| 1 |
11
|
aaronmk
|
# Database access
|
| 2 |
|
|
|
| 3 |
|
|
import random
|
| 4 |
|
|
import re
|
| 5 |
|
|
import sys
|
| 6 |
865
|
aaronmk
|
import warnings
|
| 7 |
11
|
aaronmk
|
|
| 8 |
300
|
aaronmk
|
import exc
|
| 9 |
862
|
aaronmk
|
import strings
|
| 10 |
131
|
aaronmk
|
import util
|
| 11 |
11
|
aaronmk
|
|
| 12 |
832
|
aaronmk
|
##### Exceptions
|
| 13 |
|
|
|
| 14 |
135
|
aaronmk
|
def get_cur_query(cur):
|
| 15 |
|
|
if hasattr(cur, 'query'): return cur.query
|
| 16 |
|
|
elif hasattr(cur, '_last_executed'): return cur._last_executed
|
| 17 |
|
|
else: return None
|
| 18 |
14
|
aaronmk
|
|
| 19 |
300
|
aaronmk
|
def _add_cursor_info(e, cur): exc.add_msg(e, 'query: '+get_cur_query(cur))
|
| 20 |
135
|
aaronmk
|
|
| 21 |
300
|
aaronmk
|
class DbException(exc.ExceptionWithCause):
|
| 22 |
14
|
aaronmk
|
def __init__(self, msg, cause=None, cur=None):
|
| 23 |
300
|
aaronmk
|
exc.ExceptionWithCause.__init__(self, msg, cause)
|
| 24 |
14
|
aaronmk
|
if cur != None: _add_cursor_info(self, cur)
|
| 25 |
|
|
|
| 26 |
360
|
aaronmk
|
class NameException(DbException): pass
|
| 27 |
|
|
|
| 28 |
468
|
aaronmk
|
class ExceptionWithColumns(DbException):
|
| 29 |
|
|
def __init__(self, cols, cause=None):
|
| 30 |
|
|
DbException.__init__(self, 'columns: ' + ', '.join(cols), cause)
|
| 31 |
|
|
self.cols = cols
|
| 32 |
11
|
aaronmk
|
|
| 33 |
468
|
aaronmk
|
class DuplicateKeyException(ExceptionWithColumns): pass
|
| 34 |
13
|
aaronmk
|
|
| 35 |
468
|
aaronmk
|
class NullValueException(ExceptionWithColumns): pass
|
| 36 |
13
|
aaronmk
|
|
| 37 |
89
|
aaronmk
|
class EmptyRowException(DbException): pass
|
| 38 |
|
|
|
| 39 |
865
|
aaronmk
|
##### Warnings
|
| 40 |
|
|
|
| 41 |
|
|
class DbWarning(UserWarning): pass
|
| 42 |
|
|
|
| 43 |
832
|
aaronmk
|
##### Input validation
|
| 44 |
|
|
|
| 45 |
11
|
aaronmk
|
def check_name(name):
|
| 46 |
|
|
if re.search(r'\W', name) != None: raise NameException('Name "'+name
|
| 47 |
|
|
+'" may contain only alphanumeric characters and _')
|
| 48 |
|
|
|
| 49 |
643
|
aaronmk
|
def esc_name(db, name):
|
| 50 |
|
|
module = util.root_module(db)
|
| 51 |
645
|
aaronmk
|
if module == 'psycopg2': return name
|
| 52 |
|
|
# Don't enclose in quotes because this disables case-insensitivity
|
| 53 |
643
|
aaronmk
|
elif module == 'MySQLdb': quote = '`'
|
| 54 |
645
|
aaronmk
|
else: raise NotImplementedError("Can't escape name for "+module+' database')
|
| 55 |
643
|
aaronmk
|
return quote + name.replace(quote, '') + quote
|
| 56 |
|
|
|
| 57 |
832
|
aaronmk
|
##### Querying
|
| 58 |
|
|
|
| 59 |
830
|
aaronmk
|
def run_raw_query(db, query, params=None):
|
| 60 |
11
|
aaronmk
|
cur = db.cursor()
|
| 61 |
|
