Project

General

Profile

1 11 aaronmk
# Database access
2
3 1869 aaronmk
import copy
4 2127 aaronmk
import operator
5 11 aaronmk
import re
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 1889 aaronmk
from Proxy import Proxy
13 1872 aaronmk
import rand
14 2217 aaronmk
import sql_gen
15 862 aaronmk
import strings
16 131 aaronmk
import util
17 11 aaronmk
18 832 aaronmk
##### Exceptions
19
20 2170 aaronmk
def get_cur_query(cur, input_query=None, input_params=None):
21 2168 aaronmk
    raw_query = None
22
    if hasattr(cur, 'query'): raw_query = cur.query
23
    elif hasattr(cur, '_last_executed'): raw_query = cur._last_executed
24 2170 aaronmk
25
    if raw_query != None: return raw_query
26 2371 aaronmk
    else: return '[input] '+strings.ustr(input_query)+' % '+repr(input_params)
27 14 aaronmk
28 2170 aaronmk
def _add_cursor_info(e, *args, **kw_args):
29
    '''For params, see get_cur_query()'''
30
    exc.add_msg(e, 'query: '+str(get_cur_query(*args, **kw_args)))
31 135 aaronmk
32 300 aaronmk
class DbException(exc.ExceptionWithCause):
33 14 aaronmk
    def __init__(self, msg, cause=None, cur=None):
34 2145 aaronmk
        exc.ExceptionWithCause.__init__(self, msg, cause, cause_newline=True)
35 14 aaronmk
        if cur != None: _add_cursor_info(self, cur)
36
37 2143 aaronmk
class ExceptionWithName(DbException):
38
    def __init__(self, name, cause=None):
39 2484 aaronmk
        DbException.__init__(self, 'for name: '+strings.as_tt(str(name)), cause)
40 2143 aaronmk
        self.name = name
41 360 aaronmk
42 2240 aaronmk
class ExceptionWithNameValue(DbException):
43
    def __init__(self, name, value, cause=None):
44 2484 aaronmk
        DbException.__init__(self, 'for name: '+strings.as_tt(str(name))
45
            +'; value: '+strings.as_tt(repr(value)), cause)
46 2240 aaronmk
        self.name = name
47
        self.value = value
48
49 2306 aaronmk
class ConstraintException(DbException):
50
    def __init__(self, name, cols, cause=None):
51 2484 aaronmk
        DbException.__init__(self, 'Violated '+strings.as_tt(name)
52
            +' constraint on columns: '+strings.as_tt(', '.join(cols)), cause)
53 2306 aaronmk
        self.name = name
54 468 aaronmk
        self.cols = cols
55 11 aaronmk
56 2523 aaronmk
class MissingCastException(DbException):
57
    def __init__(self, type_, col, cause=None):
58
        DbException.__init__(self, 'Missing cast to type '+strings.as_tt(type_)
59
            +' on column: '+strings.as_tt(col), cause)
60
        self.type = type_
61
        self.col = col
62
63 2143 aaronmk
class NameException(DbException): pass
64
65 2306 aaronmk
class DuplicateKeyException(ConstraintException): pass
66 13 aaronmk
67 2306 aaronmk
class NullValueException(ConstraintException): pass
68 13 aaronmk
69 2240 aaronmk
class FunctionValueException(ExceptionWithNameValue): pass
70 2239 aaronmk
71 2143 aaronmk
class DuplicateTableException(ExceptionWithName): pass
72
73 2188 aaronmk
class DuplicateFunctionException(ExceptionWithName): pass
74
75 89 aaronmk
class EmptyRowException(DbException): pass
76
77 865 aaronmk
##### Warnings
78
79
class DbWarning(UserWarning): pass
80
81 1930 aaronmk
##### Result retrieval
82
83
def col_names(cur): return (col[0] for col in cur.description)
84
85
def rows(cur): return iter(lambda: cur.fetchone(), None)
86
87
def consume_rows(cur):
88
    '''Used to fetch all rows so result will be cached'''
89
    iters.consume_iter(rows(cur))
90
91
def next_row(cur): return rows(cur).next()
92
93
def row(cur):
94
    row_ = next_row(cur)
95
    consume_rows(cur)
96
    return row_
97
98
def next_value(cur): return next_row(cur)[0]
99
100
def value(cur): return row(cur)[0]
101
102
def values(cur): return iters.func_iter(lambda: next_value(cur))
103
104
def value_or_none(cur):
105
    try: return value(cur)
106
    except StopIteration: return None
107
108 2762 aaronmk
##### Escaping
109 2101 aaronmk
110 2573 aaronmk
def esc_name_by_module(module, name):
111
    if module == 'psycopg2' or module == None: quote = '"'
112 2101 aaronmk
    elif module == 'MySQLdb': quote = '`'
113
    else: raise NotImplementedError("Can't escape name for "+module+' database')
114 2500 aaronmk
    return sql_gen.esc_name(name, quote)
115 2101 aaronmk
116
def esc_name_by_engine(engine, name, **kw_args):
117
    return esc_name_by_module(db_engines[engine][0], name, **kw_args)
118
119
def esc_name(db, name, **kw_args):
120
    return esc_name_by_module(util.root_module(db.db), name, **kw_args)
121
122
def qual_name(db, schema, table):
123
    def esc_name_(name): return esc_name(db, name)
124
    table = esc_name_(table)
125
    if schema != None: return esc_name_(schema)+'.'+table
126
    else: return table
127
128 1869 aaronmk
##### Database connections
129 1849 aaronmk
130 2097 aaronmk
db_config_names = ['engine', 'host', 'user', 'password', 'database', 'schemas']
131 1926 aaronmk
132 1869 aaronmk
db_engines = {
133
    'MySQL': ('MySQLdb', {'password': 'passwd', 'database': 'db'}),
134
    'PostgreSQL': ('psycopg2', {}),
135
}
136
137
DatabaseErrors_set = set([DbException])
138
DatabaseErrors = tuple(DatabaseErrors_set)
139
140
def _add_module(module):
141
    DatabaseErrors_set.add(module.DatabaseError)
142
    global DatabaseErrors
143
    DatabaseErrors = tuple(DatabaseErrors_set)
144
145
def db_config_str(db_config):
146
    return db_config['engine']+' database '+db_config['database']
147
148 1909 aaronmk
def _query_lookup(query, params): return (query, dicts.make_hashable(params))
149 1894 aaronmk
150 2448 aaronmk
log_debug_none = lambda msg, level=2: None
151 1901 aaronmk
152 1849 aaronmk
class DbConn:
153 2190 aaronmk
    def __init__(self, db_config, serializable=True, autocommit=False,
154
        caching=True, log_debug=log_debug_none):
155 1869 aaronmk
        self.db_config = db_config
156
        self.serializable = serializable
157 2190 aaronmk
        self.autocommit = autocommit
158
        self.caching = caching
159 1901 aaronmk
        self.log_debug = log_debug
160 2193 aaronmk
        self.debug = log_debug != log_debug_none
161 1869 aaronmk
162
        self.__db = None
163 1889 aaronmk
        self.query_results = {}
164 2139 aaronmk
        self._savepoint = 0
165 2671 aaronmk
        self._notices_seen = set()
166 1869 aaronmk
167
    def __getattr__(self, name):
168
        if name == '__dict__': raise Exception('getting __dict__')
169
        if name == 'db': return self._db()
170
        else: raise AttributeError()
171
172
    def __getstate__(self):
173
        state = copy.copy(self.__dict__) # shallow copy
174 1915 aaronmk
        state['log_debug'] = None # don't pickle the debug callback
175 1869 aaronmk
        state['_DbConn__db'] = None # don't pickle the connection
176
        return state
177
178 2165 aaronmk
    def connected(self): return self.__db != None
179
180 1869 aaronmk
    def _db(self):
181
        if self.__db == None:
182
            # Process db_config
183
            db_config = self.db_config.copy() # don't modify input!
