Project

General

Profile

1
# Database access
2

    
3
import copy
4
import re
5
import time
6
import warnings
7

    
8
import exc
9
import dicts
10
import iters
11
import lists
12
import profiling
13
from Proxy import Proxy
14
import rand
15
import sql_gen
16
import strings
17
import util
18

    
19
##### Exceptions
20

    
21
def get_cur_query(cur, input_query=None):
22
    raw_query = None
23
    if hasattr(cur, 'query'): raw_query = cur.query
24
    elif hasattr(cur, '_last_executed'): raw_query = cur._last_executed
25
    
26
    if raw_query != None: return raw_query
27
    else: return '[input] '+strings.ustr(input_query)
28

    
29
def _add_cursor_info(e, *args, **kw_args):
30
    '''For params, see get_cur_query()'''
31
    exc.add_msg(e, 'query: '+strings.ustr(get_cur_query(*args, **kw_args)))
32

    
33
class DbException(exc.ExceptionWithCause):
34
    def __init__(self, msg, cause=None, cur=None):
35
        exc.ExceptionWithCause.__init__(self, msg, cause, cause_newline=True)
36
        if cur != None: _add_cursor_info(self, cur)
37

    
38
class ExceptionWithName(DbException):
39
    def __init__(self, name, cause=None):
40
        DbException.__init__(self, 'for name: '+strings.as_tt(str(name)), cause)
41
        self.name = name
42

    
43
class ExceptionWithValue(DbException):
44
    def __init__(self, value, cause=None):
45
        DbException.__init__(self, 'for value: '+strings.as_tt(repr(value)),
46
            cause)
47
        self.value = value
48

    
49
class ExceptionWithNameType(DbException):
50
    def __init__(self, type_, name, cause=None):
51
        DbException.__init__(self, 'for type: '+strings.as_tt(str(type_))
52
            +'; name: '+strings.as_tt(name), cause)
53
        self.type = type_
54
        self.name = name
55

    
56
class ConstraintException(DbException):
57
    def __init__(self, name, cond, cols, cause=None):
58
        msg = 'Violated '+strings.as_tt(name)+' constraint'
59
        if cond != None: msg += ' with condition '+cond
60
        if cols != []: msg += ' on columns: '+strings.as_tt(', '.join(cols))
61
        DbException.__init__(self, msg, cause)
62
        self.name = name
63
        self.cond = cond
64
        self.cols = cols
65

    
66
class MissingCastException(DbException):
67
    def __init__(self, type_, col=None, cause=None):
68
        msg = 'Missing cast to type '+strings.as_tt(type_)
69
        if col != None: msg += ' on column: '+strings.as_tt(col)
70
        DbException.__init__(self, msg, cause)
71
        self.type = type_
72
        self.col = col
73

    
74
class NameException(DbException): pass
75

    
76
class DuplicateKeyException(ConstraintException): pass
77

    
78
class NullValueException(ConstraintException): pass
79

    
80
class CheckException(ConstraintException): pass
81

    
82
class InvalidValueException(ExceptionWithValue): pass
83

    
84
class DuplicateException(ExceptionWithNameType): pass
85

    
86
class DoesNotExistException(ExceptionWithNameType): pass
87

    
88
class EmptyRowException(DbException): pass
89

    
90
##### Warnings
91

    
92
class DbWarning(UserWarning): pass
93

    
94
##### Result retrieval
95

    
96
def col_names(cur): return (col[0] for col in cur.description)
97

    
98
def rows(cur): return iter(lambda: cur.fetchone(), None)
99

    
100
def consume_rows(cur):
101
    '''Used to fetch all rows so result will be cached'''
102
    iters.consume_iter(rows(cur))
103

    
104
def next_row(cur): return rows(cur).next()
105

    
106
def row(cur):
107
    row_ = next_row(cur)
108
    consume_rows(cur)
109
    return row_
110

    
111
def next_value(cur): return next_row(cur)[0]
112

    
113
def value(cur): return row(cur)[0]
114

    
115
def values(cur): return iters.func_iter(lambda: next_value(cur))
116

    
117
def value_or_none(cur):
118
    try: return value(cur)
119
    except StopIteration: return None
120

    
121
##### Escaping
122

    
123
def esc_name_by_module(module, name):
124
    if module == 'psycopg2' or module == None: quote = '"'
125
    elif module == 'MySQLdb': quote = '`'
126
    else: raise NotImplementedError("Can't escape name for "+module+' database')
127
    return sql_gen.esc_name(name, quote)
128

    
129
def esc_name_by_engine(engine, name, **kw_args):
130
    return esc_name_by_module(db_engines[engine][0], name, **kw_args)
131

    
132
def esc_name(db, name, **kw_args):
133
    return esc_name_by_module(util.root_module(db.db), name, **kw_args)
134

    
135
def qual_name(db, schema, table):
136
    def esc_name_(name): return esc_name(db, name)
137
    table = esc_name_(table)
138
    if schema != None: return esc_name_(schema)+'.'+table
139
    else: return table
140

    
141
##### Database connections
142

    
143
db_config_names = ['engine', 'host', 'user', 'password', 'database', 'schemas']
144

    
145
db_engines = {
146
    'MySQL': ('MySQLdb', {'password': 'passwd', 'database': 'db'}),
147
    'PostgreSQL': ('psycopg2', {}),
148
}
149

    
150
DatabaseErrors_set = set([DbException])
151
DatabaseErrors = tuple(DatabaseErrors_set)
152

    
153
def _add_module(module):
154
    DatabaseErrors_set.add(module.DatabaseError)
155
    global DatabaseErrors
156
    DatabaseErrors = tuple(DatabaseErrors_set)
157

    
158
def db_config_str(db_config):
159
    return db_config['engine']+' database '+db_config['database']
160

    
161
log_debug_none = lambda msg, level=2: None
162

    
163
class DbConn:
164
    def __init__(self, db_config, autocommit=True, caching=True,
165
        log_debug=log_debug_none, debug_temp=False, src=None):
166
        '''
167
        @param debug_temp Whether temporary objects should instead be permanent.
168
            This assists in debugging the internal objects used by the program.
169
        @param src In autocommit mode, will be included in a comment in every
170
            query, to help identify the data source in pg_stat_activity.
171
        '''
172
        self.db_config = db_config
173
        self.autocommit = autocommit
174
        self.caching = caching
175
        self.log_debug = log_debug
176
        self.debug = log_debug != log_debug_none
177
        self.debug_temp = debug_temp
178
        self.src = src
179
        self.autoanalyze = False
180
        self.autoexplain = False
181
        self.profile_row_ct = None
182
        
183
        self._savepoint = 0
184
        self._reset()
185
    
186
    def __getattr__(self, name):
187
        if name == '__dict__': raise Exception('getting __dict__')
188
        if name == 'db': return self._db()
189
        else: raise AttributeError()
190
    
