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 (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 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 {
        Rprintf("Error: the supplied value for arg name is not a string.\n");
    }
    return namelen;
}
 
static int _find_free_filename2(char *rname, char **iname, int *namelen, int *file_idx) {
    sqlite3_stmt *stmt;
    sqlite3_prepare(g_workspace, "select 1 from workspace where internal_name=?", -1,
            &stmt, NULL);
    do {
        if (!_file_exists(*iname)) {
            sqlite3_reset(stmt);
            sqlite3_bind_text(stmt, 1, *iname, -1, SQLITE_STATIC);
            if (sqlite3_step(stmt) != SQLITE_ROW) break;
        }
        *namelen = sprintf(*iname, "%s%d.db", rname, ++(*file_idx)) - 3;
    } while (*file_idx < 10000);
 
    sqlite3_finalize(stmt);
    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, int protect) {
    char *iname, *rname;
    int namelen, file_idx = 0, res;
 
    namelen = _check_sdf_name(name, &rname, &iname, &file_idx);
 
    if (!namelen) return NULL;
 
    _find_free_filename2(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, protect)) { /* _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; 
    }
 
    if (onamelen) *onamelen = namelen;
    return iname;
}
/* checks if a column has a corresponding factor|ordered table */
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 */
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 == STRSXP || 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;
}
 
/* set ordinary df row name as numbers */
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, FALSE);
    
    if (iname != NULL) {
        int sql_len, sql_len2;
        sqlite3_stmt *stmt;
 
        /* create sdf_data table */
        SEXP names = GET_NAMES(df), variable, levels, var_class;
        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);
            var_class = GET_CLASS(variable);
            class = (var_class == R_NilValue) ? NULL : CHAR(STRING_ELT(var_class, 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 (class != NULL && (strcmp(class, "factor") == 0 || strcmp(class, "ordered") == 0)){
                if (_create_factor_table2(iname, class, col_name)) 
                    return R_NilValue; /* dup tbl name? */
 
                _sqlite_exec("begin");
                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);
                _sqlite_exec("commit");
            }
        }
        
        _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);
 
        _sqlite_error(_sqlite_exec("begin"));
        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:
                    case STRSXP:
                        sqlite3_bind_text(stmt, j+2, CHAR_ELT(variable, i), -1, SQLITE_STATIC);
                }
                /* TODO: handle NA's & NULL's */
            }
 
            res = sqlite3_step(stmt);
            if (res != SQLITE_DONE) { 
                _sqlite_exec("ROLLBACK");
                sqlite3_finalize(stmt);
                Rprintf("SQLITE ERROR: %s\n", sqlite3_errmsg(g_workspace));
                return R_NilValue; /* why? */
            }
        }
                
        sqlite3_finalize(stmt);
        _sqlite_error(_sqlite_exec("COMMIT"));
 
        /* 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, FALSE)) 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;
    int len;
    sqlite3_stmt *stmt;
    int res;
 
    iname  = CHAR(STRING_ELT(_getListElement(sdf, "iname"),0));
    sprintf(g_sql_buf[0], "select * from [%s].sdf_data;", iname);
    if (!USE_SDF1(iname, TRUE, FALSE)) return R_NilValue;
 
    res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, NULL);
    if (_sqlite_error(res)) return R_NilValue;
 
    len = sqlite3_column_count(stmt)-1;
    sqlite3_finalize(stmt);
    return ScalarInteger(len);
}
 
/* get row count */
SEXP sdf_get_row_count(SEXP sdf) {
    char *iname = CHAR(STRING_ELT(_getListElement(sdf, "iname"),0));
    if (!USE_SDF1(iname, TRUE, FALSE)) return R_NilValue;
 
    int nrow;
    SEXP ret;
   
    nrow = _get_row_count2(iname, TRUE);
 
    if (nrow < 1) ret = R_NilValue;
    else ret = ScalarInteger(nrow);
 
    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 new_sdf) {
    SEXP ret = R_NilValue;
    char *iname = SDF_INAME(sdf);
 
 
    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, force_new_df = LOGICAL(new_sdf)[0];
 
    if (!USE_SDF1(iname, TRUE, TRUE)) return R_NilValue;
 
    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: set columns in the select statement
     */
    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: setup limit or where clause
     */
    idxlen = LENGTH(row);
    if (row == R_NilValue) {
        if (col_index_len == 1 && !force_new_df) {
            ret = sdf_get_variable(sdf, mkString(sqlite3_column_name(stmt,col_indices[0])));
            sqlite3_finalize(stmt);
            return ret;
        } if (col_index_len > 1 || (force_new_df && 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, TRUE);
 
            /* 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;
 
 
 
 
 
 
            /* remove protection */
            UNUSE_SDF2(iname2);
 
            /* 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 */
    }
 
    UNUSE_SDF2(iname);
 
    if (ret != R_NilValue && col_index_len > 1) {
        SET_CLASS(ret, mkString("data.frame"));
        _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 *iname = SDF_INAME(sdf), *colname;
    SEXP ret = R_NilValue, col, names, levels, rownames;
    int ncols, buflen, buflen2, i, j, res, nrows, *types, rownames_type;
    sqlite3_stmt *stmt;
    const char *type;
    char *rn_str; 
 
    class = CHAR_ELT(GET_CLASS(data), 0);
    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 */
            _sqlite_error(res);
            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]);
        res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, NULL);
        if (_sqlite_error(res)) return ret;
 
        /* finally, add rows */
        rownames = getAttrib(data, R_RowNamesSymbol); /* GET_ROWNAMES(data); */
        nrows = LENGTH(rownames);
        rownames_type = TYPEOF(rownames);
 
