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 4491 aaronmk
        DbException.__init__(self, 'for name: '
41
            +strings.as_tt(strings.ustr(name)), cause)
42 2143 aaronmk
        self.name = name
43 360 aaronmk
44 3109 aaronmk
class ExceptionWithValue(DbException):
45
    def __init__(self, value, cause=None):
46 4492 aaronmk
        DbException.__init__(self, 'for value: '
47
            +strings.as_tt(strings.urepr(value)), cause)
48 2240 aaronmk
        self.value = value
49
50 2945 aaronmk
class ExceptionWithNameType(DbException):
51
    def __init__(self, type_, name, cause=None):
52 4491 aaronmk
        DbException.__init__(self, 'for type: '+strings.as_tt(strings.ustr(
53
            type_))+'; name: '+strings.as_tt(name), cause)
54 2945 aaronmk
        self.type = type_
55
        self.name = name
56
57 2306 aaronmk
class ConstraintException(DbException):
58 3345 aaronmk
    def __init__(self, name, cond, cols, cause=None):
59
        msg = 'Violated '+strings.as_tt(name)+' constraint'
60 5447 aaronmk
        if cond != None: msg += ' with condition '+strings.as_tt(cond)
61 3345 aaronmk
        if cols != []: msg += ' on columns: '+strings.as_tt(', '.join(cols))
62
        DbException.__init__(self, msg, cause)
63 2306 aaronmk
        self.name = name
64 3345 aaronmk
        self.cond = cond
65 468 aaronmk
        self.cols = cols
66 11 aaronmk
67 2523 aaronmk
class MissingCastException(DbException):
68 4139 aaronmk
    def __init__(self, type_, col=None, cause=None):
69
        msg = 'Missing cast to type '+strings.as_tt(type_)
70
        if col != None: msg += ' on column: '+strings.as_tt(col)
71
        DbException.__init__(self, msg, cause)
72 2523 aaronmk
        self.type = type_
73
        self.col = col
74
75 5576 aaronmk
class EncodingException(ExceptionWithName): pass
76
77 2306 aaronmk
class DuplicateKeyException(ConstraintException): pass
78 13 aaronmk
79 2306 aaronmk
class NullValueException(ConstraintException): pass
80 13 aaronmk
81 3346 aaronmk
class CheckException(ConstraintException): pass
82
83 3109 aaronmk
class InvalidValueException(ExceptionWithValue): pass
84 2239 aaronmk
85 2945 aaronmk
class DuplicateException(ExceptionWithNameType): pass
86 2143 aaronmk
87 3419 aaronmk
class DoesNotExistException(ExceptionWithNameType): pass
88
89 89 aaronmk
class EmptyRowException(DbException): pass
90
91 865 aaronmk
##### Warnings
92
93
class DbWarning(UserWarning): pass
94
95 1930 aaronmk
##### Result retrieval
96
97
def col_names(cur): return (col[0] for col in cur.description)
98
99
def rows(cur): return iter(lambda: cur.fetchone(), None)
100
101
def consume_rows(cur):
102
    '''Used to fetch all rows so result will be cached'''
103
    iters.consume_iter(rows(cur))
104
105
def next_row(cur): return rows(cur).next()
106
107
def row(cur):
108
    row_ = next_row(cur)
109
    consume_rows(cur)
110
    return row_
111
112
def next_value(cur): return next_row(cur)[0]
113
114
def value(cur): return row(cur)[0]
115
116
def values(cur): return iters.func_iter(lambda: next_value(cur))
117
118
def value_or_none(cur):
119
    try: return value(cur)
120
    except StopIteration: return None
121
122 2762 aaronmk
##### Escaping
123 2101 aaronmk
124 2573 aaronmk
def esc_name_by_module(module, name):
125
    if module == 'psycopg2' or module == None: quote = '"'
126 2101 aaronmk
    elif module == 'MySQLdb': quote = '`'
127
    else: raise NotImplementedError("Can't escape name for "+module+' database')
128 2500 aaronmk
    return sql_gen.esc_name(name, quote)
129 2101 aaronmk
130
def esc_name_by_engine(engine, name, **kw_args):
131
    return esc_name_by_module(db_engines[engine][0], name, **kw_args)
132
133
def esc_name(db, name, **kw_args):
134
    return esc_name_by_module(util.root_module(db.db), name, **kw_args)
135
136
def qual_name(db, schema, table):
137
    def esc_name_(name): return esc_name(db, name)
138
    table = esc_name_(table)
139
    if schema != None: return esc_name_(schema)+'.'+table
140
    else: return table
141
142 1869 aaronmk
##### Database connections
143 1849 aaronmk
144 2097 aaronmk
db_config_names = ['engine', 'host', 'user', 'password', 'database', 'schemas']
145 1926 aaronmk
146 1869 aaronmk
db_engines = {
147
    'MySQL': ('MySQLdb', {'password': 'passwd', 'database': 'db'}),
148
    'PostgreSQL': ('psycopg2', {}),
149
}
150
151
DatabaseErrors_set = set([DbException])
152
DatabaseErrors = tuple(DatabaseErrors_set)
153
154
def _add_module(module):
155
    DatabaseErrors_set.add(module.DatabaseError)
156
    global DatabaseErrors
157
    DatabaseErrors = tuple(DatabaseErrors_set)
158
159
def db_config_str(db_config):
160
    return db_config['engine']+' database '+db_config['database']
161
162 2448 aaronmk
log_debug_none = lambda msg, level=2: None
163 1901 aaronmk
164 1849 aaronmk
class DbConn:
165 2923 aaronmk
    def __init__(self, db_config, autocommit=True, caching=True,
166 3183 aaronmk
        log_debug=log_debug_none, debug_temp=False, src=None):
167 2915 aaronmk
        '''
168
        @param debug_temp Whether temporary objects should instead be permanent.
169
            This assists in debugging the internal objects used by the program.
170 3183 aaronmk
        @param src In autocommit mode, will be included in a comment in every
171
            query, to help identify the data source in pg_stat_activity.
172 2915 aaronmk
        '''
173 1869 aaronmk
        self.db_config = db_config
174 2190 aaronmk
        self.autocommit = autocommit
175
        self.caching = caching
176 1901 aaronmk
        self.log_debug = log_debug
177 2193 aaronmk
        self.debug = log_debug != log_debug_none
178 2915 aaronmk
        self.debug_temp = debug_temp
179 3183 aaronmk
        self.src = src
180 3074 aaronmk
        self.autoanalyze = False
181 3269 aaronmk
        self.autoexplain = False
182
        self.profile_row_ct = None
183 1869 aaronmk
184 3124 aaronmk
        self._savepoint = 0
185 3120 aaronmk
        self._reset()
186 1869 aaronmk
187
    def __getattr__(self, name):
188
        if name == '__dict__': raise Exception('getting __dict__')
189
        if name == 'db': return self._db()
190
        else: raise AttributeError()
191
192
    def __getstate__(self):
193
        state = copy.copy(self.__dict__) # shallow copy
194 1915 aaronmk
        state['log_debug'] = None # don't pickle the debug callback
195 1869 aaronmk
        state['_DbConn__db'] = None # don't pickle the connection
196
        return state
197
198 3118 aaronmk
    def clear_cache(self): self.query_results = {}
199
200 3120 aaronmk
    def _reset(self):
201 3118 aaronmk
        self.clear_cache()
202 3124 aaronmk
        assert self._savepoint == 0
203 3118 aaronmk
        self._notices_seen = set()
204
        self.__db = None
205
206 2165 aaronmk
    def connected(self): return self.__db != None
207
208 3116 aaronmk
    def close(self):
209 3119 aaronmk
        if not self.connected(): return
210
211 3135 aaronmk
        # Record that the automatic transaction is now closed
212 3136 aaronmk
        self._savepoint -= 1
213 3135 aaronmk
214 3119 aaronmk
        self.db.close()
215 3120 aaronmk
        self._reset()
216 3116 aaronmk
217 3125 aaronmk
    def reconnect(self):
218
        # Do not do this in test mode as it would roll back everything
219
        if self.autocommit: self.close()
220
        # Connection will be reopened automatically on first query
221
222 1869 aaronmk
    def _db(self):
223
        if self.__db == None:
224
            # Process db_config
225
            db_config = self.db_config.copy() # don't modify input!
