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 2771 aaronmk
    exc.add_msg(e, 'query: '+strings.ustr(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 2792 aaronmk
                self.run_query("SELECT set_config('search_path', "
202
                    +self.esc_value(schemas_)
203
                    +" || current_setting('search_path'), false)", log_level=3)
204 1869 aaronmk
205
        return self.__db
206 1889 aaronmk
207 1891 aaronmk
    class DbCursor(Proxy):
208 1927 aaronmk
        def __init__(self, outer):
209 1891 aaronmk
            Proxy.__init__(self, outer.db.cursor())
210 2191 aaronmk
            self.outer = outer
211 1927 aaronmk
            self.query_results = outer.query_results
212 1894 aaronmk
            self.query_lookup = None
213 1891 aaronmk
            self.result = []
214 1889 aaronmk
215 1894 aaronmk
        def execute(self, query, params=None):
216 2784 aaronmk
            assert params == None or params == [] or params == ()
217 2763 aaronmk
            if params == [] or params == (): params = None
218
                # None turns off the mogrifier
219 2762 aaronmk
220 2764 aaronmk
            self._is_insert = query.startswith('INSERT')
221 1894 aaronmk
            self.query_lookup = _query_lookup(query, params)
222 2148 aaronmk
            try:
223 2191 aaronmk
                try:
224 2763 aaronmk
                    cur = self.inner.execute(query, params)
225 2191 aaronmk
                    self.outer.do_autocommit()
226 2648 aaronmk
                finally: self.query = get_cur_query(self.inner, query, params)
227 1904 aaronmk
            except Exception, e:
228 2170 aaronmk
                _add_cursor_info(e, self, query, params)
229 1904 aaronmk
                self.result = e # cache the exception as the result
230
                self._cache_result()
231
                raise
232 1930 aaronmk
            # Fetch all rows so result will be cached
233
            if self.rowcount == 0 and not self._is_insert: consume_rows(self)
234 2762 aaronmk
            return cur
235 1894 aaronmk
236 1891 aaronmk
        def fetchone(self):
237
            row = self.inner.fetchone()
238 1899 aaronmk
            if row != None: self.result.append(row)
239
            # otherwise, fetched all rows
240 1904 aaronmk
            else: self._cache_result()
241
            return row
242
243
        def _cache_result(self):
244 1906 aaronmk
            # For inserts, only cache exceptions since inserts are not
245
            # idempotent, but an invalid insert will always be invalid
246 1930 aaronmk
            if self.query_results != None and (not self._is_insert
247 1906 aaronmk
                or isinstance(self.result, Exception)):
248
249 1894 aaronmk
                assert self.query_lookup != None
250 1916 aaronmk
                self.query_results[self.query_lookup] = self.CacheCursor(
251
                    util.dict_subset(dicts.AttrsDictView(self),
252
                    ['query', 'result', 'rowcount', 'description']))
253 1906 aaronmk
254 1916 aaronmk
        class CacheCursor:
255
            def __init__(self, cached_result): self.__dict__ = cached_result
256
257 1927 aaronmk
            def execute(self, *args, **kw_args):
258 1916 aaronmk
                if isinstance(self.result, Exception): raise self.result
259
                # otherwise, result is a rows list
260
                self.iter = iter(self.result)
261
262
            def fetchone(self):
263
                try: return self.iter.next()
264
                except StopIteration: return None
265 1891 aaronmk
266 2212 aaronmk
    def esc_value(self, value):
267 2663 aaronmk
        try: str_ = self.mogrify('%s', [value])
268
        except NotImplementedError, e:
269
            module = util.root_module(self.db)
270
            if module == 'MySQLdb':
271
                import _mysql
272
                str_ = _mysql.escape_string(value)
273
            else: raise e
274 2374 aaronmk
        return strings.to_unicode(str_)
275 2212 aaronmk
276 2347 aaronmk
    def esc_name(self, name): return esc_name(self, name) # calls global func
277
278 2665 aaronmk
    def can_mogrify(self):
279 2663 aaronmk
        module = util.root_module(self.db)
280 2665 aaronmk
        return module == 'psycopg2'
281 2663 aaronmk
282 2665 aaronmk
    def mogrify(self, query, params=None):
283
        if self.can_mogrify(): return self.db.cursor().mogrify(query, params)
284
        else: raise NotImplementedError("Can't mogrify query")
285
286 2671 aaronmk
    def print_notices(self):
287 2725 aaronmk
        if hasattr(self.db, 'notices'):
288
            for msg in self.db.notices:
289
                if msg not in self._notices_seen:
290
                    self._notices_seen.add(msg)
291
                    self.log_debug(msg, level=2)
292 2671 aaronmk
293 2793 aaronmk
    def run_query(self, query, cacheable=False, log_level=2,
294 2464 aaronmk
        debug_msg_ref=None):
295 2445 aaronmk
        '''
296 2464 aaronmk
        @param log_ignore_excs The log_level will be increased by 2 if the query
297
            throws one of these exceptions.
298 2664 aaronmk
        @param debug_msg_ref If specified, the log message will be returned in
299
            this instead of being output. This allows you to filter log messages
300
            depending on the result of the query.