        _sqlite_begin;
        for (i = 0; i < nrows; i++) {
            sqlite3_reset(stmt);
            if (rownames_type == STRSXP) {
                rn_str = CHAR_ELT(rownames, i);
            } else if (rownames_type == INTSXP) {
                sprintf(g_sql_buf[1], "%d", INTEGER(rownames)[i]);
                rn_str = g_sql_buf[1];
            } else if (rownames_type == REALSXP) {
                sprintf(g_sql_buf[1], "%f", REAL(rownames)[i]);
                rn_str = g_sql_buf[1];
            } else {
                sprintf(g_sql_buf[1], "%d", i);
                rn_str = g_sql_buf[1];
            }
            sqlite3_bind_text(stmt, 1, rn_str, -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 */
                for (j = 1; res != SQLITE_DONE; j++) { /* _normally_, j shouldn't overflow */
                    sqlite3_reset(stmt);
                    sprintf(g_sql_buf[2], "%s-%d", rn_str, j);
                    sqlite3_bind_text(stmt, 1, g_sql_buf[2], -1, SQLITE_STATIC);
                    res = sqlite3_step(stmt);
                }
            } 
        }
 
        sqlite3_finalize(stmt);
        _sqlite_commit;
        ret = sdf;
    } else if (strcmp(class,"sqlite.data.frame") == 0) {
    }
 
    return ret;
}
 
 
SEXP sdf_get_iname(SEXP sdf) {
    char *iname, *sql;
    sqlite3_stmt *stmt;
    SEXP ret, names;
 
    iname = SDF_INAME(sdf);
    if (!USE_SDF1(iname, TRUE, FALSE)) return R_NilValue;
 
    sql = "select internal_name, rel_filename from workspace where internal_name=?";
    sqlite3_prepare(g_workspace, sql, -1, &stmt, NULL);
    sqlite3_bind_text(stmt, 1, iname, -1, SQLITE_STATIC);
    sqlite3_step(stmt);
 
    PROTECT(ret = NEW_CHARACTER(2));
    SET_STRING_ELT(ret, 0, mkChar((char *)sqlite3_column_text(stmt, 0)));
    SET_STRING_ELT(ret, 1, mkChar((char *)sqlite3_column_text(stmt, 1)));
 
    PROTECT(names = NEW_CHARACTER(2));
    SET_STRING_ELT(names, 0, mkChar("Internal Name"));
    SET_STRING_ELT(names, 1, mkChar("File"));
 
    SET_NAMES(ret, names);
 
    sqlite3_finalize(stmt);
    UNPROTECT(2);
 
