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 3345 aaronmk
    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 2306 aaronmk
        self.name = name
63 3345 aaronmk
        self.cond = cond
64 468 aaronmk
        self.cols = cols
65 11 aaronmk
66 2523 aaronmk
class MissingCastException(DbException):
67
    def __init__(self, type_, col, cause=None):
68
        DbException.__init__(self, 'Missing cast to type '+strings.as_tt(type_)
69
            +' on column: '+strings.as_tt(col), cause)
70
        self.type = type_
71
        self.col = col
72
73 2143 aaronmk
class NameException(DbException): pass
74
75 2306 aaronmk
class DuplicateKeyException(ConstraintException): pass
76 13 aaronmk
77 2306 aaronmk
class NullValueException(ConstraintException): pass
78 13 aaronmk
79 3346 aaronmk
class CheckException(ConstraintException): pass
80
81 3109 aaronmk
class InvalidValueException(ExceptionWithValue): pass
82 2239 aaronmk
83 2945 aaronmk
class DuplicateException(ExceptionWithNameType): pass
84 2143 aaronmk
85 3419 aaronmk
class DoesNotExistException(ExceptionWithNameType): pass
86
87 89 aaronmk
class EmptyRowException(DbException): pass
88
89 865 aaronmk
##### Warnings
90
91
class DbWarning(UserWarning): pass
92
93 1930 aaronmk
##### Result retrieval
94
95
def col_names(cur): return (col[0] for col in cur.description)
96
97
def rows(cur): return iter(lambda: cur.fetchone(), None)
98
99
def consume_rows(cur):
100
    '''Used to fetch all rows so result will be cached'''
101
    iters.consume_iter(rows(cur))
102
103
def next_row(cur): return rows(cur).next()
104
105
def row(cur):
106
    row_ = next_row(cur)
107
    consume_rows(cur)
108
    return row_
109
110
def next_value(cur): return next_row(cur)[0]
111
112
def value(cur): return row(cur)[0]
113
114
def values(cur): return iters.func_iter(lambda: next_value(cur))
115
116
def value_or_none(cur):
117
    try: return value(cur)
118
    except StopIteration: return None
119
120 2762 aaronmk
##### Escaping
121 2101 aaronmk
122 2573 aaronmk
def esc_name_by_module(module, name):
123
    if module == 'psycopg2' or module == None: quote = '"'
124 2101 aaronmk
    elif module == 'MySQLdb': quote = '`'
125
    else: raise NotImplementedError("Can't escape name for "+module+' database')
126 2500 aaronmk
    return sql_gen.esc_name(name, quote)
127 2101 aaronmk
128
def esc_name_by_engine(engine, name, **kw_args):
129
    return esc_name_by_module(db_engines[engine][0], name, **kw_args)
130
131
def esc_name(db, name, **kw_args):
132
    return esc_name_by_module(util.root_module(db.db), name, **kw_args)
133
134
def qual_name(db, schema, table):
135
    def esc_name_(name): return esc_name(db, name)
136
    table = esc_name_(table)
137
    if schema != None: return esc_name_(schema)+'.'+table
138
    else: return table
139
140 1869 aaronmk
##### Database connections
141 1849 aaronmk
142 2097 aaronmk
db_config_names = ['engine', 'host', 'user', 'password', 'database', 'schemas']
143 1926 aaronmk
144 1869 aaronmk
db_engines = {
145
    'MySQL': ('MySQLdb', {'password': 'passwd', 'database': 'db'}),
146
    'PostgreSQL': ('psycopg2', {}),
147
}
148
149
DatabaseErrors_set = set([DbException])
150
DatabaseErrors = tuple(DatabaseErrors_set)
151
152
def _add_module(module):
153
    DatabaseErrors_set.add(module.DatabaseError)
154
    global DatabaseErrors
155
    DatabaseErrors = tuple(DatabaseErrors_set)
156
157
def db_config_str(db_config):
158
    return db_config['engine']+' database '+db_config['database']
159
160 2448 aaronmk
log_debug_none = lambda msg, level=2: None
161 1901 aaronmk
162 1849 aaronmk
class DbConn:
163 2923 aaronmk
    def __init__(self, db_config, autocommit=True, caching=True,
164 3183 aaronmk
        log_debug=log_debug_none, debug_temp=False, src=None):
165 2915 aaronmk
        '''
166
        @param debug_temp Whether temporary objects should instead be permanent.
167
            This assists in debugging the internal objects used by the program.
168 3183 aaronmk
        @param src In autocommit mode, will be included in a comment in every
169
            query, to help identify the data source in pg_stat_activity.
170 2915 aaronmk
        '''
171 1869 aaronmk
        self.db_config = db_config
172 2190 aaronmk
        self.autocommit = autocommit
173
        self.caching = caching
174 1901 aaronmk
        self.log_debug = log_debug
175 2193 aaronmk
        self.debug = log_debug != log_debug_none
176 2915 aaronmk
        self.debug_temp = debug_temp
177 3183 aaronmk
        self.src = src
178 3074 aaronmk
        self.autoanalyze = False
179 3269 aaronmk
        self.autoexplain = False
180
        self.profile_row_ct = None
181 1869 aaronmk
182 3124 aaronmk
        self._savepoint = 0
183 3120 aaronmk
        self._reset()
184 1869 aaronmk
185
    def __getattr__(self, name):
186
        if name == '__dict__': raise Exception('getting __dict__')
187
        if name == 'db': return self._db()
188
        else: raise AttributeError()
189
190
    def __getstate__(self):
191
        state = copy.copy(self.__dict__) # shallow copy
192 1915 aaronmk
        state['log_debug'] = None # don't pickle the debug callback
193 1869 aaronmk
        state['_DbConn__db'] = None # don't pickle the connection
194
        return state
195
196 3118 aaronmk
    def clear_cache(self): self.query_results = {}
197
198 3120 aaronmk
    def _reset(self):
199 3118 aaronmk
        self.clear_cache()
200 3124 aaronmk
        assert self._savepoint == 0
201 3118 aaronmk
        self._notices_seen = set()
202
        self.__db = None
203
204 2165 aaronmk
    def connected(self): return self.__db != None
205
206 3116 aaronmk
    def close(self):
207 3119 aaronmk
        if not self.connected(): return
208
209 3135 aaronmk
        # Record that the automatic transaction is now closed
210 3136 aaronmk
        self._savepoint -= 1
211 3135 aaronmk
212 3119 aaronmk
        self.db.close()
213 3120 aaronmk
        self._reset()
214 3116 aaronmk
215 3125 aaronmk
    def reconnect(self):
216
        # Do not do this in test mode as it would roll back everything
217
        if self.autocommit: self.close()
218
        # Connection will be reopened automatically on first query
219
220 1869 aaronmk
    def _db(self):
221
        if self.__db == None:
222
            # Process db_config
223
            db_config = self.db_config.copy() # don't modify input!
