Project

General

Profile

1 11 aaronmk
# Database access
2
3 1869 aaronmk
import copy
4 11 aaronmk
import re
5 865 aaronmk
import warnings
6 11 aaronmk
7 300 aaronmk
import exc
8 1909 aaronmk
import dicts
9 1893 aaronmk
import iters
10 1960 aaronmk
import lists
11 1889 aaronmk
from Proxy import Proxy
12 1872 aaronmk
import rand
13 2217 aaronmk
import sql_gen
14 862 aaronmk
import strings
15 131 aaronmk
import util
16 11 aaronmk
17 832 aaronmk
##### Exceptions
18
19 2804 aaronmk
def get_cur_query(cur, input_query=None):
20 2168 aaronmk
    raw_query = None
21
    if hasattr(cur, 'query'): raw_query = cur.query
22
    elif hasattr(cur, '_last_executed'): raw_query = cur._last_executed
23 2170 aaronmk
24
    if raw_query != None: return raw_query
25 2804 aaronmk
    else: return '[input] '+strings.ustr(input_query)
26 14 aaronmk
27 2170 aaronmk
def _add_cursor_info(e, *args, **kw_args):
28
    '''For params, see get_cur_query()'''
29 2771 aaronmk
    exc.add_msg(e, 'query: '+strings.ustr(get_cur_query(*args, **kw_args)))
30 135 aaronmk
31 300 aaronmk
class DbException(exc.ExceptionWithCause):
32 14 aaronmk
    def __init__(self, msg, cause=None, cur=None):
33 2145 aaronmk
        exc.ExceptionWithCause.__init__(self, msg, cause, cause_newline=True)
34 14 aaronmk
        if cur != None: _add_cursor_info(self, cur)
35
36 2143 aaronmk
class ExceptionWithName(DbException):
37
    def __init__(self, name, cause=None):
38 2484 aaronmk
        DbException.__init__(self, 'for name: '+strings.as_tt(str(name)), cause)
39 2143 aaronmk
        self.name = name
40 360 aaronmk
41 2240 aaronmk
class ExceptionWithNameValue(DbException):
42
    def __init__(self, name, value, cause=None):
43 2484 aaronmk
        DbException.__init__(self, 'for name: '+strings.as_tt(str(name))
44
            +'; value: '+strings.as_tt(repr(value)), cause)
45 2240 aaronmk
        self.name = name
46
        self.value = value
47
48 2945 aaronmk
class ExceptionWithNameType(DbException):
49
    def __init__(self, type_, name, cause=None):
50
        DbException.__init__(self, 'for type: '+strings.as_tt(str(type_))
51
            +'; name: '+strings.as_tt(name), cause)
52
        self.type = type_
53
        self.name = name
54
55 2306 aaronmk
class ConstraintException(DbException):
56
    def __init__(self, name, cols, cause=None):
57 2484 aaronmk
        DbException.__init__(self, 'Violated '+strings.as_tt(name)
58
            +' constraint on columns: '+strings.as_tt(', '.join(cols)), cause)
59 2306 aaronmk
        self.name = name
60 468 aaronmk
        self.cols = cols
61 11 aaronmk
62 2523 aaronmk
class MissingCastException(DbException):
63
    def __init__(self, type_, col, cause=None):
64
        DbException.__init__(self, 'Missing cast to type '+strings.as_tt(type_)
65
            +' on column: '+strings.as_tt(col), cause)
66
        self.type = type_
67
        self.col = col
68
69 2143 aaronmk
class NameException(DbException): pass
70
71 2306 aaronmk
class DuplicateKeyException(ConstraintException): pass
72 13 aaronmk
73 2306 aaronmk
class NullValueException(ConstraintException): pass
74 13 aaronmk
75 2240 aaronmk
class FunctionValueException(ExceptionWithNameValue): pass
76 2239 aaronmk
77 2945 aaronmk
class DuplicateException(ExceptionWithNameType): pass
78 2143 aaronmk
79 89 aaronmk
class EmptyRowException(DbException): pass
80
81 865 aaronmk
##### Warnings
82
83
class DbWarning(UserWarning): pass
84
85 1930 aaronmk
##### Result retrieval
86
87
def col_names(cur): return (col[0] for col in cur.description)
88
89
def rows(cur): return iter(lambda: cur.fetchone(), None)
90
91
def consume_rows(cur):
92
    '''Used to fetch all rows so result will be cached'''
93
    iters.consume_iter(rows(cur))
94
95
def next_row(cur): return rows(cur).next()
96
97
def row(cur):
98
    row_ = next_row(cur)
99
    consume_rows(cur)
100
    return row_
101
102
def next_value(cur): return next_row(cur)[0]
103
104
def value(cur): return row(cur)[0]
105
106
def values(cur): return iters.func_iter(lambda: next_value(cur))
107
108
def value_or_none(cur):
109
    try: return value(cur)
110
    except StopIteration: return None
111
112 2762 aaronmk
##### Escaping
113 2101 aaronmk
114 2573 aaronmk
def esc_name_by_module(module, name):
115
    if module == 'psycopg2' or module == None: quote = '"'
116 2101 aaronmk
    elif module == 'MySQLdb': quote = '`'
117
    else: raise NotImplementedError("Can't escape name for "+module+' database')
118 2500 aaronmk
    return sql_gen.esc_name(name, quote)
119 2101 aaronmk
120
def esc_name_by_engine(engine, name, **kw_args):
121
    return esc_name_by_module(db_engines[engine][0], name, **kw_args)
122
123
def esc_name(db, name, **kw_args):
124
    return esc_name_by_module(util.root_module(db.db), name, **kw_args)
125
126
def qual_name(db, schema, table):
127
    def esc_name_(name): return esc_name(db, name)
128
    table = esc_name_(table)
129
    if schema != None: return esc_name_(schema)+'.'+table
130
    else: return table
131
132 1869 aaronmk
##### Database connections
133 1849 aaronmk
134 2097 aaronmk
db_config_names = ['engine', 'host', 'user', 'password', 'database', 'schemas']
135 1926 aaronmk
136 1869 aaronmk
db_engines = {
137
    'MySQL': ('MySQLdb', {'password': 'passwd', 'database': 'db'}),
138
    'PostgreSQL': ('psycopg2', {}),
139
}
140
141
DatabaseErrors_set = set([DbException])
142
DatabaseErrors = tuple(DatabaseErrors_set)
143
144
def _add_module(module):
145
    DatabaseErrors_set.add(module.DatabaseError)
146
    global DatabaseErrors
147
    DatabaseErrors = tuple(DatabaseErrors_set)
148
149
def db_config_str(db_config):
150
    return db_config['engine']+' database '+db_config['database']
151
152 2448 aaronmk
log_debug_none = lambda msg, level=2: None
153 1901 aaronmk
154 1849 aaronmk
class DbConn:
155 2923 aaronmk
    def __init__(self, db_config, autocommit=True, caching=True,
156 2915 aaronmk
        log_debug=log_debug_none, debug_temp=False):
157
        '''
158
        @param debug_temp Whether temporary objects should instead be permanent.
159
            This assists in debugging the internal objects used by the program.
160
        '''
161 1869 aaronmk
        self.db_config = db_config
162 2190 aaronmk
        self.autocommit = autocommit
163
        self.caching = caching
164 1901 aaronmk
        self.log_debug = log_debug
165 2193 aaronmk
        self.debug = log_debug != log_debug_none
166 2915 aaronmk
        self.debug_temp = debug_temp
167 3074 aaronmk
        self.autoanalyze = False
168 1869 aaronmk
169
        self.__db = None
170 1889 aaronmk
        self.query_results = {}
171 2139 aaronmk
        self._savepoint = 0
172 2671 aaronmk
        self._notices_seen = set()
173 1869 aaronmk
174
    def __getattr__(self, name):
175
        if name == '__dict__': raise Exception('getting __dict__')
176
        if name == 'db': return self._db()
177
        else: raise AttributeError()
178
179
    def __getstate__(self):
180
        state = copy.copy(self.__dict__) # shallow copy
181 1915 aaronmk
        state['log_debug'] = None # don't pickle the debug callback
182 1869 aaronmk
        state['_DbConn__db'] = None # don't pickle the connection
183
        return state
184
185 2165 aaronmk
    def connected(self): return self.__db != None
186
187 1869 aaronmk
    def _db(self):
188
        if self.__db == None:
189
            # Process db_config
190
            db_config = self.db_config.copy() # don't modify input!
