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 2764 aaronmk
            self._is_insert = query.startswith('INSERT')
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 2765 aaronmk
    else:
542
        assert fields != []
543
        query += '\n, '.join(map(parse_col, fields))
544 2200 aaronmk
545
    # Main table
546 2467 aaronmk
    query += '\nFROM '+table0.to_str(db)
547 865 aaronmk
548 2122 aaronmk
    # Add joins
549 2271 aaronmk
    left_table = table0
550 2263 aaronmk
    for join_ in tables:
551
        table = join_.table
552 2238 aaronmk
553 2343 aaronmk
        # Parse special values
554
        if join_.type_ is sql_gen.filter_out: # filter no match
555 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
556
                None))
557 2343 aaronmk
558 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
559 2122 aaronmk
560
        left_table = table
561
562 865 aaronmk
    missing = True
563 2376 aaronmk
    if conds != []:
564 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
565
        else: whitespace = '\n'
566 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
567
            .to_str(db) for l, r in conds], 'WHERE')
568 865 aaronmk
        missing = False
569 2227 aaronmk
    if order_by != None:
570 2467 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
571
    if limit != None: query += '\nLIMIT '+str(limit); missing = False
572 865 aaronmk
    if start != None:
573 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
574 865 aaronmk
        missing = False
575
    if missing: warnings.warn(DbWarning(
576
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
577
578 2315 aaronmk
    return (query, [])
579 11 aaronmk
580 2054 aaronmk
def select(db, *args, **kw_args):
581
    '''For params, see mk_select() and run_query()'''
582
    recover = kw_args.pop('recover', None)
583
    cacheable = kw_args.pop('cacheable', True)
584 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
585 2054 aaronmk
586
    query, params = mk_select(db, *args, **kw_args)
587 2442 aaronmk
    return run_query(db, query, params, recover, cacheable, log_level=log_level)
588 2054 aaronmk
589 2066 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, params=None,
590 2292 aaronmk
    returning=None, embeddable=False):
591 1960 aaronmk
    '''
592
    @param returning str|None An inserted column (such as pkey) to return
593 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
594 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
595
        query will be fully cached, not just if it raises an exception.
596 1960 aaronmk
    '''
597 2754 aaronmk
    table = sql_gen.remove_table_rename(sql_gen.as_Table(table))
598 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
599 2646 aaronmk
    if cols != None:
600
        cols = [sql_gen.to_name_only_col(v, table).to_str(db) for v in cols]
601 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
602 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
603 2063 aaronmk
604
    # Build query
605 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
606
    query = first_line
607 2467 aaronmk
    if cols != None: query += '\n('+', '.join(cols)+')'
608
    query += '\n'+select_query
609 2063 aaronmk
610
    if returning != None:
611 2743 aaronmk
        query += '\nRETURNING '+sql_gen.to_name_only_col(returning).to_str(db)
612 2063 aaronmk
613 2070 aaronmk
    if embeddable:
614 2327 aaronmk
        assert returning != None
615
616 2070 aaronmk
        # Create function
617 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
618 2327 aaronmk
        return_type = 'SETOF '+returning.to_str(db)+'%TYPE'
619 2189 aaronmk
        while True:
620
            try:
621 2698 aaronmk
                function = sql_gen.TempFunction(function_name, db.autocommit)
622 2194 aaronmk
623 2189 aaronmk
                function_query = '''\
624 2698 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''()
625 2467 aaronmk
RETURNS '''+return_type+'''
626
LANGUAGE sql
627
AS $$
628
'''+mogrify(db, query, params)+''';
629
$$;
630 2070 aaronmk
'''
631 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
632 2464 aaronmk
                    log_ignore_excs=(DuplicateFunctionException,))
633 2189 aaronmk
                break # this version was successful
634
            except DuplicateFunctionException, e:
635
                function_name = next_version(function_name)
636
                # try again with next version of name
637 2070 aaronmk
638 2337 aaronmk
        # Return query that uses function
639 2698 aaronmk
        func_table = sql_gen.NamedTable('f', sql_gen.FunctionCall(function),
640 2742 aaronmk
            [returning]) # AS clause requires function alias
641 2337 aaronmk
        return mk_select(db, func_table, start=0, order_by=None)
642 2070 aaronmk
643 2066 aaronmk
    return (query, params)
644
645
def insert_select(db, *args, **kw_args):
646 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
647 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
648
        values in
649 2072 aaronmk
    '''
650 2386 aaronmk
    into = kw_args.pop('into', None)
651
    if into != None: kw_args['embeddable'] = True
652 2066 aaronmk
    recover = kw_args.pop('recover', None)
653
    cacheable = kw_args.pop('cacheable', True)
654 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
655 2066 aaronmk
656
    query, params = mk_insert_select(db, *args, **kw_args)
657 2386 aaronmk
    return run_query_into(db, query, params, into, recover=recover,
658 2673 aaronmk
        cacheable=cacheable, log_level=log_level)
659 2063 aaronmk
660 2738 aaronmk
default = sql_gen.default # tells insert() to use the default value for a column
661 2066 aaronmk
662 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
663 2085 aaronmk
    '''For params, see insert_select()'''
664 1960 aaronmk
    if lists.is_seq(row): cols = None
665
    else:
666
        cols = row.keys()
667
        row = row.values()
668 2738 aaronmk
    row = list(row) # ensure that "== []" works
669 1960 aaronmk
670 2738 aaronmk
    if row == []: query = None
671
    else: query = sql_gen.Values(row).to_str(db)
672 1961 aaronmk
673 2738 aaronmk
    return insert_select(db, table, cols, query, [], *args, **kw_args)
674 11 aaronmk
675 2402 aaronmk
def mk_update(db, table, changes=None, cond=None):
676
    '''
677
    @param changes [(col, new_value),...]