191
    def __getstate__(self):
192
        state = copy.copy(self.__dict__) # shallow copy
193
        state['log_debug'] = None # don't pickle the debug callback
194
        state['_DbConn__db'] = None # don't pickle the connection
195
        return state
196
    
197
    def clear_cache(self): self.query_results = {}
198
    
199
    def _reset(self):
200
        self.clear_cache()
201
        assert self._savepoint == 0
202
        self._notices_seen = set()
203
        self.__db = None
204
    
205
    def connected(self): return self.__db != None
206
    
207
    def close(self):
208
        if not self.connected(): return
209
        
210
        # Record that the automatic transaction is now closed
211
        self._savepoint -= 1
212
        
213
        self.db.close()
214
        self._reset()
215
    
216
    def reconnect(self):
217
        # Do not do this in test mode as it would roll back everything
218
        if self.autocommit: self.close()
219
        # Connection will be reopened automatically on first query
220
    
221
    def _db(self):
222
        if self.__db == None:
223
            # Process db_config
224
            db_config = self.db_config.copy() # don't modify input!
225
            schemas = db_config.pop('schemas', None)
226
            module_name, mappings = db_engines[db_config.pop('engine')]
227
            module = __import__(module_name)
228
            _add_module(module)
229
            for orig, new in mappings.iteritems():
230
                try: util.rename_key(db_config, orig, new)
231
                except KeyError: pass
232
            
233
            # Connect
234
            self.__db = module.connect(**db_config)
235
            
236
            # Record that a transaction is already open
237
            self._savepoint += 1
238
            
239
            # Configure connection
240
            if hasattr(self.db, 'set_isolation_level'):
241
                import psycopg2.extensions
242
                self.db.set_isolation_level(
243
                    psycopg2.extensions.ISOLATION_LEVEL_READ_COMMITTED)
244
            if schemas != None:
245
                search_path = [self.esc_name(s) for s in schemas.split(',')]
246
                search_path.append(value(run_query(self, 'SHOW search_path',
247
                    log_level=4)))
248
                run_query(self, 'SET search_path TO '+(','.join(search_path)),
249
                    log_level=3)
250
        
251
        return self.__db
252
    
253
    class DbCursor(Proxy):
254
        def __init__(self, outer):
255
            Proxy.__init__(self, outer.db.cursor())
256
            self.outer = outer
257
            self.query_results = outer.query_results
258
            self.query_lookup = None
259
            self.result = []
260
        
261
        def execute(self, query):
262
            self._is_insert = query.startswith('INSERT')
263
            self.query_lookup = query
264
            try:
265
                try: cur = self.inner.execute(query)
266
                finally: self.query = get_cur_query(self.inner, query)
267
            except Exception, e:
268
                self.result = e # cache the exception as the result
269
                self._cache_result()
270
                raise
271
            
272
            # Always cache certain queries
273
            query = sql_gen.lstrip(query)
274
            if query.startswith('CREATE') or query.startswith('ALTER'):
275
                # structural changes
276
                # Rest of query must be unique in the face of name collisions,
277
                # so don't cache ADD COLUMN unless it has distinguishing comment
278
                if query.find('ADD COLUMN') < 0 or query.endswith('*/'):
279
                    self._cache_result()
280
            elif self.rowcount == 0 and query.startswith('SELECT'): # empty
281
                consume_rows(self) # fetch all rows so result will be cached
282
            
283
            return cur
284
        
285
        def fetchone(self):
286
            row = self.inner.fetchone()
287
            if row != None: self.result.append(row)
288
            # otherwise, fetched all rows
289
            else: self._cache_result()
290
            return row
291
        
292
        def _cache_result(self):
293
            # For inserts that return a result set, don't cache result set since
294
            # inserts are not idempotent. Other non-SELECT queries don't have
295
            # their result set read, so only exceptions will be cached (an
296
            # invalid query will always be invalid).
297
            if self.query_results != None and (not self._is_insert
298
                or isinstance(self.result, Exception)):
299
                
300
                assert self.query_lookup != None
301
                self.query_results[self.query_lookup] = self.CacheCursor(
302
                    util.dict_subset(dicts.AttrsDictView(self),
303
                    ['query', 'result', 'rowcount', 'description']))
304
        
305
        class CacheCursor:
306
            def __init__(self, cached_result): self.__dict__ = cached_result
307
            
308
            def execute(self, *args, **kw_args):
309
                if isinstance(self.result, Exception): raise self.result
310
                # otherwise, result is a rows list
311
                self.iter = iter(self.result)
312
            
313
            def fetchone(self):
314
                try: return self.iter.next()
315
                except StopIteration: return None
316
    
317
    def esc_value(self, value):
318
        try: str_ = self.mogrify('%s', [value])
319
        except NotImplementedError, e:
320
            module = util.root_module(self.db)
321
            if module == 'MySQLdb':
322
                import _mysql
323
                str_ = _mysql.escape_string(value)
324
            else: raise e
325
        return strings.to_unicode(str_)
326
    
327
    def esc_name(self, name): return esc_name(self, name) # calls global func
328
    
329
    def std_code(self, str_):
330
        '''Standardizes SQL code.
331
        * Ensures that string literals are prefixed by `E`
332
        '''
333
        if str_.startswith("'"): str_ = 'E'+str_
334
        return str_
335
    
336
    def can_mogrify(self):
337
        module = util.root_module(self.db)
338
        return module == 'psycopg2'
339
    
340
    def mogrify(self, query, params=None):
341
        if self.can_mogrify(): return self.db.cursor().mogrify(query, params)
342
        else: raise NotImplementedError("Can't mogrify query")
343
    
344
    def print_notices(self):
345
        if hasattr(self.db, 'notices'):
346
            for msg in self.db.notices:
347
                if msg not in self._notices_seen:
348
                    self._notices_seen.add(msg)
349
                    self.log_debug(msg, level=2)
350
    
351
    def run_query(self, query, cacheable=False, log_level=2,
352
        debug_msg_ref=None):
353
        '''
354
        @param log_ignore_excs The log_level will be increased by 2 if the query
355
            throws one of these exceptions.
356
        @param debug_msg_ref If specified, the log message will be returned in
357
            this instead of being output. This allows you to filter log messages
358
            depending on the result of the query.
359
        '''
360
        assert query != None
361
        
362
        if self.autocommit and self.src != None:
363
            query = sql_gen.esc_comment(self.src)+'\t'+query
364
        
365
        if not self.caching: cacheable = False
366
        used_cache = False
367
        
368
        if self.debug:
369
            profiler = profiling.ItersProfiler(start_now=True, iter_text='row')
370
        try:
371
            # Get cursor
372
            if cacheable:
373
                try: cur = self.query_results[query]
374
                except KeyError: cur = self.DbCursor(self)
375
                else: used_cache = True
376
            else: cur = self.db.cursor()
377
            