|
try: cur.execute(query, params)
|
| 62 |
46
|
aaronmk
|
except Exception, e:
|
| 63 |
|
|
_add_cursor_info(e, cur)
|
| 64 |
11
|
aaronmk
|
raise
|
| 65 |
867
|
aaronmk
|
if run_raw_query.debug:
|
| 66 |
|
|
sys.stderr.write(strings.one_line(get_cur_query(cur))+'\n')
|
| 67 |
11
|
aaronmk
|
return cur
|
| 68 |
|
|
|
| 69 |
832
|
aaronmk
|
##### Recoverable querying
|
| 70 |
15
|
aaronmk
|
|
| 71 |
11
|
aaronmk
|
def with_savepoint(db, func):
|
| 72 |
|
|
savepoint = 'savepoint_'+str(random.randint(0, sys.maxint)) # must be unique
|
| 73 |
830
|
aaronmk
|
run_raw_query(db, 'SAVEPOINT '+savepoint)
|
| 74 |
11
|
aaronmk
|
try: return_val = func()
|
| 75 |
|
|
except:
|
| 76 |
830
|
aaronmk
|
run_raw_query(db, 'ROLLBACK TO SAVEPOINT '+savepoint)
|
| 77 |
11
|
aaronmk
|
raise
|
| 78 |
|
|
else:
|
| 79 |
830
|
aaronmk
|
run_raw_query(db, 'RELEASE SAVEPOINT '+savepoint)
|
| 80 |
11
|
aaronmk
|
return return_val
|
| 81 |
|
|
|
| 82 |
830
|
aaronmk
|
def run_query(db, query, params=None, recover=None):
|
| 83 |
|
|
if recover == None: recover = False
|
| 84 |
|
|
|
| 85 |
|
|
def run(): return run_raw_query(db, query, params)
|
| 86 |
|
|
if recover: return with_savepoint(db, run)
|
| 87 |
|
|
else: return run()
|
| 88 |
|
|
|
| 89 |
832
|
aaronmk
|
##### Result retrieval
|
| 90 |
|
|
|
| 91 |
1135
|
aaronmk
|
def col_names(cur): return (col[0] for col in cur.description)
|
| 92 |
832
|
aaronmk
|
|
| 93 |
|
|
def rows(cur): return iter(lambda: cur.fetchone(), None)
|
| 94 |
|
|
|
| 95 |
|
|
def row(cur): return rows(cur).next()
|
| 96 |
|
|
|
| 97 |
|
|
def value(cur): return row(cur)[0]
|
| 98 |
|
|
|
| 99 |
|
|
def values(cur): return iter(lambda: value(cur), None)
|
| 100 |
|
|
|
| 101 |
|
|
def value_or_none(cur):
|
| 102 |
|
|
try: return value(cur)
|
| 103 |
|
|
except StopIteration: return None
|
| 104 |
|
|
|
| 105 |
|
|
##### Basic queries
|
| 106 |
|
|
|
| 107 |
1135
|
aaronmk
|
def select(db, table, fields=None, conds=None, limit=None, start=None,
|
| 108 |
|
|
recover=None):
|
| 109 |
|
|
'''@param fields Use None to select all fields in the table'''
|
| 110 |
|
|
if conds == None: conds = {}
|
| 111 |
135
|
aaronmk
|
assert limit == None or type(limit) == int
|
| 112 |
865
|
aaronmk
|
assert start == None or type(start) == int
|
| 113 |
15
|
aaronmk
|
check_name(table)
|
| 114 |
1135
|
aaronmk
|
if fields != None: map(check_name, fields)
|
| 115 |
15
|
aaronmk
|
map(check_name, conds.keys())
|
| 116 |
865
|
aaronmk
|
|
| 117 |
11
|
aaronmk
|
def cond(entry):
|
| 118 |
13
|
aaronmk
|
col, value = entry
|
| 119 |
644
|
aaronmk
|
cond_ = esc_name(db, col)+' '
|
| 120 |
11
|
aaronmk
|
if value == None: cond_ += 'IS'
|
| 121 |
|
|
