Project

General

Profile

1 11 aaronmk
# Database access
2
3 1869 aaronmk
import copy
4 11 aaronmk
import re
5 865 aaronmk
import warnings
6 11 aaronmk
7 300 aaronmk
import exc
8 1909 aaronmk
import dicts
9 1893 aaronmk
import iters
10 1960 aaronmk
import lists
11 1889 aaronmk
from Proxy import Proxy
12 1872 aaronmk
import rand
13 862 aaronmk
import strings
14 131 aaronmk
import util
15 11 aaronmk
16 832 aaronmk
##### Exceptions
17
18 135 aaronmk
def get_cur_query(cur):
19
    if hasattr(cur, 'query'): return cur.query
20
    elif hasattr(cur, '_last_executed'): return cur._last_executed
21
    else: return None
22 14 aaronmk
23 300 aaronmk
def _add_cursor_info(e, cur): exc.add_msg(e, 'query: '+get_cur_query(cur))
24 135 aaronmk
25 300 aaronmk
class DbException(exc.ExceptionWithCause):
26 14 aaronmk
    def __init__(self, msg, cause=None, cur=None):
27 300 aaronmk
        exc.ExceptionWithCause.__init__(self, msg, cause)
28 14 aaronmk
        if cur != None: _add_cursor_info(self, cur)
29
30 360 aaronmk
class NameException(DbException): pass
31
32 468 aaronmk
class ExceptionWithColumns(DbException):
33
    def __init__(self, cols, cause=None):
34
        DbException.__init__(self, 'columns: ' + ', '.join(cols), cause)
35
        self.cols = cols
36 11 aaronmk
37 468 aaronmk
class DuplicateKeyException(ExceptionWithColumns): pass
38 13 aaronmk
39 468 aaronmk
class NullValueException(ExceptionWithColumns): pass
40 13 aaronmk
41 89 aaronmk
class EmptyRowException(DbException): pass
42
43 865 aaronmk
##### Warnings
44
45
class DbWarning(UserWarning): pass
46
47 1930 aaronmk
##### Result retrieval
48
49
def col_names(cur): return (col[0] for col in cur.description)
50
51
def rows(cur): return iter(lambda: cur.fetchone(), None)
52
53
def consume_rows(cur):
54
    '''Used to fetch all rows so result will be cached'''
55
    iters.consume_iter(rows(cur))
56
57
def next_row(cur): return rows(cur).next()
58
59
def row(cur):
60
    row_ = next_row(cur)
61
    consume_rows(cur)
62
    return row_
63
64
def next_value(cur): return next_row(cur)[0]
65
66
def value(cur): return row(cur)[0]
67
68
def values(cur): return iters.func_iter(lambda: next_value(cur))
69
70
def value_or_none(cur):
71
    try: return value(cur)
72
    except StopIteration: return None
73
74 1869 aaronmk
##### Database connections
75 1849 aaronmk
76 1926 aaronmk
db_config_names = ['engine', 'host', 'user', 'password', 'database']
77
78 1869 aaronmk
db_engines = {
79
    'MySQL': ('MySQLdb', {'password': 'passwd', 'database': 'db'}),
80
    'PostgreSQL': ('psycopg2', {}),
81
}
82
83
DatabaseErrors_set = set([DbException])
84
DatabaseErrors = tuple(DatabaseErrors_set)
85
86
def _add_module(module):
87
    DatabaseErrors_set.add(module.DatabaseError)
88
    global DatabaseErrors
89
    DatabaseErrors = tuple(DatabaseErrors_set)
90
91
def db_config_str(db_config):
92
    return db_config['engine']+' database '+db_config['database']
93
94 1909 aaronmk
def _query_lookup(query, params): return (query, dicts.make_hashable(params))
95 1894 aaronmk
96 1901 aaronmk
log_debug_none = lambda msg: None
97
98 1849 aaronmk
class DbConn:
99 2047 aaronmk
    def __init__(self, db_config, serializable=True, log_debug=log_debug_none,
100 2050 aaronmk
        caching=True):
101 1869 aaronmk
        self.db_config = db_config
102
        self.serializable = serializable
103 1901 aaronmk
        self.log_debug = log_debug
104 2047 aaronmk
        self.caching = caching
105 1869 aaronmk
106
        self.__db = None
107 1889 aaronmk
        self.query_results = {}
108 1869 aaronmk
109
    def __getattr__(self, name):
110
        if name == '__dict__': raise Exception('getting __dict__')
111
        if name == 'db': return self._db()
112
        else: raise AttributeError()
113
114
    def __getstate__(self):
115
        state = copy.copy(self.__dict__) # shallow copy
116 1915 aaronmk
        state['log_debug'] = None # don't pickle the debug callback
117 1869 aaronmk
        state['_DbConn__db'] = None # don't pickle the connection
118
        return state
119
120
    def _db(self):
121
        if self.__db == None:
122
            # Process db_config
123
            db_config = self.db_config.copy() # don't modify input!