184 2097 aaronmk
            schemas = db_config.pop('schemas', None)
185 1869 aaronmk
            module_name, mappings = db_engines[db_config.pop('engine')]
186
            module = __import__(module_name)
187
            _add_module(module)
188
            for orig, new in mappings.iteritems():
189
                try: util.rename_key(db_config, orig, new)
190
                except KeyError: pass
191
192
            # Connect
193
            self.__db = module.connect(**db_config)
194
195
            # Configure connection
196 2676 aaronmk
            if self.serializable and not self.autocommit: self.run_query(
197 2678 aaronmk
                'SET TRANSACTION ISOLATION LEVEL SERIALIZABLE', log_level=4)
198 2101 aaronmk
            if schemas != None:
199
                schemas_ = ''.join((esc_name(self, s)+', '
200
                    for s in schemas.split(',')))
201 2676 aaronmk
                self.run_query("SELECT set_config('search_path', \
202
%s || current_setting('search_path'), false)", [schemas_], log_level=3)
203 1869 aaronmk
204
        return self.__db
205 1889 aaronmk
206 1891 aaronmk
    class DbCursor(Proxy):
207 1927 aaronmk
        def __init__(self, outer):
208 1891 aaronmk
            Proxy.__init__(self, outer.db.cursor())
209 2191 aaronmk
            self.outer = outer
210 1927 aaronmk
            self.query_results = outer.query_results
211 1894 aaronmk
            self.query_lookup = None
212 1891 aaronmk
            self.result = []
213 1889 aaronmk
214 1894 aaronmk
        def execute(self, query, params=None):
215 2763 aaronmk
            if params == [] or params == (): params = None
216
                # None turns off the mogrifier
217 2762 aaronmk
218 1930 aaronmk
            self._is_insert = query.upper().find('INSERT') >= 0
219 1894 aaronmk
            self.query_lookup = _query_lookup(query, params)
220 2148 aaronmk
            try:
221 2191 aaronmk
                try:
222 2763 aaronmk
                    cur = self.inner.execute(query, params)
223 2191 aaronmk
                    self.outer.do_autocommit()
224 2648 aaronmk
                finally: self.query = get_cur_query(self.inner, query, params)
225 1904 aaronmk
            except Exception, e:
226 2170 aaronmk
                _add_cursor_info(e, self, query, params)
227 1904 aaronmk
                self.result = e # cache the exception as the result
228
                self._cache_result()
229
                raise
230 1930 aaronmk
            # Fetch all rows so result will be cached
231
            if self.rowcount == 0 and not self._is_insert: consume_rows(self)
232 2762 aaronmk
            return cur
233 1894 aaronmk
234 1891 aaronmk
        def fetchone(self):
235
            row = self.inner.fetchone()
236 1899 aaronmk
            if row != None: self.result.append(row)
237
            # otherwise, fetched all rows
238 1904 aaronmk
            else: self._cache_result()
239
            return row
240
241
        def _cache_result(self):
242 1906 aaronmk
            # For inserts, only cache exceptions since inserts are not
243
            # idempotent, but an invalid insert will always be invalid
244 1930 aaronmk
            if self.query_results != None and (not self._is_insert
245 1906 aaronmk
                or isinstance(self.result, Exception)):
246
247 1894 aaronmk
                assert self.query_lookup != None
248 1916 aaronmk
                self.query_results[self.query_lookup] = self.CacheCursor(
249
                    util.dict_subset(dicts.AttrsDictView(self),
250
                    ['query', 'result', 'rowcount', 'description']))
251 1906 aaronmk
252 1916 aaronmk
        class CacheCursor:
253
            def __init__(self, cached_result): self.__dict__ = cached_result
254
255 1927 aaronmk
            def execute(self, *args, **kw_args):
256 1916 aaronmk
                if isinstance(self.result, Exception): raise self.result
257
                # otherwise, result is a rows list
258
                self.iter = iter(self.result)
259
260
            def fetchone(self):
261
                try: return self.iter.next()
262
                except StopIteration: return None
263 1891 aaronmk
264 2212 aaronmk
    def esc_value(self, value):
265 2663 aaronmk
        try: str_ = self.mogrify('%s', [value])
266
        except NotImplementedError, e:
267
            module = util.root_module(self.db)
268
            if module == 'MySQLdb':
269
                import _mysql
270
                str_ = _mysql.escape_string(value)
271
            else: raise e
272 2374 aaronmk
        return strings.to_unicode(str_)
273 2212 aaronmk
274 2347 aaronmk
    def esc_name(self, name): return esc_name(self, name) # calls global func
275
276 2665 aaronmk
    def can_mogrify(self):
277 2663 aaronmk
        module = util.root_module(self.db)
278 2665 aaronmk
        return module == 'psycopg2'
279 2663 aaronmk
280 2665 aaronmk
    def mogrify(self, query, params=None):
281
        if self.can_mogrify(): return self.db.cursor().mogrify(query, params)
282
        else: raise NotImplementedError("Can't mogrify query")
283
284 2671 aaronmk
    def print_notices(self):
285 2725 aaronmk
        if hasattr(self.db, 'notices'):
286
            for msg in self.db.notices:
287
                if msg not in self._notices_seen:
288
                    self._notices_seen.add(msg)
289
                    self.log_debug(msg, level=2)
290 2671 aaronmk
291 2445 aaronmk
    def run_query(self, query, params=None, cacheable=False, log_level=2,
292 2464 aaronmk
        debug_msg_ref=None):
293 2445 aaronmk
        '''
294 2464 aaronmk
        @param log_ignore_excs The log_level will be increased by 2 if the query
295
            throws one of these exceptions.
296 2664 aaronmk
        @param debug_msg_ref If specified, the log message will be returned in
297
            this instead of being output. This allows you to filter log messages
298
            depending on the result of the query.