191 2097 aaronmk
            schemas = db_config.pop('schemas', None)
192 1869 aaronmk
            module_name, mappings = db_engines[db_config.pop('engine')]
193
            module = __import__(module_name)
194
            _add_module(module)
195
            for orig, new in mappings.iteritems():
196
                try: util.rename_key(db_config, orig, new)
197
                except KeyError: pass
198
199
            # Connect
200
            self.__db = module.connect(**db_config)
201
202
            # Configure connection
203 2906 aaronmk
            if hasattr(self.db, 'set_isolation_level'):
204
                import psycopg2.extensions
205
                self.db.set_isolation_level(
206
                    psycopg2.extensions.ISOLATION_LEVEL_READ_COMMITTED)
207 2101 aaronmk
            if schemas != None:
208 2893 aaronmk
                search_path = [self.esc_name(s) for s in schemas.split(',')]
209
                search_path.append(value(run_query(self, 'SHOW search_path',
210
                    log_level=4)))
211
                run_query(self, 'SET search_path TO '+(','.join(search_path)),
212
                    log_level=3)
213 1869 aaronmk
214
        return self.__db
215 1889 aaronmk
216 1891 aaronmk
    class DbCursor(Proxy):
217 1927 aaronmk
        def __init__(self, outer):
218 1891 aaronmk
            Proxy.__init__(self, outer.db.cursor())
219 2191 aaronmk
            self.outer = outer
220 1927 aaronmk
            self.query_results = outer.query_results
221 1894 aaronmk
            self.query_lookup = None
222 1891 aaronmk
            self.result = []
223 1889 aaronmk
224 2802 aaronmk
        def execute(self, query):
225 2764 aaronmk
            self._is_insert = query.startswith('INSERT')
226 2797 aaronmk
            self.query_lookup = query
227 2148 aaronmk
            try:
228 2191 aaronmk
                try:
229 2802 aaronmk
                    cur = self.inner.execute(query)
230 2191 aaronmk
                    self.outer.do_autocommit()
231 2802 aaronmk
                finally: self.query = get_cur_query(self.inner, query)
232 1904 aaronmk
            except Exception, e:
233 2802 aaronmk
                _add_cursor_info(e, self, query)
234 1904 aaronmk
                self.result = e # cache the exception as the result
235
                self._cache_result()
236
                raise
237 3004 aaronmk
238
            # Always cache certain queries
239
            if query.startswith('CREATE') or query.startswith('ALTER'):
240 3007 aaronmk
                # structural changes
241 3040 aaronmk
                # Rest of query must be unique in the face of name collisions,
242
                # so don't cache ADD COLUMN unless it has distinguishing comment
243
                if query.find('ADD COLUMN') < 0 or query.endswith('*/'):
244 3007 aaronmk
                    self._cache_result()
245 3004 aaronmk
            elif self.rowcount == 0 and query.startswith('SELECT'): # empty
246 2800 aaronmk
                consume_rows(self) # fetch all rows so result will be cached
247 3004 aaronmk
248 2762 aaronmk
            return cur
249 1894 aaronmk
250 1891 aaronmk
        def fetchone(self):
251
            row = self.inner.fetchone()
252 1899 aaronmk
            if row != None: self.result.append(row)
253
            # otherwise, fetched all rows
254 1904 aaronmk
            else: self._cache_result()
255
            return row
256
257
        def _cache_result(self):
258 2948 aaronmk
            # For inserts that return a result set, don't cache result set since
259
            # inserts are not idempotent. Other non-SELECT queries don't have
260
            # their result set read, so only exceptions will be cached (an
261
            # invalid query will always be invalid).
262 1930 aaronmk
            if self.query_results != None and (not self._is_insert
263 1906 aaronmk
                or isinstance(self.result, Exception)):
264
265 1894 aaronmk
                assert self.query_lookup != None
266 1916 aaronmk
                self.query_results[self.query_lookup] = self.CacheCursor(
267
                    util.dict_subset(dicts.AttrsDictView(self),
268
                    ['query', 'result', 'rowcount', 'description']))
269 1906 aaronmk
270 1916 aaronmk
        class CacheCursor:
271
            def __init__(self, cached_result): self.__dict__ = cached_result
272
273 1927 aaronmk
            def execute(self, *args, **kw_args):
274 1916 aaronmk
                if isinstance(self.result, Exception): raise self.result
275
                # otherwise, result is a rows list
276
                self.iter = iter(self.result)
277
278
            def fetchone(self):
279
                try: return self.iter.next()
280
                except StopIteration: return None
281 1891 aaronmk
282 2212 aaronmk
    def esc_value(self, value):
283 2663 aaronmk
        try: str_ = self.mogrify('%s', [value])
284
        except NotImplementedError, e:
285
            module = util.root_module(self.db)
286
            if module == 'MySQLdb':
287
                import _mysql
288
                str_ = _mysql.escape_string(value)
289
            else: raise e
290 2374 aaronmk
        return strings.to_unicode(str_)
291 2212 aaronmk
292 2347 aaronmk
    def esc_name(self, name): return esc_name(self, name) # calls global func
293
294 2814 aaronmk
    def std_code(self, str_):
295
        '''Standardizes SQL code.
296
        * Ensures that string literals are prefixed by `E`
297
        '''
298
        if str_.startswith("'"): str_ = 'E'+str_
299
        return str_
300
301 2665 aaronmk
    def can_mogrify(self):
302 2663 aaronmk
        module = util.root_module(self.db)
303 2665 aaronmk
        return module == 'psycopg2'
304 2663 aaronmk
305 2665 aaronmk
    def mogrify(self, query, params=None):
306
        if self.can_mogrify(): return self.db.cursor().mogrify(query, params)
307
        else: raise NotImplementedError("Can't mogrify query")
308
309 2671 aaronmk
    def print_notices(self):
310 2725 aaronmk
        if hasattr(self.db, 'notices'):
311
            for msg in self.db.notices:
312
                if msg not in self._notices_seen:
313
                    self._notices_seen.add(msg)
314
                    self.log_debug(msg, level=2)
315 2671 aaronmk
316 2793 aaronmk
    def run_query(self, query, cacheable=False, log_level=2,
317 2464 aaronmk
        debug_msg_ref=None):
318 2445 aaronmk
        '''
319 2464 aaronmk
        @param log_ignore_excs The log_level will be increased by 2 if the query
320
            throws one of these exceptions.
321 2664 aaronmk
        @param debug_msg_ref If specified, the log message will be returned in
322
            this instead of being output. This allows you to filter log messages
323
            depending on the result of the query.
