT(ret, i), row_index_len));
            }
        }
 
        UNPROTECT(1); /* for the ret from _setup_df_sexp1 */
    }
 
    if (ret != R_NilValue && col_index_len > 1) {
        SEXP class = mkString("data.frame");
        SET_CLASS(ret, class);
        _set_rownames2(ret);
    } else if (ret != R_NilValue && col_index_len == 1) {
        ret = VECTOR_ELT(ret, 0);
    }
 
    return ret;
}
 
SEXP sdf_rbind(SEXP sdf, SEXP data) {
    char *class = CHAR_ELT(GET_CLASS(data), 0);
    char *iname = SDF_INAME(sdf), *colname, *rowname;
    SEXP ret = R_NilValue, col, names, levels, rownames;
    int ncols, buflen, buflen2, i, j, res, nrows, *types;
    sqlite3_stmt *stmt;
    const char *type;
 
    if (strcmp(class,"data.frame") == 0) {
        /* check table names, types. variables may not be in same order, as
         * long as names and types are identical. */
        names = GET_NAMES(data);
        ncols = GET_LENGTH(names);
 
        buflen = sprintf(g_sql_buf[0], "select 1");
        for (i = 0; i < ncols; i++) {
            _expand_buf(0, buflen + 100);
            buflen += sprintf(g_sql_buf[0]+buflen, ",[%s]", CHAR_ELT(names, i));
        }
        _expand_buf(0, buflen + 100);
        sprintf(g_sql_buf[0]+buflen, " from [%s].sdf_data", iname);
 
        res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, NULL);
        if (res != SQLITE_OK) { /* column name mismatch */
            Rprintf("Error: column names should exactly match the names of the SDF.\n");
            return ret;
        }
 
        /* now check the type */
        types = (int *)R_alloc(ncols, sizeof(int));
        for (i = 0; i< ncols; i++) {
 
            type = sqlite3_column_decltype(stmt, i+1); /* +1 bec 1st col is "1" */
 
 
 
 
 
 
 
 
 
            col = VECTOR_ELT(data, i);
            class = CHAR_ELT(GET_CLASS(col),0);
 
 
 
 
 
 
 
 
            if (strcmp(type, "double") == 0 && IS_NUMERIC(col)) types[i] = SQLITE_FLOAT;
            else if (strcmp(type, "text") == 0 && IS_CHARACTER(col)) types[i] = SQLITE_TEXT;
            else if (strcmp(type, "int") == 0) { 
                types[i] = SQLITE_INTEGER;
                if (isUnordered(col)) {  /* test if factor */
                    colname = CHAR_ELT(names, i);
                    sqlite3_stmt *stmt2;
                    if (!_is_factor2(iname, "factor", colname)) break;
                    sprintf(g_sql_buf[1], "[%s].[factor %s]", iname, colname);
                    nrows = _get_row_count2(g_sql_buf[1], 0);
 
                    /* levels in sdf must be >= levels in df */
                    levels = GET_LEVELS(col);
                    if (nrows < GET_LENGTH(levels)) {
                        Rprintf("Error: The data frame variable %s has more levels than"
                                " its SDF counterpart.\n", colname);
                        break;
                    }
 
                    sprintf(g_sql_buf[2], "select label from %s order by level", g_sql_buf[1]);
                    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt2, NULL);
                    _sqlite_error(res);
                    
                    for (j = 0; j < GET_LENGTH(levels); j++)  {
                        sqlite3_step(stmt2);
                        if (strcmp((char *)sqlite3_column_text(stmt2, 0), CHAR_ELT(levels, j)) != 0) {
                            Rprintf("Error: Level label mismatch in variable %s, [%s] in data frame"
                                    " vs [%s] in SDF.\n", colname, (char *)sqlite3_column_text(stmt2, 0),
                                    CHAR_ELT(levels, j));
                            sqlite3_finalize(stmt2);
                            break;
                        }
                    }
 
                    sqlite3_finalize(stmt2);
 
                } else if (isOrdered(col)) { /* same with ordered */
                    colname = CHAR_ELT(names, i);
                    sqlite3_stmt *stmt2;
                    if (!_is_factor2(iname, "ordered", colname)) break;
                    sprintf(g_sql_buf[1], "[%s].[ordered %s]", iname, colname);
                    nrows = _get_row_count2(g_sql_buf[1], 0);
 
                    /* levels in sdf must be >= levels in df */
                    levels = GET_LEVELS(col);
                    if (nrows < GET_LENGTH(levels)) {
                        Rprintf("Error: The data frame variable %s has more levels than"
                                " its SDF counterpart.\n", colname);
                        break;
                    }
 
                    sprintf(g_sql_buf[2], "select label from %s order by level", g_sql_buf[1]);
                    sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt2, NULL);
                    
                    for (j = 0; j < GET_LENGTH(levels); j++)  {
                        sqlite3_step(stmt2);
                        if (strcmp((char *)sqlite3_column_text(stmt2, 0), CHAR_ELT(levels, j)) != 0) {
                            Rprintf("Error: Level label mismatch in variable %s, [%s] in data frame"
                                    " vs [%s] in SDF.\n", colname, (char *)sqlite3_column_text(stmt2, 0),
                                    CHAR_ELT(levels, j));
                            sqlite3_finalize(stmt2);
                            break;
                        }
                    }
 
                    sqlite3_finalize(stmt2);
 
                }  /* else if class == "ordered" */
            } /* if integer */
        } /* for loop for column type checking */
 
        sqlite3_finalize(stmt);
        if (i < ncols) {
            Rprintf("Error: type mismatch at variable %s\n", CHAR_ELT(names, i));
            return ret;
        }
 
        /* create the insert statement */
        buflen = sprintf(g_sql_buf[0], "insert into [%s].sdf_data([row name]", iname);
        buflen2 = sprintf(g_sql_buf[1], "values(?");
        for (i = 0; i < ncols; i++) {
            _expand_buf(0, buflen+100);
            buflen += sprintf(g_sql_buf[0]+buflen, ",[%s]", CHAR_ELT(names, i));
            _expand_buf(1, buflen2+5);
            buflen2 += sprintf(g_sql_buf[1]+buflen2, ",?");
        }
 
        _expand_buf(2, buflen + buflen2 + 10);
        sprintf(g_sql_buf[2], "%s) %s)", g_sql_buf[0], g_sql_buf[1]);
        Rprintf("%s\n", g_sql_buf[2]);
        res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, NULL);
        if (_sqlite_error(res)) return ret;
 
