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 9757 aaronmk
            except DatabaseErrors, e2: # e.g. "current transaction is aborted"
544
                exc.print_ex(e2)
545
                raise e
546 5763 aaronmk
        return DuplicateKeyException(constraint, cond, cols, e)
547 5577 aaronmk
548 5763 aaronmk
    match = re.match(r'^duplicate key value violates unique constraint "(.+?)"',
549
        msg)
550
    if match:
551
        constraint, = match.groups()
552
        raise make_DuplicateKeyException(constraint, e)
553
554 5764 aaronmk
    match = re.match(r'^could not create unique index "(.+?)"\n'
555
        r'DETAIL:  Key .+? is duplicated', msg)
556
    if match:
557
        constraint, = match.groups()
558 5823 aaronmk
        raise DuplicateKeyException(constraint, None, [], e)
559 5764 aaronmk
560 5577 aaronmk
    match = re.match(r'^null value in column "(.+?)" violates not-null'
561
        r' constraint', msg)
562
    if match:
563
        col, = match.groups()
564
        raise NullValueException('NOT NULL', None, [col], e)
565
566
    match = re.match(r'^new row for relation "(.+?)" violates check '
567
        r'constraint "(.+?)"', msg)
568
    if match:
569
        table, constraint = match.groups()
570
        constraint = sql_gen.Col(constraint, table)
571
        cond = None
572
        if recover: # need auto-rollback to run constraint_cond()
573
            try: cond = constraint_cond(db, constraint)
574
            except NotImplementedError: pass
575
        raise CheckException(constraint.to_str(db), cond, [], e)
576
577 5723 aaronmk
    match = re.match(r'^(?:invalid input (?:syntax|value)\b[^:]*'
578 5717 aaronmk
        r'|.+? out of range)(?:: "(.+?)")?', msg)
579 5577 aaronmk
    if match:
580
        value, = match.groups()
581 5717 aaronmk
        value = util.do_ignore_none(strings.to_unicode, value)
582
        raise InvalidValueException(value, e)
583 5577 aaronmk
584
    match = re.match(r'^column "(.+?)" is of type (.+?) but expression '
585
        r'is of type', msg)
586
    if match:
587
        col, type_ = match.groups()
588
        raise MissingCastException(type_, col, e)
589
590
    match = re.match(r'^could not determine polymorphic type because '
591
        r'input has type "unknown"', msg)
592
    if match: raise MissingCastException('text', None, e)
593
594 5724 aaronmk
    match = re.match(r'^.+? types (.+?) and (.+?) cannot be matched', msg)
595
    if match:
596
        type0, type1 = match.groups()
597
        raise MissingCastException(type0, None, e)
598 5577 aaronmk
599 7112 aaronmk
    match = re.match(r'^.*?\brelation "(.+?)" is not a (table)', msg)
600
    if match:
601
        name, type_ = match.groups()
602
        raise InvalidTypeException(type_, name, e)
603
604 9458 aaronmk
    typed_name_re = r'^(\S+) ("?)(.+?)\2(?: of relation ".+?")?'
605 9457 aaronmk
        # regexp must be followed with text for .*? to match properly
606 5577 aaronmk
607 9458 aaronmk
    match = re.match(typed_name_re+r'(?:\(.*?)? already exists', msg)
608 5577 aaronmk
    if match:
609 9458 aaronmk
        type_, quote, name = match.groups()
610 5577 aaronmk
        raise DuplicateException(type_, name, e)
611
612
    match = re.match(r'more than one (\S+) named ""(.+?)""', msg)
613
    if match:
614
        type_, name = match.groups()
615
        raise DuplicateException(type_, name, e)
616
617
    match = re.match(typed_name_re+r' does not exist', msg)
618
    if match:
619 9458 aaronmk
        type_, quote, name = match.groups()
620 5683 aaronmk
        if type_ == 'function':
621 5709 aaronmk
            match = re.match(r'^(.+?)\(.*\)$', name)
622 5710 aaronmk
            if match: # includes params, so is call rather than cast to regproc
623 5708 aaronmk
                function_name, = match.groups()
624 5711 aaronmk
                func = sql_gen.Function(function_name)
625 5712 aaronmk
                if function_exists(db, func) and msg.find('CAST') < 0:
626 5683 aaronmk
                    # not found only because of a missing cast
627 5715 aaronmk
                    type_ = function_param0_type(db, func)
628 6301 aaronmk
                    col = None
629 5792 aaronmk
                    if type_ == 'anyelement': type_ = 'text'
630 6301 aaronmk
                    elif type_ == 'hstore': # cast just the value param
631
                        type_ = 'text'
632
                        col = 'value'
633
                    raise MissingCastException(type_, col, e)
634 5577 aaronmk
        raise DoesNotExistException(type_, name, e)
635
636
    raise # no specific exception raised
637
638 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
639 11 aaronmk
640 2791 aaronmk
def run_query(db, query, recover=None, cacheable=False, log_level=2,
641
    log_ignore_excs=None, **kw_args):
642 2794 aaronmk
    '''For params, see DbConn.run_query()'''
643 830 aaronmk
    if recover == None: recover = False
644 2464 aaronmk
    if log_ignore_excs == None: log_ignore_excs = ()
645
    log_ignore_excs = tuple(log_ignore_excs)
646 3236 aaronmk
    debug_msg_ref = [None]
647 830 aaronmk
648 3267 aaronmk
    query = with_explain_comment(db, query)
649 3258 aaronmk
650 2148 aaronmk
    try:
651 2464 aaronmk
        try:
652 2794 aaronmk
            def run(): return db.run_query(query, cacheable, log_level,
653 2793 aaronmk
                debug_msg_ref, **kw_args)
654 2796 aaronmk
            if recover and not db.is_cached(query):
655 2464 aaronmk
                return with_savepoint(db, run)
656
            else: return run() # don't need savepoint if cached
657 5890 aaronmk
        except Exception, e:
658
            # Give failed EXPLAIN approximately the log_level of its query
659
            if query.startswith('EXPLAIN'): log_level -= 1
660
661
            parse_exception(db, e, recover)
662 2464 aaronmk
    except log_ignore_excs:
663
        log_level += 2
664
        raise
665
    finally:
666 3236 aaronmk
        if debug_msg_ref[0] != None: db.log_debug(debug_msg_ref[0], log_level)
667 830 aaronmk
668 832 aaronmk
##### Basic queries
669
670 3256 aaronmk
def is_explainable(query):
671
    # See <http://www.postgresql.org/docs/8.3/static/sql-explain.html#AEN57749>
672 3257 aaronmk
    return re.match(r'^(?:SELECT|INSERT|UPDATE|DELETE|VALUES|EXECUTE|DECLARE)\b'
673
        , query)
674 3256 aaronmk
675 3263 aaronmk
def explain(db, query, **kw_args):
676
    '''
677
    For params, see run_query().
