Project

General

Profile

1 11 aaronmk
# Database access
2
3 1869 aaronmk
import copy
4 11 aaronmk
import re
5 3238 aaronmk
import time
6 865 aaronmk
import warnings
7 11 aaronmk
8 300 aaronmk
import exc
9 1909 aaronmk
import dicts
10 1893 aaronmk
import iters
11 1960 aaronmk
import lists
12 3241 aaronmk
import profiling
13 1889 aaronmk
from Proxy import Proxy
14 1872 aaronmk
import rand
15 2217 aaronmk
import sql_gen
16 862 aaronmk
import strings
17 131 aaronmk
import util
18 11 aaronmk
19 832 aaronmk
##### Exceptions
20
21 2804 aaronmk
def get_cur_query(cur, input_query=None):
22 2168 aaronmk
    raw_query = None
23
    if hasattr(cur, 'query'): raw_query = cur.query
24
    elif hasattr(cur, '_last_executed'): raw_query = cur._last_executed
25 2170 aaronmk
26
    if raw_query != None: return raw_query
27 2804 aaronmk
    else: return '[input] '+strings.ustr(input_query)
28 14 aaronmk
29 2170 aaronmk
def _add_cursor_info(e, *args, **kw_args):
30
    '''For params, see get_cur_query()'''
31 2771 aaronmk
    exc.add_msg(e, 'query: '+strings.ustr(get_cur_query(*args, **kw_args)))
32 135 aaronmk
33 300 aaronmk
class DbException(exc.ExceptionWithCause):
34 14 aaronmk
    def __init__(self, msg, cause=None, cur=None):
35 2145 aaronmk
        exc.ExceptionWithCause.__init__(self, msg, cause, cause_newline=True)
36 14 aaronmk
        if cur != None: _add_cursor_info(self, cur)
37
38 2143 aaronmk
class ExceptionWithName(DbException):
39
    def __init__(self, name, cause=None):
40 2484 aaronmk
        DbException.__init__(self, 'for name: '+strings.as_tt(str(name)), cause)
41 2143 aaronmk
        self.name = name
42 360 aaronmk
43 3109 aaronmk
class ExceptionWithValue(DbException):
44
    def __init__(self, value, cause=None):
45
        DbException.__init__(self, 'for value: '+strings.as_tt(repr(value)),
46
            cause)
47 2240 aaronmk
        self.value = value
48
49 2945 aaronmk
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 2306 aaronmk
class ConstraintException(DbException):
57
    def __init__(self, name, cols, cause=None):
58 2484 aaronmk
        DbException.__init__(self, 'Violated '+strings.as_tt(name)
59
            +' constraint on columns: '+strings.as_tt(', '.join(cols)), cause)
60 2306 aaronmk
        self.name = name
61 468 aaronmk
        self.cols = cols
62 11 aaronmk
63 2523 aaronmk
class MissingCastException(DbException):
64
    def __init__(self, type_, col, cause=None):
65
        DbException.__init__(self, 'Missing cast to type '+strings.as_tt(type_)
66
            +' on column: '+strings.as_tt(col), cause)
67
        self.type = type_
68
        self.col = col
69
70 2143 aaronmk
class NameException(DbException): pass
71
72 2306 aaronmk
class DuplicateKeyException(ConstraintException): pass
73 13 aaronmk
74 2306 aaronmk
class NullValueException(ConstraintException): pass
75 13 aaronmk
76 3109 aaronmk
class InvalidValueException(ExceptionWithValue): pass
77 2239 aaronmk
78 2945 aaronmk
class DuplicateException(ExceptionWithNameType): pass
79 2143 aaronmk
80 89 aaronmk
class EmptyRowException(DbException): pass
81
82 865 aaronmk
##### Warnings
83
84
class DbWarning(UserWarning): pass
85
86 1930 aaronmk
##### Result retrieval
87
88
def col_names(cur): return (col[0] for col in cur.description)
89
90
def rows(cur): return iter(lambda: cur.fetchone(), None)
91
92
def consume_rows(cur):
93
    '''Used to fetch all rows so result will be cached'''
94
    iters.consume_iter(rows(cur))
95
96
def next_row(cur): return rows(cur).next()
97
98
def row(cur):
99
    row_ = next_row(cur)
100
    consume_rows(cur)
101
    return row_
102
103
def next_value(cur): return next_row(cur)[0]
104
105
def value(cur): return row(cur)[0]
106
107
def values(cur): return iters.func_iter(lambda: next_value(cur))
108
109
def value_or_none(cur):
110
    try: return value(cur)
111
    except StopIteration: return None
112
113 2762 aaronmk
##### Escaping
114 2101 aaronmk
115 2573 aaronmk
def esc_name_by_module(module, name):
116
    if module == 'psycopg2' or module == None: quote = '"'
117 2101 aaronmk
    elif module == 'MySQLdb': quote = '`'
118
    else: raise NotImplementedError("Can't escape name for "+module+' database')
119 2500 aaronmk
    return sql_gen.esc_name(name, quote)
120 2101 aaronmk
121
def esc_name_by_engine(engine, name, **kw_args):
122
    return esc_name_by_module(db_engines[engine][0], name, **kw_args)
123
124
def esc_name(db, name, **kw_args):
125
    return esc_name_by_module(util.root_module(db.db), name, **kw_args)
126
127
def qual_name(db, schema, table):
128
    def esc_name_(name): return esc_name(db, name)
129
    table = esc_name_(table)
130
    if schema != None: return esc_name_(schema)+'.'+table
131
    else: return table
132
133 1869 aaronmk
##### Database connections
134 1849 aaronmk
135 2097 aaronmk
db_config_names = ['engine', 'host', 'user', 'password', 'database', 'schemas']
136 1926 aaronmk
137 1869 aaronmk
db_engines = {
138
    'MySQL': ('MySQLdb', {'password': 'passwd', 'database': 'db'}),
139
    'PostgreSQL': ('psycopg2', {}),
140
}
141
142
DatabaseErrors_set = set([DbException])
143
DatabaseErrors = tuple(DatabaseErrors_set)
144
145
def _add_module(module):
146
    DatabaseErrors_set.add(module.DatabaseError)
147
    global DatabaseErrors
148
    DatabaseErrors = tuple(DatabaseErrors_set)
149
150
def db_config_str(db_config):
151
    return db_config['engine']+' database '+db_config['database']
152
153 2448 aaronmk
log_debug_none = lambda msg, level=2: None
154 1901 aaronmk
155 1849 aaronmk
class DbConn:
156 2923 aaronmk
    def __init__(self, db_config, autocommit=True, caching=True,
157 3183 aaronmk
        log_debug=log_debug_none, debug_temp=False, src=None):
158 2915 aaronmk
        '''
159
        @param debug_temp Whether temporary objects should instead be permanent.
160
            This assists in debugging the internal objects used by the program.
161 3183 aaronmk
        @param src In autocommit mode, will be included in a comment in every
162
            query, to help identify the data source in pg_stat_activity.