324 2445 aaronmk
        '''
325 2167 aaronmk
        assert query != None
326
327 2047 aaronmk
        if not self.caching: cacheable = False
328 1903 aaronmk
        used_cache = False
329 2664 aaronmk
330
        def log_msg(query):
331
            if used_cache: cache_status = 'cache hit'
332
            elif cacheable: cache_status = 'cache miss'
333
            else: cache_status = 'non-cacheable'
334
            return 'DB query: '+cache_status+':\n'+strings.as_code(query, 'SQL')
335
336 1903 aaronmk
        try:
337 1927 aaronmk
            # Get cursor
338
            if cacheable:
339
                try:
340 2797 aaronmk
                    cur = self.query_results[query]
341 1927 aaronmk
                    used_cache = True
342
                except KeyError: cur = self.DbCursor(self)
343
            else: cur = self.db.cursor()
344
345 2664 aaronmk
            # Log query
346
            if self.debug and debug_msg_ref == None: # log before running
347
                self.log_debug(log_msg(query), log_level)
348
349 1927 aaronmk
            # Run query
350 2793 aaronmk
            cur.execute(query)
351 1903 aaronmk
        finally:
352 2671 aaronmk
            self.print_notices()
353 2664 aaronmk
            if self.debug and debug_msg_ref != None: # return after running
354 2793 aaronmk
                debug_msg_ref[0] = log_msg(str(get_cur_query(cur, query)))
355 1903 aaronmk
356
        return cur
357 1914 aaronmk
358 2797 aaronmk
    def is_cached(self, query): return query in self.query_results
359 2139 aaronmk
360 2907 aaronmk
    def with_autocommit(self, func):
361 2801 aaronmk
        import psycopg2.extensions
362
363
        prev_isolation_level = self.db.isolation_level
364 2907 aaronmk
        self.db.set_isolation_level(
365
            psycopg2.extensions.ISOLATION_LEVEL_AUTOCOMMIT)
366 2683 aaronmk
        try: return func()
367 2801 aaronmk
        finally: self.db.set_isolation_level(prev_isolation_level)
368 2683 aaronmk
369 2139 aaronmk
    def with_savepoint(self, func):
370 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
371 2443 aaronmk
        self.run_query('SAVEPOINT '+savepoint, log_level=4)
372 2139 aaronmk
        self._savepoint += 1
373 2930 aaronmk
        try: return func()
374 2139 aaronmk
        except:
375 2443 aaronmk
            self.run_query('ROLLBACK TO SAVEPOINT '+savepoint, log_level=4)
376 2139 aaronmk
            raise
377 2930 aaronmk
        finally:
378
            # Always release savepoint, because after ROLLBACK TO SAVEPOINT,
379
            # "The savepoint remains valid and can be rolled back to again"
380
            # (http://www.postgresql.org/docs/8.3/static/sql-rollback-to.html).
381 2443 aaronmk
            self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
382 2930 aaronmk
383
            self._savepoint -= 1
384
            assert self._savepoint >= 0
385
386
            self.do_autocommit() # OK to do this after ROLLBACK TO SAVEPOINT
387 2191 aaronmk
388
    def do_autocommit(self):
389
        '''Autocommits if outside savepoint'''
390
        assert self._savepoint >= 0
391
        if self.autocommit and self._savepoint == 0:
392 2924 aaronmk
            self.log_debug('Autocommitting', level=4)
393 2191 aaronmk
            self.db.commit()
394 2643 aaronmk
395 2819 aaronmk
    def col_info(self, col):
396 2643 aaronmk
        table = sql_gen.Table('columns', 'information_schema')
397 3063 aaronmk
        type_ = sql_gen.Coalesce(sql_gen.Nullif(sql_gen.Col('data_type'),
398
            'USER-DEFINED'), sql_gen.Col('udt_name'))
399
        cols = [type_, 'column_default', cast(self, 'boolean', 'is_nullable')]
400 2643 aaronmk
401
        conds = [('table_name', col.table.name), ('column_name', col.name)]
402
        schema = col.table.schema
403
        if schema != None: conds.append(('table_schema', schema))
404
405 2819 aaronmk
        type_, default, nullable = row(select(self, table, cols, conds,
406 3059 aaronmk
            order_by='table_schema', limit=1, cacheable=False, log_level=4))
407 2643 aaronmk
            # TODO: order_by search_path schema order
408 2819 aaronmk
        default = sql_gen.as_Code(default, self)
409
410
        return sql_gen.TypedCol(col.name, type_, default, nullable)
411 2917 aaronmk
412
    def TempFunction(self, name):
413
        if self.debug_temp: schema = None
414
        else: schema = 'pg_temp'
415
        return sql_gen.Function(name, schema)
416 1849 aaronmk
417 1869 aaronmk
connect = DbConn
418
419 832 aaronmk
##### Recoverable querying
420 15 aaronmk
421 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
422 11 aaronmk
423 2791 aaronmk
def run_query(db, query, recover=None, cacheable=False, log_level=2,
424
    log_ignore_excs=None, **kw_args):
425 2794 aaronmk
    '''For params, see DbConn.run_query()'''
426 830 aaronmk
    if recover == None: recover = False
427 2464 aaronmk
    if log_ignore_excs == None: log_ignore_excs = ()
428
    log_ignore_excs = tuple(log_ignore_excs)
429 830 aaronmk
430 2666 aaronmk
    debug_msg_ref = None # usually, db.run_query() logs query before running it
431
    # But if filtering with log_ignore_excs, wait until after exception parsing
432 2984 aaronmk
    if log_ignore_excs != () or not db.can_mogrify(): debug_msg_ref = [None]
433 2666 aaronmk
434 2148 aaronmk
    try:
435 2464 aaronmk
        try:
436 2794 aaronmk
            def run(): return db.run_query(query, cacheable, log_level,
437 2793 aaronmk
                debug_msg_ref, **kw_args)
438 2796 aaronmk
            if recover and not db.is_cached(query):
439 2464 aaronmk
                return with_savepoint(db, run)
440
            else: return run() # don't need savepoint if cached
441
        except Exception, e:
442
            msg = exc.str_(e)
443
444
            match = re.search(r'duplicate key value violates unique constraint '
445 2493 aaronmk
                r'"((_?[^\W_]+)_.+?)"', msg)
446 2464 aaronmk
            if match:
447
                constraint, table = match.groups()
448 3025 aaronmk
                cols = []
449
                if recover: # need auto-rollback to run index_cols()
450
                    try: cols = index_cols(db, table, constraint)
451
                    except NotImplementedError: pass
452
                raise DuplicateKeyException(constraint, cols, e)
453 2464 aaronmk
454 2493 aaronmk
            match = re.search(r'null value in column "(.+?)" violates not-null'
455 2464 aaronmk
                r' constraint', msg)
456
            if match: raise NullValueException('NOT NULL', [match.group(1)], e)
457
458
            match = re.search(r'\b(?:invalid input (?:syntax|value)\b.*?'
459
                r'|date/time field value out of range): "(.+?)"\n'
460 2535 aaronmk
                r'(?:(?s).*?)\bfunction "(.+?)"', msg)
461 2464 aaronmk
            if match:
462
                value, name = match.groups()
463
                raise FunctionValueException(name, strings.to_unicode(value), e)
464
465 2526 aaronmk
            match = re.search(r'column "(.+?)" is of type (.+?) but expression '
466 2523 aaronmk
                r'is of type', msg)
467
            if match:
468
                col, type_ = match.groups()
469
                raise MissingCastException(type_, col, e)
470
471 2946 aaronmk
            match = re.search(r'\b(\S+) "(.+?)".*? already exists', msg)
472 2945 aaronmk
            if match:
473
                type_, name = match.groups()
474
                raise DuplicateException(type_, name, e)
475 2464 aaronmk
476
            raise # no specific exception raised
477
    except log_ignore_excs:
478
        log_level += 2
479
        raise
480
    finally:
481 2666 aaronmk
        if debug_msg_ref != None and debug_msg_ref[0] != None:
482
            db.log_debug(debug_msg_ref[0], log_level)
483 830 aaronmk
484 832 aaronmk
##### Basic queries
485
486 2153 aaronmk
def next_version(name):
487 2163 aaronmk
    version = 1 # first existing name was version 0
488 2586 aaronmk
    match = re.match(r'^(.*)#(\d+)$', name)
489 2153 aaronmk
    if match:
490 2586 aaronmk
        name, version = match.groups()
491
        version = int(version)+1
492 2932 aaronmk
    return sql_gen.concat(name, '#'+str(version))
493 2153 aaronmk
494 2899 aaronmk
def lock_table(db, table, mode):
495
    table = sql_gen.as_Table(table)
496
    run_query(db, 'LOCK TABLE '+table.to_str(db)+' IN '+mode+' MODE')
497
498 2789 aaronmk
def run_query_into(db, query, into=None, add_indexes_=False, **kw_args):
499 2085 aaronmk
    '''Outputs a query to a temp table.