678
    '''
679 3267 aaronmk
    kw_args.setdefault('log_level', 4)
680 3263 aaronmk
681 3750 aaronmk
    return strings.ustr(strings.join_lines(values(run_query(db,
682
        'EXPLAIN '+query, recover=True, cacheable=True, **kw_args))))
683 3256 aaronmk
        # not a higher log_level because it's useful to see what query is being
684
        # run before it's executed, which EXPLAIN effectively provides
685
686 3265 aaronmk
def has_comment(query): return query.endswith('*/')
687
688
def with_explain_comment(db, query, **kw_args):
689 3269 aaronmk
    if db.autoexplain and not has_comment(query) and is_explainable(query):
690 3265 aaronmk
        query += '\n'+sql_gen.esc_comment(' EXPLAIN:\n'
691
            +explain(db, query, **kw_args))
692
    return query
693
694 2153 aaronmk
def next_version(name):
695 2163 aaronmk
    version = 1 # first existing name was version 0
696 2586 aaronmk
    match = re.match(r'^(.*)#(\d+)$', name)
697 2153 aaronmk
    if match:
698 2586 aaronmk
        name, version = match.groups()
699
        version = int(version)+1
700 2932 aaronmk
    return sql_gen.concat(name, '#'+str(version))
701 2153 aaronmk
702 2899 aaronmk
def lock_table(db, table, mode):
703
    table = sql_gen.as_Table(table)
704
    run_query(db, 'LOCK TABLE '+table.to_str(db)+' IN '+mode+' MODE')
705
706 3303 aaronmk
def run_query_into(db, query, into=None, add_pkey_=False, **kw_args):
707 2085 aaronmk
    '''Outputs a query to a temp table.
708
    For params, see run_query().
709
    '''
710 2789 aaronmk
    if into == None: return run_query(db, query, **kw_args)
711 2790 aaronmk
712
    assert isinstance(into, sql_gen.Table)
713
714 2992 aaronmk
    into.is_temp = True
715 3008 aaronmk
    # "temporary tables cannot specify a schema name", so remove schema
716
    into.schema = None
717 2992 aaronmk
718 2790 aaronmk
    kw_args['recover'] = True
719 2945 aaronmk
    kw_args.setdefault('log_ignore_excs', (DuplicateException,))
720 2790 aaronmk
721 2916 aaronmk
    temp = not db.debug_temp # tables are permanent in debug_temp mode
722 2790 aaronmk
723
    # Create table
724
    while True:
725
        create_query = 'CREATE'
726
        if temp: create_query += ' TEMP'
727
        create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
728 2385 aaronmk
729 2790 aaronmk
        try:
730
            cur = run_query(db, create_query, **kw_args)
731
                # CREATE TABLE AS sets rowcount to # rows in query
732
            break
733 2945 aaronmk
        except DuplicateException, e:
734 2790 aaronmk
            into.name = next_version(into.name)
735
            # try again with next version of name
736
737 5967 aaronmk
    if add_pkey_: add_pkey_or_index(db, into, warn=True)
738 3075 aaronmk
739
    # According to the PostgreSQL doc, "The autovacuum daemon cannot access and
740
    # therefore cannot vacuum or analyze temporary tables. [...] if a temporary
741
    # table is going to be used in complex queries, it is wise to run ANALYZE on
742
    # the temporary table after it is populated."
743
    # (http://www.postgresql.org/docs/9.1/static/sql-createtable.html)
744
    # If into is not a temp table, ANALYZE is useful but not required.
745 3073 aaronmk
    analyze(db, into)
746 2790 aaronmk
747
    return cur
748 2085 aaronmk
749 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
750
751 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
752
753 8333 aaronmk
def has_subset_func(db, table):
754
    return sql_gen.is_underlying_table(table) and function_exists(db, table)
755
756 3420 aaronmk
def mk_select(db, tables=None, fields=None, conds=None, distinct_on=[],
757 3494 aaronmk
    limit=None, start=None, order_by=order_by_pkey, default_table=None,
758
    explain=True):
759 1981 aaronmk
    '''
760 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
761 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
762 1981 aaronmk
    @param fields Use None to select all fields in the table
763 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
764 2379 aaronmk
        * container can be any iterable type
765 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
766
        * compare_right_side: sql_gen.ValueCond|literal value
767 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
768
        use all columns
769 2786 aaronmk
    @return query
770 1981 aaronmk
    '''
771 2315 aaronmk
    # Parse tables param
772 2964 aaronmk
    tables = lists.mk_seq(tables)
773 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
774 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
775 2121 aaronmk
776 2315 aaronmk
    # Parse other params
777 2376 aaronmk
    if conds == None: conds = []
778 2650 aaronmk
    elif dicts.is_dict(conds): conds = conds.items()
779 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
780 3129 aaronmk
    assert limit == None or isinstance(limit, (int, long))
781
    assert start == None or isinstance(start, (int, long))
782 5527 aaronmk
    if limit == 0: order_by = None
783 8334 aaronmk
    if order_by in (None, order_by_pkey) and has_subset_func(db, table0):
784
        # can use subset function for fast querying at large OFFSET values
785 8335 aaronmk
        run_query(db, 'SET LOCAL enable_sort TO off')
786 8334 aaronmk
        table0 = sql_gen.FunctionCall(table0, limit_=limit, offset_=start)
787
        if limit != 0: limit = None # done by function
788
        start = None # done by function
789
        order_by = None # done by function
790
    elif order_by is order_by_pkey:
791 5641 aaronmk
        if lists.is_seq(distinct_on) and distinct_on: order_by = distinct_on[0]
792
        elif table0 != None: order_by = table_order_by(db, table0, recover=True)
793
        else: order_by = None
794 865 aaronmk
795 2315 aaronmk
    query = 'SELECT'
796 2056 aaronmk
797 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
798 2056 aaronmk
799 2200 aaronmk
    # DISTINCT ON columns
800 2233 aaronmk
    if distinct_on != []:
801 2467 aaronmk
        query += '\nDISTINCT'
802 2254 aaronmk
        if distinct_on is not distinct_on_all:
803 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
804
805
    # Columns
806 3185 aaronmk
    if query.find('\n') >= 0: whitespace = '\n'
807
    else: whitespace = ' '
808
    if fields == None: query += whitespace+'*'
809 2765 aaronmk
    else:
