django1/django/db/backends/mysql/introspection.py

216 lines
8.9 KiB
Python

from collections import namedtuple
from MySQLdb.constants import FIELD_TYPE
from django.db.backends.base.introspection import (
BaseDatabaseIntrospection, FieldInfo, TableInfo,
)
from django.utils.datastructures import OrderedSet
from django.utils.encoding import force_text
FieldInfo = namedtuple('FieldInfo', FieldInfo._fields + ('extra', 'default'))
InfoLine = namedtuple('InfoLine', 'col_name data_type max_len num_prec num_scale extra column_default')
class DatabaseIntrospection(BaseDatabaseIntrospection):
data_types_reverse = {
FIELD_TYPE.BLOB: 'TextField',
FIELD_TYPE.CHAR: 'CharField',
FIELD_TYPE.DECIMAL: 'DecimalField',
FIELD_TYPE.NEWDECIMAL: 'DecimalField',
FIELD_TYPE.DATE: 'DateField',
FIELD_TYPE.DATETIME: 'DateTimeField',
FIELD_TYPE.DOUBLE: 'FloatField',
FIELD_TYPE.FLOAT: 'FloatField',
FIELD_TYPE.INT24: 'IntegerField',
FIELD_TYPE.LONG: 'IntegerField',
FIELD_TYPE.LONGLONG: 'BigIntegerField',
FIELD_TYPE.SHORT: 'SmallIntegerField',
FIELD_TYPE.STRING: 'CharField',
FIELD_TYPE.TIME: 'TimeField',
FIELD_TYPE.TIMESTAMP: 'DateTimeField',
FIELD_TYPE.TINY: 'IntegerField',
FIELD_TYPE.TINY_BLOB: 'TextField',
FIELD_TYPE.MEDIUM_BLOB: 'TextField',
FIELD_TYPE.LONG_BLOB: 'TextField',
FIELD_TYPE.VAR_STRING: 'CharField',
}
def get_field_type(self, data_type, description):
field_type = super(DatabaseIntrospection, self).get_field_type(data_type, description)
if 'auto_increment' in description.extra:
if field_type == 'IntegerField':
return 'AutoField'
elif field_type == 'BigIntegerField':
return 'BigAutoField'
return field_type
def get_table_list(self, cursor):
"""
Returns a list of table and view names in the current database.
"""
cursor.execute("SHOW FULL TABLES")
return [TableInfo(row[0], {'BASE TABLE': 't', 'VIEW': 'v'}.get(row[1]))
for row in cursor.fetchall()]
def get_table_description(self, cursor, table_name):
"""
Returns a description of the table, with the DB-API cursor.description interface."
"""
# information_schema database gives more accurate results for some figures:
# - varchar length returned by cursor.description is an internal length,
# not visible length (#5725)
# - precision and scale (for decimal fields) (#5014)
# - auto_increment is not available in cursor.description
cursor.execute("""
SELECT column_name, data_type, character_maximum_length, numeric_precision,
numeric_scale, extra, column_default
FROM information_schema.columns
WHERE table_name = %s AND table_schema = DATABASE()""", [table_name])
field_info = {line[0]: InfoLine(*line) for line in cursor.fetchall()}
cursor.execute("SELECT * FROM %s LIMIT 1" % self.connection.ops.quote_name(table_name))
def to_int(i):
return int(i) if i is not None else i
fields = []
for line in cursor.description:
col_name = force_text(line[0])
fields.append(
FieldInfo(*((col_name,)
+ line[1:3]
+ (to_int(field_info[col_name].max_len) or line[3],
to_int(field_info[col_name].num_prec) or line[4],
to_int(field_info[col_name].num_scale) or line[5])
+ (line[6],)
+ (field_info[col_name].extra,)
+ (field_info[col_name].column_default,)))
)
return fields
def get_relations(self, cursor, table_name):
"""
Returns a dictionary of {field_name: (field_name_other_table, other_table)}
representing all relationships to the given table.