    return ret;
}

Rev 3457	Rev 3471
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 (name == R_NilValue) {`	15	`if (name == R_NilValue) {`
16	`*rname = "data";`	16	`*rname = "data";`
17	`namelen = 5;`	17	`namelen = 5;`
18	`iname = (char)R_alloc(13, sizeof(char)); /* data10000.db\0 */`	18	`iname = (char)R_alloc(13, sizeof(char)); /* data10000.db\0 */`
19	`*file_idx = 1;`	19	`*file_idx = 1;`
20	`sprintf(iname, "%s%d.db", rname, *file_idx);`	20	`sprintf(iname, "%s%d.db", rname, *file_idx);`
21	`} else if (IS_CHARACTER(name)) {`	21	`} else if (IS_CHARACTER(name)) {`
22	`*rname = CHAR_ELT(name,0);`	22	`*rname = CHAR_ELT(name,0);`
23	`if (!_is_r_sym(*rname)) {`	23	`if (!_is_r_sym(*rname)) {`
24	`Rprintf("Error: supplied name \"%s\"is not a valid R symbol.\n",`	24	`Rprintf("Error: supplied name \"%s\"is not a valid R symbol.\n",`
25	`*rname);`	25	`*rname);`
26	`return FALSE;`	26	`return FALSE;`
27	`}`	27	`}`
28	`namelen = strlen(*rname);`	28	`namelen = strlen(*rname);`
29	`iname = (char)R_alloc(namelen + 9, sizeof(char)); /* <name>10000.db\0 */`	29	`iname = (char)R_alloc(namelen + 9, sizeof(char)); /* <name>10000.db\0 */`
30	`sprintf(iname, "%s.db", rname);`	30	`sprintf(iname, "%s.db", rname);`
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_filename2(char rname, char iname, int namelen, int *file_idx) {`
-		38	`sqlite3_stmt *stmt;`
-		39	`sqlite3_prepare(g_workspace, "select 1 from workspace where internal_name=?", -1,`
-		40	`&stmt, NULL);`
38	`do {`	41	`do {`
39	`if (!_file_exists(*iname)) break;`	42	`if (!_file_exists(*iname)) {`
-		43	`sqlite3_reset(stmt);`
-		44	`sqlite3_bind_text(stmt, 1, *iname, -1, SQLITE_STATIC);`
-		45	`if (sqlite3_step(stmt) != SQLITE_ROW) break;`
-		46	`}`
40	`namelen = sprintf(iname, "%s%d.db", rname, ++(*file_idx)) - 3;`	47	`namelen = sprintf(iname, "%s%d.db", rname, ++(*file_idx)) - 3;`
41	`} while (*file_idx < 10000);`	48	`} while (*file_idx < 10000);`
42		49
-		50	`sqlite3_finalize(stmt);`
43	`return *file_idx;`	51	`return *file_idx;`
44	`}`	52	`}`
45		53
46	`/* creates an sdf attribute table */`	54	`/* creates an sdf attribute table */`
47	`static int _create_sdf_attribute2(const char *iname) {`	55	`static int _create_sdf_attribute2(const char *iname) {`
48	`sprintf(g_sql_buf[2], "create table [%s].sdf_attributes(attr text, "`	56	`sprintf(g_sql_buf[2], "create table [%s].sdf_attributes(attr text, "`
49	`"value text, primary key (attr))", iname);`	57	`"value text, primary key (attr))", iname);`
50	`int res;`	58	`int res;`
51	`res = _sqlite_exec(g_sql_buf[2]);`	59	`res = _sqlite_exec(g_sql_buf[2]);`
52	`if (res == SQLITE_OK) {`	60	`if (res == SQLITE_OK) {`
53	`sprintf(g_sql_buf[2], "insert into [%s].sdf_attributes values ('name',"`	61	`sprintf(g_sql_buf[2], "insert into [%s].sdf_attributes values ('name',"`
54	`"'%s');", iname, iname);`	62	`"'%s');", iname, iname);`
55	`res = _sqlite_exec(g_sql_buf[2]);`	63	`res = _sqlite_exec(g_sql_buf[2]);`
56	`}`	64	`}`
57	`return res;`	65	`return res;`
58	`}`	66	`}`
59		67
60	`char _create_sdf_skeleton1(SEXP name, int onamelen, int protect) {`	68	`char _create_sdf_skeleton1(SEXP name, int onamelen, int protect) {`
61	`char iname, rname;`	69	`char iname, rname;`
62	`int namelen, file_idx = 0, res;`	70	`int namelen, file_idx = 0, res;`
63		71
64	`namelen = _check_sdf_name(name, &rname, &iname, &file_idx);`	72	`namelen = _check_sdf_name(name, &rname, &iname, &file_idx);`
65		73
66	`if (!namelen) return NULL;`	74	`if (!namelen) return NULL;`
67		75
68	`_find_free_filename(rname, &iname, &namelen, &file_idx);`	76	`_find_free_filename2(rname, &iname, &namelen, &file_idx);`
69		77
70	`if (file_idx >= 10000) {`	78	`if (file_idx >= 10000) {`
71	`Rprintf("Error: cannot find free SDF name.\n");`	79	`Rprintf("Error: cannot find free SDF name.\n");`
72	`return NULL;`	80	`return NULL;`
73	`}`	81	`}`
74		82
75	`/* add to workspace */`	83	`/* add to workspace */`
76	`strcpy(g_sql_buf[3], iname);`	84	`strcpy(g_sql_buf[3], iname);`
77	`iname[namelen] = 0; /* remove ".db" */`	85	`iname[namelen] = 0; /* remove ".db" */`
78	`res = _add_sdf1(g_sql_buf[3], iname);`	86	`res = _add_sdf1(g_sql_buf[3], iname);`
79	`if (_sqlite_error(res)) return NULL;`	87	`if (_sqlite_error(res)) return NULL;`
80		88
81	`/* detach SDF's if necessary to attach this one. if file does not`	89	`/* 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 */`	90	`* exist, then a sqlite db will be created after we make our 1st table */`
83	`if (!USE_SDF1(iname, FALSE, protect)) { /* _delete_sdf2(iname); */ return NULL; }`	91	`if (!USE_SDF1(iname, FALSE, protect)) { /* _delete_sdf2(iname); */ return NULL; }`
84		92
85		93
86	`/* create attributes table */`	94	`/* create attributes table */`
87	`res = _create_sdf_attribute2(iname);`	95	`res = _create_sdf_attribute2(iname);`
88	`if (_sqlite_error(res)) {`	96	`if (_sqlite_error(res)) {`
89	`sprintf(g_sql_buf[2], "detach '%s'", iname);`	97	`sprintf(g_sql_buf[2], "detach '%s'", iname);`
90	`_delete_sdf2(iname);`	98	`_delete_sdf2(iname);`
91	`return NULL;`	99	`return NULL;`
92	`}`	100	`}`
93		101
94	`if (onamelen) *onamelen = namelen;`	102	`if (onamelen) *onamelen = namelen;`
95	`return iname;`	103	`return iname;`
96	`}`	104	`}`
