datasette_pytables/__init__.py

   1 from collections import OrderedDict
   2 from moz_sql_parser import parse
   3 import re
   4 import tables
   5
   6 _connector_type = 'pytables'
   7
   8 def inspect(path):
   9     "Open file and return tables info"
  10     h5tables = {}
  11     views = []
  12     h5file = tables.open_file(path)
  13
  14     for table in filter(lambda node: not(isinstance(node, tables.group.Group)), h5file):
  15         colnames = []
  16         if isinstance(table, tables.table.Table):
  17             colnames = table.colnames
  18
  19         h5tables[table._v_pathname] = {
  20             'name': table._v_pathname,
  21             'columns': colnames,
  22             'primary_keys': [],
  23             'count': int(table.nrows),
  24             'label_column': None,
  25             'hidden': False,
  26             'fts_table': None,
  27             'foreign_keys': {'incoming': [], 'outgoing': []},
  28         }
  29
  30     h5file.close()
  31     return h5tables, views, _connector_type
  32
  33 def _parse_sql(sql, params):
  34     # Table name
  35     sql = re.sub('(?i)from \[(.*)]', 'from "\g<1>"', sql)
  36     # Params
  37     for param in params:
  38         sql = sql.replace(":" + param, param)
  39
  40     parsed = parse(sql)
  41     if type(parsed['select']) is not list:
  42         parsed['select'] = [parsed['select']]
  43
  44     return parsed
  45
  46 _operators = {
  47     'eq': '==',
  48     'neq': '!=',
  49     'gt': '>',
  50     'gte': '>=',
  51     'lt': '<',
  52     'lte': '<=',
  53     'and': '&',
  54     'or': '|',
  55 }
  56
  57 class Connection:
  58     def __init__(self, path):
  59         self.path = path
  60         self.h5file = tables.open_file(path)
  61
  62     def execute(self, sql, params=None, truncate=False):
  63         if params is None:
  64             params = {}
  65         rows = []
  66         truncated = False
  67         description = []
  68
  69         parsed_sql = _parse_sql(sql, params)
  70         table = self.h5file.get_node(parsed_sql['from'])
  71         table_rows = []
  72         fields = parsed_sql['select']
  73
  74         query = ''
  75         start = 0
  76         end = table.nrows
  77
  78         # Use 'where' statement or get all the rows
  79         def _cast_param(field, pname):
  80             # Cast value to the column type
  81             coltype = table.coltypes[field]
  82             fcast = None
  83             if coltype == 'string':
  84                 fcast = str
  85             elif coltype.startswith('int'):
  86                 fcast = int
  87             elif coltype.startswith('float'):
  88                 fcast = float
  89             if fcast:
  90                 params[pname] = fcast(params[pname])
  91
  92         def _translate_where(where):
  93             # Translate SQL to PyTables expression
  94             expr = ''
  95             operator = list(where)[0]
  96
  97             if operator in ['and', 'or']:
  98                 subexpr = ["({})".format(_translate_where(q)) for q in where[operator]]
  99                 expr = " {} ".format(_operators[operator]).join(subexpr)
 100             elif where == {'eq': ['rowid', 'p0']}:
 101                 nonlocal start, end
 102                 start = int(params['p0'])
 103                 end = start + 1
 104             else:
 105                 left, right = where[operator]
 106                 if left in params:
 107                     _cast_param(right, left)
 108                 elif right in params:
 109                     _cast_param(left, right)
 110
 111                 expr = "{left} {operator} {right}".format(left=left, operator=_operators.get(operator, operator), right=right)
 112
 113             return expr
 114
 115         if 'where' in parsed_sql:
 116             try:
 117                 query = _translate_where(parsed_sql['where'])
 118             except:
 119                 # Probably it's a PyTables query
 120                 query = str(parsed_sql['where'])[6:]    # without where keyword
 121
 122         # Limit number of rows
 123         if 'limit' in parsed_sql:
 124             max_rows = int(parsed_sql['limit'])
 125             if end - start > max_rows:
 126                 end = start + max_rows
 127
 128         # Execute query
 129         if query:
 130             table_rows = table.where(query, params, start, end)
 131         else:
 132             table_rows = table.iterrows(start, end)
 133
 134         # Prepare rows
 135         if len(fields) == 1 and type(fields[0]['value']) is dict and \
 136            fields[0]['value'].get('count') == '*':
 137             rows.append(Row({'count(*)': int(table.nrows)}))
 138         else:
 139             for table_row in table_rows:
 140                 row = Row()
 141                 for field in fields:
 142                     if field['value'] == 'rowid':
 143                         row['rowid'] = int(table_row.nrow)
 144                     elif field['value'] == '*':
 145                         for col in table.colnames:
 146                             value = table_row[col]
 147                             if type(value) is bytes:
 148                                 value = value.decode('utf-8')
 149                             row[col] = value
 150                     else:
 151                         row[field['value']] = table_row[field['value']]
 152                 rows.append(row)
 153
 154         # Prepare query description
 155         for field in [f['value'] for f in fields]:
 156             if field == '*':
 157                 for col in table.colnames:
 158                     description.append((col,))
 159             else:
 160                 description.append((field,))
 161
 162         # Return the rows
 163         if truncate:
 164             return rows, truncated, tuple(description)
 165         else:
 166             return rows
 167
 168 class Row(OrderedDict):
 169     def __getitem__(self, label):
 170         if type(label) is int:
 171             return super(OrderedDict, self).__getitem__(list(self.keys())[label])
 172         else:
 173             return super(OrderedDict, self).__getitem__(label)
 174
 175     def __iter__(self):
 176         return self.values().__iter__()