301 2445 aaronmk
        '''
302 2167 aaronmk
        assert query != None
303
304 2047 aaronmk
        if not self.caching: cacheable = False
305 1903 aaronmk
        used_cache = False
306 2664 aaronmk
307
        def log_msg(query):
308
            if used_cache: cache_status = 'cache hit'
309
            elif cacheable: cache_status = 'cache miss'
310
            else: cache_status = 'non-cacheable'
311
            return 'DB query: '+cache_status+':\n'+strings.as_code(query, 'SQL')
312
313 1903 aaronmk
        try:
314 1927 aaronmk
            # Get cursor
315
            if cacheable:
316 2793 aaronmk
                query_lookup = _query_lookup(query, [])
317 1927 aaronmk
                try:
318
                    cur = self.query_results[query_lookup]
319
                    used_cache = True
320
                except KeyError: cur = self.DbCursor(self)
321
            else: cur = self.db.cursor()
322
323 2664 aaronmk
            # Log query
324
            if self.debug and debug_msg_ref == None: # log before running
325
                self.log_debug(log_msg(query), log_level)
326
327 1927 aaronmk
            # Run query
328 2793 aaronmk
            cur.execute(query)
329 1903 aaronmk
        finally:
330 2671 aaronmk
            self.print_notices()
331 2664 aaronmk
            if self.debug and debug_msg_ref != None: # return after running
332 2793 aaronmk
                debug_msg_ref[0] = log_msg(str(get_cur_query(cur, query)))
333 1903 aaronmk
334
        return cur
335 1914 aaronmk
336
    def is_cached(self, query, params=None):
337
        return _query_lookup(query, params) in self.query_results
338 2139 aaronmk
339 2683 aaronmk
    def with_autocommit(self, func, autocommit=True):
340
        prev_autocommit = self.db.autocommit
341
        self.db.autocommit = autocommit
342
        try: return func()
343
        finally: self.db.autocommit = prev_autocommit
344
345 2139 aaronmk
    def with_savepoint(self, func):
346 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
347 2443 aaronmk
        self.run_query('SAVEPOINT '+savepoint, log_level=4)
348 2139 aaronmk
        self._savepoint += 1
349 2683 aaronmk
        try:
350 2139 aaronmk
            try: return_val = func()
351
            finally:
352
                self._savepoint -= 1
353
                assert self._savepoint >= 0
354
        except:
355 2443 aaronmk
            self.run_query('ROLLBACK TO SAVEPOINT '+savepoint, log_level=4)
356 2139 aaronmk
            raise
357
        else:
358 2443 aaronmk
            self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
359 2191 aaronmk
            self.do_autocommit()
360 2139 aaronmk
            return return_val
361 2191 aaronmk
362
    def do_autocommit(self):
363
        '''Autocommits if outside savepoint'''
364
        assert self._savepoint >= 0
365
        if self.autocommit and self._savepoint == 0:
366 2668 aaronmk
            self.log_debug('Autocommitting')
367 2191 aaronmk
            self.db.commit()
368 2643 aaronmk
369
    def col_default(self, col):
370
        table = sql_gen.Table('columns', 'information_schema')
371
372
        conds = [('table_name', col.table.name), ('column_name', col.name)]
373
        schema = col.table.schema
374
        if schema != None: conds.append(('table_schema', schema))
375
376 2660 aaronmk
        return sql_gen.as_Code(value(select(self, table, ['column_default'],
377
            conds, order_by='table_schema', limit=1, log_level=3)))
378 2643 aaronmk
            # TODO: order_by search_path schema order
379 1849 aaronmk
380 1869 aaronmk
connect = DbConn
381
382 832 aaronmk
##### Querying
383
384 2068 aaronmk
def mogrify(db, query, params):
385
    module = util.root_module(db.db)
386
    if module == 'psycopg2': return db.db.cursor().mogrify(query, params)
387
    else: raise NotImplementedError("Can't mogrify query for "+module+
388
        ' database')
389
390 832 aaronmk
##### Recoverable querying
391 15 aaronmk
392 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
393 11 aaronmk
394 2791 aaronmk
def run_query(db, query, recover=None, cacheable=False, log_level=2,
395
    log_ignore_excs=None, **kw_args):
396 2794 aaronmk
    '''For params, see DbConn.run_query()'''
397 830 aaronmk
    if recover == None: recover = False
398 2464 aaronmk
    if log_ignore_excs == None: log_ignore_excs = ()
399
    log_ignore_excs = tuple(log_ignore_excs)
400 830 aaronmk
401 2666 aaronmk
    debug_msg_ref = None # usually, db.run_query() logs query before running it
402
    # But if filtering with log_ignore_excs, wait until after exception parsing
403
    if log_ignore_excs != () or not db.can_mogrify(): debug_msg_ref = [None]
404
405 2148 aaronmk
    try:
406 2464 aaronmk
        try:
407 2794 aaronmk
            def run(): return db.run_query(query, cacheable, log_level,
408 2793 aaronmk
                debug_msg_ref, **kw_args)
409 2791 aaronmk
            if recover and not db.is_cached(query, []):
410 2464 aaronmk
                return with_savepoint(db, run)
411
            else: return run() # don't need savepoint if cached
412
        except Exception, e:
413
            if not recover: raise # need savepoint to run index_cols()
414
            msg = exc.str_(e)
415
416
            match = re.search(r'duplicate key value violates unique constraint '
417 2493 aaronmk
                r'"((_?[^\W_]+)_.+?)"', msg)
418 2464 aaronmk
            if match:
419
                constraint, table = match.groups()
420
                try: cols = index_cols(db, table, constraint)
421
                except NotImplementedError: raise e
422
                else: raise DuplicateKeyException(constraint, cols, e)
423
424 2493 aaronmk
            match = re.search(r'null value in column "(.+?)" violates not-null'
425 2464 aaronmk
                r' constraint', msg)
426
            if match: raise NullValueException('NOT NULL', [match.group(1)], e)
427
428
            match = re.search(r'\b(?:invalid input (?:syntax|value)\b.*?'
429
                r'|date/time field value out of range): "(.+?)"\n'
430 2535 aaronmk
                r'(?:(?s).*?)\bfunction "(.+?)"', msg)
431 2464 aaronmk
            if match:
432
                value, name = match.groups()
433
                raise FunctionValueException(name, strings.to_unicode(value), e)
434
435 2526 aaronmk
            match = re.search(r'column "(.+?)" is of type (.+?) but expression '
436 2523 aaronmk
                r'is of type', msg)
437
            if match:
438
                col, type_ = match.groups()
439
                raise MissingCastException(type_, col, e)
440
441 2493 aaronmk
            match = re.search(r'relation "(.+?)" already exists', msg)
442 2464 aaronmk
            if match: raise DuplicateTableException(match.group(1), e)
443
444 2493 aaronmk
            match = re.search(r'function "(.+?)" already exists', msg)
445 2464 aaronmk
            if match: raise DuplicateFunctionException(match.group(1), e)
446
447
            raise # no specific exception raised
448
    except log_ignore_excs:
449
        log_level += 2
450
        raise
451
    finally:
452 2666 aaronmk
        if debug_msg_ref != None and debug_msg_ref[0] != None:
453
            db.log_debug(debug_msg_ref[0], log_level)
454 830 aaronmk
455 832 aaronmk
##### Basic queries
456
457 2153 aaronmk
def next_version(name):
458 2163 aaronmk
    version = 1 # first existing name was version 0
459 2586 aaronmk
    match = re.match(r'^(.*)#(\d+)$', name)
460 2153 aaronmk
    if match:
461 2586 aaronmk
        name, version = match.groups()
462
        version = int(version)+1
463 2588 aaronmk
    return sql_gen.add_suffix(name, '#'+str(version))
464 2153 aaronmk
465 2789 aaronmk
def run_query_into(db, query, into=None, add_indexes_=False, **kw_args):
466 2085 aaronmk
    '''Outputs a query to a temp table.
467
    For params, see run_query().