        /* finally, add rows */
        nrows = LENGTH(GET_ROWNAMES(data));
        rownames = GET_ROWNAMES(data);
 
        for (i = 0; i < nrows; i++) {
            sqlite3_bind_text(stmt, 1, CHAR_ELT(rownames, i), -1, SQLITE_STATIC);
            for (j = 0; j < ncols; j++) {
                col = VECTOR_ELT(data, j);
                switch(types[j]) {
                    case SQLITE_FLOAT:
                        sqlite3_bind_double(stmt, j+2, REAL(col)[i]); break;
                    case SQLITE_TEXT:
                        sqlite3_bind_text(stmt, j+2, CHAR_ELT(col, i), -1, SQLITE_STATIC); break;
                    case SQLITE_INTEGER:
                    default:  /* runtime error if we're wrong */
                        sqlite3_bind_int(stmt, j+2, INTEGER(col)[i]); break;
                }
            }
 
            res = sqlite3_step(stmt);
            if (res != SQLITE_DONE) { /* row name pk conflict */
                rowname = CHAR_ELT(rownames, i);
                for (j = 1; res != SQLITE_DONE; j++) { /* _normally_, j shouldn't overflow */
                    sqlite3_reset(stmt);
                    sprintf(g_sql_buf[2], "%s-%d", rowname, j);
                    sqlite3_bind_text(stmt, 1, g_sql_buf[2], -1, SQLITE_STATIC);
                    res = sqlite3_step(stmt);
                }
            }
            sqlite3_reset(stmt);
        }
 
        sqlite3_finalize(stmt);
        ret = sdf;
    } else if (strcmp(class,"sqlite.data.frame") == 0) {
    }
 
    return ret;
}
 
 
SEXP sdf_get_iname(SEXP sdf) {
    return _getListElement(sdf, "iname");
}
 
SEXP sopen(SEXP name) {
    char *filename;
    
    if (IS_CHARACTER(name)) {
        filename = CHAR(STRING_ELT(name,0));
        Rprintf("%s\n", filename);
    }
    /* sqlite3 *db;
    int res = sqlite3_open(filename, &db); */
    SEXP ret;
    PROTECT(ret = NEW_LOGICAL(1));
    LOGICAL(ret)[0] = IS_CHARACTER(name);
    /* sqlite3_close(db); */ 
    UNPROTECT(1);
    return ret;
}