97	`/* checks if a column has a corresponding factor\|ordered table */`	105	`/* checks if a column has a corresponding factor\|ordered table */`
98	`int _is_factor2(const char iname, const char factor_type, const char *colname) {`	106	`int _is_factor2(const char iname, const char factor_type, const char *colname) {`
99	`sqlite3_stmt *stmt;`	107	`sqlite3_stmt *stmt;`
100	`sprintf(g_sql_buf[2], "select * from [%s].[%s %s]", iname, factor_type,`	108	`sprintf(g_sql_buf[2], "select * from [%s].[%s %s]", iname, factor_type,`
101	`colname);`	109	`colname);`
102	`int res;`	110	`int res;`
103	`res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);`	111	`res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);`
104	`sqlite3_finalize(stmt);`	112	`sqlite3_finalize(stmt);`
105	`return res == SQLITE_OK;`	113	`return res == SQLITE_OK;`
106	`}`	114	`}`
107		115
108	`/* create a factor\|ordered table */`	116	`/* create a factor\|ordered table */`
109	`int _create_factor_table2(const char iname, const char factor_type,`	117	`int _create_factor_table2(const char iname, const char factor_type,`
110	`const char *colname) {`	118	`const char *colname) {`
111	`sqlite3_stmt *stmt;`	119	`sqlite3_stmt *stmt;`
112	`sprintf(g_sql_buf[2], "create table [%s].[%s %s] (level int, label text, "`	120	`sprintf(g_sql_buf[2], "create table [%s].[%s %s] (level int, label text, "`
113	`"primary key(level), unique(label));", iname, factor_type, colname);`	121	`"primary key(level), unique(label));", iname, factor_type, colname);`
114	`int res;`	122	`int res;`
115	`res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);`	123	`res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);`
116	`if (res == SQLITE_OK) sqlite3_step(stmt);`	124	`if (res == SQLITE_OK) sqlite3_step(stmt);`
117	`sqlite3_finalize(stmt);`	125	`sqlite3_finalize(stmt);`
118	`return res; /* error on dup name? */`	126	`return res; /* error on dup name? */`
119	`}`	127	`}`
120		128
121	`/* copy a factor\|ordered table from a sdf(db) to another sdf(db) */`	129	`/* copy a factor\|ordered table from a sdf(db) to another sdf(db) */`
122	`int _copy_factor_levels2(const char factor_type, const char iname_src,`	130	`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) {`	131	`const char colname_src, const char iname_dst, const char *colname_dst) {`
124	`sqlite3_stmt *stmt;`	132	`sqlite3_stmt *stmt;`
125	`int res;`	133	`int res;`
126	`res = _create_factor_table2(iname_dst, factor_type, colname_dst);`	134	`res = _create_factor_table2(iname_dst, factor_type, colname_dst);`
127	`if (res == SQLITE_OK) {`	135	`if (res == SQLITE_OK) {`
128	`sprintf(g_sql_buf[2], "insert into [%s].[%s %s] select * from [%s].[%s %s]",`	136	`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,`	137	`iname_src, factor_type, colname_src, iname_dst, factor_type,`
130	`colname_dst);`	138	`colname_dst);`
131	`res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);`	139	`res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);`
132	`if (res == SQLITE_OK) sqlite3_step(stmt);`	140	`if (res == SQLITE_OK) sqlite3_step(stmt);`
133	`sqlite3_finalize(stmt);`	141	`sqlite3_finalize(stmt);`
134	`}`	142	`}`
135	`return res; /* error on dup name? */`	143	`return res; /* error on dup name? */`
136	`}`	144	`}`
137		145
138		146
139	`/* assumes stmt has col_cnt + 1 columns, col 0 is [row name]. returned SEXP is`	147	`/* 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 &`	148	`* not UNPROTECT-ed, user will have to do that. will create the names &`
141	`* attach factor infos */`	149	`* attach factor infos */`
142	`static SEXP _setup_df_sexp1(sqlite3_stmt stmt, const char iname,`	150	`static SEXP _setup_df_sexp1(sqlite3_stmt stmt, const char iname,`
143	`int col_cnt, int row_cnt, int *dup_indices) {`	151	`int col_cnt, int row_cnt, int *dup_indices) {`
144	`SEXP ret, names, value, class = R_NilValue;`	152	`SEXP ret, names, value, class = R_NilValue;`
145	`int i, type;`	153	`int i, type;`
146	`const char colname, coltype;`	154	`const char colname, coltype;`
147		155
148	`PROTECT(ret = NEW_LIST(col_cnt));`	156	`PROTECT(ret = NEW_LIST(col_cnt));`
149		157
150	`/* set up names. */`	158	`/* set up names. */`
151	`PROTECT(names = NEW_CHARACTER(col_cnt));`	159	`PROTECT(names = NEW_CHARACTER(col_cnt));`
152	`for (i = 0; i < col_cnt; i++) {`	160	`for (i = 0; i < col_cnt; i++) {`
153	`colname = sqlite3_column_name(stmt, i+1); /* +1 bec [row name is 0] */`	161	`colname = sqlite3_column_name(stmt, i+1); /* +1 bec [row name is 0] */`
154	`coltype = sqlite3_column_decltype(stmt, i+1);`	162	`coltype = sqlite3_column_decltype(stmt, i+1);`
155		163
156	`if (dup_indices[i] == 0) {`	164	`if (dup_indices[i] == 0) {`
157	`strcpy(g_sql_buf[1], colname);`	165	`strcpy(g_sql_buf[1], colname);`
158	`} else {`	166	`} else {`
159	`sprintf(g_sql_buf[1], "%s.%d", colname, dup_indices[i]);`	167	`sprintf(g_sql_buf[1], "%s.%d", colname, dup_indices[i]);`
160	`}`	168	`}`
161		169
162	`SET_STRING_ELT(names, i, mkChar(g_sql_buf[1]));`	170	`SET_STRING_ELT(names, i, mkChar(g_sql_buf[1]));`
163		171
164	`if (strcmp(coltype, "text") == 0) {`	172	`if (strcmp(coltype, "text") == 0) {`
165	`PROTECT(value = NEW_CHARACTER(row_cnt));`	173	`PROTECT(value = NEW_CHARACTER(row_cnt));`