299 2445 aaronmk
        '''
300 2167 aaronmk
        assert query != None
301
302 2047 aaronmk
        if not self.caching: cacheable = False
303 1903 aaronmk
        used_cache = False
304 2664 aaronmk
305
        def log_msg(query):
306
            if used_cache: cache_status = 'cache hit'
307
            elif cacheable: cache_status = 'cache miss'
308
            else: cache_status = 'non-cacheable'
309
            return 'DB query: '+cache_status+':\n'+strings.as_code(query, 'SQL')
310
311 1903 aaronmk
        try:
312 1927 aaronmk
            # Get cursor
313
            if cacheable:
314
                query_lookup = _query_lookup(query, params)
315
                try:
316
                    cur = self.query_results[query_lookup]
317
                    used_cache = True
318
                except KeyError: cur = self.DbCursor(self)
319
            else: cur = self.db.cursor()
320
321 2664 aaronmk
            # Log query
322
            if self.debug and debug_msg_ref == None: # log before running
323
                self.log_debug(log_msg(query), log_level)
324
325 1927 aaronmk
            # Run query
326 2148 aaronmk
            cur.execute(query, params)
327 1903 aaronmk
        finally:
328 2671 aaronmk
            self.print_notices()
329 2664 aaronmk
            if self.debug and debug_msg_ref != None: # return after running
330
                debug_msg_ref[0] = log_msg(str(get_cur_query(cur, query,
331
                    params)))
332 1903 aaronmk
333
        return cur
334 1914 aaronmk
335
    def is_cached(self, query, params=None):
336
        return _query_lookup(query, params) in self.query_results
337 2139 aaronmk
338 2683 aaronmk
    def with_autocommit(self, func, autocommit=True):
339
        prev_autocommit = self.db.autocommit
340
        self.db.autocommit = autocommit
341
        try: return func()
342
        finally: self.db.autocommit = prev_autocommit
343
344 2139 aaronmk
    def with_savepoint(self, func):
345 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
346 2443 aaronmk
        self.run_query('SAVEPOINT '+savepoint, log_level=4)
347 2139 aaronmk
        self._savepoint += 1
348 2683 aaronmk
        try:
349 2139 aaronmk
            try: return_val = func()
350
            finally:
351
                self._savepoint -= 1
352
                assert self._savepoint >= 0
353
        except:
354 2443 aaronmk
            self.run_query('ROLLBACK TO SAVEPOINT '+savepoint, log_level=4)
355 2139 aaronmk
            raise
356
        else:
357 2443 aaronmk
            self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
358 2191 aaronmk
            self.do_autocommit()
359 2139 aaronmk
            return return_val
360 2191 aaronmk
361
    def do_autocommit(self):
362
        '''Autocommits if outside savepoint'''
363
        assert self._savepoint >= 0
364
        if self.autocommit and self._savepoint == 0:
365 2668 aaronmk
            self.log_debug('Autocommitting')
366 2191 aaronmk
            self.db.commit()
367 2643 aaronmk
368
    def col_default(self, col):
369
        table = sql_gen.Table('columns', 'information_schema')
370
371
        conds = [('table_name', col.table.name), ('column_name', col.name)]
372
        schema = col.table.schema
373
        if schema != None: conds.append(('table_schema', schema))
374
375 2660 aaronmk
        return sql_gen.as_Code(value(select(self, table, ['column_default'],
376
            conds, order_by='table_schema', limit=1, log_level=3)))
377 2643 aaronmk
            # TODO: order_by search_path schema order
378 1849 aaronmk
379 1869 aaronmk
connect = DbConn
380
381 832 aaronmk
##### Querying
382
383 1894 aaronmk
def run_raw_query(db, *args, **kw_args):
384 2085 aaronmk
    '''For params, see DbConn.run_query()'''
385 1894 aaronmk
    return db.run_query(*args, **kw_args)
386 11 aaronmk
387 2068 aaronmk
def mogrify(db, query, params):
388
    module = util.root_module(db.db)
389
    if module == 'psycopg2': return db.db.cursor().mogrify(query, params)
390
    else: raise NotImplementedError("Can't mogrify query for "+module+
391
        ' database')
392
393 832 aaronmk
##### Recoverable querying
394 15 aaronmk
395 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
396 11 aaronmk
397 2464 aaronmk
def run_query(db, query, params=None, recover=None, cacheable=False,
398
    log_level=2, log_ignore_excs=None, **kw_args):
399 2441 aaronmk
    '''For params, see run_raw_query()'''
400 830 aaronmk
    if recover == None: recover = False
401 2464 aaronmk
    if log_ignore_excs == None: log_ignore_excs = ()
402
    log_ignore_excs = tuple(log_ignore_excs)
403 830 aaronmk
404 2666 aaronmk
    debug_msg_ref = None # usually, db.run_query() logs query before running it
405
    # But if filtering with log_ignore_excs, wait until after exception parsing
406
    if log_ignore_excs != () or not db.can_mogrify(): debug_msg_ref = [None]
407
408 2148 aaronmk
    try:
409 2464 aaronmk
        try:
410
            def run(): return run_raw_query(db, query, params, cacheable,
411
                log_level, debug_msg_ref, **kw_args)
412
            if recover and not db.is_cached(query, params):
413
                return with_savepoint(db, run)
414
            else: return run() # don't need savepoint if cached
415
        except Exception, e:
416
            if not recover: raise # need savepoint to run index_cols()
417
            msg = exc.str_(e)
418
419
            match = re.search(r'duplicate key value violates unique constraint '
420 2493 aaronmk
                r'"((_?[^\W_]+)_.+?)"', msg)
421 2464 aaronmk
            if match:
422
                constraint, table = match.groups()
423
                try: cols = index_cols(db, table, constraint)
424
                except NotImplementedError: raise e
425
                else: raise DuplicateKeyException(constraint, cols, e)
426
427 2493 aaronmk
            match = re.search(r'null value in column "(.+?)" violates not-null'
428 2464 aaronmk
                r' constraint', msg)
429
            if match: raise NullValueException('NOT NULL', [match.group(1)], e)
430
431
            match = re.search(r'\b(?:invalid input (?:syntax|value)\b.*?'
432
                r'|date/time field value out of range): "(.+?)"\n'
433 2535 aaronmk
                r'(?:(?s).*?)\bfunction "(.+?)"', msg)
434 2464 aaronmk
            if match:
435
                value, name = match.groups()
436
                raise FunctionValueException(name, strings.to_unicode(value), e)
437
438 2526 aaronmk
            match = re.search(r'column "(.+?)" is of type (.+?) but expression '
439 2523 aaronmk
                r'is of type', msg)
440
            if match:
441
                col, type_ = match.groups()
442
                raise MissingCastException(type_, col, e)
443
444 2493 aaronmk
            match = re.search(r'relation "(.+?)" already exists', msg)
445 2464 aaronmk
            if match: raise DuplicateTableException(match.group(1), e)
446
447 2493 aaronmk
            match = re.search(r'function "(.+?)" already exists', msg)
448 2464 aaronmk
            if match: raise DuplicateFunctionException(match.group(1), e)
449
450
            raise # no specific exception raised
451
    except log_ignore_excs:
452
        log_level += 2
453
        raise
454
    finally:
455 2666 aaronmk
        if debug_msg_ref != None and debug_msg_ref[0] != None:
456
            db.log_debug(debug_msg_ref[0], log_level)
457 830 aaronmk
458 832 aaronmk
##### Basic queries
459
460 2153 aaronmk
def next_version(name):
461 2163 aaronmk
    version = 1 # first existing name was version 0
462 2586 aaronmk
    match = re.match(r'^(.*)#(\d+)$', name)
463 2153 aaronmk
    if match:
464 2586 aaronmk
        name, version = match.groups()
465
        version = int(version)+1
466 2588 aaronmk
    return sql_gen.add_suffix(name, '#'+str(version))
467 2153 aaronmk
468 2386 aaronmk
def run_query_into(db, query, params, into=None, *args, **kw_args):
469 2085 aaronmk
    '''Outputs a query to a temp table.
470
    For params, see run_query().