678
        * container can be any iterable type
679
        * col: sql_gen.Code|str (for col name)
680
        * new_value: sql_gen.Code|literal value
681
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
682
    @return str query
683
    '''
684
    query = 'UPDATE '+sql_gen.as_Table(table).to_str(db)+'\nSET\n'
685 2405 aaronmk
    query += ',\n'.join((sql_gen.to_name_only_col(col, table).to_str(db)+' = '
686 2402 aaronmk
        +sql_gen.as_Value(new_value).to_str(db) for col, new_value in changes))
687 2467 aaronmk
    if cond != None: query += '\nWHERE\n'+cond.to_str(db)
688 2402 aaronmk
689
    return query
690
691
def update(db, *args, **kw_args):
692
    '''For params, see mk_update() and run_query()'''
693
    recover = kw_args.pop('recover', None)
694
695
    return run_query(db, mk_update(db, *args, **kw_args), [], recover)
696
697 135 aaronmk
def last_insert_id(db):
698 1849 aaronmk
    module = util.root_module(db.db)
699 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
700
    elif module == 'MySQLdb': return db.insert_id()
701
    else: return None
702 13 aaronmk
703 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
704 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
705 2415 aaronmk
    to names that will be distinct among the columns' tables.
706 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
707 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
708
    @param into The table for the new columns.
709 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
710
        columns will be included in the mapping even if they are not in cols.
711
        The tables of the provided Col objects will be changed to into, so make
712
        copies of them if you want to keep the original tables.
713
    @param as_items Whether to return a list of dict items instead of a dict
714 2383 aaronmk
    @return dict(orig_col=new_col, ...)
715
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
716 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
717
        * All mappings use the into table so its name can easily be
718 2383 aaronmk
          changed for all columns at once
719
    '''
720 2415 aaronmk
    cols = lists.uniqify(cols)
721
722 2394 aaronmk
    items = []
723 2389 aaronmk
    for col in preserve:
724 2390 aaronmk
        orig_col = copy.copy(col)
725 2392 aaronmk
        col.table = into
726 2394 aaronmk
        items.append((orig_col, col))
727
    preserve = set(preserve)
728
    for col in cols:
729 2716 aaronmk
        if col not in preserve:
730
            items.append((col, sql_gen.Col(str(col), into, col.srcs)))
731 2394 aaronmk
732
    if not as_items: items = dict(items)
733
    return items
734 2383 aaronmk
735 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
736 2391 aaronmk
    '''For params, see mk_flatten_mapping()
737
    @return See return value of mk_flatten_mapping()
738
    '''
739 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
740
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
741 2391 aaronmk
    run_query_into(db, *mk_select(db, joins, cols, limit=limit, start=start),
742 2392 aaronmk
        into=into)
743 2394 aaronmk
    return dict(items)
744 2391 aaronmk
745 2751 aaronmk
def mk_track_data_error(db, errors_table, cols, value, error_code, error):
746
    cols = map(sql_gen.to_name_only_col, cols)
747
748
    columns_cols = ['column']
749
    columns = sql_gen.NamedValues('columns', columns_cols,
750
        [[c.name] for c in cols])
751
    values_cols = ['value', 'error_code', 'error']
752
    values = sql_gen.NamedValues('values', values_cols,
753
        [value, error_code, error])
754
755
    select_cols = columns_cols+values_cols
756
    name_only_cols = map(sql_gen.to_name_only_col, select_cols)
757 2755 aaronmk
    errors_table = sql_gen.NamedTable('errors', errors_table)
758 2751 aaronmk
    joins = [columns, sql_gen.Join(values, type_='CROSS'),
759
        sql_gen.Join(errors_table, dict(zip(name_only_cols, select_cols)),
760
        sql_gen.filter_out)]
761
762
    return mk_insert_select(db, errors_table, name_only_cols,
763
        *mk_select(db, joins, select_cols, order_by=None))[0]
764
765 2758 aaronmk
def track_data_error(db, errors_table, *args, **kw_args):
766
    '''
767
    @param errors_table If None, does nothing.