468
    '''
469 2789 aaronmk
    if into == None: return run_query(db, query, **kw_args)
470 2790 aaronmk
471
    assert isinstance(into, sql_gen.Table)
472
473
    kw_args['recover'] = True
474
    kw_args.setdefault('log_ignore_excs', (DuplicateTableException,))
475
476
    temp = not db.autocommit # tables are permanent in autocommit mode
477
    # "temporary tables cannot specify a schema name", so remove schema
478
    if temp: into.schema = None
479
480
    # Create table
481
    while True:
482
        create_query = 'CREATE'
483
        if temp: create_query += ' TEMP'
484
        create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
485 2385 aaronmk
486 2790 aaronmk
        try:
487
            cur = run_query(db, create_query, **kw_args)
488
                # CREATE TABLE AS sets rowcount to # rows in query
489
            break
490
        except DuplicateTableException, e:
491
            into.name = next_version(into.name)
492
            # try again with next version of name
493
494
    if add_indexes_: add_indexes(db, into)
495
496
    return cur
497 2085 aaronmk
498 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
499
500 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
501
502 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
503 2293 aaronmk
    start=None, order_by=order_by_pkey, default_table=None):
504 1981 aaronmk
    '''
505 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
506 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
507 1981 aaronmk
    @param fields Use None to select all fields in the table
508 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
509 2379 aaronmk
        * container can be any iterable type
510 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
511
        * compare_right_side: sql_gen.ValueCond|literal value
512 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
513
        use all columns
514 2786 aaronmk
    @return query
515 1981 aaronmk
    '''
516 2315 aaronmk
    # Parse tables param
517 2121 aaronmk
    if not lists.is_seq(tables): tables = [tables]
518 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
519 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
520 2121 aaronmk
521 2315 aaronmk
    # Parse other params
522 2376 aaronmk
    if conds == None: conds = []
523 2650 aaronmk
    elif dicts.is_dict(conds): conds = conds.items()
524 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
525 135 aaronmk
    assert limit == None or type(limit) == int
526 865 aaronmk
    assert start == None or type(start) == int
527 2315 aaronmk
    if order_by is order_by_pkey:
528
        if distinct_on != []: order_by = None
529
        else: order_by = pkey(db, table0, recover=True)
530 865 aaronmk
531 2315 aaronmk
    query = 'SELECT'
532 2056 aaronmk
533 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
534 2056 aaronmk
535 2200 aaronmk
    # DISTINCT ON columns
536 2233 aaronmk
    if distinct_on != []:
537 2467 aaronmk
        query += '\nDISTINCT'
538 2254 aaronmk
        if distinct_on is not distinct_on_all:
539 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
540
541
    # Columns
542 2467 aaronmk
    query += '\n'
543 1135 aaronmk
    if fields == None: query += '*'
544 2765 aaronmk
    else:
545
        assert fields != []
546
        query += '\n, '.join(map(parse_col, fields))
547 2200 aaronmk
548
    # Main table
549 2467 aaronmk
    query += '\nFROM '+table0.to_str(db)
550 865 aaronmk
551 2122 aaronmk
    # Add joins
552 2271 aaronmk
    left_table = table0
553 2263 aaronmk
    for join_ in tables:
554
        table = join_.table
555 2238 aaronmk
556 2343 aaronmk
        # Parse special values
557
        if join_.type_ is sql_gen.filter_out: # filter no match
558 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
559
                None))
560 2343 aaronmk
561 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
562 2122 aaronmk
563
        left_table = table
564
565 865 aaronmk
    missing = True
566 2376 aaronmk
    if conds != []:
567 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
568
        else: whitespace = '\n'
569 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
570
            .to_str(db) for l, r in conds], 'WHERE')
571 865 aaronmk
        missing = False
572 2227 aaronmk
    if order_by != None:
573 2467 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
574
    if limit != None: query += '\nLIMIT '+str(limit); missing = False
575 865 aaronmk
    if start != None:
576 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
577 865 aaronmk
        missing = False
578
    if missing: warnings.warn(DbWarning(
579
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
580
581 2786 aaronmk
    return query
582 11 aaronmk
583 2054 aaronmk
def select(db, *args, **kw_args):
584
    '''For params, see mk_select() and run_query()'''
585
    recover = kw_args.pop('recover', None)
586
    cacheable = kw_args.pop('cacheable', True)
587 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
588 2054 aaronmk
589 2791 aaronmk
    return run_query(db, mk_select(db, *args, **kw_args), recover, cacheable,
590
        log_level=log_level)
591 2054 aaronmk
592 2788 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, returning=None,
593
    embeddable=False):
594 1960 aaronmk
    '''
595
    @param returning str|None An inserted column (such as pkey) to return
596 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
597 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
598
        query will be fully cached, not just if it raises an exception.
599 1960 aaronmk
    '''
600 2754 aaronmk
    table = sql_gen.remove_table_rename(sql_gen.as_Table(table))
601 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
602 2646 aaronmk
    if cols != None:
603
        cols = [sql_gen.to_name_only_col(v, table).to_str(db) for v in cols]
604 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
605 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
606 2063 aaronmk
607
    # Build query
608 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
609
    query = first_line
610 2467 aaronmk
    if cols != None: query += '\n('+', '.join(cols)+')'
611
    query += '\n'+select_query
612 2063 aaronmk
613
    if returning != None:
614 2743 aaronmk
        query += '\nRETURNING '+sql_gen.to_name_only_col(returning).to_str(db)
615 2063 aaronmk
616 2070 aaronmk
    if embeddable:
617 2327 aaronmk
        assert returning != None
618
619 2070 aaronmk
        # Create function
620 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
621 2327 aaronmk
        return_type = 'SETOF '+returning.to_str(db)+'%TYPE'
622 2189 aaronmk
        while True:
623
            try:
624 2698 aaronmk
                function = sql_gen.TempFunction(function_name, db.autocommit)
625 2194 aaronmk
626 2189 aaronmk
                function_query = '''\
627 2698 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''()
628 2467 aaronmk
RETURNS '''+return_type+'''
629
LANGUAGE sql
630
AS $$
631 2788 aaronmk
'''+query+''';
632 2467 aaronmk
$$;
633 2070 aaronmk
'''
634 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
635 2464 aaronmk
                    log_ignore_excs=(DuplicateFunctionException,))
636 2189 aaronmk
                break # this version was successful
637
            except DuplicateFunctionException, e:
638
                function_name = next_version(function_name)
639
                # try again with next version of name
640 2070 aaronmk
641 2337 aaronmk
        # Return query that uses function
642 2698 aaronmk
        func_table = sql_gen.NamedTable('f', sql_gen.FunctionCall(function),
643 2742 aaronmk
            [returning]) # AS clause requires function alias
644 2787 aaronmk
        return mk_select(db, func_table, start=0, order_by=None)
645 2070 aaronmk
646 2787 aaronmk
    return query
647 2066 aaronmk
648
def insert_select(db, *args, **kw_args):
649 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
650 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
651
        values in
652 2072 aaronmk
    '''
653 2386 aaronmk
    into = kw_args.pop('into', None)
654
    if into != None: kw_args['embeddable'] = True
655 2066 aaronmk
    recover = kw_args.pop('recover', None)
656
    cacheable = kw_args.pop('cacheable', True)
657 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
658 2066 aaronmk
659 2789 aaronmk
    return run_query_into(db, mk_insert_select(db, *args, **kw_args), into,
660 2787 aaronmk
        recover=recover, cacheable=cacheable, log_level=log_level)
661 2063 aaronmk
662 2738 aaronmk
default = sql_gen.default # tells insert() to use the default value for a column
663 2066 aaronmk
664 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
665 2085 aaronmk
    '''For params, see insert_select()'''
666 1960 aaronmk
    if lists.is_seq(row): cols = None
667
    else:
668
        cols = row.keys()
669
        row = row.values()
670 2738 aaronmk
    row = list(row) # ensure that "== []" works
671 1960 aaronmk
672 2738 aaronmk
    if row == []: query = None
673
    else: query = sql_gen.Values(row).to_str(db)
674 1961 aaronmk
675 2788 aaronmk
    return insert_select(db, table, cols, query, *args, **kw_args)
676 11 aaronmk
677 2402 aaronmk
def mk_update(db, table, changes=None, cond=None):
678
    '''
679
    @param changes [(col, new_value),...]