163 2915 aaronmk
        '''
164 1869 aaronmk
        self.db_config = db_config
165 2190 aaronmk
        self.autocommit = autocommit
166
        self.caching = caching
167 1901 aaronmk
        self.log_debug = log_debug
168 2193 aaronmk
        self.debug = log_debug != log_debug_none
169 2915 aaronmk
        self.debug_temp = debug_temp
170 3183 aaronmk
        self.src = src
171 3074 aaronmk
        self.autoanalyze = False
172 3269 aaronmk
        self.autoexplain = False
173
        self.profile_row_ct = None
174 1869 aaronmk
175 3124 aaronmk
        self._savepoint = 0
176 3120 aaronmk
        self._reset()
177 1869 aaronmk
178
    def __getattr__(self, name):
179
        if name == '__dict__': raise Exception('getting __dict__')
180
        if name == 'db': return self._db()
181
        else: raise AttributeError()
182
183
    def __getstate__(self):
184
        state = copy.copy(self.__dict__) # shallow copy
185 1915 aaronmk
        state['log_debug'] = None # don't pickle the debug callback
186 1869 aaronmk
        state['_DbConn__db'] = None # don't pickle the connection
187
        return state
188
189 3118 aaronmk
    def clear_cache(self): self.query_results = {}
190
191 3120 aaronmk
    def _reset(self):
192 3118 aaronmk
        self.clear_cache()
193 3124 aaronmk
        assert self._savepoint == 0
194 3118 aaronmk
        self._notices_seen = set()
195
        self.__db = None
196
197 2165 aaronmk
    def connected(self): return self.__db != None
198
199 3116 aaronmk
    def close(self):
200 3119 aaronmk
        if not self.connected(): return
201
202 3135 aaronmk
        # Record that the automatic transaction is now closed
203 3136 aaronmk
        self._savepoint -= 1
204 3135 aaronmk
205 3119 aaronmk
        self.db.close()
206 3120 aaronmk
        self._reset()
207 3116 aaronmk
208 3125 aaronmk
    def reconnect(self):
209
        # Do not do this in test mode as it would roll back everything
210
        if self.autocommit: self.close()
211
        # Connection will be reopened automatically on first query
212
213 1869 aaronmk
    def _db(self):
214
        if self.__db == None:
215
            # Process db_config
216
            db_config = self.db_config.copy() # don't modify input!
217 2097 aaronmk
            schemas = db_config.pop('schemas', None)
218 1869 aaronmk
            module_name, mappings = db_engines[db_config.pop('engine')]
219
            module = __import__(module_name)
220
            _add_module(module)
221
            for orig, new in mappings.iteritems():
222
                try: util.rename_key(db_config, orig, new)
223
                except KeyError: pass
224
225
            # Connect
226
            self.__db = module.connect(**db_config)
227
228 3161 aaronmk
            # Record that a transaction is already open
229
            self._savepoint += 1
230
231 1869 aaronmk
            # Configure connection
232 2906 aaronmk
            if hasattr(self.db, 'set_isolation_level'):
233
                import psycopg2.extensions
234
                self.db.set_isolation_level(
235
                    psycopg2.extensions.ISOLATION_LEVEL_READ_COMMITTED)
236 2101 aaronmk
            if schemas != None:
237 2893 aaronmk
                search_path = [self.esc_name(s) for s in schemas.split(',')]
238
                search_path.append(value(run_query(self, 'SHOW search_path',
239
                    log_level=4)))
240
                run_query(self, 'SET search_path TO '+(','.join(search_path)),
241
                    log_level=3)
242 1869 aaronmk
243
        return self.__db
244 1889 aaronmk
245 1891 aaronmk
    class DbCursor(Proxy):
246 1927 aaronmk
        def __init__(self, outer):
247 1891 aaronmk
            Proxy.__init__(self, outer.db.cursor())
248 2191 aaronmk
            self.outer = outer
249 1927 aaronmk
            self.query_results = outer.query_results
250 1894 aaronmk
            self.query_lookup = None
251 1891 aaronmk
            self.result = []
252 1889 aaronmk
253 2802 aaronmk
        def execute(self, query):
254 2764 aaronmk
            self._is_insert = query.startswith('INSERT')
255 2797 aaronmk
            self.query_lookup = query
256 2148 aaronmk
            try:
257 3162 aaronmk
                try: cur = self.inner.execute(query)
258 2802 aaronmk
                finally: self.query = get_cur_query(self.inner, query)
259 1904 aaronmk
            except Exception, e:
260
                self.result = e # cache the exception as the result
261
                self._cache_result()
262
                raise
263 3004 aaronmk
264
            # Always cache certain queries
265 3183 aaronmk
            query = sql_gen.lstrip(query)
266 3004 aaronmk
            if query.startswith('CREATE') or query.startswith('ALTER'):
267 3007 aaronmk
                # structural changes
268 3040 aaronmk
                # Rest of query must be unique in the face of name collisions,
269
                # so don't cache ADD COLUMN unless it has distinguishing comment
270
                if query.find('ADD COLUMN') < 0 or query.endswith('*/'):
271 3007 aaronmk
                    self._cache_result()
272 3004 aaronmk
            elif self.rowcount == 0 and query.startswith('SELECT'): # empty
273 2800 aaronmk
                consume_rows(self) # fetch all rows so result will be cached
274 3004 aaronmk
275 2762 aaronmk
            return cur
276 1894 aaronmk
277 1891 aaronmk
        def fetchone(self):
278
            row = self.inner.fetchone()
279 1899 aaronmk
            if row != None: self.result.append(row)
280
            # otherwise, fetched all rows
281 1904 aaronmk
            else: self._cache_result()
282
            return row
283
284
        def _cache_result(self):
285 2948 aaronmk
            # For inserts that return a result set, don't cache result set since
286
            # inserts are not idempotent. Other non-SELECT queries don't have
287
            # their result set read, so only exceptions will be cached (an
288
            # invalid query will always be invalid).
289 1930 aaronmk
            if self.query_results != None and (not self._is_insert
290 1906 aaronmk
                or isinstance(self.result, Exception)):
291
292 1894 aaronmk
                assert self.query_lookup != None
293 1916 aaronmk
                self.query_results[self.query_lookup] = self.CacheCursor(
294
                    util.dict_subset(dicts.AttrsDictView(self),
295
                    ['query', 'result', 'rowcount', 'description']))
296 1906 aaronmk
297 1916 aaronmk
        class CacheCursor:
298
            def __init__(self, cached_result): self.__dict__ = cached_result
299
300 1927 aaronmk
            def execute(self, *args, **kw_args):
301 1916 aaronmk
                if isinstance(self.result, Exception): raise self.result
302
                # otherwise, result is a rows list
303
                self.iter = iter(self.result)
304
305
            def fetchone(self):
306
                try: return self.iter.next()
307
                except StopIteration: return None
308 1891 aaronmk
309 2212 aaronmk
    def esc_value(self, value):
310 2663 aaronmk
        try: str_ = self.mogrify('%s', [value])
311
        except NotImplementedError, e:
312
            module = util.root_module(self.db)
313
            if module == 'MySQLdb':
314
                import _mysql
315
                str_ = _mysql.escape_string(value)
316
            else: raise e
317 2374 aaronmk
        return strings.to_unicode(str_)
318 2212 aaronmk
319 2347 aaronmk
    def esc_name(self, name): return esc_name(self, name) # calls global func
320
321 2814 aaronmk
    def std_code(self, str_):
322
        '''Standardizes SQL code.
323
        * Ensures that string literals are prefixed by `E`
324
        '''
325
        if str_.startswith("'"): str_ = 'E'+str_
326
        return str_
327
328 2665 aaronmk
    def can_mogrify(self):
329 2663 aaronmk
        module = util.root_module(self.db)
330 2665 aaronmk
        return module == 'psycopg2'
331 2663 aaronmk
332 2665 aaronmk
    def mogrify(self, query, params=None):
333
        if self.can_mogrify(): return self.db.cursor().mogrify(query, params)
334
        else: raise NotImplementedError("Can't mogrify query")
335
336 2671 aaronmk
    def print_notices(self):
337 2725 aaronmk
        if hasattr(self.db, 'notices'):
338
            for msg in self.db.notices:
339
                if msg not in self._notices_seen:
340
                    self._notices_seen.add(msg)
341
                    self.log_debug(msg, level=2)
342 2671 aaronmk
343 2793 aaronmk
    def run_query(self, query, cacheable=False, log_level=2,
344 2464 aaronmk
        debug_msg_ref=None):
345 2445 aaronmk
        '''
346 2464 aaronmk
        @param log_ignore_excs The log_level will be increased by 2 if the query
347
            throws one of these exceptions.