768
    '''
769
    if errors_table == None: return
770
    run_query(db, mk_track_data_error(db, errors_table, *args, **kw_args),
771
        cacheable=True, log_level=4)
772
773 2704 aaronmk
def cast(db, type_, col, errors_table=None):
774
    '''Casts an (unrenamed) column or value.
775
    If a column, converts any errors to warnings.
776
    @param col sql_gen.Col|sql_gen.Literal
777
    @param errors_table None|sql_gen.Table|str
778
        If set and col is a column with srcs, saves any errors in this table,
779
        using column's srcs attr as the source columns.
780
    '''
781
    if isinstance(col, sql_gen.Literal): # literal value, so just cast
782
        return sql_gen.CustomCode(col.to_str(db)+'::'+type_)
783
784
    assert isinstance(col, sql_gen.Col)
785
    assert not isinstance(col, sql_gen.NamedCol)
786
787
    save_errors = errors_table != None and col.srcs != ()
788
    if save_errors:
789
        errors_table = sql_gen.as_Table(errors_table)
790
        srcs = map(sql_gen.to_name_only_col, col.srcs)
791 2717 aaronmk
        function_name = str(sql_gen.FunctionCall(type_, *srcs))
792 2704 aaronmk
    else: function_name = type_
793
    function = sql_gen.TempFunction(function_name, db.autocommit)
794
795 2718 aaronmk
    while True:
796
        # Create function definition
797
        query = '''\
798 2704 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''(value text)
799
RETURNS '''+type_+'''
800
LANGUAGE plpgsql
801
'''
802 2718 aaronmk
        if not save_errors: query += 'IMMUTABLE\n'
803
        query += '''\
804 2704 aaronmk
STRICT
805
AS $$
806
BEGIN
807
    /* The explicit cast to the return type is needed to make the cast happen
808
    inside the try block. (Implicit casts to the return type happen at the end
809
    of the function, outside any block.) */
810
    RETURN value::'''+type_+''';
811
EXCEPTION
812
    WHEN data_exception THEN
813
'''
814 2718 aaronmk
        if save_errors:
815
            query += '''\
816 2704 aaronmk
        -- Save error in errors table.
817 2751 aaronmk
        -- Insert the value and error for *each* source column.
818
'''+mk_track_data_error(db, errors_table, srcs,
819
    *map(sql_gen.CustomCode, ['value', 'SQLSTATE', 'SQLERRM']))+''';
820 2704 aaronmk
821
'''
822 2718 aaronmk
        query += '''\
823 2704 aaronmk
        RAISE WARNING '%', SQLERRM;
824
        RETURN NULL;
825
END;
826
$$;
827
'''
828 2718 aaronmk
829
        # Create function
830
        try:
831
            run_query(db, query, recover=True, cacheable=True,
832
                log_ignore_excs=(DuplicateFunctionException,))
833
            break # successful
834
        except DuplicateFunctionException:
835
            if save_errors: function.name = next_version(function.name)
836
                # try again with next version of name
837
            else: break # plain cast function, so only need one version
838 2704 aaronmk
839
    return sql_gen.FunctionCall(function, col)
840
841 2414 aaronmk
##### Database structure queries
842
843 2426 aaronmk
def table_row_count(db, table, recover=None):
844
    return value(run_query(db, *mk_select(db, table, [sql_gen.row_count],
845 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
846 2426 aaronmk
847 2414 aaronmk
def table_cols(db, table, recover=None):
848
    return list(col_names(select(db, table, limit=0, order_by=None,
849 2443 aaronmk
        recover=recover, log_level=4)))
850 2414 aaronmk
851 2291 aaronmk
def pkey(db, table, recover=None):
852 832 aaronmk
    '''Assumed to be first column in table'''
853 2339 aaronmk
    return table_cols(db, table, recover)[0]
854 832 aaronmk
855 2559 aaronmk
not_null_col = 'not_null_col'
856 2340 aaronmk
857
def table_not_null_col(db, table, recover=None):
858
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
859
    if not_null_col in table_cols(db, table, recover): return not_null_col
860
    else: return pkey(db, table, recover)
861
862 853 aaronmk
def index_cols(db, table, index):
863
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
864
    automatically created. When you don't know whether something is a UNIQUE
865
    constraint or a UNIQUE index, use this function.'''