226 2097 aaronmk
            schemas = db_config.pop('schemas', None)
227 1869 aaronmk
            module_name, mappings = db_engines[db_config.pop('engine')]
228
            module = __import__(module_name)
229
            _add_module(module)
230
            for orig, new in mappings.iteritems():
231
                try: util.rename_key(db_config, orig, new)
232
                except KeyError: pass
233
234
            # Connect
235
            self.__db = module.connect(**db_config)
236
237 3161 aaronmk
            # Record that a transaction is already open
238
            self._savepoint += 1
239
240 1869 aaronmk
            # Configure connection
241 2906 aaronmk
            if hasattr(self.db, 'set_isolation_level'):
242
                import psycopg2.extensions
243
                self.db.set_isolation_level(
244
                    psycopg2.extensions.ISOLATION_LEVEL_READ_COMMITTED)
245 2101 aaronmk
            if schemas != None:
246 2893 aaronmk
                search_path = [self.esc_name(s) for s in schemas.split(',')]
247
                search_path.append(value(run_query(self, 'SHOW search_path',
248
                    log_level=4)))
249
                run_query(self, 'SET search_path TO '+(','.join(search_path)),
250
                    log_level=3)
251 1869 aaronmk
252
        return self.__db
253 1889 aaronmk
254 1891 aaronmk
    class DbCursor(Proxy):
255 1927 aaronmk
        def __init__(self, outer):
256 1891 aaronmk
            Proxy.__init__(self, outer.db.cursor())
257 2191 aaronmk
            self.outer = outer
258 1927 aaronmk
            self.query_results = outer.query_results
259 1894 aaronmk
            self.query_lookup = None
260 1891 aaronmk
            self.result = []
261 1889 aaronmk
262 2802 aaronmk
        def execute(self, query):
263 2764 aaronmk
            self._is_insert = query.startswith('INSERT')
264 2797 aaronmk
            self.query_lookup = query
265 2148 aaronmk
            try:
266 3162 aaronmk
                try: cur = self.inner.execute(query)
267 2802 aaronmk
                finally: self.query = get_cur_query(self.inner, query)
268 1904 aaronmk
            except Exception, e:
269
                self.result = e # cache the exception as the result
270
                self._cache_result()
271
                raise
272 3004 aaronmk
273
            # Always cache certain queries
274 3183 aaronmk
            query = sql_gen.lstrip(query)
275 3004 aaronmk
            if query.startswith('CREATE') or query.startswith('ALTER'):
276 3007 aaronmk
                # structural changes
277 3040 aaronmk
                # Rest of query must be unique in the face of name collisions,
278
                # so don't cache ADD COLUMN unless it has distinguishing comment
279
                if query.find('ADD COLUMN') < 0 or query.endswith('*/'):
280 3007 aaronmk
                    self._cache_result()
281 3004 aaronmk
            elif self.rowcount == 0 and query.startswith('SELECT'): # empty
282 2800 aaronmk
                consume_rows(self) # fetch all rows so result will be cached
283 3004 aaronmk
284 2762 aaronmk
            return cur
285 1894 aaronmk
286 1891 aaronmk
        def fetchone(self):
287
            row = self.inner.fetchone()
288 1899 aaronmk
            if row != None: self.result.append(row)
289
            # otherwise, fetched all rows
290 1904 aaronmk
            else: self._cache_result()
291
            return row
292
293
        def _cache_result(self):
294 2948 aaronmk
            # For inserts that return a result set, don't cache result set since
295
            # inserts are not idempotent. Other non-SELECT queries don't have
296
            # their result set read, so only exceptions will be cached (an
297
            # invalid query will always be invalid).
298 1930 aaronmk
            if self.query_results != None and (not self._is_insert
299 1906 aaronmk
                or isinstance(self.result, Exception)):
300
301 1894 aaronmk
                assert self.query_lookup != None
302 1916 aaronmk
                self.query_results[self.query_lookup] = self.CacheCursor(
303
                    util.dict_subset(dicts.AttrsDictView(self),
304
                    ['query', 'result', 'rowcount', 'description']))
305 1906 aaronmk
306 1916 aaronmk
        class CacheCursor:
307
            def __init__(self, cached_result): self.__dict__ = cached_result
308
309 1927 aaronmk
            def execute(self, *args, **kw_args):
310 1916 aaronmk
                if isinstance(self.result, Exception): raise self.result
311
                # otherwise, result is a rows list
312
                self.iter = iter(self.result)
313
314
            def fetchone(self):
315
                try: return self.iter.next()
316
                except StopIteration: return None
317 1891 aaronmk
318 2212 aaronmk
    def esc_value(self, value):
319 2663 aaronmk
        try: str_ = self.mogrify('%s', [value])
320
        except NotImplementedError, e:
321
            module = util.root_module(self.db)
322
            if module == 'MySQLdb':
323
                import _mysql
324
                str_ = _mysql.escape_string(value)
325
            else: raise e
326 2374 aaronmk
        return strings.to_unicode(str_)
327 2212 aaronmk
328 2347 aaronmk
    def esc_name(self, name): return esc_name(self, name) # calls global func
329
330 2814 aaronmk
    def std_code(self, str_):
331
        '''Standardizes SQL code.
332
        * Ensures that string literals are prefixed by `E`
333
        '''
334
        if str_.startswith("'"): str_ = 'E'+str_
335
        return str_
336
337 2665 aaronmk
    def can_mogrify(self):
338 2663 aaronmk
        module = util.root_module(self.db)
339 2665 aaronmk
        return module == 'psycopg2'
340 2663 aaronmk
341 2665 aaronmk
    def mogrify(self, query, params=None):
342
        if self.can_mogrify(): return self.db.cursor().mogrify(query, params)
343
        else: raise NotImplementedError("Can't mogrify query")
344
345 5579 aaronmk
    def set_encoding(self, encoding):
346
        encoding_str = sql_gen.Literal(encoding)
347
        run_query(self, 'SET NAMES '+encoding_str.to_str(self))
348
349 2671 aaronmk
    def print_notices(self):
350 2725 aaronmk
        if hasattr(self.db, 'notices'):
351
            for msg in self.db.notices:
352
                if msg not in self._notices_seen:
353
                    self._notices_seen.add(msg)
354
                    self.log_debug(msg, level=2)
355 2671 aaronmk
356 2793 aaronmk
    def run_query(self, query, cacheable=False, log_level=2,
357 2464 aaronmk
        debug_msg_ref=None):
358 2445 aaronmk
        '''
359 2464 aaronmk
        @param log_ignore_excs The log_level will be increased by 2 if the query
360
            throws one of these exceptions.
361 2664 aaronmk
        @param debug_msg_ref If specified, the log message will be returned in
362
            this instead of being output. This allows you to filter log messages
363
            depending on the result of the query.
364 2445 aaronmk
        '''
365 2167 aaronmk
        assert query != None
366
367 3183 aaronmk
        if self.autocommit and self.src != None:
368 3206 aaronmk
            query = sql_gen.esc_comment(self.src)+'\t'+query
369 3183 aaronmk
370 2047 aaronmk
        if not self.caching: cacheable = False
371 1903 aaronmk
        used_cache = False
372 2664 aaronmk
373 3242 aaronmk
        if self.debug:
374
            profiler = profiling.ItersProfiler(start_now=True, iter_text='row')
375 1903 aaronmk
        try:
376 1927 aaronmk
            # Get cursor
377
            if cacheable:
378 3238 aaronmk
                try: cur = self.query_results[query]
379 1927 aaronmk
                except KeyError: cur = self.DbCursor(self)
380 3238 aaronmk
                else: used_cache = True
381 1927 aaronmk
            else: cur = self.db.cursor()
382
383
            # Run query
384 3238 aaronmk
            try: cur.execute(query)
385 3162 aaronmk
            except Exception, e:
386
                _add_cursor_info(e, self, query)
387
                raise
388 3238 aaronmk
            else: self.do_autocommit()
389 1903 aaronmk
        finally:
390 3242 aaronmk
            if self.debug:
391 3244 aaronmk
                profiler.stop(self.profile_row_ct)
392 3242 aaronmk
393
                ## Log or return query
394
395 4491 aaronmk
                query = strings.ustr(get_cur_query(cur, query))
396 3281 aaronmk
                # Put the src comment on a separate line in the log file
397
                query = query.replace('\t', '\n', 1)
398 3239 aaronmk
399 3240 aaronmk
                msg = 'DB query: '
400 3239 aaronmk
401 3240 aaronmk
                if used_cache: msg += 'cache hit'
402
                elif cacheable: msg += 'cache miss'
403
                else: msg += 'non-cacheable'
404 3239 aaronmk
405 3241 aaronmk
                msg += ':\n'+profiler.msg()+'\n'+strings.as_code(query, 'SQL')
406 3240 aaronmk
407 3237 aaronmk
                if debug_msg_ref != None: debug_msg_ref[0] = msg
408
                else: self.log_debug(msg, log_level)
409 3245 aaronmk
410
                self.print_notices()
411 1903 aaronmk
412
        return cur
413 1914 aaronmk
414 2797 aaronmk
    def is_cached(self, query): return query in self.query_results
415 2139 aaronmk
416 2907 aaronmk
    def with_autocommit(self, func):
417 2801 aaronmk
        import psycopg2.extensions
418
419
        prev_isolation_level = self.db.isolation_level
420 2907 aaronmk
        self.db.set_isolation_level(
421
            psycopg2.extensions.ISOLATION_LEVEL_AUTOCOMMIT)
422 2683 aaronmk
        try: return func()
423 2801 aaronmk
        finally: self.db.set_isolation_level(prev_isolation_level)
424 2683 aaronmk
425 2139 aaronmk
    def with_savepoint(self, func):
426 3137 aaronmk
        top = self._savepoint == 0
427 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
428 3137 aaronmk
429 3272 aaronmk
        if self.debug:
430 3273 aaronmk
            self.log_debug('Begin transaction', level=4)
431 3272 aaronmk
            profiler = profiling.ItersProfiler(start_now=True, iter_text='row')
432
433 3160 aaronmk
        # Must happen before running queries so they don't get autocommitted
434
        self._savepoint += 1
435
436 3137 aaronmk
        if top: query = 'START TRANSACTION ISOLATION LEVEL READ COMMITTED'
437
        else: query = 'SAVEPOINT '+savepoint
438
        self.run_query(query, log_level=4)
439
        try:
440
            return func()
441
            if top: self.run_query('COMMIT', log_level=4)
442 2139 aaronmk
        except:
443 3137 aaronmk
            if top: query = 'ROLLBACK'
444
            else: query = 'ROLLBACK TO SAVEPOINT '+savepoint
445
            self.run_query(query, log_level=4)
446
447 2139 aaronmk
            raise
448 2930 aaronmk
        finally:
