X-Git-Url: https://git.jsancho.org/?p=datasette-pytables.git;a=blobdiff_plain;f=datasette_pytables%2F__init__.py;h=a730cb7e72daae411a7c4c791e759fdd1a4d29a1;hp=93be0f9a1206208f3667633526baca08b28ae335;hb=d46bbde71b66e3c6c76e41cd303a5d5ea23db0c4;hpb=25d0571fb4a85291df0f7cea4f1424588a7961fc diff --git a/datasette_pytables/__init__.py b/datasette_pytables/__init__.py index 93be0f9..a730cb7 100644 --- a/datasette_pytables/__init__.py +++ b/datasette_pytables/__init__.py @@ -1,97 +1,239 @@ import tables -import sqlparse -from collections import OrderedDict +import datasette_connectors as dc +from .utils import parse_sql -_connector_type = 'pytables' -def inspect(path): - "Open file and return tables info" - h5tables = {} - views = [] - h5file = tables.open_file(path) +class PyTablesConnection(dc.Connection): + def __init__(self, path, connector): + super().__init__(path, connector) + self.h5file = tables.open_file(path) + + +class PyTablesConnector(dc.Connector): + connector_type = 'pytables' + connection_class = PyTablesConnection + + operators = { + 'eq': '==', + 'neq': '!=', + 'gt': '>', + 'gte': '>=', + 'lt': '<', + 'lte': '<=', + 'and': '&', + 'or': '|', + } + + def table_names(self): + return [ + node._v_pathname + for node in self.conn.h5file + if not(isinstance(node, tables.group.Group)) + ] - for table in filter(lambda node: not(isinstance(node, tables.group.Group)), h5file): - colnames = [] + def table_count(self, table_name): + table = self.conn.h5file.get_node(table_name) + return int(table.nrows) + + def table_info(self, table_name): + table = self.conn.h5file.get_node(table_name) + colnames = ['value'] if isinstance(table, tables.table.Table): colnames = table.colnames - h5tables[table._v_pathname] = { - 'name': table._v_pathname, - 'columns': colnames, - 'primary_keys': [], - 'count': table.nrows, - 'label_column': None, - 'hidden': False, - 'fts_table': None, - 'foreign_keys': {'incoming': [], 'outgoing': []}, - } - - h5file.close() - return h5tables, views, _connector_type - -def _parse_sql(sql): - parsed = sqlparse.parse(sql) - stmt = parsed[0] - parsed_sql = {} - current_keyword = "" - for token in stmt.tokens: - if token.is_keyword: - if current_keyword in parsed_sql and parsed_sql[current_keyword] == '': - # Check composed keywords like 'order by' - del parsed_sql[current_keyword] - current_keyword += " " + str(token) - else: - current_keyword = str(token) - parsed_sql[current_keyword] = "" - else: - if not token.is_whitespace: - parsed_sql[current_keyword] += str(token) - return parsed_sql + return [ + { + 'idx': idx, + 'name': colname, + 'primary_key': False, + } + for idx, colname in enumerate(colnames) + ] -class Connection: - def __init__(self, path): - self.path = path - self.h5file = tables.open_file(path) + def hidden_table_names(self): + return [] - def execute(self, sql, params=None, truncate=False): - rows = [] + def detect_spatialite(self): + return False + + def view_names(self): + return [] + + def detect_fts(self, table_name): + return False + + def foreign_keys(self, table_name): + return [] + + def execute( + self, + sql, + params=None, + truncate=False, + custom_time_limit=None, + page_size=None, + log_sql_errors=True, + ): + results = [] truncated = False - description = [] + description = () + + parsed_sql = parse_sql(sql, params) - parsed_sql = _parse_sql(sql) - table = self.h5file.get_node(parsed_sql['from'][1:-1]) + table = self.conn.h5file.get_node(parsed_sql['from']) table_rows = [] - fields = parsed_sql['select'].split(',') + fields = parsed_sql['select'] + colnames = ['value'] + if type(table) is tables.table.Table: + colnames = table.colnames + + query = '' + start = 0 + end = table.nrows # Use 'where' statement or get all the rows + def _cast_param(field, pname): + # Cast value to the column type + coltype = table.dtype.name + if type(table) is tables.table.Table: + coltype = table.coltypes[field] + fcast = None + if coltype == 'string': + fcast = str + elif coltype.startswith('int'): + fcast = int + elif coltype.startswith('float'): + fcast = float + if fcast: + params[pname] = fcast(params[pname]) + + def _translate_where(where): + # Translate SQL to PyTables expression + nonlocal start, end + expr = '' + operator = list(where)[0] + + if operator in ['and', 'or']: + subexpr = [_translate_where(e) for e in where[operator]] + subexpr = filter(lambda e: e, subexpr) + subexpr = ["({})".format(e) for e in subexpr] + expr = " {} ".format(self.operators[operator]).join(subexpr) + elif operator == 'exists': + pass + elif where == {'eq': ['rowid', 'p0']}: + start = int(params['p0']) + end = start + 1 + elif where == {'gt': ['rowid', 'p0']}: + start = int(params['p0']) + 1 + else: + left, right = where[operator] + if left in params: + _cast_param(right, left) + elif right in params: + _cast_param(left, right) + + expr = "{left} {operator} {right}".format(left=left, operator=self.operators.get(operator, operator), right=right) + + return expr + if 'where' in parsed_sql: - pass + if type(parsed_sql['where']) is dict: + query = _translate_where(parsed_sql['where']) + else: + query = parsed_sql['where'] + + # Sort by column + orderby = '' + if 'orderby' in parsed_sql: + orderby = parsed_sql['orderby'] + if type(orderby) is list: + orderby = orderby[0] + orderby = orderby['value'] + if orderby == 'rowid': + orderby = '' + + # Limit number of rows + limit = None + if 'limit' in parsed_sql: + limit = int(parsed_sql['limit']) + + # Truncate if needed + if page_size and max_returned_rows and truncate: + if max_returned_rows == page_size: + max_returned_rows += 1 + + # Execute query + if query: + table_rows = table.where(query, params, start, end) + elif orderby: + table_rows = table.itersorted(orderby, start=start, stop=end) else: - table_rows = table.iterrows() + table_rows = table.iterrows(start, end) - if len(fields) == 1 and fields[0] == 'count(*)': - rows.append(Row({'count(*)': table.nrows})) + # Prepare rows + def normalize_field_value(value): + if type(value) is bytes: + return value.decode('utf-8') + elif not type(value) in (int, float, complex): + return str(value) + else: + return value + + def make_get_rowid(): + if type(table) is tables.table.Table: + def get_rowid(row): + return int(row.nrow) + else: + rowid = start - 1 + def get_rowid(row): + nonlocal rowid + rowid += 1 + return rowid + return get_rowid + + def make_get_row_value(): + if type(table) is tables.table.Table: + def get_row_value(row, field): + return row[field] + else: + def get_row_value(row, field): + return row + return get_row_value + + if len(fields) == 1 and type(fields[0]['value']) is dict and \ + fields[0]['value'].get('count') == '*': + results.append({'count(*)': int(table.nrows)}) else: + get_rowid = make_get_rowid() + get_row_value = make_get_row_value() + count = 0 for table_row in table_rows: - row = Row() + count += 1 + if limit and count > limit: + break + if truncate and max_returned_rows and count > max_returned_rows: + truncated = True + break + row = {} for field in fields: - if field == 'rowid': - row[field] = table_row.nrow - elif field == '*': - for col in table.colnames: - row[col] = table_row[col] + field_name = field['value'] + if type(field_name) is dict and 'distinct' in field_name: + field_name = field_name['distinct'] + if field_name == 'rowid': + row['rowid'] = get_rowid(table_row) + elif field_name == '*': + for col in colnames: + row[col] = normalize_field_value(get_row_value(table_row, col)) else: - row[field] = table_row[field] - rows.append(row) - - description = ((col,) for col in table.colnames) + row[field_name] = normalize_field_value(get_row_value(table_row, field_name)) + results.append(row) - if truncate: - return rows, truncated, description - else: - return rows + # Prepare query description + for field in [f['value'] for f in fields]: + if field == '*': + for col in colnames: + description += ((col,),) + else: + description += ((field,),) -class Row(OrderedDict): - def __getitem__(self, label): - if type(label) is int: - return super(OrderedDict, self).__getitem__(list(self.keys())[label]) + return results, truncated, description