866 1909 aaronmk
    module = util.root_module(db.db)
867
    if module == 'psycopg2':
868
        return list(values(run_query(db, '''\
869 853 aaronmk
SELECT attname
870 866 aaronmk
FROM
871
(
872
        SELECT attnum, attname
873
        FROM pg_index
874
        JOIN pg_class index ON index.oid = indexrelid
875
        JOIN pg_class table_ ON table_.oid = indrelid
876
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
877
        WHERE
878
            table_.relname = %(table)s
879
            AND index.relname = %(index)s
880
    UNION
881
        SELECT attnum, attname
882
        FROM
883
        (
884
            SELECT
885
                indrelid
886
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
887
                    AS indkey
888
            FROM pg_index
889
            JOIN pg_class index ON index.oid = indexrelid
890
            JOIN pg_class table_ ON table_.oid = indrelid
891
            WHERE
892
                table_.relname = %(table)s
893
                AND index.relname = %(index)s
894
        ) s
895
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
896
) s
897 853 aaronmk
ORDER BY attnum
898
''',
899 2443 aaronmk
            {'table': table, 'index': index}, cacheable=True, log_level=4)))
900 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
901
        ' database')
902 853 aaronmk
903 464 aaronmk
def constraint_cols(db, table, constraint):
904 1849 aaronmk
    module = util.root_module(db.db)
905 464 aaronmk
    if module == 'psycopg2':
906
        return list(values(run_query(db, '''\
907
SELECT attname
908
FROM pg_constraint
909
JOIN pg_class ON pg_class.oid = conrelid
910
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
911
WHERE
912
    relname = %(table)s
913
    AND conname = %(constraint)s
914
ORDER BY attnum
915
''',
916
            {'table': table, 'constraint': constraint})))
917
    else: raise NotImplementedError("Can't list constraint columns for "+module+
918
        ' database')
919
920 2096 aaronmk
row_num_col = '_row_num'
921
922 2688 aaronmk
def add_index(db, exprs, table=None, unique=False):
923
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
924 2538 aaronmk
    Currently, only function calls are supported as expressions.
925
    '''
926 2688 aaronmk
    if not lists.is_seq(exprs): exprs = [exprs]
927 2538 aaronmk
928 2688 aaronmk
    # Parse exprs
929
    old_exprs = exprs[:]
930
    exprs = []
931
    cols = []
932
    for i, expr in enumerate(old_exprs):
933
        expr = copy.deepcopy(expr) # don't modify input!
934
        expr = sql_gen.as_Col(expr)
935
936
        # Extract col
937
        if isinstance(expr, sql_gen.FunctionCall):
938
            col = expr.args[0]
939
            expr = sql_gen.Expr(expr)
940
        else: col = expr
941
942
        # Extract table
943
        if table == None:
944
            assert sql_gen.is_table_col(col)
945
            table = col.table
946
947
        col.table = None
948
949
        exprs.append(expr)
950
        cols.append(col)
951 2408 aaronmk
952 2688 aaronmk
    table = sql_gen.as_Table(table)
953
    index = sql_gen.Table(str(sql_gen.Col(','.join(map(str, cols)), table)))
954
955
    str_ = 'CREATE'
956
    if unique: str_ += ' UNIQUE'
957
    str_ += ' INDEX '+index.to_str(db)+' ON '+table.to_str(db)+' ('+(
958
        ', '.join((v.to_str(db) for v in exprs)))+')'
959
960
    try: run_query(db, str_, recover=True, cacheable=True, log_level=3)
961 2408 aaronmk
    except DuplicateTableException: pass # index already existed
962
963 2686 aaronmk
def add_pkey(db, table, cols=None, recover=None):
964
    '''Adds a primary key.
965
    @param cols [sql_gen.Col,...] The columns in the primary key.
966
        Defaults to the first column in the table.
967 2406 aaronmk
    @pre The table must not already have a primary key.
968
    '''
969
    table = sql_gen.as_Table(table)
970 2686 aaronmk
    if cols == None: cols = [pkey(db, table, recover)]
971
    col_strs = [sql_gen.to_name_only_col(v).to_str(db) for v in cols]
972 2406 aaronmk
973 2590 aaronmk
    index = sql_gen.as_Table(sql_gen.add_suffix(table.name, '_pkey'))
974 2614 aaronmk
    try:
975
        run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD CONSTRAINT '
976 2686 aaronmk
            +index.to_str(db)+' PRIMARY KEY ('+(', '.join(col_strs))+')',
977 2614 aaronmk
            recover=True, cacheable=True, log_level=3,
978
            log_ignore_excs=(DuplicateTableException,))
979
    except DuplicateTableException, e:
980
        index.name = next_version(index.name)
981
        # try again with next version of name
982 2406 aaronmk
983 2086 aaronmk
def add_row_num(db, table):
984 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
985
    be the primary key.'''
986 2320 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
987 2096 aaronmk
    run_query(db, 'ALTER TABLE '+table+' ADD COLUMN '+row_num_col
988 2443 aaronmk
        +' serial NOT NULL PRIMARY KEY', log_level=3)
989 2086 aaronmk
990 2760 aaronmk
def create_table(db, table, cols, has_pkey=True, col_indexes=True):
991 2675 aaronmk
    '''Creates a table.
992 2681 aaronmk
    @param cols [sql_gen.TypedCol,...] The column names and types
993
    @param has_pkey If set, the first column becomes the primary key.
994 2760 aaronmk
    @param col_indexes bool|[ref]
995
        * If True, indexes will be added on all non-pkey columns.
996
        * If a list reference, [0] will be set to a function to do this.
997
          This can be used to delay index creation until the table is populated.
