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 7112 aaronmk
class InvalidTypeException(ExceptionWithNameType): pass
86
87 2945 aaronmk
class DuplicateException(ExceptionWithNameType): pass
88 2143 aaronmk
89 3419 aaronmk
class DoesNotExistException(ExceptionWithNameType): pass
90
91 89 aaronmk
class EmptyRowException(DbException): pass
92
93 865 aaronmk
##### Warnings
94
95
class DbWarning(UserWarning): pass
96
97 1930 aaronmk
##### Result retrieval
98
99
def col_names(cur): return (col[0] for col in cur.description)
100
101
def rows(cur): return iter(lambda: cur.fetchone(), None)
102
103
def consume_rows(cur):
104
    '''Used to fetch all rows so result will be cached'''
105
    iters.consume_iter(rows(cur))
106
107
def next_row(cur): return rows(cur).next()
108
109
def row(cur):
110
    row_ = next_row(cur)
111
    consume_rows(cur)
112
    return row_
113
114
def next_value(cur): return next_row(cur)[0]
115
116
def value(cur): return row(cur)[0]
117
118
def values(cur): return iters.func_iter(lambda: next_value(cur))
119
120
def value_or_none(cur):
121
    try: return value(cur)
122
    except StopIteration: return None
123
124 2762 aaronmk
##### Escaping
125 2101 aaronmk
126 2573 aaronmk
def esc_name_by_module(module, name):
127
    if module == 'psycopg2' or module == None: quote = '"'
128 2101 aaronmk
    elif module == 'MySQLdb': quote = '`'
129
    else: raise NotImplementedError("Can't escape name for "+module+' database')
130 2500 aaronmk
    return sql_gen.esc_name(name, quote)
131 2101 aaronmk
132
def esc_name_by_engine(engine, name, **kw_args):
133
    return esc_name_by_module(db_engines[engine][0], name, **kw_args)
134
135
def esc_name(db, name, **kw_args):
136
    return esc_name_by_module(util.root_module(db.db), name, **kw_args)
137
138
def qual_name(db, schema, table):
139
    def esc_name_(name): return esc_name(db, name)
140
    table = esc_name_(table)
141
    if schema != None: return esc_name_(schema)+'.'+table
142
    else: return table
143
144 1869 aaronmk
##### Database connections
145 1849 aaronmk
146 2097 aaronmk
db_config_names = ['engine', 'host', 'user', 'password', 'database', 'schemas']
147 1926 aaronmk
148 1869 aaronmk
db_engines = {
149
    'MySQL': ('MySQLdb', {'password': 'passwd', 'database': 'db'}),
150
    'PostgreSQL': ('psycopg2', {}),
151
}
152
153
DatabaseErrors_set = set([DbException])
154
DatabaseErrors = tuple(DatabaseErrors_set)
155
156
def _add_module(module):
157
    DatabaseErrors_set.add(module.DatabaseError)
158
    global DatabaseErrors
159
    DatabaseErrors = tuple(DatabaseErrors_set)
160
161
def db_config_str(db_config):
162
    return db_config['engine']+' database '+db_config['database']
163
164 2448 aaronmk
log_debug_none = lambda msg, level=2: None
165 1901 aaronmk
166 1849 aaronmk
class DbConn:
167 2923 aaronmk
    def __init__(self, db_config, autocommit=True, caching=True,
168 3183 aaronmk
        log_debug=log_debug_none, debug_temp=False, src=None):
169 2915 aaronmk
        '''
170
        @param debug_temp Whether temporary objects should instead be permanent.
171
            This assists in debugging the internal objects used by the program.
172 3183 aaronmk
        @param src In autocommit mode, will be included in a comment in every
173
            query, to help identify the data source in pg_stat_activity.
174 2915 aaronmk
        '''
175 1869 aaronmk
        self.db_config = db_config
176 2190 aaronmk
        self.autocommit = autocommit
177
        self.caching = caching
178 1901 aaronmk
        self.log_debug = log_debug
179 2193 aaronmk
        self.debug = log_debug != log_debug_none
180 2915 aaronmk
        self.debug_temp = debug_temp
181 3183 aaronmk
        self.src = src
182 3074 aaronmk
        self.autoanalyze = False
183 3269 aaronmk
        self.autoexplain = False
184
        self.profile_row_ct = None
185 1869 aaronmk
186 3124 aaronmk
        self._savepoint = 0
187 3120 aaronmk
        self._reset()
188 1869 aaronmk
189
    def __getattr__(self, name):
190
        if name == '__dict__': raise Exception('getting __dict__')
191
        if name == 'db': return self._db()
192
        else: raise AttributeError()
193
194
    def __getstate__(self):
195
        state = copy.copy(self.__dict__) # shallow copy
196 1915 aaronmk
        state['log_debug'] = None # don't pickle the debug callback
197 1869 aaronmk
        state['_DbConn__db'] = None # don't pickle the connection
198
        return state
199
200 3118 aaronmk
    def clear_cache(self): self.query_results = {}
201
202 3120 aaronmk
    def _reset(self):
203 3118 aaronmk
        self.clear_cache()
204 3124 aaronmk
        assert self._savepoint == 0
205 3118 aaronmk
        self._notices_seen = set()
206
        self.__db = None
207
208 2165 aaronmk
    def connected(self): return self.__db != None
209
210 3116 aaronmk
    def close(self):
211 3119 aaronmk
        if not self.connected(): return
212
213 3135 aaronmk
        # Record that the automatic transaction is now closed
214 3136 aaronmk
        self._savepoint -= 1
215 3135 aaronmk
216 3119 aaronmk
        self.db.close()
217 3120 aaronmk
        self._reset()
218 3116 aaronmk
219 3125 aaronmk
    def reconnect(self):
220
        # Do not do this in test mode as it would roll back everything
221
        if self.autocommit: self.close()
222
        # Connection will be reopened automatically on first query
223
224 1869 aaronmk
    def _db(self):
225
        if self.__db == None:
226
            # Process db_config
227
            db_config = self.db_config.copy() # don't modify input!
228 2097 aaronmk
            schemas = db_config.pop('schemas', None)
229 1869 aaronmk
            module_name, mappings = db_engines[db_config.pop('engine')]
230
            module = __import__(module_name)
231
            _add_module(module)
232
            for orig, new in mappings.iteritems():
233
                try: util.rename_key(db_config, orig, new)
234
                except KeyError: pass
235
236
            # Connect
237
            self.__db = module.connect(**db_config)
238
239 3161 aaronmk
            # Record that a transaction is already open
240
            self._savepoint += 1
241
242 1869 aaronmk
            # Configure connection
243 2906 aaronmk
            if hasattr(self.db, 'set_isolation_level'):
244
                import psycopg2.extensions
245
                self.db.set_isolation_level(
246
                    psycopg2.extensions.ISOLATION_LEVEL_READ_COMMITTED)
247 2101 aaronmk
            if schemas != None:
248 2893 aaronmk
                search_path = [self.esc_name(s) for s in schemas.split(',')]
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 7112 aaronmk
    match = re.match(r'^.*?\brelation "(.+?)" is not a (table)', msg)
597
    if match:
598
        name, type_ = match.groups()
599
        raise InvalidTypeException(type_, name, e)
600
601 8819 aaronmk
    typed_name_re = r'^(\S+) "?(.+?)"?(?: of relation ".+?")?'
