-from collections import OrderedDict
from moz_sql_parser import parse
import re
+
import tables
+import datasette_connectors as dc
+
+
+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
+
+ 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
+
+ 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
-_connector_type = 'pytables'
def inspect(path):
"Open file and return tables info"
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
def _parse_sql(sql, params):
# Table name
- sql = re.sub('(?i)from \[(.*)]', 'from "\g<1>"', sql)
+ sql = re.sub(r'(?i)from \[(.*)]', r'from "\g<1>"', sql)
# Params
for param in params:
sql = sql.replace(":" + param, param)
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)
+ modified_sql = re.sub('(?i)where (.*)(' + token + ')', r'\g<2>', sql)
parsed = parse(modified_sql)
parsed['where'] = res.group(1).strip()
break
self.path = path
self.h5file = tables.open_file(path)
- def execute(self, sql, params=None, truncate=False, page_size=None):
+ def execute(self, sql, params=None, truncate=False, page_size=None, max_returned_rows=None):
if params is None:
params = {}
rows = []
parsed_sql = _parse_sql(sql, params)
if parsed_sql['from'] == 'sqlite_master':
- return self._execute_datasette_query(sql, params)
+ 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]
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:
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
+
+ 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:
+ 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))
+ rows.append(row)
# 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',))
+ for col in colnames:
+ description.append((col,))
else:
description.append((field,))
# Return the rows
- if truncate:
- return rows, truncated, tuple(description)
- else:
- return rows
+ 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:
+ if sql == 'select sql from sqlite_master where name = :n and type=:t':
+ if params['t'] == 'view':
return []
+ else:
+ try:
+ table = self.h5file.get_node(params['n'])
+ colnames = ['value']
+ if type(table) is tables.table.Table:
+ colnames = table.colnames
+ row = Row()
+ row['sql'] = 'CREATE TABLE {} ({})'.format(params['n'], ", ".join(colnames))
+ return [row]
+ except:
+ return []
else:
- raise Exception("SQLite queries cannot be executed with this connector")
+ raise Exception("SQLite queries cannot be executed with this connector: %s, %s" % (sql, params))
+
class Row(list):
def __init__(self, values=None):
projects / datasette-pytables.git / blobdiff