Rev 3324	Rev 3358
1	`#include <stdio.h>`	1	`#include <stdio.h>`
2	`#include <string.h>`	2	`#include <string.h>`
3	`#include "sqlite_dataframe.h"`	3	`#include "sqlite_dataframe.h"`
4		4
5	`/****************************************************************************`	5	`/****************************************************************************`
6	`* UTILITY FUNCTIONS`	6	`* UTILITY FUNCTIONS`
7	`****************************************************************************/`	7	`****************************************************************************/`
8		8
9	`/* if user supplied a name, return that name. otherwise, use the`	9	`/* if user supplied a name, return that name. otherwise, use the`
10	`* default "data" */`	10	`* default "data" */`
11	`static int _check_sdf_name(SEXP name, char rname, char iname, int *file_idx) {`	11	`static int _check_sdf_name(SEXP name, char rname, char iname, int *file_idx) {`
12	`int namelen = 0;`	12	`int namelen = 0;`
13		13
14	`/* check if arg name is supplied */`	14	`/* check if arg name is supplied */`
15	`if (IS_CHARACTER(name)) {`	15	`if (IS_CHARACTER(name)) {`
16	`*rname = CHAR_ELT(name,0);`	16	`*rname = CHAR_ELT(name,0);`
17	`if (!_is_r_sym(*rname)) {`	17	`if (!_is_r_sym(*rname)) {`
18	`Rprintf("Error: supplied name \"%s\"is not a valid R symbol.\n",`	18	`Rprintf("Error: supplied name \"%s\"is not a valid R symbol.\n",`
19	`*rname);`	19	`*rname);`
20	`return FALSE;`	20	`return FALSE;`
21	`}`	21	`}`
22	`namelen = strlen(*rname);`	22	`namelen = strlen(*rname);`
23	`iname = (char)R_alloc(namelen + 9, sizeof(char)); /* <name>10000.db\0 */`	23	`iname = (char)R_alloc(namelen + 9, sizeof(char)); /* <name>10000.db\0 */`
24	`sprintf(iname, "%s.db", rname);`	24	`sprintf(iname, "%s.db", rname);`
25	`} else if (name == R_NilValue) {`	25	`} else if (name == R_NilValue) {`
26	`*rname = "data";`	26	`*rname = "data";`
27	`namelen = 5;`	27	`namelen = 5;`
28	`iname = (char)R_alloc(13, sizeof(char)); /* data10000.db\0 */`	28	`iname = (char)R_alloc(13, sizeof(char)); /* data10000.db\0 */`
29	`*file_idx = 1;`	29	`*file_idx = 1;`
30	`sprintf(iname, "%s%d.db", rname, *file_idx);`	30	`sprintf(iname, "%s%d.db", rname, *file_idx);`
31	`} else {`	31	`} else {`
32	`Rprintf("Error: the supplied value for arg name is not a string.\n");`	32	`Rprintf("Error: the supplied value for arg name is not a string.\n");`
33	`}`	33	`}`
34	`return namelen;`	34	`return namelen;`
35	`}`	35	`}`
36		36
37	`static int _find_free_filename(char rname, char iname, int namelen, int *file_idx) {`	37	`static int _find_free_filename(char rname, char iname, int namelen, int *file_idx) {`
38	`do {`	38	`do {`
39	`if (!_file_exists(*iname)) break;`	39	`if (!_file_exists(*iname)) break;`
40	`namelen = sprintf(iname, "%s%d.db", rname, ++(*file_idx)) - 3;`	40	`namelen = sprintf(iname, "%s%d.db", rname, ++(*file_idx)) - 3;`
41	`} while (*file_idx < 10000);`	41	`} while (*file_idx < 10000);`
42		42
43	`return *file_idx;`	43	`return *file_idx;`
44	`}`	44	`}`
45		45
46	`/* creates an sdf attribute table */`	46	`/* creates an sdf attribute table */`
47	`static int _create_sdf_attribute2(const char *iname) {`	47	`static int _create_sdf_attribute2(const char *iname) {`
48	`sprintf(g_sql_buf[2], "create table [%s].sdf_attributes(attr text, "`	48	`sprintf(g_sql_buf[2], "create table [%s].sdf_attributes(attr text, "`
49	`"value text, primary key (attr))", iname);`	49	`"value text, primary key (attr))", iname);`
50	`int res;`	50	`int res;`
51	`res = _sqlite_exec(g_sql_buf[2]);`	51	`res = _sqlite_exec(g_sql_buf[2]);`
52	`if (res == SQLITE_OK) {`	52	`if (res == SQLITE_OK) {`
53	`sprintf(g_sql_buf[2], "insert into [%s].sdf_attributes values ('name',"`	53	`sprintf(g_sql_buf[2], "insert into [%s].sdf_attributes values ('name',"`
54	`"'%s');", iname, iname);`	54	`"'%s');", iname, iname);`
55	`res = _sqlite_exec(g_sql_buf[2]);`	55	`res = _sqlite_exec(g_sql_buf[2]);`
56	`}`	56	`}`
57	`return res;`	57	`return res;`
58	`}`	58	`}`
59		59
60	`char _create_sdf_skeleton2(SEXP name, int onamelen) {`	60	`char _create_sdf_skeleton1(SEXP name, int onamelen) {`
61	`char iname, rname;`	61	`char iname, rname;`
62	`int namelen, file_idx = 0, res;`	62	`int namelen, file_idx = 0, res;`
63		63
64	`namelen = _check_sdf_name(name, &rname, &iname, &file_idx);`	64	`namelen = _check_sdf_name(name, &rname, &iname, &file_idx);`
65		65
66	`if (!namelen) return NULL;`	66	`if (!namelen) return NULL;`
67		67
68	`_find_free_filename(rname, &iname, &namelen, &file_idx);`	68	`_find_free_filename(rname, &iname, &namelen, &file_idx);`
69		69
70	`if (file_idx >= 10000) {`	70	`if (file_idx >= 10000) {`
71	`Rprintf("Error: cannot find free SDF name.\n");`	71	`Rprintf("Error: cannot find free SDF name.\n");`
72	`return NULL;`	72	`return NULL;`
73	`}`	73	`}`
74		74
75	`/* add to workspace */`	75	`/* add to workspace */`
76	`strcpy(g_sql_buf[3], iname);`	76	`strcpy(g_sql_buf[3], iname);`
77	`iname[namelen] = 0; /* remove ".db" */`	77	`iname[namelen] = 0; /* remove ".db" */`
78	`res = _add_sdf1(g_sql_buf[3], iname);`	78	`res = _add_sdf1(g_sql_buf[3], iname);`
79	`if (_sqlite_error(res)) return NULL;`	79	`if (_sqlite_error(res)) return NULL;`
80		80
81	`/* detach SDF's if necessary to attach this one. if file does not`	81	`/* detach SDF's if necessary to attach this one. if file does not`
82	`* exist, then a sqlite db will be created after we make our 1st table */`	82	`* exist, then a sqlite db will be created after we make our 1st table */`