810
        assert fields != []
811 3185 aaronmk
        if len(fields) > 1: whitespace = '\n'
812
        query += whitespace+('\n, '.join(map(parse_col, fields)))
813 2200 aaronmk
814
    # Main table
815 3185 aaronmk
    if query.find('\n') >= 0 or len(tables) > 0: whitespace = '\n'
816
    else: whitespace = ' '
817 3420 aaronmk
    if table0 != None: query += whitespace+'FROM '+table0.to_str(db)
818 865 aaronmk
819 2122 aaronmk
    # Add joins
820 2271 aaronmk
    left_table = table0
821 2263 aaronmk
    for join_ in tables:
822
        table = join_.table
823 2238 aaronmk
824 2343 aaronmk
        # Parse special values
825
        if join_.type_ is sql_gen.filter_out: # filter no match
826 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
827 2853 aaronmk
                sql_gen.CompareCond(None, '~=')))
828 2343 aaronmk
829 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
830 2122 aaronmk
831
        left_table = table
832
833 865 aaronmk
    missing = True
834 2376 aaronmk
    if conds != []:
835 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
836
        else: whitespace = '\n'
837 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
838
            .to_str(db) for l, r in conds], 'WHERE')
839 2227 aaronmk
    if order_by != None:
840 5642 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by).to_str(db)
841 3297 aaronmk
    if limit != None: query += '\nLIMIT '+str(limit)
842 865 aaronmk
    if start != None:
843 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
844 865 aaronmk
845 3494 aaronmk
    if explain: query = with_explain_comment(db, query)
846 3266 aaronmk
847 2786 aaronmk
    return query
848 11 aaronmk
849 2054 aaronmk
def select(db, *args, **kw_args):
850
    '''For params, see mk_select() and run_query()'''
851
    recover = kw_args.pop('recover', None)
852
    cacheable = kw_args.pop('cacheable', True)
853 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
854 2054 aaronmk
855 2791 aaronmk
    return run_query(db, mk_select(db, *args, **kw_args), recover, cacheable,
856
        log_level=log_level)
857 2054 aaronmk
858 2788 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, returning=None,
859 3181 aaronmk
    embeddable=False, ignore=False, src=None):
860 1960 aaronmk
    '''
861
    @param returning str|None An inserted column (such as pkey) to return
862 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
863 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
864
        query will be fully cached, not just if it raises an exception.
865 3009 aaronmk
    @param ignore Whether to ignore duplicate keys.
866 3181 aaronmk
    @param src Will be included in the name of any created function, to help
867
        identify the data source in pg_stat_activity.
868 1960 aaronmk
    '''
869 2754 aaronmk
    table = sql_gen.remove_table_rename(sql_gen.as_Table(table))
870 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
871 3010 aaronmk
    if cols != None: cols = [sql_gen.to_name_only_col(c, table) for c in cols]
872 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
873 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
874 2063 aaronmk
875 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
876 2063 aaronmk
877 3009 aaronmk
    def mk_insert(select_query):
878
        query = first_line
879 3014 aaronmk
        if cols != None:
880
            query += '\n('+(', '.join((c.to_str(db) for c in cols)))+')'
881 3009 aaronmk
        query += '\n'+select_query
882
883
        if returning != None:
884
            returning_name_col = sql_gen.to_name_only_col(returning)
885
            query += '\nRETURNING '+returning_name_col.to_str(db)
886
887
        return query
888 2063 aaronmk
889 3489 aaronmk
    return_type = sql_gen.CustomCode('unknown')
890
    if returning != None: return_type = sql_gen.ColType(returning)
891 3017 aaronmk
892 3009 aaronmk
    if ignore:
893 3017 aaronmk
        # Always return something to set the correct rowcount
894
        if returning == None: returning = sql_gen.NamedCol('NULL', None)
895
896 3009 aaronmk
        embeddable = True # must use function
897 3010 aaronmk
898 3450 aaronmk
        if cols == None: row = [sql_gen.Col(sql_gen.all_cols, 'row')]
899
        else: row = [sql_gen.Col(c.name, 'row') for c in cols]
900 3092 aaronmk
901 3484 aaronmk
        query = sql_gen.RowExcIgnore(sql_gen.RowType(table), select_query,
902 3497 aaronmk
            sql_gen.ReturnQuery(mk_insert(sql_gen.Values(row).to_str(db))),
903
            cols)
904 3009 aaronmk
    else: query = mk_insert(select_query)
905
906 2070 aaronmk
    if embeddable:
907
        # Create function
908 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
909 3181 aaronmk
        if src != None: function_name = src+': '+function_name
910 2189 aaronmk
        while True:
911
            try:
912 3451 aaronmk
                func = db.TempFunction(function_name)
913 3489 aaronmk
                def_ = sql_gen.FunctionDef(func, sql_gen.SetOf(return_type),
914
                    query)
915 2194 aaronmk
916 3443 aaronmk
                run_query(db, def_.to_str(db), recover=True, cacheable=True,
917 2945 aaronmk
                    log_ignore_excs=(DuplicateException,))
918 2189 aaronmk
                break # this version was successful
919 2945 aaronmk
            except DuplicateException, e:
920 2189 aaronmk
                function_name = next_version(function_name)
921
                # try again with next version of name
922 2070 aaronmk
923 2337 aaronmk
        # Return query that uses function
924 3009 aaronmk
        cols = None
925
        if returning != None: cols = [returning]
926 3451 aaronmk
        func_table = sql_gen.NamedTable('f', sql_gen.FunctionCall(func), cols)
927
            # AS clause requires function alias
928 3298 aaronmk
        return mk_select(db, func_table, order_by=None)
929 2070 aaronmk
930 2787 aaronmk
    return query
931 2066 aaronmk
932 3074 aaronmk
def insert_select(db, table, *args, **kw_args):
933 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
934 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
935
        values in
936 2072 aaronmk
    '''
937 3141 aaronmk
    returning = kw_args.get('returning', None)
938
    ignore = kw_args.get('ignore', False)
939
940 2386 aaronmk
    into = kw_args.pop('into', None)
941