348 2664 aaronmk
        @param debug_msg_ref If specified, the log message will be returned in
349
            this instead of being output. This allows you to filter log messages
350
            depending on the result of the query.
351 2445 aaronmk
        '''
352 2167 aaronmk
        assert query != None
353
354 3183 aaronmk
        if self.autocommit and self.src != None:
355 3206 aaronmk
            query = sql_gen.esc_comment(self.src)+'\t'+query
356 3183 aaronmk
357 2047 aaronmk
        if not self.caching: cacheable = False
358 1903 aaronmk
        used_cache = False
359 2664 aaronmk
360 3242 aaronmk
        if self.debug:
361
            profiler = profiling.ItersProfiler(start_now=True, iter_text='row')
362 1903 aaronmk
        try:
363 1927 aaronmk
            # Get cursor
364
            if cacheable:
365 3238 aaronmk
                try: cur = self.query_results[query]
366 1927 aaronmk
                except KeyError: cur = self.DbCursor(self)
367 3238 aaronmk
                else: used_cache = True
368 1927 aaronmk
            else: cur = self.db.cursor()
369
370
            # Run query
371 3238 aaronmk
            try: cur.execute(query)
372 3162 aaronmk
            except Exception, e:
373
                _add_cursor_info(e, self, query)
374
                raise
375 3238 aaronmk
            else: self.do_autocommit()
376 1903 aaronmk
        finally:
377 3242 aaronmk
            if self.debug:
378 3244 aaronmk
                profiler.stop(self.profile_row_ct)
379 3242 aaronmk
380
                ## Log or return query
381
382 3239 aaronmk
                query = str(get_cur_query(cur, query))
383 3281 aaronmk
                # Put the src comment on a separate line in the log file
384
                query = query.replace('\t', '\n', 1)
385 3239 aaronmk
386 3240 aaronmk
                msg = 'DB query: '
387 3239 aaronmk
388 3240 aaronmk
                if used_cache: msg += 'cache hit'
389
                elif cacheable: msg += 'cache miss'
390
                else: msg += 'non-cacheable'
391 3239 aaronmk
392 3241 aaronmk
                msg += ':\n'+profiler.msg()+'\n'+strings.as_code(query, 'SQL')
393 3240 aaronmk
394 3237 aaronmk
                if debug_msg_ref != None: debug_msg_ref[0] = msg
395
                else: self.log_debug(msg, log_level)
396 3245 aaronmk
397
                self.print_notices()
398 1903 aaronmk
399
        return cur
400 1914 aaronmk
401 2797 aaronmk
    def is_cached(self, query): return query in self.query_results
402 2139 aaronmk
403 2907 aaronmk
    def with_autocommit(self, func):
404 2801 aaronmk
        import psycopg2.extensions
405
406
        prev_isolation_level = self.db.isolation_level
407 2907 aaronmk
        self.db.set_isolation_level(
408
            psycopg2.extensions.ISOLATION_LEVEL_AUTOCOMMIT)
409 2683 aaronmk
        try: return func()
410 2801 aaronmk
        finally: self.db.set_isolation_level(prev_isolation_level)
411 2683 aaronmk
412 2139 aaronmk
    def with_savepoint(self, func):
413 3137 aaronmk
        top = self._savepoint == 0
414 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
415 3137 aaronmk
416 3272 aaronmk
        if self.debug:
417 3273 aaronmk
            self.log_debug('Begin transaction', level=4)
418 3272 aaronmk
            profiler = profiling.ItersProfiler(start_now=True, iter_text='row')
419
420 3160 aaronmk
        # Must happen before running queries so they don't get autocommitted
421
        self._savepoint += 1
422
423 3137 aaronmk
        if top: query = 'START TRANSACTION ISOLATION LEVEL READ COMMITTED'
424
        else: query = 'SAVEPOINT '+savepoint
425
        self.run_query(query, log_level=4)
426
        try:
427
            return func()
428
            if top: self.run_query('COMMIT', log_level=4)
429 2139 aaronmk
        except:
430 3137 aaronmk
            if top: query = 'ROLLBACK'
431
            else: query = 'ROLLBACK TO SAVEPOINT '+savepoint
432
            self.run_query(query, log_level=4)
433
434 2139 aaronmk
            raise
435 2930 aaronmk
        finally:
436
            # Always release savepoint, because after ROLLBACK TO SAVEPOINT,
437
            # "The savepoint remains valid and can be rolled back to again"
438
            # (http://www.postgresql.org/docs/8.3/static/sql-rollback-to.html).
439 3137 aaronmk
            if not top:
440
                self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
441 2930 aaronmk
442
            self._savepoint -= 1
443
            assert self._savepoint >= 0
444
445 3272 aaronmk
            if self.debug:
446
                profiler.stop(self.profile_row_ct)
447 3273 aaronmk
                self.log_debug('End transaction\n'+profiler.msg(), level=4)
448 3272 aaronmk
449 2930 aaronmk
            self.do_autocommit() # OK to do this after ROLLBACK TO SAVEPOINT
450 2191 aaronmk
451
    def do_autocommit(self):
452
        '''Autocommits if outside savepoint'''
453 3135 aaronmk
        assert self._savepoint >= 1
454
        if self.autocommit and self._savepoint == 1:
455 2924 aaronmk
            self.log_debug('Autocommitting', level=4)
456 2191 aaronmk
            self.db.commit()
457 2643 aaronmk
458 3155 aaronmk
    def col_info(self, col, cacheable=True):
459 2643 aaronmk
        table = sql_gen.Table('columns', 'information_schema')
460 3063 aaronmk
        type_ = sql_gen.Coalesce(sql_gen.Nullif(sql_gen.Col('data_type'),
461
            'USER-DEFINED'), sql_gen.Col('udt_name'))
462 3078 aaronmk
        cols = [type_, 'column_default',
463
            sql_gen.Cast('boolean', sql_gen.Col('is_nullable'))]
464 2643 aaronmk
465
        conds = [('table_name', col.table.name), ('column_name', col.name)]
466
        schema = col.table.schema
467
        if schema != None: conds.append(('table_schema', schema))
468
469 2819 aaronmk
        type_, default, nullable = row(select(self, table, cols, conds,
470 3150 aaronmk
            order_by='table_schema', limit=1, cacheable=cacheable, log_level=4))
471 2643 aaronmk
            # TODO: order_by search_path schema order
472 2819 aaronmk
        default = sql_gen.as_Code(default, self)
473
474
        return sql_gen.TypedCol(col.name, type_, default, nullable)
475 2917 aaronmk
476
    def TempFunction(self, name):
477
        if self.debug_temp: schema = None
478
        else: schema = 'pg_temp'
479
        return sql_gen.Function(name, schema)
480 1849 aaronmk
481 1869 aaronmk
connect = DbConn
482
483 832 aaronmk
##### Recoverable querying
484 15 aaronmk
485 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
486 11 aaronmk
487 2791 aaronmk
def run_query(db, query, recover=None, cacheable=False, log_level=2,
488
    log_ignore_excs=None, **kw_args):
489 2794 aaronmk
    '''For params, see DbConn.run_query()'''
490 830 aaronmk
    if recover == None: recover = False
491 2464 aaronmk
    if log_ignore_excs == None: log_ignore_excs = ()
492
    log_ignore_excs = tuple(log_ignore_excs)
493 3236 aaronmk
    debug_msg_ref = [None]
494 830 aaronmk
495 3267 aaronmk
    query = with_explain_comment(db, query)
496 3258 aaronmk
497 2148 aaronmk
    try:
498 2464 aaronmk
        try:
499 2794 aaronmk
            def run(): return db.run_query(query, cacheable, log_level,
500 2793 aaronmk
                debug_msg_ref, **kw_args)
501 2796 aaronmk
            if recover and not db.is_cached(query):
502 2464 aaronmk
                return with_savepoint(db, run)
503
            else: return run() # don't need savepoint if cached
504
        except Exception, e:
505 3095 aaronmk
            msg = strings.ustr(e.args[0])
506 2464 aaronmk
507 3095 aaronmk
            match = re.match(r'^duplicate key value violates unique constraint '
508 3096 aaronmk
                r'"((_?[^\W_]+(?=[._]))?.+?)"', msg)
509 2464 aaronmk
            if match:
510
                constraint, table = match.groups()
511 3025 aaronmk
                cols = []
512
                if recover: # need auto-rollback to run index_cols()
513
                    try: cols = index_cols(db, table, constraint)
514
                    except NotImplementedError: pass
515
                raise DuplicateKeyException(constraint, cols, e)
516 2464 aaronmk
517 3095 aaronmk
            match = re.match(r'^null value in column "(.+?)" violates not-null'
518 2464 aaronmk
                r' constraint', msg)
519
            if match: raise NullValueException('NOT NULL', [match.group(1)], e)
520
521 3095 aaronmk
            match = re.match(r'^(?:invalid input (?:syntax|value)\b.*?'