83	`if (!USE_SDF(iname, FALSE)) { /* _delete_sdf2(iname); */ return NULL; }`	83	`if (!USE_SDF1(iname, FALSE)) { /* _delete_sdf2(iname); */ return NULL; }`
84		84
85		85
86	`/* create attributes table */`	86	`/* create attributes table */`
87	`res = _create_sdf_attribute2(iname);`	87	`res = _create_sdf_attribute2(iname);`
88	`if (_sqlite_error(res)) {`	88	`if (_sqlite_error(res)) {`
89	`sprintf(g_sql_buf[2], "detach '%s'", iname);`	89	`sprintf(g_sql_buf[2], "detach '%s'", iname);`
90	`_delete_sdf2(iname);`	90	`_delete_sdf2(iname);`
91	`return NULL;`	91	`return NULL;`
92	`}`	92	`}`
93		93
94	`*onamelen = namelen;`	94	`*onamelen = namelen;`
95	`return iname;`	95	`return iname;`
96	`}`	96	`}`
97	`/* checks if a column has a corresponding factor\|ordered table */`	97	`/* checks if a column has a corresponding factor\|ordered table */`
98	`static int _is_factor2(const char iname, const char factor_type, const char *colname) {`	98	`static int _is_factor2(const char iname, const char factor_type, const char *colname) {`
99	`sqlite3_stmt *stmt;`	99	`sqlite3_stmt *stmt;`
100	`sprintf(g_sql_buf[2], "select * from [%s].[%s %s]", iname, factor_type,`	100	`sprintf(g_sql_buf[2], "select * from [%s].[%s %s]", iname, factor_type,`
101	`colname);`	101	`colname);`
102	`int res;`	102	`int res;`
103	`res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);`	103	`res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);`
104	`sqlite3_finalize(stmt);`	104	`sqlite3_finalize(stmt);`
105	`return res == SQLITE_OK;`	105	`return res == SQLITE_OK;`
106	`}`	106	`}`
107		107
108	`/* create a factor\|ordered table */`	108	`/* create a factor\|ordered table */`
109	`static int _create_factor_table2(const char iname, const char factor_type,`	109	`static int _create_factor_table2(const char iname, const char factor_type,`
110	`const char *colname) {`	110	`const char *colname) {`
111	`sqlite3_stmt *stmt;`	111	`sqlite3_stmt *stmt;`
112	`sprintf(g_sql_buf[2], "create table [%s].[%s %s] (level int, label text, "`	112	`sprintf(g_sql_buf[2], "create table [%s].[%s %s] (level int, label text, "`
113	`"primary key(level), unique(label));", iname, factor_type, colname);`	113	`"primary key(level), unique(label));", iname, factor_type, colname);`
114	`int res;`	114	`int res;`
115	`res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);`	115	`res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);`
116	`if (res == SQLITE_OK) sqlite3_step(stmt);`	116	`if (res == SQLITE_OK) sqlite3_step(stmt);`
117	`sqlite3_finalize(stmt);`	117	`sqlite3_finalize(stmt);`
118	`return res; /* error on dup name? */`	118	`return res; /* error on dup name? */`
119	`}`	119	`}`
120		120
121	`/* copy a factor\|ordered table from a sdf(db) to another sdf(db) */`	121	`/* copy a factor\|ordered table from a sdf(db) to another sdf(db) */`
122	`static int _copy_factor_levels2(const char factor_type, const char iname_src,`	122	`int _copy_factor_levels2(const char factor_type, const char iname_src,`
123	`const char colname_src, const char iname_dst, const char *colname_dst) {`	123	`const char colname_src, const char iname_dst, const char *colname_dst) {`
124	`sqlite3_stmt *stmt;`	124	`sqlite3_stmt *stmt;`
125	`int res;`	125	`int res;`
126	`res = _create_factor_table2(iname_dst, factor_type, colname_dst);`	126	`res = _create_factor_table2(iname_dst, factor_type, colname_dst);`
127	`if (res == SQLITE_OK) {`	127	`if (res == SQLITE_OK) {`
128	`sprintf(g_sql_buf[2], "insert into [%s].[%s %s] select * from [%s].[%s %s]",`	128	`sprintf(g_sql_buf[2], "insert into [%s].[%s %s] select * from [%s].[%s %s]",`
129	`iname_src, factor_type, colname_src, iname_dst, factor_type,`	129	`iname_src, factor_type, colname_src, iname_dst, factor_type,`
130	`colname_dst);`	130	`colname_dst);`
131	`res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);`	131	`res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);`
132	`if (res == SQLITE_OK) sqlite3_step(stmt);`	132	`if (res == SQLITE_OK) sqlite3_step(stmt);`
133	`sqlite3_finalize(stmt);`	133	`sqlite3_finalize(stmt);`
134	`}`	134	`}`
135	`return res; /* error on dup name? */`	135	`return res; /* error on dup name? */`
136	`}`	136	`}`
137		137
138		138
139	`/* assumes stmt has col_cnt + 1 columns, col 0 is [row name]. returned SEXP is`	139	`/* assumes stmt has col_cnt + 1 columns, col 0 is [row name]. returned SEXP is`
140	`* not UNPROTECT-ed, user will have to do that. will create the names &`	140	`* not UNPROTECT-ed, user will have to do that. will create the names &`
141	`* attach factor infos */`	141	`* attach factor infos */`
142	`static SEXP _setup_df_sexp1(sqlite3_stmt stmt, const char iname,`	142	`static SEXP _setup_df_sexp1(sqlite3_stmt stmt, const char iname,`
143	`int col_cnt, int row_cnt, int *dup_indices) {`	143	`int col_cnt, int row_cnt, int *dup_indices) {`
144	`SEXP ret, names, value, class = R_NilValue;`	144	`SEXP ret, names, value, class = R_NilValue;`
145	`int i, type;`	145	`int i, type;`
146	`const char colname, coltype;`	146	`const char colname, coltype;`
147		147
148	`PROTECT(ret = NEW_LIST(col_cnt));`	148	`PROTECT(ret = NEW_LIST(col_cnt));`
149		149
150	`/* set up names. */`	150	`/* set up names. */`
151	`PROTECT(names = NEW_CHARACTER(col_cnt));`	151	`PROTECT(names = NEW_CHARACTER(col_cnt));`
152	`for (i = 0; i < col_cnt; i++) {`	152	`for (i = 0; i < col_cnt; i++) {`