602 5577 aaronmk
603
    match = re.match(typed_name_re+r'.*? already exists', msg)
604
    if match:
605
        type_, name = match.groups()
606
        raise DuplicateException(type_, name, e)
607
608
    match = re.match(r'more than one (\S+) named ""(.+?)""', msg)
609
    if match:
610
        type_, name = match.groups()
611
        raise DuplicateException(type_, name, e)
612
613
    match = re.match(typed_name_re+r' does not exist', msg)
614
    if match:
615
        type_, name = match.groups()
616 5683 aaronmk
        if type_ == 'function':
617 5709 aaronmk
            match = re.match(r'^(.+?)\(.*\)$', name)
618 5710 aaronmk
            if match: # includes params, so is call rather than cast to regproc
619 5708 aaronmk
                function_name, = match.groups()
620 5711 aaronmk
                func = sql_gen.Function(function_name)
621 5712 aaronmk
                if function_exists(db, func) and msg.find('CAST') < 0:
622 5683 aaronmk
                    # not found only because of a missing cast
623 5715 aaronmk
                    type_ = function_param0_type(db, func)
624 6301 aaronmk
                    col = None
625 5792 aaronmk
                    if type_ == 'anyelement': type_ = 'text'
626 6301 aaronmk
                    elif type_ == 'hstore': # cast just the value param
627
                        type_ = 'text'
628
                        col = 'value'
629
                    raise MissingCastException(type_, col, e)
630 5577 aaronmk
        raise DoesNotExistException(type_, name, e)
631
632
    raise # no specific exception raised
633
634 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
635 11 aaronmk
636 2791 aaronmk
def run_query(db, query, recover=None, cacheable=False, log_level=2,
637
    log_ignore_excs=None, **kw_args):
638 2794 aaronmk
    '''For params, see DbConn.run_query()'''
639 830 aaronmk
    if recover == None: recover = False
640 2464 aaronmk
    if log_ignore_excs == None: log_ignore_excs = ()
641
    log_ignore_excs = tuple(log_ignore_excs)
642 3236 aaronmk
    debug_msg_ref = [None]
643 830 aaronmk
644 3267 aaronmk
    query = with_explain_comment(db, query)
645 3258 aaronmk
646 2148 aaronmk
    try:
647 2464 aaronmk
        try:
648 2794 aaronmk
            def run(): return db.run_query(query, cacheable, log_level,
649 2793 aaronmk
                debug_msg_ref, **kw_args)
650 2796 aaronmk
            if recover and not db.is_cached(query):
651 2464 aaronmk
                return with_savepoint(db, run)
652
            else: return run() # don't need savepoint if cached
653 5890 aaronmk
        except Exception, e:
654
            # Give failed EXPLAIN approximately the log_level of its query
655
            if query.startswith('EXPLAIN'): log_level -= 1
656
657
            parse_exception(db, e, recover)
658 2464 aaronmk
    except log_ignore_excs:
659
        log_level += 2
660
        raise
661
    finally:
662 3236 aaronmk
        if debug_msg_ref[0] != None: db.log_debug(debug_msg_ref[0], log_level)
663 830 aaronmk
664 832 aaronmk
##### Basic queries
665
666 3256 aaronmk
def is_explainable(query):
667
    # See <http://www.postgresql.org/docs/8.3/static/sql-explain.html#AEN57749>
668 3257 aaronmk
    return re.match(r'^(?:SELECT|INSERT|UPDATE|DELETE|VALUES|EXECUTE|DECLARE)\b'
669
        , query)
670 3256 aaronmk
671 3263 aaronmk
def explain(db, query, **kw_args):
672
    '''
673
    For params, see run_query().
674
    '''
675 3267 aaronmk
    kw_args.setdefault('log_level', 4)
676 3263 aaronmk
677 3750 aaronmk
    return strings.ustr(strings.join_lines(values(run_query(db,
678
        'EXPLAIN '+query, recover=True, cacheable=True, **kw_args))))
679 3256 aaronmk
        # not a higher log_level because it's useful to see what query is being
680
        # run before it's executed, which EXPLAIN effectively provides
681
682 3265 aaronmk
def has_comment(query): return query.endswith('*/')
683
684
def with_explain_comment(db, query, **kw_args):
685 3269 aaronmk
    if db.autoexplain and not has_comment(query) and is_explainable(query):
686 3265 aaronmk
        query += '\n'+sql_gen.esc_comment(' EXPLAIN:\n'
687
            +explain(db, query, **kw_args))
688
    return query
689
690 2153 aaronmk
def next_version(name):
691 2163 aaronmk
    version = 1 # first existing name was version 0
692 2586 aaronmk
    match = re.match(r'^(.*)#(\d+)$', name)
693 2153 aaronmk
    if match:
694 2586 aaronmk
        name, version = match.groups()
695
        version = int(version)+1
696 2932 aaronmk
    return sql_gen.concat(name, '#'+str(version))
697 2153 aaronmk
698 2899 aaronmk
def lock_table(db, table, mode):
699
    table = sql_gen.as_Table(table)
700
    run_query(db, 'LOCK TABLE '+table.to_str(db)+' IN '+mode+' MODE')
701
702 3303 aaronmk
def run_query_into(db, query, into=None, add_pkey_=False, **kw_args):
703 2085 aaronmk
    '''Outputs a query to a temp table.
704
    For params, see run_query().
705
    '''
706 2789 aaronmk
    if into == None: return run_query(db, query, **kw_args)
707 2790 aaronmk
708
    assert isinstance(into, sql_gen.Table)
709
710 2992 aaronmk
    into.is_temp = True
711 3008 aaronmk
    # "temporary tables cannot specify a schema name", so remove schema
712
    into.schema = None
713 2992 aaronmk
714 2790 aaronmk
    kw_args['recover'] = True
715 2945 aaronmk
    kw_args.setdefault('log_ignore_excs', (DuplicateException,))
716 2790 aaronmk
717 2916 aaronmk
    temp = not db.debug_temp # tables are permanent in debug_temp mode
718 2790 aaronmk
719
    # Create table
720
    while True:
721
        create_query = 'CREATE'
722
        if temp: create_query += ' TEMP'
723
        create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
724 2385 aaronmk
725 2790 aaronmk
        try:
726
            cur = run_query(db, create_query, **kw_args)
727
                # CREATE TABLE AS sets rowcount to # rows in query
728
            break
729 2945 aaronmk
        except DuplicateException, e:
730 2790 aaronmk
            into.name = next_version(into.name)
731
            # try again with next version of name
732
733 5967 aaronmk
    if add_pkey_: add_pkey_or_index(db, into, warn=True)
