-from collections import OrderedDict
-from moz_sql_parser import parse
-import re
import tables
+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)
+
- for table in filter(lambda node: not(isinstance(node, tables.group.Group)), h5file):
- colnames = []
+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))
+ ]
+
+ 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': int(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, params):
- # Table name
- sql = re.sub('(?i)from \[(.*)]', 'from "\g<1>"', sql)
- # Params
- for param in params:
- sql = sql.replace(":" + param, param)
-
- try:
- parsed = parse(sql)
- except:
- # Propably it's a PyTables expression
- for token in ['group by', 'order by', 'limit', '']:
- res = re.search('(?i)where (.*)' + token, sql)
- if res:
- modified_sql = re.sub('(?i)where (.*)(' + token + ')', '\g<2>', sql)
- parsed = parse(modified_sql)
- parsed['where'] = res.group(1).strip()
- break
-
- # Always a list of fields
- if type(parsed['select']) is not list:
- parsed['select'] = [parsed['select']]
-
- return parsed
-
-_operators = {
- 'eq': '==',
- 'neq': '!=',
- 'gt': '>',
- 'gte': '>=',
- 'lt': '<',
- 'lte': '<=',
- 'and': '&',
- 'or': '|',
-}
-
-class Connection:
- def __init__(self, path):
- self.path = path
- self.h5file = tables.open_file(path)
-
- def execute(self, sql, params=None, truncate=False, page_size=None):
- if params is None:
- params = {}
- rows = []
+ return [
+ {
+ 'idx': idx,
+ 'name': colname,
+ 'primary_key': False,
+ }
+ for idx, colname in enumerate(colnames)
+ ]
+
+ def hidden_table_names(self):
+ return []
+
+ 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, params)
- table = self.h5file.get_node(parsed_sql['from'])
+ table = self.conn.h5file.get_node(parsed_sql['from'])
table_rows = []
fields = parsed_sql['select']
+ colnames = ['value']
+ if type(table) is tables.table.Table:
+ colnames = table.colnames
query = ''
start = 0
# Use 'where' statement or get all the rows
def _cast_param(field, pname):
# Cast value to the column type
- coltype = table.coltypes[field]
+ coltype = table.dtype.name
+ if type(table) is tables.table.Table:
+ coltype = table.coltypes[field]
fcast = None
if coltype == 'string':
fcast = str
def _translate_where(where):
# Translate SQL to PyTables expression
+ nonlocal start, end
expr = ''
operator = list(where)[0]
subexpr = [_translate_where(e) for e in where[operator]]
subexpr = filter(lambda e: e, subexpr)
subexpr = ["({})".format(e) for e in subexpr]
- expr = " {} ".format(_operators[operator]).join(subexpr)
+ expr = " {} ".format(self.operators[operator]).join(subexpr)
elif operator == 'exists':
pass
elif where == {'eq': ['rowid', 'p0']}:
- nonlocal start, end
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:
elif right in params:
_cast_param(left, right)
- expr = "{left} {operator} {right}".format(left=left, operator=_operators.get(operator, operator), right=right)
+ expr = "{left} {operator} {right}".format(left=left, operator=self.operators.get(operator, operator), right=right)
return expr
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:
- max_rows = int(parsed_sql['limit'])
- if end - start > max_rows:
- end = start + max_rows
+ limit = int(parsed_sql['limit'])
# Truncate if needed
- if page_size and truncate:
- if end - start > page_size:
- end = start + page_size
- truncated = True
+ 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(start, end)
# Prepare rows
- if len(fields) == 1 and type(fields[0]['value']) is dict and \
- fields[0]['value'].get('count') == '*':
- rows.append(Row({'count(*)': int(table.nrows)}))
- else:
+ 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:
- for table_row in table_rows:
- row = Row()
- for field in fields:
- 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'] = int(table_row.nrow)
- elif field_name == '*':
- for col in table.colnames:
- value = table_row[col]
- if type(value) is bytes:
- value = value.decode('utf-8')
- row[col] = value
- else:
- row[field_name] = table_row[field_name]
- rows.append(row)
+ def get_rowid(row):
+ return int(row.nrow)
else:
- # Any kind of array
rowid = start - 1
- for table_row in table_rows:
- row = Row()
+ def get_rowid(row):
+ nonlocal rowid
rowid += 1
- for field in fields:
- 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'] = rowid
- else:
- value = table_row
- if type(value) is bytes:
- value = value.decode('utf-8')
- row['value'] = value
- rows.append(row)
+ return rowid
+ return get_rowid
- # Prepare query description
- for field in [f['value'] for f in fields]:
- if field == '*':
- if type(table) is tables.table.Table:
- for col in table.colnames:
- description.append((col,))
- else:
- description.append(('value',))
+ def make_get_row_value():
+ if type(table) is tables.table.Table:
+ def get_row_value(row, field):
+ return row[field]
else:
- description.append((field,))
+ def get_row_value(row, field):
+ return row
+ return get_row_value
- # Return the rows
- if truncate:
- return rows, truncated, tuple(description)
+ if len(fields) == 1 and type(fields[0]['value']) is dict and \
+ fields[0]['value'].get('count') == '*':
+ results.append({'count(*)': int(table.nrows)})
else:
- return rows
+ get_rowid = make_get_rowid()
+ get_row_value = make_get_row_value()
+ count = 0
+ for table_row in table_rows:
+ 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:
+ 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_name] = normalize_field_value(get_row_value(table_row, field_name))
+ results.append(row)
-class Row(OrderedDict):
- def __getitem__(self, label):
- if type(label) is int:
- return super(OrderedDict, self).__getitem__(list(self.keys())[label])
- else:
- return super(OrderedDict, self).__getitem__(label)
+ # Prepare query description
+ for field in [f['value'] for f in fields]:
+ if field == '*':
+ for col in colnames:
+ description += ((col,),)
+ else:
+ description += ((field,),)
- def __iter__(self):
- return self.values().__iter__()
+ return results, truncated, description
projects / datasette-pytables.git / blobdiff