    if into != None: kw_args['embeddable'] = True
942 2066 aaronmk
    recover = kw_args.pop('recover', None)
943 3141 aaronmk
    if ignore: recover = True
944 2066 aaronmk
    cacheable = kw_args.pop('cacheable', True)
945 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
946 2066 aaronmk
947 3141 aaronmk
    rowcount_only = ignore and returning == None # keep NULL rows on server
948
    if rowcount_only: into = sql_gen.Table('rowcount')
949
950 3074 aaronmk
    cur = run_query_into(db, mk_insert_select(db, table, *args, **kw_args),
951
        into, recover=recover, cacheable=cacheable, log_level=log_level)
952 3141 aaronmk
    if rowcount_only: empty_temp(db, into)
953 3074 aaronmk
    autoanalyze(db, table)
954
    return cur
955 2063 aaronmk
956 2738 aaronmk
default = sql_gen.default # tells insert() to use the default value for a column
957 2066 aaronmk
958 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
959 2085 aaronmk
    '''For params, see insert_select()'''
960 5050 aaronmk
    ignore = kw_args.pop('ignore', False)
961 5094 aaronmk
    if ignore: kw_args.setdefault('recover', True)
962 5050 aaronmk
963 1960 aaronmk
    if lists.is_seq(row): cols = None
964
    else:
965
        cols = row.keys()
966
        row = row.values()
967 2738 aaronmk
    row = list(row) # ensure that "== []" works
968 1960 aaronmk
969 2738 aaronmk
    if row == []: query = None
970
    else: query = sql_gen.Values(row).to_str(db)
971 1961 aaronmk
972 5050 aaronmk
    try: return insert_select(db, table, cols, query, *args, **kw_args)
973 5057 aaronmk
    except (DuplicateKeyException, NullValueException):
974 5050 aaronmk
        if not ignore: raise
975 5163 aaronmk
        return None
976 11 aaronmk
977 3152 aaronmk
def mk_update(db, table, changes=None, cond=None, in_place=False,
978 3153 aaronmk
    cacheable_=True):
979 2402 aaronmk
    '''
980
    @param changes [(col, new_value),...]
981
        * container can be any iterable type
982
        * col: sql_gen.Code|str (for col name)
983
        * new_value: sql_gen.Code|literal value
984
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
985 3056 aaronmk
    @param in_place If set, locks the table and updates rows in place.
986
        This avoids creating dead rows in PostgreSQL.
987
        * cond must be None
988 3153 aaronmk
    @param cacheable_ Whether column structure information used to generate the
989 3152 aaronmk
        query can be cached
990 2402 aaronmk
    @return str query
991
    '''
992 3057 aaronmk
    table = sql_gen.as_Table(table)
993
    changes = [(sql_gen.to_name_only_col(c, table), sql_gen.as_Value(v))
994
        for c, v in changes]
995
996 3056 aaronmk
    if in_place:
997
        assert cond == None
998 3058 aaronmk
999 5398 aaronmk
        def col_type(col):
1000
            return sql_gen.canon_type(db.col_info(
1001
                sql_gen.with_default_table(c, table), cacheable_).type)
1002
        changes = [(c, v, col_type(c)) for c, v in changes]
1003 3065 aaronmk
        query = 'ALTER TABLE '+table.to_str(db)+'\n'
1004 5396 aaronmk
        query += ',\n'.join(('ALTER COLUMN '+c.to_str(db)+' TYPE '+t+'\nUSING '
1005
            +v.to_str(db) for c, v, t in changes))
1006 3058 aaronmk
    else:
1007
        query = 'UPDATE '+table.to_str(db)+'\nSET\n'
1008
        query += ',\n'.join((c.to_str(db)+' = '+v.to_str(db)
1009
            for c, v in changes))
1010
        if cond != None: query += '\nWHERE\n'+cond.to_str(db)
1011 3056 aaronmk
1012 3266 aaronmk
    query = with_explain_comment(db, query)
1013
1014 2402 aaronmk
    return query
1015
1016 3074 aaronmk
def update(db, table, *args, **kw_args):
1017 2402 aaronmk
    '''For params, see mk_update() and run_query()'''
1018
    recover = kw_args.pop('recover', None)
1019 3043 aaronmk
    cacheable = kw_args.pop('cacheable', False)
1020 3030 aaronmk
    log_level = kw_args.pop('log_level', 2)
1021 2402 aaronmk
1022 3074 aaronmk
    cur = run_query(db, mk_update(db, table, *args, **kw_args), recover,
1023
        cacheable, log_level=log_level)
1024
    autoanalyze(db, table)
1025
    return cur
1026 2402 aaronmk
1027 3286 aaronmk
def mk_delete(db, table, cond=None):
1028
    '''
1029
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
1030
    @return str query
1031
    '''
1032
    query = 'DELETE FROM '+table.to_str(db)
1033
    if cond != None: query += '\nWHERE '+cond.to_str(db)
1034
1035
    query = with_explain_comment(db, query)
1036
1037
    return query
1038
1039
def delete(db, table, *args, **kw_args):
1040
    '''For params, see mk_delete() and run_query()'''
1041
    recover = kw_args.pop('recover', None)
1042 3295 aaronmk
    cacheable = kw_args.pop('cacheable', True)
1043 3286 aaronmk
    log_level = kw_args.pop('log_level', 2)
1044
1045
    cur = run_query(db, mk_delete(db, table, *args, **kw_args), recover,
1046
        cacheable, log_level=log_level)
1047
    autoanalyze(db, table)
1048
    return cur
1049
1050 135 aaronmk
def last_insert_id(db):
1051 1849 aaronmk
    module = util.root_module(db.db)
1052 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
1053
    elif module == 'MySQLdb': return db.insert_id()
1054
    else: return None
1055 13 aaronmk
1056 3490 aaronmk
def define_func(db, def_):
1057
    func = def_.function
1058
    while True:
1059
        try:
1060
            run_query(db, def_.to_str(db), recover=True, cacheable=True,
1061
                log_ignore_excs=(DuplicateException,))
1062
            break # successful
1063
        except DuplicateException:
1064 3495 aaronmk
            func.name = next_version(func.name)
1065 3490 aaronmk
            # try again with next version of name
1066
1067 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
1068 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
1069 2415 aaronmk
    to names that will be distinct among the columns' tables.
1070 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
1071 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
1072
    @param into The table for the new columns.
1073 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
1074
        columns will be included in the mapping even if they are not in cols.
1075
        The tables of the provided Col objects will be changed to into, so make
1076
        copies of them if you want to keep the original tables.