153	`colname = sqlite3_column_name(stmt, i+1); /* +1 bec [row name is 0] */`	153	`colname = sqlite3_column_name(stmt, i+1); /* +1 bec [row name is 0] */`
154	`coltype = sqlite3_column_decltype(stmt, i+1);`	154	`coltype = sqlite3_column_decltype(stmt, i+1);`
155		155
156	`if (dup_indices[i] == 0) {`	156	`if (dup_indices[i] == 0) {`
157	`strcpy(g_sql_buf[1], colname);`	157	`strcpy(g_sql_buf[1], colname);`
158	`} else {`	158	`} else {`
159	`sprintf(g_sql_buf[1], "%s.%d", colname, dup_indices[i]);`	159	`sprintf(g_sql_buf[1], "%s.%d", colname, dup_indices[i]);`
160	`}`	160	`}`
161		161
162	`SET_STRING_ELT(names, i, mkChar(g_sql_buf[1]));`	162	`SET_STRING_ELT(names, i, mkChar(g_sql_buf[1]));`
163		163
164	`if (strcmp(coltype, "text") == 0) {`	164	`if (strcmp(coltype, "text") == 0) {`
165	`PROTECT(value = NEW_CHARACTER(row_cnt));`	165	`PROTECT(value = NEW_CHARACTER(row_cnt));`
166	`} else if (strcmp(coltype, "double") == 0) {`	166	`} else if (strcmp(coltype, "double") == 0) {`
167	`PROTECT(value = NEW_NUMERIC(row_cnt));`	167	`PROTECT(value = NEW_NUMERIC(row_cnt));`
168	`} else if (strcmp(coltype, "bit") == 0) {`	168	`} else if (strcmp(coltype, "bit") == 0) {`
169	`PROTECT(value = NEW_LOGICAL(row_cnt));`	169	`PROTECT(value = NEW_LOGICAL(row_cnt));`
170	`} else if (strcmp(coltype, "integer") == 0 \|\|`	170	`} else if (strcmp(coltype, "integer") == 0 \|\|`
171	`strcmp(coltype, "int") == 0) {`	171	`strcmp(coltype, "int") == 0) {`
172	`PROTECT(value = NEW_INTEGER(row_cnt));`	172	`PROTECT(value = NEW_INTEGER(row_cnt));`
173	`type = _get_factor_levels1(iname, colname, value);`	173	`type = _get_factor_levels1(iname, colname, value);`
174	`if (type == VAR_FACTOR) {`	174	`if (type == VAR_FACTOR) {`
175	`PROTECT(class = mkString("factor"));`	175	`PROTECT(class = mkString("factor"));`
176	`SET_CLASS(value, class);`	176	`SET_CLASS(value, class);`
177	`UNPROTECT(1);`	177	`UNPROTECT(1);`
178	`} else if (type == VAR_ORDERED) {`	178	`} else if (type == VAR_ORDERED) {`
179	`PROTECT(class = NEW_CHARACTER(2));`	179	`PROTECT(class = NEW_CHARACTER(2));`
180	`SET_STRING_ELT(class, 0, mkChar("ordered"));`	180	`SET_STRING_ELT(class, 0, mkChar("ordered"));`
181	`SET_STRING_ELT(class, 1, mkChar("factor"));`	181	`SET_STRING_ELT(class, 1, mkChar("factor"));`
182	`SET_CLASS(value, class);`	182	`SET_CLASS(value, class);`
183	`UNPROTECT(1);`	183	`UNPROTECT(1);`
184	`}`	184	`}`
185	`} else {`	185	`} else {`
186	`Rprintf("Error: not supported type %s for %s\n", coltype, colname);`	186	`Rprintf("Error: not supported type %s for %s\n", coltype, colname);`
187	`UNPROTECT(2); /* unprotect ret, names */`	187	`UNPROTECT(2); /* unprotect ret, names */`
188	`return R_NilValue;`	188	`return R_NilValue;`
189	`}`	189	`}`
190		190
191	`SET_VECTOR_ELT(ret, i, value);`	191	`SET_VECTOR_ELT(ret, i, value);`
192	`UNPROTECT(1); /* unprotect value */`	192	`UNPROTECT(1); /* unprotect value */`
193		193
194	`}`	194	`}`
195	`SET_NAMES(ret, names);`	195	`SET_NAMES(ret, names);`
196	`UNPROTECT(1); /* unprotect names only */`	196	`UNPROTECT(1); /* unprotect names only */`
197	`return ret;`	197	`return ret;`
198	`}`	198	`}`
199		199
200	`/* expected that 1st col of stmt is [row name], so we start with index 1 */`	200	`/* expected that 1st col of stmt is [row name], so we start with index 1 */`
201	`static int _add_row_to_df(SEXP df, sqlite3_stmt *stmt, int row, int ncols) {`	201	`static int _add_row_to_df(SEXP df, sqlite3_stmt *stmt, int row, int ncols) {`
202	`SEXP vec;`	202	`SEXP vec;`
203	`int type = -1, is_null, i;`	203	`int type = -1, is_null, i;`
204	`for (i = 1; i <= ncols; i++) {`	204	`for (i = 1; i <= ncols; i++) {`
205	`vec = VECTOR_ELT(df, i-1);`	205	`vec = VECTOR_ELT(df, i-1);`
206	`type = TYPEOF(vec);`	206	`type = TYPEOF(vec);`
207		207
208	`is_null = (sqlite3_column_type(stmt, row) == SQLITE_NULL);`	208	`is_null = (sqlite3_column_type(stmt, row) == SQLITE_NULL);`
209		209
210	`if (type == CHARSXP) {`	210	`if (type == CHARSXP) {`
211	`SET_STRING_ELT(vec, row, mkChar((char *)sqlite3_column_text(stmt, i)));`	211	`SET_STRING_ELT(vec, row, mkChar((char *)sqlite3_column_text(stmt, i)));`
212	`} else if (type == INTSXP) {`	212	`} else if (type == INTSXP) {`
213	`INTEGER(vec)[row] = sqlite3_column_int(stmt, i);`	213	`INTEGER(vec)[row] = sqlite3_column_int(stmt, i);`
214	`} else if (type == REALSXP) {`	214	`} else if (type == REALSXP) {`
215	`REAL(vec)[row] = sqlite3_column_double(stmt, i);`	215	`REAL(vec)[row] = sqlite3_column_double(stmt, i);`
216	`} else if (type == LGLSXP) {`	216	`} else if (type == LGLSXP) {`
217	`LOGICAL(vec)[row] = sqlite3_column_int(stmt, i);`	217	`LOGICAL(vec)[row] = sqlite3_column_int(stmt, i);`
218	`}`	218	`}`
219	`}`	219	`}`
220	`return type;`	220	`return type;`
221	`}`	221	`}`
222		222
223	`static void _set_rownames2(SEXP df) {`	223	`static void _set_rownames2(SEXP df) {`
224	`SEXP value = VECTOR_ELT(df, 0);`	224	`SEXP value = VECTOR_ELT(df, 0);`
225	`int len = LENGTH(value), i;`	225	`int len = LENGTH(value), i;`
226	`PROTECT(value = NEW_CHARACTER(len));`	226	`PROTECT(value = NEW_CHARACTER(len));`
227	`for (i = 0; i < len; i++) {`	227	`for (i = 0; i < len; i++) {`
228	`sprintf(g_sql_buf[2], "%d", i+1);`	228	`sprintf(g_sql_buf[2], "%d", i+1);`
229	`SET_STRING_ELT(value, i, mkChar(g_sql_buf[2]));`	229	`SET_STRING_ELT(value, i, mkChar(g_sql_buf[2]));`
230	`}`	230	`}`
231		231
232	`SET_ROWNAMES(df, value);`	232	`SET_ROWNAMES(df, value);`