449
            # Always release savepoint, because after ROLLBACK TO SAVEPOINT,
450
            # "The savepoint remains valid and can be rolled back to again"
451
            # (http://www.postgresql.org/docs/8.3/static/sql-rollback-to.html).
452 3137 aaronmk
            if not top:
453
                self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
454 2930 aaronmk
455
            self._savepoint -= 1
456
            assert self._savepoint >= 0
457
458 3272 aaronmk
            if self.debug:
459
                profiler.stop(self.profile_row_ct)
460 3273 aaronmk
                self.log_debug('End transaction\n'+profiler.msg(), level=4)
461 3272 aaronmk
462 2930 aaronmk
            self.do_autocommit() # OK to do this after ROLLBACK TO SAVEPOINT
463 2191 aaronmk
464
    def do_autocommit(self):
465
        '''Autocommits if outside savepoint'''
466 3135 aaronmk
        assert self._savepoint >= 1
467
        if self.autocommit and self._savepoint == 1:
468 2924 aaronmk
            self.log_debug('Autocommitting', level=4)
469 2191 aaronmk
            self.db.commit()
470 2643 aaronmk
471 3155 aaronmk
    def col_info(self, col, cacheable=True):
472 6900 aaronmk
        module = util.root_module(self.db)
473 6899 aaronmk
        if module == 'psycopg2':
474 6900 aaronmk
            qual_table = sql_gen.Literal(col.table.to_str(self))
475 6899 aaronmk
            col_name_str = sql_gen.Literal(col.name)
476
            try:
477 6900 aaronmk
                type_, is_array, default, nullable = row(run_query(self, '''\
478 6899 aaronmk
SELECT
479
format_type(COALESCE(NULLIF(typelem, 0), pg_type.oid), -1) AS type
480
, typcategory = 'A' AS type_is_array
481 6903 aaronmk
, pg_get_expr(pg_attrdef.adbin, attrelid, true) AS default
482 6899 aaronmk
, NOT pg_attribute.attnotnull AS nullable
483
FROM pg_attribute
484
LEFT JOIN pg_type ON pg_type.oid = atttypid
485
LEFT JOIN pg_attrdef ON adrelid = attrelid AND adnum = attnum
486
WHERE
487 6901 aaronmk
attrelid = '''+qual_table.to_str(self)+'''::regclass
488
AND attname = '''+col_name_str.to_str(self)+'''
489 6899 aaronmk
'''
490 6904 aaronmk
                    , recover=True, cacheable=cacheable, log_level=4))
491 6905 aaronmk
            except (DoesNotExistException, StopIteration):
492
                raise sql_gen.NoUnderlyingTableException(col)
493 6900 aaronmk
            if is_array: type_ = sql_gen.ArrayType(type_)
494 6899 aaronmk
        else:
495
            table = sql_gen.Table('columns', 'information_schema')
496
            cols = [sql_gen.Col('data_type'), sql_gen.Col('udt_name'),
497
                'column_default', sql_gen.Cast('boolean',
498
                sql_gen.Col('is_nullable'))]
499
500
            conds = [('table_name', col.table.name),
501
                ('column_name', strings.ustr(col.name))]
502
            schema = col.table.schema
503
            if schema != None: conds.append(('table_schema', schema))
504
505
            cur = select(self, table, cols, conds, order_by='table_schema',
506
                limit=1, cacheable=cacheable, log_level=4)
507
            try: type_, extra_type, default, nullable = row(cur)
508
            except StopIteration: raise sql_gen.NoUnderlyingTableException(col)
509
            if type_ == 'USER-DEFINED': type_ = extra_type
510
            elif type_ == 'ARRAY':
511
                type_ = sql_gen.ArrayType(strings.remove_prefix('_', extra_type,
512
                    require=True))
513 2643 aaronmk
514 6899 aaronmk
        if default != None: default = sql_gen.as_Code(default, self)
515 2819 aaronmk
        return sql_gen.TypedCol(col.name, type_, default, nullable)
516 2917 aaronmk
517
    def TempFunction(self, name):
518
        if self.debug_temp: schema = None
519
        else: schema = 'pg_temp'
520
        return sql_gen.Function(name, schema)
521 1849 aaronmk
522 1869 aaronmk
connect = DbConn
523
524 832 aaronmk
##### Recoverable querying
525 15 aaronmk
526 5577 aaronmk
def parse_exception(db, e, recover=False):
527
    msg = strings.ustr(e.args[0])
528
    msg = re.sub(r'^(?:PL/Python: )?ValueError: ', r'', msg)
529
530
    match = re.match(r'^invalid byte sequence for encoding "(.+?)":', msg)
531
    if match:
532
        encoding, = match.groups()
533
        raise EncodingException(encoding, e)
534
535 5763 aaronmk
    def make_DuplicateKeyException(constraint, e):
536 5577 aaronmk
        cols = []
537
        cond = None
538
        if recover: # need auto-rollback to run index_cols()
539
            try:
540
                cols = index_cols(db, constraint)
541
                cond = index_cond(db, constraint)
542
            except NotImplementedError: pass
543 5763 aaronmk
        return DuplicateKeyException(constraint, cond, cols, e)
544 5577 aaronmk
545 5763 aaronmk
    match = re.match(r'^duplicate key value violates unique constraint "(.+?)"',
546
        msg)
547
    if match:
548
        constraint, = match.groups()
549
        raise make_DuplicateKeyException(constraint, e)
550
551 5764 aaronmk
    match = re.match(r'^could not create unique index "(.+?)"\n'
552
        r'DETAIL:  Key .+? is duplicated', msg)
553
    if match:
554
        constraint, = match.groups()
555 5823 aaronmk
        raise DuplicateKeyException(constraint, None, [], e)
556 5764 aaronmk
557 5577 aaronmk
    match = re.match(r'^null value in column "(.+?)" violates not-null'
558
        r' constraint', msg)
559
    if match:
560
        col, = match.groups()
561
        raise NullValueException('NOT NULL', None, [col], e)
562
563
    match = re.match(r'^new row for relation "(.+?)" violates check '
564
        r'constraint "(.+?)"', msg)
565
    if match:
566
        table, constraint = match.groups()
567
        constraint = sql_gen.Col(constraint, table)
568
        cond = None
569
        if recover: # need auto-rollback to run constraint_cond()
570
            try: cond = constraint_cond(db, constraint)
571
            except NotImplementedError: pass
572
        raise CheckException(constraint.to_str(db), cond, [], e)
573
574 5723 aaronmk
    match = re.match(r'^(?:invalid input (?:syntax|value)\b[^:]*'
575 5717 aaronmk
        r'|.+? out of range)(?:: "(.+?)")?', msg)
576 5577 aaronmk
    if match:
577
        value, = match.groups()
578 5717 aaronmk
        value = util.do_ignore_none(strings.to_unicode, value)
579
        raise InvalidValueException(value, e)
580 5577 aaronmk
581
    match = re.match(r'^column "(.+?)" is of type (.+?) but expression '
582
        r'is of type', msg)
583
    if match:
584
        col, type_ = match.groups()
585
        raise MissingCastException(type_, col, e)
586
587
    match = re.match(r'^could not determine polymorphic type because '
588
        r'input has type "unknown"', msg)
589
    if match: raise MissingCastException('text', None, e)
590
591 5724 aaronmk
    match = re.match(r'^.+? types (.+?) and (.+?) cannot be matched', msg)
592
    if match:
593
        type0, type1 = match.groups()
594
        raise MissingCastException(type0, None, e)
595 5577 aaronmk
596 5707 aaronmk
    typed_name_re = r'^(\S+) "?(.+?)"?(?: of relation ".+?")?'