734 3075 aaronmk
735
    # According to the PostgreSQL doc, "The autovacuum daemon cannot access and
736
    # therefore cannot vacuum or analyze temporary tables. [...] if a temporary
737
    # table is going to be used in complex queries, it is wise to run ANALYZE on
738
    # the temporary table after it is populated."
739
    # (http://www.postgresql.org/docs/9.1/static/sql-createtable.html)
740
    # If into is not a temp table, ANALYZE is useful but not required.
741 3073 aaronmk
    analyze(db, into)
742 2790 aaronmk
743
    return cur
744 2085 aaronmk
745 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
746
747 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
748
749 8333 aaronmk
def has_subset_func(db, table):
750
    return sql_gen.is_underlying_table(table) and function_exists(db, table)
751
752 3420 aaronmk
def mk_select(db, tables=None, fields=None, conds=None, distinct_on=[],
753 3494 aaronmk
    limit=None, start=None, order_by=order_by_pkey, default_table=None,
754
    explain=True):
755 1981 aaronmk
    '''
756 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
757 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
758 1981 aaronmk
    @param fields Use None to select all fields in the table
759 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
760 2379 aaronmk
        * container can be any iterable type
761 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
762
        * compare_right_side: sql_gen.ValueCond|literal value
763 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
764
        use all columns
765 2786 aaronmk
    @return query
766 1981 aaronmk
    '''
767 2315 aaronmk
    # Parse tables param
768 2964 aaronmk
    tables = lists.mk_seq(tables)
769 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
770 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
771 2121 aaronmk
772 2315 aaronmk
    # Parse other params
773 2376 aaronmk
    if conds == None: conds = []
774 2650 aaronmk
    elif dicts.is_dict(conds): conds = conds.items()
775 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
776 3129 aaronmk
    assert limit == None or isinstance(limit, (int, long))
777
    assert start == None or isinstance(start, (int, long))
778 5527 aaronmk
    if limit == 0: order_by = None
779 8334 aaronmk
    if order_by in (None, order_by_pkey) and has_subset_func(db, table0):
780
        # can use subset function for fast querying at large OFFSET values
781 8335 aaronmk
        run_query(db, 'SET LOCAL enable_sort TO off')
782 8334 aaronmk
        table0 = sql_gen.FunctionCall(table0, limit_=limit, offset_=start)
783
        if limit != 0: limit = None # done by function
784
        start = None # done by function
785
        order_by = None # done by function
786
    elif order_by is order_by_pkey:
787 5641 aaronmk
        if lists.is_seq(distinct_on) and distinct_on: order_by = distinct_on[0]
788
        elif table0 != None: order_by = table_order_by(db, table0, recover=True)
789
        else: order_by = None
790 865 aaronmk
791 2315 aaronmk
    query = 'SELECT'
792 2056 aaronmk
793 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
794 2056 aaronmk
795 2200 aaronmk
    # DISTINCT ON columns
796 2233 aaronmk
    if distinct_on != []:
797 2467 aaronmk
        query += '\nDISTINCT'
798 2254 aaronmk
        if distinct_on is not distinct_on_all:
799 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
800
801
    # Columns
802 3185 aaronmk
    if query.find('\n') >= 0: whitespace = '\n'
803
    else: whitespace = ' '
804
    if fields == None: query += whitespace+'*'
805 2765 aaronmk
    else:
806
        assert fields != []
807 3185 aaronmk
        if len(fields) > 1: whitespace = '\n'
808
        query += whitespace+('\n, '.join(map(parse_col, fields)))
809 2200 aaronmk
810
    # Main table
811 3185 aaronmk
    if query.find('\n') >= 0 or len(tables) > 0: whitespace = '\n'
812
    else: whitespace = ' '
813 3420 aaronmk
    if table0 != None: query += whitespace+'FROM '+table0.to_str(db)
814 865 aaronmk
815 2122 aaronmk
    # Add joins
816 2271 aaronmk
    left_table = table0
817 2263 aaronmk
    for join_ in tables:
818
        table = join_.table
819 2238 aaronmk
820 2343 aaronmk
        # Parse special values
821
        if join_.type_ is sql_gen.filter_out: # filter no match
822 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
823 2853 aaronmk
                sql_gen.CompareCond(None, '~=')))
824 2343 aaronmk
825 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
826 2122 aaronmk
827
        left_table = table
828
829 865 aaronmk
    missing = True
830 2376 aaronmk
    if conds != []:
831 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
832
        else: whitespace = '\n'
833 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
834
            .to_str(db) for l, r in conds], 'WHERE')
835 2227 aaronmk
    if order_by != None:
836 5642 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by).to_str(db)
837 3297 aaronmk
    if limit != None: query += '\nLIMIT '+str(limit)
838 865 aaronmk
    if start != None:
839 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
840 865 aaronmk
841 3494 aaronmk
    if explain: query = with_explain_comment(db, query)
842 3266 aaronmk
843 2786 aaronmk
    return query
844 11 aaronmk
845 2054 aaronmk
def select(db, *args, **kw_args):
846
    '''For params, see mk_select() and run_query()'''
847
    recover = kw_args.pop('recover', None)
848
    cacheable = kw_args.pop('cacheable', True)
849 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
850 2054 aaronmk
851 2791 aaronmk
    return run_query(db, mk_select(db, *args, **kw_args), recover, cacheable,
852
        log_level=log_level)
853 2054 aaronmk
854 2788 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, returning=None,
855 3181 aaronmk
    embeddable=False, ignore=False, src=None):
856 1960 aaronmk
    '''
857
    @param returning str|None An inserted column (such as pkey) to return
858 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
859 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
860
        query will be fully cached, not just if it raises an exception.
861 3009 aaronmk
    @param ignore Whether to ignore duplicate keys.
862 3181 aaronmk
    @param src Will be included in the name of any created function, to help
863
        identify the data source in pg_stat_activity.