1077
    @param as_items Whether to return a list of dict items instead of a dict
1078 2383 aaronmk
    @return dict(orig_col=new_col, ...)
1079
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
1080 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
1081
        * All mappings use the into table so its name can easily be
1082 2383 aaronmk
          changed for all columns at once
1083
    '''
1084 2415 aaronmk
    cols = lists.uniqify(cols)
1085
1086 2394 aaronmk
    items = []
1087 2389 aaronmk
    for col in preserve:
1088 2390 aaronmk
        orig_col = copy.copy(col)
1089 2392 aaronmk
        col.table = into
1090 2394 aaronmk
        items.append((orig_col, col))
1091
    preserve = set(preserve)
1092
    for col in cols:
1093 2716 aaronmk
        if col not in preserve:
1094 3750 aaronmk
            items.append((col, sql_gen.Col(strings.ustr(col), into, col.srcs)))
1095 2394 aaronmk
1096
    if not as_items: items = dict(items)
1097
    return items
1098 2383 aaronmk
1099 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
1100 2391 aaronmk
    '''For params, see mk_flatten_mapping()
1101
    @return See return value of mk_flatten_mapping()
1102
    '''
1103 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
1104
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
1105 5523 aaronmk
    run_query_into(db, mk_select(db, joins, cols, limit=limit, start=start),
1106
        into=into, add_pkey_=True)
1107 3708 aaronmk
        # don't cache because the temp table will usually be truncated after use
1108 2394 aaronmk
    return dict(items)
1109 2391 aaronmk
1110 3079 aaronmk
##### Database structure introspection
1111 2414 aaronmk
1112 3079 aaronmk
#### Tables
1113
1114 4555 aaronmk
def tables(db, schema_like='public', table_like='%', exact=False,
1115
    cacheable=True):
1116 3079 aaronmk
    if exact: compare = '='
1117
    else: compare = 'LIKE'
1118
1119
    module = util.root_module(db.db)
1120
    if module == 'psycopg2':
1121
        conds = [('schemaname', sql_gen.CompareCond(schema_like, compare)),
1122
            ('tablename', sql_gen.CompareCond(table_like, compare))]
1123
        return values(select(db, 'pg_tables', ['tablename'], conds,
1124 4555 aaronmk
            order_by='tablename', cacheable=cacheable, log_level=4))
1125 3079 aaronmk
    elif module == 'MySQLdb':
1126
        return values(run_query(db, 'SHOW TABLES LIKE '+db.esc_value(table_like)
1127
            , cacheable=True, log_level=4))
1128
    else: raise NotImplementedError("Can't list tables for "+module+' database')
1129
1130 4556 aaronmk
def table_exists(db, table, cacheable=True):
1131 3079 aaronmk
    table = sql_gen.as_Table(table)
1132 4556 aaronmk
    return list(tables(db, table.schema, table.name, True, cacheable)) != []
1133 3079 aaronmk
1134 2426 aaronmk
def table_row_count(db, table, recover=None):
1135 2786 aaronmk
    return value(run_query(db, mk_select(db, table, [sql_gen.row_count],
1136 3298 aaronmk
        order_by=None), recover=recover, log_level=3))
1137 2426 aaronmk
1138 5337 aaronmk
def table_col_names(db, table, recover=None):
1139 5528 aaronmk
    return list(col_names(select(db, table, limit=0, recover=recover,
1140
        log_level=4)))
1141 2414 aaronmk
1142 5383 aaronmk
def table_cols(db, table, *args, **kw_args):
1143
    return [sql_gen.as_Col(strings.ustr(c), table)
1144
        for c in table_col_names(db, table, *args, **kw_args)]
1145
1146 5521 aaronmk
def table_pkey_index(db, table, recover=None):
1147
    table_str = sql_gen.Literal(table.to_str(db))
1148
    try:
1149
        return sql_gen.Table(value(run_query(db, '''\
1150
SELECT relname
1151
FROM pg_index
1152
JOIN pg_class index ON index.oid = indexrelid
1153
WHERE
1154
indrelid = '''+table_str.to_str(db)+'''::regclass
1155
AND indisprimary
1156
'''
1157
            , recover, cacheable=True, log_level=4)), table.schema)
1158
    except StopIteration: raise DoesNotExistException('primary key', '')
1159
1160 5389 aaronmk
def table_pkey_col(db, table, recover=None):
1161 5061 aaronmk
    table = sql_gen.as_Table(table)
1162
1163 5989 aaronmk
    module = util.root_module(db.db)
1164
    if module == 'psycopg2':
1165
        return sql_gen.Col(index_cols(db, table_pkey_index(db, table,
1166
            recover))[0], table)
1167
    else:
1168
        join_cols = ['table_schema', 'table_name', 'constraint_schema',
1169
            'constraint_name']
1170
        tables = [sql_gen.Table('key_column_usage', 'information_schema'),
1171
            sql_gen.Join(
1172
                sql_gen.Table('table_constraints', 'information_schema'),
1173
                dict(((c, sql_gen.join_same_not_null) for c in join_cols)))]
1174
        cols = [sql_gen.Col('column_name')]
1175
1176
        conds = [('constraint_type', 'PRIMARY KEY'), ('table_name', table.name)]
1177
        schema = table.schema
1178
        if schema != None: conds.append(('table_schema', schema))
1179
        order_by = 'position_in_unique_constraint'
1180
1181
        try: return sql_gen.Col(value(select(db, tables, cols, conds,
1182
            order_by=order_by, limit=1, log_level=4)), table)
1183
        except StopIteration: raise DoesNotExistException('primary key', '')
1184 5389 aaronmk
1185 5990 aaronmk
def table_has_pkey(db, table, recover=None):
1186
    try: table_pkey_col(db, table, recover)
1187
    except DoesNotExistException: return False
1188
    else: return True
1189
1190 5389 aaronmk
def pkey_name(db, table, recover=None):
1191
    '''If no pkey, returns the first column in the table.'''
1192 5392 aaronmk
    return pkey_col(db, table, recover).name
1193 832 aaronmk
1194 5390 aaronmk
def pkey_col(db, table, recover=None):
1195 5391 aaronmk
    '''If no pkey, returns the first column in the table.'''