597 5577 aaronmk
598
    match = re.match(typed_name_re+r'.*? already exists', msg)
599
    if match:
600
        type_, name = match.groups()
601
        raise DuplicateException(type_, name, e)
602
603
    match = re.match(r'more than one (\S+) named ""(.+?)""', msg)
604
    if match:
605
        type_, name = match.groups()
606
        raise DuplicateException(type_, name, e)
607
608
    match = re.match(typed_name_re+r' does not exist', msg)
609
    if match:
610
        type_, name = match.groups()
611 5683 aaronmk
        if type_ == 'function':
612 5709 aaronmk
            match = re.match(r'^(.+?)\(.*\)$', name)
613 5710 aaronmk
            if match: # includes params, so is call rather than cast to regproc
614 5708 aaronmk
                function_name, = match.groups()
615 5711 aaronmk
                func = sql_gen.Function(function_name)
616 5712 aaronmk
                if function_exists(db, func) and msg.find('CAST') < 0:
617 5683 aaronmk
                    # not found only because of a missing cast
618 5715 aaronmk
                    type_ = function_param0_type(db, func)
619 6301 aaronmk
                    col = None
620 5792 aaronmk
                    if type_ == 'anyelement': type_ = 'text'
621 6301 aaronmk
                    elif type_ == 'hstore': # cast just the value param
622
                        type_ = 'text'
623
                        col = 'value'
624
                    raise MissingCastException(type_, col, e)
625 5577 aaronmk
        raise DoesNotExistException(type_, name, e)
626
627
    raise # no specific exception raised
628
629 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
630 11 aaronmk
631 2791 aaronmk
def run_query(db, query, recover=None, cacheable=False, log_level=2,
632
    log_ignore_excs=None, **kw_args):
633 2794 aaronmk
    '''For params, see DbConn.run_query()'''
634 830 aaronmk
    if recover == None: recover = False
635 2464 aaronmk
    if log_ignore_excs == None: log_ignore_excs = ()
636
    log_ignore_excs = tuple(log_ignore_excs)
637 3236 aaronmk
    debug_msg_ref = [None]
638 830 aaronmk
639 3267 aaronmk
    query = with_explain_comment(db, query)
640 3258 aaronmk
641 2148 aaronmk
    try:
642 2464 aaronmk
        try:
643 2794 aaronmk
            def run(): return db.run_query(query, cacheable, log_level,
644 2793 aaronmk
                debug_msg_ref, **kw_args)
645 2796 aaronmk
            if recover and not db.is_cached(query):
646 2464 aaronmk
                return with_savepoint(db, run)
647
            else: return run() # don't need savepoint if cached
648 5890 aaronmk
        except Exception, e:
649
            # Give failed EXPLAIN approximately the log_level of its query
650
            if query.startswith('EXPLAIN'): log_level -= 1
651
652
            parse_exception(db, e, recover)
653 2464 aaronmk
    except log_ignore_excs:
654
        log_level += 2
655
        raise
656
    finally:
657 3236 aaronmk
        if debug_msg_ref[0] != None: db.log_debug(debug_msg_ref[0], log_level)
658 830 aaronmk
659 832 aaronmk
##### Basic queries
660
661 3256 aaronmk
def is_explainable(query):
662
    # See <http://www.postgresql.org/docs/8.3/static/sql-explain.html#AEN57749>
663 3257 aaronmk
    return re.match(r'^(?:SELECT|INSERT|UPDATE|DELETE|VALUES|EXECUTE|DECLARE)\b'
664
        , query)
665 3256 aaronmk
666 3263 aaronmk
def explain(db, query, **kw_args):
667
    '''
668
    For params, see run_query().
669
    '''
670 3267 aaronmk
    kw_args.setdefault('log_level', 4)
671 3263 aaronmk
672 3750 aaronmk
    return strings.ustr(strings.join_lines(values(run_query(db,
673
        'EXPLAIN '+query, recover=True, cacheable=True, **kw_args))))
674 3256 aaronmk
        # not a higher log_level because it's useful to see what query is being
675
        # run before it's executed, which EXPLAIN effectively provides
676
677 3265 aaronmk
def has_comment(query): return query.endswith('*/')
678
679
def with_explain_comment(db, query, **kw_args):
680 3269 aaronmk
    if db.autoexplain and not has_comment(query) and is_explainable(query):
681 3265 aaronmk
        query += '\n'+sql_gen.esc_comment(' EXPLAIN:\n'
682
            +explain(db, query, **kw_args))
683
    return query
684
685 2153 aaronmk
def next_version(name):
686 2163 aaronmk
    version = 1 # first existing name was version 0
687 2586 aaronmk
    match = re.match(r'^(.*)#(\d+)$', name)
688 2153 aaronmk
    if match:
689 2586 aaronmk
        name, version = match.groups()
690
        version = int(version)+1
691 2932 aaronmk
    return sql_gen.concat(name, '#'+str(version))
692 2153 aaronmk
693 2899 aaronmk
def lock_table(db, table, mode):
694
    table = sql_gen.as_Table(table)
695
    run_query(db, 'LOCK TABLE '+table.to_str(db)+' IN '+mode+' MODE')
696
697 3303 aaronmk
def run_query_into(db, query, into=None, add_pkey_=False, **kw_args):
698 2085 aaronmk
    '''Outputs a query to a temp table.
699
    For params, see run_query().
700
    '''
701 2789 aaronmk
    if into == None: return run_query(db, query, **kw_args)
702 2790 aaronmk
703
    assert isinstance(into, sql_gen.Table)
704
705 2992 aaronmk
    into.is_temp = True
706 3008 aaronmk
    # "temporary tables cannot specify a schema name", so remove schema
707
    into.schema = None
708 2992 aaronmk
709 2790 aaronmk
    kw_args['recover'] = True
710 2945 aaronmk
    kw_args.setdefault('log_ignore_excs', (DuplicateException,))
711 2790 aaronmk
712 2916 aaronmk
    temp = not db.debug_temp # tables are permanent in debug_temp mode
713 2790 aaronmk
714
    # Create table
715
    while True:
716
        create_query = 'CREATE'
717
        if temp: create_query += ' TEMP'
718
        create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
719 2385 aaronmk
720 2790 aaronmk
        try:
721
            cur = run_query(db, create_query, **kw_args)
722
                # CREATE TABLE AS sets rowcount to # rows in query
723
            break
724 2945 aaronmk
        except DuplicateException, e:
725 2790 aaronmk
            into.name = next_version(into.name)
726
            # try again with next version of name
727
728 5967 aaronmk
    if add_pkey_: add_pkey_or_index(db, into, warn=True)
729 3075 aaronmk
730
    # According to the PostgreSQL doc, "The autovacuum daemon cannot access and
731
    # therefore cannot vacuum or analyze temporary tables. [...] if a temporary
732
    # table is going to be used in complex queries, it is wise to run ANALYZE on
733
    # the temporary table after it is populated."
734
    # (http://www.postgresql.org/docs/9.1/static/sql-createtable.html)
735
    # If into is not a temp table, ANALYZE is useful but not required.
736 3073 aaronmk
    analyze(db, into)
737 2790 aaronmk
738
    return cur
739 2085 aaronmk
740 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
741
742 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
743
744 3420 aaronmk
def mk_select(db, tables=None, fields=None, conds=None, distinct_on=[],
745 3494 aaronmk
    limit=None, start=None, order_by=order_by_pkey, default_table=None,
746
    explain=True):
747 1981 aaronmk
    '''
748 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
749 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
750 1981 aaronmk
    @param fields Use None to select all fields in the table
751 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
752 2379 aaronmk
        * container can be any iterable type
753 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
754
        * compare_right_side: sql_gen.ValueCond|literal value
755 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
756
        use all columns
757 2786 aaronmk
    @return query
758 1981 aaronmk
    '''
759 2315 aaronmk
    # Parse tables param
760 2964 aaronmk
    tables = lists.mk_seq(tables)
761 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
762 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
763 2121 aaronmk
764 2315 aaronmk
    # Parse other params
765 2376 aaronmk
    if conds == None: conds = []
766 2650 aaronmk
    elif dicts.is_dict(conds): conds = conds.items()
767 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
768 3129 aaronmk
    assert limit == None or isinstance(limit, (int, long))
769
    assert start == None or isinstance(start, (int, long))
770 5527 aaronmk
    if limit == 0: order_by = None
771 2315 aaronmk
    if order_by is order_by_pkey:
772 5641 aaronmk
        if lists.is_seq(distinct_on) and distinct_on: order_by = distinct_on[0]
773
        elif table0 != None: order_by = table_order_by(db, table0, recover=True)
774
        else: order_by = None
775 865 aaronmk
776 2315 aaronmk
    query = 'SELECT'
777 2056 aaronmk
778 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
779 2056 aaronmk
780 2200 aaronmk
    # DISTINCT ON columns
781 2233 aaronmk
    if distinct_on != []:
782 2467 aaronmk
        query += '\nDISTINCT'
783 2254 aaronmk
        if distinct_on is not distinct_on_all:
784 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
785
786
    # Columns
787 3185 aaronmk
    if query.find('\n') >= 0: whitespace = '\n'
788
    else: whitespace = ' '
789
    if fields == None: query += whitespace+'*'
790 2765 aaronmk
    else:
791
        assert fields != []
792 3185 aaronmk
        if len(fields) > 1: whitespace = '\n'
793
        query += whitespace+('\n, '.join(map(parse_col, fields)))
794 2200 aaronmk
795
    # Main table
796 3185 aaronmk
    if query.find('\n') >= 0 or len(tables) > 0: whitespace = '\n'
797
    else: whitespace = ' '
798 3420 aaronmk
    if table0 != None: query += whitespace+'FROM '+table0.to_str(db)
799 865 aaronmk
800 2122 aaronmk
    # Add joins
801 2271 aaronmk
    left_table = table0
802 2263 aaronmk
    for join_ in tables:
803
        table = join_.table
804 2238 aaronmk
805 2343 aaronmk
        # Parse special values
806
        if join_.type_ is sql_gen.filter_out: # filter no match
807 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
808 2853 aaronmk
                sql_gen.CompareCond(None, '~=')))
809 2343 aaronmk
810 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
811 2122 aaronmk
812
        left_table = table
813
814 865 aaronmk
    missing = True
815 2376 aaronmk
    if conds != []:
816 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
817
        else: whitespace = '\n'
818 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
819
            .to_str(db) for l, r in conds], 'WHERE')
820 2227 aaronmk
    if order_by != None:
821 5642 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by).to_str(db)
822 3297 aaronmk
    if limit != None: query += '\nLIMIT '+str(limit)
823 865 aaronmk
    if start != None:
824 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
825 865 aaronmk
826 3494 aaronmk
    if explain: query = with_explain_comment(db, query)
827 3266 aaronmk
828 2786 aaronmk
    return query
829 11 aaronmk
830 2054 aaronmk
def select(db, *args, **kw_args):
831
    '''For params, see mk_select() and run_query()'''
832
    recover = kw_args.pop('recover', None)
833
    cacheable = kw_args.pop('cacheable', True)
834 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
835 2054 aaronmk
836 2791 aaronmk
    return run_query(db, mk_select(db, *args, **kw_args), recover, cacheable,
837
        log_level=log_level)
838 2054 aaronmk
839 2788 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, returning=None,
840 3181 aaronmk
    embeddable=False, ignore=False, src=None):
841 1960 aaronmk
    '''
842
    @param returning str|None An inserted column (such as pkey) to return
843 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
844 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
845
        query will be fully cached, not just if it raises an exception.