864 1960 aaronmk
    '''
865 2754 aaronmk
    table = sql_gen.remove_table_rename(sql_gen.as_Table(table))
866 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
867 3010 aaronmk
    if cols != None: cols = [sql_gen.to_name_only_col(c, table) for c in cols]
868 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
869 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
870 2063 aaronmk
871 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
872 2063 aaronmk
873 3009 aaronmk
    def mk_insert(select_query):
874
        query = first_line
875 3014 aaronmk
        if cols != None:
876
            query += '\n('+(', '.join((c.to_str(db) for c in cols)))+')'
877 3009 aaronmk
        query += '\n'+select_query
878
879
        if returning != None:
880
            returning_name_col = sql_gen.to_name_only_col(returning)
881
            query += '\nRETURNING '+returning_name_col.to_str(db)
882
883
        return query
884 2063 aaronmk
885 3489 aaronmk
    return_type = sql_gen.CustomCode('unknown')
886
    if returning != None: return_type = sql_gen.ColType(returning)
887 3017 aaronmk
888 3009 aaronmk
    if ignore:
889 3017 aaronmk
        # Always return something to set the correct rowcount
890
        if returning == None: returning = sql_gen.NamedCol('NULL', None)
891
892 3009 aaronmk
        embeddable = True # must use function
893 3010 aaronmk
894 3450 aaronmk
        if cols == None: row = [sql_gen.Col(sql_gen.all_cols, 'row')]
895
        else: row = [sql_gen.Col(c.name, 'row') for c in cols]
896 3092 aaronmk
897 3484 aaronmk
        query = sql_gen.RowExcIgnore(sql_gen.RowType(table), select_query,
898 3497 aaronmk
            sql_gen.ReturnQuery(mk_insert(sql_gen.Values(row).to_str(db))),
899
            cols)
900 3009 aaronmk
    else: query = mk_insert(select_query)
901
902 2070 aaronmk
    if embeddable:
903
        # Create function
904 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
905 3181 aaronmk
        if src != None: function_name = src+': '+function_name
906 2189 aaronmk
        while True:
907
            try:
908 3451 aaronmk
                func = db.TempFunction(function_name)
909 3489 aaronmk
                def_ = sql_gen.FunctionDef(func, sql_gen.SetOf(return_type),
910
                    query)
911 2194 aaronmk
912 3443 aaronmk
                run_query(db, def_.to_str(db), recover=True, cacheable=True,
913 2945 aaronmk
                    log_ignore_excs=(DuplicateException,))
914 2189 aaronmk
                break # this version was successful
915 2945 aaronmk
            except DuplicateException, e:
916 2189 aaronmk
                function_name = next_version(function_name)
917
                # try again with next version of name
918 2070 aaronmk
919 2337 aaronmk
        # Return query that uses function
920 3009 aaronmk
        cols = None
921
        if returning != None: cols = [returning]
922 3451 aaronmk
        func_table = sql_gen.NamedTable('f', sql_gen.FunctionCall(func), cols)
923
            # AS clause requires function alias
924 3298 aaronmk
        return mk_select(db, func_table, order_by=None)
925 2070 aaronmk
926 2787 aaronmk
    return query
927 2066 aaronmk
928 3074 aaronmk
def insert_select(db, table, *args, **kw_args):
929 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
930 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
931
        values in
932 2072 aaronmk
    '''
933 3141 aaronmk
    returning = kw_args.get('returning', None)
934
    ignore = kw_args.get('ignore', False)
935
936 2386 aaronmk
    into = kw_args.pop('into', None)
937
    if into != None: kw_args['embeddable'] = True
938 2066 aaronmk
    recover = kw_args.pop('recover', None)
939 3141 aaronmk
    if ignore: recover = True
940 2066 aaronmk
    cacheable = kw_args.pop('cacheable', True)
941 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
942 2066 aaronmk
943 3141 aaronmk
    rowcount_only = ignore and returning == None # keep NULL rows on server
944
    if rowcount_only: into = sql_gen.Table('rowcount')
945
946 3074 aaronmk
    cur = run_query_into(db, mk_insert_select(db, table, *args, **kw_args),
947
        into, recover=recover, cacheable=cacheable, log_level=log_level)
948 3141 aaronmk
    if rowcount_only: empty_temp(db, into)
949 3074 aaronmk
    autoanalyze(db, table)
950
    return cur
951 2063 aaronmk
952 2738 aaronmk
default = sql_gen.default # tells insert() to use the default value for a column
953 2066 aaronmk
954 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
955 2085 aaronmk
    '''For params, see insert_select()'''
956 5050 aaronmk
    ignore = kw_args.pop('ignore', False)
957 5094 aaronmk
    if ignore: kw_args.setdefault('recover', True)
958 5050 aaronmk
959 1960 aaronmk
    if lists.is_seq(row): cols = None
960
    else:
961
        cols = row.keys()
962
        row = row.values()
963 2738 aaronmk
    row = list(row) # ensure that "== []" works
964 1960 aaronmk
965 2738 aaronmk
    if row == []: query = None
966
    else: query = sql_gen.Values(row).to_str(db)
967 1961 aaronmk
968 5050 aaronmk
    try: return insert_select(db, table, cols, query, *args, **kw_args)
969 5057 aaronmk
    except (DuplicateKeyException, NullValueException):
970 5050 aaronmk
        if not ignore: raise
971 5163 aaronmk
        return None
972 11 aaronmk
973 3152 aaronmk
def mk_update(db, table, changes=None, cond=None, in_place=False,
974 3153 aaronmk
    cacheable_=True):
975 2402 aaronmk
    '''
976
    @param changes [(col, new_value),...]
977
        * container can be any iterable type
978
        * col: sql_gen.Code|str (for col name)
979
        * new_value: sql_gen.Code|literal value
980
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
981 3056 aaronmk
    @param in_place If set, locks the table and updates rows in place.
982
        This avoids creating dead rows in PostgreSQL.
983
        * cond must be None
984 3153 aaronmk
    @param cacheable_ Whether column structure information used to generate the
985 3152 aaronmk
        query can be cached
986 2402 aaronmk
    @return str query
987
    '''
988 3057 aaronmk
    table = sql_gen.as_Table(table)
989
    changes = [(sql_gen.to_name_only_col(c, table), sql_gen.as_Value(v))
990
        for c, v in changes]
991
992 3056 aaronmk
    if in_place:
993
        assert cond == None
994 3058 aaronmk
995 5398 aaronmk
        def col_type(col):
996
            return sql_gen.canon_type(db.col_info(
997
                sql_gen.with_default_table(c, table), cacheable_).type)
998
        changes = [(c, v, col_type(c)) for c, v in changes]
999 3065 aaronmk
        query = 'ALTER TABLE '+table.to_str(db)+'\n'
1000 5396 aaronmk
        query += ',\n'.join(('ALTER COLUMN '+c.to_str(db)+' TYPE '+t+'\nUSING '
1001
            +v.to_str(db) for c, v, t in changes))
1002 3058 aaronmk
    else:
1003
        query = 'UPDATE '+table.to_str(db)+'\nSET\n'
1004
        query += ',\n'.join((c.to_str(db)+' = '+v.to_str(db)
1005
            for c, v in changes))
1006
        if cond != None: query += '\nWHERE\n'+cond.to_str(db)
1007 3056 aaronmk
1008 3266 aaronmk
    query = with_explain_comment(db, query)
1009
1010 2402 aaronmk
    return query
1011
1012 3074 aaronmk
def update(db, table, *args, **kw_args):
1013 2402 aaronmk
    '''For params, see mk_update() and run_query()'''
1014
    recover = kw_args.pop('recover', None)
1015 3043 aaronmk
    cacheable = kw_args.pop('cacheable', False)
1016 3030 aaronmk
    log_level = kw_args.pop('log_level', 2)
1017 2402 aaronmk
1018 3074 aaronmk
    cur = run_query(db, mk_update(db, table, *args, **kw_args), recover,
1019
        cacheable, log_level=log_level)
1020
    autoanalyze(db, table)
1021
    return cur
1022 2402 aaronmk
1023 3286 aaronmk
def mk_delete(db, table, cond=None):
1024
    '''
1025
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
1026
    @return str query
1027
    '''
1028
    query = 'DELETE FROM '+table.to_str(db)
1029
    if cond != None: query += '\nWHERE '+cond.to_str(db)
1030
1031
    query = with_explain_comment(db, query)
1032
1033
    return query
1034
1035
def delete(db, table, *args, **kw_args):
1036
    '''For params, see mk_delete() and run_query()'''
1037
    recover = kw_args.pop('recover', None)
1038 3295 aaronmk
    cacheable = kw_args.pop('cacheable', True)
1039 3286 aaronmk
    log_level = kw_args.pop('log_level', 2)
1040
1041
    cur = run_query(db, mk_delete(db, table, *args, **kw_args), recover,
1042
        cacheable, log_level=log_level)
1043
    autoanalyze(db, table)
1044
    return cur
1045
1046 135 aaronmk
def last_insert_id(db):
1047 1849 aaronmk
    module = util.root_module(db.db)
1048 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
1049
    elif module == 'MySQLdb': return db.insert_id()
1050
    else: return None
1051 13 aaronmk
1052 3490 aaronmk
def define_func(db, def_):
1053
    func = def_.function
1054
    while True:
1055
        try:
1056
            run_query(db, def_.to_str(db), recover=True, cacheable=True,
1057
                log_ignore_excs=(DuplicateException,))
1058
            break # successful
1059
        except DuplicateException:
1060 3495 aaronmk
            func.name = next_version(func.name)
1061 3490 aaronmk
            # try again with next version of name
1062
1063 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
1064 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
1065 2415 aaronmk
    to names that will be distinct among the columns' tables.