522 3109 aaronmk
                r'|.+? field value out of range): "(.+?)"', msg)
523 2464 aaronmk
            if match:
524 3109 aaronmk
                value, = match.groups()
525
                raise InvalidValueException(strings.to_unicode(value), e)
526 2464 aaronmk
527 3095 aaronmk
            match = re.match(r'^column "(.+?)" is of type (.+?) but expression '
528 2523 aaronmk
                r'is of type', msg)
529
            if match:
530
                col, type_ = match.groups()
531
                raise MissingCastException(type_, col, e)
532
533 3095 aaronmk
            match = re.match(r'^(\S+) "(.+?)".*? already exists', msg)
534 2945 aaronmk
            if match:
535
                type_, name = match.groups()
536
                raise DuplicateException(type_, name, e)
537 2464 aaronmk
538
            raise # no specific exception raised
539
    except log_ignore_excs:
540
        log_level += 2
541
        raise
542
    finally:
543 3236 aaronmk
        if debug_msg_ref[0] != None: db.log_debug(debug_msg_ref[0], log_level)
544 830 aaronmk
545 832 aaronmk
##### Basic queries
546
547 3256 aaronmk
def is_explainable(query):
548
    # See <http://www.postgresql.org/docs/8.3/static/sql-explain.html#AEN57749>
549 3257 aaronmk
    return re.match(r'^(?:SELECT|INSERT|UPDATE|DELETE|VALUES|EXECUTE|DECLARE)\b'
550
        , query)
551 3256 aaronmk
552 3263 aaronmk
def explain(db, query, **kw_args):
553
    '''
554
    For params, see run_query().
555
    '''
556 3267 aaronmk
    kw_args.setdefault('log_level', 4)
557 3263 aaronmk
558 3256 aaronmk
    return strings.join_lines(values(run_query(db, 'EXPLAIN '+query,
559 3263 aaronmk
        recover=True, cacheable=True, **kw_args)))
560 3256 aaronmk
        # not a higher log_level because it's useful to see what query is being
561
        # run before it's executed, which EXPLAIN effectively provides
562
563 3265 aaronmk
def has_comment(query): return query.endswith('*/')
564
565
def with_explain_comment(db, query, **kw_args):
566 3269 aaronmk
    if db.autoexplain and not has_comment(query) and is_explainable(query):
567 3265 aaronmk
        query += '\n'+sql_gen.esc_comment(' EXPLAIN:\n'
568
            +explain(db, query, **kw_args))
569
    return query
570
571 2153 aaronmk
def next_version(name):
572 2163 aaronmk
    version = 1 # first existing name was version 0
573 2586 aaronmk
    match = re.match(r'^(.*)#(\d+)$', name)
574 2153 aaronmk
    if match:
575 2586 aaronmk
        name, version = match.groups()
576
        version = int(version)+1
577 2932 aaronmk
    return sql_gen.concat(name, '#'+str(version))
578 2153 aaronmk
579 2899 aaronmk
def lock_table(db, table, mode):
580
    table = sql_gen.as_Table(table)
581
    run_query(db, 'LOCK TABLE '+table.to_str(db)+' IN '+mode+' MODE')
582
583 2789 aaronmk
def run_query_into(db, query, into=None, add_indexes_=False, **kw_args):
584 2085 aaronmk
    '''Outputs a query to a temp table.
585
    For params, see run_query().
586
    '''
587 2789 aaronmk
    if into == None: return run_query(db, query, **kw_args)
588 2790 aaronmk
589
    assert isinstance(into, sql_gen.Table)
590
591 2992 aaronmk
    into.is_temp = True
592 3008 aaronmk
    # "temporary tables cannot specify a schema name", so remove schema
593
    into.schema = None
594 2992 aaronmk
595 2790 aaronmk
    kw_args['recover'] = True
596 2945 aaronmk
    kw_args.setdefault('log_ignore_excs', (DuplicateException,))
597 2790 aaronmk
598 2916 aaronmk
    temp = not db.debug_temp # tables are permanent in debug_temp mode
599 2790 aaronmk
600
    # Create table
601
    while True:
602
        create_query = 'CREATE'
603
        if temp: create_query += ' TEMP'
604
        create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
605 2385 aaronmk
606 2790 aaronmk
        try:
607
            cur = run_query(db, create_query, **kw_args)
608
                # CREATE TABLE AS sets rowcount to # rows in query
609
            break
610 2945 aaronmk
        except DuplicateException, e:
611 2790 aaronmk
            into.name = next_version(into.name)
612
            # try again with next version of name
613
614
    if add_indexes_: add_indexes(db, into)