1196 5392 aaronmk
    try: return table_pkey_col(db, table, recover)
1197 5393 aaronmk
    except DoesNotExistException: return table_cols(db, table, recover)[0]
1198 5128 aaronmk
1199 2559 aaronmk
not_null_col = 'not_null_col'
1200 2340 aaronmk
1201
def table_not_null_col(db, table, recover=None):
1202
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
1203 5337 aaronmk
    if not_null_col in table_col_names(db, table, recover): return not_null_col
1204 5388 aaronmk
    else: return pkey_name(db, table, recover)
1205 2340 aaronmk
1206 3348 aaronmk
def constraint_cond(db, constraint):
1207
    module = util.root_module(db.db)
1208
    if module == 'psycopg2':
1209
        table_str = sql_gen.Literal(constraint.table.to_str(db))
1210
        name_str = sql_gen.Literal(constraint.name)
1211
        return value(run_query(db, '''\
1212
SELECT consrc
1213
FROM pg_constraint
1214
WHERE
1215
conrelid = '''+table_str.to_str(db)+'''::regclass
1216
AND conname = '''+name_str.to_str(db)+'''
1217
'''
1218
            , cacheable=True, log_level=4))
1219 5443 aaronmk
    else: raise NotImplementedError("Can't get constraint condition for "
1220
        +module+' database')
1221 3348 aaronmk
1222 5520 aaronmk
def index_exprs(db, index):
1223 3322 aaronmk
    index = sql_gen.as_Table(index)
1224 1909 aaronmk
    module = util.root_module(db.db)
1225
    if module == 'psycopg2':
1226 3322 aaronmk
        qual_index = sql_gen.Literal(index.to_str(db))
1227 5520 aaronmk
        return list(values(run_query(db, '''\
1228 3322 aaronmk
SELECT pg_get_indexdef(indexrelid, generate_series(1, indnatts), true)
1229
FROM pg_index
1230
WHERE indexrelid = '''+qual_index.to_str(db)+'''::regclass
1231 2782 aaronmk
'''
1232
            , cacheable=True, log_level=4)))
1233 5520 aaronmk
    else: raise NotImplementedError()
1234 853 aaronmk
1235 5520 aaronmk
def index_cols(db, index):
1236
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
1237
    automatically created. When you don't know whether something is a UNIQUE
1238
    constraint or a UNIQUE index, use this function.'''
1239
    return map(sql_gen.parse_expr_col, index_exprs(db, index))
1240
1241 5445 aaronmk
def index_cond(db, index):
1242
    index = sql_gen.as_Table(index)
1243
    module = util.root_module(db.db)
1244
    if module == 'psycopg2':
1245
        qual_index = sql_gen.Literal(index.to_str(db))
1246
        return value(run_query(db, '''\
1247
SELECT pg_get_expr(indpred, indrelid, true)
1248
FROM pg_index
1249
WHERE indexrelid = '''+qual_index.to_str(db)+'''::regclass
1250
'''
1251
            , cacheable=True, log_level=4))
1252
    else: raise NotImplementedError()
1253
1254 5521 aaronmk
def index_order_by(db, index):
1255
    return sql_gen.CustomCode(', '.join(index_exprs(db, index)))
1256
1257
def table_cluster_on(db, table, recover=None):
1258
    '''
1259
    @return The table's cluster index, or its pkey if none is set
1260
    '''
1261
    table_str = sql_gen.Literal(table.to_str(db))
1262
    try:
1263
        return sql_gen.Table(value(run_query(db, '''\
1264
SELECT relname
1265
FROM pg_index
1266
JOIN pg_class index ON index.oid = indexrelid
1267
WHERE
1268
indrelid = '''+table_str.to_str(db)+'''::regclass
1269
AND indisclustered
1270
'''
1271
            , recover, cacheable=True, log_level=4)), table.schema)
1272
    except StopIteration: return table_pkey_index(db, table, recover)
1273
1274
def table_order_by(db, table, recover=None):
1275 8223 aaronmk
    '''
1276
    @return None if table is view, because table_cluster_on() would return None
1277
    '''
1278 5525 aaronmk
    if table.order_by == None:
1279
        try: table.order_by = index_order_by(db, table_cluster_on(db, table,
1280
            recover))
1281
        except DoesNotExistException: pass
1282
    return table.order_by
1283 5521 aaronmk
1284 7242 aaronmk
#### Views
1285
1286
def view_exists(db, view):
1287
    view_str = sql_gen.Literal(view.to_str(db))
1288
    try:
1289
        return value(run_query(db, '''\
1290
SELECT relkind = 'v'
1291
FROM pg_class
1292
WHERE oid = '''+view_str.to_str(db)+'''::regclass
1293
'''
1294
            , cacheable=True, log_level=4))
1295
    except DoesNotExistException: return False
1296
1297 3079 aaronmk
#### Functions
1298
1299
def function_exists(db, function):
1300 3423 aaronmk
    qual_function = sql_gen.Literal(function.to_str(db))
1301
    try:
1302 3425 aaronmk
        select(db, fields=[sql_gen.Cast('regproc', qual_function)],
1303
            recover=True, cacheable=True, log_level=4)
1304 3423 aaronmk
    except DoesNotExistException: return False
1305 4146 aaronmk
    except DuplicateException: return True # overloaded function
1306 3423 aaronmk
    else: return True
1307 3079 aaronmk
1308 5713 aaronmk
def function_param0_type(db, function):
1309
    qual_function = sql_gen.Literal(function.to_str(db))
1310
    return value(run_query(db, '''\
1311
SELECT proargtypes[0]::regtype
1312
FROM pg_proc
1313
WHERE oid = '''+qual_function.to_str(db)+'''::regproc
1314
'''
1315
        , cacheable=True, log_level=4))
1316
1317 3079 aaronmk
##### Structural changes
1318
1319
#### Columns
1320
1321 5020 aaronmk
def add_col(db, table, col, comment=None, if_not_exists=False, **kw_args):
1322 3079 aaronmk
    '''
1323
    @param col TypedCol Name may be versioned, so be sure to propagate any
1324
        renaming back to any source column for the TypedCol.
1325
    @param comment None|str SQL comment used to distinguish columns of the same
1326
        name from each other when they contain different data, to allow the
1327
        ADD COLUMN query to be cached. If not set, query will not be cached.