else: cond_ += '='
|
| 122 |
|
|
cond_ += ' %s'
|
| 123 |
|
|
return cond_
|
| 124 |
1135
|
aaronmk
|
query = 'SELECT '
|
| 125 |
|
|
if fields == None: query += '*'
|
| 126 |
|
|
else: query += ', '.join([esc_name(db, field) for field in fields])
|
| 127 |
|
|
query += ' FROM '+esc_name(db, table)
|
| 128 |
865
|
aaronmk
|
|
| 129 |
|
|
missing = True
|
| 130 |
89
|
aaronmk
|
if conds != {}:
|
| 131 |
|
|
query += ' WHERE '+' AND '.join(map(cond, conds.iteritems()))
|
| 132 |
865
|
aaronmk
|
missing = False
|
| 133 |
|
|
if limit != None: query += ' LIMIT '+str(limit); missing = False
|
| 134 |
|
|
if start != None:
|
| 135 |
|
|
if start != 0: query += ' OFFSET '+str(start)
|
| 136 |
|
|
missing = False
|
| 137 |
|
|
if missing: warnings.warn(DbWarning(
|
| 138 |
|
|
'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
|
| 139 |
|
|
|
| 140 |
830
|
aaronmk
|
return run_query(db, query, conds.values(), recover)
|
| 141 |
11
|
aaronmk
|
|
| 142 |
830
|
aaronmk
|
def insert(db, table, row, recover=None):
|
| 143 |
11
|
aaronmk
|
check_name(table)
|
| 144 |
13
|
aaronmk
|
cols = row.keys()
|
| 145 |
15
|
aaronmk
|
map(check_name, cols)
|
| 146 |
89
|
aaronmk
|
query = 'INSERT INTO '+table
|
| 147 |
|
|
if row != {}: query += ' ('+', '.join(cols)+') VALUES ('\
|
| 148 |
|
|
+', '.join(['%s']*len(cols))+')'
|
| 149 |
|
|
else: query += ' DEFAULT VALUES'
|
| 150 |
830
|
aaronmk
|
return run_query(db, query, row.values(), recover)
|
| 151 |
11
|
aaronmk
|
|
| 152 |
135
|
aaronmk
|
def last_insert_id(db):
|
| 153 |
|
|
module = util.root_module(db)
|
| 154 |
|
|
if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
|
| 155 |
|
|
elif module == 'MySQLdb': return db.insert_id()
|
| 156 |
|
|
else: return None
|
| 157 |
13
|
aaronmk
|
|
| 158 |
832
|
aaronmk
|
def truncate(db, table):
|
| 159 |
|
|
check_name(table)
|
| 160 |
869
|
aaronmk
|
return run_raw_query(db, 'TRUNCATE '+table+' CASCADE')
|
| 161 |
832
|
aaronmk
|
|
| 162 |
|
|
##### Database structure queries
|
| 163 |
|
|
|
| 164 |
|
|
def pkey(db, cache, table, recover=None):
|
| 165 |
|
|
'''Assumed to be first column in table'''
|
| 166 |
|
|
check_name(table)
|
| 167 |
|
|
if table not in cache:
|
| 168 |
1135
|
aaronmk
|
cache[table] = col_names(run_query(db,
|
| 169 |
|
|
'SELECT * FROM '+table+' LIMIT 0', recover=recover)).next()
|
| 170 |
832
|
aaronmk
|
return cache[table]
|
| 171 |
|
|
|
| 172 |
853
|
aaronmk
|
def index_cols(db, table, index):
|
| 173 |
|
|
'''Can also use this for UNIQUE constraints, because a UNIQUE index is
|
| 174 |
|
|
automatically created. When you don't know whether something is a UNIQUE
|
| 175 |
|
|
constraint or a UNIQUE index, use this function.'''
