Esempio n. 1
0
def process_column_remapping(remap_column):
    """

    :param remap_column:
    :return:
    """
    column_remapping = {}
    remap_column = remap_column or []
    for remap in remap_column:
        src, trg = remap.split('=', 1)
        column_remapping[apply_char_replacements(
            src)] = apply_char_replacements(trg)
    return column_remapping
Esempio n. 2
0
def process_table_remapping(remap_table):
    """

    :param remap_table:
    :return:
    """
    table_remapping = {}
    remap_table = remap_table or []
    for remap in remap_table:
        src, trg = remap.split('=', 1)  # TODO: Error handling
        table_remapping[apply_char_replacements(
            src)] = apply_char_replacements(trg)
    return table_remapping
Esempio n. 3
0
def preprocess_filters(filter_args):
    """
    Check and organize filters
    :param filter_args:
    :return:
    """
    filters = {}
    if filter_args:
        for filter_combo in filter_args:
            parts = filter_combo.split('|')
            if len(parts) < 2:
                raise FilterError(f"Invalid filter combo: {filter_combo}")
            col = parts[0]
            if col in filters:
                raise FilterError(f"Multiple filters for column: {col}")
            filters[col] = []
            for f in parts[1:]:
                filter_parts = f.split(":", 1)
                if len(filter_parts) > 1:
                    args = filter_parts.pop(-1)
                    filter_parts.extend(args.split(","))
                filter_name = filter_parts[0]
                if filter_name not in FILTERS:
                    raise FilterError(f"Invalid filter name: {filter_name}")
                filter_parts = [
                    apply_char_replacements(p) for p in filter_parts
                ]
                filters[col].append(filter_parts)
    return filters
Esempio n. 4
0
def main(args=None):
    global DEBUG
    if args is None:
        args = sys.argv[1:]

    parser = build_args_parser()
    args = parser.parse_args(args=args)
    DEBUG = args.debug
    debug(args, 'args=')

    if args.filters_list:
        print_filter_list_table(args.output_format)
        return 0

    if args.replacements_list:
        print_replacements_table(args.output_format)
        return 0

    if not args.sql:
        raise Error("You must specify the SQL to execute.")

    # Process table re-mappings, if any
    table_remapping = process_table_remapping(args.remap_table)
    debug(table_remapping, 'table_remapping=')

    # Pre-process the filters
    filters = preprocess_filters(args.filter)
    debug(filters, 'filters=')

    # Process the column re-mappings, if any
    column_remapping = process_column_remapping(args.remap_column)
    debug(column_remapping, 'column_remapping=')

    # Process delimiters
    input_delimiter = apply_char_replacements(args.input_delimiter)

    execute(args.sql,
            headers=args.headers,
            filters=filters,
            output=args.output,
            output_format=args.output_format,
            skip_lines=args.skip_lines,
            output_delimiter=',',
            column_remapping=column_remapping,
            table_remapping=table_remapping,
            auto_filter=args.auto_filter,
            save_db=args.save_db,
            load_db=args.load_db,
            input_format=args.input_format,
            input_delimiter=input_delimiter,
            input_encoding=args.input_encoding,
            debug_=args.debug)
    return 0
Esempio n. 5
0
def rewrite_sql(sql, table_remap=None):
    """
    Re-write the SQL, replacing @filenames with table names.
    Leave non-@ prefixed table names as-is.
    Handle stdin - and @-
    :param sql:
    :param table_remap:
    :return:
    """
    table_remap = table_remap or {}
    tables, rewrite, i = {}, [], 0
    for s in sql:
        s = apply_char_replacements(s)
        for m in FROM_PATTERN.finditer(s):
            if m.group(2):
                grp, path = 2, m.group(2)
            elif m.group(3):
                grp, path = 3, m.group(3)
            elif m.group(4):
                grp, path = 4, m.group(4)
            else:
                raise Error("Path parsing error.")

            if path != '-':
                path, exists = expand_path_and_exists(path)
                if not exists:
                    raise FileNotFoundError(f"File not found: {path}")

            rewrite.append(s[i:m.start(grp) -
                             (2 if grp == 2 else 1 if grp == 3 else 0)])
            i = m.end(grp) + (1 if grp == 2 else 0)

            if path != '-':
                filename = os.path.basename(path)
                tablename = os.path.splitext(filename)[0]
            else:
                filename = '-'
                tablename = 'stdin'

            if path in table_remap:
                tablename = table_remap[path]
            elif filename in table_remap:
                tablename = table_remap[filename]
            elif tablename in table_remap:
                tablename = table_remap[tablename]

            rewrite.append(f'"{tablename}"')
            tables[tablename] = path

        rewrite.append(s[i:])

    return ''.join(rewrite), tables
Esempio n. 6
0
def rewrite_sql(sql, table_remap=None):
    """
    Re-write the SQL, replacing @filenames with table names.
    Leave non-@ prefixed table names as-is.
    Handle stdin - and @-
    :param sql:
    :param table_remap:
    :return:
    """
    table_remap = table_remap or {}
    tables, rewrite, i = {}, [], 0
    for s in sql:
        # print(s)
        s = apply_char_replacements(s)
        for m in FROM_PATTERN.finditer(s):
            # print(m, m.groups())
            if m.group(2):
                grp, path = 2, m.group(2)
            elif m.group(3):
                grp, path = 3, m.group(3)
            elif m.group(4):
                grp, path = 4, m.group(4)
            else:
                raise Error("Path parsing error.")

            # print(path)
            if path != '-':
                parse_result = urlparse(path)
                scheme = parse_result.scheme
                # print(repr(scheme))
                if scheme in {'http', 'https'}:
                    pass
                elif scheme == 's3':
                    pass
                elif scheme == 'gs':
                    pass
                elif scheme in {'file', ''}:
                    path = parse_result.path
                    path, exists = expand_path_and_exists(path)
                    if not exists:
                        raise FileNotFoundError(f"File not found: {path}")
                else:
                    raise Error("Invalid URL scheme: {scheme}")

            rewrite.append(s[i:m.start(grp) -
                             (2 if grp == 2 else 1 if grp == 3 else 0)])
            i = m.end(grp) + (1 if grp == 2 else 0)

            if path != '-':
                filename = os.path.basename(path)
                tablename = os.path.splitext(filename)[0]
            else:
                filename = '-'
                tablename = 'stdin'

            if path in table_remap:
                tablename = table_remap[path]
            elif filename in table_remap:
                tablename = table_remap[filename]
            elif tablename in table_remap:
                tablename = table_remap[tablename]

            if tablename.upper() in RESERVED_WORDS:
                sys.stderr.write(
                    f"Warning: Table name {tablename} is a SQLite reserved word."
                )

            rewrite.append(f'"{tablename}"')
            tables[tablename] = path

        rewrite.append(s[i:])

    return ''.join(rewrite), tables