998 2675 aaronmk
    '''
999
    table = sql_gen.as_Table(table)
1000
1001 2681 aaronmk
    if has_pkey:
1002
        cols[0] = pkey = copy.copy(cols[0]) # don't modify input!
1003
        pkey.type += ' NOT NULL PRIMARY KEY'
1004
1005 2675 aaronmk
    str_ = 'CREATE TABLE '+table.to_str(db)+' (\n'
1006 2681 aaronmk
    str_ += '\n, '.join(v.to_str(db) for v in cols)
1007
    str_ += '\n);\n'
1008 2675 aaronmk
    run_query(db, str_, cacheable=True, log_level=2)
1009 2760 aaronmk
1010
    # Add indexes
1011
    if has_pkey: index_cols = cols[1:]
1012
    else: index_cols = cols
1013
    def add_indexes():
1014
        for col in index_cols: add_index(db, col.to_Col(), table)
1015
    if isinstance(col_indexes, list): col_indexes[0] = add_indexes # defer
1016
    elif col_indexes: add_indexes() # add now
1017 2675 aaronmk
1018 2684 aaronmk
def vacuum(db, table):
1019
    table = sql_gen.as_Table(table)
1020
    db.with_autocommit(lambda: run_query(db, 'VACUUM ANALYZE '+table.to_str(db),
1021
        log_level=3))
1022
1023 2732 aaronmk
def truncate(db, table, schema='public'):
1024 2733 aaronmk
    return run_query(db, 'TRUNCATE '+table.to_str(db)+' CASCADE')
1025 2732 aaronmk
1026 2729 aaronmk
def tables(db, schema_like='public', table_like='%', exact=False):
1027
    if exact: compare = '='
1028
    else: compare = 'LIKE'
1029
1030 1849 aaronmk
    module = util.root_module(db.db)
1031 2548 aaronmk
    params = {'schema_like': schema_like, 'table_like': table_like}
1032 832 aaronmk
    if module == 'psycopg2':
1033 1968 aaronmk
        return values(run_query(db, '''\
1034
SELECT tablename
1035
FROM pg_tables
1036
WHERE
1037 2729 aaronmk
    schemaname '''+compare+''' %(schema_like)s
1038
    AND tablename '''+compare+''' %(table_like)s
1039 1968 aaronmk
ORDER BY tablename
1040
''',
1041 2731 aaronmk
            params, cacheable=True, log_level=4))
1042 1968 aaronmk
    elif module == 'MySQLdb':
1043
        return values(run_query(db, 'SHOW TABLES LIKE %(table_like)s', params,
1044 2731 aaronmk
            cacheable=True, log_level=4))
1045 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
1046 830 aaronmk
1047 2726 aaronmk
def table_exists(db, table):
1048
    table = sql_gen.as_Table(table)
1049 2730 aaronmk
    return list(tables(db, table.schema, table.name, exact=True)) != []
1050 2726 aaronmk
1051
def errors_table(db, table, if_exists=True):
1052
    '''
1053
    @param if_exists If set, returns None if the errors table doesn't exist
1054
    @return None|sql_gen.Table
1055
    '''
1056
    table = sql_gen.as_Table(table)
1057
    if table.srcs != (): table = table.srcs[0]
1058
1059
    errors_table = sql_gen.suffixed_table(table, '.errors')
1060
    if if_exists and not table_exists(db, errors_table): return None
1061
    return errors_table
1062
1063 833 aaronmk
##### Database management
1064
1065 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
1066
    '''For kw_args, see tables()'''
1067
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
1068 833 aaronmk
1069 832 aaronmk
##### Heuristic queries
1070
1071 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
1072 1554 aaronmk
    '''Recovers from errors.
1073 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
1074
    '''
1075 2642 aaronmk
    row = sql_gen.ColDict(db, table, row)
1076 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
1077
1078 471 aaronmk
    try:
1079 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
1080 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
1081
            row_ct_ref[0] += cur.rowcount
1082
        return value(cur)
1083 471 aaronmk
    except DuplicateKeyException, e:
1084 2653 aaronmk
        row = sql_gen.ColDict(db, table,
1085
            util.dict_subset_right_join(row, e.cols))
1086
        return value(select(db, table, [pkey_], row, recover=True))
1087 471 aaronmk
1088 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
1089 830 aaronmk
    '''Recovers from errors'''
1090 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
1091 14 aaronmk
    except StopIteration:
1092 40 aaronmk
        if not create: raise
1093 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
1094 2078 aaronmk
1095 2593 aaronmk
def is_func_result(col):
1096
    return col.table.name.find('(') >= 0 and col.name == 'result'
1097
1098 2592 aaronmk
def into_table_name(out_table, in_tables0, mapping, is_func):
1099 2615 aaronmk
    def in_col_str(in_col):
1100
        in_col = sql_gen.remove_col_rename(in_col)
1101
        if isinstance(in_col, sql_gen.Col):
1102
            table = in_col.table
1103
            if table == in_tables0:
1104
                in_col = sql_gen.to_name_only_col(in_col)
1105
            elif is_func_result(in_col): in_col = table # omit col name
1106
        return str(in_col)
1107
1108 2580 aaronmk
    str_ = str(out_table)
1109
    if is_func:
1110 2615 aaronmk
        str_ += '('
1111 2580 aaronmk
1112 2615 aaronmk
        try: value_in_col = mapping['value']
1113
        except KeyError:
1114 2654 aaronmk
            str_ += ', '.join((str(k)+'='+in_col_str(v)
1115 2615 aaronmk
                for k, v in mapping.iteritems()))
1116
        else: str_ += in_col_str(value_in_col)
1117
1118
        str_ += ')'
1119 2580 aaronmk
    else: str_ += '_pkeys'
1120
    return str_
1121
1122 2508 aaronmk
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None, into=None,
1123 2552 aaronmk
    default=None, is_func=False):
1124 2078 aaronmk
    '''Recovers from errors.