1328
    '''
1329
    assert isinstance(col, sql_gen.TypedCol)
1330
1331
    while True:
1332
        str_ = 'ALTER TABLE '+table.to_str(db)+' ADD COLUMN '+col.to_str(db)
1333
        if comment != None: str_ += ' '+sql_gen.esc_comment(comment)
1334
1335
        try:
1336
            run_query(db, str_, recover=True, cacheable=True, **kw_args)
1337
            break
1338
        except DuplicateException:
1339 5020 aaronmk
            if if_not_exists: raise
1340 3079 aaronmk
            col.name = next_version(col.name)
1341
            # try again with next version of name
1342
1343
def add_not_null(db, col):
1344
    table = col.table
1345
    col = sql_gen.to_name_only_col(col)
1346
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ALTER COLUMN '
1347
        +col.to_str(db)+' SET NOT NULL', cacheable=True, log_level=3)
1348
1349 4443 aaronmk
def drop_not_null(db, col):
1350
    table = col.table
1351
    col = sql_gen.to_name_only_col(col)
1352
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ALTER COLUMN '
1353
        +col.to_str(db)+' DROP NOT NULL', cacheable=True, log_level=3)
1354
1355 2096 aaronmk
row_num_col = '_row_num'
1356
1357 4997 aaronmk
row_num_col_def = sql_gen.TypedCol('', 'serial', nullable=False,
1358 3079 aaronmk
    constraints='PRIMARY KEY')
1359
1360 4997 aaronmk
def add_row_num(db, table, name=row_num_col):
1361
    '''Adds a row number column to a table. Its definition is in
1362
    row_num_col_def. It will be the primary key.'''
1363
    col_def = copy.copy(row_num_col_def)
1364
    col_def.name = name
1365 5021 aaronmk
    add_col(db, table, col_def, comment='', if_not_exists=True, log_level=3)
1366 3079 aaronmk
1367
#### Indexes
1368
1369
def add_pkey(db, table, cols=None, recover=None):
1370
    '''Adds a primary key.
1371
    @param cols [sql_gen.Col,...] The columns in the primary key.
1372
        Defaults to the first column in the table.
1373
    @pre The table must not already have a primary key.
1374
    '''
1375
    table = sql_gen.as_Table(table)
1376 5388 aaronmk
    if cols == None: cols = [pkey_name(db, table, recover)]
1377 3079 aaronmk
    col_strs = [sql_gen.to_name_only_col(v).to_str(db) for v in cols]
1378
1379
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD PRIMARY KEY ('
1380
        +(', '.join(col_strs))+')', recover=True, cacheable=True, log_level=3,
1381
        log_ignore_excs=(DuplicateException,))
1382
1383 2998 aaronmk
def add_index(db, exprs, table=None, unique=False, ensure_not_null_=True):
1384 2688 aaronmk
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
1385 3356 aaronmk
    Currently, only function calls and literal values are supported expressions.
1386 2998 aaronmk
    @param ensure_not_null_ If set, translates NULL values to sentinel values.
1387 2847 aaronmk
        This allows indexes to be used for comparisons where NULLs are equal.
1388 2538 aaronmk
    '''
1389 2964 aaronmk
    exprs = lists.mk_seq(exprs)
1390 2538 aaronmk
1391 2688 aaronmk
    # Parse exprs
1392
    old_exprs = exprs[:]
1393
    exprs = []
1394
    cols = []
1395
    for i, expr in enumerate(old_exprs):
1396 2823 aaronmk
        expr = sql_gen.as_Col(expr, table)
1397 2688 aaronmk
1398 2823 aaronmk
        # Handle nullable columns
1399 2998 aaronmk
        if ensure_not_null_:
1400 3164 aaronmk
            try: expr = sql_gen.ensure_not_null(db, expr)
1401 2860 aaronmk
            except KeyError: pass # unknown type, so just create plain index
1402 2823 aaronmk
1403 2688 aaronmk
        # Extract col
1404 3002 aaronmk
        expr = copy.deepcopy(expr) # don't modify input!
1405 3356 aaronmk
        col = expr
1406
        if isinstance(expr, sql_gen.FunctionCall): col = expr.args[0]
1407
        expr = sql_gen.cast_literal(expr)
1408
        if not isinstance(expr, (sql_gen.Expr, sql_gen.Col)):
1409 2688 aaronmk
            expr = sql_gen.Expr(expr)
1410 3356 aaronmk
1411 2688 aaronmk
1412
        # Extract table
1413
        if table == None:
1414
            assert sql_gen.is_table_col(col)
1415
            table = col.table
1416
1417 3356 aaronmk
        if isinstance(col, sql_gen.Col): col.table = None
1418 2688 aaronmk
1419
        exprs.append(expr)
1420
        cols.append(col)
1421 2408 aaronmk
1422 2688 aaronmk
    table = sql_gen.as_Table(table)
1423
1424 3005 aaronmk
    # Add index
1425 3148 aaronmk
    str_ = 'CREATE'
1426
    if unique: str_ += ' UNIQUE'
1427
    str_ += ' INDEX ON '+table.to_str(db)+' ('+(
1428
        ', '.join((v.to_str(db) for v in exprs)))+')'
1429
    run_query(db, str_, recover=True, cacheable=True, log_level=3)
1430 2408 aaronmk
1431 5987 aaronmk
def add_pkey_index(db, table): add_index(db, pkey_col(db, table), table)
1432
1433 5765 aaronmk
def add_pkey_or_index(db, table, cols=None, recover=None, warn=False):
1434
    try: add_pkey(db, table, cols, recover)
1435
    except DuplicateKeyException, e:
1436
        if warn: warnings.warn(UserWarning(exc.str_(e)))
1437 5988 aaronmk
        add_pkey_index(db, table)
1438 5765 aaronmk
1439 3083 aaronmk
already_indexed = object() # tells add_indexes() the pkey has already been added
1440
1441
def add_indexes(db, table, has_pkey=True):
1442
    '''Adds an index on all columns in a table.
1443
    @param has_pkey bool|already_indexed Whether a pkey instead of a regular
1444
        index should be added on the first column.