124
            module_name, mappings = db_engines[db_config.pop('engine')]
125
            module = __import__(module_name)
126
            _add_module(module)
127
            for orig, new in mappings.iteritems():
128
                try: util.rename_key(db_config, orig, new)
129
                except KeyError: pass
130
131
            # Connect
132
            self.__db = module.connect(**db_config)
133
134
            # Configure connection
135
            if self.serializable: run_raw_query(self,
136
                'SET TRANSACTION ISOLATION LEVEL SERIALIZABLE')
137
138
        return self.__db
139 1889 aaronmk
140 1891 aaronmk
    class DbCursor(Proxy):
141 1927 aaronmk
        def __init__(self, outer):
142 1891 aaronmk
            Proxy.__init__(self, outer.db.cursor())
143 1927 aaronmk
            self.query_results = outer.query_results
144 1894 aaronmk
            self.query_lookup = None
145 1891 aaronmk
            self.result = []
146 1889 aaronmk
147 1894 aaronmk
        def execute(self, query, params=None):
148 1930 aaronmk
            self._is_insert = query.upper().find('INSERT') >= 0
149 1894 aaronmk
            self.query_lookup = _query_lookup(query, params)
150 1904 aaronmk
            try: return_value = self.inner.execute(query, params)
151
            except Exception, e:
152
                self.result = e # cache the exception as the result
153
                self._cache_result()
154
                raise
155
            finally: self.query = get_cur_query(self.inner)
156 1930 aaronmk
            # Fetch all rows so result will be cached
157
            if self.rowcount == 0 and not self._is_insert: consume_rows(self)
158 1894 aaronmk
            return return_value
159
160 1891 aaronmk
        def fetchone(self):
161
            row = self.inner.fetchone()
162 1899 aaronmk
            if row != None: self.result.append(row)
163
            # otherwise, fetched all rows
164 1904 aaronmk
            else: self._cache_result()
165
            return row
166
167
        def _cache_result(self):
168 1906 aaronmk
            # For inserts, only cache exceptions since inserts are not
169
            # idempotent, but an invalid insert will always be invalid
170 1930 aaronmk
            if self.query_results != None and (not self._is_insert
171 1906 aaronmk
                or isinstance(self.result, Exception)):
172
173 1894 aaronmk
                assert self.query_lookup != None
174 1916 aaronmk
                self.query_results[self.query_lookup] = self.CacheCursor(
175
                    util.dict_subset(dicts.AttrsDictView(self),
176
                    ['query', 'result', 'rowcount', 'description']))
177 1906 aaronmk
178 1916 aaronmk
        class CacheCursor:
179
            def __init__(self, cached_result): self.__dict__ = cached_result
180
181 1927 aaronmk
            def execute(self, *args, **kw_args):
182 1916 aaronmk
                if isinstance(self.result, Exception): raise self.result
183
                # otherwise, result is a rows list
184
                self.iter = iter(self.result)
185
186
            def fetchone(self):
187
                try: return self.iter.next()
188
                except StopIteration: return None
189 1891 aaronmk
190 1894 aaronmk
    def run_query(self, query, params=None, cacheable=False):
191 2047 aaronmk
        if not self.caching: cacheable = False
192 1903 aaronmk
        used_cache = False
193
        try:
194 1927 aaronmk
            # Get cursor
195
            if cacheable:
196
                query_lookup = _query_lookup(query, params)
197
                try:
198
                    cur = self.query_results[query_lookup]
199
                    used_cache = True
200
                except KeyError: cur = self.DbCursor(self)
201
            else: cur = self.db.cursor()
202
203
            # Run query