1125
    Only works under PostgreSQL (uses INSERT RETURNING).
1126 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
1127
        tables to join with it using the main input table's pkey
1128 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
1129 2612 aaronmk
        * out_table_col: str (*not* sql_gen.Col)
1130
        * in_table_col: sql_gen.Col|literal-value
1131 2489 aaronmk
    @param into The table to contain the output and input pkeys.
1132 2574 aaronmk
        Defaults to `out_table.name+'_pkeys'`.
1133 2509 aaronmk
    @param default The *output* column to use as the pkey for missing rows.
1134
        If this output column does not exist in the mapping, uses None.
1135 2552 aaronmk
    @param is_func Whether out_table is the name of a SQL function, not a table
1136 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
1137 2078 aaronmk
    '''
1138 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
1139 2766 aaronmk
    assert mapping != {}
1140 2312 aaronmk
1141 2450 aaronmk
    def log_debug(msg): db.log_debug(msg, level=1.5)
1142 2505 aaronmk
    def col_ustr(str_):
1143 2567 aaronmk
        return strings.repr_no_u(sql_gen.remove_col_rename(str_))
1144 2450 aaronmk
1145 2486 aaronmk
    log_debug('********** New iteration **********')
1146 2505 aaronmk
    log_debug('Inserting these input columns into '+strings.as_tt(
1147
        out_table.to_str(db))+':\n'+strings.as_table(mapping, ustr=col_ustr))
1148 2463 aaronmk
1149 2382 aaronmk
    # Create input joins from list of input tables
1150
    in_tables_ = in_tables[:] # don't modify input!
1151
    in_tables0 = in_tables_.pop(0) # first table is separate
1152 2735 aaronmk
    errors_table_ = errors_table(db, in_tables0)
1153 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
1154 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
1155 2460 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v,
1156
        {in_pkey: sql_gen.join_same_not_null}) for v in in_tables_]
1157 2131 aaronmk
1158 2592 aaronmk
    if into == None:
1159
        into = into_table_name(out_table, in_tables0, mapping, is_func)
1160
    into = sql_gen.as_Table(into)
1161
1162 2702 aaronmk
    # Set column sources
1163
    in_cols = filter(sql_gen.is_table_col, mapping.values())
1164
    for col in in_cols:
1165
        if col.table == in_tables0: col.set_srcs(sql_gen.src_self)
1166
1167 2486 aaronmk
    log_debug('Joining together input tables into temp table')
1168 2395 aaronmk
    # Place in new table for speed and so don't modify input if values edited
1169 2584 aaronmk
    in_table = sql_gen.Table('in')
1170 2702 aaronmk
    mapping = dicts.join(mapping, flatten(db, in_table, input_joins, in_cols,
1171
        preserve=[in_pkey_col], start=0))
1172 2395 aaronmk
    input_joins = [in_table]
1173 2486 aaronmk
    db.log_debug('Temp table: '+strings.as_tt(in_table.to_str(db)), level=2)
1174 2395 aaronmk
1175 2656 aaronmk
    mapping = sql_gen.ColDict(db, out_table, mapping)
1176
        # after applying dicts.join() because that returns a plain dict
1177
1178 2509 aaronmk
    # Resolve default value column
1179
    try: default = mapping[default]
1180
    except KeyError:
1181
        if default != None:
1182
            db.log_debug('Default value column '
1183
                +strings.as_tt(strings.repr_no_u(default))
1184 2511 aaronmk
                +' does not exist in mapping, falling back to None', level=2.1)
1185 2509 aaronmk
            default = None
1186
1187 2279 aaronmk
    out_pkey = pkey(db, out_table, recover=True)