224 2097 aaronmk
            schemas = db_config.pop('schemas', None)
225 1869 aaronmk
            module_name, mappings = db_engines[db_config.pop('engine')]
226
            module = __import__(module_name)
227
            _add_module(module)
228
            for orig, new in mappings.iteritems():
229
                try: util.rename_key(db_config, orig, new)
230
                except KeyError: pass
231
232
            # Connect
233
            self.__db = module.connect(**db_config)
234
235 3161 aaronmk
            # Record that a transaction is already open
236
            self._savepoint += 1
237
238 1869 aaronmk
            # Configure connection
239 2906 aaronmk
            if hasattr(self.db, 'set_isolation_level'):
240
                import psycopg2.extensions
241
                self.db.set_isolation_level(
242
                    psycopg2.extensions.ISOLATION_LEVEL_READ_COMMITTED)
243 2101 aaronmk
            if schemas != None:
244 2893 aaronmk
                search_path = [self.esc_name(s) for s in schemas.split(',')]
245
                search_path.append(value(run_query(self, 'SHOW search_path',
246
                    log_level=4)))
247
                run_query(self, 'SET search_path TO '+(','.join(search_path)),
248
                    log_level=3)
249 1869 aaronmk
250
        return self.__db
251 1889 aaronmk
252 1891 aaronmk
    class DbCursor(Proxy):
253 1927 aaronmk
        def __init__(self, outer):
254 1891 aaronmk
            Proxy.__init__(self, outer.db.cursor())
255 2191 aaronmk
            self.outer = outer
256 1927 aaronmk
            self.query_results = outer.query_results
257 1894 aaronmk
            self.query_lookup = None
258 1891 aaronmk
            self.result = []
259 1889 aaronmk
260 2802 aaronmk
        def execute(self, query):
261 2764 aaronmk
            self._is_insert = query.startswith('INSERT')
262 2797 aaronmk
            self.query_lookup = query
263 2148 aaronmk
            try:
264 3162 aaronmk
                try: cur = self.inner.execute(query)
265 2802 aaronmk
                finally: self.query = get_cur_query(self.inner, query)
266 1904 aaronmk
            except Exception, e:
267
                self.result = e # cache the exception as the result
268
                self._cache_result()
269
                raise
270 3004 aaronmk
271
            # Always cache certain queries
272 3183 aaronmk
            query = sql_gen.lstrip(query)
273 3004 aaronmk
            if query.startswith('CREATE') or query.startswith('ALTER'):
274 3007 aaronmk
                # structural changes
275 3040 aaronmk
                # Rest of query must be unique in the face of name collisions,
276
                # so don't cache ADD COLUMN unless it has distinguishing comment
277
                if query.find('ADD COLUMN') < 0 or query.endswith('*/'):
278 3007 aaronmk
                    self._cache_result()
279 3004 aaronmk
            elif self.rowcount == 0 and query.startswith('SELECT'): # empty
280 2800 aaronmk
                consume_rows(self) # fetch all rows so result will be cached
281 3004 aaronmk
282 2762 aaronmk
            return cur
283 1894 aaronmk
284 1891 aaronmk
        def fetchone(self):
285
            row = self.inner.fetchone()
286 1899 aaronmk
            if row != None: self.result.append(row)
287
            # otherwise, fetched all rows
288 1904 aaronmk
            else: self._cache_result()
289
            return row
290
291
        def _cache_result(self):
292 2948 aaronmk
            # For inserts that return a result set, don't cache result set since
293
            # inserts are not idempotent. Other non-SELECT queries don't have
294
            # their result set read, so only exceptions will be cached (an
295
            # invalid query will always be invalid).
296 1930 aaronmk
            if self.query_results != None and (not self._is_insert
297 1906 aaronmk
                or isinstance(self.result, Exception)):
298
299 1894 aaronmk
                assert self.query_lookup != None
300 1916 aaronmk
                self.query_results[self.query_lookup] = self.CacheCursor(
301
                    util.dict_subset(dicts.AttrsDictView(self),
302
                    ['query', 'result', 'rowcount', 'description']))
303 1906 aaronmk
304 1916 aaronmk
        class CacheCursor:
305
            def __init__(self, cached_result): self.__dict__ = cached_result
306
307 1927 aaronmk
            def execute(self, *args, **kw_args):
308 1916 aaronmk
                if isinstance(self.result, Exception): raise self.result
309
                # otherwise, result is a rows list
310
                self.iter = iter(self.result)
311
312
            def fetchone(self):
313
                try: return self.iter.next()
314
                except StopIteration: return None
315 1891 aaronmk
316 2212 aaronmk
    def esc_value(self, value):
317 2663 aaronmk
        try: str_ = self.mogrify('%s', [value])
318
        except NotImplementedError, e:
319
            module = util.root_module(self.db)
320
            if module == 'MySQLdb':
321
                import _mysql
322
                str_ = _mysql.escape_string(value)
323
            else: raise e
324 2374 aaronmk
        return strings.to_unicode(str_)
325 2212 aaronmk
326 2347 aaronmk
    def esc_name(self, name): return esc_name(self, name) # calls global func
327
328 2814 aaronmk
    def std_code(self, str_):
329
        '''Standardizes SQL code.
330
        * Ensures that string literals are prefixed by `E`
331
        '''
332
        if str_.startswith("'"): str_ = 'E'+str_
333
        return str_
334
335 2665 aaronmk
    def can_mogrify(self):
336 2663 aaronmk
        module = util.root_module(self.db)
337 2665 aaronmk
        return module == 'psycopg2'
338 2663 aaronmk
339 2665 aaronmk
    def mogrify(self, query, params=None):
340
        if self.can_mogrify(): return self.db.cursor().mogrify(query, params)
341
        else: raise NotImplementedError("Can't mogrify query")
342
343 2671 aaronmk
    def print_notices(self):
344 2725 aaronmk
        if hasattr(self.db, 'notices'):
345
            for msg in self.db.notices:
346
                if msg not in self._notices_seen:
347
                    self._notices_seen.add(msg)
348
                    self.log_debug(msg, level=2)
349 2671 aaronmk
350 2793 aaronmk
    def run_query(self, query, cacheable=False, log_level=2,
351 2464 aaronmk
        debug_msg_ref=None):
352 2445 aaronmk
        '''
353 2464 aaronmk
        @param log_ignore_excs The log_level will be increased by 2 if the query
354
            throws one of these exceptions.
355 2664 aaronmk
        @param debug_msg_ref If specified, the log message will be returned in
356
            this instead of being output. This allows you to filter log messages
357
            depending on the result of the query.