615 3075 aaronmk
616
    # According to the PostgreSQL doc, "The autovacuum daemon cannot access and
617
    # therefore cannot vacuum or analyze temporary tables. [...] if a temporary
618
    # table is going to be used in complex queries, it is wise to run ANALYZE on
619
    # the temporary table after it is populated."
620
    # (http://www.postgresql.org/docs/9.1/static/sql-createtable.html)
621
    # If into is not a temp table, ANALYZE is useful but not required.
622 3073 aaronmk
    analyze(db, into)
623 2790 aaronmk
624
    return cur
625 2085 aaronmk
626 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
627
628 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
629
630 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
631 2293 aaronmk
    start=None, order_by=order_by_pkey, default_table=None):
632 1981 aaronmk
    '''
633 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
634 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
635 1981 aaronmk
    @param fields Use None to select all fields in the table
636 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
637 2379 aaronmk
        * container can be any iterable type
638 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
639
        * compare_right_side: sql_gen.ValueCond|literal value
640 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
641
        use all columns
642 2786 aaronmk
    @return query
643 1981 aaronmk
    '''
644 2315 aaronmk
    # Parse tables param
645 2964 aaronmk
    tables = lists.mk_seq(tables)
646 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
647 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
648 2121 aaronmk
649 2315 aaronmk
    # Parse other params
650 2376 aaronmk
    if conds == None: conds = []
651 2650 aaronmk
    elif dicts.is_dict(conds): conds = conds.items()
652 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
653 3129 aaronmk
    assert limit == None or isinstance(limit, (int, long))
654
    assert start == None or isinstance(start, (int, long))
655 2315 aaronmk
    if order_by is order_by_pkey:
656
        if distinct_on != []: order_by = None
657
        else: order_by = pkey(db, table0, recover=True)
658 865 aaronmk
659 2315 aaronmk
    query = 'SELECT'
660 2056 aaronmk
661 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
662 2056 aaronmk
663 2200 aaronmk
    # DISTINCT ON columns
664 2233 aaronmk
    if distinct_on != []:
665 2467 aaronmk
        query += '\nDISTINCT'
666 2254 aaronmk
        if distinct_on is not distinct_on_all:
667 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
668
669
    # Columns
670 3185 aaronmk
    if query.find('\n') >= 0: whitespace = '\n'
671
    else: whitespace = ' '
672
    if fields == None: query += whitespace+'*'
673 2765 aaronmk
    else:
674
        assert fields != []
675 3185 aaronmk
        if len(fields) > 1: whitespace = '\n'
676
        query += whitespace+('\n, '.join(map(parse_col, fields)))
677 2200 aaronmk
678
    # Main table
679 3185 aaronmk
    if query.find('\n') >= 0 or len(tables) > 0: whitespace = '\n'
680
    else: whitespace = ' '
681
    query += whitespace+'FROM '+table0.to_str(db)
682 865 aaronmk
683 2122 aaronmk
    # Add joins
684 2271 aaronmk
    left_table = table0
685 2263 aaronmk
    for join_ in tables:
686
        table = join_.table
687 2238 aaronmk
688 2343 aaronmk
        # Parse special values
689
        if join_.type_ is sql_gen.filter_out: # filter no match
690 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
691 2853 aaronmk
                sql_gen.CompareCond(None, '~=')))
692 2343 aaronmk
693 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
694 2122 aaronmk
695
        left_table = table
696
697 865 aaronmk
    missing = True
698 2376 aaronmk
    if conds != []:
699 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
700
        else: whitespace = '\n'
701 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
702
            .to_str(db) for l, r in conds], 'WHERE')
703 865 aaronmk
        missing = False
704 2227 aaronmk
    if order_by != None:
705 2467 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
706
    if limit != None: query += '\nLIMIT '+str(limit); missing = False
707 865 aaronmk
    if start != None:
708 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
709 865 aaronmk
        missing = False
710
    if missing: warnings.warn(DbWarning(
711
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
712
713 3266 aaronmk
    query = with_explain_comment(db, query)
714
715 2786 aaronmk
    return query
716 11 aaronmk
717 2054 aaronmk
def select(db, *args, **kw_args):
718
    '''For params, see mk_select() and run_query()'''
719
    recover = kw_args.pop('recover', None)
720
    cacheable = kw_args.pop('cacheable', True)
721 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
722 2054 aaronmk
723 2791 aaronmk
    return run_query(db, mk_select(db, *args, **kw_args), recover, cacheable,
724
        log_level=log_level)
725 2054 aaronmk
726 2788 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, returning=None,
727 3181 aaronmk
    embeddable=False, ignore=False, src=None):
728 1960 aaronmk
    '''
729
    @param returning str|None An inserted column (such as pkey) to return
730 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
731 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
732
        query will be fully cached, not just if it raises an exception.
733 3009 aaronmk
    @param ignore Whether to ignore duplicate keys.
734 3181 aaronmk
    @param src Will be included in the name of any created function, to help
735
        identify the data source in pg_stat_activity.
736 1960 aaronmk
    '''
737 2754 aaronmk
    table = sql_gen.remove_table_rename(sql_gen.as_Table(table))
738 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
739 3010 aaronmk
    if cols != None: cols = [sql_gen.to_name_only_col(c, table) for c in cols]
740 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
741 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
742 2063 aaronmk
743 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
744 2063 aaronmk
745 3009 aaronmk
    def mk_insert(select_query):
746
        query = first_line
747 3014 aaronmk
        if cols != None:
748
            query += '\n('+(', '.join((c.to_str(db) for c in cols)))+')'
749 3009 aaronmk
        query += '\n'+select_query
750
751
        if returning != None:
752
            returning_name_col = sql_gen.to_name_only_col(returning)
753
            query += '\nRETURNING '+returning_name_col.to_str(db)
754
755
        return query
756 2063 aaronmk
757 3017 aaronmk
    return_type = 'unknown'
758
    if returning != None: return_type = returning.to_str(db)+'%TYPE'
759
760 3009 aaronmk
    lang = 'sql'
761
    if ignore:
762 3017 aaronmk
        # Always return something to set the correct rowcount
763
        if returning == None: returning = sql_gen.NamedCol('NULL', None)
764
765 3009 aaronmk
        embeddable = True # must use function
766
        lang = 'plpgsql'
767 3010 aaronmk
768 3092 aaronmk
        if cols == None:
769
            row = [sql_gen.Col(sql_gen.all_cols, 'row')]
770
            row_vars = [sql_gen.Table('row')]
771
        else:
772
            row_vars = row = [sql_gen.Col(c.name, 'row') for c in cols]
773
774 3009 aaronmk
        query = '''\
775 3010 aaronmk
DECLARE
776 3014 aaronmk
    row '''+table.to_str(db)+'''%ROWTYPE;
777 3009 aaronmk
BEGIN
778 3019 aaronmk
    /* Need an EXCEPTION block for each individual row because "When an error is
779
    caught by an EXCEPTION clause, [...] all changes to persistent database
780
    state within the block are rolled back."
781
    This is unfortunate because "A block containing an EXCEPTION clause is
782
    significantly more expensive to enter and exit than a block without one."
