-from collections import OrderedDict
from moz_sql_parser import parse
import re
import tables
parsed = parse(sql)
except:
# Propably it's a PyTables expression
- for token in ['group by', 'order by', 'limit']:
+ 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)
+ parsed['where'] = res.group(1).strip()
break
# Always a list of fields
self.path = path
self.h5file = tables.open_file(path)
- def execute(self, sql, params=None, truncate=False):
+ def execute(self, sql, params=None, truncate=False, page_size=None, max_returned_rows=None):
if params is None:
params = {}
rows = []
description = []
parsed_sql = _parse_sql(sql, params)
+
+ if parsed_sql['from'] == 'sqlite_master':
+ rows = self._execute_datasette_query(sql, params)
+ description = (('value',))
+ return rows, truncated, description
+
table = self.h5file.get_node(parsed_sql['from'])
table_rows = []
fields = parsed_sql['select']
# Use 'where' statement or get all the rows
def _cast_param(field, pname):
# Cast value to the column type
- coltype = table.coltypes[field]
+ if type(table) is tables.table.Table:
+ coltype = table.coltypes[field]
+ else:
+ coltype = table.dtype.name
fcast = None
if coltype == 'string':
fcast = str
def _translate_where(where):
# Translate SQL to PyTables expression
+ nonlocal start, end
expr = ''
operator = list(where)[0]
if operator in ['and', 'or']:
- subexpr = ["({})".format(_translate_where(q)) for q in where[operator]]
+ 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)
+ 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:
if end - start > max_rows:
end = start + max_rows
+ # Truncate if needed
+ if page_size and max_returned_rows and truncate:
+ if max_returned_rows == page_size:
+ max_returned_rows += 1
+ if end - start > max_returned_rows:
+ end = start + max_returned_rows
+ truncated = True
+
# Execute query
if query:
table_rows = table.where(query, params, start, end)
fields[0]['value'].get('count') == '*':
rows.append(Row({'count(*)': int(table.nrows)}))
else:
- for table_row in table_rows:
- row = Row()
- for field in fields:
- if field['value'] == 'rowid':
- row['rowid'] = int(table_row.nrow)
- elif field['value'] == '*':
- for col in table.colnames:
- value = table_row[col]
+ 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)
+ else:
+ # Any kind of array
+ rowid = start - 1
+ for table_row in table_rows:
+ row = Row()
+ 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[col] = value
- else:
- row[field['value']] = table_row[field['value']]
- rows.append(row)
+ row['value'] = value
+ rows.append(row)
# Prepare query description
for field in [f['value'] for f in fields]:
if field == '*':
- for col in table.colnames:
- description.append((col,))
+ if type(table) is tables.table.Table:
+ for col in table.colnames:
+ description.append((col,))
+ else:
+ description.append(('value',))
else:
description.append((field,))
# Return the rows
- if truncate:
- return rows, truncated, tuple(description)
+ return rows, truncated, tuple(description)
+
+ def _execute_datasette_query(self, sql, params):
+ "Datasette special queries for getting tables info"
+ if sql == "SELECT count(*) from sqlite_master WHERE type = 'view' and name=:n":
+ row = Row()
+ row['count(*)'] = 0
+ return [row]
+ elif sql == 'select sql from sqlite_master where name = :n and type="table"':
+ try:
+ table = self.h5file.get_node(params['n'])
+ row = Row()
+ row['sql'] = 'CREATE TABLE {} ()'.format(params['n'])
+ return [row]
+ except:
+ return []
+ else:
+ raise Exception("SQLite queries cannot be executed with this connector")
+
+class Row(list):
+ def __init__(self, values=None):
+ self.labels = []
+ self.values = []
+ if values:
+ for idx in values:
+ self.__setitem__(idx, values[idx])
+
+ def __setitem__(self, idx, value):
+ if type(idx) is str:
+ if idx in self.labels:
+ self.values[self.labels.index(idx)] = value
+ else:
+ self.labels.append(idx)
+ self.values.append(value)
else:
- return rows
+ self.values[idx] = value
-class Row(OrderedDict):
- def __getitem__(self, label):
- if type(label) is int:
- return super(OrderedDict, self).__getitem__(list(self.keys())[label])
+ def __getitem__(self, idx):
+ if type(idx) is str:
+ return self.values[self.labels.index(idx)]
else:
- return super(OrderedDict, self).__getitem__(label)
+ return self.values[idx]
def __iter__(self):
- return self.values().__iter__()
+ return self.values.__iter__()