358 2445 aaronmk
        '''
359 2167 aaronmk
        assert query != None
360
361 3183 aaronmk
        if self.autocommit and self.src != None:
362 3206 aaronmk
            query = sql_gen.esc_comment(self.src)+'\t'+query
363 3183 aaronmk
364 2047 aaronmk
        if not self.caching: cacheable = False
365 1903 aaronmk
        used_cache = False
366 2664 aaronmk
367 3242 aaronmk
        if self.debug:
368
            profiler = profiling.ItersProfiler(start_now=True, iter_text='row')
369 1903 aaronmk
        try:
370 1927 aaronmk
            # Get cursor
371
            if cacheable:
372 3238 aaronmk
                try: cur = self.query_results[query]
373 1927 aaronmk
                except KeyError: cur = self.DbCursor(self)
374 3238 aaronmk
                else: used_cache = True
375 1927 aaronmk
            else: cur = self.db.cursor()
376
377
            # Run query
378 3238 aaronmk
            try: cur.execute(query)
379 3162 aaronmk
            except Exception, e:
380
                _add_cursor_info(e, self, query)
381
                raise
382 3238 aaronmk
            else: self.do_autocommit()
383 1903 aaronmk
        finally:
384 3242 aaronmk
            if self.debug:
385 3244 aaronmk
                profiler.stop(self.profile_row_ct)
386 3242 aaronmk
387
                ## Log or return query
388
389 3239 aaronmk
                query = str(get_cur_query(cur, query))
390 3281 aaronmk
                # Put the src comment on a separate line in the log file
391
                query = query.replace('\t', '\n', 1)
392 3239 aaronmk
393 3240 aaronmk
                msg = 'DB query: '
394 3239 aaronmk
395 3240 aaronmk
                if used_cache: msg += 'cache hit'
396
                elif cacheable: msg += 'cache miss'
397
                else: msg += 'non-cacheable'
398 3239 aaronmk
399 3241 aaronmk
                msg += ':\n'+profiler.msg()+'\n'+strings.as_code(query, 'SQL')
400 3240 aaronmk
401 3237 aaronmk
                if debug_msg_ref != None: debug_msg_ref[0] = msg
402
                else: self.log_debug(msg, log_level)
403 3245 aaronmk
404
                self.print_notices()
405 1903 aaronmk
406
        return cur
407 1914 aaronmk
408 2797 aaronmk
    def is_cached(self, query): return query in self.query_results
409 2139 aaronmk
410 2907 aaronmk
    def with_autocommit(self, func):
411 2801 aaronmk
        import psycopg2.extensions
412
413
        prev_isolation_level = self.db.isolation_level
414 2907 aaronmk
        self.db.set_isolation_level(
415
            psycopg2.extensions.ISOLATION_LEVEL_AUTOCOMMIT)
416 2683 aaronmk
        try: return func()
417 2801 aaronmk
        finally: self.db.set_isolation_level(prev_isolation_level)
418 2683 aaronmk
419 2139 aaronmk
    def with_savepoint(self, func):
420 3137 aaronmk
        top = self._savepoint == 0
421 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
422 3137 aaronmk
423 3272 aaronmk
        if self.debug:
424 3273 aaronmk
            self.log_debug('Begin transaction', level=4)
425 3272 aaronmk
            profiler = profiling.ItersProfiler(start_now=True, iter_text='row')
426
427 3160 aaronmk
        # Must happen before running queries so they don't get autocommitted
428
        self._savepoint += 1
429
430 3137 aaronmk
        if top: query = 'START TRANSACTION ISOLATION LEVEL READ COMMITTED'
431
        else: query = 'SAVEPOINT '+savepoint
432
        self.run_query(query, log_level=4)
433
        try:
434
            return func()
435
            if top: self.run_query('COMMIT', log_level=4)
436 2139 aaronmk
        except:
437 3137 aaronmk
            if top: query = 'ROLLBACK'
438
            else: query = 'ROLLBACK TO SAVEPOINT '+savepoint
439
            self.run_query(query, log_level=4)
440
441 2139 aaronmk
            raise
442 2930 aaronmk
        finally:
443
            # Always release savepoint, because after ROLLBACK TO SAVEPOINT,
444
            # "The savepoint remains valid and can be rolled back to again"
445
            # (http://www.postgresql.org/docs/8.3/static/sql-rollback-to.html).
446 3137 aaronmk
            if not top:
447
                self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
448 2930 aaronmk
449
            self._savepoint -= 1
450
            assert self._savepoint >= 0
451
452 3272 aaronmk
            if self.debug:
453
                profiler.stop(self.profile_row_ct)
454 3273 aaronmk
                self.log_debug('End transaction\n'+profiler.msg(), level=4)
455 3272 aaronmk
456 2930 aaronmk
            self.do_autocommit() # OK to do this after ROLLBACK TO SAVEPOINT
457 2191 aaronmk
458
    def do_autocommit(self):
459
        '''Autocommits if outside savepoint'''
460 3135 aaronmk
        assert self._savepoint >= 1
461
        if self.autocommit and self._savepoint == 1:
462 2924 aaronmk
            self.log_debug('Autocommitting', level=4)
463 2191 aaronmk
            self.db.commit()
464 2643 aaronmk
465 3155 aaronmk
    def col_info(self, col, cacheable=True):
466 2643 aaronmk
        table = sql_gen.Table('columns', 'information_schema')
467 3063 aaronmk
        type_ = sql_gen.Coalesce(sql_gen.Nullif(sql_gen.Col('data_type'),
468
            'USER-DEFINED'), sql_gen.Col('udt_name'))
469 3078 aaronmk
        cols = [type_, 'column_default',
470
            sql_gen.Cast('boolean', sql_gen.Col('is_nullable'))]
471 2643 aaronmk
472
        conds = [('table_name', col.table.name), ('column_name', col.name)]
473
        schema = col.table.schema
474
        if schema != None: conds.append(('table_schema', schema))
475
476 2819 aaronmk
        type_, default, nullable = row(select(self, table, cols, conds,
477 3150 aaronmk
            order_by='table_schema', limit=1, cacheable=cacheable, log_level=4))
478 2643 aaronmk
            # TODO: order_by search_path schema order
479 2819 aaronmk
        default = sql_gen.as_Code(default, self)
480
481
        return sql_gen.TypedCol(col.name, type_, default, nullable)
482 2917 aaronmk
483
    def TempFunction(self, name):
484
        if self.debug_temp: schema = None
485
        else: schema = 'pg_temp'
486
        return sql_gen.Function(name, schema)
487 1849 aaronmk
488 1869 aaronmk
connect = DbConn
489
490 832 aaronmk
##### Recoverable querying
491 15 aaronmk
492 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
493 11 aaronmk
494 2791 aaronmk
def run_query(db, query, recover=None, cacheable=False, log_level=2,
495
    log_ignore_excs=None, **kw_args):
496 2794 aaronmk
    '''For params, see DbConn.run_query()'''
497 830 aaronmk
    if recover == None: recover = False
498 2464 aaronmk
    if log_ignore_excs == None: log_ignore_excs = ()
499
    log_ignore_excs = tuple(log_ignore_excs)
500 3236 aaronmk
    debug_msg_ref = [None]
501 830 aaronmk
502 3267 aaronmk
    query = with_explain_comment(db, query)
503 3258 aaronmk
504 2148 aaronmk
    try:
505 2464 aaronmk
        try:
506 2794 aaronmk
            def run(): return db.run_query(query, cacheable, log_level,
507 2793 aaronmk
                debug_msg_ref, **kw_args)
508 2796 aaronmk
            if recover and not db.is_cached(query):
509 2464 aaronmk
                return with_savepoint(db, run)
510
            else: return run() # don't need savepoint if cached
511
        except Exception, e:
512 3095 aaronmk
            msg = strings.ustr(e.args[0])
513 3414 aaronmk
            msg = re.sub(r'^PL/Python: \w+: ', r'', msg)
514 2464 aaronmk
515 3095 aaronmk
            match = re.match(r'^duplicate key value violates unique constraint '
516 3338 aaronmk
                r'"(.+?)"', msg)
517 2464 aaronmk
            if match:
518 3338 aaronmk
                constraint, = match.groups()
519 3025 aaronmk
                cols = []
520
                if recover: # need auto-rollback to run index_cols()
521 3319 aaronmk
                    try: cols = index_cols(db, constraint)
522 3025 aaronmk
                    except NotImplementedError: pass
523 3345 aaronmk
                raise DuplicateKeyException(constraint, None, cols, e)
524 2464 aaronmk
525 3095 aaronmk
            match = re.match(r'^null value in column "(.+?)" violates not-null'
526 2464 aaronmk
                r' constraint', msg)
527 3345 aaronmk
            if match:
528
                col, = match.groups()
529
                raise NullValueException('NOT NULL', None, [col], e)
530 2464 aaronmk
531 3346 aaronmk
            match = re.match(r'^new row for relation "(.+?)" violates check '
532
                r'constraint "(.+?)"', msg)
533
            if match:
534
                table, constraint = match.groups()
535 3347 aaronmk
                constraint = sql_gen.Col(constraint, table)
536 3349 aaronmk
                cond = None
537
                if recover: # need auto-rollback to run constraint_cond()
538
                    try: cond = constraint_cond(db, constraint)
539
                    except NotImplementedError: pass
540
                raise CheckException(constraint.to_str(db), cond, [], e)
541 3346 aaronmk
542 3095 aaronmk
            match = re.match(r'^(?:invalid input (?:syntax|value)\b.*?'