233	`UNPROTECT(1);`	233	`UNPROTECT(1);`
234	`}`	234	`}`
235		235
236		236
237	`/****************************************************************************`	237	`/****************************************************************************`
238	`* SDF FUNCTIONS`	238	`* SDF FUNCTIONS`
239	`****************************************************************************/`	239	`****************************************************************************/`
240		240
241	`SEXP sdf_create_sdf(SEXP df, SEXP name) {`	241	`SEXP sdf_create_sdf(SEXP df, SEXP name) {`
242	`SEXP ret;`	242	`SEXP ret;`
243	`char *iname;`	243	`char *iname;`
244	`int namelen, res, i, j;`	244	`int namelen, res, i, j;`
245		245
246	`/* find free name, attach sdf, create sdf_attributes */`	246	`/* find free name, attach sdf, create sdf_attributes */`
247	`iname = _create_sdf_skeleton2(name, &namelen);`	247	`iname = _create_sdf_skeleton1(name, &namelen);`
248		248
249	`if (iname != NULL) {`	249	`if (iname != NULL) {`
250	`int sql_len, sql_len2;`	250	`int sql_len, sql_len2;`
251	`sqlite3_stmt *stmt;`	251	`sqlite3_stmt *stmt;`
252		252
253	`/* create sdf_data table */`	253	`/* create sdf_data table */`
254	`SEXP names = GET_NAMES(df), variable, levels;`	254	`SEXP names = GET_NAMES(df), variable, levels;`
255	`int ncols = GET_LENGTH(names), type, *types;`	255	`int ncols = GET_LENGTH(names), type, *types;`
256	`char col_name, class, *factor;`	256	`char col_name, class, *factor;`
257		257
258	`/* variables for adding rownames */`	258	`/* variables for adding rownames */`
259	`SEXP rownames;`	259	`SEXP rownames;`
260	`int nrows;`	260	`int nrows;`
261	`char *row_name;`	261	`char *row_name;`
262		262
263	`/* TODO: put constraints on table after inserting everything? */`	263	`/* TODO: put constraints on table after inserting everything? */`
264		264
265		265
266	`/*`	266	`/*`
267	`* create the create table and insert sql scripts by looping through`	267	`* create the create table and insert sql scripts by looping through`
268	`* the columns of df`	268	`* the columns of df`
269	`*/`	269	`*/`
270	`types = (int *)R_alloc(ncols, sizeof(int));`	270	`types = (int *)R_alloc(ncols, sizeof(int));`
271	`sql_len = sprintf(g_sql_buf[0], "create table [%s].sdf_data ([row name] text", iname);`	271	`sql_len = sprintf(g_sql_buf[0], "create table [%s].sdf_data ([row name] text", iname);`
272	`sql_len2 = sprintf(g_sql_buf[1], "insert into [%s].sdf_data values(?", iname);`	272	`sql_len2 = sprintf(g_sql_buf[1], "insert into [%s].sdf_data values(?", iname);`
273		273
274	`for (i = 0; i < ncols; i++) {`	274	`for (i = 0; i < ncols; i++) {`
275	`col_name = CHAR(STRING_ELT(names, i));`	275	`col_name = CHAR(STRING_ELT(names, i));`
276		276
277	`/* add column definition to the create table sql */`	277	`/* add column definition to the create table sql */`
278	`_expand_buf(0, sql_len+strlen(col_name)+10);`	278	`_expand_buf(0, sql_len+strlen(col_name)+10);`
279		279
280	`variable = _getListElement(df, col_name);`	280	`variable = _getListElement(df, col_name);`
281	`class = CHAR(STRING_ELT(GET_CLASS(variable), 0));`	281	`class = CHAR(STRING_ELT(GET_CLASS(variable), 0));`
282	`type = TYPEOF(variable);`	282	`type = TYPEOF(variable);`
283	`types[i] = type;`	283	`types[i] = type;`
284		284
285	`sql_len += sprintf(g_sql_buf[0]+sql_len, ", [%s] %s", col_name,`	285	`sql_len += sprintf(g_sql_buf[0]+sql_len, ", [%s] %s", col_name,`
286	`_get_column_type(class, type));`	286	`_get_column_type(class, type));`
287		287
288	`/* add handler to insert table sql */`	288	`/* add handler to insert table sql */`
289	`_expand_buf(1, sql_len+5);`	289	`_expand_buf(1, sql_len+5);`
290	`strcpy(g_sql_buf[1]+sql_len2, ",?");`	290	`strcpy(g_sql_buf[1]+sql_len2, ",?");`
291	`sql_len2 += 2;`	291	`sql_len2 += 2;`
292		292
293	`/* create separate table for factors decode */`	293	`/* create separate table for factors decode */`
294	`if (strcmp(class, "factor") == 0 \|\| strcmp(class, "ordered") == 0){`	294	`if (strcmp(class, "factor") == 0 \|\| strcmp(class, "ordered") == 0){`
295	`if (_create_factor_table2(iname, class, col_name))`	295	`if (_create_factor_table2(iname, class, col_name))`
296	`return R_NilValue; /* dup tbl name? */`	296	`return R_NilValue; /* dup tbl name? */`
297		297
298	`levels = GET_LEVELS(variable);`	298	`levels = GET_LEVELS(variable);`
299	`sprintf(g_sql_buf[2], "insert into [%s].[%s %s] values(?, ?);",`	299	`sprintf(g_sql_buf[2], "insert into [%s].[%s %s] values(?, ?);",`
300	`iname, class, col_name);`	300	`iname, class, col_name);`
301	`res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, NULL);`	301	`res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, NULL);`
302	`if (_sqlite_error(res)) return R_NilValue; /* dup tbl name? */`	302	`if (_sqlite_error(res)) return R_NilValue; /* dup tbl name? */`
303		303
304	`for (j = 0; j < GET_LENGTH(levels); j++) {`	304	`for (j = 0; j < GET_LENGTH(levels); j++) {`
305	`sqlite3_reset(stmt);`	305	`sqlite3_reset(stmt);`
306	`factor = CHAR(STRING_ELT(levels, j));`	306	`factor = CHAR(STRING_ELT(levels, j));`
307	`sqlite3_bind_int(stmt, 1, j+1);`	307	`sqlite3_bind_int(stmt, 1, j+1);`
308	`sqlite3_bind_text(stmt, 2, factor, -1, SQLITE_STATIC);`	308	`sqlite3_bind_text(stmt, 2, factor, -1, SQLITE_STATIC);`
309	`sqlite3_step(stmt);`	309	`sqlite3_step(stmt);`
310	`}`	310	`}`
311	`sqlite3_finalize(stmt);`	311	`sqlite3_finalize(stmt);`
312	`}`	312	`}`