680
        * container can be any iterable type
681
        * col: sql_gen.Code|str (for col name)
682
        * new_value: sql_gen.Code|literal value
683
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
684
    @return str query
685
    '''
686
    query = 'UPDATE '+sql_gen.as_Table(table).to_str(db)+'\nSET\n'
687 2405 aaronmk
    query += ',\n'.join((sql_gen.to_name_only_col(col, table).to_str(db)+' = '
688 2402 aaronmk
        +sql_gen.as_Value(new_value).to_str(db) for col, new_value in changes))
689 2467 aaronmk
    if cond != None: query += '\nWHERE\n'+cond.to_str(db)
690 2402 aaronmk
691
    return query
692
693
def update(db, *args, **kw_args):
694
    '''For params, see mk_update() and run_query()'''
695
    recover = kw_args.pop('recover', None)
696
697 2791 aaronmk
    return run_query(db, mk_update(db, *args, **kw_args), recover)
698 2402 aaronmk
699 135 aaronmk
def last_insert_id(db):
700 1849 aaronmk
    module = util.root_module(db.db)
701 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
702
    elif module == 'MySQLdb': return db.insert_id()
703
    else: return None
704 13 aaronmk
705 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
706 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
707 2415 aaronmk
    to names that will be distinct among the columns' tables.
708 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
709 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
710
    @param into The table for the new columns.
711 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
712
        columns will be included in the mapping even if they are not in cols.
713
        The tables of the provided Col objects will be changed to into, so make
714
        copies of them if you want to keep the original tables.
715
    @param as_items Whether to return a list of dict items instead of a dict
716 2383 aaronmk
    @return dict(orig_col=new_col, ...)
717
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
718 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
719
        * All mappings use the into table so its name can easily be
720 2383 aaronmk
          changed for all columns at once
721
    '''
722 2415 aaronmk
    cols = lists.uniqify(cols)
723
724 2394 aaronmk
    items = []
725 2389 aaronmk
    for col in preserve:
726 2390 aaronmk
        orig_col = copy.copy(col)
727 2392 aaronmk
        col.table = into
728 2394 aaronmk
        items.append((orig_col, col))
729
    preserve = set(preserve)
730
    for col in cols:
731 2716 aaronmk
        if col not in preserve:
732
            items.append((col, sql_gen.Col(str(col), into, col.srcs)))
733 2394 aaronmk
734
    if not as_items: items = dict(items)
735
    return items
736 2383 aaronmk
737 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
738 2391 aaronmk
    '''For params, see mk_flatten_mapping()
739
    @return See return value of mk_flatten_mapping()
740
    '''
741 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
742
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
743 2786 aaronmk
    run_query_into(db, mk_select(db, joins, cols, limit=limit, start=start),
744 2392 aaronmk
        into=into)
745 2394 aaronmk
    return dict(items)
746 2391 aaronmk
747 2751 aaronmk
def mk_track_data_error(db, errors_table, cols, value, error_code, error):
748
    cols = map(sql_gen.to_name_only_col, cols)
749
750
    columns_cols = ['column']
751
    columns = sql_gen.NamedValues('columns', columns_cols,
752
        [[c.name] for c in cols])
753
    values_cols = ['value', 'error_code', 'error']
754
    values = sql_gen.NamedValues('values', values_cols,
755
        [value, error_code, error])
756
757
    select_cols = columns_cols+values_cols
758
    name_only_cols = map(sql_gen.to_name_only_col, select_cols)
759 2755 aaronmk
    errors_table = sql_gen.NamedTable('errors', errors_table)
760 2751 aaronmk
    joins = [columns, sql_gen.Join(values, type_='CROSS'),
761
        sql_gen.Join(errors_table, dict(zip(name_only_cols, select_cols)),
762
        sql_gen.filter_out)]
763
764
    return mk_insert_select(db, errors_table, name_only_cols,
765 2787 aaronmk
        mk_select(db, joins, select_cols, order_by=None))
766 2751 aaronmk
767 2758 aaronmk
def track_data_error(db, errors_table, *args, **kw_args):
768
    '''
769
    @param errors_table If None, does nothing.
770
    '''
771
    if errors_table == None: return
772
    run_query(db, mk_track_data_error(db, errors_table, *args, **kw_args),
773
        cacheable=True, log_level=4)
774
775 2704 aaronmk
def cast(db, type_, col, errors_table=None):
776
    '''Casts an (unrenamed) column or value.
777 2769 aaronmk
    If errors_table set and col has srcs, saves errors in errors_table (using
778
    col's srcs attr as the source columns) and converts errors to warnings.
779 2704 aaronmk
    @param col sql_gen.Col|sql_gen.Literal
780
    @param errors_table None|sql_gen.Table|str
781
    '''
782 2769 aaronmk
    save_errors = (errors_table != None and isinstance(col, sql_gen.Col)
783
        and col.srcs != ())
784
    if not save_errors: # can't save errors
785
        return sql_gen.CustomCode(col.to_str(db)+'::'+type_) # just cast
786 2704 aaronmk
787
    assert not isinstance(col, sql_gen.NamedCol)
788
789 2770 aaronmk
    errors_table = sql_gen.as_Table(errors_table)
790
    srcs = map(sql_gen.to_name_only_col, col.srcs)
791
    function_name = str(sql_gen.FunctionCall(type_, *srcs))
792 2704 aaronmk
    function = sql_gen.TempFunction(function_name, db.autocommit)
793
794 2718 aaronmk
    while True:
795
        # Create function definition
796
        query = '''\
797 2704 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''(value text)
798
RETURNS '''+type_+'''
799
LANGUAGE plpgsql
800
STRICT
801
AS $$
802
BEGIN
803
    /* The explicit cast to the return type is needed to make the cast happen
804
    inside the try block. (Implicit casts to the return type happen at the end
805
    of the function, outside any block.) */
806
    RETURN value::'''+type_+''';
807
EXCEPTION
808
    WHEN data_exception THEN
809
        -- Save error in errors table.