783 3015 aaronmk
    (http://www.postgresql.org/docs/8.3/static/plpgsql-control-structures.html\
784
#PLPGSQL-ERROR-TRAPPING)
785
    */
786 3092 aaronmk
    FOR '''+(', '.join((v.to_str(db) for v in row_vars)))+''' IN
787 3034 aaronmk
'''+select_query+'''
788
    LOOP
789 3015 aaronmk
        BEGIN
790 3019 aaronmk
            RETURN QUERY
791 3014 aaronmk
'''+mk_insert(sql_gen.Values(row).to_str(db))+'''
792 3010 aaronmk
;
793 3015 aaronmk
        EXCEPTION
794 3019 aaronmk
            WHEN unique_violation THEN NULL; -- continue to next row
795 3015 aaronmk
        END;
796 3010 aaronmk
    END LOOP;
797
END;\
798 3009 aaronmk
'''
799
    else: query = mk_insert(select_query)
800
801 2070 aaronmk
    if embeddable:
802
        # Create function
803 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
804 3181 aaronmk
        if src != None: function_name = src+': '+function_name
805 2189 aaronmk
        while True:
806
            try:
807 2918 aaronmk
                function = db.TempFunction(function_name)
808 2194 aaronmk
809 2189 aaronmk
                function_query = '''\
810 2698 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''()
811 3017 aaronmk
RETURNS SETOF '''+return_type+'''
812 3009 aaronmk
LANGUAGE '''+lang+'''
813 2467 aaronmk
AS $$
814 3009 aaronmk
'''+query+'''
815 2467 aaronmk
$$;
816 2070 aaronmk
'''
817 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
818 2945 aaronmk
                    log_ignore_excs=(DuplicateException,))
819 2189 aaronmk
                break # this version was successful
820 2945 aaronmk
            except DuplicateException, e:
821 2189 aaronmk
                function_name = next_version(function_name)
822
                # try again with next version of name
823 2070 aaronmk
824 2337 aaronmk
        # Return query that uses function
825 3009 aaronmk
        cols = None
826
        if returning != None: cols = [returning]
827 2698 aaronmk
        func_table = sql_gen.NamedTable('f', sql_gen.FunctionCall(function),
828 3009 aaronmk
            cols) # AS clause requires function alias
829 2787 aaronmk
        return mk_select(db, func_table, start=0, order_by=None)
830 2070 aaronmk
831 2787 aaronmk
    return query
832 2066 aaronmk
833 3074 aaronmk
def insert_select(db, table, *args, **kw_args):
834 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
835 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
836
        values in
837 2072 aaronmk
    '''
838 3141 aaronmk
    returning = kw_args.get('returning', None)
839
    ignore = kw_args.get('ignore', False)
840
841 2386 aaronmk
    into = kw_args.pop('into', None)
842
    if into != None: kw_args['embeddable'] = True
843 2066 aaronmk
    recover = kw_args.pop('recover', None)
844 3141 aaronmk
    if ignore: recover = True
845 2066 aaronmk
    cacheable = kw_args.pop('cacheable', True)
846 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
847 2066 aaronmk
848 3141 aaronmk
    rowcount_only = ignore and returning == None # keep NULL rows on server
849
    if rowcount_only: into = sql_gen.Table('rowcount')
850
851 3074 aaronmk
    cur = run_query_into(db, mk_insert_select(db, table, *args, **kw_args),
852
        into, recover=recover, cacheable=cacheable, log_level=log_level)
853 3141 aaronmk
    if rowcount_only: empty_temp(db, into)
854 3074 aaronmk
    autoanalyze(db, table)
855
    return cur
856 2063 aaronmk
857 2738 aaronmk
default = sql_gen.default # tells insert() to use the default value for a column
858 2066 aaronmk
859 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
860 2085 aaronmk
    '''For params, see insert_select()'''
861 1960 aaronmk
    if lists.is_seq(row): cols = None
862
    else:
863
        cols = row.keys()
864
        row = row.values()
865 2738 aaronmk
    row = list(row) # ensure that "== []" works
866 1960 aaronmk
867 2738 aaronmk
    if row == []: query = None
868
    else: query = sql_gen.Values(row).to_str(db)
869 1961 aaronmk
870 2788 aaronmk
    return insert_select(db, table, cols, query, *args, **kw_args)
871 11 aaronmk
872 3152 aaronmk
def mk_update(db, table, changes=None, cond=None, in_place=False,
873 3153 aaronmk
    cacheable_=True):
874 2402 aaronmk
    '''
875
    @param changes [(col, new_value),...]
876
        * container can be any iterable type
877
        * col: sql_gen.Code|str (for col name)
878
        * new_value: sql_gen.Code|literal value
879
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
880 3056 aaronmk
    @param in_place If set, locks the table and updates rows in place.
881
        This avoids creating dead rows in PostgreSQL.
882
        * cond must be None
883 3153 aaronmk
    @param cacheable_ Whether column structure information used to generate the
884 3152 aaronmk
        query can be cached
885 2402 aaronmk
    @return str query
886
    '''
887 3057 aaronmk
    table = sql_gen.as_Table(table)
888
    changes = [(sql_gen.to_name_only_col(c, table), sql_gen.as_Value(v))
889
        for c, v in changes]
890
891 3056 aaronmk
    if in_place:
892
        assert cond == None
893 3058 aaronmk
894 3065 aaronmk
        query = 'ALTER TABLE '+table.to_str(db)+'\n'
895
        query += ',\n'.join(('ALTER COLUMN '+c.to_str(db)+' TYPE '
896 3153 aaronmk
            +db.col_info(sql_gen.with_default_table(c, table), cacheable_).type
897 3065 aaronmk
            +'\nUSING '+v.to_str(db) for c, v in changes))
898 3058 aaronmk
    else:
899
        query = 'UPDATE '+table.to_str(db)+'\nSET\n'
900
        query += ',\n'.join((c.to_str(db)+' = '+v.to_str(db)
901
            for c, v in changes))
902
        if cond != None: query += '\nWHERE\n'+cond.to_str(db)
903 3056 aaronmk
904 3266 aaronmk
    query = with_explain_comment(db, query)
905
906 2402 aaronmk
    return query
907
908 3074 aaronmk
def update(db, table, *args, **kw_args):
909 2402 aaronmk
    '''For params, see mk_update() and run_query()'''
910
    recover = kw_args.pop('recover', None)
911 3043 aaronmk
    cacheable = kw_args.pop('cacheable', False)
912 3030 aaronmk
    log_level = kw_args.pop('log_level', 2)
913 2402 aaronmk
914 3074 aaronmk
    cur = run_query(db, mk_update(db, table, *args, **kw_args), recover,
915
        cacheable, log_level=log_level)
916
    autoanalyze(db, table)
917
    return cur
918 2402 aaronmk
919 135 aaronmk
def last_insert_id(db):
920 1849 aaronmk
    module = util.root_module(db.db)
921 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
922
    elif module == 'MySQLdb': return db.insert_id()
923
    else: return None
924 13 aaronmk
925 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
926 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
927 2415 aaronmk
    to names that will be distinct among the columns' tables.
928 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
929 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
930
    @param into The table for the new columns.
931 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
932
        columns will be included in the mapping even if they are not in cols.
933
        The tables of the provided Col objects will be changed to into, so make
934
        copies of them if you want to keep the original tables.
935
    @param as_items Whether to return a list of dict items instead of a dict
936 2383 aaronmk
    @return dict(orig_col=new_col, ...)
937
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
938 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
939
        * All mappings use the into table so its name can easily be
940 2383 aaronmk
          changed for all columns at once
941
    '''
942 2415 aaronmk
    cols = lists.uniqify(cols)
943
944 2394 aaronmk
    items = []
945 2389 aaronmk
    for col in preserve:
946 2390 aaronmk
        orig_col = copy.copy(col)
947 2392 aaronmk
        col.table = into
948 2394 aaronmk
        items.append((orig_col, col))
949
    preserve = set(preserve)
950
    for col in cols:
951 2716 aaronmk
        if col not in preserve:
952
            items.append((col, sql_gen.Col(str(col), into, col.srcs)))
953 2394 aaronmk
954
    if not as_items: items = dict(items)
955
    return items
956 2383 aaronmk
957 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
958 2391 aaronmk
    '''For params, see mk_flatten_mapping()
959
    @return See return value of mk_flatten_mapping()
960
    '''
961 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
962
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
963 2786 aaronmk
    run_query_into(db, mk_select(db, joins, cols, limit=limit, start=start),
964 3166 aaronmk
        into=into)
965 2394 aaronmk
    return dict(items)
966 2391 aaronmk
967 3079 aaronmk
##### Database structure introspection
968 2414 aaronmk
969 3079 aaronmk
#### Tables
970
971
def tables(db, schema_like='public', table_like='%', exact=False):
972
    if exact: compare = '='
973
    else: compare = 'LIKE'
974
975
    module = util.root_module(db.db)
976
    if module == 'psycopg2':
977
        conds = [('schemaname', sql_gen.CompareCond(schema_like, compare)),
978
            ('tablename', sql_gen.CompareCond(table_like, compare))]
979
        return values(select(db, 'pg_tables', ['tablename'], conds,
980
            order_by='tablename', log_level=4))
981
    elif module == 'MySQLdb':
982
        return values(run_query(db, 'SHOW TABLES LIKE '+db.esc_value(table_like)
983
            , cacheable=True, log_level=4))
984
    else: raise NotImplementedError("Can't list tables for "+module+' database')
985
986
def table_exists(db, table):
987
    table = sql_gen.as_Table(table)
988
    return list(tables(db, table.schema, table.name, exact=True)) != []
989
990 2426 aaronmk
def table_row_count(db, table, recover=None):
991 2786 aaronmk
    return value(run_query(db, mk_select(db, table, [sql_gen.row_count],
992 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
993 2426 aaronmk
994 2414 aaronmk
def table_cols(db, table, recover=None):
995
    return list(col_names(select(db, table, limit=0, order_by=None,
996 2443 aaronmk
        recover=recover, log_level=4)))
997 2414 aaronmk
998 2291 aaronmk
def pkey(db, table, recover=None):
999 832 aaronmk
    '''Assumed to be first column in table'''
1000 2339 aaronmk
    return table_cols(db, table, recover)[0]
1001 832 aaronmk
1002 2559 aaronmk
not_null_col = 'not_null_col'
1003 2340 aaronmk
1004
def table_not_null_col(db, table, recover=None):
1005
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
1006
    if not_null_col in table_cols(db, table, recover): return not_null_col
1007
    else: return pkey(db, table, recover)
1008
1009 853 aaronmk
def index_cols(db, table, index):
1010
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
1011
    automatically created. When you don't know whether something is a UNIQUE
1012
    constraint or a UNIQUE index, use this function.'''