378
            # Run query
379
            try: cur.execute(query)
380
            except Exception, e:
381
                _add_cursor_info(e, self, query)
382
                raise
383
            else: self.do_autocommit()
384
        finally:
385
            if self.debug:
386
                profiler.stop(self.profile_row_ct)
387
                
388
                ## Log or return query
389
                
390
                query = str(get_cur_query(cur, query))
391
                # Put the src comment on a separate line in the log file
392
                query = query.replace('\t', '\n', 1)
393
                
394
                msg = 'DB query: '
395
                
396
                if used_cache: msg += 'cache hit'
397
                elif cacheable: msg += 'cache miss'
398
                else: msg += 'non-cacheable'
399
                
400
                msg += ':\n'+profiler.msg()+'\n'+strings.as_code(query, 'SQL')
401
                
402
                if debug_msg_ref != None: debug_msg_ref[0] = msg
403
                else: self.log_debug(msg, log_level)
404
                
405
                self.print_notices()
406
        
407
        return cur
408
    
409
    def is_cached(self, query): return query in self.query_results
410
    
411
    def with_autocommit(self, func):
412
        import psycopg2.extensions
413
        
414
        prev_isolation_level = self.db.isolation_level
415
        self.db.set_isolation_level(
416
            psycopg2.extensions.ISOLATION_LEVEL_AUTOCOMMIT)
417
        try: return func()
418
        finally: self.db.set_isolation_level(prev_isolation_level)
419
    
420
    def with_savepoint(self, func):
421
        top = self._savepoint == 0
422
        savepoint = 'level_'+str(self._savepoint)
423
        
424
        if self.debug:
425
            self.log_debug('Begin transaction', level=4)
426
            profiler = profiling.ItersProfiler(start_now=True, iter_text='row')
427
        
428
        # Must happen before running queries so they don't get autocommitted
429
        self._savepoint += 1
430
        
431
        if top: query = 'START TRANSACTION ISOLATION LEVEL READ COMMITTED'
432
        else: query = 'SAVEPOINT '+savepoint
433
        self.run_query(query, log_level=4)
434
        try:
435
            return func()
436
            if top: self.run_query('COMMIT', log_level=4)
437
        except:
438
            if top: query = 'ROLLBACK'
439
            else: query = 'ROLLBACK TO SAVEPOINT '+savepoint
440
            self.run_query(query, log_level=4)
441
            
442
            raise
443
        finally:
444
            # Always release savepoint, because after ROLLBACK TO SAVEPOINT,
445
            # "The savepoint remains valid and can be rolled back to again"
446
            # (http://www.postgresql.org/docs/8.3/static/sql-rollback-to.html).
447
            if not top:
448
                self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
449
            
450
            self._savepoint -= 1
451
            assert self._savepoint >= 0
452
            
453
            if self.debug:
454
                profiler.stop(self.profile_row_ct)
455
                self.log_debug('End transaction\n'+profiler.msg(), level=4)
456
            
457
            self.do_autocommit() # OK to do this after ROLLBACK TO SAVEPOINT
458
    
459
    def do_autocommit(self):
460
        '''Autocommits if outside savepoint'''
461
        assert self._savepoint >= 1
462
        if self.autocommit and self._savepoint == 1:
463
            self.log_debug('Autocommitting', level=4)
464
            self.db.commit()
465
    
466
    def col_info(self, col, cacheable=True):
467
        table = sql_gen.Table('columns', 'information_schema')
468
        type_ = sql_gen.Coalesce(sql_gen.Nullif(sql_gen.Col('data_type'),
469
            'USER-DEFINED'), sql_gen.Col('udt_name'))
470
        cols = [type_, 'column_default',
471
            sql_gen.Cast('boolean', sql_gen.Col('is_nullable'))]
472
        
473
        conds = [('table_name', col.table.name),
474
            ('column_name', strings.ustr(col.name))]
475
        schema = col.table.schema
476
        if schema != None: conds.append(('table_schema', schema))
477
        
478
        cur = select(self, table, cols, conds, order_by='table_schema', limit=1,
479
            cacheable=cacheable, log_level=4) # TODO: order by search_path order
480
        try: type_, default, nullable = row(cur)
481
        except StopIteration: raise sql_gen.NoUnderlyingTableException(col)
482
        default = sql_gen.as_Code(default, self)
483
        
484
        return sql_gen.TypedCol(col.name, type_, default, nullable)
485
    
486
    def TempFunction(self, name):
487
        if self.debug_temp: schema = None
488
        else: schema = 'pg_temp'
489
        return sql_gen.Function(name, schema)
490

    
491
connect = DbConn
492

    
493
##### Recoverable querying
494

    
495
def with_savepoint(db, func): return db.with_savepoint(func)
496

    
497
def run_query(db, query, recover=None, cacheable=False, log_level=2,
498
    log_ignore_excs=None, **kw_args):
499
    '''For params, see DbConn.run_query()'''
500
    if recover == None: recover = False
501
    if log_ignore_excs == None: log_ignore_excs = ()
502
    log_ignore_excs = tuple(log_ignore_excs)
503
    debug_msg_ref = [None]
504
    
505
    query = with_explain_comment(db, query)
506
    
507
    try:
508
        try:
509
            def run(): return db.run_query(query, cacheable, log_level,
510
                debug_msg_ref, **kw_args)
511
            if recover and not db.is_cached(query):
512
                return with_savepoint(db, run)
513
            else: return run() # don't need savepoint if cached
514
        except Exception, e:
515
            msg = strings.ustr(e.args[0])
516
            msg = re.sub(r'^(?:PL/Python: )?ValueError: ', r'', msg)
517
            
518
            match = re.match(r'^duplicate key value violates unique constraint '
519
                r'"(.+?)"', msg)
520
            if match:
521
                constraint, = match.groups()
522
                cols = []
523
                if recover: # need auto-rollback to run index_cols()
524
                    try: cols = index_cols(db, constraint)
525
                    except NotImplementedError: pass
526
                raise DuplicateKeyException(constraint, None, cols, e)
527
            
528
            match = re.match(r'^null value in column "(.+?)" violates not-null'
529
                r' constraint', msg)
530
            if match:
531
                col, = match.groups()
532
                raise NullValueException('NOT NULL', None, [col], e)
533
            
534
            match = re.match(r'^new row for relation "(.+?)" violates check '
535
                r'constraint "(.+?)"', msg)
536
            if match:
537
                table, constraint = match.groups()
538
                constraint = sql_gen.Col(constraint, table)
539
                cond = None
540
                if recover: # need auto-rollback to run constraint_cond()
541
                    try: cond = constraint_cond(db, constraint)
542
                    except NotImplementedError: pass
543
                raise CheckException(constraint.to_str(db), cond, [], e)
544
            