810 2751 aaronmk
        -- Insert the value and error for *each* source column.
811
'''+mk_track_data_error(db, errors_table, srcs,
812
    *map(sql_gen.CustomCode, ['value', 'SQLSTATE', 'SQLERRM']))+''';
813 2704 aaronmk
814
        RAISE WARNING '%', SQLERRM;
815
        RETURN NULL;
816
END;
817
$$;
818
'''
819 2718 aaronmk
820
        # Create function
821
        try:
822
            run_query(db, query, recover=True, cacheable=True,
823
                log_ignore_excs=(DuplicateFunctionException,))
824
            break # successful
825
        except DuplicateFunctionException:
826 2770 aaronmk
            function.name = next_version(function.name)
827
            # try again with next version of name
828 2704 aaronmk
829
    return sql_gen.FunctionCall(function, col)
830
831 2414 aaronmk
##### Database structure queries
832
833 2426 aaronmk
def table_row_count(db, table, recover=None):
834 2786 aaronmk
    return value(run_query(db, mk_select(db, table, [sql_gen.row_count],
835 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
836 2426 aaronmk
837 2414 aaronmk
def table_cols(db, table, recover=None):
838
    return list(col_names(select(db, table, limit=0, order_by=None,
839 2443 aaronmk
        recover=recover, log_level=4)))
840 2414 aaronmk
841 2291 aaronmk
def pkey(db, table, recover=None):
842 832 aaronmk
    '''Assumed to be first column in table'''
843 2339 aaronmk
    return table_cols(db, table, recover)[0]
844 832 aaronmk
845 2559 aaronmk
not_null_col = 'not_null_col'
846 2340 aaronmk
847
def table_not_null_col(db, table, recover=None):
848
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
849
    if not_null_col in table_cols(db, table, recover): return not_null_col
850
    else: return pkey(db, table, recover)
851
852 853 aaronmk
def index_cols(db, table, index):
853
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
854
    automatically created. When you don't know whether something is a UNIQUE
855
    constraint or a UNIQUE index, use this function.'''
856 1909 aaronmk
    module = util.root_module(db.db)
857
    if module == 'psycopg2':
858
        return list(values(run_query(db, '''\
859 853 aaronmk
SELECT attname
860 866 aaronmk
FROM
861
(
862
        SELECT attnum, attname
863
        FROM pg_index
864
        JOIN pg_class index ON index.oid = indexrelid
865
        JOIN pg_class table_ ON table_.oid = indrelid
866
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
867
        WHERE
868 2782 aaronmk
            table_.relname = '''+db.esc_value(table)+'''
869
            AND index.relname = '''+db.esc_value(index)+'''
870 866 aaronmk
    UNION
871
        SELECT attnum, attname
872
        FROM
873
        (
874
            SELECT
875
                indrelid
876
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
877
                    AS indkey
878
            FROM pg_index
879
            JOIN pg_class index ON index.oid = indexrelid
880
            JOIN pg_class table_ ON table_.oid = indrelid
881
            WHERE
882 2782 aaronmk
                table_.relname = '''+db.esc_value(table)+'''
883
                AND index.relname = '''+db.esc_value(index)+'''
884 866 aaronmk
        ) s
885
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
886
) s
887 853 aaronmk
ORDER BY attnum
888 2782 aaronmk
'''
889
            , cacheable=True, log_level=4)))
890 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
891
        ' database')
892 853 aaronmk
893 464 aaronmk
def constraint_cols(db, table, constraint):
894 1849 aaronmk
    module = util.root_module(db.db)
895 464 aaronmk
    if module == 'psycopg2':
896
        return list(values(run_query(db, '''\
897
SELECT attname
898
FROM pg_constraint
899
JOIN pg_class ON pg_class.oid = conrelid
900
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
901
WHERE
902 2783 aaronmk
    relname = '''+db.esc_value(table)+'''
903
    AND conname = '''+db.esc_value(constraint)+'''
904 464 aaronmk
ORDER BY attnum
905 2783 aaronmk
'''
906
            )))
907 464 aaronmk
    else: raise NotImplementedError("Can't list constraint columns for "+module+
908
        ' database')
909
910 2096 aaronmk
row_num_col = '_row_num'
911
912 2688 aaronmk
def add_index(db, exprs, table=None, unique=False):
913
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
914 2538 aaronmk
    Currently, only function calls are supported as expressions.
915
    '''
916 2688 aaronmk
    if not lists.is_seq(exprs): exprs = [exprs]
917 2538 aaronmk
918 2688 aaronmk
    # Parse exprs
919
    old_exprs = exprs[:]
920
    exprs = []
921
    cols = []
922
    for i, expr in enumerate(old_exprs):
923
        expr = copy.deepcopy(expr) # don't modify input!
924
        expr = sql_gen.as_Col(expr)
925
926
        # Extract col
927
        if isinstance(expr, sql_gen.FunctionCall):
928
            col = expr.args[0]
929
            expr = sql_gen.Expr(expr)
930
        else: col = expr
931
932
        # Extract table
933
        if table == None:
934
            assert sql_gen.is_table_col(col)
935
            table = col.table
936
937
        col.table = None
938
939
        exprs.append(expr)
940
        cols.append(col)
941 2408 aaronmk
942 2688 aaronmk
    table = sql_gen.as_Table(table)
943
    index = sql_gen.Table(str(sql_gen.Col(','.join(map(str, cols)), table)))
944
945
    str_ = 'CREATE'
946
    if unique: str_ += ' UNIQUE'
947
    str_ += ' INDEX '+index.to_str(db)+' ON '+table.to_str(db)+' ('+(
948
        ', '.join((v.to_str(db) for v in exprs)))+')'
949
950
    try: run_query(db, str_, recover=True, cacheable=True, log_level=3)
951 2408 aaronmk
    except DuplicateTableException: pass # index already existed
952
953 2686 aaronmk
def add_pkey(db, table, cols=None, recover=None):
954
    '''Adds a primary key.
955
    @param cols [sql_gen.Col,...] The columns in the primary key.
956
        Defaults to the first column in the table.
957 2406 aaronmk
    @pre The table must not already have a primary key.