471
    '''
472 2386 aaronmk
    if into == None: return run_query(db, query, params, *args, **kw_args)
473 2085 aaronmk
    else: # place rows in temp table
474 2386 aaronmk
        assert isinstance(into, sql_gen.Table)
475 2385 aaronmk
476 2153 aaronmk
        kw_args['recover'] = True
477 2464 aaronmk
        kw_args.setdefault('log_ignore_excs', (DuplicateTableException,))
478 2440 aaronmk
479 2468 aaronmk
        temp = not db.autocommit # tables are permanent in autocommit mode
480 2440 aaronmk
        # "temporary tables cannot specify a schema name", so remove schema
481
        if temp: into.schema = None
482
483 2153 aaronmk
        while True:
484
            try:
485 2194 aaronmk
                create_query = 'CREATE'
486 2440 aaronmk
                if temp: create_query += ' TEMP'
487 2467 aaronmk
                create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
488 2194 aaronmk
489
                return run_query(db, create_query, params, *args, **kw_args)
490 2153 aaronmk
                    # CREATE TABLE AS sets rowcount to # rows in query
491
            except DuplicateTableException, e:
492 2386 aaronmk
                into.name = next_version(into.name)
493 2153 aaronmk
                # try again with next version of name
494 2085 aaronmk
495 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
496
497 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
498
499 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
500 2293 aaronmk
    start=None, order_by=order_by_pkey, default_table=None):
501 1981 aaronmk
    '''
502 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
503 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
504 1981 aaronmk
    @param fields Use None to select all fields in the table
505 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
506 2379 aaronmk
        * container can be any iterable type
507 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
508
        * compare_right_side: sql_gen.ValueCond|literal value
509 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
510
        use all columns
511 2054 aaronmk
    @return tuple(query, params)
512 1981 aaronmk
    '''
513 2315 aaronmk
    # Parse tables param
514 2121 aaronmk
    if not lists.is_seq(tables): tables = [tables]
515 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
516 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
517 2121 aaronmk
518 2315 aaronmk
    # Parse other params
519 2376 aaronmk
    if conds == None: conds = []
520 2650 aaronmk
    elif dicts.is_dict(conds): conds = conds.items()
521 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
522 135 aaronmk
    assert limit == None or type(limit) == int
523 865 aaronmk
    assert start == None or type(start) == int
524 2315 aaronmk
    if order_by is order_by_pkey:
525
        if distinct_on != []: order_by = None
526
        else: order_by = pkey(db, table0, recover=True)
527 865 aaronmk
528 2315 aaronmk
    query = 'SELECT'
529 2056 aaronmk
530 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
531 2056 aaronmk
532 2200 aaronmk
    # DISTINCT ON columns
533 2233 aaronmk
    if distinct_on != []:
534 2467 aaronmk
        query += '\nDISTINCT'
535 2254 aaronmk
        if distinct_on is not distinct_on_all:
536 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
537
538
    # Columns
539 2467 aaronmk
    query += '\n'
540 1135 aaronmk
    if fields == None: query += '*'
541 2479 aaronmk
    else: query += '\n, '.join(map(parse_col, fields))
542 2200 aaronmk
543
    # Main table
544 2467 aaronmk
    query += '\nFROM '+table0.to_str(db)
545 865 aaronmk
546 2122 aaronmk
    # Add joins
547 2271 aaronmk
    left_table = table0
548 2263 aaronmk
    for join_ in tables:
549
        table = join_.table
550 2238 aaronmk
551 2343 aaronmk
        # Parse special values
552
        if join_.type_ is sql_gen.filter_out: # filter no match
553 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
554
                None))
555 2343 aaronmk
556 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
557 2122 aaronmk
558
        left_table = table
559
560 865 aaronmk
    missing = True
561 2376 aaronmk
    if conds != []:
562 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
563
        else: whitespace = '\n'
564 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
565
            .to_str(db) for l, r in conds], 'WHERE')
566 865 aaronmk
        missing = False
567 2227 aaronmk
    if order_by != None:
568 2467 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
569
    if limit != None: query += '\nLIMIT '+str(limit); missing = False
570 865 aaronmk
    if start != None:
571 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
572 865 aaronmk
        missing = False
573
    if missing: warnings.warn(DbWarning(
574
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
575
576 2315 aaronmk
    return (query, [])
577 11 aaronmk
578 2054 aaronmk
def select(db, *args, **kw_args):
579
    '''For params, see mk_select() and run_query()'''
580
    recover = kw_args.pop('recover', None)
581
    cacheable = kw_args.pop('cacheable', True)
582 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
583 2054 aaronmk
584
    query, params = mk_select(db, *args, **kw_args)
585 2442 aaronmk
    return run_query(db, query, params, recover, cacheable, log_level=log_level)
586 2054 aaronmk
587 2066 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, params=None,
588 2292 aaronmk
    returning=None, embeddable=False):
589 1960 aaronmk
    '''
590
    @param returning str|None An inserted column (such as pkey) to return
591 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
592 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
593
        query will be fully cached, not just if it raises an exception.
594 1960 aaronmk
    '''
595 2754 aaronmk
    table = sql_gen.remove_table_rename(sql_gen.as_Table(table))
596 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
597 2646 aaronmk
    if cols != None:
598
        cols = [sql_gen.to_name_only_col(v, table).to_str(db) for v in cols]
599 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
600 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
601 2063 aaronmk
602
    # Build query
603 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
604
    query = first_line
605 2467 aaronmk
    if cols != None: query += '\n('+', '.join(cols)+')'
606
    query += '\n'+select_query
607 2063 aaronmk
608
    if returning != None:
609 2743 aaronmk
        query += '\nRETURNING '+sql_gen.to_name_only_col(returning).to_str(db)
610 2063 aaronmk
611 2070 aaronmk
    if embeddable:
612 2327 aaronmk
        assert returning != None
613
614 2070 aaronmk
        # Create function
615 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
616 2327 aaronmk
        return_type = 'SETOF '+returning.to_str(db)+'%TYPE'
617 2189 aaronmk
        while True:
618
            try:
619 2698 aaronmk
                function = sql_gen.TempFunction(function_name, db.autocommit)
620 2194 aaronmk
621 2189 aaronmk
                function_query = '''\
622 2698 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''()
623 2467 aaronmk
RETURNS '''+return_type+'''
624
LANGUAGE sql
625
AS $$
626
'''+mogrify(db, query, params)+''';
627
$$;
628 2070 aaronmk
'''
629 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
630 2464 aaronmk
                    log_ignore_excs=(DuplicateFunctionException,))
631 2189 aaronmk
                break # this version was successful
632
            except DuplicateFunctionException, e:
633
                function_name = next_version(function_name)
634
                # try again with next version of name
635 2070 aaronmk
636 2337 aaronmk
        # Return query that uses function
637 2698 aaronmk
        func_table = sql_gen.NamedTable('f', sql_gen.FunctionCall(function),
638 2742 aaronmk
            [returning]) # AS clause requires function alias
639 2337 aaronmk
        return mk_select(db, func_table, start=0, order_by=None)
640 2070 aaronmk
641 2066 aaronmk
    return (query, params)
642
643
def insert_select(db, *args, **kw_args):
644 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
645 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
646
        values in
647 2072 aaronmk
    '''
648 2386 aaronmk
    into = kw_args.pop('into', None)
649
    if into != None: kw_args['embeddable'] = True
650 2066 aaronmk
    recover = kw_args.pop('recover', None)
651
    cacheable = kw_args.pop('cacheable', True)
652 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
653 2066 aaronmk
654
    query, params = mk_insert_select(db, *args, **kw_args)
655 2386 aaronmk
    return run_query_into(db, query, params, into, recover=recover,
656 2673 aaronmk
        cacheable=cacheable, log_level=log_level)
657 2063 aaronmk
658 2738 aaronmk
default = sql_gen.default # tells insert() to use the default value for a column
659 2066 aaronmk
660 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
661 2085 aaronmk
    '''For params, see insert_select()'''
662 1960 aaronmk
    if lists.is_seq(row): cols = None
663
    else:
664
        cols = row.keys()
665
        row = row.values()
666 2738 aaronmk
    row = list(row) # ensure that "== []" works
667 1960 aaronmk
668 2738 aaronmk
    if row == []: query = None
669
    else: query = sql_gen.Values(row).to_str(db)
670 1961 aaronmk
671 2738 aaronmk
    return insert_select(db, table, cols, query, [], *args, **kw_args)
672 11 aaronmk
673 2402 aaronmk
def mk_update(db, table, changes=None, cond=None):
674
    '''
675
    @param changes [(col, new_value),...]
676
        * container can be any iterable type
677
        * col: sql_gen.Code|str (for col name)
678
        * new_value: sql_gen.Code|literal value
679
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
680
    @return str query
681
    '''
682
    query = 'UPDATE '+sql_gen.as_Table(table).to_str(db)+'\nSET\n'
683 2405 aaronmk
    query += ',\n'.join((sql_gen.to_name_only_col(col, table).to_str(db)+' = '
684 2402 aaronmk
        +sql_gen.as_Value(new_value).to_str(db) for col, new_value in changes))
685 2467 aaronmk
    if cond != None: query += '\nWHERE\n'+cond.to_str(db)
686 2402 aaronmk
687
    return query
688
689
def update(db, *args, **kw_args):
690
    '''For params, see mk_update() and run_query()'''
691
    recover = kw_args.pop('recover', None)
692
693
    return run_query(db, mk_update(db, *args, **kw_args), [], recover)
694
695 135 aaronmk
def last_insert_id(db):
696 1849 aaronmk
    module = util.root_module(db.db)
697 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
698
    elif module == 'MySQLdb': return db.insert_id()
699
    else: return None
700 13 aaronmk
701 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
702 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
703 2415 aaronmk
    to names that will be distinct among the columns' tables.
704 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
705 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
706
    @param into The table for the new columns.
707 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
708
        columns will be included in the mapping even if they are not in cols.
709
        The tables of the provided Col objects will be changed to into, so make
710
        copies of them if you want to keep the original tables.
711
    @param as_items Whether to return a list of dict items instead of a dict
712 2383 aaronmk
    @return dict(orig_col=new_col, ...)