545
            match = re.match(r'^(?:invalid input (?:syntax|value)\b.*?'
546
                r'|.+? out of range): "(.+?)"', msg)
547
            if match:
548
                value, = match.groups()
549
                raise InvalidValueException(strings.to_unicode(value), e)
550
            
551
            match = re.match(r'^column "(.+?)" is of type (.+?) but expression '
552
                r'is of type', msg)
553
            if match:
554
                col, type_ = match.groups()
555
                raise MissingCastException(type_, col, e)
556
            
557
            match = re.match(r'^could not determine polymorphic type because '
558
                r'input has type "unknown"', msg)
559
            if match: raise MissingCastException('text', None, e)
560
            
561
            match = re.match(r'^.+? types .+? and .+? cannot be matched', msg)
562
            if match: raise MissingCastException('text', None, e)
563
            
564
            typed_name_re = r'^(\S+) "(.+?)"'
565
            
566
            match = re.match(typed_name_re+r'.*? already exists', msg)
567
            if match:
568
                type_, name = match.groups()
569
                raise DuplicateException(type_, name, e)
570
            
571
            match = re.match(r'more than one (\S+) named ""(.+?)""', msg)
572
            if match:
573
                type_, name = match.groups()
574
                raise DuplicateException(type_, name, e)
575
            
576
            match = re.match(typed_name_re+r' does not exist', msg)
577
            if match:
578
                type_, name = match.groups()
579
                raise DoesNotExistException(type_, name, e)
580
            
581
            raise # no specific exception raised
582
    except log_ignore_excs:
583
        log_level += 2
584
        raise
585
    finally:
586
        if debug_msg_ref[0] != None: db.log_debug(debug_msg_ref[0], log_level)
587

    
588
##### Basic queries
589

    
590
def is_explainable(query):
591
    # See <http://www.postgresql.org/docs/8.3/static/sql-explain.html#AEN57749>
592
    return re.match(r'^(?:SELECT|INSERT|UPDATE|DELETE|VALUES|EXECUTE|DECLARE)\b'
593
        , query)
594

    
595
def explain(db, query, **kw_args):
596
    '''
597
    For params, see run_query().
598
    '''
599
    kw_args.setdefault('log_level', 4)
600
    
601
    return strings.ustr(strings.join_lines(values(run_query(db,
602
        'EXPLAIN '+query, recover=True, cacheable=True, **kw_args))))
603
        # not a higher log_level because it's useful to see what query is being
604
        # run before it's executed, which EXPLAIN effectively provides
605

    
606
def has_comment(query): return query.endswith('*/')
607

    
608
def with_explain_comment(db, query, **kw_args):
609
    if db.autoexplain and not has_comment(query) and is_explainable(query):
610
        query += '\n'+sql_gen.esc_comment(' EXPLAIN:\n'
611
            +explain(db, query, **kw_args))
612
    return query
613

    
614
def next_version(name):
615
    version = 1 # first existing name was version 0
616
    match = re.match(r'^(.*)#(\d+)$', name)
617
    if match:
618
        name, version = match.groups()
619
        version = int(version)+1
620
    return sql_gen.concat(name, '#'+str(version))
621

    
622
def lock_table(db, table, mode):
623
    table = sql_gen.as_Table(table)
624
    run_query(db, 'LOCK TABLE '+table.to_str(db)+' IN '+mode+' MODE')
625

    
626
def run_query_into(db, query, into=None, add_pkey_=False, **kw_args):
627
    '''Outputs a query to a temp table.
628
    For params, see run_query().
629
    '''
630
    if into == None: return run_query(db, query, **kw_args)
631
    
632
    assert isinstance(into, sql_gen.Table)
633
    
634
    into.is_temp = True
635
    # "temporary tables cannot specify a schema name", so remove schema
636
    into.schema = None
637
    
638
    kw_args['recover'] = True
639
    kw_args.setdefault('log_ignore_excs', (DuplicateException,))
640
    
641
    temp = not db.debug_temp # tables are permanent in debug_temp mode
642
    
643
    # Create table
644
    while True:
645
        create_query = 'CREATE'
646
        if temp: create_query += ' TEMP'
647
        create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
648
        
649
        try:
650
            cur = run_query(db, create_query, **kw_args)
651
                # CREATE TABLE AS sets rowcount to # rows in query
652
            break
653
        except DuplicateException, e:
654
            into.name = next_version(into.name)
655
            # try again with next version of name
656
    
657
    if add_pkey_: add_pkey(db, into)
658
    
659
    # According to the PostgreSQL doc, "The autovacuum daemon cannot access and
660
    # therefore cannot vacuum or analyze temporary tables. [...] if a temporary
661
    # table is going to be used in complex queries, it is wise to run ANALYZE on
662
    # the temporary table after it is populated."
663
    # (http://www.postgresql.org/docs/9.1/static/sql-createtable.html)
664
    # If into is not a temp table, ANALYZE is useful but not required.
665
    analyze(db, into)
666
    
667
    return cur
668

    
669
order_by_pkey = object() # tells mk_select() to order by the pkey
670

    
671
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
672

    
673
def mk_select(db, tables=None, fields=None, conds=None, distinct_on=[],
674
    limit=None, start=None, order_by=order_by_pkey, default_table=None,
675
    explain=True):
676
    '''
677
    @param tables The single table to select from, or a list of tables to join
678
        together, with tables after the first being sql_gen.Join objects
679
    @param fields Use None to select all fields in the table
680
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
681
        * container can be any iterable type
682
        * compare_left_side: sql_gen.Code|str (for col name)
683
        * compare_right_side: sql_gen.ValueCond|literal value
684
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
685
        use all columns
686
    @return query
687
    '''
688
    # Parse tables param
689
    tables = lists.mk_seq(tables)
690
    tables = list(tables) # don't modify input! (list() copies input)
691
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
692
    
693
    # Parse other params
694
    if conds == None: conds = []
695
    elif dicts.is_dict(conds): conds = conds.items()
696
    conds = list(conds) # don't modify input! (list() copies input)
697
    assert limit == None or isinstance(limit, (int, long))
698
    assert start == None or isinstance(start, (int, long))
699
    if order_by is order_by_pkey:
700
        if table0 == None or distinct_on != []: order_by = None
701
        else: order_by = pkey(db, table0, recover=True)
702
    
703
    query = 'SELECT'
704
    
705
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
706
    
707
    # DISTINCT ON columns
708
    if distinct_on != []:
709
        query += '\nDISTINCT'
710
        if distinct_on is not distinct_on_all:
711
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
712
    
713
    # Columns
714
    if query.find('\n') >= 0: whitespace = '\n'
715
    else: whitespace = ' '
716
    if fields == None: query += whitespace+'*'
717
    else:
718
        assert fields != []
719
        if len(fields) > 1: whitespace = '\n'
720
        query += whitespace+('\n, '.join(map(parse_col, fields)))
721
    