543 3109 aaronmk
                r'|.+? field value out of range): "(.+?)"', msg)
544 2464 aaronmk
            if match:
545 3109 aaronmk
                value, = match.groups()
546
                raise InvalidValueException(strings.to_unicode(value), e)
547 2464 aaronmk
548 3095 aaronmk
            match = re.match(r'^column "(.+?)" is of type (.+?) but expression '
549 2523 aaronmk
                r'is of type', msg)
550
            if match:
551
                col, type_ = match.groups()
552
                raise MissingCastException(type_, col, e)
553
554 3419 aaronmk
            typed_name_re = r'^(\S+) "(.+?)"'
555
556
            match = re.match(typed_name_re+r'.*? already exists', msg)
557 2945 aaronmk
            if match:
558
                type_, name = match.groups()
559
                raise DuplicateException(type_, name, e)
560 2464 aaronmk
561 3419 aaronmk
            match = re.match(typed_name_re+r' does not exist', msg)
562
            if match:
563
                type_, name = match.groups()
564
                raise DoesNotExistException(type_, name, e)
565
566 2464 aaronmk
            raise # no specific exception raised
567
    except log_ignore_excs:
568
        log_level += 2
569
        raise
570
    finally:
571 3236 aaronmk
        if debug_msg_ref[0] != None: db.log_debug(debug_msg_ref[0], log_level)
572 830 aaronmk
573 832 aaronmk
##### Basic queries
574
575 3256 aaronmk
def is_explainable(query):
576
    # See <http://www.postgresql.org/docs/8.3/static/sql-explain.html#AEN57749>
577 3257 aaronmk
    return re.match(r'^(?:SELECT|INSERT|UPDATE|DELETE|VALUES|EXECUTE|DECLARE)\b'
578
        , query)
579 3256 aaronmk
580 3263 aaronmk
def explain(db, query, **kw_args):
581
    '''
582
    For params, see run_query().
583
    '''
584 3267 aaronmk
    kw_args.setdefault('log_level', 4)
585 3263 aaronmk
586 3256 aaronmk
    return strings.join_lines(values(run_query(db, 'EXPLAIN '+query,
587 3263 aaronmk
        recover=True, cacheable=True, **kw_args)))
588 3256 aaronmk
        # not a higher log_level because it's useful to see what query is being
589
        # run before it's executed, which EXPLAIN effectively provides
590
591 3265 aaronmk
def has_comment(query): return query.endswith('*/')
592
593
def with_explain_comment(db, query, **kw_args):
594 3269 aaronmk
    if db.autoexplain and not has_comment(query) and is_explainable(query):
595 3265 aaronmk
        query += '\n'+sql_gen.esc_comment(' EXPLAIN:\n'
596
            +explain(db, query, **kw_args))
597
    return query
598
599 2153 aaronmk
def next_version(name):
600 2163 aaronmk
    version = 1 # first existing name was version 0
601 2586 aaronmk
    match = re.match(r'^(.*)#(\d+)$', name)
602 2153 aaronmk
    if match:
603 2586 aaronmk
        name, version = match.groups()
604
        version = int(version)+1
605 2932 aaronmk
    return sql_gen.concat(name, '#'+str(version))
606 2153 aaronmk
607 2899 aaronmk
def lock_table(db, table, mode):
608
    table = sql_gen.as_Table(table)
609
    run_query(db, 'LOCK TABLE '+table.to_str(db)+' IN '+mode+' MODE')
610
611 3303 aaronmk
def run_query_into(db, query, into=None, add_pkey_=False, **kw_args):
612 2085 aaronmk
    '''Outputs a query to a temp table.
613
    For params, see run_query().
614
    '''
615 2789 aaronmk
    if into == None: return run_query(db, query, **kw_args)
616 2790 aaronmk
617
    assert isinstance(into, sql_gen.Table)
618
619 2992 aaronmk
    into.is_temp = True
620 3008 aaronmk
    # "temporary tables cannot specify a schema name", so remove schema
621
    into.schema = None
622 2992 aaronmk
623 2790 aaronmk
    kw_args['recover'] = True
624 2945 aaronmk
    kw_args.setdefault('log_ignore_excs', (DuplicateException,))
625 2790 aaronmk
626 2916 aaronmk
    temp = not db.debug_temp # tables are permanent in debug_temp mode
627 2790 aaronmk
628
    # Create table
629
    while True:
630
        create_query = 'CREATE'
631
        if temp: create_query += ' TEMP'
632
        create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
633 2385 aaronmk
634 2790 aaronmk
        try:
635
            cur = run_query(db, create_query, **kw_args)
636
                # CREATE TABLE AS sets rowcount to # rows in query
637
            break
638 2945 aaronmk
        except DuplicateException, e:
639 2790 aaronmk
            into.name = next_version(into.name)
640
            # try again with next version of name
641
642 3303 aaronmk
    if add_pkey_: add_pkey(db, into)