204
            try: cur.execute(query, params)
205
            except Exception, e:
206
                _add_cursor_info(e, cur)
207
                raise
208 1903 aaronmk
        finally:
209
            if self.log_debug != log_debug_none: # only compute msg if needed
210
                if used_cache: cache_status = 'Cache hit'
211
                elif cacheable: cache_status = 'Cache miss'
212
                else: cache_status = 'Non-cacheable'
213 1927 aaronmk
                self.log_debug(cache_status+': '
214
                    +strings.one_line(get_cur_query(cur)))
215 1903 aaronmk
216
        return cur
217 1914 aaronmk
218
    def is_cached(self, query, params=None):
219
        return _query_lookup(query, params) in self.query_results
220 1849 aaronmk
221 1869 aaronmk
connect = DbConn
222
223 1919 aaronmk
##### Input validation
224
225
def check_name(name):
226
    if re.search(r'\W', name) != None: raise NameException('Name "'+name
227
        +'" may contain only alphanumeric characters and _')
228
229
def esc_name_by_module(module, name, preserve_case=False):
230
    if module == 'psycopg2':
231
        if preserve_case: quote = '"'
232
        # Don't enclose in quotes because this disables case-insensitivity
233
        else: return name
234
    elif module == 'MySQLdb': quote = '`'
235
    else: raise NotImplementedError("Can't escape name for "+module+' database')
236
    return quote + name.replace(quote, '') + quote
237
238
def esc_name_by_engine(engine, name, **kw_args):
239
    return esc_name_by_module(db_engines[engine][0], name, **kw_args)
240
241
def esc_name(db, name, **kw_args):
242
    return esc_name_by_module(util.root_module(db.db), name, **kw_args)
243
244 1968 aaronmk
def qual_name(db, schema, table):
245 1981 aaronmk
    def esc_name_(name): return esc_name(db, name, preserve_case=True)
246 2051 aaronmk
    table = esc_name_(table)
247
    if schema != None: return esc_name_(schema)+'.'+table
248
    else: return table
249 1968 aaronmk
250 832 aaronmk
##### Querying
251
252 1894 aaronmk
def run_raw_query(db, *args, **kw_args):
253
    '''For args, see DbConn.run_query()'''
254
    return db.run_query(*args, **kw_args)
255 11 aaronmk
256 832 aaronmk
##### Recoverable querying
257 15 aaronmk
258 11 aaronmk
def with_savepoint(db, func):
259 1872 aaronmk
    savepoint = 'savepoint_'+str(rand.rand_int()) # must be unique
260 830 aaronmk
    run_raw_query(db, 'SAVEPOINT '+savepoint)
261 11 aaronmk
    try: return_val = func()
262
    except:
263 830 aaronmk
        run_raw_query(db, 'ROLLBACK TO SAVEPOINT '+savepoint)
264 11 aaronmk
        raise
265
    else:
266 830 aaronmk
        run_raw_query(db, 'RELEASE SAVEPOINT '+savepoint)
267 11 aaronmk
        return return_val
268
269 1894 aaronmk
def run_query(db, query, params=None, recover=None, cacheable=False):
270 830 aaronmk
    if recover == None: recover = False
271
272 1894 aaronmk
    def run(): return run_raw_query(db, query, params, cacheable)
273 1914 aaronmk
    if recover and not db.is_cached(query, params):
274
        return with_savepoint(db, run)
275
    else: return run() # don't need savepoint if cached
276 830 aaronmk
277 832 aaronmk
##### Basic queries
278
279 2054 aaronmk
def mk_select(db, table, fields=None, conds=None, limit=None, start=None,
280
    table_is_esc=False):
281 1981 aaronmk
    '''
282
    @param fields Use None to select all fields in the table
283
    @param table_is_esc Whether the table name has already been escaped
284 2054 aaronmk
    @return tuple(query, params)
285 1981 aaronmk
    '''
286 1135 aaronmk
    if conds == None: conds = {}
287 135 aaronmk
    assert limit == None or type(limit) == int
288 865 aaronmk
    assert start == None or type(start) == int
289 1981 aaronmk
    if not table_is_esc: check_name(table)
290 1135 aaronmk