958
    '''
959
    table = sql_gen.as_Table(table)
960 2686 aaronmk
    if cols == None: cols = [pkey(db, table, recover)]
961
    col_strs = [sql_gen.to_name_only_col(v).to_str(db) for v in cols]
962 2406 aaronmk
963 2781 aaronmk
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD PRIMARY KEY ('
964
        +(', '.join(col_strs))+')', recover=True, cacheable=True, log_level=3,
965
        log_ignore_excs=(DuplicateTableException,))
966 2406 aaronmk
967 2772 aaronmk
already_indexed = object() # tells add_indexes() the pkey has already been added
968
969
def add_indexes(db, table, has_pkey=True):
970
    '''Adds an index on all columns in a table.
971
    @param has_pkey bool|already_indexed Whether a pkey instead of a regular
972
        index should be added on the first column.
973
        * If already_indexed, the pkey is assumed to have already been added
974
    '''
975
    cols = table_cols(db, table)
976
    if has_pkey:
977
        if has_pkey is not already_indexed: add_pkey(db, table)
978
        cols = cols[1:]
979
    for col in cols: add_index(db, col, table)
980
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 2773 aaronmk
    if has_pkey: has_pkey = already_indexed
1010
    def add_indexes_(): add_indexes(db, table, has_pkey)
1011
    if isinstance(col_indexes, list): col_indexes[0] = add_indexes_ # defer
1012
    elif col_indexes: add_indexes_() # add now
1013 2675 aaronmk
1014 2684 aaronmk
def vacuum(db, table):
1015
    table = sql_gen.as_Table(table)
1016
    db.with_autocommit(lambda: run_query(db, 'VACUUM ANALYZE '+table.to_str(db),
1017
        log_level=3))
1018
1019 2732 aaronmk
def truncate(db, table, schema='public'):
1020 2777 aaronmk
    table = sql_gen.as_Table(table, schema)
1021 2733 aaronmk
    return run_query(db, 'TRUNCATE '+table.to_str(db)+' CASCADE')
1022 2732 aaronmk
1023 2729 aaronmk
def tables(db, schema_like='public', table_like='%', exact=False):
1024
    if exact: compare = '='
1025
    else: compare = 'LIKE'
1026
1027 1849 aaronmk
    module = util.root_module(db.db)
1028 832 aaronmk
    if module == 'psycopg2':
1029 2779 aaronmk
        conds = [('schemaname', sql_gen.CompareCond(schema_like, compare)),
1030
            ('tablename', sql_gen.CompareCond(table_like, compare))]
1031
        return values(select(db, 'pg_tables', ['tablename'], conds,
1032
            order_by='tablename', log_level=4))
1033 1968 aaronmk
    elif module == 'MySQLdb':
1034 2779 aaronmk
        return values(run_query(db, 'SHOW TABLES LIKE '+db.esc_value(table_like)
1035
            , cacheable=True, log_level=4))
1036 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
1037 830 aaronmk
1038 2726 aaronmk
def table_exists(db, table):
1039
    table = sql_gen.as_Table(table)
1040 2730 aaronmk
    return list(tables(db, table.schema, table.name, exact=True)) != []
1041 2726 aaronmk
1042
def errors_table(db, table, if_exists=True):
1043
    '''
1044
    @param if_exists If set, returns None if the errors table doesn't exist
1045
    @return None|sql_gen.Table
1046
    '''
1047
    table = sql_gen.as_Table(table)
1048
    if table.srcs != (): table = table.srcs[0]
1049
1050
    errors_table = sql_gen.suffixed_table(table, '.errors')
1051
    if if_exists and not table_exists(db, errors_table): return None
1052
    return errors_table
1053
1054 833 aaronmk
##### Database management
1055
1056 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
1057
    '''For kw_args, see tables()'''
1058
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
1059 833 aaronmk
1060 832 aaronmk
##### Heuristic queries
1061
1062 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
1063 1554 aaronmk
    '''Recovers from errors.
1064 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
1065
    '''
1066 2642 aaronmk
    row = sql_gen.ColDict(db, table, row)
1067 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
1068
1069 471 aaronmk
    try:
1070 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
1071 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
1072
            row_ct_ref[0] += cur.rowcount
1073
        return value(cur)
1074 471 aaronmk
    except DuplicateKeyException, e:
1075 2653 aaronmk
        row = sql_gen.ColDict(db, table,
1076
            util.dict_subset_right_join(row, e.cols))
1077
        return value(select(db, table, [pkey_], row, recover=True))
1078 471 aaronmk
1079 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
1080 830 aaronmk
    '''Recovers from errors'''
1081 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
1082 14 aaronmk
    except StopIteration:
1083 40 aaronmk
        if not create: raise
1084 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
1085 2078 aaronmk
1086 2593 aaronmk
def is_func_result(col):
1087
    return col.table.name.find('(') >= 0 and col.name == 'result'
1088
1089 2592 aaronmk
def into_table_name(out_table, in_tables0, mapping, is_func):
1090 2615 aaronmk
    def in_col_str(in_col):
1091
        in_col = sql_gen.remove_col_rename(in_col)
1092
        if isinstance(in_col, sql_gen.Col):
1093
            table = in_col.table
1094
            if table == in_tables0:
1095
                in_col = sql_gen.to_name_only_col(in_col)
1096
            elif is_func_result(in_col): in_col = table # omit col name
1097
        return str(in_col)
1098
1099 2580 aaronmk
    str_ = str(out_table)
1100
    if is_func:
1101 2615 aaronmk
        str_ += '('
1102 2580 aaronmk
1103 2615 aaronmk
        try: value_in_col = mapping['value']
1104
        except KeyError:
1105 2654 aaronmk
            str_ += ', '.join((str(k)+'='+in_col_str(v)
1106 2615 aaronmk
                for k, v in mapping.iteritems()))
1107
        else: str_ += in_col_str(value_in_col)
1108
1109
        str_ += ')'
1110 2580 aaronmk
    else: str_ += '_pkeys'
1111
    return str_
1112
1113 2508 aaronmk
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None, into=None,
1114 2552 aaronmk
    default=None, is_func=False):
1115 2078 aaronmk
    '''Recovers from errors.
1116
    Only works under PostgreSQL (uses INSERT RETURNING).
1117 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
1118
        tables to join with it using the main input table's pkey
1119 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
1120 2612 aaronmk
        * out_table_col: str (*not* sql_gen.Col)
1121
        * in_table_col: sql_gen.Col|literal-value
1122 2489 aaronmk
    @param into The table to contain the output and input pkeys.
1123 2574 aaronmk
        Defaults to `out_table.name+'_pkeys'`.
1124 2509 aaronmk
    @param default The *output* column to use as the pkey for missing rows.