313	`}`	313	`}`
314		314
315	`_expand_buf(0,sql_len+35);`	315	`_expand_buf(0,sql_len+35);`
316	`sql_len += sprintf(g_sql_buf[0]+sql_len, ", primary key([row name]));");`	316	`sql_len += sprintf(g_sql_buf[0]+sql_len, ", primary key([row name]));");`
317	`/* sql_len += sprintf(g_sql_buf[0]+sql_len, ");"); */`	317	`/* sql_len += sprintf(g_sql_buf[0]+sql_len, ");"); */`
318	`res = _sqlite_exec(g_sql_buf[0]);`	318	`res = _sqlite_exec(g_sql_buf[0]);`
319	`if (_sqlite_error(res)) return R_NilValue; /* why? */`	319	`if (_sqlite_error(res)) return R_NilValue; /* why? */`
320		320
321	`/*`	321	`/*`
322	`* add the data in df to the sdf`	322	`* add the data in df to the sdf`
323	`*/`	323	`*/`
324	`rownames = getAttrib(df, R_RowNamesSymbol);`	324	`rownames = getAttrib(df, R_RowNamesSymbol);`
325	`nrows = GET_LENGTH(rownames);`	325	`nrows = GET_LENGTH(rownames);`
326		326
327	`sprintf(g_sql_buf[1]+sql_len2, ")");`	327	`sprintf(g_sql_buf[1]+sql_len2, ")");`
328	`res = sqlite3_prepare(g_workspace, g_sql_buf[1], -1, &stmt, NULL);`	328	`res = sqlite3_prepare(g_workspace, g_sql_buf[1], -1, &stmt, NULL);`
329		329
330	`for (i = 0; i < nrows; i++) {`	330	`for (i = 0; i < nrows; i++) {`
331	`sqlite3_reset(stmt);`	331	`sqlite3_reset(stmt);`
332		332
333	`/* since this is coming from a real dataframe, we're assured that`	333	`/* since this is coming from a real dataframe, we're assured that`
334	`* the rownames are unique */`	334	`* the rownames are unique */`
335	`if (IS_CHARACTER(rownames)) {`	335	`if (IS_CHARACTER(rownames)) {`
336	`row_name = CHAR(STRING_ELT(rownames, i));`	336	`row_name = CHAR(STRING_ELT(rownames, i));`
337	`if (row_name) / if not empty string */`	337	`if (row_name) / if not empty string */`
338	`sqlite3_bind_text(stmt, 1, row_name, strlen(row_name), SQLITE_STATIC);`	338	`sqlite3_bind_text(stmt, 1, row_name, strlen(row_name), SQLITE_STATIC);`
339	`else sqlite3_bind_int(stmt, 1, i);`	339	`else sqlite3_bind_int(stmt, 1, i);`
340	`} else if (IS_INTEGER(rownames)) {`	340	`} else if (IS_INTEGER(rownames)) {`
341	`sqlite3_bind_int(stmt, 1, INTEGER(rownames)[i]);`	341	`sqlite3_bind_int(stmt, 1, INTEGER(rownames)[i]);`
342	`} else sqlite3_bind_int(stmt, 1, i);`	342	`} else sqlite3_bind_int(stmt, 1, i);`
343		343
344	`for (j = 0; j < ncols; j++) {`	344	`for (j = 0; j < ncols; j++) {`
345	`variable = VECTOR_ELT(df, j);`	345	`variable = VECTOR_ELT(df, j);`
346	`switch(types[j]) {`	346	`switch(types[j]) {`
347	`case INTSXP :`	347	`case INTSXP :`
348	`sqlite3_bind_int(stmt, j+2, INTEGER(variable)[i]);`	348	`sqlite3_bind_int(stmt, j+2, INTEGER(variable)[i]);`
349	`break;`	349	`break;`
350	`case REALSXP:`	350	`case REALSXP:`
351	`sqlite3_bind_double(stmt, j+2, REAL(variable)[i]);`	351	`sqlite3_bind_double(stmt, j+2, REAL(variable)[i]);`
352	`break;`	352	`break;`
353	`case CHARSXP:`	353	`case CHARSXP:`
354	`col_name = CHAR(STRING_ELT(variable,i));`	354	`col_name = CHAR(STRING_ELT(variable,i));`
355	`sqlite3_bind_text(stmt, j+2, col_name, strlen(col_name), SQLITE_STATIC);`	355	`sqlite3_bind_text(stmt, j+2, col_name, strlen(col_name), SQLITE_STATIC);`
356	`}`	356	`}`
357	`/* TODO: handle NA's & NULL's */`	357	`/* TODO: handle NA's & NULL's */`
358	`}`	358	`}`
359		359
360	`res = sqlite3_step(stmt);`	360	`res = sqlite3_step(stmt);`
361	`if (res != SQLITE_DONE) {`	361	`if (res != SQLITE_DONE) {`
362	`sqlite3_finalize(stmt);`	362	`sqlite3_finalize(stmt);`
363	`Rprintf("ERROR: %s\n", sqlite3_errmsg(g_workspace));`	363	`Rprintf("ERROR: %s\n", sqlite3_errmsg(g_workspace));`
364	`return R_NilValue; /* why? */`	364	`return R_NilValue; /* why? */`
365	`}`	365	`}`
366	`}`	366	`}`
367		367
368	`sqlite3_finalize(stmt);`	368	`sqlite3_finalize(stmt);`
369		369
370	`/* create a new object representing the sdf */`	370	`/* create a new object representing the sdf */`
371	`ret = _create_sdf_sexp(iname);`	371	`ret = _create_sdf_sexp(iname);`
372		372
373	`} else {`	373	`} else {`
374	`Rprintf("ERROR: unable to create a sqlite data frame.\n");`	374	`Rprintf("ERROR: unable to create a sqlite data frame.\n");`
375	`ret = R_NilValue;`	375	`ret = R_NilValue;`
376	`}`	376	`}`
377		377
378	`return ret;`	378	`return ret;`
379	`}`	379	`}`
380		380
381	`SEXP sdf_get_names(SEXP sdf) {`	381	`SEXP sdf_get_names(SEXP sdf) {`
382	`char *iname;`	382	`char *iname;`
383	`iname = SDF_INAME(sdf);`	383	`iname = SDF_INAME(sdf);`
384	`if (!USE_SDF(iname, TRUE)) return R_NilValue;`	384	`if (!USE_SDF1(iname, TRUE)) return R_NilValue;`
385		385
386	`int len;`	386	`int len;`
387	`len = sprintf(g_sql_buf[0], "select * from [%s].sdf_data;", iname);`	387	`len = sprintf(g_sql_buf[0], "select * from [%s].sdf_data;", iname);`
388		388
389	`sqlite3_stmt *stmt;`	389	`sqlite3_stmt *stmt;`
390	`int res, i;`	390	`int res, i;`
391		391
392	`res = sqlite3_prepare(g_workspace, g_sql_buf[0], len, &stmt, NULL);`	392	`res = sqlite3_prepare(g_workspace, g_sql_buf[0], len, &stmt, NULL);`
393	`if (_sqlite_error(res)) return R_NilValue;`	393	`if (_sqlite_error(res)) return R_NilValue;`
394		394
395	`SEXP ret;`	395	`SEXP ret;`
396		396
397	`len = sqlite3_column_count(stmt)-1;`	397	`len = sqlite3_column_count(stmt)-1;`
398	`PROTECT(ret = NEW_CHARACTER(len));`	398	`PROTECT(ret = NEW_CHARACTER(len));`
399		399
400	`for (i = 0; i < len; i++) {`	400	`for (i = 0; i < len; i++) {`
401	`SET_STRING_ELT(ret, i, mkChar(sqlite3_column_name(stmt, i+1)));`	401	`SET_STRING_ELT(ret, i, mkChar(sqlite3_column_name(stmt, i+1)));`