1066 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
1067 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
1068
    @param into The table for the new columns.
1069 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
1070
        columns will be included in the mapping even if they are not in cols.
1071
        The tables of the provided Col objects will be changed to into, so make
1072
        copies of them if you want to keep the original tables.
1073
    @param as_items Whether to return a list of dict items instead of a dict
1074 2383 aaronmk
    @return dict(orig_col=new_col, ...)
1075
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
1076 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
1077
        * All mappings use the into table so its name can easily be
1078 2383 aaronmk
          changed for all columns at once
1079
    '''
1080 2415 aaronmk
    cols = lists.uniqify(cols)
1081
1082 2394 aaronmk
    items = []
1083 2389 aaronmk
    for col in preserve:
1084 2390 aaronmk
        orig_col = copy.copy(col)
1085 2392 aaronmk
        col.table = into
1086 2394 aaronmk
        items.append((orig_col, col))
1087
    preserve = set(preserve)
1088
    for col in cols:
1089 2716 aaronmk
        if col not in preserve:
1090 3750 aaronmk
            items.append((col, sql_gen.Col(strings.ustr(col), into, col.srcs)))
1091 2394 aaronmk
1092
    if not as_items: items = dict(items)
1093
    return items
1094 2383 aaronmk
1095 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
1096 2391 aaronmk
    '''For params, see mk_flatten_mapping()
1097
    @return See return value of mk_flatten_mapping()
1098
    '''
1099 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
1100
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
1101 5523 aaronmk
    run_query_into(db, mk_select(db, joins, cols, limit=limit, start=start),
1102
        into=into, add_pkey_=True)
1103 3708 aaronmk
        # don't cache because the temp table will usually be truncated after use
1104 2394 aaronmk
    return dict(items)
1105 2391 aaronmk
1106 3079 aaronmk
##### Database structure introspection
1107 2414 aaronmk
1108 3079 aaronmk
#### Tables
1109
1110 4555 aaronmk
def tables(db, schema_like='public', table_like='%', exact=False,
1111
    cacheable=True):
1112 3079 aaronmk
    if exact: compare = '='
1113
    else: compare = 'LIKE'
1114
1115
    module = util.root_module(db.db)
1116
    if module == 'psycopg2':
1117
        conds = [('schemaname', sql_gen.CompareCond(schema_like, compare)),
1118
            ('tablename', sql_gen.CompareCond(table_like, compare))]
1119
        return values(select(db, 'pg_tables', ['tablename'], conds,
1120 4555 aaronmk
            order_by='tablename', cacheable=cacheable, log_level=4))
1121 3079 aaronmk
    elif module == 'MySQLdb':
1122
        return values(run_query(db, 'SHOW TABLES LIKE '+db.esc_value(table_like)
1123
            , cacheable=True, log_level=4))
1124
    else: raise NotImplementedError("Can't list tables for "+module+' database')
1125
1126 4556 aaronmk
def table_exists(db, table, cacheable=True):
1127 3079 aaronmk
    table = sql_gen.as_Table(table)
1128 4556 aaronmk
    return list(tables(db, table.schema, table.name, True, cacheable)) != []
1129 3079 aaronmk
1130 2426 aaronmk
def table_row_count(db, table, recover=None):
1131 2786 aaronmk
    return value(run_query(db, mk_select(db, table, [sql_gen.row_count],
1132 3298 aaronmk
        order_by=None), recover=recover, log_level=3))
1133 2426 aaronmk
1134 5337 aaronmk
def table_col_names(db, table, recover=None):
1135 5528 aaronmk
    return list(col_names(select(db, table, limit=0, recover=recover,
1136
        log_level=4)))
1137 2414 aaronmk
1138 5383 aaronmk
def table_cols(db, table, *args, **kw_args):
1139
    return [sql_gen.as_Col(strings.ustr(c), table)
1140
        for c in table_col_names(db, table, *args, **kw_args)]
1141
1142 5521 aaronmk
def table_pkey_index(db, table, recover=None):
1143
    table_str = sql_gen.Literal(table.to_str(db))
1144
    try:
1145
        return sql_gen.Table(value(run_query(db, '''\
1146
SELECT relname
1147
FROM pg_index
1148
JOIN pg_class index ON index.oid = indexrelid
1149
WHERE
1150
indrelid = '''+table_str.to_str(db)+'''::regclass
1151
AND indisprimary
1152
'''
1153
            , recover, cacheable=True, log_level=4)), table.schema)
1154
    except StopIteration: raise DoesNotExistException('primary key', '')
1155
1156 5389 aaronmk
def table_pkey_col(db, table, recover=None):
1157 5061 aaronmk
    table = sql_gen.as_Table(table)
1158
1159 5989 aaronmk
    module = util.root_module(db.db)
1160
    if module == 'psycopg2':
1161
        return sql_gen.Col(index_cols(db, table_pkey_index(db, table,
1162
            recover))[0], table)
1163
    else:
1164
        join_cols = ['table_schema', 'table_name', 'constraint_schema',
1165
            'constraint_name']
1166
        tables = [sql_gen.Table('key_column_usage', 'information_schema'),
1167
            sql_gen.Join(
1168
                sql_gen.Table('table_constraints', 'information_schema'),
1169
                dict(((c, sql_gen.join_same_not_null) for c in join_cols)))]
1170
        cols = [sql_gen.Col('column_name')]
1171
1172
        conds = [('constraint_type', 'PRIMARY KEY'), ('table_name', table.name)]
1173
        schema = table.schema
1174
        if schema != None: conds.append(('table_schema', schema))
1175
        order_by = 'position_in_unique_constraint'
1176
1177
        try: return sql_gen.Col(value(select(db, tables, cols, conds,
1178
            order_by=order_by, limit=1, log_level=4)), table)
1179
        except StopIteration: raise DoesNotExistException('primary key', '')
1180 5389 aaronmk
1181 5990 aaronmk
def table_has_pkey(db, table, recover=None):
1182
    try: table_pkey_col(db, table, recover)
1183
    except DoesNotExistException: return False
1184
    else: return True
1185
1186 5389 aaronmk
def pkey_name(db, table, recover=None):
1187
    '''If no pkey, returns the first column in the table.'''