1125
        If this output column does not exist in the mapping, uses None.
1126 2552 aaronmk
    @param is_func Whether out_table is the name of a SQL function, not a table
1127 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
1128 2078 aaronmk
    '''
1129 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
1130 2312 aaronmk
1131 2450 aaronmk
    def log_debug(msg): db.log_debug(msg, level=1.5)
1132 2505 aaronmk
    def col_ustr(str_):
1133 2567 aaronmk
        return strings.repr_no_u(sql_gen.remove_col_rename(str_))
1134 2450 aaronmk
1135 2768 aaronmk
    out_pkey = pkey(db, out_table, recover=True)
1136
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
1137
1138
    if mapping == {}: # need at least one column for INSERT SELECT
1139
        mapping = {out_pkey: None} # ColDict will replace with default value
1140
1141 2486 aaronmk
    log_debug('********** New iteration **********')
1142 2505 aaronmk
    log_debug('Inserting these input columns into '+strings.as_tt(
1143
        out_table.to_str(db))+':\n'+strings.as_table(mapping, ustr=col_ustr))
1144 2463 aaronmk
1145 2382 aaronmk
    # Create input joins from list of input tables
1146
    in_tables_ = in_tables[:] # don't modify input!
1147
    in_tables0 = in_tables_.pop(0) # first table is separate
1148 2735 aaronmk
    errors_table_ = errors_table(db, in_tables0)
1149 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
1150 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
1151 2460 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v,
1152
        {in_pkey: sql_gen.join_same_not_null}) for v in in_tables_]
1153 2131 aaronmk
1154 2592 aaronmk
    if into == None:
1155
        into = into_table_name(out_table, in_tables0, mapping, is_func)
1156
    into = sql_gen.as_Table(into)
1157
1158 2702 aaronmk
    # Set column sources
1159
    in_cols = filter(sql_gen.is_table_col, mapping.values())
1160
    for col in in_cols:
1161
        if col.table == in_tables0: col.set_srcs(sql_gen.src_self)
1162
1163 2486 aaronmk
    log_debug('Joining together input tables into temp table')
1164 2395 aaronmk
    # Place in new table for speed and so don't modify input if values edited
1165 2584 aaronmk
    in_table = sql_gen.Table('in')
1166 2702 aaronmk
    mapping = dicts.join(mapping, flatten(db, in_table, input_joins, in_cols,
1167
        preserve=[in_pkey_col], start=0))
1168 2395 aaronmk
    input_joins = [in_table]
1169 2486 aaronmk
    db.log_debug('Temp table: '+strings.as_tt(in_table.to_str(db)), level=2)
1170 2395 aaronmk
1171 2656 aaronmk
    mapping = sql_gen.ColDict(db, out_table, mapping)
1172
        # after applying dicts.join() because that returns a plain dict
1173
1174 2509 aaronmk
    # Resolve default value column
1175
    try: default = mapping[default]
1176
    except KeyError:
1177
        if default != None:
1178
            db.log_debug('Default value column '
1179
                +strings.as_tt(strings.repr_no_u(default))
1180 2511 aaronmk
                +' does not exist in mapping, falling back to None', level=2.1)
1181 2509 aaronmk
            default = None
1182
1183 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
1184 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
1185
1186 2201 aaronmk
    pkeys_table_exists_ref = [False]
1187 2420 aaronmk
    def insert_into_pkeys(joins, cols):
1188 2786 aaronmk
        query = mk_select(db, joins, cols, order_by=None, start=0)
1189 2201 aaronmk
        if pkeys_table_exists_ref[0]:
1190 2786 aaronmk
            insert_select(db, into, pkeys_names, query)
1191 2201 aaronmk
        else:
1192 2786 aaronmk
            run_query_into(db, query, into=into)
1193 2201 aaronmk
            pkeys_table_exists_ref[0] = True
1194
1195 2429 aaronmk
    limit_ref = [None]
1196 2380 aaronmk
    conds = set()
1197 2233 aaronmk
    distinct_on = []
1198 2325 aaronmk
    def mk_main_select(joins, cols):
1199 2429 aaronmk
        return mk_select(db, joins, cols, conds, distinct_on,
1200
            limit=limit_ref[0], start=0)
1201 2132 aaronmk
1202 2519 aaronmk
    exc_strs = set()
1203 2309 aaronmk
    def log_exc(e):
1204 2519 aaronmk
        e_str = exc.str_(e, first_line_only=True)
1205
        log_debug('Caught exception: '+e_str)
1206
        assert e_str not in exc_strs # avoid infinite loops
1207
        exc_strs.add(e_str)
1208 2735 aaronmk
1209 2451 aaronmk
    def remove_all_rows():
1210 2450 aaronmk
        log_debug('Returning NULL for all rows')
1211 2429 aaronmk
        limit_ref[0] = 0 # just create an empty pkeys table
1212 2735 aaronmk
1213 2758 aaronmk
    def ignore(in_col, value, e):
1214
        track_data_error(db, errors_table_, in_col.srcs, value, e.cause.pgcode,
1215
            e.cause.pgerror)
1216
1217 2545 aaronmk
        in_col_str = strings.as_tt(repr(in_col))
1218 2544 aaronmk
        db.log_debug('Adding index on '+in_col_str+' to enable fast filtering',
1219
            level=2.5)
1220 2537 aaronmk
        add_index(db, in_col)
1221 2735 aaronmk
1222 2545 aaronmk
        log_debug('Ignoring rows with '+in_col_str+' = '
1223
            +strings.as_tt(repr(value)))
1224 2758 aaronmk
    def remove_rows(in_col, value, e):
1225
        ignore(in_col, value, e)
1226 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
1227
        assert cond not in conds # avoid infinite loops
1228 2380 aaronmk
        conds.add(cond)
1229 2758 aaronmk
    def invalid2null(in_col, value, e):
1230
        ignore(in_col, value, e)
1231 2403 aaronmk
        update(db, in_table, [(in_col, None)],
1232
            sql_gen.ColValueCond(in_col, value))
1233 2245 aaronmk
1234 2589 aaronmk
    def insert_pkeys_table(which):
1235
        return sql_gen.Table(sql_gen.add_suffix(in_table.name,
1236
            '_insert_'+which+'_pkeys'))
1237
    insert_out_pkeys = insert_pkeys_table('out')
1238
    insert_in_pkeys = insert_pkeys_table('in')
1239
1240 2206 aaronmk
    # Do inserts and selects
1241 2642 aaronmk
    join_cols = sql_gen.ColDict(db, out_table)
1242 2206 aaronmk
    while True:
1243 2521 aaronmk
        if limit_ref[0] == 0: # special case
1244
            log_debug('Creating an empty pkeys table')
1245 2786 aaronmk
            cur = run_query_into(db, mk_select(db, out_table, [out_pkey],
1246 2521 aaronmk
                limit=limit_ref[0]), into=insert_out_pkeys)
1247
            break # don't do main case
1248
1249 2303 aaronmk
        has_joins = join_cols != {}
1250
1251 2305 aaronmk
        # Prepare to insert new rows
1252 2325 aaronmk
        insert_joins = input_joins[:] # don't modify original!