722
    # Main table
723
    if query.find('\n') >= 0 or len(tables) > 0: whitespace = '\n'
724
    else: whitespace = ' '
725
    if table0 != None: query += whitespace+'FROM '+table0.to_str(db)
726
    
727
    # Add joins
728
    left_table = table0
729
    for join_ in tables:
730
        table = join_.table
731
        
732
        # Parse special values
733
        if join_.type_ is sql_gen.filter_out: # filter no match
734
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
735
                sql_gen.CompareCond(None, '~=')))
736
        
737
        query += '\n'+join_.to_str(db, left_table)
738
        
739
        left_table = table
740
    
741
    missing = True
742
    if conds != []:
743
        if len(conds) == 1: whitespace = ' '
744
        else: whitespace = '\n'
745
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
746
            .to_str(db) for l, r in conds], 'WHERE')
747
    if order_by != None:
748
        query += '\nORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
749
    if limit != None: query += '\nLIMIT '+str(limit)
750
    if start != None:
751
        if start != 0: query += '\nOFFSET '+str(start)
752
    
753
    if explain: query = with_explain_comment(db, query)
754
    
755
    return query
756

    
757
def select(db, *args, **kw_args):
758
    '''For params, see mk_select() and run_query()'''
759
    recover = kw_args.pop('recover', None)
760
    cacheable = kw_args.pop('cacheable', True)
761
    log_level = kw_args.pop('log_level', 2)
762
    
763
    return run_query(db, mk_select(db, *args, **kw_args), recover, cacheable,
764
        log_level=log_level)
765

    
766
def mk_insert_select(db, table, cols=None, select_query=None, returning=None,
767
    embeddable=False, ignore=False, src=None):
768
    '''
769
    @param returning str|None An inserted column (such as pkey) to return
770
    @param embeddable Whether the query should be embeddable as a nested SELECT.
771
        Warning: If you set this and cacheable=True when the query is run, the
772
        query will be fully cached, not just if it raises an exception.
773
    @param ignore Whether to ignore duplicate keys.
774
    @param src Will be included in the name of any created function, to help
775
        identify the data source in pg_stat_activity.
776
    '''
777
    table = sql_gen.remove_table_rename(sql_gen.as_Table(table))
778
    if cols == []: cols = None # no cols (all defaults) = unknown col names
779
    if cols != None: cols = [sql_gen.to_name_only_col(c, table) for c in cols]
780
    if select_query == None: select_query = 'DEFAULT VALUES'
781
    if returning != None: returning = sql_gen.as_Col(returning, table)
782
    
783
    first_line = 'INSERT INTO '+table.to_str(db)
784
    
785
    def mk_insert(select_query):
786
        query = first_line
787
        if cols != None:
788
            query += '\n('+(', '.join((c.to_str(db) for c in cols)))+')'
789
        query += '\n'+select_query
790
        
791
        if returning != None:
792
            returning_name_col = sql_gen.to_name_only_col(returning)
793
            query += '\nRETURNING '+returning_name_col.to_str(db)
794
        
795
        return query
796
    
797
    return_type = sql_gen.CustomCode('unknown')
798
    if returning != None: return_type = sql_gen.ColType(returning)
799
    
800
    if ignore:
801
        # Always return something to set the correct rowcount
802
        if returning == None: returning = sql_gen.NamedCol('NULL', None)
803
        
804
        embeddable = True # must use function
805
        
806
        if cols == None: row = [sql_gen.Col(sql_gen.all_cols, 'row')]
807
        else: row = [sql_gen.Col(c.name, 'row') for c in cols]
808
        
809
        query = sql_gen.RowExcIgnore(sql_gen.RowType(table), select_query,
810
            sql_gen.ReturnQuery(mk_insert(sql_gen.Values(row).to_str(db))),
811
            cols)
812
    else: query = mk_insert(select_query)
813
    
814
    if embeddable:
815
        # Create function
816
        function_name = sql_gen.clean_name(first_line)
817
        if src != None: function_name = src+': '+function_name
818
        while True:
819
            try:
820
                func = db.TempFunction(function_name)
821
                def_ = sql_gen.FunctionDef(func, sql_gen.SetOf(return_type),
822
                    query)
823
                
824
                run_query(db, def_.to_str(db), recover=True, cacheable=True,
825
                    log_ignore_excs=(DuplicateException,))
826
                break # this version was successful
827
            except DuplicateException, e:
828
                function_name = next_version(function_name)
829
                # try again with next version of name
830
        
831
        # Return query that uses function
832
        cols = None
833
        if returning != None: cols = [returning]
834
        func_table = sql_gen.NamedTable('f', sql_gen.FunctionCall(func), cols)
835
            # AS clause requires function alias
836
        return mk_select(db, func_table, order_by=None)
837
    
838
    return query
839

    
840
def insert_select(db, table, *args, **kw_args):
841
    '''For params, see mk_insert_select() and run_query_into()
842
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
843
        values in
844
    '''
845
    returning = kw_args.get('returning', None)
846
    ignore = kw_args.get('ignore', False)
847
    
848
    into = kw_args.pop('into', None)
849
    if into != None: kw_args['embeddable'] = True
850
    recover = kw_args.pop('recover', None)
851
    if ignore: recover = True
852
    cacheable = kw_args.pop('cacheable', True)
853
    log_level = kw_args.pop('log_level', 2)
854
    
855
    rowcount_only = ignore and returning == None # keep NULL rows on server
856
    if rowcount_only: into = sql_gen.Table('rowcount')
857
    
858
    cur = run_query_into(db, mk_insert_select(db, table, *args, **kw_args),
859
        into, recover=recover, cacheable=cacheable, log_level=log_level)
860
    if rowcount_only: empty_temp(db, into)
861
    autoanalyze(db, table)
862
    return cur
863

    
864
default = sql_gen.default # tells insert() to use the default value for a column
865

    
866
def insert(db, table, row, *args, **kw_args):
867
    '''For params, see insert_select()'''
868
    if lists.is_seq(row): cols = None
869
    else:
870
        cols = row.keys()
871
        row = row.values()
872
    row = list(row) # ensure that "== []" works
873
    
874
    if row == []: query = None
875
    else: query = sql_gen.Values(row).to_str(db)
876
    
877
    return insert_select(db, table, cols, query, *args, **kw_args)
878

    
879
def mk_update(db, table, changes=None, cond=None, in_place=False,
880
    cacheable_=True):
881
    '''
882
    @param changes [(col, new_value),...]
883
        * container can be any iterable type
884
        * col: sql_gen.Code|str (for col name)
885
        * new_value: sql_gen.Code|literal value
886
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
887
    @param in_place If set, locks the table and updates rows in place.
888
        This avoids creating dead rows in PostgreSQL.
889
        * cond must be None
890
    @param cacheable_ Whether column structure information used to generate the
891
        query can be cached
892
    @return str query
893
    '''
894
    table = sql_gen.as_Table(table)
895
    changes = [(sql_gen.to_name_only_col(c, table), sql_gen.as_Value(v))
896
        for c, v in changes]
897
    