166	`} else if (strcmp(coltype, "double") == 0) {`	174	`} else if (strcmp(coltype, "double") == 0) {`
167	`PROTECT(value = NEW_NUMERIC(row_cnt));`	175	`PROTECT(value = NEW_NUMERIC(row_cnt));`
168	`} else if (strcmp(coltype, "bit") == 0) {`	176	`} else if (strcmp(coltype, "bit") == 0) {`
169	`PROTECT(value = NEW_LOGICAL(row_cnt));`	177	`PROTECT(value = NEW_LOGICAL(row_cnt));`
170	`} else if (strcmp(coltype, "integer") == 0 \|\|`	178	`} else if (strcmp(coltype, "integer") == 0 \|\|`
171	`strcmp(coltype, "int") == 0) {`	179	`strcmp(coltype, "int") == 0) {`
172	`PROTECT(value = NEW_INTEGER(row_cnt));`	180	`PROTECT(value = NEW_INTEGER(row_cnt));`
173	`type = _get_factor_levels1(iname, colname, value);`	181	`type = _get_factor_levels1(iname, colname, value);`
174	`if (type == VAR_FACTOR) {`	182	`if (type == VAR_FACTOR) {`
175	`PROTECT(class = mkString("factor"));`	183	`PROTECT(class = mkString("factor"));`
176	`SET_CLASS(value, class);`	184	`SET_CLASS(value, class);`
177	`UNPROTECT(1);`	185	`UNPROTECT(1);`
178	`} else if (type == VAR_ORDERED) {`	186	`} else if (type == VAR_ORDERED) {`
179	`PROTECT(class = NEW_CHARACTER(2));`	187	`PROTECT(class = NEW_CHARACTER(2));`
180	`SET_STRING_ELT(class, 0, mkChar("ordered"));`	188	`SET_STRING_ELT(class, 0, mkChar("ordered"));`
181	`SET_STRING_ELT(class, 1, mkChar("factor"));`	189	`SET_STRING_ELT(class, 1, mkChar("factor"));`
182	`SET_CLASS(value, class);`	190	`SET_CLASS(value, class);`
183	`UNPROTECT(1);`	191	`UNPROTECT(1);`
184	`}`	192	`}`
185	`} else {`	193	`} else {`
186	`Rprintf("Error: not supported type %s for %s\n", coltype, colname);`	194	`Rprintf("Error: not supported type %s for %s\n", coltype, colname);`
187	`UNPROTECT(2); /* unprotect ret, names */`	195	`UNPROTECT(2); /* unprotect ret, names */`
188	`return R_NilValue;`	196	`return R_NilValue;`
189	`}`	197	`}`
190		198
191	`SET_VECTOR_ELT(ret, i, value);`	199	`SET_VECTOR_ELT(ret, i, value);`
192	`UNPROTECT(1); /* unprotect value */`	200	`UNPROTECT(1); /* unprotect value */`
193		201
194	`}`	202	`}`
195	`SET_NAMES(ret, names);`	203	`SET_NAMES(ret, names);`
196	`UNPROTECT(1); /* unprotect names only */`	204	`UNPROTECT(1); /* unprotect names only */`
197	`return ret;`	205	`return ret;`
198	`}`	206	`}`
199		207
200	`/* expected that 1st col of stmt is [row name], so we start with index 1 */`	208	`/* 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) {`	209	`static int _add_row_to_df(SEXP df, sqlite3_stmt *stmt, int row, int ncols) {`
202	`SEXP vec;`	210	`SEXP vec;`
203	`int type = -1, is_null, i;`	211	`int type = -1, is_null, i;`
204	`for (i = 1; i <= ncols; i++) {`	212	`for (i = 1; i <= ncols; i++) {`
205	`vec = VECTOR_ELT(df, i-1);`	213	`vec = VECTOR_ELT(df, i-1);`
206	`type = TYPEOF(vec);`	214	`type = TYPEOF(vec);`
207		215
208	`is_null = (sqlite3_column_type(stmt, row) == SQLITE_NULL);`	216	`is_null = (sqlite3_column_type(stmt, row) == SQLITE_NULL);`
209		217
210	`if (type == STRSXP \|\| type == CHARSXP) {`	218	`if (type == STRSXP \|\| type == CHARSXP) {`
211	`SET_STRING_ELT(vec, row, mkChar((char *)sqlite3_column_text(stmt, i)));`	219	`SET_STRING_ELT(vec, row, mkChar((char *)sqlite3_column_text(stmt, i)));`
212	`} else if (type == INTSXP) {`	220	`} else if (type == INTSXP) {`
213	`INTEGER(vec)[row] = sqlite3_column_int(stmt, i);`	221	`INTEGER(vec)[row] = sqlite3_column_int(stmt, i);`
214	`} else if (type == REALSXP) {`	222	`} else if (type == REALSXP) {`
215	`REAL(vec)[row] = sqlite3_column_double(stmt, i);`	223	`REAL(vec)[row] = sqlite3_column_double(stmt, i);`
216	`} else if (type == LGLSXP) {`	224	`} else if (type == LGLSXP) {`
217	`LOGICAL(vec)[row] = sqlite3_column_int(stmt, i);`	225	`LOGICAL(vec)[row] = sqlite3_column_int(stmt, i);`
218	`}`	226	`}`
219	`}`	227	`}`
220	`return type;`	228	`return type;`
221	`}`	229	`}`
222		230
223	`/* set ordinary df row name as numbers */`	231	`/* set ordinary df row name as numbers */`
224	`static void _set_rownames2(SEXP df) {`	232	`static void _set_rownames2(SEXP df) {`
225	`SEXP value = VECTOR_ELT(df, 0);`	233	`SEXP value = VECTOR_ELT(df, 0);`
226	`int len = LENGTH(value), i;`	234	`int len = LENGTH(value), i;`
227	`PROTECT(value = NEW_CHARACTER(len));`	235	`PROTECT(value = NEW_CHARACTER(len));`
228	`for (i = 0; i < len; i++) {`	236	`for (i = 0; i < len; i++) {`
229	`sprintf(g_sql_buf[2], "%d", i+1);`	237	`sprintf(g_sql_buf[2], "%d", i+1);`
230	`SET_STRING_ELT(value, i, mkChar(g_sql_buf[2]));`	238	`SET_STRING_ELT(value, i, mkChar(g_sql_buf[2]));`
231	`}`	239	`}`
232		240
233	`SET_ROWNAMES(df, value);`	241	`SET_ROWNAMES(df, value);`
234	`UNPROTECT(1);`	242	`UNPROTECT(1);`
235	`}`	243	`}`
236		244
237		245
238	`/****************************************************************************`	246	`/****************************************************************************`
239	`* SDF FUNCTIONS`	247	`* SDF FUNCTIONS`
240	`****************************************************************************/`	248	`****************************************************************************/`
241		249
242	`SEXP sdf_create_sdf(SEXP df, SEXP name) {`	250	`SEXP sdf_create_sdf(SEXP df, SEXP name) {`
243	`SEXP ret;`	251	`SEXP ret;`