500
    For params, see run_query().
501
    '''
502 2789 aaronmk
    if into == None: return run_query(db, query, **kw_args)
503 2790 aaronmk
504
    assert isinstance(into, sql_gen.Table)
505
506 2992 aaronmk
    into.is_temp = True
507 3008 aaronmk
    # "temporary tables cannot specify a schema name", so remove schema
508
    into.schema = None
509 2992 aaronmk
510 2790 aaronmk
    kw_args['recover'] = True
511 2945 aaronmk
    kw_args.setdefault('log_ignore_excs', (DuplicateException,))
512 2790 aaronmk
513 2916 aaronmk
    temp = not db.debug_temp # tables are permanent in debug_temp mode
514 2790 aaronmk
515
    # Create table
516
    while True:
517
        create_query = 'CREATE'
518
        if temp: create_query += ' TEMP'
519
        create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
520 2385 aaronmk
521 2790 aaronmk
        try:
522
            cur = run_query(db, create_query, **kw_args)
523
                # CREATE TABLE AS sets rowcount to # rows in query
524
            break
525 2945 aaronmk
        except DuplicateException, e:
526 2790 aaronmk
            into.name = next_version(into.name)
527
            # try again with next version of name
528
529
    if add_indexes_: add_indexes(db, into)
530 3075 aaronmk
531
    # According to the PostgreSQL doc, "The autovacuum daemon cannot access and
532
    # therefore cannot vacuum or analyze temporary tables. [...] if a temporary
533
    # table is going to be used in complex queries, it is wise to run ANALYZE on
534
    # the temporary table after it is populated."
535
    # (http://www.postgresql.org/docs/9.1/static/sql-createtable.html)
536
    # If into is not a temp table, ANALYZE is useful but not required.
537 3073 aaronmk
    analyze(db, into)
538 2790 aaronmk
539
    return cur
540 2085 aaronmk
541 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
542
543 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
544
545 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
546 2293 aaronmk
    start=None, order_by=order_by_pkey, default_table=None):
547 1981 aaronmk
    '''
548 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
549 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
550 1981 aaronmk
    @param fields Use None to select all fields in the table
551 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
552 2379 aaronmk
        * container can be any iterable type
553 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
554
        * compare_right_side: sql_gen.ValueCond|literal value
555 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
556
        use all columns
557 2786 aaronmk
    @return query
558 1981 aaronmk
    '''
559 2315 aaronmk
    # Parse tables param
560 2964 aaronmk
    tables = lists.mk_seq(tables)
561 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
562 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
563 2121 aaronmk
564 2315 aaronmk
    # Parse other params
565 2376 aaronmk
    if conds == None: conds = []
566 2650 aaronmk
    elif dicts.is_dict(conds): conds = conds.items()
567 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
568 135 aaronmk
    assert limit == None or type(limit) == int
569 865 aaronmk
    assert start == None or type(start) == int
570 2315 aaronmk
    if order_by is order_by_pkey:
571
        if distinct_on != []: order_by = None
572
        else: order_by = pkey(db, table0, recover=True)
573 865 aaronmk
574 2315 aaronmk
    query = 'SELECT'
575 2056 aaronmk
576 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
577 2056 aaronmk
578 2200 aaronmk
    # DISTINCT ON columns
579 2233 aaronmk
    if distinct_on != []:
580 2467 aaronmk
        query += '\nDISTINCT'
581 2254 aaronmk
        if distinct_on is not distinct_on_all:
582 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
583
584
    # Columns
585 3027 aaronmk
    if fields == None:
586
        if query.find('\n') >= 0: whitespace = '\n'
587
        else: whitespace = ' '
588
        query += whitespace+'*'
589 2765 aaronmk
    else:
590
        assert fields != []
591 3027 aaronmk
        query += '\n'+('\n, '.join(map(parse_col, fields)))
592 2200 aaronmk
593
    # Main table
594 2467 aaronmk
    query += '\nFROM '+table0.to_str(db)
595 865 aaronmk
596 2122 aaronmk
    # Add joins
597 2271 aaronmk
    left_table = table0
598 2263 aaronmk
    for join_ in tables:
599
        table = join_.table
600 2238 aaronmk
601 2343 aaronmk
        # Parse special values
602
        if join_.type_ is sql_gen.filter_out: # filter no match
603 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
604 2853 aaronmk
                sql_gen.CompareCond(None, '~=')))
605 2343 aaronmk
606 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
607 2122 aaronmk
608
        left_table = table
609
610 865 aaronmk
    missing = True
611 2376 aaronmk
    if conds != []:
612 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
613
        else: whitespace = '\n'
614 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
615
            .to_str(db) for l, r in conds], 'WHERE')
616 865 aaronmk
        missing = False
617 2227 aaronmk
    if order_by != None:
618 2467 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
619
    if limit != None: query += '\nLIMIT '+str(limit); missing = False
620 865 aaronmk
    if start != None:
621 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
622 865 aaronmk
        missing = False
623
    if missing: warnings.warn(DbWarning(
624
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
625
626 2786 aaronmk
    return query
627 11 aaronmk
628 2054 aaronmk
def select(db, *args, **kw_args):
629
    '''For params, see mk_select() and run_query()'''
630
    recover = kw_args.pop('recover', None)
631
    cacheable = kw_args.pop('cacheable', True)
632 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
633 2054 aaronmk
634 2791 aaronmk
    return run_query(db, mk_select(db, *args, **kw_args), recover, cacheable,
635
        log_level=log_level)
636 2054 aaronmk
637 2788 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, returning=None,
638 3009 aaronmk
    embeddable=False, ignore=False):
639 1960 aaronmk
    '''
640
    @param returning str|None An inserted column (such as pkey) to return
641 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
642 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
643
        query will be fully cached, not just if it raises an exception.
644 3009 aaronmk
    @param ignore Whether to ignore duplicate keys.
645 1960 aaronmk
    '''
646 2754 aaronmk
    table = sql_gen.remove_table_rename(sql_gen.as_Table(table))
647 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
648 3010 aaronmk
    if cols != None: cols = [sql_gen.to_name_only_col(c, table) for c in cols]
649 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
650 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
651 2063 aaronmk
652 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
653 2063 aaronmk
654 3009 aaronmk
    def mk_insert(select_query):
655
        query = first_line
656 3014 aaronmk
        if cols != None:
657
            query += '\n('+(', '.join((c.to_str(db) for c in cols)))+')'
658 3009 aaronmk
        query += '\n'+select_query
659
660
        if returning != None:
661
            returning_name_col = sql_gen.to_name_only_col(returning)
662
            query += '\nRETURNING '+returning_name_col.to_str(db)
663
664
        return query
665 2063 aaronmk
666 3017 aaronmk
    return_type = 'unknown'
667
    if returning != None: return_type = returning.to_str(db)+'%TYPE'
668
669 3009 aaronmk
    lang = 'sql'
670
    if ignore:
671 3017 aaronmk
        assert cols != None
672
        # Always return something to set the correct rowcount
673
        if returning == None: returning = sql_gen.NamedCol('NULL', None)
674
675 3009 aaronmk
        embeddable = True # must use function
676
        lang = 'plpgsql'
677 3014 aaronmk
        row = [sql_gen.Col(c.name, 'row') for c in cols]
678 3010 aaronmk
679 3009 aaronmk
        query = '''\
680 3010 aaronmk
DECLARE
681 3014 aaronmk
    row '''+table.to_str(db)+'''%ROWTYPE;
682 3009 aaronmk
BEGIN
683 3019 aaronmk
    /* Need an EXCEPTION block for each individual row because "When an error is
684
    caught by an EXCEPTION clause, [...] all changes to persistent database
685
    state within the block are rolled back."