"""
constraints = self.get_key_columns(cursor, table_name)
relations = {}
for my_fieldname, other_table, other_field in constraints:
relations[my_fieldname] = (other_field, other_table)
return relations
def get_key_columns(self, cursor, table_name):
"""
Returns a list of (column_name, referenced_table_name, referenced_column_name) for all
key columns in given table.
"""
key_columns = []
cursor.execute("""
SELECT column_name, referenced_table_name, referenced_column_name
FROM information_schema.key_column_usage
WHERE table_name = %s
AND table_schema = DATABASE()
AND referenced_table_name IS NOT NULL
AND referenced_column_name IS NOT NULL""", [table_name])
key_columns.extend(cursor.fetchall())
return key_columns
def get_indexes(self, cursor, table_name):
cursor.execute("SHOW INDEX FROM %s" % self.connection.ops.quote_name(table_name))
# Do a two-pass search for indexes: on first pass check which indexes
# are multicolumn, on second pass check which single-column indexes
# are present.
rows = list(cursor.fetchall())
multicol_indexes = set()
for row in rows:
if row[3] > 1:
multicol_indexes.add(row[2])
indexes = {}
for row in rows:
if row[2] in multicol_indexes:
continue
if row[4] not in indexes:
indexes[row[4]] = {'primary_key': False, 'unique': False}
# It's possible to have the unique and PK constraints in separate indexes.
if row[2] == 'PRIMARY':
indexes[row[4]]['primary_key'] = True
if not row[1]:
indexes[row[4]]['unique'] = True
return indexes
def get_storage_engine(self, cursor, table_name):
"""
Retrieves the storage engine for a given table. Returns the default
storage engine if the table doesn't exist.
"""
cursor.execute(
"SELECT engine "
"FROM information_schema.tables "
"WHERE table_name = %s", [table_name])
result = cursor.fetchone()
if not result:
return self.connection.features._mysql_storage_engine
return result[0]
def get_constraints(self, cursor, table_name):
"""
Retrieves any constraints or keys (unique, pk, fk, check, index) across one or more columns.
"""
constraints = {}
# Get the actual constraint names and columns
name_query = """
SELECT kc.`constraint_name`, kc.`column_name`,
kc.`referenced_table_name`, kc.`referenced_column_name`
FROM information_schema.key_column_usage AS kc
WHERE
kc.table_schema = %s AND
kc.table_name = %s
"""
cursor.execute(name_query, [self.connection.settings_dict['NAME'], table_name])
for constraint, column, ref_table, ref_column in cursor.fetchall():
if constraint not in constraints:
constraints[constraint] = {
'columns': OrderedSet(),
'primary_key': False,
'unique': False,
'index': False,
'check': False,
'foreign_key': (ref_table, ref_column) if ref_column else None,
}
constraints[constraint]['columns'].add(column)
# Now get the constraint types
type_query = """
SELECT c.constraint_name, c.constraint_type
FROM information_schema.table_constraints AS c
WHERE
c.table_schema = %s AND
c.table_name = %s
"""
cursor.execute(type_query, [self.connection.settings_dict['NAME'], table_name])
for constraint, kind in cursor.fetchall():
if kind.lower() == "primary key":
constraints[constraint]['primary_key'] = True
constraints[constraint]['unique'] = True
elif kind.lower() == "unique":
constraints[constraint]['unique'] = True
# Now add in the indexes
cursor.execute("SHOW INDEX FROM %s" % self.connection.ops.quote_name(table_name))
for table, non_unique, index, colseq, column in [x[:5] for x in cursor.fetchall()]:
if index not in constraints:
constraints[index] = {
'columns': OrderedSet(),
'primary_key': False,
'unique': False,
'index': True,
'check': False,
'foreign_key': None,
}
constraints[index]['index'] = True
constraints[index]['columns'].add(column)
# Convert the sorted sets to lists
for constraint in constraints.values():
constraint['columns'] = list(constraint['columns'])
return constraints