244	`char *iname;`	252	`char *iname;`
245	`int namelen, res, i, j;`	253	`int namelen, res, i, j;`
246		254
247	`/* find free name, attach sdf, create sdf_attributes */`	255	`/* find free name, attach sdf, create sdf_attributes */`
248	`iname = _create_sdf_skeleton1(name, &namelen, FALSE);`	256	`iname = _create_sdf_skeleton1(name, &namelen, FALSE);`
249		257
250	`if (iname != NULL) {`	258	`if (iname != NULL) {`
251	`int sql_len, sql_len2;`	259	`int sql_len, sql_len2;`
252	`sqlite3_stmt *stmt;`	260	`sqlite3_stmt *stmt;`
253		261
254	`/* create sdf_data table */`	262	`/* create sdf_data table */`
255	`SEXP names = GET_NAMES(df), variable, levels, var_class;`	263	`SEXP names = GET_NAMES(df), variable, levels, var_class;`
256	`int ncols = GET_LENGTH(names), type, *types;`	264	`int ncols = GET_LENGTH(names), type, *types;`
257	`char col_name, class, *factor;`	265	`char col_name, class, *factor;`
258		266
259	`/* variables for adding rownames */`	267	`/* variables for adding rownames */`
260	`SEXP rownames;`	268	`SEXP rownames;`
261	`int nrows;`	269	`int nrows;`
262	`char *row_name;`	270	`char *row_name;`
263		271
264	`/* TODO: put constraints on table after inserting everything? */`	272	`/* TODO: put constraints on table after inserting everything? */`
265		273
266		274
267	`/*`	275	`/*`
268	`* create the create table and insert sql scripts by looping through`	276	`* create the create table and insert sql scripts by looping through`
269	`* the columns of df`	277	`* the columns of df`
270	`*/`	278	`*/`
271	`types = (int *)R_alloc(ncols, sizeof(int));`	279	`types = (int *)R_alloc(ncols, sizeof(int));`
272	`sql_len = sprintf(g_sql_buf[0], "create table [%s].sdf_data ([row name] text", iname);`	280	`sql_len = sprintf(g_sql_buf[0], "create table [%s].sdf_data ([row name] text", iname);`
273	`sql_len2 = sprintf(g_sql_buf[1], "insert into [%s].sdf_data values(?", iname);`	281	`sql_len2 = sprintf(g_sql_buf[1], "insert into [%s].sdf_data values(?", iname);`
274		282
275	`for (i = 0; i < ncols; i++) {`	283	`for (i = 0; i < ncols; i++) {`
276	`col_name = CHAR(STRING_ELT(names, i));`	284	`col_name = CHAR(STRING_ELT(names, i));`
277		285
278	`/* add column definition to the create table sql */`	286	`/* add column definition to the create table sql */`
279	`_expand_buf(0, sql_len+strlen(col_name)+10);`	287	`_expand_buf(0, sql_len+strlen(col_name)+10);`
280		288
281	`variable = _getListElement(df, col_name);`	289	`variable = _getListElement(df, col_name);`
282	`var_class = GET_CLASS(variable);`	290	`var_class = GET_CLASS(variable);`
283	`class = (var_class == R_NilValue) ? NULL : CHAR(STRING_ELT(var_class, 0));`	291	`class = (var_class == R_NilValue) ? NULL : CHAR(STRING_ELT(var_class, 0));`
284	`type = TYPEOF(variable);`	292	`type = TYPEOF(variable);`
285	`types[i] = type;`	293	`types[i] = type;`
286		294
287	`sql_len += sprintf(g_sql_buf[0]+sql_len, ", [%s] %s", col_name,`	295	`sql_len += sprintf(g_sql_buf[0]+sql_len, ", [%s] %s", col_name,`
288	`_get_column_type(class, type));`	296	`_get_column_type(class, type));`
289		297
290	`/* add handler to insert table sql */`	298	`/* add handler to insert table sql */`
291	`_expand_buf(1, sql_len+5);`	299	`_expand_buf(1, sql_len+5);`
292	`strcpy(g_sql_buf[1]+sql_len2, ",?");`	300	`strcpy(g_sql_buf[1]+sql_len2, ",?");`
293	`sql_len2 += 2;`	301	`sql_len2 += 2;`
294		302
295	`/* create separate table for factors decode */`	303	`/* create separate table for factors decode */`
296	`if (class != NULL && (strcmp(class, "factor") == 0 \|\| strcmp(class, "ordered") == 0)){`	304	`if (class != NULL && (strcmp(class, "factor") == 0 \|\| strcmp(class, "ordered") == 0)){`
297	`if (_create_factor_table2(iname, class, col_name))`	305	`if (_create_factor_table2(iname, class, col_name))`
298	`return R_NilValue; /* dup tbl name? */`	306	`return R_NilValue; /* dup tbl name? */`
299		307
300	`_sqlite_exec("begin");`	308	`_sqlite_exec("begin");`
301	`levels = GET_LEVELS(variable);`	309	`levels = GET_LEVELS(variable);`
302	`sprintf(g_sql_buf[2], "insert into [%s].[%s %s] values(?, ?);",`	310	`sprintf(g_sql_buf[2], "insert into [%s].[%s %s] values(?, ?);",`
303	`iname, class, col_name);`	311	`iname, class, col_name);`
304	`res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, NULL);`	312	`res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, NULL);`
305	`if (_sqlite_error(res)) return R_NilValue; /* dup tbl name? */`	313	`if (_sqlite_error(res)) return R_NilValue; /* dup tbl name? */`
306		314
307	`for (j = 0; j < GET_LENGTH(levels); j++) {`	315	`for (j = 0; j < GET_LENGTH(levels); j++) {`
308	`sqlite3_reset(stmt);`	316	`sqlite3_reset(stmt);`
309	`factor = CHAR(STRING_ELT(levels, j));`	317	`factor = CHAR(STRING_ELT(levels, j));`
310	`sqlite3_bind_int(stmt, 1, j+1);`	318	`sqlite3_bind_int(stmt, 1, j+1);`
311	`sqlite3_bind_text(stmt, 2, factor, -1, SQLITE_STATIC);`	319	`sqlite3_bind_text(stmt, 2, factor, -1, SQLITE_STATIC);`
312	`sqlite3_step(stmt);`	320	`sqlite3_step(stmt);`
313	`}`	321	`}`
314	`sqlite3_finalize(stmt);`	322	`sqlite3_finalize(stmt);`
315	`_sqlite_exec("commit");`	323	`_sqlite_exec("commit");`
316	`}`	324	`}`
317	`}`	325	`}`
318		326
319	`_expand_buf(0,sql_len+35);`	327	`_expand_buf(0,sql_len+35);`