402	`}`	402	`}`
403		403
404	`sqlite3_finalize(stmt);`	404	`sqlite3_finalize(stmt);`
405	`UNPROTECT(1);`	405	`UNPROTECT(1);`
406	`return ret;`	406	`return ret;`
407	`}`	407	`}`
408		408
409	`SEXP sdf_get_length(SEXP sdf) {`	409	`SEXP sdf_get_length(SEXP sdf) {`
410	`char *iname;`	410	`char *iname;`
411	`iname = CHAR(STRING_ELT(_getListElement(sdf, "iname"),0));`	411	`iname = CHAR(STRING_ELT(_getListElement(sdf, "iname"),0));`
412	`if (!USE_SDF(iname, TRUE)) return R_NilValue;`	412	`if (!USE_SDF1(iname, TRUE)) return R_NilValue;`
413		413
414	`int len = sprintf(g_sql_buf[0], "select * from [%s].sdf_data;", iname);`	414	`int len = sprintf(g_sql_buf[0], "select * from [%s].sdf_data;", iname);`
415		415
416	`sqlite3_stmt *stmt;`	416	`sqlite3_stmt *stmt;`
417	`int res;`	417	`int res;`
418		418
419	`res = sqlite3_prepare(g_workspace, g_sql_buf[0], len, &stmt, NULL);`	419	`res = sqlite3_prepare(g_workspace, g_sql_buf[0], len, &stmt, NULL);`
420	`if (_sqlite_error(res)) return R_NilValue;`	420	`if (_sqlite_error(res)) return R_NilValue;`
421		421
422	`SEXP ret;`	422	`SEXP ret;`
423		423
424	`len = sqlite3_column_count(stmt)-1;`	424	`len = sqlite3_column_count(stmt)-1;`
425	`PROTECT(ret = NEW_INTEGER(1));`	425	`PROTECT(ret = NEW_INTEGER(1));`
426	`INTEGER(ret)[0] = len;`	426	`INTEGER(ret)[0] = len;`
427		427
428	`sqlite3_finalize(stmt);`	428	`sqlite3_finalize(stmt);`
429	`UNPROTECT(1);`	429	`UNPROTECT(1);`
430	`return ret;`	430	`return ret;`
431	`}`	431	`}`
432		432
433	`/* get row count */`	433	`/* get row count */`
434	`SEXP sdf_get_row_count(SEXP sdf) {`	434	`SEXP sdf_get_row_count(SEXP sdf) {`
435	`char *iname = CHAR(STRING_ELT(_getListElement(sdf, "iname"),0));`	435	`char *iname = CHAR(STRING_ELT(_getListElement(sdf, "iname"),0));`
436	`if (!USE_SDF(iname, TRUE)) return R_NilValue;`	436	`if (!USE_SDF1(iname, TRUE)) return R_NilValue;`
437		437
438	`char **out;`	438	`int nrow;`
439	`int res, ncol, nrow;`	-
440	`SEXP ret;`	439	`SEXP ret;`
441		440
442	`sprintf(g_sql_buf[0], "select count(*) from [%s].sdf_data;", iname);`	-
443	`res = sqlite3_get_table(g_workspace, g_sql_buf[0], &out, &nrow, &ncol, NULL);`	-
444	`if (_sqlite_error(res)) return R_NilValue;`	441	`nrow = _get_row_count2(iname, TRUE);`
445		-
446		442
447	`if (nrow != 1) ret = R_NilValue;`	443	`if (nrow < 1) ret = R_NilValue;`
448	`else {`	444	`else {`
449	`PROTECT(ret = NEW_INTEGER(1));`	445	`PROTECT(ret = NEW_INTEGER(1));`
450	`INTEGER(ret)[0] = atoi(out[1]);`	446	`INTEGER(ret)[0] = nrow;`
451	`UNPROTECT(1);`	447	`UNPROTECT(1);`
452	`}`	448	`}`
453		449
454	`sqlite3_free_table(out);`	-
455	`return ret;`	450	`return ret;`
456	`}`	451	`}`
457		452
458		453
459	`SEXP sdf_import_table(SEXP _filename, SEXP _name, SEXP _sep, SEXP _quote,`	454	`SEXP sdf_import_table(SEXP _filename, SEXP _name, SEXP _sep, SEXP _quote,`
460	`SEXP _rownames, SEXP _colnames) {`	455	`SEXP _rownames, SEXP _colnames) {`
461	`char *filename = CHAR_ELT(_filename, 0);`	456	`char *filename = CHAR_ELT(_filename, 0);`
462	`FILE *f = fopen(filename, "r");`	457	`FILE *f = fopen(filename, "r");`
463		458
464	`if (f == NULL) {`	459	`if (f == NULL) {`
465	`Rprintf("Error: File %s does not exist.", filename);`	460	`Rprintf("Error: File %s does not exist.", filename);`
466	`return R_NilValue;`	461	`return R_NilValue;`
467	`}`	462	`}`
468		463
469	`/char sep = CHAR_ELT(_sep, 0), *quote = CHAR_ELT(_quote,0);`	464	`/char sep = CHAR_ELT(_sep, 0), *quote = CHAR_ELT(_quote,0);`
470	`char name = CHAR_ELT(_name, 0);/`	465	`char name = CHAR_ELT(_name, 0);/`
471		466
472	`/* create the table */`	467	`/* create the table */`
473	`/* insert the stuffs */`	468	`/* insert the stuffs */`
474	`/* register to workspace */`	469	`/* register to workspace */`
475		470
476	`return R_NilValue;`	471	`return R_NilValue;`
477	`}`	472	`}`
478		473
479	`SEXP sdf_get_index(SEXP sdf, SEXP row, SEXP col) {`	474	`SEXP sdf_get_index(SEXP sdf, SEXP row, SEXP col) {`
480	`SEXP ret = R_NilValue;`	475	`SEXP ret = R_NilValue;`
481	`char *iname = SDF_INAME(sdf);`	476	`char *iname = SDF_INAME(sdf);`
482	`if (!USE_SDF(iname, TRUE)) return R_NilValue;`	477	`if (!USE_SDF1(iname, TRUE)) return R_NilValue;`
483		478
484	`sqlite3_stmt *stmt;`	479	`sqlite3_stmt *stmt;`
485	`int buflen = 0, idxlen, col_cnt, row_cnt, index;`	480	`int buflen = 0, idxlen, col_cnt, row_cnt, index;`
486	`int i, j,res;`	481	`int i, j,res;`
487	`int col_indices, col_index_len, dup_indices;`	482	`int col_indices, col_index_len, dup_indices;`
488	`int row_index_len = 0;`	483	`int row_index_len = 0;`
489		484
490	`sprintf(g_sql_buf[0], "select * from [%s].sdf_data ", iname);`	485	`sprintf(g_sql_buf[0], "select * from [%s].sdf_data ", iname);`
491	`res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0);`	486	`res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0);`
492	`if (_sqlite_error(res)) return R_NilValue;`	487	`if (_sqlite_error(res)) return R_NilValue;`
493		488
494	`buflen = sprintf(g_sql_buf[0], "select [row name]");`	489	`buflen = sprintf(g_sql_buf[0], "select [row name]");`
495	`idxlen = LENGTH(col);`	490	`idxlen = LENGTH(col);`
496	`col_cnt = sqlite3_column_count(stmt) - 1;`	491	`col_cnt = sqlite3_column_count(stmt) - 1;`
497		492
498	`/*`	493	`/*`
499	`* PROCESS COLUMN INDEX`	494	`* PROCESS COLUMN INDEX`
500	`*/`	495	`*/`

The R Project SVN R-packages

(root)/trunk/SQLiteDF/src/sqlite_dataframe.c – Rev 3324 → 3358