713
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
714 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
715
        * All mappings use the into table so its name can easily be
716 2383 aaronmk
          changed for all columns at once
717
    '''
718 2415 aaronmk
    cols = lists.uniqify(cols)
719
720 2394 aaronmk
    items = []
721 2389 aaronmk
    for col in preserve:
722 2390 aaronmk
        orig_col = copy.copy(col)
723 2392 aaronmk
        col.table = into
724 2394 aaronmk
        items.append((orig_col, col))
725
    preserve = set(preserve)
726
    for col in cols:
727 2716 aaronmk
        if col not in preserve:
728
            items.append((col, sql_gen.Col(str(col), into, col.srcs)))
729 2394 aaronmk
730
    if not as_items: items = dict(items)
731
    return items
732 2383 aaronmk
733 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
734 2391 aaronmk
    '''For params, see mk_flatten_mapping()
735
    @return See return value of mk_flatten_mapping()
736
    '''
737 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
738
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
739 2391 aaronmk
    run_query_into(db, *mk_select(db, joins, cols, limit=limit, start=start),
740 2392 aaronmk
        into=into)
741 2394 aaronmk
    return dict(items)
742 2391 aaronmk
743 2751 aaronmk
def mk_track_data_error(db, errors_table, cols, value, error_code, error):
744
    cols = map(sql_gen.to_name_only_col, cols)
745
746
    columns_cols = ['column']
747
    columns = sql_gen.NamedValues('columns', columns_cols,
748
        [[c.name] for c in cols])
749
    values_cols = ['value', 'error_code', 'error']
750
    values = sql_gen.NamedValues('values', values_cols,
751
        [value, error_code, error])
752
753
    select_cols = columns_cols+values_cols
754
    name_only_cols = map(sql_gen.to_name_only_col, select_cols)
755 2755 aaronmk
    errors_table = sql_gen.NamedTable('errors', errors_table)
756 2751 aaronmk
    joins = [columns, sql_gen.Join(values, type_='CROSS'),
757
        sql_gen.Join(errors_table, dict(zip(name_only_cols, select_cols)),
758
        sql_gen.filter_out)]
759
760
    return mk_insert_select(db, errors_table, name_only_cols,
761
        *mk_select(db, joins, select_cols, order_by=None))[0]
762
763 2758 aaronmk
def track_data_error(db, errors_table, *args, **kw_args):
764
    '''
765
    @param errors_table If None, does nothing.
766
    '''
767
    if errors_table == None: return
768
    run_query(db, mk_track_data_error(db, errors_table, *args, **kw_args),
769
        cacheable=True, log_level=4)
770
771 2704 aaronmk
def cast(db, type_, col, errors_table=None):
772
    '''Casts an (unrenamed) column or value.
773
    If a column, converts any errors to warnings.
774
    @param col sql_gen.Col|sql_gen.Literal
775
    @param errors_table None|sql_gen.Table|str
776
        If set and col is a column with srcs, saves any errors in this table,
777
        using column's srcs attr as the source columns.
778
    '''
779
    if isinstance(col, sql_gen.Literal): # literal value, so just cast
780
        return sql_gen.CustomCode(col.to_str(db)+'::'+type_)
781
782
    assert isinstance(col, sql_gen.Col)
783
    assert not isinstance(col, sql_gen.NamedCol)
784
785
    save_errors = errors_table != None and col.srcs != ()
786
    if save_errors:
787
        errors_table = sql_gen.as_Table(errors_table)
788
        srcs = map(sql_gen.to_name_only_col, col.srcs)
789 2717 aaronmk
        function_name = str(sql_gen.FunctionCall(type_, *srcs))
790 2704 aaronmk
    else: function_name = type_
791
    function = sql_gen.TempFunction(function_name, db.autocommit)
792
793 2718 aaronmk
    while True:
794
        # Create function definition
795
        query = '''\
796 2704 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''(value text)
797
RETURNS '''+type_+'''
798
LANGUAGE plpgsql
799
'''
800 2718 aaronmk
        if not save_errors: query += 'IMMUTABLE\n'
801
        query += '''\
802 2704 aaronmk
STRICT
803
AS $$
804
BEGIN
805
    /* The explicit cast to the return type is needed to make the cast happen
806
    inside the try block. (Implicit casts to the return type happen at the end
807
    of the function, outside any block.) */
808
    RETURN value::'''+type_+''';
809
EXCEPTION
810
    WHEN data_exception THEN
811
'''
812 2718 aaronmk
        if save_errors:
813
            query += '''\
814 2704 aaronmk
        -- Save error in errors table.
815 2751 aaronmk
        -- Insert the value and error for *each* source column.
816
'''+mk_track_data_error(db, errors_table, srcs,
817
    *map(sql_gen.CustomCode, ['value', 'SQLSTATE', 'SQLERRM']))+''';
818 2704 aaronmk
819
'''
820 2718 aaronmk
        query += '''\
821 2704 aaronmk
        RAISE WARNING '%', SQLERRM;
822
        RETURN NULL;
823
END;
824
$$;
825
'''
826 2718 aaronmk
827
        # Create function
828
        try:
829
            run_query(db, query, recover=True, cacheable=True,
830
                log_ignore_excs=(DuplicateFunctionException,))
831
            break # successful
832
        except DuplicateFunctionException:
833
            if save_errors: function.name = next_version(function.name)
834
                # try again with next version of name
835
            else: break # plain cast function, so only need one version
836 2704 aaronmk
837
    return sql_gen.FunctionCall(function, col)
838
839 2414 aaronmk
##### Database structure queries
840
841 2426 aaronmk
def table_row_count(db, table, recover=None):
842
    return value(run_query(db, *mk_select(db, table, [sql_gen.row_count],
843 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
844 2426 aaronmk
845 2414 aaronmk
def table_cols(db, table, recover=None):
846
    return list(col_names(select(db, table, limit=0, order_by=None,
847 2443 aaronmk
        recover=recover, log_level=4)))
848 2414 aaronmk
849 2291 aaronmk
def pkey(db, table, recover=None):
850 832 aaronmk
    '''Assumed to be first column in table'''
851 2339 aaronmk
    return table_cols(db, table, recover)[0]
852 832 aaronmk
853 2559 aaronmk
not_null_col = 'not_null_col'
854 2340 aaronmk
855
def table_not_null_col(db, table, recover=None):
856
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
857
    if not_null_col in table_cols(db, table, recover): return not_null_col
858
    else: return pkey(db, table, recover)
859
860 853 aaronmk
def index_cols(db, table, index):
861
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
862
    automatically created. When you don't know whether something is a UNIQUE
863
    constraint or a UNIQUE index, use this function.'''
864 1909 aaronmk
    module = util.root_module(db.db)
865
    if module == 'psycopg2':
866
        return list(values(run_query(db, '''\
867 853 aaronmk
SELECT attname
868 866 aaronmk
FROM
869
(
870
        SELECT attnum, attname
871
        FROM pg_index
872
        JOIN pg_class index ON index.oid = indexrelid
873
        JOIN pg_class table_ ON table_.oid = indrelid
874
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
875
        WHERE
876
            table_.relname = %(table)s
877
            AND index.relname = %(index)s
878
    UNION
879
        SELECT attnum, attname
880
        FROM
881
        (
882
            SELECT
883
                indrelid
884
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
885
                    AS indkey
886
            FROM pg_index
887
            JOIN pg_class index ON index.oid = indexrelid
888
            JOIN pg_class table_ ON table_.oid = indrelid
889
            WHERE
890
                table_.relname = %(table)s
891
                AND index.relname = %(index)s
892
        ) s
893
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
894
) s
895 853 aaronmk
ORDER BY attnum
896
''',
897 2443 aaronmk
            {'table': table, 'index': index}, cacheable=True, log_level=4)))
898 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
899
        ' database')
900 853 aaronmk
901 464 aaronmk
def constraint_cols(db, table, constraint):
902 1849 aaronmk
    module = util.root_module(db.db)
903 464 aaronmk
    if module == 'psycopg2':
904
        return list(values(run_query(db, '''\
905
SELECT attname
906
FROM pg_constraint
907
JOIN pg_class ON pg_class.oid = conrelid
908
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
909
WHERE
910
    relname = %(table)s
911
    AND conname = %(constraint)s
912
ORDER BY attnum
913
''',
914
            {'table': table, 'constraint': constraint})))
915
    else: raise NotImplementedError("Can't list constraint columns for "+module+
916
        ' database')
917
918 2096 aaronmk
row_num_col = '_row_num'
919
920 2688 aaronmk
def add_index(db, exprs, table=None, unique=False):
921
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
922 2538 aaronmk
    Currently, only function calls are supported as expressions.