320	`sql_len += sprintf(g_sql_buf[0]+sql_len, ", primary key([row name]));");`	328	`sql_len += sprintf(g_sql_buf[0]+sql_len, ", primary key([row name]));");`
321	`/* sql_len += sprintf(g_sql_buf[0]+sql_len, ");"); */`	329	`/* sql_len += sprintf(g_sql_buf[0]+sql_len, ");"); */`
322	`res = _sqlite_exec(g_sql_buf[0]);`	330	`res = _sqlite_exec(g_sql_buf[0]);`
323	`if (_sqlite_error(res)) return R_NilValue; /* why? */`	331	`if (_sqlite_error(res)) return R_NilValue; /* why? */`
324		332
325	`/*`	333	`/*`
326	`* add the data in df to the sdf`	334	`* add the data in df to the sdf`
327	`*/`	335	`*/`
328	`rownames = getAttrib(df, R_RowNamesSymbol);`	336	`rownames = getAttrib(df, R_RowNamesSymbol);`
329	`nrows = GET_LENGTH(rownames);`	337	`nrows = GET_LENGTH(rownames);`
330		338
331	`_sqlite_error(_sqlite_exec("begin"));`	339	`_sqlite_error(_sqlite_exec("begin"));`
332	`sprintf(g_sql_buf[1]+sql_len2, ")");`	340	`sprintf(g_sql_buf[1]+sql_len2, ")");`
333	`res = sqlite3_prepare(g_workspace, g_sql_buf[1], -1, &stmt, NULL);`	341	`res = sqlite3_prepare(g_workspace, g_sql_buf[1], -1, &stmt, NULL);`
334		342
335	`for (i = 0; i < nrows; i++) {`	343	`for (i = 0; i < nrows; i++) {`
336	`sqlite3_reset(stmt);`	344	`sqlite3_reset(stmt);`
337		345
338	`/* since this is coming from a real dataframe, we're assured that`	346	`/* since this is coming from a real dataframe, we're assured that`
339	`* the rownames are unique */`	347	`* the rownames are unique */`
340	`if (IS_CHARACTER(rownames)) {`	348	`if (IS_CHARACTER(rownames)) {`
341	`row_name = CHAR(STRING_ELT(rownames, i));`	349	`row_name = CHAR(STRING_ELT(rownames, i));`
342	`if (row_name) / if not empty string */`	350	`if (row_name) / if not empty string */`
343	`sqlite3_bind_text(stmt, 1, row_name, strlen(row_name), SQLITE_STATIC);`	351	`sqlite3_bind_text(stmt, 1, row_name, strlen(row_name), SQLITE_STATIC);`
344	`else sqlite3_bind_int(stmt, 1, i);`	352	`else sqlite3_bind_int(stmt, 1, i);`
345	`} else if (IS_INTEGER(rownames)) {`	353	`} else if (IS_INTEGER(rownames)) {`
346	`sqlite3_bind_int(stmt, 1, INTEGER(rownames)[i]);`	354	`sqlite3_bind_int(stmt, 1, INTEGER(rownames)[i]);`
347	`} else sqlite3_bind_int(stmt, 1, i);`	355	`} else sqlite3_bind_int(stmt, 1, i);`
348		356
349	`for (j = 0; j < ncols; j++) {`	357	`for (j = 0; j < ncols; j++) {`
350	`variable = VECTOR_ELT(df, j);`	358	`variable = VECTOR_ELT(df, j);`
351	`switch(types[j]) {`	359	`switch(types[j]) {`
352	`case INTSXP :`	360	`case INTSXP :`
353	`sqlite3_bind_int(stmt, j+2, INTEGER(variable)[i]);`	361	`sqlite3_bind_int(stmt, j+2, INTEGER(variable)[i]);`
354	`break;`	362	`break;`
355	`case REALSXP:`	363	`case REALSXP:`
356	`sqlite3_bind_double(stmt, j+2, REAL(variable)[i]);`	364	`sqlite3_bind_double(stmt, j+2, REAL(variable)[i]);`
357	`break;`	365	`break;`
358	`case CHARSXP:`	366	`case CHARSXP:`
359	`case STRSXP:`	367	`case STRSXP:`
360	`sqlite3_bind_text(stmt, j+2, CHAR_ELT(variable, i), -1, SQLITE_STATIC);`	368	`sqlite3_bind_text(stmt, j+2, CHAR_ELT(variable, i), -1, SQLITE_STATIC);`
361	`}`	369	`}`
362	`/* TODO: handle NA's & NULL's */`	370	`/* TODO: handle NA's & NULL's */`
363	`}`	371	`}`
364		372
365	`res = sqlite3_step(stmt);`	373	`res = sqlite3_step(stmt);`
366	`if (res != SQLITE_DONE) {`	374	`if (res != SQLITE_DONE) {`
367	`_sqlite_exec("ROLLBACK");`	375	`_sqlite_exec("ROLLBACK");`
368	`sqlite3_finalize(stmt);`	376	`sqlite3_finalize(stmt);`
369	`Rprintf("SQLITE ERROR: %s\n", sqlite3_errmsg(g_workspace));`	377	`Rprintf("SQLITE ERROR: %s\n", sqlite3_errmsg(g_workspace));`
370	`return R_NilValue; /* why? */`	378	`return R_NilValue; /* why? */`
371	`}`	379	`}`
372	`}`	380	`}`
373		381
374	`sqlite3_finalize(stmt);`	382	`sqlite3_finalize(stmt);`
375	`_sqlite_error(_sqlite_exec("COMMIT"));`	383	`_sqlite_error(_sqlite_exec("COMMIT"));`
376		384
377	`/* create a new object representing the sdf */`	385	`/* create a new object representing the sdf */`
378	`ret = _create_sdf_sexp(iname);`	386	`ret = _create_sdf_sexp(iname);`
379		387
380	`} else {`	388	`} else {`
381	`Rprintf("ERROR: unable to create a sqlite data frame.\n");`	389	`Rprintf("ERROR: unable to create a sqlite data frame.\n");`
382	`ret = R_NilValue;`	390	`ret = R_NilValue;`
383	`}`	391	`}`
384		392
385	`return ret;`	393	`return ret;`
386	`}`	394	`}`
387		395
388	`SEXP sdf_get_names(SEXP sdf) {`	396	`SEXP sdf_get_names(SEXP sdf) {`
389	`char *iname;`	397	`char *iname;`
390	`iname = SDF_INAME(sdf);`	398	`iname = SDF_INAME(sdf);`
391	`if (!USE_SDF1(iname, TRUE, FALSE)) return R_NilValue;`	399	`if (!USE_SDF1(iname, TRUE, FALSE)) return R_NilValue;`
392		400
393	`int len;`	401	`int len;`
394	`len = sprintf(g_sql_buf[0], "select * from [%s].sdf_data;", iname);`	402	`len = sprintf(g_sql_buf[0], "select * from [%s].sdf_data;", iname);`
395		403
396	`sqlite3_stmt *stmt;`	404	`sqlite3_stmt *stmt;`
397	`int res, i;`	405	`int res, i;`
398		406
399	`res = sqlite3_prepare(g_workspace, g_sql_buf[0], len, &stmt, NULL);`	407	`res = sqlite3_prepare(g_workspace, g_sql_buf[0], len, &stmt, NULL);`
400	`if (_sqlite_error(res)) return R_NilValue;`	408	`if (_sqlite_error(res)) return R_NilValue;`
401		409
402	`SEXP ret;`	410	`SEXP ret;`
403		411