686
    This is unfortunate because "A block containing an EXCEPTION clause is
687
    significantly more expensive to enter and exit than a block without one."
688 3015 aaronmk
    (http://www.postgresql.org/docs/8.3/static/plpgsql-control-structures.html\
689
#PLPGSQL-ERROR-TRAPPING)
690
    */
691 3034 aaronmk
    FOR '''+(', '.join((c.to_str(db) for c in row)))+''' IN
692
'''+select_query+'''
693
    LOOP
694 3015 aaronmk
        BEGIN
695 3019 aaronmk
            RETURN QUERY
696 3014 aaronmk
'''+mk_insert(sql_gen.Values(row).to_str(db))+'''
697 3010 aaronmk
;
698 3015 aaronmk
        EXCEPTION
699 3019 aaronmk
            WHEN unique_violation THEN NULL; -- continue to next row
700 3015 aaronmk
        END;
701 3010 aaronmk
    END LOOP;
702
END;\
703 3009 aaronmk
'''
704
    else: query = mk_insert(select_query)
705
706 2070 aaronmk
    if embeddable:
707
        # Create function
708 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
709 2189 aaronmk
        while True:
710
            try:
711 2918 aaronmk
                function = db.TempFunction(function_name)
712 2194 aaronmk
713 2189 aaronmk
                function_query = '''\
714 2698 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''()
715 3017 aaronmk
RETURNS SETOF '''+return_type+'''
716 3009 aaronmk
LANGUAGE '''+lang+'''
717 2467 aaronmk
AS $$
718 3009 aaronmk
'''+query+'''
719 2467 aaronmk
$$;
720 2070 aaronmk
'''
721 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
722 2945 aaronmk
                    log_ignore_excs=(DuplicateException,))
723 2189 aaronmk
                break # this version was successful
724 2945 aaronmk
            except DuplicateException, e:
725 2189 aaronmk
                function_name = next_version(function_name)
726
                # try again with next version of name
727 2070 aaronmk
728 2337 aaronmk
        # Return query that uses function
729 3009 aaronmk
        cols = None
730
        if returning != None: cols = [returning]
731 2698 aaronmk
        func_table = sql_gen.NamedTable('f', sql_gen.FunctionCall(function),
732 3009 aaronmk
            cols) # AS clause requires function alias
733 2787 aaronmk
        return mk_select(db, func_table, start=0, order_by=None)
734 2070 aaronmk
735 2787 aaronmk
    return query
736 2066 aaronmk
737 3074 aaronmk
def insert_select(db, table, *args, **kw_args):
738 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
739 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
740
        values in
741 2072 aaronmk
    '''
742 2386 aaronmk
    into = kw_args.pop('into', None)
743
    if into != None: kw_args['embeddable'] = True
744 2066 aaronmk
    recover = kw_args.pop('recover', None)
745 3011 aaronmk
    if kw_args.get('ignore', False): recover = True
746 2066 aaronmk
    cacheable = kw_args.pop('cacheable', True)
747 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
748 2066 aaronmk
749 3074 aaronmk
    cur = run_query_into(db, mk_insert_select(db, table, *args, **kw_args),
750
        into, recover=recover, cacheable=cacheable, log_level=log_level)
751
    autoanalyze(db, table)
752
    return cur
753 2063 aaronmk
754 2738 aaronmk
default = sql_gen.default # tells insert() to use the default value for a column
755 2066 aaronmk
756 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
757 2085 aaronmk
    '''For params, see insert_select()'''
758 1960 aaronmk
    if lists.is_seq(row): cols = None
759
    else:
760
        cols = row.keys()
761
        row = row.values()
762 2738 aaronmk
    row = list(row) # ensure that "== []" works
763 1960 aaronmk
764 2738 aaronmk
    if row == []: query = None
765
    else: query = sql_gen.Values(row).to_str(db)
766 1961 aaronmk
767 2788 aaronmk
    return insert_select(db, table, cols, query, *args, **kw_args)
768 11 aaronmk
769 3056 aaronmk
def mk_update(db, table, changes=None, cond=None, in_place=False):
770 2402 aaronmk
    '''
771
    @param changes [(col, new_value),...]
772
        * container can be any iterable type
773
        * col: sql_gen.Code|str (for col name)
774
        * new_value: sql_gen.Code|literal value
775
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
776 3056 aaronmk
    @param in_place If set, locks the table and updates rows in place.
777
        This avoids creating dead rows in PostgreSQL.
778
        * cond must be None
779 2402 aaronmk
    @return str query
780
    '''
781 3057 aaronmk
    table = sql_gen.as_Table(table)
782
    changes = [(sql_gen.to_name_only_col(c, table), sql_gen.as_Value(v))
783
        for c, v in changes]
784
785 3056 aaronmk
    if in_place:
786
        assert cond == None
787 3058 aaronmk
788 3065 aaronmk
        query = 'ALTER TABLE '+table.to_str(db)+'\n'
789
        query += ',\n'.join(('ALTER COLUMN '+c.to_str(db)+' TYPE '
790
            +db.col_info(sql_gen.with_default_table(c, table)).type
791
            +'\nUSING '+v.to_str(db) for c, v in changes))
792 3058 aaronmk
    else:
793
        query = 'UPDATE '+table.to_str(db)+'\nSET\n'
794
        query += ',\n'.join((c.to_str(db)+' = '+v.to_str(db)
795
            for c, v in changes))
796
        if cond != None: query += '\nWHERE\n'+cond.to_str(db)
797 3056 aaronmk
798 2402 aaronmk
    return query
799
800 3074 aaronmk
def update(db, table, *args, **kw_args):
801 2402 aaronmk
    '''For params, see mk_update() and run_query()'''
802
    recover = kw_args.pop('recover', None)
803 3043 aaronmk
    cacheable = kw_args.pop('cacheable', False)
804 3030 aaronmk
    log_level = kw_args.pop('log_level', 2)
805 2402 aaronmk
806 3074 aaronmk
    cur = run_query(db, mk_update(db, table, *args, **kw_args), recover,
807
        cacheable, log_level=log_level)
808
    autoanalyze(db, table)
809
    return cur
810 2402 aaronmk
811 135 aaronmk
def last_insert_id(db):
812 1849 aaronmk
    module = util.root_module(db.db)
813 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
814
    elif module == 'MySQLdb': return db.insert_id()
815
    else: return None
816 13 aaronmk
817 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
818 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
819 2415 aaronmk
    to names that will be distinct among the columns' tables.
820 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
821 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
822
    @param into The table for the new columns.
823 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
824
        columns will be included in the mapping even if they are not in cols.
825
        The tables of the provided Col objects will be changed to into, so make
826
        copies of them if you want to keep the original tables.
827
    @param as_items Whether to return a list of dict items instead of a dict
828 2383 aaronmk
    @return dict(orig_col=new_col, ...)
829
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
830 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
831
        * All mappings use the into table so its name can easily be
832 2383 aaronmk
          changed for all columns at once
833
    '''
834 2415 aaronmk
    cols = lists.uniqify(cols)
835
836 2394 aaronmk
    items = []
837 2389 aaronmk
    for col in preserve:
838 2390 aaronmk
        orig_col = copy.copy(col)
839 2392 aaronmk
        col.table = into
840 2394 aaronmk
        items.append((orig_col, col))
841
    preserve = set(preserve)
842
    for col in cols:
843 2716 aaronmk
        if col not in preserve:
844
            items.append((col, sql_gen.Col(str(col), into, col.srcs)))
845 2394 aaronmk
846
    if not as_items: items = dict(items)
847
    return items
848 2383 aaronmk
849 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
850 2391 aaronmk
    '''For params, see mk_flatten_mapping()
851
    @return See return value of mk_flatten_mapping()
852
    '''
853 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
854
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
855 2786 aaronmk
    run_query_into(db, mk_select(db, joins, cols, limit=limit, start=start),
856 2846 aaronmk
        into=into, add_indexes_=True)
857 2394 aaronmk
    return dict(items)
858 2391 aaronmk
859 3050 aaronmk
def track_data_error(db, errors_table, cols, value, error_code, error):
860 2758 aaronmk
    '''
861
    @param errors_table If None, does nothing.