    if fields != None: map(check_name, fields)
291 15 aaronmk
    map(check_name, conds.keys())
292 865 aaronmk
293 11 aaronmk
    def cond(entry):
294 13 aaronmk
        col, value = entry
295 644 aaronmk
        cond_ = esc_name(db, col)+' '
296 11 aaronmk
        if value == None: cond_ += 'IS'
297
        else: cond_ += '='
298
        cond_ += ' %s'
299
        return cond_
300 1135 aaronmk
    query = 'SELECT '
301
    if fields == None: query += '*'
302
    else: query += ', '.join([esc_name(db, field) for field in fields])
303
    query += ' FROM '+esc_name(db, table)
304 865 aaronmk
305
    missing = True
306 89 aaronmk
    if conds != {}:
307
        query += ' WHERE '+' AND '.join(map(cond, conds.iteritems()))
308 865 aaronmk
        missing = False
309
    if limit != None: query += ' LIMIT '+str(limit); missing = False
310
    if start != None:
311
        if start != 0: query += ' OFFSET '+str(start)
312
        missing = False
313
    if missing: warnings.warn(DbWarning(
314
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
315
316 2054 aaronmk
    return (query, conds.values())
317 11 aaronmk
318 2054 aaronmk
def select(db, *args, **kw_args):
319
    '''For params, see mk_select() and run_query()'''
320
    recover = kw_args.pop('recover', None)
321
    cacheable = kw_args.pop('cacheable', True)
322
323
    query, params = mk_select(db, *args, **kw_args)
324
    return run_query(db, query, params, recover, cacheable)
325
326 1961 aaronmk
default = object() # tells insert() to use the default value for a column
327
328 1960 aaronmk
def insert(db, table, row, returning=None, recover=None, cacheable=True,
329
    table_is_esc=False):
330
    '''
331
    @param returning str|None An inserted column (such as pkey) to return
332
    @param table_is_esc Whether the table name has already been escaped
333
    '''
334
    if not table_is_esc: check_name(table)
335
    if lists.is_seq(row): cols = None
336
    else:
337
        cols = row.keys()
338
        row = row.values()
339
        map(check_name, cols)
340
    row = list(row) # ensure that "!= []" works
341
342 1961 aaronmk
    # Check for special values
343
    labels = []
344
    values = []
345
    for value in row:
346
        if value == default: labels.append('DEFAULT')
347
        else:
348
            labels.append('%s')
349
            values.append(value)
350
351
    # Build query
352 89 aaronmk
    query = 'INSERT INTO '+table
353 1961 aaronmk
    if values != []:
354 1960 aaronmk
        if cols != None: query += ' ('+', '.join(cols)+')'
355 1961 aaronmk
        query += ' VALUES ('+(', '.join(labels))+')'
356 89 aaronmk
    else: query += ' DEFAULT VALUES'
357 1554 aaronmk
358
    if returning != None:
359
        check_name(returning)
360
        query += ' RETURNING '+returning
361
362 1961 aaronmk
    return run_query(db, query, values, recover, cacheable)
363 11 aaronmk
364 135 aaronmk
def last_insert_id(db):
365 1849 aaronmk
    module = util.root_module(db.db)
366 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
367
    elif module == 'MySQLdb': return db.insert_id()
368
    else: return None
369 13 aaronmk
370 1968 aaronmk
def truncate(db, table, schema='public'):
371
    return run_query(db, 'TRUNCATE '+qual_name(db, schema, table)+' CASCADE')
372 832 aaronmk
373
##### Database structure queries
374
375 1850 aaronmk
def pkey(db, table, recover=None):
376 832 aaronmk
    '''Assumed to be first column in table'''
377
    check_name(table)
378 1929 aaronmk
    return col_names(select(db, table, limit=0, recover=recover)).next()
379 832 aaronmk
380 853 aaronmk
def index_cols(db, table, index):
381
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
382
    automatically created. When you don't know whether something is a UNIQUE
383
    constraint or a UNIQUE index, use this function.'''