643 3075 aaronmk
644
    # According to the PostgreSQL doc, "The autovacuum daemon cannot access and
645
    # therefore cannot vacuum or analyze temporary tables. [...] if a temporary
646
    # table is going to be used in complex queries, it is wise to run ANALYZE on
647
    # the temporary table after it is populated."
648
    # (http://www.postgresql.org/docs/9.1/static/sql-createtable.html)
649
    # If into is not a temp table, ANALYZE is useful but not required.
650 3073 aaronmk
    analyze(db, into)
651 2790 aaronmk
652
    return cur
653 2085 aaronmk
654 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
655
656 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
657
658 3420 aaronmk
def mk_select(db, tables=None, fields=None, conds=None, distinct_on=[],
659
    limit=None, start=None, order_by=order_by_pkey, default_table=None):
660 1981 aaronmk
    '''
661 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
662 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
663 1981 aaronmk
    @param fields Use None to select all fields in the table
664 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
665 2379 aaronmk
        * container can be any iterable type
666 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
667
        * compare_right_side: sql_gen.ValueCond|literal value
668 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
669
        use all columns
670 2786 aaronmk
    @return query
671 1981 aaronmk
    '''
672 2315 aaronmk
    # Parse tables param
673 2964 aaronmk
    tables = lists.mk_seq(tables)
674 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
675 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
676 2121 aaronmk
677 2315 aaronmk
    # Parse other params
678 2376 aaronmk
    if conds == None: conds = []
679 2650 aaronmk
    elif dicts.is_dict(conds): conds = conds.items()
680 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
681 3129 aaronmk
    assert limit == None or isinstance(limit, (int, long))
682
    assert start == None or isinstance(start, (int, long))
683 2315 aaronmk
    if order_by is order_by_pkey:
684 3421 aaronmk
        if table0 == None or distinct_on != []: order_by = None
685 2315 aaronmk
        else: order_by = pkey(db, table0, recover=True)
686 865 aaronmk
687 2315 aaronmk
    query = 'SELECT'
688 2056 aaronmk
689 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
690 2056 aaronmk
691 2200 aaronmk
    # DISTINCT ON columns
692 2233 aaronmk
    if distinct_on != []:
693 2467 aaronmk
        query += '\nDISTINCT'
694 2254 aaronmk
        if distinct_on is not distinct_on_all:
695 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
696
697
    # Columns
698 3185 aaronmk
    if query.find('\n') >= 0: whitespace = '\n'
699
    else: whitespace = ' '
700
    if fields == None: query += whitespace+'*'
701 2765 aaronmk
    else:
702
        assert fields != []
703 3185 aaronmk
        if len(fields) > 1: whitespace = '\n'
704
        query += whitespace+('\n, '.join(map(parse_col, fields)))
705 2200 aaronmk
706
    # Main table
707 3185 aaronmk
    if query.find('\n') >= 0 or len(tables) > 0: whitespace = '\n'
708
    else: whitespace = ' '
709 3420 aaronmk
    if table0 != None: query += whitespace+'FROM '+table0.to_str(db)
710 865 aaronmk
711 2122 aaronmk
    # Add joins
712 2271 aaronmk
    left_table = table0
713 2263 aaronmk
    for join_ in tables:
714
        table = join_.table
715 2238 aaronmk
716 2343 aaronmk
        # Parse special values
717
        if join_.type_ is sql_gen.filter_out: # filter no match
718 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
719 2853 aaronmk
                sql_gen.CompareCond(None, '~=')))
720 2343 aaronmk
721 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
722 2122 aaronmk
723
        left_table = table
724
725 865 aaronmk
    missing = True
726 2376 aaronmk
    if conds != []:
727 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
728
        else: whitespace = '\n'
729 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
730
            .to_str(db) for l, r in conds], 'WHERE')
731 2227 aaronmk
    if order_by != None:
732 2467 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
733 3297 aaronmk
    if limit != None: query += '\nLIMIT '+str(limit)
734 865 aaronmk
    if start != None:
735 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
736 865 aaronmk
737 3266 aaronmk
    query = with_explain_comment(db, query)
738
739 2786 aaronmk
    return query
740 11 aaronmk
741 2054 aaronmk
def select(db, *args, **kw_args):
742
    '''For params, see mk_select() and run_query()'''
743
    recover = kw_args.pop('recover', None)
744
    cacheable = kw_args.pop('cacheable', True)
745 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
746 2054 aaronmk
747 2791 aaronmk
    return run_query(db, mk_select(db, *args, **kw_args), recover, cacheable,
748
        log_level=log_level)
749 2054 aaronmk
750 2788 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, returning=None,
751 3181 aaronmk
    embeddable=False, ignore=False, src=None):
752 1960 aaronmk
    '''
753
    @param returning str|None An inserted column (such as pkey) to return
754 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
755 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
756
        query will be fully cached, not just if it raises an exception.
757 3009 aaronmk
    @param ignore Whether to ignore duplicate keys.
758 3181 aaronmk
    @param src Will be included in the name of any created function, to help
759
        identify the data source in pg_stat_activity.
760 1960 aaronmk
    '''
761 2754 aaronmk
    table = sql_gen.remove_table_rename(sql_gen.as_Table(table))
762 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
763 3010 aaronmk
    if cols != None: cols = [sql_gen.to_name_only_col(c, table) for c in cols]
764 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
765 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
766 2063 aaronmk
767 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
768 2063 aaronmk
769 3009 aaronmk
    def mk_insert(select_query):
770
        query = first_line
771 3014 aaronmk
        if cols != None:
772
            query += '\n('+(', '.join((c.to_str(db) for c in cols)))+')'
773 3009 aaronmk
        query += '\n'+select_query
774
775
        if returning != None:
776
            returning_name_col = sql_gen.to_name_only_col(returning)
777
            query += '\nRETURNING '+returning_name_col.to_str(db)
778
779
        return query
780 2063 aaronmk
781 3017 aaronmk
    return_type = 'unknown'
782
    if returning != None: return_type = returning.to_str(db)+'%TYPE'
783
784 3009 aaronmk
    lang = 'sql'
785
    if ignore:
786 3017 aaronmk
        # Always return something to set the correct rowcount
787
        if returning == None: returning = sql_gen.NamedCol('NULL', None)
788
789 3009 aaronmk
        embeddable = True # must use function
790
        lang = 'plpgsql'
791 3010 aaronmk
792 3092 aaronmk
        if cols == None:
793
            row = [sql_gen.Col(sql_gen.all_cols, 'row')]
794
            row_vars = [sql_gen.Table('row')]
795
        else:
796
            row_vars = row = [sql_gen.Col(c.name, 'row') for c in cols]
797
798 3009 aaronmk
        query = '''\
799 3010 aaronmk
DECLARE
800 3014 aaronmk
    row '''+table.to_str(db)+'''%ROWTYPE;
801 3009 aaronmk
BEGIN
802 3019 aaronmk
    /* Need an EXCEPTION block for each individual row because "When an error is
803
    caught by an EXCEPTION clause, [...] all changes to persistent database
804
    state within the block are rolled back."
805
    This is unfortunate because "A block containing an EXCEPTION clause is
806
    significantly more expensive to enter and exit than a block without one."