846 3009 aaronmk
    @param ignore Whether to ignore duplicate keys.
847 3181 aaronmk
    @param src Will be included in the name of any created function, to help
848
        identify the data source in pg_stat_activity.
849 1960 aaronmk
    '''
850 2754 aaronmk
    table = sql_gen.remove_table_rename(sql_gen.as_Table(table))
851 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
852 3010 aaronmk
    if cols != None: cols = [sql_gen.to_name_only_col(c, table) for c in cols]
853 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
854 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
855 2063 aaronmk
856 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
857 2063 aaronmk
858 3009 aaronmk
    def mk_insert(select_query):
859
        query = first_line
860 3014 aaronmk
        if cols != None:
861
            query += '\n('+(', '.join((c.to_str(db) for c in cols)))+')'
862 3009 aaronmk
        query += '\n'+select_query
863
864
        if returning != None:
865
            returning_name_col = sql_gen.to_name_only_col(returning)
866
            query += '\nRETURNING '+returning_name_col.to_str(db)
867
868
        return query
869 2063 aaronmk
870 3489 aaronmk
    return_type = sql_gen.CustomCode('unknown')
871
    if returning != None: return_type = sql_gen.ColType(returning)
872 3017 aaronmk
873 3009 aaronmk
    if ignore:
874 3017 aaronmk
        # Always return something to set the correct rowcount
875
        if returning == None: returning = sql_gen.NamedCol('NULL', None)
876
877 3009 aaronmk
        embeddable = True # must use function
878 3010 aaronmk
879 3450 aaronmk
        if cols == None: row = [sql_gen.Col(sql_gen.all_cols, 'row')]
880
        else: row = [sql_gen.Col(c.name, 'row') for c in cols]
881 3092 aaronmk
882 3484 aaronmk
        query = sql_gen.RowExcIgnore(sql_gen.RowType(table), select_query,
883 3497 aaronmk
            sql_gen.ReturnQuery(mk_insert(sql_gen.Values(row).to_str(db))),
884
            cols)
885 3009 aaronmk
    else: query = mk_insert(select_query)
886
887 2070 aaronmk
    if embeddable:
888
        # Create function
889 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
890 3181 aaronmk
        if src != None: function_name = src+': '+function_name
891 2189 aaronmk
        while True:
892
            try:
893 3451 aaronmk
                func = db.TempFunction(function_name)
894 3489 aaronmk
                def_ = sql_gen.FunctionDef(func, sql_gen.SetOf(return_type),
895
                    query)
896 2194 aaronmk
897 3443 aaronmk
                run_query(db, def_.to_str(db), recover=True, cacheable=True,
898 2945 aaronmk
                    log_ignore_excs=(DuplicateException,))
899 2189 aaronmk
                break # this version was successful
900 2945 aaronmk
            except DuplicateException, e:
901 2189 aaronmk
                function_name = next_version(function_name)
902
                # try again with next version of name
903 2070 aaronmk
904 2337 aaronmk
        # Return query that uses function
905 3009 aaronmk
        cols = None
906
        if returning != None: cols = [returning]
907 3451 aaronmk
        func_table = sql_gen.NamedTable('f', sql_gen.FunctionCall(func), cols)
908
            # AS clause requires function alias
909 3298 aaronmk
        return mk_select(db, func_table, order_by=None)
910 2070 aaronmk
911 2787 aaronmk
    return query
912 2066 aaronmk
913 3074 aaronmk
def insert_select(db, table, *args, **kw_args):
914 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
915 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
916
        values in
917 2072 aaronmk
    '''
918 3141 aaronmk
    returning = kw_args.get('returning', None)
919
    ignore = kw_args.get('ignore', False)
920
921 2386 aaronmk
    into = kw_args.pop('into', None)
922
    if into != None: kw_args['embeddable'] = True
923 2066 aaronmk
    recover = kw_args.pop('recover', None)
924 3141 aaronmk
    if ignore: recover = True
925 2066 aaronmk
    cacheable = kw_args.pop('cacheable', True)
926 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
927 2066 aaronmk
928 3141 aaronmk
    rowcount_only = ignore and returning == None # keep NULL rows on server
929
    if rowcount_only: into = sql_gen.Table('rowcount')
930
931 3074 aaronmk
    cur = run_query_into(db, mk_insert_select(db, table, *args, **kw_args),
932
        into, recover=recover, cacheable=cacheable, log_level=log_level)
933 3141 aaronmk
    if rowcount_only: empty_temp(db, into)
934 3074 aaronmk
    autoanalyze(db, table)
935
    return cur
936 2063 aaronmk
937 2738 aaronmk
default = sql_gen.default # tells insert() to use the default value for a column
938 2066 aaronmk
939 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
940 2085 aaronmk
    '''For params, see insert_select()'''
941 5050 aaronmk
    ignore = kw_args.pop('ignore', False)
942 5094 aaronmk
    if ignore: kw_args.setdefault('recover', True)
943 5050 aaronmk
944 1960 aaronmk
    if lists.is_seq(row): cols = None
945
    else:
946
        cols = row.keys()
947
        row = row.values()
948 2738 aaronmk
    row = list(row) # ensure that "== []" works
949 1960 aaronmk
950 2738 aaronmk
    if row == []: query = None
951
    else: query = sql_gen.Values(row).to_str(db)
952 1961 aaronmk
953 5050 aaronmk
    try: return insert_select(db, table, cols, query, *args, **kw_args)
954 5057 aaronmk
    except (DuplicateKeyException, NullValueException):
955 5050 aaronmk
        if not ignore: raise
956 5163 aaronmk
        return None
957 11 aaronmk
958 3152 aaronmk
def mk_update(db, table, changes=None, cond=None, in_place=False,
959 3153 aaronmk
    cacheable_=True):
960 2402 aaronmk
    '''
961
    @param changes [(col, new_value),...]
962
        * container can be any iterable type
963
        * col: sql_gen.Code|str (for col name)
964
        * new_value: sql_gen.Code|literal value
965
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
966 3056 aaronmk
    @param in_place If set, locks the table and updates rows in place.
967
        This avoids creating dead rows in PostgreSQL.