1188 2285 aaronmk
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
1189 2142 aaronmk
1190 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
1191 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
1192
1193 2201 aaronmk
    pkeys_table_exists_ref = [False]
1194 2420 aaronmk
    def insert_into_pkeys(joins, cols):
1195
        query, params = mk_select(db, joins, cols, order_by=None, start=0)
1196 2201 aaronmk
        if pkeys_table_exists_ref[0]:
1197 2489 aaronmk
            insert_select(db, into, pkeys_names, query, params)
1198 2201 aaronmk
        else:
1199 2489 aaronmk
            run_query_into(db, query, params, into=into)
1200 2201 aaronmk
            pkeys_table_exists_ref[0] = True
1201
1202 2429 aaronmk
    limit_ref = [None]
1203 2380 aaronmk
    conds = set()
1204 2233 aaronmk
    distinct_on = []
1205 2325 aaronmk
    def mk_main_select(joins, cols):
1206 2429 aaronmk
        return mk_select(db, joins, cols, conds, distinct_on,
1207
            limit=limit_ref[0], start=0)
1208 2132 aaronmk
1209 2519 aaronmk
    exc_strs = set()
1210 2309 aaronmk
    def log_exc(e):
1211 2519 aaronmk
        e_str = exc.str_(e, first_line_only=True)
1212
        log_debug('Caught exception: '+e_str)
1213
        assert e_str not in exc_strs # avoid infinite loops
1214
        exc_strs.add(e_str)
1215 2735 aaronmk
1216 2451 aaronmk
    def remove_all_rows():
1217 2450 aaronmk
        log_debug('Returning NULL for all rows')
1218 2429 aaronmk
        limit_ref[0] = 0 # just create an empty pkeys table
1219 2735 aaronmk
1220 2758 aaronmk
    def ignore(in_col, value, e):
1221
        track_data_error(db, errors_table_, in_col.srcs, value, e.cause.pgcode,
1222
            e.cause.pgerror)
1223
1224 2545 aaronmk
        in_col_str = strings.as_tt(repr(in_col))
1225 2544 aaronmk
        db.log_debug('Adding index on '+in_col_str+' to enable fast filtering',
1226
            level=2.5)
1227 2537 aaronmk
        add_index(db, in_col)
1228 2735 aaronmk
1229 2545 aaronmk
        log_debug('Ignoring rows with '+in_col_str+' = '
1230
            +strings.as_tt(repr(value)))
1231 2758 aaronmk
    def remove_rows(in_col, value, e):
1232
        ignore(in_col, value, e)
1233 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
1234
        assert cond not in conds # avoid infinite loops
1235 2380 aaronmk
        conds.add(cond)
1236 2758 aaronmk
    def invalid2null(in_col, value, e):
1237
        ignore(in_col, value, e)
1238 2403 aaronmk
        update(db, in_table, [(in_col, None)],
1239
            sql_gen.ColValueCond(in_col, value))
1240 2245 aaronmk
1241 2589 aaronmk
    def insert_pkeys_table(which):
1242
        return sql_gen.Table(sql_gen.add_suffix(in_table.name,
1243
            '_insert_'+which+'_pkeys'))
1244
    insert_out_pkeys = insert_pkeys_table('out')
1245
    insert_in_pkeys = insert_pkeys_table('in')
1246
1247 2206 aaronmk
    # Do inserts and selects
1248 2642 aaronmk
    join_cols = sql_gen.ColDict(db, out_table)
1249 2206 aaronmk
    while True:
1250 2521 aaronmk
        if limit_ref[0] == 0: # special case
1251
            log_debug('Creating an empty pkeys table')
1252
            cur = run_query_into(db, *mk_select(db, out_table, [out_pkey],
1253
                limit=limit_ref[0]), into=insert_out_pkeys)
1254
            break # don't do main case
1255
1256 2303 aaronmk
        has_joins = join_cols != {}
1257
1258 2305 aaronmk
        # Prepare to insert new rows
1259 2325 aaronmk
        insert_joins = input_joins[:] # don't modify original!