807 3015 aaronmk
    (http://www.postgresql.org/docs/8.3/static/plpgsql-control-structures.html\
808
#PLPGSQL-ERROR-TRAPPING)
809
    */
810 3092 aaronmk
    FOR '''+(', '.join((v.to_str(db) for v in row_vars)))+''' IN
811 3034 aaronmk
'''+select_query+'''
812
    LOOP
813 3015 aaronmk
        BEGIN
814 3019 aaronmk
            RETURN QUERY
815 3014 aaronmk
'''+mk_insert(sql_gen.Values(row).to_str(db))+'''
816 3010 aaronmk
;
817 3015 aaronmk
        EXCEPTION
818 3019 aaronmk
            WHEN unique_violation THEN NULL; -- continue to next row
819 3015 aaronmk
        END;
820 3010 aaronmk
    END LOOP;
821
END;\
822 3009 aaronmk
'''
823
    else: query = mk_insert(select_query)
824
825 2070 aaronmk
    if embeddable:
826
        # Create function
827 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
828 3181 aaronmk
        if src != None: function_name = src+': '+function_name
829 2189 aaronmk
        while True:
830
            try:
831 2918 aaronmk
                function = db.TempFunction(function_name)
832 2194 aaronmk
833 2189 aaronmk
                function_query = '''\
834 2698 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''()
835 3017 aaronmk
RETURNS SETOF '''+return_type+'''
836 3009 aaronmk
LANGUAGE '''+lang+'''
837 2467 aaronmk
AS $$
838 3009 aaronmk
'''+query+'''
839 2467 aaronmk
$$;
840 2070 aaronmk
'''
841 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
842 2945 aaronmk
                    log_ignore_excs=(DuplicateException,))
843 2189 aaronmk
                break # this version was successful
844 2945 aaronmk
            except DuplicateException, e:
845 2189 aaronmk
                function_name = next_version(function_name)
846
                # try again with next version of name
847 2070 aaronmk
848 2337 aaronmk
        # Return query that uses function
849 3009 aaronmk
        cols = None
850
        if returning != None: cols = [returning]
851 2698 aaronmk
        func_table = sql_gen.NamedTable('f', sql_gen.FunctionCall(function),
852 3009 aaronmk
            cols) # AS clause requires function alias
853 3298 aaronmk
        return mk_select(db, func_table, order_by=None)
854 2070 aaronmk
855 2787 aaronmk
    return query
856 2066 aaronmk
857 3074 aaronmk
def insert_select(db, table, *args, **kw_args):
858 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
859 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
860
        values in
861 2072 aaronmk
    '''
862 3141 aaronmk
    returning = kw_args.get('returning', None)
863
    ignore = kw_args.get('ignore', False)
864
865 2386 aaronmk
    into = kw_args.pop('into', None)
866
    if into != None: kw_args['embeddable'] = True
867 2066 aaronmk
    recover = kw_args.pop('recover', None)
868 3141 aaronmk
    if ignore: recover = True
869 2066 aaronmk
    cacheable = kw_args.pop('cacheable', True)
870 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
871 2066 aaronmk
872 3141 aaronmk
    rowcount_only = ignore and returning == None # keep NULL rows on server
873
    if rowcount_only: into = sql_gen.Table('rowcount')
874
875 3074 aaronmk
    cur = run_query_into(db, mk_insert_select(db, table, *args, **kw_args),
876
        into, recover=recover, cacheable=cacheable, log_level=log_level)
877 3141 aaronmk
    if rowcount_only: empty_temp(db, into)
878 3074 aaronmk
    autoanalyze(db, table)
879
    return cur
880 2063 aaronmk
881 2738 aaronmk
default = sql_gen.default # tells insert() to use the default value for a column
882 2066 aaronmk
883 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
884 2085 aaronmk
    '''For params, see insert_select()'''
885 1960 aaronmk
    if lists.is_seq(row): cols = None
886
    else:
887
        cols = row.keys()
888
        row = row.values()
889 2738 aaronmk
    row = list(row) # ensure that "== []" works
890 1960 aaronmk
891 2738 aaronmk
    if row == []: query = None
892
    else: query = sql_gen.Values(row).to_str(db)
893 1961 aaronmk
894 2788 aaronmk
    return insert_select(db, table, cols, query, *args, **kw_args)
895 11 aaronmk
896 3152 aaronmk
def mk_update(db, table, changes=None, cond=None, in_place=False,
897 3153 aaronmk
    cacheable_=True):
898 2402 aaronmk
    '''
899
    @param changes [(col, new_value),...]
900
        * container can be any iterable type
901
        * col: sql_gen.Code|str (for col name)
902
        * new_value: sql_gen.Code|literal value
903
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
904 3056 aaronmk
    @param in_place If set, locks the table and updates rows in place.
905
        This avoids creating dead rows in PostgreSQL.
906
        * cond must be None
907 3153 aaronmk
    @param cacheable_ Whether column structure information used to generate the
908 3152 aaronmk
        query can be cached
909 2402 aaronmk
    @return str query
910
    '''
911 3057 aaronmk
    table = sql_gen.as_Table(table)
912
    changes = [(sql_gen.to_name_only_col(c, table), sql_gen.as_Value(v))
913
        for c, v in changes]
914
915 3056 aaronmk
    if in_place:
916
        assert cond == None
917 3058 aaronmk
918 3065 aaronmk
        query = 'ALTER TABLE '+table.to_str(db)+'\n'
919
        query += ',\n'.join(('ALTER COLUMN '+c.to_str(db)+' TYPE '
920 3153 aaronmk
            +db.col_info(sql_gen.with_default_table(c, table), cacheable_).type
921 3065 aaronmk
            +'\nUSING '+v.to_str(db) for c, v in changes))
922 3058 aaronmk
    else:
923
        query = 'UPDATE '+table.to_str(db)+'\nSET\n'
924
        query += ',\n'.join((c.to_str(db)+' = '+v.to_str(db)
925
            for c, v in changes))
926
        if cond != None: query += '\nWHERE\n'+cond.to_str(db)
927 3056 aaronmk
928 3266 aaronmk
    query = with_explain_comment(db, query)
929
930 2402 aaronmk
    return query
931
932 3074 aaronmk
def update(db, table, *args, **kw_args):
933 2402 aaronmk
    '''For params, see mk_update() and run_query()'''
934
    recover = kw_args.pop('recover', None)
935 3043 aaronmk
    cacheable = kw_args.pop('cacheable', False)
936 3030 aaronmk
    log_level = kw_args.pop('log_level', 2)
937 2402 aaronmk
938 3074 aaronmk
    cur = run_query(db, mk_update(db, table, *args, **kw_args), recover,
939
        cacheable, log_level=log_level)
940
    autoanalyze(db, table)
941
    return cur
942 2402 aaronmk
943 3286 aaronmk
def mk_delete(db, table, cond=None):
944
    '''
945
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
946
    @return str query
947
    '''
948
    query = 'DELETE FROM '+table.to_str(db)
949
    if cond != None: query += '\nWHERE '+cond.to_str(db)
950
951
    query = with_explain_comment(db, query)
952
953
    return query
954
955
def delete(db, table, *args, **kw_args):
956
    '''For params, see mk_delete() and run_query()'''
957
    recover = kw_args.pop('recover', None)
958 3295 aaronmk
    cacheable = kw_args.pop('cacheable', True)
959 3286 aaronmk
    log_level = kw_args.pop('log_level', 2)
960
961
    cur = run_query(db, mk_delete(db, table, *args, **kw_args), recover,
962
        cacheable, log_level=log_level)
963
    autoanalyze(db, table)
964
    return cur
965
966 135 aaronmk
def last_insert_id(db):
967 1849 aaronmk
    module = util.root_module(db.db)
968 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
969
    elif module == 'MySQLdb': return db.insert_id()
970
    else: return None
971 13 aaronmk
972 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
973 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
974 2415 aaronmk
    to names that will be distinct among the columns' tables.
975 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
976 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
977
    @param into The table for the new columns.
978 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
979
        columns will be included in the mapping even if they are not in cols.
980
        The tables of the provided Col objects will be changed to into, so make
981
        copies of them if you want to keep the original tables.
982
    @param as_items Whether to return a list of dict items instead of a dict
983 2383 aaronmk
    @return dict(orig_col=new_col, ...)