1188 5392 aaronmk
    return pkey_col(db, table, recover).name
1189 832 aaronmk
1190 5390 aaronmk
def pkey_col(db, table, recover=None):
1191 5391 aaronmk
    '''If no pkey, returns the first column in the table.'''
1192 5392 aaronmk
    try: return table_pkey_col(db, table, recover)
1193 5393 aaronmk
    except DoesNotExistException: return table_cols(db, table, recover)[0]
1194 5128 aaronmk
1195 2559 aaronmk
not_null_col = 'not_null_col'
1196 2340 aaronmk
1197
def table_not_null_col(db, table, recover=None):
1198
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
1199 5337 aaronmk
    if not_null_col in table_col_names(db, table, recover): return not_null_col
1200 5388 aaronmk
    else: return pkey_name(db, table, recover)
1201 2340 aaronmk
1202 3348 aaronmk
def constraint_cond(db, constraint):
1203
    module = util.root_module(db.db)
1204
    if module == 'psycopg2':
1205
        table_str = sql_gen.Literal(constraint.table.to_str(db))
1206
        name_str = sql_gen.Literal(constraint.name)
1207
        return value(run_query(db, '''\
1208
SELECT consrc
1209
FROM pg_constraint
1210
WHERE
1211
conrelid = '''+table_str.to_str(db)+'''::regclass
1212
AND conname = '''+name_str.to_str(db)+'''
1213
'''
1214
            , cacheable=True, log_level=4))
1215 5443 aaronmk
    else: raise NotImplementedError("Can't get constraint condition for "
1216
        +module+' database')
1217 3348 aaronmk
1218 5520 aaronmk
def index_exprs(db, index):
1219 3322 aaronmk
    index = sql_gen.as_Table(index)
1220 1909 aaronmk
    module = util.root_module(db.db)
1221
    if module == 'psycopg2':
1222 3322 aaronmk
        qual_index = sql_gen.Literal(index.to_str(db))
1223 5520 aaronmk
        return list(values(run_query(db, '''\
1224 3322 aaronmk
SELECT pg_get_indexdef(indexrelid, generate_series(1, indnatts), true)
1225
FROM pg_index
1226
WHERE indexrelid = '''+qual_index.to_str(db)+'''::regclass
1227 2782 aaronmk
'''
1228
            , cacheable=True, log_level=4)))
1229 5520 aaronmk
    else: raise NotImplementedError()
1230 853 aaronmk
1231 5520 aaronmk
def index_cols(db, index):
1232
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
1233
    automatically created. When you don't know whether something is a UNIQUE
1234
    constraint or a UNIQUE index, use this function.'''
1235
    return map(sql_gen.parse_expr_col, index_exprs(db, index))
1236
1237 5445 aaronmk
def index_cond(db, index):
1238
    index = sql_gen.as_Table(index)
1239
    module = util.root_module(db.db)
1240
    if module == 'psycopg2':
1241
        qual_index = sql_gen.Literal(index.to_str(db))
1242
        return value(run_query(db, '''\
1243
SELECT pg_get_expr(indpred, indrelid, true)
1244
FROM pg_index
1245
WHERE indexrelid = '''+qual_index.to_str(db)+'''::regclass
1246
'''
1247
            , cacheable=True, log_level=4))
1248
    else: raise NotImplementedError()
1249
1250 5521 aaronmk
def index_order_by(db, index):
1251
    return sql_gen.CustomCode(', '.join(index_exprs(db, index)))
1252
1253
def table_cluster_on(db, table, recover=None):
1254
    '''
1255
    @return The table's cluster index, or its pkey if none is set
1256
    '''
1257
    table_str = sql_gen.Literal(table.to_str(db))
1258
    try:
1259
        return sql_gen.Table(value(run_query(db, '''\
1260
SELECT relname
1261
FROM pg_index
1262
JOIN pg_class index ON index.oid = indexrelid
1263
WHERE
1264
indrelid = '''+table_str.to_str(db)+'''::regclass
1265
AND indisclustered
1266
'''
1267
            , recover, cacheable=True, log_level=4)), table.schema)
1268
    except StopIteration: return table_pkey_index(db, table, recover)
1269
1270
def table_order_by(db, table, recover=None):
1271 8223 aaronmk
    '''
1272
    @return None if table is view, because table_cluster_on() would return None
1273
    '''
1274 5525 aaronmk
    if table.order_by == None:
1275
        try: table.order_by = index_order_by(db, table_cluster_on(db, table,
1276
            recover))
1277
        except DoesNotExistException: pass
1278
    return table.order_by
1279 5521 aaronmk
1280 7242 aaronmk
#### Views
1281
1282
def view_exists(db, view):
1283
    view_str = sql_gen.Literal(view.to_str(db))
1284
    try:
1285
        return value(run_query(db, '''\
1286
SELECT relkind = 'v'
1287
FROM pg_class
1288
WHERE oid = '''+view_str.to_str(db)+'''::regclass
1289
'''
1290
            , cacheable=True, log_level=4))
1291
    except DoesNotExistException: return False
1292
1293 3079 aaronmk
#### Functions
1294
1295
def function_exists(db, function):
1296 3423 aaronmk
    qual_function = sql_gen.Literal(function.to_str(db))
1297
    try:
1298 3425 aaronmk
        select(db, fields=[sql_gen.Cast('regproc', qual_function)],
1299
            recover=True, cacheable=True, log_level=4)
1300 3423 aaronmk
    except DoesNotExistException: return False
1301 4146 aaronmk
    except DuplicateException: return True # overloaded function
1302 3423 aaronmk
    else: return True
1303 3079 aaronmk
1304 5713 aaronmk
def function_param0_type(db, function):
1305
    qual_function = sql_gen.Literal(function.to_str(db))
1306
    return value(run_query(db, '''\
1307
SELECT proargtypes[0]::regtype
1308
FROM pg_proc
1309
WHERE oid = '''+qual_function.to_str(db)+'''::regproc
1310
'''
1311
        , cacheable=True, log_level=4))
1312
1313 3079 aaronmk
##### Structural changes
1314
1315
#### Columns
1316
1317 5020 aaronmk
def add_col(db, table, col, comment=None, if_not_exists=False, **kw_args):
1318 3079 aaronmk
    '''
1319
    @param col TypedCol Name may be versioned, so be sure to propagate any
1320
        renaming back to any source column for the TypedCol.