862
    '''
863 2810 aaronmk
    if errors_table == None or cols == (): return
864 3050 aaronmk
865
    for col in cols:
866
        try:
867
            insert(db, errors_table, dict(column=col.name, value=value,
868
                error_code=error_code, error=error), recover=True,
869
                cacheable=True, log_level=4)
870
        except DuplicateKeyException: pass
871 2758 aaronmk
872 2704 aaronmk
def cast(db, type_, col, errors_table=None):
873
    '''Casts an (unrenamed) column or value.
874 2769 aaronmk
    If errors_table set and col has srcs, saves errors in errors_table (using
875
    col's srcs attr as the source columns) and converts errors to warnings.
876 2820 aaronmk
    @param col str|sql_gen.Col|sql_gen.Literal
877 2704 aaronmk
    @param errors_table None|sql_gen.Table|str
878
    '''
879 2820 aaronmk
    col = sql_gen.as_Col(col)
880 2769 aaronmk
    save_errors = (errors_table != None and isinstance(col, sql_gen.Col)
881
        and col.srcs != ())
882 2987 aaronmk
    if not save_errors: return sql_gen.Cast(type_, col) # can't save errors
883 2704 aaronmk
884
    assert not isinstance(col, sql_gen.NamedCol)
885
886 2770 aaronmk
    errors_table = sql_gen.as_Table(errors_table)
887
    srcs = map(sql_gen.to_name_only_col, col.srcs)
888
    function_name = str(sql_gen.FunctionCall(type_, *srcs))
889 2918 aaronmk
    function = db.TempFunction(function_name)
890 2704 aaronmk
891 2718 aaronmk
    while True:
892
        # Create function definition
893 3049 aaronmk
        errors_table_cols = map(sql_gen.Col,
894
            ['column', 'value', 'error_code', 'error'])
895 2718 aaronmk
        query = '''\
896 2704 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''(value text)
897
RETURNS '''+type_+'''
898
LANGUAGE plpgsql
899
STRICT
900
AS $$
901
BEGIN
902
    /* The explicit cast to the return type is needed to make the cast happen
903
    inside the try block. (Implicit casts to the return type happen at the end
904
    of the function, outside any block.) */
905
    RETURN value::'''+type_+''';
906
EXCEPTION
907
    WHEN data_exception THEN
908
        -- Save error in errors table.
909 3049 aaronmk
        DECLARE
910
            error_code text := SQLSTATE;
911
            error text := SQLERRM;
912
            "column" text;
913
        BEGIN
914
            -- Insert the value and error for *each* source column.
915
            FOR "column" IN
916
'''+mk_select(db, sql_gen.NamedValues('c', None, [[c.name] for c in srcs]),
917 3053 aaronmk
    order_by=None, start=0)+'''
918 3049 aaronmk
            LOOP
919
                BEGIN
920
'''+mk_insert_select(db, errors_table, errors_table_cols,
921
    sql_gen.Values(errors_table_cols).to_str(db))+''';
922
                EXCEPTION
923
                    WHEN unique_violation THEN NULL; -- continue to next row
924
                END;
925
            END LOOP;
926
        END;
927 2704 aaronmk
928
        RAISE WARNING '%', SQLERRM;
929
        RETURN NULL;
930
END;
931
$$;
932
'''
933 2718 aaronmk
934
        # Create function
935
        try:
936
            run_query(db, query, recover=True, cacheable=True,
937 2945 aaronmk
                log_ignore_excs=(DuplicateException,))
938 2718 aaronmk
            break # successful
939 2945 aaronmk
        except DuplicateException:
940 2770 aaronmk
            function.name = next_version(function.name)
941
            # try again with next version of name
942 2704 aaronmk
943
    return sql_gen.FunctionCall(function, col)
944
945 2414 aaronmk
##### Database structure queries
946
947 2426 aaronmk
def table_row_count(db, table, recover=None):
948 2786 aaronmk
    return value(run_query(db, mk_select(db, table, [sql_gen.row_count],
949 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
950 2426 aaronmk
951 2414 aaronmk
def table_cols(db, table, recover=None):
952
    return list(col_names(select(db, table, limit=0, order_by=None,
953 2443 aaronmk
        recover=recover, log_level=4)))
954 2414 aaronmk
955 2291 aaronmk
def pkey(db, table, recover=None):
956 832 aaronmk
    '''Assumed to be first column in table'''
957 2339 aaronmk
    return table_cols(db, table, recover)[0]
958 832 aaronmk
959 2559 aaronmk
not_null_col = 'not_null_col'
960 2340 aaronmk
961
def table_not_null_col(db, table, recover=None):
962
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
963
    if not_null_col in table_cols(db, table, recover): return not_null_col
964
    else: return pkey(db, table, recover)
965
966 853 aaronmk
def index_cols(db, table, index):
967
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
968
    automatically created. When you don't know whether something is a UNIQUE
969
    constraint or a UNIQUE index, use this function.'''
970 1909 aaronmk
    module = util.root_module(db.db)
971
    if module == 'psycopg2':
972
        return list(values(run_query(db, '''\
973 853 aaronmk
SELECT attname
974 866 aaronmk
FROM
975
(
976
        SELECT attnum, attname
977
        FROM pg_index
978
        JOIN pg_class index ON index.oid = indexrelid
979
        JOIN pg_class table_ ON table_.oid = indrelid
980
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
981
        WHERE
982 2782 aaronmk
            table_.relname = '''+db.esc_value(table)+'''
983
            AND index.relname = '''+db.esc_value(index)+'''
984 866 aaronmk
    UNION
985
        SELECT attnum, attname
986
        FROM
987
        (
988
            SELECT
989
                indrelid
990
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
991
                    AS indkey
992
            FROM pg_index
993
            JOIN pg_class index ON index.oid = indexrelid
994
            JOIN pg_class table_ ON table_.oid = indrelid
995
            WHERE
996 2782 aaronmk
                table_.relname = '''+db.esc_value(table)+'''
997
                AND index.relname = '''+db.esc_value(index)+'''
998 866 aaronmk
        ) s
999
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
1000
) s
1001 853 aaronmk
ORDER BY attnum
1002 2782 aaronmk
'''
1003
            , cacheable=True, log_level=4)))
1004 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
1005
        ' database')
1006 853 aaronmk
1007 464 aaronmk
def constraint_cols(db, table, constraint):
1008 1849 aaronmk
    module = util.root_module(db.db)
1009 464 aaronmk
    if module == 'psycopg2':
1010
        return list(values(run_query(db, '''\
1011
SELECT attname
1012
FROM pg_constraint
1013
JOIN pg_class ON pg_class.oid = conrelid
1014
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
1015
WHERE
1016 2783 aaronmk
    relname = '''+db.esc_value(table)+'''
1017
    AND conname = '''+db.esc_value(constraint)+'''
1018 464 aaronmk
ORDER BY attnum
1019 2783 aaronmk
'''
1020
            )))
1021 464 aaronmk
    else: raise NotImplementedError("Can't list constraint columns for "+module+
1022
        ' database')
1023
1024 2096 aaronmk
row_num_col = '_row_num'
1025
1026 2998 aaronmk
def add_index(db, exprs, table=None, unique=False, ensure_not_null_=True):
1027 2688 aaronmk
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
1028 2538 aaronmk
    Currently, only function calls are supported as expressions.