|
| 176 |
|
|
check_name(table)
|
| 177 |
|
|
check_name(index)
|
| 178 |
|
|
module = util.root_module(db)
|
| 179 |
|
|
if module == 'psycopg2':
|
| 180 |
|
|
return list(values(run_query(db, '''\
|
| 181 |
|
|
SELECT attname
|
| 182 |
866
|
aaronmk
|
FROM
|
| 183 |
|
|
(
|
| 184 |
|
|
SELECT attnum, attname
|
| 185 |
|
|
FROM pg_index
|
| 186 |
|
|
JOIN pg_class index ON index.oid = indexrelid
|
| 187 |
|
|
JOIN pg_class table_ ON table_.oid = indrelid
|
| 188 |
|
|
JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
|
| 189 |
|
|
WHERE
|
| 190 |
|
|
table_.relname = %(table)s
|
| 191 |
|
|
AND index.relname = %(index)s
|
| 192 |
|
|
UNION
|
| 193 |
|
|
SELECT attnum, attname
|
| 194 |
|
|
FROM
|
| 195 |
|
|
(
|
| 196 |
|
|
SELECT
|
| 197 |
|
|
indrelid
|
| 198 |
|
|
, (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
|
| 199 |
|
|
AS indkey
|
| 200 |
|
|
FROM pg_index
|
| 201 |
|
|
JOIN pg_class index ON index.oid = indexrelid
|
| 202 |
|
|
JOIN pg_class table_ ON table_.oid = indrelid
|
| 203 |
|
|
WHERE
|
| 204 |
|
|
table_.relname = %(table)s
|
| 205 |
|
|
AND index.relname = %(index)s
|
| 206 |
|
|
) s
|
| 207 |
|
|
JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
|
| 208 |
|
|
) s
|
| 209 |
853
|
aaronmk
|
ORDER BY attnum
|
| 210 |
|
|
''',
|
| 211 |
|
|
{'table': table, 'index': index})))
|
| 212 |
|
|
else: raise NotImplementedError("Can't list index columns for "+module+
|
| 213 |
|
|
' database')
|
| 214 |
|
|
|
| 215 |
464
|
aaronmk
|
def constraint_cols(db, table, constraint):
|
| 216 |
|
|
check_name(table)
|
| 217 |
|
|
check_name(constraint)
|
| 218 |
|
|
module = util.root_module(db)
|
| 219 |
|
|
if module == 'psycopg2':
|
| 220 |
|
|
return list(values(run_query(db, '''\
|
| 221 |
|
|
SELECT attname
|
| 222 |
|
|
FROM pg_constraint
|
| 223 |
|
|
JOIN pg_class ON pg_class.oid = conrelid
|
| 224 |
|
|
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
|
| 225 |
|
|
WHERE
|
| 226 |
|
|
relname = %(table)s
|
| 227 |
|
|
AND conname = %(constraint)s
|
| 228 |
|
|
ORDER BY attnum
|
| 229 |
|
|
''',
|
| 230 |
|
|
{'table': table, 'constraint': constraint})))
|
| 231 |
|
|
else: raise NotImplementedError("Can't list constraint columns for "+module+
|
| 232 |
|
|
' database')
|
| 233 |
|
|
|
| 234 |
832
|
aaronmk
|
def tables(db):
|
| 235 |
|
|
module = util.root_module(db)
|
| 236 |
|
|
if module == 'psycopg2':
|
| 237 |
|
|
return values(run_query(db, "SELECT tablename from pg_tables "
|
| 238 |
|
|
"WHERE schemaname = 'public' ORDER BY tablename"))
|
| 239 |
|
|
elif module == 'MySQLdb': return values(run_query(db, 'SHOW TABLES'))
|
| 240 |
|
|
else: raise NotImplementedError("Can't list tables for "+module+' database')
|
| 241 |
830
|
aaronmk
|
|
| 242 |
833
|
aaronmk
|
##### Database management
|
| 243 |
|
|
|
| 244 |
|
|
def empty_db(db):
|
| 245 |
|
|
for table in tables(db): truncate(db, table)
|
| 246 |
|
|
|
| 247 |
832
|
aaronmk
|
##### Heuristic queries
|
| 248 |
|
|
|
| 249 |
14
|
aaronmk
|
def try_insert(db, table, row):
|
| 250 |
830
|
aaronmk
|
'''Recovers from errors'''
|
| 251 |
|
|
try: return insert(db, table, row, recover=True)
|
| 252 |
46
|
aaronmk
|