898
    if in_place:
899
        assert cond == None
900
        
901
        query = 'ALTER TABLE '+table.to_str(db)+'\n'
902
        query += ',\n'.join(('ALTER COLUMN '+c.to_str(db)+' TYPE '
903
            +db.col_info(sql_gen.with_default_table(c, table), cacheable_).type
904
            +'\nUSING '+v.to_str(db) for c, v in changes))
905
    else:
906
        query = 'UPDATE '+table.to_str(db)+'\nSET\n'
907
        query += ',\n'.join((c.to_str(db)+' = '+v.to_str(db)
908
            for c, v in changes))
909
        if cond != None: query += '\nWHERE\n'+cond.to_str(db)
910
    
911
    query = with_explain_comment(db, query)
912
    
913
    return query
914

    
915
def update(db, table, *args, **kw_args):
916
    '''For params, see mk_update() and run_query()'''
917
    recover = kw_args.pop('recover', None)
918
    cacheable = kw_args.pop('cacheable', False)
919
    log_level = kw_args.pop('log_level', 2)
920
    
921
    cur = run_query(db, mk_update(db, table, *args, **kw_args), recover,
922
        cacheable, log_level=log_level)
923
    autoanalyze(db, table)
924
    return cur
925

    
926
def mk_delete(db, table, cond=None):
927
    '''
928
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
929
    @return str query
930
    '''
931
    query = 'DELETE FROM '+table.to_str(db)
932
    if cond != None: query += '\nWHERE '+cond.to_str(db)
933
    
934
    query = with_explain_comment(db, query)
935
    
936
    return query
937

    
938
def delete(db, table, *args, **kw_args):
939
    '''For params, see mk_delete() and run_query()'''
940
    recover = kw_args.pop('recover', None)
941
    cacheable = kw_args.pop('cacheable', True)
942
    log_level = kw_args.pop('log_level', 2)
943
    
944
    cur = run_query(db, mk_delete(db, table, *args, **kw_args), recover,
945
        cacheable, log_level=log_level)
946
    autoanalyze(db, table)
947
    return cur
948

    
949
def last_insert_id(db):
950
    module = util.root_module(db.db)
951
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
952
    elif module == 'MySQLdb': return db.insert_id()
953
    else: return None
954

    
955
def define_func(db, def_):
956
    func = def_.function
957
    while True:
958
        try:
959
            run_query(db, def_.to_str(db), recover=True, cacheable=True,
960
                log_ignore_excs=(DuplicateException,))
961
            break # successful
962
        except DuplicateException:
963
            func.name = next_version(func.name)
964
            # try again with next version of name
965

    
966
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
967
    '''Creates a mapping from original column names (which may have collisions)
968
    to names that will be distinct among the columns' tables.
969
    This is meant to be used for several tables that are being joined together.
970
    @param cols The columns to combine. Duplicates will be removed.
971
    @param into The table for the new columns.
972
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
973
        columns will be included in the mapping even if they are not in cols.
974
        The tables of the provided Col objects will be changed to into, so make
975
        copies of them if you want to keep the original tables.
976
    @param as_items Whether to return a list of dict items instead of a dict
977
    @return dict(orig_col=new_col, ...)
978
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
979
        * new_col: sql_gen.Col(orig_col_name, into)
980
        * All mappings use the into table so its name can easily be
981
          changed for all columns at once
982
    '''
983
    cols = lists.uniqify(cols)
984
    
985
    items = []
986
    for col in preserve:
987
        orig_col = copy.copy(col)
988
        col.table = into
989
        items.append((orig_col, col))
990
    preserve = set(preserve)
991
    for col in cols:
992
        if col not in preserve:
993
            items.append((col, sql_gen.Col(strings.ustr(col), into, col.srcs)))
994
    
995
    if not as_items: items = dict(items)
996
    return items
997

    
998
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
999
    '''For params, see mk_flatten_mapping()
1000
    @return See return value of mk_flatten_mapping()
1001
    '''
1002
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
1003
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
1004
    run_query_into(db, mk_select(db, joins, cols, order_by=None, limit=limit,
1005
        start=start), into=into, add_pkey_=True)
1006
        # don't cache because the temp table will usually be truncated after use
1007
    return dict(items)
1008

    
1009
##### Database structure introspection
1010

    
1011
#### Expressions
1012

    
1013
bool_re = r'(?:true|false)'
1014

    
1015
def simplify_expr(expr):
1016
    expr = expr.replace('(NULL IS NULL)', 'true')
1017
    expr = expr.replace('(NULL IS NOT NULL)', 'false')
1018
    expr = re.sub(r' OR '+bool_re, r'', expr)
1019
    expr = re.sub(bool_re+r' OR ', r'', expr)
1020
    while True:
1021
        expr, n = re.subn(r'\((\([^()]*\))\)', r'\1', expr)
1022
        if n == 0: break
1023
    return expr
1024

    
1025
name_re = r'(?:\w+|(?:"[^"]*")+)'
1026

    
1027
def parse_expr_col(str_):
1028
    match = re.match(r'^\('+name_re+r'\(('+name_re+r').*\)\)$', str_)
1029
    if match: str_ = match.group(1)
1030
    return sql_gen.unesc_name(str_)
1031

    
1032
def map_expr(db, expr, mapping, in_cols_found=None):
1033
    '''Replaces output columns with input columns in an expression.
1034
    @param in_cols_found If set, will be filled in with the expr's (input) cols
1035
    '''
1036
    for out, in_ in mapping.iteritems():
1037
        orig_expr = expr
1038
        out = sql_gen.to_name_only_col(out)
1039
        in_str = sql_gen.to_name_only_col(sql_gen.remove_col_rename(in_)
1040
            ).to_str(db)
1041
        
1042
        # Replace out both with and without quotes
1043
        expr = expr.replace(out.to_str(db), in_str)
1044
        expr = re.sub(r'(?<!\.)\b'+out.name+r'\b(?!\.)', in_str, expr)
1045
        
1046
        if in_cols_found != None and expr != orig_expr: # replaced something
1047
            in_cols_found.append(in_)
1048
    
1049
    return simplify_expr(expr)
1050

    
1051
#### Tables
1052

    
1053
def tables(db, schema_like='public', table_like='%', exact=False):
1054
    if exact: compare = '='
1055
    else: compare = 'LIKE'
1056
    
1057
    module = util.root_module(db.db)
1058
    if module == 'psycopg2':
1059
        conds = [('schemaname', sql_gen.CompareCond(schema_like, compare)),
1060
            ('tablename', sql_gen.CompareCond(table_like, compare))]
1061
        return values(select(db, 'pg_tables', ['tablename'], conds,
1062
            order_by='tablename', log_level=4))
1063
    elif module == 'MySQLdb':
1064
        return values(run_query(db, 'SHOW TABLES LIKE '+db.esc_value(table_like)
1065
            , cacheable=True, log_level=4))
1066
    else: raise NotImplementedError("Can't list tables for "+module+' database')
1067