1260 2403 aaronmk
        insert_args = dict(recover=True, cacheable=False)
1261 2303 aaronmk
        if has_joins:
1262 2317 aaronmk
            distinct_on = [v.to_Col() for v in join_cols.values()]
1263 2325 aaronmk
            insert_joins.append(sql_gen.Join(out_table, join_cols,
1264
                sql_gen.filter_out))
1265
        else:
1266 2404 aaronmk
            insert_args.update(dict(returning=out_pkey, into=insert_out_pkeys))
1267 2520 aaronmk
        main_select = mk_main_select(insert_joins, mapping.values())[0]
1268 2303 aaronmk
1269 2486 aaronmk
        log_debug('Trying to insert new rows')
1270 2206 aaronmk
        try:
1271 2518 aaronmk
            cur = insert_select(db, out_table, mapping.keys(), main_select,
1272
                **insert_args)
1273 2357 aaronmk
            break # insert successful
1274 2206 aaronmk
        except DuplicateKeyException, e:
1275 2309 aaronmk
            log_exc(e)
1276
1277 2258 aaronmk
            old_join_cols = join_cols.copy()
1278 2565 aaronmk
            join_cols.update(util.dict_subset_right_join(mapping, e.cols))
1279 2486 aaronmk
            log_debug('Ignoring existing rows, comparing on these columns:\n'
1280 2505 aaronmk
                +strings.as_inline_table(join_cols, ustr=col_ustr))
1281 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
1282 2230 aaronmk
        except NullValueException, e:
1283 2309 aaronmk
            log_exc(e)
1284
1285 2230 aaronmk
            out_col, = e.cols
1286
            try: in_col = mapping[out_col]
1287 2356 aaronmk
            except KeyError:
1288 2486 aaronmk
                log_debug('Missing mapping for NOT NULL column '+out_col)
1289 2451 aaronmk
                remove_all_rows()
1290 2758 aaronmk
            else: remove_rows(in_col, None, e)
1291 2542 aaronmk
        except FunctionValueException, e:
1292
            log_exc(e)
1293
1294
            func_name = e.name
1295
            value = e.value
1296
            for out_col, in_col in mapping.iteritems():
1297 2758 aaronmk
                in_col = sql_gen.unwrap_func_call(in_col, func_name)
1298
                invalid2null(in_col, value, e)
1299 2525 aaronmk
        except MissingCastException, e:
1300
            log_exc(e)
1301
1302
            out_col = e.col
1303 2706 aaronmk
            type_ = e.type
1304
1305 2724 aaronmk
            log_debug('Casting '+strings.as_tt(out_col)+' input to '
1306
                +strings.as_tt(type_))
1307 2735 aaronmk
            def wrap_func(col): return cast(db, type_, col, errors_table_)
1308 2705 aaronmk
            mapping[out_col] = sql_gen.wrap(wrap_func, mapping[out_col])
1309 2429 aaronmk
        except DatabaseErrors, e:
1310
            log_exc(e)
1311
1312 2531 aaronmk
            msg = 'No handler for exception: '+exc.str_(e)
1313 2451 aaronmk
            warnings.warn(DbWarning(msg))
1314
            log_debug(msg)
1315
            remove_all_rows()
1316 2358 aaronmk
        # after exception handled, rerun loop with additional constraints
1317 2132 aaronmk
1318 2357 aaronmk
    if row_ct_ref != None and cur.rowcount >= 0:
1319
        row_ct_ref[0] += cur.rowcount
1320
1321
    if has_joins:
1322
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
1323 2486 aaronmk
        log_debug('Getting output table pkeys of existing/inserted rows')
1324 2420 aaronmk
        insert_into_pkeys(select_joins, pkeys_cols)
1325 2357 aaronmk
    else:
1326 2404 aaronmk
        add_row_num(db, insert_out_pkeys) # for joining with input pkeys
1327 2357 aaronmk
1328 2486 aaronmk
        log_debug('Getting input table pkeys of inserted rows')
1329 2357 aaronmk
        run_query_into(db, *mk_main_select(input_joins, [in_pkey]),
1330 2404 aaronmk
            into=insert_in_pkeys)
1331
        add_row_num(db, insert_in_pkeys) # for joining with output pkeys
1332 2357 aaronmk
1333 2428 aaronmk
        assert table_row_count(db, insert_out_pkeys) == table_row_count(db,
1334
            insert_in_pkeys)
1335
1336 2486 aaronmk
        log_debug('Combining output and input pkeys in inserted order')
1337 2404 aaronmk
        pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
1338 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
1339 2420 aaronmk
        insert_into_pkeys(pkey_joins, pkeys_names)
1340 2357 aaronmk
1341 2486 aaronmk
    db.log_debug('Adding pkey on pkeys table to enable fast joins', level=2.5)
1342 2594 aaronmk
    add_pkey(db, into)
1343 2407 aaronmk
1344 2508 aaronmk
    log_debug('Setting pkeys of missing rows to '+strings.as_tt(repr(default)))
1345 2489 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(into,
1346 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
1347
        # must use join_same_not_null or query will take forever
1348 2420 aaronmk
    insert_into_pkeys(missing_rows_joins,
1349 2508 aaronmk
        [in_pkey_col, sql_gen.NamedCol(out_pkey, default)])
1350 2357 aaronmk
1351 2489 aaronmk
    assert table_row_count(db, into) == table_row_count(db, in_table)
1352 2428 aaronmk
1353 2702 aaronmk
    srcs = []
1354
    if is_func: srcs = sql_gen.cols_srcs(in_cols)
1355
    return sql_gen.Col(out_pkey, into, srcs)
1356 2115 aaronmk
1357
##### Data cleanup
1358
1359 2290 aaronmk
def cleanup_table(db, table, cols):
1360 2677 aaronmk
    table = sql_gen.as_Table(table)
1361
    cols = map(sql_gen.as_Col, cols)
1362 2115 aaronmk
1363 2677 aaronmk
    expr = ('nullif(nullif(trim(both from %s), '+db.esc_value('')+'), '
1364
        +db.esc_value(r'\N')+')')
1365
    changes = [(v, sql_gen.CustomCode(expr % v.to_str(db)))
1366
        for v in cols]
1367 2115 aaronmk
1368 2677 aaronmk
    update(db, table, changes)