1321
    @param comment None|str SQL comment used to distinguish columns of the same
1322
        name from each other when they contain different data, to allow the
1323
        ADD COLUMN query to be cached. If not set, query will not be cached.
1324
    '''
1325
    assert isinstance(col, sql_gen.TypedCol)
1326
1327
    while True:
1328
        str_ = 'ALTER TABLE '+table.to_str(db)+' ADD COLUMN '+col.to_str(db)
1329
        if comment != None: str_ += ' '+sql_gen.esc_comment(comment)
1330
1331
        try:
1332
            run_query(db, str_, recover=True, cacheable=True, **kw_args)
1333
            break
1334
        except DuplicateException:
1335 5020 aaronmk
            if if_not_exists: raise
1336 3079 aaronmk
            col.name = next_version(col.name)
1337
            # try again with next version of name
1338
1339
def add_not_null(db, col):
1340
    table = col.table
1341
    col = sql_gen.to_name_only_col(col)
1342
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ALTER COLUMN '
1343
        +col.to_str(db)+' SET NOT NULL', cacheable=True, log_level=3)
1344
1345 4443 aaronmk
def drop_not_null(db, col):
1346
    table = col.table
1347
    col = sql_gen.to_name_only_col(col)
1348
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ALTER COLUMN '
1349
        +col.to_str(db)+' DROP NOT NULL', cacheable=True, log_level=3)
1350
1351 2096 aaronmk
row_num_col = '_row_num'
1352
1353 4997 aaronmk
row_num_col_def = sql_gen.TypedCol('', 'serial', nullable=False,
1354 3079 aaronmk
    constraints='PRIMARY KEY')
1355
1356 4997 aaronmk
def add_row_num(db, table, name=row_num_col):
1357
    '''Adds a row number column to a table. Its definition is in
1358
    row_num_col_def. It will be the primary key.'''
1359
    col_def = copy.copy(row_num_col_def)
1360
    col_def.name = name
1361 5021 aaronmk
    add_col(db, table, col_def, comment='', if_not_exists=True, log_level=3)
1362 3079 aaronmk
1363
#### Indexes
1364
1365
def add_pkey(db, table, cols=None, recover=None):
1366
    '''Adds a primary key.
1367
    @param cols [sql_gen.Col,...] The columns in the primary key.
1368
        Defaults to the first column in the table.
1369
    @pre The table must not already have a primary key.
1370
    '''
1371
    table = sql_gen.as_Table(table)
1372 5388 aaronmk
    if cols == None: cols = [pkey_name(db, table, recover)]
1373 3079 aaronmk
    col_strs = [sql_gen.to_name_only_col(v).to_str(db) for v in cols]
1374
1375
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD PRIMARY KEY ('
1376
        +(', '.join(col_strs))+')', recover=True, cacheable=True, log_level=3,
1377
        log_ignore_excs=(DuplicateException,))
1378
1379 2998 aaronmk
def add_index(db, exprs, table=None, unique=False, ensure_not_null_=True):
1380 2688 aaronmk
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
1381 3356 aaronmk
    Currently, only function calls and literal values are supported expressions.
1382 2998 aaronmk
    @param ensure_not_null_ If set, translates NULL values to sentinel values.
1383 2847 aaronmk
        This allows indexes to be used for comparisons where NULLs are equal.
1384 2538 aaronmk
    '''
1385 2964 aaronmk
    exprs = lists.mk_seq(exprs)
1386 2538 aaronmk
1387 2688 aaronmk
    # Parse exprs
1388
    old_exprs = exprs[:]
1389
    exprs = []
1390
    cols = []
1391
    for i, expr in enumerate(old_exprs):
1392 2823 aaronmk
        expr = sql_gen.as_Col(expr, table)
1393 2688 aaronmk
1394 2823 aaronmk
        # Handle nullable columns
1395 2998 aaronmk
        if ensure_not_null_:
1396 3164 aaronmk
            try: expr = sql_gen.ensure_not_null(db, expr)
1397 2860 aaronmk
            except KeyError: pass # unknown type, so just create plain index
1398 2823 aaronmk
1399 2688 aaronmk
        # Extract col
1400 3002 aaronmk
        expr = copy.deepcopy(expr) # don't modify input!
1401 3356 aaronmk
        col = expr
1402
        if isinstance(expr, sql_gen.FunctionCall): col = expr.args[0]
1403
        expr = sql_gen.cast_literal(expr)
1404
        if not isinstance(expr, (sql_gen.Expr, sql_gen.Col)):
1405 2688 aaronmk
            expr = sql_gen.Expr(expr)
1406 3356 aaronmk
1407 2688 aaronmk
1408
        # Extract table
1409
        if table == None:
1410
            assert sql_gen.is_table_col(col)
1411
            table = col.table
1412
1413 3356 aaronmk
        if isinstance(col, sql_gen.Col): col.table = None
1414 2688 aaronmk
1415
        exprs.append(expr)
1416
        cols.append(col)
1417 2408 aaronmk
1418 2688 aaronmk
    table = sql_gen.as_Table(table)
1419
1420 3005 aaronmk
    # Add index
1421 3148 aaronmk
    str_ = 'CREATE'
1422
    if unique: str_ += ' UNIQUE'
1423
    str_ += ' INDEX ON '+table.to_str(db)+' ('+(
1424
        ', '.join((v.to_str(db) for v in exprs)))+')'
1425
    run_query(db, str_, recover=True, cacheable=True, log_level=3)
1426 2408 aaronmk
1427 5987 aaronmk
def add_pkey_index(db, table): add_index(db, pkey_col(db, table), table)
1428
1429 5765 aaronmk
def add_pkey_or_index(db, table, cols=None, recover=None, warn=False):
1430
    try: add_pkey(db, table, cols, recover)
1431
    except DuplicateKeyException, e:
1432
        if warn: warnings.warn(UserWarning(exc.str_(e)))
1433 5988 aaronmk
        add_pkey_index(db, table)
1434 5765 aaronmk
1435 3083 aaronmk
already_indexed = object() # tells add_indexes() the pkey has already been added
1436
1437
def add_indexes(db, table, has_pkey=True):
1438
    '''Adds an index on all columns in a table.
1439
    @param has_pkey bool|already_indexed Whether a pkey instead of a regular
1440
        index should be added on the first column.