except Exception, e:
|
| 253 |
|
|
msg = str(e)
|
| 254 |
465
|
aaronmk
|
match = re.search(r'duplicate key value violates unique constraint '
|
| 255 |
|
|
r'"(([^\W_]+)_[^"]+)"', msg)
|
| 256 |
|
|
if match:
|
| 257 |
|
|
constraint, table = match.groups()
|
| 258 |
854
|
aaronmk
|
try: cols = index_cols(db, table, constraint)
|
| 259 |
465
|
aaronmk
|
except NotImplementedError: raise e
|
| 260 |
851
|
aaronmk
|
else: raise DuplicateKeyException(cols, e)
|
| 261 |
13
|
aaronmk
|
match = re.search(r'null value in column "(\w+)" violates not-null '
|
| 262 |
|
|
'constraint', msg)
|
| 263 |
470
|
aaronmk
|
if match: raise NullValueException([match.group(1)], e)
|
| 264 |
13
|
aaronmk
|
raise # no specific exception raised
|
| 265 |
11
|
aaronmk
|
|
| 266 |
471
|
aaronmk
|
def put(db, table, row, pkey, row_ct_ref=None):
|
| 267 |
830
|
aaronmk
|
'''Recovers from errors'''
|
| 268 |
471
|
aaronmk
|
try:
|
| 269 |
|
|
row_ct = try_insert(db, table, row).rowcount
|
| 270 |
|
|
if row_ct_ref != None and row_ct >= 0: row_ct_ref[0] += row_ct
|
| 271 |
|
|
return last_insert_id(db)
|
| 272 |
|
|
except DuplicateKeyException, e:
|
| 273 |
1069
|
aaronmk
|
return value(select(db, table, [pkey],
|
| 274 |
|
|
util.dict_subset_right_join(row, e.cols), recover=True))
|
| 275 |
471
|
aaronmk
|
|
| 276 |
473
|
aaronmk
|
def get(db, table, row, pkey, row_ct_ref=None, create=False):
|
| 277 |
830
|
aaronmk
|
'''Recovers from errors'''
|
| 278 |
|
|
try: return value(select(db, table, [pkey], row, 1, recover=True))
|
| 279 |
14
|
aaronmk
|
except StopIteration:
|
| 280 |
40
|
aaronmk
|
if not create: raise
|
| 281 |
471
|
aaronmk
|
return put(db, table, row, pkey, row_ct_ref) # insert new row
|
| 282 |
131
|
aaronmk
|
|
| 283 |
832
|
aaronmk
|
##### Database connections
|
| 284 |
|
|
|
| 285 |
131
|
aaronmk
|
db_engines = {
|
| 286 |
|
|
'MySQL': ('MySQLdb', {'password': 'passwd', 'database': 'db'}),
|
| 287 |
|
|
'PostgreSQL': ('psycopg2', {}),
|
| 288 |
|
|
}
|
| 289 |
|
|
|
| 290 |
360
|
aaronmk
|
DatabaseErrors_set = set([DbException])
|
| 291 |
|
|
DatabaseErrors = tuple(DatabaseErrors_set)
|
| 292 |
342
|
aaronmk
|
|
| 293 |
|
|
def _add_module(module):
|
| 294 |
|
|
DatabaseErrors_set.add(module.DatabaseError)
|
| 295 |
|
|
global DatabaseErrors
|
| 296 |
|
|
DatabaseErrors = tuple(DatabaseErrors_set)
|
| 297 |
|
|
|
| 298 |
646
|
aaronmk
|
def connect(db_config, serializable=True):
|
| 299 |
131
|
aaronmk
|
db_config = db_config.copy() # don't modify input!
|
| 300 |
342
|
aaronmk
|
module_name, mappings = db_engines[db_config.pop('engine')]
|
| 301 |
|
|
module = __import__(module_name)
|
| 302 |
|
|
_add_module(module)
|
| 303 |
330
|
aaronmk
|
for orig, new in mappings.iteritems():
|
| 304 |
|
|
try: util.rename_key(db_config, orig, new)
|
| 305 |
|
|
except KeyError: pass
|
| 306 |
646
|
aaronmk
|
db = module.connect(**db_config)
|
| 307 |
|
|
if serializable:
|
| 308 |
830
|
aaronmk
|
run_raw_query(db, 'SET TRANSACTION ISOLATION LEVEL SERIALIZABLE')
|
| 309 |
646
|
aaronmk
|
return db
|
| 310 |
|
|
|
| 311 |
|
|
def db_config_str(db_config):
|
| 312 |
|
|
return db_config['engine']+' database '+db_config['database']
|
