-from collections import OrderedDict
from moz_sql_parser import parse
import re
import tables
h5file = tables.open_file(path)
for table in filter(lambda node: not(isinstance(node, tables.group.Group)), h5file):
- colnames = []
+ colnames = ['value']
if isinstance(table, tables.table.Table):
colnames = table.colnames
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']
+ 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]
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:
query = parsed_sql['where']
# 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 max_returned_rows and truncate:
+ if max_returned_rows == page_size:
+ max_returned_rows += 1
# Execute query
if query:
table_rows = table.iterrows(start, end)
# 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') == '*':
rows.append(Row({'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:
+ count += 1
+ if limit and count > limit:
+ break
+ if truncate and max_returned_rows and count > max_returned_rows:
+ truncated = True
+ break
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(value) is bytes:
- value = value.decode('utf-8')
- row[col] = value
+ 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['value']] = table_row[field['value']]
+ row[field_name] = normalize_field_value(get_row_value(table_row, field_name))
rows.append(row)
# Prepare query description
for field in [f['value'] for f in fields]:
if field == '*':
- for col in table.colnames:
+ for col in colnames:
description.append((col,))
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__()
projects / datasette-pytables.git / blobdiff