1029 2998 aaronmk
    @param ensure_not_null_ If set, translates NULL values to sentinel values.
1030 2847 aaronmk
        This allows indexes to be used for comparisons where NULLs are equal.
1031 2538 aaronmk
    '''
1032 2964 aaronmk
    exprs = lists.mk_seq(exprs)
1033 2538 aaronmk
1034 2688 aaronmk
    # Parse exprs
1035
    old_exprs = exprs[:]
1036
    exprs = []
1037
    cols = []
1038
    for i, expr in enumerate(old_exprs):
1039 2823 aaronmk
        expr = sql_gen.as_Col(expr, table)
1040 2688 aaronmk
1041 2823 aaronmk
        # Handle nullable columns
1042 2998 aaronmk
        if ensure_not_null_:
1043
            try: expr = ensure_not_null(db, expr)
1044 2860 aaronmk
            except KeyError: pass # unknown type, so just create plain index
1045 2823 aaronmk
1046 2688 aaronmk
        # Extract col
1047 3002 aaronmk
        expr = copy.deepcopy(expr) # don't modify input!
1048 2688 aaronmk
        if isinstance(expr, sql_gen.FunctionCall):
1049
            col = expr.args[0]
1050
            expr = sql_gen.Expr(expr)
1051
        else: col = expr
1052 2823 aaronmk
        assert isinstance(col, sql_gen.Col)
1053 2688 aaronmk
1054
        # Extract table
1055
        if table == None:
1056
            assert sql_gen.is_table_col(col)
1057
            table = col.table
1058
1059
        col.table = None
1060
1061
        exprs.append(expr)
1062
        cols.append(col)
1063 2408 aaronmk
1064 2688 aaronmk
    table = sql_gen.as_Table(table)
1065
    index = sql_gen.Table(str(sql_gen.Col(','.join(map(str, cols)), table)))
1066
1067 3005 aaronmk
    # Add index
1068
    while True:
1069
        str_ = 'CREATE'
1070
        if unique: str_ += ' UNIQUE'
1071
        str_ += ' INDEX '+index.to_str(db)+' ON '+table.to_str(db)+' ('+(
1072
            ', '.join((v.to_str(db) for v in exprs)))+')'
1073
1074
        try:
1075
            run_query(db, str_, recover=True, cacheable=True, log_level=3,
1076
                log_ignore_excs=(DuplicateException,))
1077
            break
1078
        except DuplicateException:
1079
            index.name = next_version(index.name)
1080
            # try again with next version of name
1081 2408 aaronmk
1082 2686 aaronmk
def add_pkey(db, table, cols=None, recover=None):
1083
    '''Adds a primary key.
1084
    @param cols [sql_gen.Col,...] The columns in the primary key.
1085
        Defaults to the first column in the table.
1086 2406 aaronmk
    @pre The table must not already have a primary key.
1087
    '''
1088
    table = sql_gen.as_Table(table)
1089 2686 aaronmk
    if cols == None: cols = [pkey(db, table, recover)]
1090
    col_strs = [sql_gen.to_name_only_col(v).to_str(db) for v in cols]
1091 2406 aaronmk
1092 2781 aaronmk
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD PRIMARY KEY ('
1093
        +(', '.join(col_strs))+')', recover=True, cacheable=True, log_level=3,
1094 2945 aaronmk
        log_ignore_excs=(DuplicateException,))
1095 2406 aaronmk
1096 2997 aaronmk
def add_not_null(db, col):
1097
    table = col.table
1098
    col = sql_gen.to_name_only_col(col)
1099
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ALTER COLUMN '
1100 3031 aaronmk
        +col.to_str(db)+' SET NOT NULL', cacheable=True, log_level=3)
1101 2997 aaronmk
1102
def add_index_col(db, col, suffix, expr, nullable=True):
1103 3000 aaronmk
    if sql_gen.index_col(col) != None: return # already has index col
1104 2997 aaronmk
1105
    new_col = sql_gen.suffixed_col(col, suffix)
1106
1107 3006 aaronmk
    # Add column
1108 3038 aaronmk
    new_typed_col = sql_gen.TypedCol(new_col.name, db.col_info(col).type)
1109 3045 aaronmk
    add_col(db, col.table, new_typed_col, comment='src: '+repr(col),
1110
        log_level=3)
1111 3037 aaronmk
    new_col.name = new_typed_col.name # propagate any renaming
1112 3006 aaronmk
1113 3064 aaronmk
    update(db, col.table, [(new_col, expr)], in_place=True, cacheable=True,
1114
        log_level=3)
1115 2997 aaronmk
    if not nullable: add_not_null(db, new_col)
1116
    add_index(db, new_col)
1117
1118 3000 aaronmk
    col.table.index_cols[col.name] = new_col
1119 2997 aaronmk
1120 3047 aaronmk
# Controls when ensure_not_null() will use index columns
1121
not_null_index_cols_min_rows = 0 # rows; initially always use index columns
1122
1123 2997 aaronmk
def ensure_not_null(db, col):
1124
    '''For params, see sql_gen.ensure_not_null()'''
1125
    expr = sql_gen.ensure_not_null(db, col)
1126
1127 3047 aaronmk
    # If a nullable column in a temp table, add separate index column instead.
1128
    # Note that for small datasources, this adds 6-25% to the total import time.
1129
    if (sql_gen.is_temp_col(col) and isinstance(expr, sql_gen.EnsureNotNull)
1130
        and table_row_count(db, col.table) >= not_null_index_cols_min_rows):
1131 2997 aaronmk
        add_index_col(db, col, '::NOT NULL', expr, nullable=False)
1132 3000 aaronmk
        expr = sql_gen.index_col(col)
1133 2997 aaronmk
1134
    return expr
1135
1136 2772 aaronmk
already_indexed = object() # tells add_indexes() the pkey has already been added
1137
1138
def add_indexes(db, table, has_pkey=True):
1139
    '''Adds an index on all columns in a table.
1140
    @param has_pkey bool|already_indexed Whether a pkey instead of a regular
1141
        index should be added on the first column.
1142
        * If already_indexed, the pkey is assumed to have already been added
1143
    '''
1144
    cols = table_cols(db, table)
1145
    if has_pkey:
1146
        if has_pkey is not already_indexed: add_pkey(db, table)
1147
        cols = cols[1:]
1148
    for col in cols: add_index(db, col, table)
1149
1150 3042 aaronmk
def add_col(db, table, col, comment=None, **kw_args):
1151 3039 aaronmk
    '''
1152 3042 aaronmk
    @param col TypedCol Name may be versioned, so be sure to propagate any
1153
        renaming back to any source column for the TypedCol.
1154
    @param comment None|str SQL comment used to distinguish columns of the same
1155
        name from each other when they contain different data, to allow the
1156
        ADD COLUMN query to be cached. If not set, query will not be cached.
1157 3039 aaronmk
    '''
1158 2934 aaronmk
    assert isinstance(col, sql_gen.TypedCol)
1159 3036 aaronmk
1160
    while True:
1161 3042 aaronmk
        str_ = 'ALTER TABLE '+table.to_str(db)+' ADD COLUMN '+col.to_str(db)
1162
        if comment != None: str_ += ' '+sql_gen.esc_comment(comment)
1163
1164 3036 aaronmk
        try:
1165 3042 aaronmk
            run_query(db, str_, recover=True, cacheable=True, **kw_args)
1166 3036 aaronmk
            break
1167
        except DuplicateException:
1168
            col.name = next_version(col.name)
1169
            # try again with next version of name
1170 2934 aaronmk
1171 2935 aaronmk
row_num_typed_col = sql_gen.TypedCol(row_num_col, 'serial', nullable=False,
1172
    constraints='PRIMARY KEY')
1173
1174 2086 aaronmk
def add_row_num(db, table):
1175 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
1176
    be the primary key.'''