1441
        * If already_indexed, the pkey is assumed to have already been added
1442
    '''
1443 5337 aaronmk
    cols = table_col_names(db, table)
1444 3083 aaronmk
    if has_pkey:
1445
        if has_pkey is not already_indexed: add_pkey(db, table)
1446
        cols = cols[1:]
1447
    for col in cols: add_index(db, col, table)
1448
1449 3079 aaronmk
#### Tables
1450 2772 aaronmk
1451 3079 aaronmk
### Maintenance
1452 2772 aaronmk
1453 3079 aaronmk
def analyze(db, table):
1454
    table = sql_gen.as_Table(table)
1455
    run_query(db, 'ANALYZE '+table.to_str(db), log_level=3)
1456 2934 aaronmk
1457 3079 aaronmk
def autoanalyze(db, table):
1458
    if db.autoanalyze: analyze(db, table)
1459 2935 aaronmk
1460 3079 aaronmk
def vacuum(db, table):
1461
    table = sql_gen.as_Table(table)
1462
    db.with_autocommit(lambda: run_query(db, 'VACUUM ANALYZE '+table.to_str(db),
1463
        log_level=3))
1464 2086 aaronmk
1465 3079 aaronmk
### Lifecycle
1466
1467 3247 aaronmk
def drop(db, type_, name):
1468
    name = sql_gen.as_Name(name)
1469
    run_query(db, 'DROP '+type_+' IF EXISTS '+name.to_str(db)+' CASCADE')
1470 2889 aaronmk
1471 3247 aaronmk
def drop_table(db, table): drop(db, 'TABLE', table)
1472
1473 3082 aaronmk
def create_table(db, table, cols=[], has_pkey=True, col_indexes=True,
1474
    like=None):
1475 2675 aaronmk
    '''Creates a table.
1476 2681 aaronmk
    @param cols [sql_gen.TypedCol,...] The column names and types
1477
    @param has_pkey If set, the first column becomes the primary key.
1478 2760 aaronmk
    @param col_indexes bool|[ref]
1479
        * If True, indexes will be added on all non-pkey columns.
1480
        * If a list reference, [0] will be set to a function to do this.
1481
          This can be used to delay index creation until the table is populated.
1482 2675 aaronmk
    '''
1483
    table = sql_gen.as_Table(table)
1484
1485 3082 aaronmk
    if like != None:
1486
        cols = [sql_gen.CustomCode('LIKE '+like.to_str(db)+' INCLUDING ALL')
1487
            ]+cols
1488 5525 aaronmk
        table.order_by = like.order_by
1489 2681 aaronmk
    if has_pkey:
1490
        cols[0] = pkey = copy.copy(cols[0]) # don't modify input!
1491 2872 aaronmk
        pkey.constraints = 'PRIMARY KEY'
1492 2681 aaronmk
1493 3085 aaronmk
    temp = table.is_temp and not db.debug_temp
1494
        # temp tables permanent in debug_temp mode
1495 2760 aaronmk
1496 3085 aaronmk
    # Create table
1497 3383 aaronmk
    def create():
1498 3085 aaronmk
        str_ = 'CREATE'
1499
        if temp: str_ += ' TEMP'
1500
        str_ += ' TABLE '+table.to_str(db)+' (\n'
1501
        str_ += '\n, '.join(c.to_str(db) for c in cols)
1502 3126 aaronmk
        str_ += '\n);'
1503 3085 aaronmk
1504 7113 aaronmk
        opts = dict(recover=True, cacheable=True, log_level=2,
1505 3383 aaronmk
            log_ignore_excs=(DuplicateException,))
1506 7113 aaronmk
        try: run_query(db, str_, **opts)
1507
        except InvalidTypeException: # try again as view
1508
            run_query_into(db, mk_select(db, like, limit=0), into=table, **opts)
1509 3383 aaronmk
    if table.is_temp:
1510
        while True:
1511
            try:
1512
                create()
1513
                break
1514
            except DuplicateException:
1515
                table.name = next_version(table.name)
1516
                # try again with next version of name
1517
    else: create()
1518 3085 aaronmk
1519 2760 aaronmk
    # Add indexes
1520 2773 aaronmk
    if has_pkey: has_pkey = already_indexed
1521
    def add_indexes_(): add_indexes(db, table, has_pkey)
1522
    if isinstance(col_indexes, list): col_indexes[0] = add_indexes_ # defer
1523
    elif col_indexes: add_indexes_() # add now
1524 2675 aaronmk
1525 3084 aaronmk
def copy_table_struct(db, src, dest):
1526
    '''Creates a structure-only copy of a table. (Does not copy data.)'''
1527 3085 aaronmk
    create_table(db, dest, has_pkey=False, col_indexes=False, like=src)
1528 3084 aaronmk
1529 5529 aaronmk
def copy_table(db, src, dest):
1530
    '''Creates a copy of a table, including data'''
1531
    copy_table_struct(db, src, dest)
1532
    insert_select(db, dest, None, mk_select(db, src))
1533
1534 3079 aaronmk
### Data
1535 2684 aaronmk
1536 2970 aaronmk
def truncate(db, table, schema='public', **kw_args):
1537
    '''For params, see run_query()'''
1538 2777 aaronmk
    table = sql_gen.as_Table(table, schema)
1539 2970 aaronmk
    return run_query(db, 'TRUNCATE '+table.to_str(db)+' CASCADE', **kw_args)
1540 2732 aaronmk
1541 2965 aaronmk
def empty_temp(db, tables):
1542
    tables = lists.mk_seq(tables)
1543 2971 aaronmk
    for table in tables: truncate(db, table, log_level=3)
1544 2965 aaronmk
1545 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
1546
    '''For kw_args, see tables()'''
1547
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
1548 3094 aaronmk
1549 7179 aaronmk
def distinct_table(db, table, distinct_on, joins=None):
1550 3094 aaronmk
    '''Creates a copy of a temp table which is distinct on the given columns.
1551 5902 aaronmk
    Adds an index on table's distinct_on columns, to facilitate merge joins.
1552 3097 aaronmk
    @param distinct_on If empty, creates a table with one row. This is useful if
1553
        your distinct_on columns are all literal values.
1554 7179 aaronmk
    @param joins The joins to use when creating the new table
1555 3099 aaronmk
    @return The new table.
1556 3094 aaronmk
    '''
1557 7179 aaronmk
    if joins == None: joins = [table]
1558
1559 3099 aaronmk
    new_table = sql_gen.suffixed_table(table, '_distinct')
1560 3411 aaronmk
    distinct_on = filter(sql_gen.is_table_col, distinct_on)
1561 3094 aaronmk
1562 3099 aaronmk
    copy_table_struct(db, table, new_table)
1563 3097 aaronmk
1564
    limit = None
1565
    if distinct_on == []: limit = 1 # one sample row
1566 5903 aaronmk
    else: add_index(db, distinct_on, table) # for join optimization
1567 3097 aaronmk
1568 7179 aaronmk
    insert_select(db, new_table, None, mk_select(db, joins,
1569
        [sql_gen.Col(sql_gen.all_cols, table)], distinct_on=distinct_on,
1570
        order_by=None, limit=limit))
1571 3099 aaronmk
    analyze(db, new_table)
1572 3094 aaronmk
1573 3099 aaronmk
    return new_table