968
        * cond must be None
969 3153 aaronmk
    @param cacheable_ Whether column structure information used to generate the
970 3152 aaronmk
        query can be cached
971 2402 aaronmk
    @return str query
972
    '''
973 3057 aaronmk
    table = sql_gen.as_Table(table)
974
    changes = [(sql_gen.to_name_only_col(c, table), sql_gen.as_Value(v))
975
        for c, v in changes]
976
977 3056 aaronmk
    if in_place:
978
        assert cond == None
979 3058 aaronmk
980 5398 aaronmk
        def col_type(col):
981
            return sql_gen.canon_type(db.col_info(
982
                sql_gen.with_default_table(c, table), cacheable_).type)
983
        changes = [(c, v, col_type(c)) for c, v in changes]
984 3065 aaronmk
        query = 'ALTER TABLE '+table.to_str(db)+'\n'
985 5396 aaronmk
        query += ',\n'.join(('ALTER COLUMN '+c.to_str(db)+' TYPE '+t+'\nUSING '
986
            +v.to_str(db) for c, v, t in changes))
987 3058 aaronmk
    else:
988
        query = 'UPDATE '+table.to_str(db)+'\nSET\n'
989
        query += ',\n'.join((c.to_str(db)+' = '+v.to_str(db)
990
            for c, v in changes))
991
        if cond != None: query += '\nWHERE\n'+cond.to_str(db)
992 3056 aaronmk
993 3266 aaronmk
    query = with_explain_comment(db, query)
994
995 2402 aaronmk
    return query
996
997 3074 aaronmk
def update(db, table, *args, **kw_args):
998 2402 aaronmk
    '''For params, see mk_update() and run_query()'''
999
    recover = kw_args.pop('recover', None)
1000 3043 aaronmk
    cacheable = kw_args.pop('cacheable', False)
1001 3030 aaronmk
    log_level = kw_args.pop('log_level', 2)
1002 2402 aaronmk
1003 3074 aaronmk
    cur = run_query(db, mk_update(db, table, *args, **kw_args), recover,
1004
        cacheable, log_level=log_level)
1005
    autoanalyze(db, table)
1006
    return cur
1007 2402 aaronmk
1008 3286 aaronmk
def mk_delete(db, table, cond=None):
1009
    '''
1010
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
1011
    @return str query
1012
    '''
1013
    query = 'DELETE FROM '+table.to_str(db)
1014
    if cond != None: query += '\nWHERE '+cond.to_str(db)
1015
1016
    query = with_explain_comment(db, query)
1017
1018
    return query
1019
1020
def delete(db, table, *args, **kw_args):
1021
    '''For params, see mk_delete() and run_query()'''
1022
    recover = kw_args.pop('recover', None)
1023 3295 aaronmk
    cacheable = kw_args.pop('cacheable', True)
1024 3286 aaronmk
    log_level = kw_args.pop('log_level', 2)
1025
1026
    cur = run_query(db, mk_delete(db, table, *args, **kw_args), recover,
1027
        cacheable, log_level=log_level)
1028
    autoanalyze(db, table)
1029
    return cur
1030
1031 135 aaronmk
def last_insert_id(db):
1032 1849 aaronmk
    module = util.root_module(db.db)
1033 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
1034
    elif module == 'MySQLdb': return db.insert_id()
1035
    else: return None
1036 13 aaronmk
1037 3490 aaronmk
def define_func(db, def_):
1038
    func = def_.function
1039
    while True:
1040
        try:
1041
            run_query(db, def_.to_str(db), recover=True, cacheable=True,
1042
                log_ignore_excs=(DuplicateException,))
1043
            break # successful
1044
        except DuplicateException:
1045 3495 aaronmk
            func.name = next_version(func.name)
1046 3490 aaronmk
            # try again with next version of name
1047
1048 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
1049 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
1050 2415 aaronmk
    to names that will be distinct among the columns' tables.
1051 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
1052 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
1053
    @param into The table for the new columns.
1054 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
1055
        columns will be included in the mapping even if they are not in cols.
1056
        The tables of the provided Col objects will be changed to into, so make
1057
        copies of them if you want to keep the original tables.
1058
    @param as_items Whether to return a list of dict items instead of a dict
1059 2383 aaronmk
    @return dict(orig_col=new_col, ...)
1060
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
1061 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
1062
        * All mappings use the into table so its name can easily be
1063 2383 aaronmk
          changed for all columns at once
1064
    '''
1065 2415 aaronmk
    cols = lists.uniqify(cols)
1066
1067 2394 aaronmk
    items = []
1068 2389 aaronmk
    for col in preserve:
1069 2390 aaronmk
        orig_col = copy.copy(col)
1070 2392 aaronmk
        col.table = into
1071 2394 aaronmk
        items.append((orig_col, col))
1072
    preserve = set(preserve)
1073
    for col in cols:
1074 2716 aaronmk
        if col not in preserve:
1075 3750 aaronmk
            items.append((col, sql_gen.Col(strings.ustr(col), into, col.srcs)))
1076 2394 aaronmk
1077
    if not as_items: items = dict(items)
1078
    return items
1079 2383 aaronmk
1080 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
1081 2391 aaronmk
    '''For params, see mk_flatten_mapping()
1082
    @return See return value of mk_flatten_mapping()
1083
    '''
1084 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
1085
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
1086 5523 aaronmk
    run_query_into(db, mk_select(db, joins, cols, limit=limit, start=start),
1087
        into=into, add_pkey_=True)
1088 3708 aaronmk
        # don't cache because the temp table will usually be truncated after use
1089 2394 aaronmk
    return dict(items)
1090 2391 aaronmk
1091 3079 aaronmk
##### Database structure introspection
1092 2414 aaronmk
1093 3079 aaronmk
#### Tables
1094
1095 4555 aaronmk
def tables(db, schema_like='public', table_like='%', exact=False,
1096
    cacheable=True):
1097 3079 aaronmk
    if exact: compare = '='
1098
    else: compare = 'LIKE'
1099
1100
    module = util.root_module(db.db)
1101
    if module == 'psycopg2':
1102
        conds = [('schemaname', sql_gen.CompareCond(schema_like, compare)),
1103
            ('tablename', sql_gen.CompareCond(table_like, compare))]
1104
        return values(select(db, 'pg_tables', ['tablename'], conds,
1105 4555 aaronmk
            order_by='tablename', cacheable=cacheable, log_level=4))
1106 3079 aaronmk
    elif module == 'MySQLdb':
1107
        return values(run_query(db, 'SHOW TABLES LIKE '+db.esc_value(table_like)
1108
            , cacheable=True, log_level=4))
1109
    else: raise NotImplementedError("Can't list tables for "+module+' database')
1110
1111 4556 aaronmk
def table_exists(db, table, cacheable=True):
1112 3079 aaronmk
    table = sql_gen.as_Table(table)
1113 4556 aaronmk
    return list(tables(db, table.schema, table.name, True, cacheable)) != []
1114 3079 aaronmk
1115 2426 aaronmk
def table_row_count(db, table, recover=None):
1116 2786 aaronmk
    return value(run_query(db, mk_select(db, table, [sql_gen.row_count],
1117 3298 aaronmk
        order_by=None), recover=recover, log_level=3))
1118 2426 aaronmk
1119 5337 aaronmk
def table_col_names(db, table, recover=None):
1120 5528 aaronmk
    return list(col_names(select(db, table, limit=0, recover=recover,
1121
        log_level=4)))
1122 2414 aaronmk
1123 5383 aaronmk
def table_cols(db, table, *args, **kw_args):
1124
    return [sql_gen.as_Col(strings.ustr(c), table)
1125
        for c in table_col_names(db, table, *args, **kw_args)]
1126
1127 5521 aaronmk
def table_pkey_index(db, table, recover=None):
1128
    table_str = sql_gen.Literal(table.to_str(db))
1129
    try:
1130
        return sql_gen.Table(value(run_query(db, '''\
1131
SELECT relname
1132
FROM pg_index
1133
JOIN pg_class index ON index.oid = indexrelid
1134
WHERE
1135
indrelid = '''+table_str.to_str(db)+'''::regclass
1136
AND indisprimary
1137
'''
1138
            , recover, cacheable=True, log_level=4)), table.schema)
1139
    except StopIteration: raise DoesNotExistException('primary key', '')
1140
1141 5389 aaronmk
def table_pkey_col(db, table, recover=None):
1142 5061 aaronmk
    table = sql_gen.as_Table(table)
1143
1144 5989 aaronmk
    module = util.root_module(db.db)
1145
    if module == 'psycopg2':
1146
        return sql_gen.Col(index_cols(db, table_pkey_index(db, table,
1147
            recover))[0], table)
1148
    else:
1149
        join_cols = ['table_schema', 'table_name', 'constraint_schema',
1150
            'constraint_name']
1151
        tables = [sql_gen.Table('key_column_usage', 'information_schema'),
1152
            sql_gen.Join(
1153
                sql_gen.Table('table_constraints', 'information_schema'),
1154
                dict(((c, sql_gen.join_same_not_null) for c in join_cols)))]
1155
        cols = [sql_gen.Col('column_name')]
1156
1157
        conds = [('constraint_type', 'PRIMARY KEY'), ('table_name', table.name)]
1158
        schema = table.schema
1159
        if schema != None: conds.append(('table_schema', schema))
1160
        order_by = 'position_in_unique_constraint'
1161
1162
        try: return sql_gen.Col(value(select(db, tables, cols, conds,
1163
            order_by=order_by, limit=1, log_level=4)), table)
1164
        except StopIteration: raise DoesNotExistException('primary key', '')
1165 5389 aaronmk
1166 5990 aaronmk
def table_has_pkey(db, table, recover=None):
1167
    try: table_pkey_col(db, table, recover)
1168
    except DoesNotExistException: return False
1169
    else: return True
1170
1171 5389 aaronmk
def pkey_name(db, table, recover=None):
1172
    '''If no pkey, returns the first column in the table.'''
1173 5392 aaronmk
    return pkey_col(db, table, recover).name
1174 832 aaronmk
1175 5390 aaronmk
def pkey_col(db, table, recover=None):
1176 5391 aaronmk
    '''If no pkey, returns the first column in the table.'''