923
    '''
924 2688 aaronmk
    if not lists.is_seq(exprs): exprs = [exprs]
925 2538 aaronmk
926 2688 aaronmk
    # Parse exprs
927
    old_exprs = exprs[:]
928
    exprs = []
929
    cols = []
930
    for i, expr in enumerate(old_exprs):
931
        expr = copy.deepcopy(expr) # don't modify input!
932
        expr = sql_gen.as_Col(expr)
933
934
        # Extract col
935
        if isinstance(expr, sql_gen.FunctionCall):
936
            col = expr.args[0]
937
            expr = sql_gen.Expr(expr)
938
        else: col = expr
939
940
        # Extract table
941
        if table == None:
942
            assert sql_gen.is_table_col(col)
943
            table = col.table
944
945
        col.table = None
946
947
        exprs.append(expr)
948
        cols.append(col)
949 2408 aaronmk
950 2688 aaronmk
    table = sql_gen.as_Table(table)
951
    index = sql_gen.Table(str(sql_gen.Col(','.join(map(str, cols)), table)))
952
953
    str_ = 'CREATE'
954
    if unique: str_ += ' UNIQUE'
955
    str_ += ' INDEX '+index.to_str(db)+' ON '+table.to_str(db)+' ('+(
956
        ', '.join((v.to_str(db) for v in exprs)))+')'
957
958
    try: run_query(db, str_, recover=True, cacheable=True, log_level=3)
959 2408 aaronmk
    except DuplicateTableException: pass # index already existed
960
961 2686 aaronmk
def add_pkey(db, table, cols=None, recover=None):
962
    '''Adds a primary key.
963
    @param cols [sql_gen.Col,...] The columns in the primary key.
964
        Defaults to the first column in the table.
965 2406 aaronmk
    @pre The table must not already have a primary key.
966
    '''
967
    table = sql_gen.as_Table(table)
968 2686 aaronmk
    if cols == None: cols = [pkey(db, table, recover)]
969
    col_strs = [sql_gen.to_name_only_col(v).to_str(db) for v in cols]
970 2406 aaronmk
971 2590 aaronmk
    index = sql_gen.as_Table(sql_gen.add_suffix(table.name, '_pkey'))
972 2614 aaronmk
    try:
973
        run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD CONSTRAINT '
974 2686 aaronmk
            +index.to_str(db)+' PRIMARY KEY ('+(', '.join(col_strs))+')',
975 2614 aaronmk
            recover=True, cacheable=True, log_level=3,
976
            log_ignore_excs=(DuplicateTableException,))
977
    except DuplicateTableException, e:
978
        index.name = next_version(index.name)
979
        # try again with next version of name
980 2406 aaronmk
981 2086 aaronmk
def add_row_num(db, table):
982 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
983
    be the primary key.'''
984 2320 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
985 2096 aaronmk
    run_query(db, 'ALTER TABLE '+table+' ADD COLUMN '+row_num_col
986 2443 aaronmk
        +' serial NOT NULL PRIMARY KEY', log_level=3)
987 2086 aaronmk
988 2760 aaronmk
def create_table(db, table, cols, has_pkey=True, col_indexes=True):
989 2675 aaronmk
    '''Creates a table.
990 2681 aaronmk
    @param cols [sql_gen.TypedCol,...] The column names and types
991
    @param has_pkey If set, the first column becomes the primary key.
992 2760 aaronmk
    @param col_indexes bool|[ref]
993
        * If True, indexes will be added on all non-pkey columns.
994
        * If a list reference, [0] will be set to a function to do this.
995
          This can be used to delay index creation until the table is populated.
996 2675 aaronmk
    '''
997
    table = sql_gen.as_Table(table)
998
999 2681 aaronmk
    if has_pkey:
1000
        cols[0] = pkey = copy.copy(cols[0]) # don't modify input!
1001
        pkey.type += ' NOT NULL PRIMARY KEY'
1002
1003 2675 aaronmk
    str_ = 'CREATE TABLE '+table.to_str(db)+' (\n'
1004 2681 aaronmk
    str_ += '\n, '.join(v.to_str(db) for v in cols)
1005
    str_ += '\n);\n'
1006 2675 aaronmk
    run_query(db, str_, cacheable=True, log_level=2)
1007 2760 aaronmk
1008
    # Add indexes
1009
    if has_pkey: index_cols = cols[1:]
1010
    else: index_cols = cols
1011
    def add_indexes():
1012
        for col in index_cols: add_index(db, col.to_Col(), table)
1013
    if isinstance(col_indexes, list): col_indexes[0] = add_indexes # defer
1014
    elif col_indexes: add_indexes() # add now
1015 2675 aaronmk
1016 2684 aaronmk
def vacuum(db, table):
1017
    table = sql_gen.as_Table(table)
1018
    db.with_autocommit(lambda: run_query(db, 'VACUUM ANALYZE '+table.to_str(db),
1019
        log_level=3))
1020
1021 2732 aaronmk
def truncate(db, table, schema='public'):
1022 2733 aaronmk
    return run_query(db, 'TRUNCATE '+table.to_str(db)+' CASCADE')
1023 2732 aaronmk
1024 2729 aaronmk
def tables(db, schema_like='public', table_like='%', exact=False):
1025
    if exact: compare = '='
1026
    else: compare = 'LIKE'
1027
1028 1849 aaronmk
    module = util.root_module(db.db)
1029 2548 aaronmk
    params = {'schema_like': schema_like, 'table_like': table_like}
1030 832 aaronmk
    if module == 'psycopg2':
1031 1968 aaronmk
        return values(run_query(db, '''\
1032
SELECT tablename
1033
FROM pg_tables
1034
WHERE
1035 2729 aaronmk
    schemaname '''+compare+''' %(schema_like)s
1036
    AND tablename '''+compare+''' %(table_like)s
1037 1968 aaronmk
ORDER BY tablename
1038
''',
1039 2731 aaronmk
            params, cacheable=True, log_level=4))
1040 1968 aaronmk
    elif module == 'MySQLdb':
1041
        return values(run_query(db, 'SHOW TABLES LIKE %(table_like)s', params,
1042 2731 aaronmk
            cacheable=True, log_level=4))
1043 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
1044 830 aaronmk
1045 2726 aaronmk
def table_exists(db, table):
1046
    table = sql_gen.as_Table(table)
1047 2730 aaronmk
    return list(tables(db, table.schema, table.name, exact=True)) != []
1048 2726 aaronmk
1049
def errors_table(db, table, if_exists=True):
1050
    '''
1051
    @param if_exists If set, returns None if the errors table doesn't exist
1052
    @return None|sql_gen.Table
1053
    '''
1054
    table = sql_gen.as_Table(table)
1055
    if table.srcs != (): table = table.srcs[0]
1056
1057
    errors_table = sql_gen.suffixed_table(table, '.errors')
1058
    if if_exists and not table_exists(db, errors_table): return None
1059
    return errors_table
1060
1061 833 aaronmk
##### Database management
1062
1063 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
1064
    '''For kw_args, see tables()'''
1065
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
1066 833 aaronmk
1067 832 aaronmk
##### Heuristic queries
1068
1069 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
1070 1554 aaronmk
    '''Recovers from errors.