    
1068
def table_exists(db, table):
1069
    table = sql_gen.as_Table(table)
1070
    return list(tables(db, table.schema, table.name, exact=True)) != []
1071

    
1072
def table_row_count(db, table, recover=None):
1073
    return value(run_query(db, mk_select(db, table, [sql_gen.row_count],
1074
        order_by=None), recover=recover, log_level=3))
1075

    
1076
def table_cols(db, table, recover=None):
1077
    return list(col_names(select(db, table, limit=0, order_by=None,
1078
        recover=recover, log_level=4)))
1079

    
1080
pkey_col = 'row_num'
1081

    
1082
def pkey(db, table, recover=None):
1083
    '''Uses pkey_col, or if not found, the first column in the table.'''
1084
    cols = table_cols(db, table, recover)
1085
    if pkey_col in cols: return pkey_col
1086
    else: return cols[0]
1087

    
1088
not_null_col = 'not_null_col'
1089

    
1090
def table_not_null_col(db, table, recover=None):
1091
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
1092
    if not_null_col in table_cols(db, table, recover): return not_null_col
1093
    else: return pkey(db, table, recover)
1094

    
1095
def constraint_cond(db, constraint):
1096
    module = util.root_module(db.db)
1097
    if module == 'psycopg2':
1098
        table_str = sql_gen.Literal(constraint.table.to_str(db))
1099
        name_str = sql_gen.Literal(constraint.name)
1100
        return value(run_query(db, '''\
1101
SELECT consrc
1102
FROM pg_constraint
1103
WHERE
1104
conrelid = '''+table_str.to_str(db)+'''::regclass
1105
AND conname = '''+name_str.to_str(db)+'''
1106
'''
1107
            , cacheable=True, log_level=4))
1108
    else: raise NotImplementedError("Can't list index columns for "+module+
1109
        ' database')
1110

    
1111
def index_cols(db, index):
1112
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
1113
    automatically created. When you don't know whether something is a UNIQUE
1114
    constraint or a UNIQUE index, use this function.'''
1115
    index = sql_gen.as_Table(index)
1116
    module = util.root_module(db.db)
1117
    if module == 'psycopg2':
1118
        qual_index = sql_gen.Literal(index.to_str(db))
1119
        return map(parse_expr_col, values(run_query(db, '''\
1120
SELECT pg_get_indexdef(indexrelid, generate_series(1, indnatts), true)
1121
FROM pg_index
1122
WHERE indexrelid = '''+qual_index.to_str(db)+'''::regclass
1123
'''
1124
            , cacheable=True, log_level=4)))
1125
    else: raise NotImplementedError("Can't list index columns for "+module+
1126
        ' database')
1127

    
1128
#### Functions
1129

    
1130
def function_exists(db, function):
1131
    qual_function = sql_gen.Literal(function.to_str(db))
1132
    try:
1133
        select(db, fields=[sql_gen.Cast('regproc', qual_function)],
1134
            recover=True, cacheable=True, log_level=4)
1135
    except DoesNotExistException: return False
1136
    except DuplicateException: return True # overloaded function
1137
    else: return True
1138

    
1139
##### Structural changes
1140

    
1141
#### Columns
1142

    
1143
def add_col(db, table, col, comment=None, **kw_args):
1144
    '''
1145
    @param col TypedCol Name may be versioned, so be sure to propagate any
1146
        renaming back to any source column for the TypedCol.
1147
    @param comment None|str SQL comment used to distinguish columns of the same
1148
        name from each other when they contain different data, to allow the
1149
        ADD COLUMN query to be cached. If not set, query will not be cached.
1150
    '''
1151
    assert isinstance(col, sql_gen.TypedCol)
1152
    
1153
    while True:
1154
        str_ = 'ALTER TABLE '+table.to_str(db)+' ADD COLUMN '+col.to_str(db)
1155
        if comment != None: str_ += ' '+sql_gen.esc_comment(comment)
1156
        
1157
        try:
1158
            run_query(db, str_, recover=True, cacheable=True, **kw_args)
1159
            break
1160
        except DuplicateException:
1161
            col.name = next_version(col.name)
1162
            # try again with next version of name
1163

    
1164
def add_not_null(db, col):
1165
    table = col.table
1166
    col = sql_gen.to_name_only_col(col)
1167
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ALTER COLUMN '
1168
        +col.to_str(db)+' SET NOT NULL', cacheable=True, log_level=3)
1169

    
1170
def drop_not_null(db, col):
1171
    table = col.table
1172
    col = sql_gen.to_name_only_col(col)
1173
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ALTER COLUMN '
1174
        +col.to_str(db)+' DROP NOT NULL', cacheable=True, log_level=3)
1175

    
1176
row_num_col = '_row_num'
1177

    
1178
row_num_typed_col = sql_gen.TypedCol(row_num_col, 'serial', nullable=False,
1179
    constraints='PRIMARY KEY')
1180

    
1181
def add_row_num(db, table):
1182
    '''Adds a row number column to a table. Its name is in row_num_col. It will
1183
    be the primary key.'''
1184
    add_col(db, table, row_num_typed_col, log_level=3)
1185

    
1186
#### Indexes
1187

    
1188
def add_pkey(db, table, cols=None, recover=None):
1189
    '''Adds a primary key.
1190
    @param cols [sql_gen.Col,...] The columns in the primary key.
1191
        Defaults to the first column in the table.
1192
    @pre The table must not already have a primary key.
1193
    '''
1194
    table = sql_gen.as_Table(table)
1195
    if cols == None: cols = [pkey(db, table, recover)]
1196
    col_strs = [sql_gen.to_name_only_col(v).to_str(db) for v in cols]
1197
    
1198
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD PRIMARY KEY ('
1199
        +(', '.join(col_strs))+')', recover=True, cacheable=True, log_level=3,
1200
        log_ignore_excs=(DuplicateException,))
1201

    
1202
def add_index(db, exprs, table=None, unique=False, ensure_not_null_=True):
1203
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
1204
    Currently, only function calls and literal values are supported expressions.
1205
    @param ensure_not_null_ If set, translates NULL values to sentinel values.
1206
        This allows indexes to be used for comparisons where NULLs are equal.
1207
    '''
1208
    exprs = lists.mk_seq(exprs)
1209
    
1210
    # Parse exprs
1211
    old_exprs = exprs[:]
1212
    exprs = []
1213
    cols = []
1214
    for i, expr in enumerate(old_exprs):
1215
        expr = sql_gen.as_Col(expr, table)
1216
        