1177 5392 aaronmk
    try: return table_pkey_col(db, table, recover)
1178 5393 aaronmk
    except DoesNotExistException: return table_cols(db, table, recover)[0]
1179 5128 aaronmk
1180 2559 aaronmk
not_null_col = 'not_null_col'
1181 2340 aaronmk
1182
def table_not_null_col(db, table, recover=None):
1183
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
1184 5337 aaronmk
    if not_null_col in table_col_names(db, table, recover): return not_null_col
1185 5388 aaronmk
    else: return pkey_name(db, table, recover)
1186 2340 aaronmk
1187 3348 aaronmk
def constraint_cond(db, constraint):
1188
    module = util.root_module(db.db)
1189
    if module == 'psycopg2':
1190
        table_str = sql_gen.Literal(constraint.table.to_str(db))
1191
        name_str = sql_gen.Literal(constraint.name)
1192
        return value(run_query(db, '''\
1193
SELECT consrc
1194
FROM pg_constraint
1195
WHERE
1196
conrelid = '''+table_str.to_str(db)+'''::regclass
1197
AND conname = '''+name_str.to_str(db)+'''
1198
'''
1199
            , cacheable=True, log_level=4))
1200 5443 aaronmk
    else: raise NotImplementedError("Can't get constraint condition for "
1201
        +module+' database')
1202 3348 aaronmk
1203 5520 aaronmk
def index_exprs(db, index):
1204 3322 aaronmk
    index = sql_gen.as_Table(index)
1205 1909 aaronmk
    module = util.root_module(db.db)
1206
    if module == 'psycopg2':
1207 3322 aaronmk
        qual_index = sql_gen.Literal(index.to_str(db))
1208 5520 aaronmk
        return list(values(run_query(db, '''\
1209 3322 aaronmk
SELECT pg_get_indexdef(indexrelid, generate_series(1, indnatts), true)
1210
FROM pg_index
1211
WHERE indexrelid = '''+qual_index.to_str(db)+'''::regclass
1212 2782 aaronmk
'''
1213
            , cacheable=True, log_level=4)))
1214 5520 aaronmk
    else: raise NotImplementedError()
1215 853 aaronmk
1216 5520 aaronmk
def index_cols(db, index):
1217
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
1218
    automatically created. When you don't know whether something is a UNIQUE
1219
    constraint or a UNIQUE index, use this function.'''
1220
    return map(sql_gen.parse_expr_col, index_exprs(db, index))
1221
1222 5445 aaronmk
def index_cond(db, index):
1223
    index = sql_gen.as_Table(index)
1224
    module = util.root_module(db.db)
1225
    if module == 'psycopg2':
1226
        qual_index = sql_gen.Literal(index.to_str(db))
1227
        return value(run_query(db, '''\
1228
SELECT pg_get_expr(indpred, indrelid, true)
1229
FROM pg_index
1230
WHERE indexrelid = '''+qual_index.to_str(db)+'''::regclass
1231
'''
1232
            , cacheable=True, log_level=4))
1233
    else: raise NotImplementedError()
1234
1235 5521 aaronmk
def index_order_by(db, index):
1236
    return sql_gen.CustomCode(', '.join(index_exprs(db, index)))
1237
1238
def table_cluster_on(db, table, recover=None):
1239
    '''
1240
    @return The table's cluster index, or its pkey if none is set
1241
    '''
1242
    table_str = sql_gen.Literal(table.to_str(db))
1243
    try:
1244
        return sql_gen.Table(value(run_query(db, '''\
1245
SELECT relname
1246
FROM pg_index
1247
JOIN pg_class index ON index.oid = indexrelid
1248
WHERE
1249
indrelid = '''+table_str.to_str(db)+'''::regclass
1250
AND indisclustered
1251
'''
1252
            , recover, cacheable=True, log_level=4)), table.schema)
1253
    except StopIteration: return table_pkey_index(db, table, recover)
1254
1255
def table_order_by(db, table, recover=None):
1256 5525 aaronmk
    if table.order_by == None:
1257
        try: table.order_by = index_order_by(db, table_cluster_on(db, table,
1258
            recover))
1259
        except DoesNotExistException: pass
1260
    return table.order_by
1261 5521 aaronmk
1262 3079 aaronmk
#### Functions
1263
1264
def function_exists(db, function):
1265 3423 aaronmk
    qual_function = sql_gen.Literal(function.to_str(db))
1266
    try:
1267 3425 aaronmk
        select(db, fields=[sql_gen.Cast('regproc', qual_function)],
1268
            recover=True, cacheable=True, log_level=4)
1269 3423 aaronmk
    except DoesNotExistException: return False
1270 4146 aaronmk
    except DuplicateException: return True # overloaded function
1271 3423 aaronmk
    else: return True
1272 3079 aaronmk
1273 5713 aaronmk
def function_param0_type(db, function):
1274
    qual_function = sql_gen.Literal(function.to_str(db))
1275
    return value(run_query(db, '''\
1276
SELECT proargtypes[0]::regtype
1277
FROM pg_proc
1278
WHERE oid = '''+qual_function.to_str(db)+'''::regproc
1279
'''
1280
        , cacheable=True, log_level=4))
1281
1282 3079 aaronmk
##### Structural changes
1283
1284
#### Columns
1285
1286 5020 aaronmk
def add_col(db, table, col, comment=None, if_not_exists=False, **kw_args):
1287 3079 aaronmk
    '''
1288
    @param col TypedCol Name may be versioned, so be sure to propagate any
1289
        renaming back to any source column for the TypedCol.
1290
    @param comment None|str SQL comment used to distinguish columns of the same
1291
        name from each other when they contain different data, to allow the
1292
        ADD COLUMN query to be cached. If not set, query will not be cached.
1293
    '''
1294
    assert isinstance(col, sql_gen.TypedCol)
1295
1296
    while True:
1297
        str_ = 'ALTER TABLE '+table.to_str(db)+' ADD COLUMN '+col.to_str(db)
1298
        if comment != None: str_ += ' '+sql_gen.esc_comment(comment)
1299
1300
        try:
1301
            run_query(db, str_, recover=True, cacheable=True, **kw_args)
1302
            break
1303
        except DuplicateException:
1304 5020 aaronmk
            if if_not_exists: raise
1305 3079 aaronmk
            col.name = next_version(col.name)
1306
            # try again with next version of name
1307
1308
def add_not_null(db, col):
1309
    table = col.table
1310
    col = sql_gen.to_name_only_col(col)
1311
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ALTER COLUMN '
1312
        +col.to_str(db)+' SET NOT NULL', cacheable=True, log_level=3)
1313
1314 4443 aaronmk
def drop_not_null(db, col):
1315
    table = col.table
1316
    col = sql_gen.to_name_only_col(col)
1317
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ALTER COLUMN '
1318
        +col.to_str(db)+' DROP NOT NULL', cacheable=True, log_level=3)
1319
1320 2096 aaronmk
row_num_col = '_row_num'
1321
1322 4997 aaronmk
row_num_col_def = sql_gen.TypedCol('', 'serial', nullable=False,
1323 3079 aaronmk
    constraints='PRIMARY KEY')
1324
1325 4997 aaronmk
def add_row_num(db, table, name=row_num_col):
1326
    '''Adds a row number column to a table. Its definition is in
1327
    row_num_col_def. It will be the primary key.'''
1328
    col_def = copy.copy(row_num_col_def)
1329
    col_def.name = name
1330 5021 aaronmk
    add_col(db, table, col_def, comment='', if_not_exists=True, log_level=3)
1331 3079 aaronmk
1332
#### Indexes
1333
1334
def add_pkey(db, table, cols=None, recover=None):
1335
    '''Adds a primary key.
1336
    @param cols [sql_gen.Col,...] The columns in the primary key.
1337
        Defaults to the first column in the table.
1338
    @pre The table must not already have a primary key.
1339
    '''
1340
    table = sql_gen.as_Table(table)
1341 5388 aaronmk
    if cols == None: cols = [pkey_name(db, table, recover)]
1342 3079 aaronmk
    col_strs = [sql_gen.to_name_only_col(v).to_str(db) for v in cols]
1343
1344
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD PRIMARY KEY ('
1345
        +(', '.join(col_strs))+')', recover=True, cacheable=True, log_level=3,
1346
        log_ignore_excs=(DuplicateException,))
1347
1348 2998 aaronmk
def add_index(db, exprs, table=None, unique=False, ensure_not_null_=True):
1349 2688 aaronmk
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
1350 3356 aaronmk
    Currently, only function calls and literal values are supported expressions.
1351 2998 aaronmk
    @param ensure_not_null_ If set, translates NULL values to sentinel values.
1352 2847 aaronmk
        This allows indexes to be used for comparisons where NULLs are equal.
1353 2538 aaronmk
    '''
1354 2964 aaronmk
    exprs = lists.mk_seq(exprs)
1355 2538 aaronmk
1356 2688 aaronmk
    # Parse exprs
1357
    old_exprs = exprs[:]
1358
    exprs = []
1359
    cols = []
1360
    for i, expr in enumerate(old_exprs):
1361 2823 aaronmk
        expr = sql_gen.as_Col(expr, table)
1362 2688 aaronmk
1363 2823 aaronmk
        # Handle nullable columns
1364 2998 aaronmk
        if ensure_not_null_:
1365 3164 aaronmk
            try: expr = sql_gen.ensure_not_null(db, expr)
1366 2860 aaronmk
            except KeyError: pass # unknown type, so just create plain index
1367 2823 aaronmk
1368 2688 aaronmk
        # Extract col
1369 3002 aaronmk
        expr = copy.deepcopy(expr) # don't modify input!