1071 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
1072
    '''
1073 2642 aaronmk
    row = sql_gen.ColDict(db, table, row)
1074 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
1075
1076 471 aaronmk
    try:
1077 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
1078 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
1079
            row_ct_ref[0] += cur.rowcount
1080
        return value(cur)
1081 471 aaronmk
    except DuplicateKeyException, e:
1082 2653 aaronmk
        row = sql_gen.ColDict(db, table,
1083
            util.dict_subset_right_join(row, e.cols))
1084
        return value(select(db, table, [pkey_], row, recover=True))
1085 471 aaronmk
1086 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
1087 830 aaronmk
    '''Recovers from errors'''
1088 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
1089 14 aaronmk
    except StopIteration:
1090 40 aaronmk
        if not create: raise
1091 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
1092 2078 aaronmk
1093 2593 aaronmk
def is_func_result(col):
1094
    return col.table.name.find('(') >= 0 and col.name == 'result'
1095
1096 2592 aaronmk
def into_table_name(out_table, in_tables0, mapping, is_func):
1097 2615 aaronmk
    def in_col_str(in_col):
1098
        in_col = sql_gen.remove_col_rename(in_col)
1099
        if isinstance(in_col, sql_gen.Col):
1100
            table = in_col.table
1101
            if table == in_tables0:
1102
                in_col = sql_gen.to_name_only_col(in_col)
1103
            elif is_func_result(in_col): in_col = table # omit col name
1104
        return str(in_col)
1105
1106 2580 aaronmk
    str_ = str(out_table)
1107
    if is_func:
1108 2615 aaronmk
        str_ += '('
1109 2580 aaronmk
1110 2615 aaronmk
        try: value_in_col = mapping['value']
1111
        except KeyError:
1112 2654 aaronmk
            str_ += ', '.join((str(k)+'='+in_col_str(v)
1113 2615 aaronmk
                for k, v in mapping.iteritems()))
1114
        else: str_ += in_col_str(value_in_col)
1115
1116
        str_ += ')'
1117 2580 aaronmk
    else: str_ += '_pkeys'
1118
    return str_
1119
1120 2508 aaronmk
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None, into=None,
1121 2552 aaronmk
    default=None, is_func=False):
1122 2078 aaronmk
    '''Recovers from errors.
1123
    Only works under PostgreSQL (uses INSERT RETURNING).
1124 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
1125
        tables to join with it using the main input table's pkey
1126 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
1127 2612 aaronmk
        * out_table_col: str (*not* sql_gen.Col)
1128
        * in_table_col: sql_gen.Col|literal-value
1129 2489 aaronmk
    @param into The table to contain the output and input pkeys.
1130 2574 aaronmk
        Defaults to `out_table.name+'_pkeys'`.
1131 2509 aaronmk
    @param default The *output* column to use as the pkey for missing rows.
1132
        If this output column does not exist in the mapping, uses None.
1133 2552 aaronmk
    @param is_func Whether out_table is the name of a SQL function, not a table
1134 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
1135 2078 aaronmk
    '''
1136 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
1137 2312 aaronmk
1138 2450 aaronmk
    def log_debug(msg): db.log_debug(msg, level=1.5)
1139 2505 aaronmk
    def col_ustr(str_):
1140 2567 aaronmk
        return strings.repr_no_u(sql_gen.remove_col_rename(str_))
1141 2450 aaronmk
1142 2486 aaronmk
    log_debug('********** New iteration **********')
1143 2505 aaronmk
    log_debug('Inserting these input columns into '+strings.as_tt(
1144
        out_table.to_str(db))+':\n'+strings.as_table(mapping, ustr=col_ustr))
1145 2463 aaronmk
1146 2382 aaronmk
    # Create input joins from list of input tables
1147
    in_tables_ = in_tables[:] # don't modify input!
1148
    in_tables0 = in_tables_.pop(0) # first table is separate
1149 2735 aaronmk
    errors_table_ = errors_table(db, in_tables0)
1150 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
1151 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
1152 2460 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v,
1153
        {in_pkey: sql_gen.join_same_not_null}) for v in in_tables_]
1154 2131 aaronmk
1155 2592 aaronmk
    if into == None:
1156
        into = into_table_name(out_table, in_tables0, mapping, is_func)
1157
    into = sql_gen.as_Table(into)
1158
1159 2702 aaronmk
    # Set column sources
1160
    in_cols = filter(sql_gen.is_table_col, mapping.values())
1161
    for col in in_cols:
1162
        if col.table == in_tables0: col.set_srcs(sql_gen.src_self)
1163
1164 2486 aaronmk
    log_debug('Joining together input tables into temp table')
1165 2395 aaronmk
    # Place in new table for speed and so don't modify input if values edited
1166 2584 aaronmk
    in_table = sql_gen.Table('in')
1167 2702 aaronmk
    mapping = dicts.join(mapping, flatten(db, in_table, input_joins, in_cols,
1168
        preserve=[in_pkey_col], start=0))
1169 2395 aaronmk
    input_joins = [in_table]
1170 2486 aaronmk
    db.log_debug('Temp table: '+strings.as_tt(in_table.to_str(db)), level=2)
1171 2395 aaronmk
1172 2656 aaronmk
    mapping = sql_gen.ColDict(db, out_table, mapping)
1173
        # after applying dicts.join() because that returns a plain dict
1174
1175 2509 aaronmk
    # Resolve default value column
1176
    try: default = mapping[default]
1177
    except KeyError:
1178
        if default != None:
1179
            db.log_debug('Default value column '
1180
                +strings.as_tt(strings.repr_no_u(default))
1181 2511 aaronmk
                +' does not exist in mapping, falling back to None', level=2.1)
1182 2509 aaronmk
            default = None
1183
1184 2279 aaronmk
    out_pkey = pkey(db, out_table, recover=True)
1185 2285 aaronmk
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
1186 2142 aaronmk
1187 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
1188 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
1189
1190 2201 aaronmk
    pkeys_table_exists_ref = [False]
1191 2420 aaronmk
    def insert_into_pkeys(joins, cols):
1192
        query, params = mk_select(db, joins, cols, order_by=None, start=0)
1193 2201 aaronmk
        if pkeys_table_exists_ref[0]:
1194 2489 aaronmk
            insert_select(db, into, pkeys_names, query, params)
1195 2201 aaronmk
        else:
1196 2489 aaronmk
            run_query_into(db, query, params, into=into)
1197 2201 aaronmk
            pkeys_table_exists_ref[0] = True
1198
1199 2429 aaronmk
    limit_ref = [None]
1200 2380 aaronmk
    conds = set()
1201 2233 aaronmk
    distinct_on = []
1202 2325 aaronmk
    def mk_main_select(joins, cols):
1203 2429 aaronmk
        return mk_select(db, joins, cols, conds, distinct_on,
1204
            limit=limit_ref[0], start=0)
1205 2132 aaronmk
1206 2519 aaronmk
    exc_strs = set()
1207 2309 aaronmk
    def log_exc(e):
1208 2519 aaronmk
        e_str = exc.str_(e, first_line_only=True)
1209
        log_debug('Caught exception: '+e_str)
1210
        assert e_str not in exc_strs # avoid infinite loops
1211
        exc_strs.add(e_str)
1212 2735 aaronmk
1213 2451 aaronmk
    def remove_all_rows():
1214 2450 aaronmk
        log_debug('Returning NULL for all rows')
1215 2429 aaronmk
        limit_ref[0] = 0 # just create an empty pkeys table
1216 2735 aaronmk
1217 2758 aaronmk
    def ignore(in_col, value, e):
1218
        track_data_error(db, errors_table_, in_col.srcs, value, e.cause.pgcode,
1219
            e.cause.pgerror)
1220
1221 2545 aaronmk
        in_col_str = strings.as_tt(repr(in_col))
1222 2544 aaronmk
        db.log_debug('Adding index on '+in_col_str+' to enable fast filtering',
1223
            level=2.5)
1224 2537 aaronmk
        add_index(db, in_col)
1225 2735 aaronmk
1226 2545 aaronmk
        log_debug('Ignoring rows with '+in_col_str+' = '
1227
            +strings.as_tt(repr(value)))
1228 2758 aaronmk
    def remove_rows(in_col, value, e):
1229
        ignore(in_col, value, e)
1230 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
1231
        assert cond not in conds # avoid infinite loops
1232 2380 aaronmk
        conds.add(cond)
1233 2758 aaronmk
    def invalid2null(in_col, value, e):
1234
        ignore(in_col, value, e)
1235 2403 aaronmk
        update(db, in_table, [(in_col, None)],
1236
            sql_gen.ColValueCond(in_col, value))
1237 2245 aaronmk
1238 2589 aaronmk
    def insert_pkeys_table(which):
1239
        return sql_gen.Table(sql_gen.add_suffix(in_table.name,
1240
            '_insert_'+which+'_pkeys'))
1241
    insert_out_pkeys = insert_pkeys_table('out')
1242
    insert_in_pkeys = insert_pkeys_table('in')
1243
1244 2206 aaronmk
    # Do inserts and selects
1245 2642 aaronmk
    join_cols = sql_gen.ColDict(db, out_table)
1246 2206 aaronmk
    while True:
1247 2521 aaronmk
        if limit_ref[0] == 0: # special case
1248
            log_debug('Creating an empty pkeys table')
1249
            cur = run_query_into(db, *mk_select(db, out_table, [out_pkey],
1250
                limit=limit_ref[0]), into=insert_out_pkeys)
1251
            break # don't do main case
1252
1253 2303 aaronmk
        has_joins = join_cols != {}
1254
1255 2305 aaronmk
        # Prepare to insert new rows
1256 2325 aaronmk
        insert_joins = input_joins[:] # don't modify original!