1013 1909 aaronmk
    module = util.root_module(db.db)
1014
    if module == 'psycopg2':
1015
        return list(values(run_query(db, '''\
1016 853 aaronmk
SELECT attname
1017 866 aaronmk
FROM
1018
(
1019
        SELECT attnum, attname
1020
        FROM pg_index
1021
        JOIN pg_class index ON index.oid = indexrelid
1022
        JOIN pg_class table_ ON table_.oid = indrelid
1023
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
1024
        WHERE
1025 2782 aaronmk
            table_.relname = '''+db.esc_value(table)+'''
1026
            AND index.relname = '''+db.esc_value(index)+'''
1027 866 aaronmk
    UNION
1028
        SELECT attnum, attname
1029
        FROM
1030
        (
1031
            SELECT
1032
                indrelid
1033
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
1034
                    AS indkey
1035
            FROM pg_index
1036
            JOIN pg_class index ON index.oid = indexrelid
1037
            JOIN pg_class table_ ON table_.oid = indrelid
1038
            WHERE
1039 2782 aaronmk
                table_.relname = '''+db.esc_value(table)+'''
1040
                AND index.relname = '''+db.esc_value(index)+'''
1041 866 aaronmk
        ) s
1042
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
1043
) s
1044 853 aaronmk
ORDER BY attnum
1045 2782 aaronmk
'''
1046
            , cacheable=True, log_level=4)))
1047 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
1048
        ' database')
1049 853 aaronmk
1050 464 aaronmk
def constraint_cols(db, table, constraint):
1051 1849 aaronmk
    module = util.root_module(db.db)
1052 464 aaronmk
    if module == 'psycopg2':
1053
        return list(values(run_query(db, '''\
1054
SELECT attname
1055
FROM pg_constraint
1056
JOIN pg_class ON pg_class.oid = conrelid
1057
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
1058
WHERE
1059 2783 aaronmk
    relname = '''+db.esc_value(table)+'''
1060
    AND conname = '''+db.esc_value(constraint)+'''
1061 464 aaronmk
ORDER BY attnum
1062 2783 aaronmk
'''
1063
            )))
1064 464 aaronmk
    else: raise NotImplementedError("Can't list constraint columns for "+module+
1065
        ' database')
1066
1067 3079 aaronmk
#### Functions
1068
1069
def function_exists(db, function):
1070
    function = sql_gen.as_Function(function)
1071
1072
    info_table = sql_gen.Table('routines', 'information_schema')
1073
    conds = [('routine_name', function.name)]
1074
    schema = function.schema
1075
    if schema != None: conds.append(('routine_schema', schema))
1076
    # Exclude trigger functions, since they cannot be called directly
1077
    conds.append(('data_type', sql_gen.CompareCond('trigger', '!=')))
1078
1079
    return list(values(select(db, info_table, ['routine_name'], conds,
1080
        order_by='routine_schema', limit=1, log_level=4))) != []
1081
        # TODO: order_by search_path schema order
1082
1083
##### Structural changes
1084
1085
#### Columns
1086
1087
def add_col(db, table, col, comment=None, **kw_args):
1088
    '''
1089
    @param col TypedCol Name may be versioned, so be sure to propagate any
1090
        renaming back to any source column for the TypedCol.
1091
    @param comment None|str SQL comment used to distinguish columns of the same
1092
        name from each other when they contain different data, to allow the
1093
        ADD COLUMN query to be cached. If not set, query will not be cached.
1094
    '''
1095
    assert isinstance(col, sql_gen.TypedCol)
1096
1097
    while True:
1098
        str_ = 'ALTER TABLE '+table.to_str(db)+' ADD COLUMN '+col.to_str(db)
1099
        if comment != None: str_ += ' '+sql_gen.esc_comment(comment)
1100
1101
        try:
1102
            run_query(db, str_, recover=True, cacheable=True, **kw_args)
1103
            break
1104
        except DuplicateException:
1105
            col.name = next_version(col.name)
1106
            # try again with next version of name
1107
1108
def add_not_null(db, col):
1109
    table = col.table
1110
    col = sql_gen.to_name_only_col(col)
1111
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ALTER COLUMN '
1112
        +col.to_str(db)+' SET NOT NULL', cacheable=True, log_level=3)
1113
1114 2096 aaronmk
row_num_col = '_row_num'
1115
1116 3079 aaronmk
row_num_typed_col = sql_gen.TypedCol(row_num_col, 'serial', nullable=False,
1117
    constraints='PRIMARY KEY')
1118
1119
def add_row_num(db, table):
1120
    '''Adds a row number column to a table. Its name is in row_num_col. It will
1121
    be the primary key.'''
1122
    add_col(db, table, row_num_typed_col, log_level=3)
1123
1124
#### Indexes
1125
1126
def add_pkey(db, table, cols=None, recover=None):
1127
    '''Adds a primary key.
1128
    @param cols [sql_gen.Col,...] The columns in the primary key.
1129
        Defaults to the first column in the table.
1130
    @pre The table must not already have a primary key.
1131
    '''
1132
    table = sql_gen.as_Table(table)
1133
    if cols == None: cols = [pkey(db, table, recover)]
1134
    col_strs = [sql_gen.to_name_only_col(v).to_str(db) for v in cols]
1135
1136
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD PRIMARY KEY ('
1137
        +(', '.join(col_strs))+')', recover=True, cacheable=True, log_level=3,
1138
        log_ignore_excs=(DuplicateException,))
1139
1140 2998 aaronmk
def add_index(db, exprs, table=None, unique=False, ensure_not_null_=True):
1141 2688 aaronmk
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
1142 2538 aaronmk
    Currently, only function calls are supported as expressions.
1143 2998 aaronmk
    @param ensure_not_null_ If set, translates NULL values to sentinel values.
1144 2847 aaronmk
        This allows indexes to be used for comparisons where NULLs are equal.
1145 2538 aaronmk
    '''
1146 2964 aaronmk
    exprs = lists.mk_seq(exprs)
1147 2538 aaronmk
1148 2688 aaronmk
    # Parse exprs
1149
    old_exprs = exprs[:]
1150
    exprs = []
1151
    cols = []
1152
    for i, expr in enumerate(old_exprs):
1153 2823 aaronmk
        expr = sql_gen.as_Col(expr, table)
1154 2688 aaronmk
1155 2823 aaronmk
        # Handle nullable columns
1156 2998 aaronmk
        if ensure_not_null_:
1157 3164 aaronmk
            try: expr = sql_gen.ensure_not_null(db, expr)
1158 2860 aaronmk
            except KeyError: pass # unknown type, so just create plain index
1159 2823 aaronmk
1160 2688 aaronmk
        # Extract col
1161 3002 aaronmk
        expr = copy.deepcopy(expr) # don't modify input!