404	`len = sqlite3_column_count(stmt)-1;`	412	`len = sqlite3_column_count(stmt)-1;`
405	`PROTECT(ret = NEW_CHARACTER(len));`	413	`PROTECT(ret = NEW_CHARACTER(len));`
406		414
407	`for (i = 0; i < len; i++) {`	415	`for (i = 0; i < len; i++) {`
408	`SET_STRING_ELT(ret, i, mkChar(sqlite3_column_name(stmt, i+1)));`	416	`SET_STRING_ELT(ret, i, mkChar(sqlite3_column_name(stmt, i+1)));`
409	`}`	417	`}`
410		418
411	`sqlite3_finalize(stmt);`	419	`sqlite3_finalize(stmt);`
412	`UNPROTECT(1);`	420	`UNPROTECT(1);`
413	`return ret;`	421	`return ret;`
414	`}`	422	`}`
415		423
416	`SEXP sdf_get_length(SEXP sdf) {`	424	`SEXP sdf_get_length(SEXP sdf) {`
417	`char *iname;`	425	`char *iname;`
418	`int len;`	426	`int len;`
419	`sqlite3_stmt *stmt;`	427	`sqlite3_stmt *stmt;`
420	`int res;`	428	`int res;`
421		429
422	`iname = CHAR(STRING_ELT(_getListElement(sdf, "iname"),0));`	430	`iname = CHAR(STRING_ELT(_getListElement(sdf, "iname"),0));`
423	`sprintf(g_sql_buf[0], "select * from [%s].sdf_data;", iname);`	431	`sprintf(g_sql_buf[0], "select * from [%s].sdf_data;", iname);`
424	`if (!USE_SDF1(iname, TRUE, FALSE)) return R_NilValue;`	432	`if (!USE_SDF1(iname, TRUE, FALSE)) return R_NilValue;`
425		433
426	`res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, NULL);`	434	`res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, NULL);`
427	`if (_sqlite_error(res)) return R_NilValue;`	435	`if (_sqlite_error(res)) return R_NilValue;`
428		436
429	`len = sqlite3_column_count(stmt)-1;`	437	`len = sqlite3_column_count(stmt)-1;`
430	`sqlite3_finalize(stmt);`	438	`sqlite3_finalize(stmt);`
431	`return ScalarInteger(len);`	439	`return ScalarInteger(len);`
432	`}`	440	`}`
433		441
434	`/* get row count */`	442	`/* get row count */`
435	`SEXP sdf_get_row_count(SEXP sdf) {`	443	`SEXP sdf_get_row_count(SEXP sdf) {`
436	`char *iname = CHAR(STRING_ELT(_getListElement(sdf, "iname"),0));`	444	`char *iname = CHAR(STRING_ELT(_getListElement(sdf, "iname"),0));`
437	`if (!USE_SDF1(iname, TRUE, FALSE)) return R_NilValue;`	445	`if (!USE_SDF1(iname, TRUE, FALSE)) return R_NilValue;`
438		446
439	`int nrow;`	447	`int nrow;`
440	`SEXP ret;`	448	`SEXP ret;`
441		449
442	`nrow = _get_row_count2(iname, TRUE);`	450	`nrow = _get_row_count2(iname, TRUE);`
443		451
444	`if (nrow < 1) ret = R_NilValue;`	452	`if (nrow < 1) ret = R_NilValue;`
445	`else ret = ScalarInteger(nrow);`	453	`else ret = ScalarInteger(nrow);`
446		454
447	`return ret;`	455	`return ret;`
448	`}`	456	`}`
449		457
450		458
451	`SEXP sdf_import_table(SEXP _filename, SEXP _name, SEXP _sep, SEXP _quote,`	459	`SEXP sdf_import_table(SEXP _filename, SEXP _name, SEXP _sep, SEXP _quote,`
452	`SEXP _rownames, SEXP _colnames) {`	460	`SEXP _rownames, SEXP _colnames) {`
453	`char *filename = CHAR_ELT(_filename, 0);`	461	`char *filename = CHAR_ELT(_filename, 0);`
454	`FILE *f = fopen(filename, "r");`	462	`FILE *f = fopen(filename, "r");`
455		463
456	`if (f == NULL) {`	464	`if (f == NULL) {`
457	`Rprintf("Error: File %s does not exist.", filename);`	465	`Rprintf("Error: File %s does not exist.", filename);`
458	`return R_NilValue;`	466	`return R_NilValue;`
459	`}`	467	`}`
460		468
461	`/char sep = CHAR_ELT(_sep, 0), *quote = CHAR_ELT(_quote,0);`	469	`/char sep = CHAR_ELT(_sep, 0), *quote = CHAR_ELT(_quote,0);`
462	`char name = CHAR_ELT(_name, 0);/`	470	`char name = CHAR_ELT(_name, 0);/`
463		471
464	`/* create the table */`	472	`/* create the table */`
465	`/* insert the stuffs */`	473	`/* insert the stuffs */`
466	`/* register to workspace */`	474	`/* register to workspace */`
467		475
468	`return R_NilValue;`	476	`return R_NilValue;`
469	`}`	477	`}`
470		478
471	`SEXP sdf_get_index(SEXP sdf, SEXP row, SEXP col) {`	479	`SEXP sdf_get_index(SEXP sdf, SEXP row, SEXP col, SEXP new_sdf) {`
472	`SEXP ret = R_NilValue;`	480	`SEXP ret = R_NilValue;`
473	`char *iname = SDF_INAME(sdf);`	481	`char *iname = SDF_INAME(sdf);`
474	`if (!USE_SDF1(iname, TRUE, TRUE)) return R_NilValue;`	-
475		-
476	`sqlite3_stmt *stmt;`	482	`sqlite3_stmt *stmt;`
477	`int buflen = 0, idxlen, col_cnt, row_cnt, index;`	483	`int buflen = 0, idxlen, col_cnt, row_cnt, index;`
478	`int i, j,res;`	484	`int i, j,res;`
479	`int col_indices, col_index_len, dup_indices;`	485	`int col_indices, col_index_len, dup_indices;`
480	`int row_index_len = 0;`	486	`int row_index_len = 0, force_new_df = LOGICAL(new_sdf)[0];`
-		487
-		488	`if (!USE_SDF1(iname, TRUE, TRUE)) return R_NilValue;`
481		489
482	`sprintf(g_sql_buf[0], "select * from [%s].sdf_data ", iname);`	490	`sprintf(g_sql_buf[0], "select * from [%s].sdf_data ", iname);`
483	`res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0);`	491	`res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0);`
484	`if (_sqlite_error(res)) return R_NilValue;`	492	`if (_sqlite_error(res)) return R_NilValue;`
485		493
486	`buflen = sprintf(g_sql_buf[0], "select [row name]");`	494	`buflen = sprintf(g_sql_buf[0], "select [row name]");`
487	`idxlen = LENGTH(col);`	495	`idxlen = LENGTH(col);`
488	`col_cnt = sqlite3_column_count(stmt) - 1;`	496	`col_cnt = sqlite3_column_count(stmt) - 1;`
489		497
490	`/*`	498	`/*`

The R Project SVN R-packages

(root)/trunk/SQLiteDF/src/sqlite_dataframe.c – Rev 3457 → 3471