1257 2403 aaronmk
        insert_args = dict(recover=True, cacheable=False)
1258 2303 aaronmk
        if has_joins:
1259 2317 aaronmk
            distinct_on = [v.to_Col() for v in join_cols.values()]
1260 2325 aaronmk
            insert_joins.append(sql_gen.Join(out_table, join_cols,
1261
                sql_gen.filter_out))
1262
        else:
1263 2404 aaronmk
            insert_args.update(dict(returning=out_pkey, into=insert_out_pkeys))
1264 2520 aaronmk
        main_select = mk_main_select(insert_joins, mapping.values())[0]
1265 2303 aaronmk
1266 2486 aaronmk
        log_debug('Trying to insert new rows')
1267 2206 aaronmk
        try:
1268 2518 aaronmk
            cur = insert_select(db, out_table, mapping.keys(), main_select,
1269
                **insert_args)
1270 2357 aaronmk
            break # insert successful
1271 2206 aaronmk
        except DuplicateKeyException, e:
1272 2309 aaronmk
            log_exc(e)
1273
1274 2258 aaronmk
            old_join_cols = join_cols.copy()
1275 2565 aaronmk
            join_cols.update(util.dict_subset_right_join(mapping, e.cols))
1276 2486 aaronmk
            log_debug('Ignoring existing rows, comparing on these columns:\n'
1277 2505 aaronmk
                +strings.as_inline_table(join_cols, ustr=col_ustr))
1278 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
1279 2230 aaronmk
        except NullValueException, e:
1280 2309 aaronmk
            log_exc(e)
1281
1282 2230 aaronmk
            out_col, = e.cols
1283
            try: in_col = mapping[out_col]
1284 2356 aaronmk
            except KeyError:
1285 2486 aaronmk
                log_debug('Missing mapping for NOT NULL column '+out_col)
1286 2451 aaronmk
                remove_all_rows()
1287 2758 aaronmk
            else: remove_rows(in_col, None, e)
1288 2542 aaronmk
        except FunctionValueException, e:
1289
            log_exc(e)
1290
1291
            func_name = e.name
1292
            value = e.value
1293
            for out_col, in_col in mapping.iteritems():
1294 2758 aaronmk
                in_col = sql_gen.unwrap_func_call(in_col, func_name)
1295
                invalid2null(in_col, value, e)
1296 2525 aaronmk
        except MissingCastException, e:
1297
            log_exc(e)
1298
1299
            out_col = e.col
1300 2706 aaronmk
            type_ = e.type
1301
1302 2724 aaronmk
            log_debug('Casting '+strings.as_tt(out_col)+' input to '
1303
                +strings.as_tt(type_))
1304 2735 aaronmk
            def wrap_func(col): return cast(db, type_, col, errors_table_)
1305 2705 aaronmk
            mapping[out_col] = sql_gen.wrap(wrap_func, mapping[out_col])
1306 2429 aaronmk
        except DatabaseErrors, e:
1307
            log_exc(e)
1308
1309 2531 aaronmk
            msg = 'No handler for exception: '+exc.str_(e)
1310 2451 aaronmk
            warnings.warn(DbWarning(msg))
1311
            log_debug(msg)
1312
            remove_all_rows()
1313 2358 aaronmk
        # after exception handled, rerun loop with additional constraints
1314 2132 aaronmk
1315 2357 aaronmk
    if row_ct_ref != None and cur.rowcount >= 0:
1316
        row_ct_ref[0] += cur.rowcount
1317
1318
    if has_joins:
1319
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
1320 2486 aaronmk
        log_debug('Getting output table pkeys of existing/inserted rows')
1321 2420 aaronmk
        insert_into_pkeys(select_joins, pkeys_cols)
1322 2357 aaronmk
    else:
1323 2404 aaronmk
        add_row_num(db, insert_out_pkeys) # for joining with input pkeys
1324 2357 aaronmk
1325 2486 aaronmk
        log_debug('Getting input table pkeys of inserted rows')
1326 2357 aaronmk
        run_query_into(db, *mk_main_select(input_joins, [in_pkey]),
1327 2404 aaronmk
            into=insert_in_pkeys)
1328
        add_row_num(db, insert_in_pkeys) # for joining with output pkeys
1329 2357 aaronmk
1330 2428 aaronmk
        assert table_row_count(db, insert_out_pkeys) == table_row_count(db,
1331
            insert_in_pkeys)
1332
1333 2486 aaronmk
        log_debug('Combining output and input pkeys in inserted order')
1334 2404 aaronmk
        pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
1335 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
1336 2420 aaronmk
        insert_into_pkeys(pkey_joins, pkeys_names)
1337 2357 aaronmk
1338 2486 aaronmk
    db.log_debug('Adding pkey on pkeys table to enable fast joins', level=2.5)
1339 2594 aaronmk
    add_pkey(db, into)
1340 2407 aaronmk
1341 2508 aaronmk
    log_debug('Setting pkeys of missing rows to '+strings.as_tt(repr(default)))
1342 2489 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(into,
1343 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
1344
        # must use join_same_not_null or query will take forever
1345 2420 aaronmk
    insert_into_pkeys(missing_rows_joins,
1346 2508 aaronmk
        [in_pkey_col, sql_gen.NamedCol(out_pkey, default)])
1347 2357 aaronmk
1348 2489 aaronmk
    assert table_row_count(db, into) == table_row_count(db, in_table)
1349 2428 aaronmk
1350 2702 aaronmk
    srcs = []
1351
    if is_func: srcs = sql_gen.cols_srcs(in_cols)
1352
    return sql_gen.Col(out_pkey, into, srcs)
1353 2115 aaronmk
1354
##### Data cleanup
1355
1356 2290 aaronmk
def cleanup_table(db, table, cols):
1357 2677 aaronmk
    table = sql_gen.as_Table(table)
1358
    cols = map(sql_gen.as_Col, cols)
1359 2115 aaronmk
1360 2677 aaronmk
    expr = ('nullif(nullif(trim(both from %s), '+db.esc_value('')+'), '
1361
        +db.esc_value(r'\N')+')')
1362
    changes = [(v, sql_gen.CustomCode(expr % v.to_str(db)))
1363
        for v in cols]
1364 2115 aaronmk
1365 2677 aaronmk
    update(db, table, changes)