1162 2688 aaronmk
        if isinstance(expr, sql_gen.FunctionCall):
1163
            col = expr.args[0]
1164
            expr = sql_gen.Expr(expr)
1165
        else: col = expr
1166 2823 aaronmk
        assert isinstance(col, sql_gen.Col)
1167 2688 aaronmk
1168
        # Extract table
1169
        if table == None:
1170
            assert sql_gen.is_table_col(col)
1171
            table = col.table
1172
1173
        col.table = None
1174
1175
        exprs.append(expr)
1176
        cols.append(col)
1177 2408 aaronmk
1178 2688 aaronmk
    table = sql_gen.as_Table(table)
1179
1180 3005 aaronmk
    # Add index
1181 3148 aaronmk
    str_ = 'CREATE'
1182
    if unique: str_ += ' UNIQUE'
1183
    str_ += ' INDEX ON '+table.to_str(db)+' ('+(
1184
        ', '.join((v.to_str(db) for v in exprs)))+')'
1185
    run_query(db, str_, recover=True, cacheable=True, log_level=3)
1186 2408 aaronmk
1187 3083 aaronmk
already_indexed = object() # tells add_indexes() the pkey has already been added
1188
1189
def add_indexes(db, table, has_pkey=True):
1190
    '''Adds an index on all columns in a table.
1191
    @param has_pkey bool|already_indexed Whether a pkey instead of a regular
1192
        index should be added on the first column.
1193
        * If already_indexed, the pkey is assumed to have already been added
1194
    '''
1195
    cols = table_cols(db, table)
1196
    if has_pkey:
1197
        if has_pkey is not already_indexed: add_pkey(db, table)
1198
        cols = cols[1:]
1199
    for col in cols: add_index(db, col, table)
1200
1201 3079 aaronmk
#### Tables
1202 2772 aaronmk
1203 3079 aaronmk
### Maintenance
1204 2772 aaronmk
1205 3079 aaronmk
def analyze(db, table):
1206
    table = sql_gen.as_Table(table)
1207
    run_query(db, 'ANALYZE '+table.to_str(db), log_level=3)
1208 2934 aaronmk
1209 3079 aaronmk
def autoanalyze(db, table):
1210
    if db.autoanalyze: analyze(db, table)
1211 2935 aaronmk
1212 3079 aaronmk
def vacuum(db, table):
1213
    table = sql_gen.as_Table(table)
1214
    db.with_autocommit(lambda: run_query(db, 'VACUUM ANALYZE '+table.to_str(db),
1215
        log_level=3))
1216 2086 aaronmk
1217 3079 aaronmk
### Lifecycle
1218
1219 3247 aaronmk
def drop(db, type_, name):
1220
    name = sql_gen.as_Name(name)
1221
    run_query(db, 'DROP '+type_+' IF EXISTS '+name.to_str(db)+' CASCADE')
1222 2889 aaronmk
1223 3247 aaronmk
def drop_table(db, table): drop(db, 'TABLE', table)
1224
1225 3082 aaronmk
def create_table(db, table, cols=[], has_pkey=True, col_indexes=True,
1226
    like=None):
1227 2675 aaronmk
    '''Creates a table.
1228 2681 aaronmk
    @param cols [sql_gen.TypedCol,...] The column names and types
1229
    @param has_pkey If set, the first column becomes the primary key.
1230 2760 aaronmk
    @param col_indexes bool|[ref]
1231
        * If True, indexes will be added on all non-pkey columns.
1232
        * If a list reference, [0] will be set to a function to do this.
1233
          This can be used to delay index creation until the table is populated.
1234 2675 aaronmk
    '''
1235
    table = sql_gen.as_Table(table)
1236
1237 3082 aaronmk
    if like != None:
1238
        cols = [sql_gen.CustomCode('LIKE '+like.to_str(db)+' INCLUDING ALL')
1239
            ]+cols
1240 2681 aaronmk
    if has_pkey:
1241
        cols[0] = pkey = copy.copy(cols[0]) # don't modify input!
1242 2872 aaronmk
        pkey.constraints = 'PRIMARY KEY'
1243 2681 aaronmk
1244 3085 aaronmk
    temp = table.is_temp and not db.debug_temp
1245
        # temp tables permanent in debug_temp mode
1246 2760 aaronmk
1247 3085 aaronmk
    # Create table
1248
    while True:
1249
        str_ = 'CREATE'
1250
        if temp: str_ += ' TEMP'
1251
        str_ += ' TABLE '+table.to_str(db)+' (\n'
1252
        str_ += '\n, '.join(c.to_str(db) for c in cols)
1253 3126 aaronmk
        str_ += '\n);'
1254 3085 aaronmk
1255
        try:
1256 3127 aaronmk
            run_query(db, str_, recover=True, cacheable=True, log_level=2,
1257 3085 aaronmk
                log_ignore_excs=(DuplicateException,))
1258
            break
1259
        except DuplicateException:
1260
            table.name = next_version(table.name)
1261
            # try again with next version of name
1262
1263 2760 aaronmk
    # Add indexes
1264 2773 aaronmk
    if has_pkey: has_pkey = already_indexed
1265
    def add_indexes_(): add_indexes(db, table, has_pkey)
1266
    if isinstance(col_indexes, list): col_indexes[0] = add_indexes_ # defer
1267
    elif col_indexes: add_indexes_() # add now
1268 2675 aaronmk
1269 3084 aaronmk
def copy_table_struct(db, src, dest):
1270
    '''Creates a structure-only copy of a table. (Does not copy data.)'''
1271 3085 aaronmk
    create_table(db, dest, has_pkey=False, col_indexes=False, like=src)
1272 3084 aaronmk
1273 3079 aaronmk
### Data
1274 2684 aaronmk
1275 2970 aaronmk
def truncate(db, table, schema='public', **kw_args):
1276
    '''For params, see run_query()'''
1277 2777 aaronmk
    table = sql_gen.as_Table(table, schema)
1278 2970 aaronmk
    return run_query(db, 'TRUNCATE '+table.to_str(db)+' CASCADE', **kw_args)
1279 2732 aaronmk
1280 2965 aaronmk
def empty_temp(db, tables):
1281
    tables = lists.mk_seq(tables)
1282 2971 aaronmk
    for table in tables: truncate(db, table, log_level=3)
1283 2965 aaronmk
1284 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
1285
    '''For kw_args, see tables()'''
1286
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
1287 3094 aaronmk
1288
def distinct_table(db, table, distinct_on):
1289
    '''Creates a copy of a temp table which is distinct on the given columns.
1290 3099 aaronmk
    The old and new tables will both get an index on these columns, to
1291
    facilitate merge joins.
1292 3097 aaronmk
    @param distinct_on If empty, creates a table with one row. This is useful if
1293
        your distinct_on columns are all literal values.
1294 3099 aaronmk
    @return The new table.
1295 3094 aaronmk
    '''
1296 3099 aaronmk
    new_table = sql_gen.suffixed_table(table, '_distinct')
1297 3094 aaronmk
1298 3099 aaronmk
    copy_table_struct(db, table, new_table)
1299 3097 aaronmk
1300
    limit = None
1301
    if distinct_on == []: limit = 1 # one sample row
1302 3099 aaronmk
    else:
1303
        add_index(db, distinct_on, new_table, unique=True)
1304
        add_index(db, distinct_on, table) # for join optimization
1305 3097 aaronmk
1306 3099 aaronmk
    insert_select(db, new_table, None, mk_select(db, table, start=0,
1307 3097 aaronmk
        limit=limit), ignore=True)
1308 3099 aaronmk
    analyze(db, new_table)
1309 3094 aaronmk
1310 3099 aaronmk
    return new_table