984
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
985 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
986
        * All mappings use the into table so its name can easily be
987 2383 aaronmk
          changed for all columns at once
988
    '''
989 2415 aaronmk
    cols = lists.uniqify(cols)
990
991 2394 aaronmk
    items = []
992 2389 aaronmk
    for col in preserve:
993 2390 aaronmk
        orig_col = copy.copy(col)
994 2392 aaronmk
        col.table = into
995 2394 aaronmk
        items.append((orig_col, col))
996
    preserve = set(preserve)
997
    for col in cols:
998 2716 aaronmk
        if col not in preserve:
999
            items.append((col, sql_gen.Col(str(col), into, col.srcs)))
1000 2394 aaronmk
1001
    if not as_items: items = dict(items)
1002
    return items
1003 2383 aaronmk
1004 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
1005 2391 aaronmk
    '''For params, see mk_flatten_mapping()
1006
    @return See return value of mk_flatten_mapping()
1007
    '''
1008 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
1009
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
1010 3296 aaronmk
    run_query_into(db, mk_select(db, joins, cols, order_by=None, limit=limit,
1011 3305 aaronmk
        start=start), into=into, add_pkey_=True)
1012 2394 aaronmk
    return dict(items)
1013 2391 aaronmk
1014 3079 aaronmk
##### Database structure introspection
1015 2414 aaronmk
1016 3321 aaronmk
#### Expressions
1017
1018 3353 aaronmk
bool_re = r'(?:true|false)'
1019
1020
def simplify_expr(expr):
1021
    expr = expr.replace('(NULL IS NULL)', 'true')
1022
    expr = expr.replace('(NULL IS NOT NULL)', 'false')
1023
    expr = re.sub(r' OR '+bool_re, r'', expr)
1024
    expr = re.sub(bool_re+r' OR ', r'', expr)
1025
    while True:
1026
        expr, n = re.subn(r'\((\([^()]*\))\)', r'\1', expr)
1027
        if n == 0: break
1028
    return expr
1029
1030 3321 aaronmk
name_re = r'(?:\w+|(?:"[^"]*")+)'
1031
1032
def parse_expr_col(str_):
1033
    match = re.match(r'^\('+name_re+r'\(('+name_re+r').*\)\)$', str_)
1034
    if match: str_ = match.group(1)
1035
    return sql_gen.unesc_name(str_)
1036
1037 3351 aaronmk
def map_expr(db, expr, mapping, in_cols_found=None):
1038
    '''Replaces output columns with input columns in an expression.
1039
    @param in_cols_found If set, will be filled in with the expr's (input) cols
1040
    '''
1041
    for out, in_ in mapping.iteritems():
1042
        orig_expr = expr
1043
        out = sql_gen.to_name_only_col(out)
1044
        in_str = sql_gen.to_name_only_col(sql_gen.remove_col_rename(in_)
1045
            ).to_str(db)
1046
1047
        # Replace out both with and without quotes
1048
        expr = expr.replace(out.to_str(db), in_str)
1049
        expr = re.sub(r'\b'+out.name+r'\b', in_str, expr)
1050
1051
        if in_cols_found != None and expr != orig_expr: # replaced something
1052
            in_cols_found.append(in_)
1053 3353 aaronmk
1054
    return simplify_expr(expr)
1055 3351 aaronmk
1056 3079 aaronmk
#### Tables
1057
1058
def tables(db, schema_like='public', table_like='%', exact=False):
1059
    if exact: compare = '='
1060
    else: compare = 'LIKE'
1061
1062
    module = util.root_module(db.db)
1063
    if module == 'psycopg2':
1064
        conds = [('schemaname', sql_gen.CompareCond(schema_like, compare)),
1065
            ('tablename', sql_gen.CompareCond(table_like, compare))]
1066
        return values(select(db, 'pg_tables', ['tablename'], conds,
1067
            order_by='tablename', log_level=4))
1068
    elif module == 'MySQLdb':
1069
        return values(run_query(db, 'SHOW TABLES LIKE '+db.esc_value(table_like)
1070
            , cacheable=True, log_level=4))
1071
    else: raise NotImplementedError("Can't list tables for "+module+' database')
1072
1073
def table_exists(db, table):
1074
    table = sql_gen.as_Table(table)
1075
    return list(tables(db, table.schema, table.name, exact=True)) != []
1076
1077 2426 aaronmk
def table_row_count(db, table, recover=None):
1078 2786 aaronmk
    return value(run_query(db, mk_select(db, table, [sql_gen.row_count],
1079 3298 aaronmk
        order_by=None), recover=recover, log_level=3))
1080 2426 aaronmk
1081 2414 aaronmk
def table_cols(db, table, recover=None):
1082
    return list(col_names(select(db, table, limit=0, order_by=None,
1083 2443 aaronmk
        recover=recover, log_level=4)))
1084 2414 aaronmk
1085 2291 aaronmk
def pkey(db, table, recover=None):
1086 832 aaronmk
    '''Assumed to be first column in table'''
1087 2339 aaronmk
    return table_cols(db, table, recover)[0]
1088 832 aaronmk
1089 2559 aaronmk
not_null_col = 'not_null_col'
1090 2340 aaronmk
1091
def table_not_null_col(db, table, recover=None):
1092
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
1093
    if not_null_col in table_cols(db, table, recover): return not_null_col
1094
    else: return pkey(db, table, recover)
1095
1096 3348 aaronmk
def constraint_cond(db, constraint):
1097
    module = util.root_module(db.db)
1098
    if module == 'psycopg2':
1099
        table_str = sql_gen.Literal(constraint.table.to_str(db))
1100
        name_str = sql_gen.Literal(constraint.name)
1101
        return value(run_query(db, '''\
1102
SELECT consrc
1103
FROM pg_constraint
1104
WHERE
1105
conrelid = '''+table_str.to_str(db)+'''::regclass
1106
AND conname = '''+name_str.to_str(db)+'''
1107
'''
1108
            , cacheable=True, log_level=4))
1109
    else: raise NotImplementedError("Can't list index columns for "+module+
1110
        ' database')
1111
1112 3319 aaronmk
def index_cols(db, index):
1113 853 aaronmk
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
1114
    automatically created. When you don't know whether something is a UNIQUE
1115
    constraint or a UNIQUE index, use this function.'''
1116 3322 aaronmk
    index = sql_gen.as_Table(index)
1117 1909 aaronmk
    module = util.root_module(db.db)
1118
    if module == 'psycopg2':
1119 3322 aaronmk
        qual_index = sql_gen.Literal(index.to_str(db))
1120
        return map(parse_expr_col, values(run_query(db, '''\
1121
SELECT pg_get_indexdef(indexrelid, generate_series(1, indnatts), true)
1122
FROM pg_index
1123
WHERE indexrelid = '''+qual_index.to_str(db)+'''::regclass
1124 2782 aaronmk
'''
1125
            , cacheable=True, log_level=4)))
1126 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
1127
        ' database')
1128 853 aaronmk
1129 3079 aaronmk
#### Functions
1130
1131
def function_exists(db, function):
1132
    function = sql_gen.as_Function(function)
1133
1134
    info_table = sql_gen.Table('routines', 'information_schema')
1135
    conds = [('routine_name', function.name)]
1136
    schema = function.schema
1137
    if schema != None: conds.append(('routine_schema', schema))
1138
    # Exclude trigger functions, since they cannot be called directly
1139
    conds.append(('data_type', sql_gen.CompareCond('trigger', '!=')))
1140
1141
    return list(values(select(db, info_table, ['routine_name'], conds,
1142
        order_by='routine_schema', limit=1, log_level=4))) != []
1143
        # TODO: order_by search_path schema order
1144
1145
##### Structural changes
1146
1147
#### Columns
1148
1149
def add_col(db, table, col, comment=None, **kw_args):
1150
    '''
1151
    @param col TypedCol Name may be versioned, so be sure to propagate any
1152
        renaming back to any source column for the TypedCol.
1153
    @param comment None|str SQL comment used to distinguish columns of the same
1154
        name from each other when they contain different data, to allow the
1155
        ADD COLUMN query to be cached. If not set, query will not be cached.