1445
        * If already_indexed, the pkey is assumed to have already been added
1446
    '''
1447 5337 aaronmk
    cols = table_col_names(db, table)
1448 3083 aaronmk
    if has_pkey:
1449
        if has_pkey is not already_indexed: add_pkey(db, table)
1450
        cols = cols[1:]
1451
    for col in cols: add_index(db, col, table)
1452
1453 3079 aaronmk
#### Tables
1454 2772 aaronmk
1455 3079 aaronmk
### Maintenance
1456 2772 aaronmk
1457 3079 aaronmk
def analyze(db, table):
1458
    table = sql_gen.as_Table(table)
1459
    run_query(db, 'ANALYZE '+table.to_str(db), log_level=3)
1460 2934 aaronmk
1461 3079 aaronmk
def autoanalyze(db, table):
1462
    if db.autoanalyze: analyze(db, table)
1463 2935 aaronmk
1464 3079 aaronmk
def vacuum(db, table):
1465
    table = sql_gen.as_Table(table)
1466
    db.with_autocommit(lambda: run_query(db, 'VACUUM ANALYZE '+table.to_str(db),
1467
        log_level=3))
1468 2086 aaronmk
1469 3079 aaronmk
### Lifecycle
1470
1471 3247 aaronmk
def drop(db, type_, name):
1472
    name = sql_gen.as_Name(name)
1473
    run_query(db, 'DROP '+type_+' IF EXISTS '+name.to_str(db)+' CASCADE')
1474 2889 aaronmk
1475 3247 aaronmk
def drop_table(db, table): drop(db, 'TABLE', table)
1476
1477 3082 aaronmk
def create_table(db, table, cols=[], has_pkey=True, col_indexes=True,
1478
    like=None):
1479 2675 aaronmk
    '''Creates a table.
1480 2681 aaronmk
    @param cols [sql_gen.TypedCol,...] The column names and types
1481
    @param has_pkey If set, the first column becomes the primary key.
1482 2760 aaronmk
    @param col_indexes bool|[ref]
1483
        * If True, indexes will be added on all non-pkey columns.
1484
        * If a list reference, [0] will be set to a function to do this.
1485
          This can be used to delay index creation until the table is populated.
1486 2675 aaronmk
    '''
1487
    table = sql_gen.as_Table(table)
1488
1489 3082 aaronmk
    if like != None:
1490
        cols = [sql_gen.CustomCode('LIKE '+like.to_str(db)+' INCLUDING ALL')
1491
            ]+cols
1492 5525 aaronmk
        table.order_by = like.order_by
1493 2681 aaronmk
    if has_pkey:
1494
        cols[0] = pkey = copy.copy(cols[0]) # don't modify input!
1495 2872 aaronmk
        pkey.constraints = 'PRIMARY KEY'
1496 2681 aaronmk
1497 3085 aaronmk
    temp = table.is_temp and not db.debug_temp
1498
        # temp tables permanent in debug_temp mode
1499 2760 aaronmk
1500 3085 aaronmk
    # Create table
1501 3383 aaronmk
    def create():
1502 3085 aaronmk
        str_ = 'CREATE'
1503
        if temp: str_ += ' TEMP'
1504
        str_ += ' TABLE '+table.to_str(db)+' (\n'
1505
        str_ += '\n, '.join(c.to_str(db) for c in cols)
1506 3126 aaronmk
        str_ += '\n);'
1507 3085 aaronmk
1508 7113 aaronmk
        opts = dict(recover=True, cacheable=True, log_level=2,
1509 3383 aaronmk
            log_ignore_excs=(DuplicateException,))
1510 7113 aaronmk
        try: run_query(db, str_, **opts)
1511
        except InvalidTypeException: # try again as view
1512
            run_query_into(db, mk_select(db, like, limit=0), into=table, **opts)
1513 3383 aaronmk
    if table.is_temp:
1514
        while True:
1515
            try:
1516
                create()
1517
                break
1518
            except DuplicateException:
1519
                table.name = next_version(table.name)
1520
                # try again with next version of name
1521
    else: create()
1522 3085 aaronmk
1523 2760 aaronmk
    # Add indexes
1524 2773 aaronmk
    if has_pkey: has_pkey = already_indexed
1525
    def add_indexes_(): add_indexes(db, table, has_pkey)
1526
    if isinstance(col_indexes, list): col_indexes[0] = add_indexes_ # defer
1527
    elif col_indexes: add_indexes_() # add now
1528 2675 aaronmk
1529 3084 aaronmk
def copy_table_struct(db, src, dest):
1530
    '''Creates a structure-only copy of a table. (Does not copy data.)'''
1531 3085 aaronmk
    create_table(db, dest, has_pkey=False, col_indexes=False, like=src)
1532 3084 aaronmk
1533 5529 aaronmk
def copy_table(db, src, dest):
1534
    '''Creates a copy of a table, including data'''
1535
    copy_table_struct(db, src, dest)
1536
    insert_select(db, dest, None, mk_select(db, src))
1537
1538 3079 aaronmk
### Data
1539 2684 aaronmk
1540 2970 aaronmk
def truncate(db, table, schema='public', **kw_args):
1541
    '''For params, see run_query()'''
1542 2777 aaronmk
    table = sql_gen.as_Table(table, schema)
1543 2970 aaronmk
    return run_query(db, 'TRUNCATE '+table.to_str(db)+' CASCADE', **kw_args)
1544 2732 aaronmk
1545 2965 aaronmk
def empty_temp(db, tables):
1546
    tables = lists.mk_seq(tables)
1547 2971 aaronmk
    for table in tables: truncate(db, table, log_level=3)
1548 2965 aaronmk
1549 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
1550
    '''For kw_args, see tables()'''
1551
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
1552 3094 aaronmk
1553 7179 aaronmk
def distinct_table(db, table, distinct_on, joins=None):
1554 3094 aaronmk
    '''Creates a copy of a temp table which is distinct on the given columns.
1555 5902 aaronmk
    Adds an index on table's distinct_on columns, to facilitate merge joins.
1556 3097 aaronmk
    @param distinct_on If empty, creates a table with one row. This is useful if
1557
        your distinct_on columns are all literal values.
1558 7179 aaronmk
    @param joins The joins to use when creating the new table
1559 3099 aaronmk
    @return The new table.
1560 3094 aaronmk
    '''
1561 7179 aaronmk
    if joins == None: joins = [table]
1562
1563 3099 aaronmk
    new_table = sql_gen.suffixed_table(table, '_distinct')
1564 3411 aaronmk
    distinct_on = filter(sql_gen.is_table_col, distinct_on)
1565 3094 aaronmk
1566 3099 aaronmk
    copy_table_struct(db, table, new_table)
1567 3097 aaronmk
1568
    limit = None
1569
    if distinct_on == []: limit = 1 # one sample row
1570 5903 aaronmk
    else: add_index(db, distinct_on, table) # for join optimization
1571 3097 aaronmk
1572 7179 aaronmk
    insert_select(db, new_table, None, mk_select(db, joins,
1573
        [sql_gen.Col(sql_gen.all_cols, table)], distinct_on=distinct_on,
1574
        order_by=None, limit=limit))
1575 3099 aaronmk
    analyze(db, new_table)
1576 3094 aaronmk
1577 3099 aaronmk
    return new_table