1370 3356 aaronmk
        col = expr
1371
        if isinstance(expr, sql_gen.FunctionCall): col = expr.args[0]
1372
        expr = sql_gen.cast_literal(expr)
1373
        if not isinstance(expr, (sql_gen.Expr, sql_gen.Col)):
1374 2688 aaronmk
            expr = sql_gen.Expr(expr)
1375 3356 aaronmk
1376 2688 aaronmk
1377
        # Extract table
1378
        if table == None:
1379
            assert sql_gen.is_table_col(col)
1380
            table = col.table
1381
1382 3356 aaronmk
        if isinstance(col, sql_gen.Col): col.table = None
1383 2688 aaronmk
1384
        exprs.append(expr)
1385
        cols.append(col)
1386 2408 aaronmk
1387 2688 aaronmk
    table = sql_gen.as_Table(table)
1388
1389 3005 aaronmk
    # Add index
1390 3148 aaronmk
    str_ = 'CREATE'
1391
    if unique: str_ += ' UNIQUE'
1392
    str_ += ' INDEX ON '+table.to_str(db)+' ('+(
1393
        ', '.join((v.to_str(db) for v in exprs)))+')'
1394
    run_query(db, str_, recover=True, cacheable=True, log_level=3)
1395 2408 aaronmk
1396 5987 aaronmk
def add_pkey_index(db, table): add_index(db, pkey_col(db, table), table)
1397
1398 5765 aaronmk
def add_pkey_or_index(db, table, cols=None, recover=None, warn=False):
1399
    try: add_pkey(db, table, cols, recover)
1400
    except DuplicateKeyException, e:
1401
        if warn: warnings.warn(UserWarning(exc.str_(e)))
1402 5988 aaronmk
        add_pkey_index(db, table)
1403 5765 aaronmk
1404 3083 aaronmk
already_indexed = object() # tells add_indexes() the pkey has already been added
1405
1406
def add_indexes(db, table, has_pkey=True):
1407
    '''Adds an index on all columns in a table.
1408
    @param has_pkey bool|already_indexed Whether a pkey instead of a regular
1409
        index should be added on the first column.
1410
        * If already_indexed, the pkey is assumed to have already been added
1411
    '''
1412 5337 aaronmk
    cols = table_col_names(db, table)
1413 3083 aaronmk
    if has_pkey:
1414
        if has_pkey is not already_indexed: add_pkey(db, table)
1415
        cols = cols[1:]
1416
    for col in cols: add_index(db, col, table)
1417
1418 3079 aaronmk
#### Tables
1419 2772 aaronmk
1420 3079 aaronmk
### Maintenance
1421 2772 aaronmk
1422 3079 aaronmk
def analyze(db, table):
1423
    table = sql_gen.as_Table(table)
1424
    run_query(db, 'ANALYZE '+table.to_str(db), log_level=3)
1425 2934 aaronmk
1426 3079 aaronmk
def autoanalyze(db, table):
1427
    if db.autoanalyze: analyze(db, table)
1428 2935 aaronmk
1429 3079 aaronmk
def vacuum(db, table):
1430
    table = sql_gen.as_Table(table)
1431
    db.with_autocommit(lambda: run_query(db, 'VACUUM ANALYZE '+table.to_str(db),
1432
        log_level=3))
1433 2086 aaronmk
1434 3079 aaronmk
### Lifecycle
1435
1436 3247 aaronmk
def drop(db, type_, name):
1437
    name = sql_gen.as_Name(name)
1438
    run_query(db, 'DROP '+type_+' IF EXISTS '+name.to_str(db)+' CASCADE')
1439 2889 aaronmk
1440 3247 aaronmk
def drop_table(db, table): drop(db, 'TABLE', table)
1441
1442 3082 aaronmk
def create_table(db, table, cols=[], has_pkey=True, col_indexes=True,
1443
    like=None):
1444 2675 aaronmk
    '''Creates a table.
1445 2681 aaronmk
    @param cols [sql_gen.TypedCol,...] The column names and types
1446
    @param has_pkey If set, the first column becomes the primary key.
1447 2760 aaronmk
    @param col_indexes bool|[ref]
1448
        * If True, indexes will be added on all non-pkey columns.
1449
        * If a list reference, [0] will be set to a function to do this.
1450
          This can be used to delay index creation until the table is populated.
1451 2675 aaronmk
    '''
1452
    table = sql_gen.as_Table(table)
1453
1454 3082 aaronmk
    if like != None:
1455
        cols = [sql_gen.CustomCode('LIKE '+like.to_str(db)+' INCLUDING ALL')
1456
            ]+cols
1457 5525 aaronmk
        table.order_by = like.order_by
1458 2681 aaronmk
    if has_pkey:
1459
        cols[0] = pkey = copy.copy(cols[0]) # don't modify input!
1460 2872 aaronmk
        pkey.constraints = 'PRIMARY KEY'
1461 2681 aaronmk
1462 3085 aaronmk
    temp = table.is_temp and not db.debug_temp
1463
        # temp tables permanent in debug_temp mode
1464 2760 aaronmk
1465 3085 aaronmk
    # Create table
1466 3383 aaronmk
    def create():
1467 3085 aaronmk
        str_ = 'CREATE'
1468
        if temp: str_ += ' TEMP'
1469
        str_ += ' TABLE '+table.to_str(db)+' (\n'
1470
        str_ += '\n, '.join(c.to_str(db) for c in cols)
1471 3126 aaronmk
        str_ += '\n);'
1472 3085 aaronmk
1473 3383 aaronmk
        run_query(db, str_, recover=True, cacheable=True, log_level=2,
1474
            log_ignore_excs=(DuplicateException,))
1475
    if table.is_temp:
1476
        while True:
1477
            try:
1478
                create()
1479
                break
1480
            except DuplicateException:
1481
                table.name = next_version(table.name)
1482
                # try again with next version of name
1483
    else: create()
1484 3085 aaronmk
1485 2760 aaronmk
    # Add indexes
1486 2773 aaronmk
    if has_pkey: has_pkey = already_indexed
1487
    def add_indexes_(): add_indexes(db, table, has_pkey)
1488
    if isinstance(col_indexes, list): col_indexes[0] = add_indexes_ # defer
1489
    elif col_indexes: add_indexes_() # add now
1490 2675 aaronmk
1491 3084 aaronmk
def copy_table_struct(db, src, dest):
1492
    '''Creates a structure-only copy of a table. (Does not copy data.)'''
1493 3085 aaronmk
    create_table(db, dest, has_pkey=False, col_indexes=False, like=src)
1494 3084 aaronmk
1495 5529 aaronmk
def copy_table(db, src, dest):
1496
    '''Creates a copy of a table, including data'''
1497
    copy_table_struct(db, src, dest)
1498
    insert_select(db, dest, None, mk_select(db, src))
1499
1500 3079 aaronmk
### Data
1501 2684 aaronmk
1502 2970 aaronmk
def truncate(db, table, schema='public', **kw_args):
1503
    '''For params, see run_query()'''
1504 2777 aaronmk
    table = sql_gen.as_Table(table, schema)
1505 2970 aaronmk
    return run_query(db, 'TRUNCATE '+table.to_str(db)+' CASCADE', **kw_args)
1506 2732 aaronmk
1507 2965 aaronmk
def empty_temp(db, tables):
1508
    tables = lists.mk_seq(tables)
1509 2971 aaronmk
    for table in tables: truncate(db, table, log_level=3)
1510 2965 aaronmk
1511 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
1512
    '''For kw_args, see tables()'''
1513
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
1514 3094 aaronmk
1515
def distinct_table(db, table, distinct_on):
1516
    '''Creates a copy of a temp table which is distinct on the given columns.
1517 5902 aaronmk
    Adds an index on table's distinct_on columns, to facilitate merge joins.
1518 3097 aaronmk
    @param distinct_on If empty, creates a table with one row. This is useful if
1519
        your distinct_on columns are all literal values.
1520 3099 aaronmk
    @return The new table.
1521 3094 aaronmk
    '''
1522 3099 aaronmk
    new_table = sql_gen.suffixed_table(table, '_distinct')
1523 3411 aaronmk
    distinct_on = filter(sql_gen.is_table_col, distinct_on)
1524 3094 aaronmk
1525 3099 aaronmk
    copy_table_struct(db, table, new_table)
1526 3097 aaronmk
1527
    limit = None
1528
    if distinct_on == []: limit = 1 # one sample row
1529 5903 aaronmk
    else: add_index(db, distinct_on, table) # for join optimization
1530 3097 aaronmk
1531 5903 aaronmk
    insert_select(db, new_table, None, mk_select(db, table,
1532
        distinct_on=distinct_on, order_by=None, limit=limit))
1533 3099 aaronmk
    analyze(db, new_table)
1534 3094 aaronmk
1535 3099 aaronmk
    return new_table