384
    check_name(table)
385
    check_name(index)
386 1909 aaronmk
    module = util.root_module(db.db)
387
    if module == 'psycopg2':
388
        return list(values(run_query(db, '''\
389 853 aaronmk
SELECT attname
390 866 aaronmk
FROM
391
(
392
        SELECT attnum, attname
393
        FROM pg_index
394
        JOIN pg_class index ON index.oid = indexrelid
395
        JOIN pg_class table_ ON table_.oid = indrelid
396
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
397
        WHERE
398
            table_.relname = %(table)s
399
            AND index.relname = %(index)s
400
    UNION
401
        SELECT attnum, attname
402
        FROM
403
        (
404
            SELECT
405
                indrelid
406
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
407
                    AS indkey
408
            FROM pg_index
409
            JOIN pg_class index ON index.oid = indexrelid
410
            JOIN pg_class table_ ON table_.oid = indrelid
411
            WHERE
412
                table_.relname = %(table)s
413
                AND index.relname = %(index)s
414
        ) s
415
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
416
) s
417 853 aaronmk
ORDER BY attnum
418
''',
419 1909 aaronmk
            {'table': table, 'index': index}, cacheable=True)))
420
    else: raise NotImplementedError("Can't list index columns for "+module+
421
        ' database')
422 853 aaronmk
423 464 aaronmk
def constraint_cols(db, table, constraint):
424
    check_name(table)
425
    check_name(constraint)
426 1849 aaronmk
    module = util.root_module(db.db)
427 464 aaronmk
    if module == 'psycopg2':
428
        return list(values(run_query(db, '''\
429
SELECT attname
430
FROM pg_constraint
431
JOIN pg_class ON pg_class.oid = conrelid
432
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
433
WHERE
434
    relname = %(table)s
435
    AND conname = %(constraint)s
436
ORDER BY attnum
437
''',
438
            {'table': table, 'constraint': constraint})))
439
    else: raise NotImplementedError("Can't list constraint columns for "+module+
440
        ' database')
441
442 1968 aaronmk
def tables(db, schema='public', table_like='%'):
443 1849 aaronmk
    module = util.root_module(db.db)
444 1968 aaronmk
    params = {'schema': schema, 'table_like': table_like}
445 832 aaronmk
    if module == 'psycopg2':
446 1968 aaronmk
        return values(run_query(db, '''\
447
SELECT tablename
448
FROM pg_tables
449
WHERE
450
    schemaname = %(schema)s
451
    AND tablename LIKE %(table_like)s
452
ORDER BY tablename
453
''',
454
            params, cacheable=True))
455
    elif module == 'MySQLdb':
456
        return values(run_query(db, 'SHOW TABLES LIKE %(table_like)s', params,
457
            cacheable=True))
458 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
459 830 aaronmk
460 833 aaronmk
##### Database management
461
462 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
463
    '''For kw_args, see tables()'''
464
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
465 833 aaronmk
466 832 aaronmk
##### Heuristic queries
467
468 1554 aaronmk
def try_insert(db, table, row, returning=None):
469 830 aaronmk
    '''Recovers from errors'''
470 1554 aaronmk
    try: return insert(db, table, row, returning, recover=True)
471 46 aaronmk
    except Exception, e:
472
        msg = str(e)
473 465 aaronmk
        match = re.search(r'duplicate key value violates unique constraint '
474
            r'"(([^\W_]+)_[^"]+)"', msg)
475
        if match:
476
            constraint, table = match.groups()
477 854 aaronmk
            try: cols = index_cols(db, table, constraint)
478 465 aaronmk
            except NotImplementedError: raise e
479 851 aaronmk
            else: raise DuplicateKeyException(cols, e)
480 13 aaronmk
        match = re.search(r'null value in column "(\w+)" violates not-null '
481
            'constraint', msg)
482 470 aaronmk
        if match: raise NullValueException([match.group(1)], e)
483 13 aaronmk
        raise # no specific exception raised
484 11 aaronmk
485 471 aaronmk
def put(db, table, row, pkey, row_ct_ref=None):
486 1554 aaronmk
    '''Recovers from errors.
487
    Only works under PostgreSQL (uses `INSERT ... RETURNING`)'''
488 471 aaronmk
    try:
489 1554 aaronmk
        cur = try_insert(db, table, row, pkey)
490
        if row_ct_ref != None and cur.rowcount >= 0:
491
            row_ct_ref[0] += cur.rowcount
492
        return value(cur)
493 471 aaronmk
    except DuplicateKeyException, e:
494 1069 aaronmk
        return value(select(db, table, [pkey],
495
            util.dict_subset_right_join(row, e.cols), recover=True))
496 471 aaronmk
497 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
498 830 aaronmk
    '''Recovers from errors'''
499
    try: return value(select(db, table, [pkey], row, 1, recover=True))
500 14 aaronmk
    except StopIteration:
501 40 aaronmk
        if not create: raise
502 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row