1253 2403 aaronmk
        insert_args = dict(recover=True, cacheable=False)
1254 2303 aaronmk
        if has_joins:
1255 2317 aaronmk
            distinct_on = [v.to_Col() for v in join_cols.values()]
1256 2325 aaronmk
            insert_joins.append(sql_gen.Join(out_table, join_cols,
1257
                sql_gen.filter_out))
1258
        else:
1259 2404 aaronmk
            insert_args.update(dict(returning=out_pkey, into=insert_out_pkeys))
1260 2786 aaronmk
        main_select = mk_main_select(insert_joins, mapping.values())
1261 2303 aaronmk
1262 2486 aaronmk
        log_debug('Trying to insert new rows')
1263 2206 aaronmk
        try:
1264 2518 aaronmk
            cur = insert_select(db, out_table, mapping.keys(), main_select,
1265
                **insert_args)
1266 2357 aaronmk
            break # insert successful
1267 2206 aaronmk
        except DuplicateKeyException, e:
1268 2309 aaronmk
            log_exc(e)
1269
1270 2258 aaronmk
            old_join_cols = join_cols.copy()
1271 2565 aaronmk
            join_cols.update(util.dict_subset_right_join(mapping, e.cols))
1272 2486 aaronmk
            log_debug('Ignoring existing rows, comparing on these columns:\n'
1273 2505 aaronmk
                +strings.as_inline_table(join_cols, ustr=col_ustr))
1274 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
1275 2230 aaronmk
        except NullValueException, e:
1276 2309 aaronmk
            log_exc(e)
1277
1278 2230 aaronmk
            out_col, = e.cols
1279
            try: in_col = mapping[out_col]
1280 2356 aaronmk
            except KeyError:
1281 2486 aaronmk
                log_debug('Missing mapping for NOT NULL column '+out_col)
1282 2451 aaronmk
                remove_all_rows()
1283 2758 aaronmk
            else: remove_rows(in_col, None, e)
1284 2542 aaronmk
        except FunctionValueException, e:
1285
            log_exc(e)
1286
1287
            func_name = e.name
1288
            value = e.value
1289
            for out_col, in_col in mapping.iteritems():
1290 2758 aaronmk
                in_col = sql_gen.unwrap_func_call(in_col, func_name)
1291
                invalid2null(in_col, value, e)
1292 2525 aaronmk
        except MissingCastException, e:
1293
            log_exc(e)
1294
1295
            out_col = e.col
1296 2706 aaronmk
            type_ = e.type
1297
1298 2724 aaronmk
            log_debug('Casting '+strings.as_tt(out_col)+' input to '
1299
                +strings.as_tt(type_))
1300 2735 aaronmk
            def wrap_func(col): return cast(db, type_, col, errors_table_)
1301 2705 aaronmk
            mapping[out_col] = sql_gen.wrap(wrap_func, mapping[out_col])
1302 2429 aaronmk
        except DatabaseErrors, e:
1303
            log_exc(e)
1304
1305 2531 aaronmk
            msg = 'No handler for exception: '+exc.str_(e)
1306 2451 aaronmk
            warnings.warn(DbWarning(msg))
1307
            log_debug(msg)
1308
            remove_all_rows()
1309 2358 aaronmk
        # after exception handled, rerun loop with additional constraints
1310 2132 aaronmk
1311 2357 aaronmk
    if row_ct_ref != None and cur.rowcount >= 0:
1312
        row_ct_ref[0] += cur.rowcount
1313
1314
    if has_joins:
1315
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
1316 2486 aaronmk
        log_debug('Getting output table pkeys of existing/inserted rows')
1317 2420 aaronmk
        insert_into_pkeys(select_joins, pkeys_cols)
1318 2357 aaronmk
    else:
1319 2404 aaronmk
        add_row_num(db, insert_out_pkeys) # for joining with input pkeys
1320 2357 aaronmk
1321 2486 aaronmk
        log_debug('Getting input table pkeys of inserted rows')
1322 2786 aaronmk
        run_query_into(db, mk_main_select(input_joins, [in_pkey]),
1323 2404 aaronmk
            into=insert_in_pkeys)
1324
        add_row_num(db, insert_in_pkeys) # for joining with output pkeys
1325 2357 aaronmk
1326 2428 aaronmk
        assert table_row_count(db, insert_out_pkeys) == table_row_count(db,
1327
            insert_in_pkeys)
1328
1329 2486 aaronmk
        log_debug('Combining output and input pkeys in inserted order')
1330 2404 aaronmk
        pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
1331 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
1332 2420 aaronmk
        insert_into_pkeys(pkey_joins, pkeys_names)
1333 2357 aaronmk
1334 2486 aaronmk
    db.log_debug('Adding pkey on pkeys table to enable fast joins', level=2.5)
1335 2594 aaronmk
    add_pkey(db, into)
1336 2407 aaronmk
1337 2508 aaronmk
    log_debug('Setting pkeys of missing rows to '+strings.as_tt(repr(default)))
1338 2489 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(into,
1339 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
1340
        # must use join_same_not_null or query will take forever
1341 2420 aaronmk
    insert_into_pkeys(missing_rows_joins,
1342 2508 aaronmk
        [in_pkey_col, sql_gen.NamedCol(out_pkey, default)])
1343 2357 aaronmk
1344 2489 aaronmk
    assert table_row_count(db, into) == table_row_count(db, in_table)
1345 2428 aaronmk
1346 2702 aaronmk
    srcs = []
1347
    if is_func: srcs = sql_gen.cols_srcs(in_cols)
1348
    return sql_gen.Col(out_pkey, into, srcs)
1349 2115 aaronmk
1350
##### Data cleanup
1351
1352 2290 aaronmk
def cleanup_table(db, table, cols):
1353 2677 aaronmk
    table = sql_gen.as_Table(table)
1354
    cols = map(sql_gen.as_Col, cols)
1355 2115 aaronmk
1356 2677 aaronmk
    expr = ('nullif(nullif(trim(both from %s), '+db.esc_value('')+'), '
1357
        +db.esc_value(r'\N')+')')
1358
    changes = [(v, sql_gen.CustomCode(expr % v.to_str(db)))
1359
        for v in cols]
1360 2115 aaronmk
1361 2677 aaronmk
    update(db, table, changes)