1217
        # Handle nullable columns
1218
        if ensure_not_null_:
1219
            try: expr = sql_gen.ensure_not_null(db, expr)
1220
            except KeyError: pass # unknown type, so just create plain index
1221
        
1222
        # Extract col
1223
        expr = copy.deepcopy(expr) # don't modify input!
1224
        col = expr
1225
        if isinstance(expr, sql_gen.FunctionCall): col = expr.args[0]
1226
        expr = sql_gen.cast_literal(expr)
1227
        if not isinstance(expr, (sql_gen.Expr, sql_gen.Col)):
1228
            expr = sql_gen.Expr(expr)
1229
            
1230
        
1231
        # Extract table
1232
        if table == None:
1233
            assert sql_gen.is_table_col(col)
1234
            table = col.table
1235
        
1236
        if isinstance(col, sql_gen.Col): col.table = None
1237
        
1238
        exprs.append(expr)
1239
        cols.append(col)
1240
    
1241
    table = sql_gen.as_Table(table)
1242
    
1243
    # Add index
1244
    str_ = 'CREATE'
1245
    if unique: str_ += ' UNIQUE'
1246
    str_ += ' INDEX ON '+table.to_str(db)+' ('+(
1247
        ', '.join((v.to_str(db) for v in exprs)))+')'
1248
    run_query(db, str_, recover=True, cacheable=True, log_level=3)
1249

    
1250
already_indexed = object() # tells add_indexes() the pkey has already been added
1251

    
1252
def add_indexes(db, table, has_pkey=True):
1253
    '''Adds an index on all columns in a table.
1254
    @param has_pkey bool|already_indexed Whether a pkey instead of a regular
1255
        index should be added on the first column.
1256
        * If already_indexed, the pkey is assumed to have already been added
1257
    '''
1258
    cols = table_cols(db, table)
1259
    if has_pkey:
1260
        if has_pkey is not already_indexed: add_pkey(db, table)
1261
        cols = cols[1:]
1262
    for col in cols: add_index(db, col, table)
1263

    
1264
#### Tables
1265

    
1266
### Maintenance
1267

    
1268
def analyze(db, table):
1269
    table = sql_gen.as_Table(table)
1270
    run_query(db, 'ANALYZE '+table.to_str(db), log_level=3)
1271

    
1272
def autoanalyze(db, table):
1273
    if db.autoanalyze: analyze(db, table)
1274

    
1275
def vacuum(db, table):
1276
    table = sql_gen.as_Table(table)
1277
    db.with_autocommit(lambda: run_query(db, 'VACUUM ANALYZE '+table.to_str(db),
1278
        log_level=3))
1279

    
1280
### Lifecycle
1281

    
1282
def drop(db, type_, name):
1283
    name = sql_gen.as_Name(name)
1284
    run_query(db, 'DROP '+type_+' IF EXISTS '+name.to_str(db)+' CASCADE')
1285

    
1286
def drop_table(db, table): drop(db, 'TABLE', table)
1287

    
1288
def create_table(db, table, cols=[], has_pkey=True, col_indexes=True,
1289
    like=None):
1290
    '''Creates a table.
1291
    @param cols [sql_gen.TypedCol,...] The column names and types
1292
    @param has_pkey If set, the first column becomes the primary key.
1293
    @param col_indexes bool|[ref]
1294
        * If True, indexes will be added on all non-pkey columns.
1295
        * If a list reference, [0] will be set to a function to do this.
1296
          This can be used to delay index creation until the table is populated.
1297
    '''
1298
    table = sql_gen.as_Table(table)
1299
    
1300
    if like != None:
1301
        cols = [sql_gen.CustomCode('LIKE '+like.to_str(db)+' INCLUDING ALL')
1302
            ]+cols
1303
    if has_pkey:
1304
        cols[0] = pkey = copy.copy(cols[0]) # don't modify input!
1305
        pkey.constraints = 'PRIMARY KEY'
1306
    
1307
    temp = table.is_temp and not db.debug_temp
1308
        # temp tables permanent in debug_temp mode
1309
    
1310
    # Create table
1311
    def create():
1312
        str_ = 'CREATE'
1313
        if temp: str_ += ' TEMP'
1314
        str_ += ' TABLE '+table.to_str(db)+' (\n'
1315
        str_ += '\n, '.join(c.to_str(db) for c in cols)
1316
        str_ += '\n);'
1317
        
1318
        run_query(db, str_, recover=True, cacheable=True, log_level=2,
1319
            log_ignore_excs=(DuplicateException,))
1320
    if table.is_temp:
1321
        while True:
1322
            try:
1323
                create()
1324
                break
1325
            except DuplicateException:
1326
                table.name = next_version(table.name)
1327
                # try again with next version of name
1328
    else: create()
1329
    
1330
    # Add indexes
1331
    if has_pkey: has_pkey = already_indexed
1332
    def add_indexes_(): add_indexes(db, table, has_pkey)
1333
    if isinstance(col_indexes, list): col_indexes[0] = add_indexes_ # defer
1334
    elif col_indexes: add_indexes_() # add now
1335

    
1336
def copy_table_struct(db, src, dest):
1337
    '''Creates a structure-only copy of a table. (Does not copy data.)'''
1338
    create_table(db, dest, has_pkey=False, col_indexes=False, like=src)
1339

    
1340
### Data
1341

    
1342
def truncate(db, table, schema='public', **kw_args):
1343
    '''For params, see run_query()'''
1344
    table = sql_gen.as_Table(table, schema)
1345
    return run_query(db, 'TRUNCATE '+table.to_str(db)+' CASCADE', **kw_args)
1346

    
1347
def empty_temp(db, tables):
1348
    tables = lists.mk_seq(tables)
1349
    for table in tables: truncate(db, table, log_level=3)
1350

    
1351
def empty_db(db, schema='public', **kw_args):
1352
    '''For kw_args, see tables()'''
1353
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
1354

    
1355
def distinct_table(db, table, distinct_on):
1356
    '''Creates a copy of a temp table which is distinct on the given columns.
1357
    The old and new tables will both get an index on these columns, to
1358
    facilitate merge joins.
1359
    @param distinct_on If empty, creates a table with one row. This is useful if
1360
        your distinct_on columns are all literal values.
1361
    @return The new table.
1362
    '''
1363
    new_table = sql_gen.suffixed_table(table, '_distinct')
1364
    distinct_on = filter(sql_gen.is_table_col, distinct_on)
1365
    
1366
    copy_table_struct(db, table, new_table)
1367
    
1368
    limit = None
1369
    if distinct_on == []: limit = 1 # one sample row
1370
    else:
1371
        add_index(db, distinct_on, new_table, unique=True)
1372
        add_index(db, distinct_on, table) # for join optimization
1373
    
1374
    insert_select(db, new_table, None, mk_select(db, table, order_by=None,
1375
        limit=limit), ignore=True)
1376
    analyze(db, new_table)
1377
    
1378
    return new_table
(27-27/40)