1156
    '''
1157
    assert isinstance(col, sql_gen.TypedCol)
1158
1159
    while True:
1160
        str_ = 'ALTER TABLE '+table.to_str(db)+' ADD COLUMN '+col.to_str(db)
1161
        if comment != None: str_ += ' '+sql_gen.esc_comment(comment)
1162
1163
        try:
1164
            run_query(db, str_, recover=True, cacheable=True, **kw_args)
1165
            break
1166
        except DuplicateException:
1167
            col.name = next_version(col.name)
1168
            # try again with next version of name
1169
1170
def add_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)+' SET NOT NULL', cacheable=True, log_level=3)
1175
1176 2096 aaronmk
row_num_col = '_row_num'
1177
1178 3079 aaronmk
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 2998 aaronmk
def add_index(db, exprs, table=None, unique=False, ensure_not_null_=True):
1203 2688 aaronmk
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
1204 3356 aaronmk
    Currently, only function calls and literal values are supported expressions.
1205 2998 aaronmk
    @param ensure_not_null_ If set, translates NULL values to sentinel values.
1206 2847 aaronmk
        This allows indexes to be used for comparisons where NULLs are equal.
1207 2538 aaronmk
    '''
1208 2964 aaronmk
    exprs = lists.mk_seq(exprs)
1209 2538 aaronmk
1210 2688 aaronmk
    # Parse exprs
1211
    old_exprs = exprs[:]
1212
    exprs = []
1213
    cols = []
1214
    for i, expr in enumerate(old_exprs):
1215 2823 aaronmk
        expr = sql_gen.as_Col(expr, table)
1216 2688 aaronmk
1217 2823 aaronmk
        # Handle nullable columns
1218 2998 aaronmk
        if ensure_not_null_:
1219 3164 aaronmk
            try: expr = sql_gen.ensure_not_null(db, expr)
1220 2860 aaronmk
            except KeyError: pass # unknown type, so just create plain index
1221 2823 aaronmk
1222 2688 aaronmk
        # Extract col
1223 3002 aaronmk
        expr = copy.deepcopy(expr) # don't modify input!
1224 3356 aaronmk
        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 2688 aaronmk
            expr = sql_gen.Expr(expr)
1229 3356 aaronmk
1230 2688 aaronmk
1231
        # Extract table
1232
        if table == None:
1233
            assert sql_gen.is_table_col(col)
1234
            table = col.table
1235
1236 3356 aaronmk
        if isinstance(col, sql_gen.Col): col.table = None
1237 2688 aaronmk
1238
        exprs.append(expr)
1239
        cols.append(col)
1240 2408 aaronmk
1241 2688 aaronmk
    table = sql_gen.as_Table(table)
1242
1243 3005 aaronmk
    # Add index
1244 3148 aaronmk
    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 2408 aaronmk
1250 3083 aaronmk
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 3079 aaronmk
#### Tables
1265 2772 aaronmk
1266 3079 aaronmk
### Maintenance
1267 2772 aaronmk
1268 3079 aaronmk
def analyze(db, table):
1269
    table = sql_gen.as_Table(table)
1270
    run_query(db, 'ANALYZE '+table.to_str(db), log_level=3)
1271 2934 aaronmk
1272 3079 aaronmk
def autoanalyze(db, table):
1273
    if db.autoanalyze: analyze(db, table)
1274 2935 aaronmk
1275 3079 aaronmk
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 2086 aaronmk
1280 3079 aaronmk
### Lifecycle
1281
1282 3247 aaronmk
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 2889 aaronmk
1286 3247 aaronmk
def drop_table(db, table): drop(db, 'TABLE', table)
1287
1288 3082 aaronmk
def create_table(db, table, cols=[], has_pkey=True, col_indexes=True,
1289
    like=None):
1290 2675 aaronmk
    '''Creates a table.
1291 2681 aaronmk
    @param cols [sql_gen.TypedCol,...] The column names and types
1292
    @param has_pkey If set, the first column becomes the primary key.
1293 2760 aaronmk
    @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 2675 aaronmk
    '''
1298
    table = sql_gen.as_Table(table)
1299
1300 3082 aaronmk
    if like != None:
1301
        cols = [sql_gen.CustomCode('LIKE '+like.to_str(db)+' INCLUDING ALL')
1302
            ]+cols
1303 2681 aaronmk
    if has_pkey:
1304
        cols[0] = pkey = copy.copy(cols[0]) # don't modify input!
1305 2872 aaronmk
        pkey.constraints = 'PRIMARY KEY'
1306 2681 aaronmk
1307 3085 aaronmk
    temp = table.is_temp and not db.debug_temp
1308
        # temp tables permanent in debug_temp mode
1309 2760 aaronmk
1310 3085 aaronmk
    # Create table
1311 3383 aaronmk
    def create():
1312 3085 aaronmk
        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 3126 aaronmk
        str_ += '\n);'
1317 3085 aaronmk
1318 3383 aaronmk
        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 3085 aaronmk
1330 2760 aaronmk
    # Add indexes
1331 2773 aaronmk
    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 2675 aaronmk
1336 3084 aaronmk
def copy_table_struct(db, src, dest):
1337
    '''Creates a structure-only copy of a table. (Does not copy data.)'''
1338 3085 aaronmk
    create_table(db, dest, has_pkey=False, col_indexes=False, like=src)
1339 3084 aaronmk
1340 3079 aaronmk
### Data
1341 2684 aaronmk
1342 2970 aaronmk
def truncate(db, table, schema='public', **kw_args):
1343
    '''For params, see run_query()'''
1344 2777 aaronmk
    table = sql_gen.as_Table(table, schema)
1345 2970 aaronmk
    return run_query(db, 'TRUNCATE '+table.to_str(db)+' CASCADE', **kw_args)
1346 2732 aaronmk
1347 2965 aaronmk
def empty_temp(db, tables):
1348
    tables = lists.mk_seq(tables)
1349 2971 aaronmk
    for table in tables: truncate(db, table, log_level=3)
1350 2965 aaronmk
1351 1968 aaronmk
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 3094 aaronmk
1355
def distinct_table(db, table, distinct_on):
1356
    '''Creates a copy of a temp table which is distinct on the given columns.
1357 3099 aaronmk
    The old and new tables will both get an index on these columns, to
1358
    facilitate merge joins.
1359 3097 aaronmk
    @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 3099 aaronmk
    @return The new table.
1362 3094 aaronmk
    '''
1363 3099 aaronmk
    new_table = sql_gen.suffixed_table(table, '_distinct')
1364 3411 aaronmk
    distinct_on = filter(sql_gen.is_table_col, distinct_on)
1365 3094 aaronmk
1366 3099 aaronmk
    copy_table_struct(db, table, new_table)
1367 3097 aaronmk
1368
    limit = None
1369
    if distinct_on == []: limit = 1 # one sample row
1370 3099 aaronmk
    else:
1371
        add_index(db, distinct_on, new_table, unique=True)
1372
        add_index(db, distinct_on, table) # for join optimization
1373 3097 aaronmk
1374 3313 aaronmk
    insert_select(db, new_table, None, mk_select(db, table, order_by=None,
1375
        limit=limit), ignore=True)
1376 3099 aaronmk
    analyze(db, new_table)
1377 3094 aaronmk
1378 3099 aaronmk
    return new_table