1177 2935 aaronmk
    add_col(db, table, row_num_typed_col, log_level=3)
1178 2086 aaronmk
1179 2936 aaronmk
def cast_temp_col(db, type_, col, errors_table=None):
1180
    '''Like cast(), but creates a new column with the cast values if the input
1181
    is a column.
1182
    @return The new column or cast value
1183
    '''
1184
    def cast_(col): return cast(db, type_, col, errors_table)
1185
1186
    try: col = sql_gen.underlying_col(col)
1187
    except sql_gen.NoUnderlyingTableException: return sql_gen.wrap(cast_, col)
1188
1189
    table = col.table
1190
    new_col = sql_gen.Col(sql_gen.concat(col.name, '::'+type_), table, col.srcs)
1191
    expr = cast_(col)
1192 3035 aaronmk
1193
    # Add column
1194 3038 aaronmk
    new_typed_col = sql_gen.TypedCol(new_col.name, type_)
1195 3045 aaronmk
    add_col(db, table, new_typed_col, comment='src: '+repr(col))
1196 3037 aaronmk
    new_col.name = new_typed_col.name # propagate any renaming
1197 3035 aaronmk
1198 3064 aaronmk
    update(db, table, [(new_col, expr)], in_place=True, cacheable=True)
1199 2955 aaronmk
    add_index(db, new_col)
1200 2936 aaronmk
1201
    return new_col
1202
1203 2889 aaronmk
def drop_table(db, table):
1204
    table = sql_gen.as_Table(table)
1205
    return run_query(db, 'DROP TABLE IF EXISTS '+table.to_str(db)+' CASCADE')
1206
1207 2760 aaronmk
def create_table(db, table, cols, has_pkey=True, col_indexes=True):
1208 2675 aaronmk
    '''Creates a table.
1209 2681 aaronmk
    @param cols [sql_gen.TypedCol,...] The column names and types
1210
    @param has_pkey If set, the first column becomes the primary key.
1211 2760 aaronmk
    @param col_indexes bool|[ref]
1212
        * If True, indexes will be added on all non-pkey columns.
1213
        * If a list reference, [0] will be set to a function to do this.
1214
          This can be used to delay index creation until the table is populated.
1215 2675 aaronmk
    '''
1216
    table = sql_gen.as_Table(table)
1217
1218 2681 aaronmk
    if has_pkey:
1219
        cols[0] = pkey = copy.copy(cols[0]) # don't modify input!
1220 2872 aaronmk
        pkey.constraints = 'PRIMARY KEY'
1221 2681 aaronmk
1222 2675 aaronmk
    str_ = 'CREATE TABLE '+table.to_str(db)+' (\n'
1223 2681 aaronmk
    str_ += '\n, '.join(v.to_str(db) for v in cols)
1224
    str_ += '\n);\n'
1225 2675 aaronmk
    run_query(db, str_, cacheable=True, log_level=2)
1226 2760 aaronmk
1227
    # Add indexes
1228 2773 aaronmk
    if has_pkey: has_pkey = already_indexed
1229
    def add_indexes_(): add_indexes(db, table, has_pkey)
1230
    if isinstance(col_indexes, list): col_indexes[0] = add_indexes_ # defer
1231
    elif col_indexes: add_indexes_() # add now
1232 2675 aaronmk
1233 3068 aaronmk
def analyze(db, table):
1234
    table = sql_gen.as_Table(table)
1235
    run_query(db, 'ANALYZE '+table.to_str(db), log_level=3)
1236
1237 3074 aaronmk
def autoanalyze(db, table):
1238
    if db.autoanalyze: analyze(db, table)
1239
1240 2684 aaronmk
def vacuum(db, table):
1241
    table = sql_gen.as_Table(table)
1242
    db.with_autocommit(lambda: run_query(db, 'VACUUM ANALYZE '+table.to_str(db),
1243
        log_level=3))
1244
1245 2970 aaronmk
def truncate(db, table, schema='public', **kw_args):
1246
    '''For params, see run_query()'''
1247 2777 aaronmk
    table = sql_gen.as_Table(table, schema)
1248 2970 aaronmk
    return run_query(db, 'TRUNCATE '+table.to_str(db)+' CASCADE', **kw_args)
1249 2732 aaronmk
1250 2965 aaronmk
def empty_temp(db, tables):
1251 2972 aaronmk
    if db.debug_temp: return # leave temp tables there for debugging
1252 2965 aaronmk
    tables = lists.mk_seq(tables)
1253 2971 aaronmk
    for table in tables: truncate(db, table, log_level=3)
1254 2965 aaronmk
1255 2729 aaronmk
def tables(db, schema_like='public', table_like='%', exact=False):
1256
    if exact: compare = '='
1257
    else: compare = 'LIKE'
1258
1259 1849 aaronmk
    module = util.root_module(db.db)
1260 832 aaronmk
    if module == 'psycopg2':
1261 2779 aaronmk
        conds = [('schemaname', sql_gen.CompareCond(schema_like, compare)),
1262
            ('tablename', sql_gen.CompareCond(table_like, compare))]
1263
        return values(select(db, 'pg_tables', ['tablename'], conds,
1264
            order_by='tablename', log_level=4))
1265 1968 aaronmk
    elif module == 'MySQLdb':
1266 2779 aaronmk
        return values(run_query(db, 'SHOW TABLES LIKE '+db.esc_value(table_like)
1267
            , cacheable=True, log_level=4))
1268 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
1269 830 aaronmk
1270 2726 aaronmk
def table_exists(db, table):
1271
    table = sql_gen.as_Table(table)
1272 2730 aaronmk
    return list(tables(db, table.schema, table.name, exact=True)) != []
1273 2726 aaronmk
1274 2913 aaronmk
def function_exists(db, function):
1275
    function = sql_gen.as_Function(function)
1276
1277
    info_table = sql_gen.Table('routines', 'information_schema')
1278
    conds = [('routine_name', function.name)]
1279
    schema = function.schema
1280
    if schema != None: conds.append(('routine_schema', schema))
1281 2914 aaronmk
    # Exclude trigger functions, since they cannot be called directly
1282
    conds.append(('data_type', sql_gen.CompareCond('trigger', '!=')))
1283 2913 aaronmk
1284
    return list(values(select(db, info_table, ['routine_name'], conds,
1285
        order_by='routine_schema', limit=1, log_level=4))) != []
1286
        # TODO: order_by search_path schema order
1287
1288 2726 aaronmk
def errors_table(db, table, if_exists=True):
1289
    '''
1290
    @param if_exists If set, returns None if the errors table doesn't exist
1291
    @return None|sql_gen.Table
1292
    '''
1293
    table = sql_gen.as_Table(table)
1294
    if table.srcs != (): table = table.srcs[0]
1295
1296
    errors_table = sql_gen.suffixed_table(table, '.errors')
1297
    if if_exists and not table_exists(db, errors_table): return None
1298
    return errors_table
1299
1300 833 aaronmk
##### Database management
1301
1302 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
1303
    '''For kw_args, see tables()'''
1304
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
1305 833 aaronmk
1306 2115 aaronmk
##### Data cleanup
1307
1308 2290 aaronmk
def cleanup_table(db, table, cols):
1309 2677 aaronmk
    table = sql_gen.as_Table(table)
1310
    cols = map(sql_gen.as_Col, cols)
1311 2115 aaronmk
1312 2677 aaronmk
    expr = ('nullif(nullif(trim(both from %s), '+db.esc_value('')+'), '
1313
        +db.esc_value(r'\N')+')')
1314
    changes = [(v, sql_gen.CustomCode(expr % v.to_str(db)))
1315
        for v in cols]
1316 2115 aaronmk
1317 3066 aaronmk
    update(db, table, changes, in_place=True)