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 2804 aaronmk
def get_cur_query(cur, input_query=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 2804 aaronmk
    else: return '[input] '+strings.ustr(input_query)
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 2448 aaronmk
log_debug_none = lambda msg, level=2: None
149 1901 aaronmk
150 1849 aaronmk
class DbConn:
151 2190 aaronmk
    def __init__(self, db_config, serializable=True, autocommit=False,
152
        caching=True, log_debug=log_debug_none):
153 1869 aaronmk
        self.db_config = db_config
154
        self.serializable = serializable
155 2190 aaronmk
        self.autocommit = autocommit
156
        self.caching = caching
157 1901 aaronmk
        self.log_debug = log_debug
158 2193 aaronmk
        self.debug = log_debug != log_debug_none
159 1869 aaronmk
160
        self.__db = None
161 1889 aaronmk
        self.query_results = {}
162 2139 aaronmk
        self._savepoint = 0
163 2671 aaronmk
        self._notices_seen = set()
164 1869 aaronmk
165
    def __getattr__(self, name):
166
        if name == '__dict__': raise Exception('getting __dict__')
167
        if name == 'db': return self._db()
168
        else: raise AttributeError()
169
170
    def __getstate__(self):
171
        state = copy.copy(self.__dict__) # shallow copy
172 1915 aaronmk
        state['log_debug'] = None # don't pickle the debug callback
173 1869 aaronmk
        state['_DbConn__db'] = None # don't pickle the connection
174
        return state
175
176 2165 aaronmk
    def connected(self): return self.__db != None
177
178 1869 aaronmk
    def _db(self):
179
        if self.__db == None:
180
            # Process db_config
181
            db_config = self.db_config.copy() # don't modify input!
182 2097 aaronmk
            schemas = db_config.pop('schemas', None)
183 1869 aaronmk
            module_name, mappings = db_engines[db_config.pop('engine')]
184
            module = __import__(module_name)
185
            _add_module(module)
186
            for orig, new in mappings.iteritems():
187
                try: util.rename_key(db_config, orig, new)
188
                except KeyError: pass
189
190
            # Connect
191
            self.__db = module.connect(**db_config)
192
193
            # Configure connection
194 2676 aaronmk
            if self.serializable and not self.autocommit: self.run_query(
195 2678 aaronmk
                'SET TRANSACTION ISOLATION LEVEL SERIALIZABLE', log_level=4)
196 2101 aaronmk
            if schemas != None:
197
                schemas_ = ''.join((esc_name(self, s)+', '
198
                    for s in schemas.split(',')))
199 2792 aaronmk
                self.run_query("SELECT set_config('search_path', "
200
                    +self.esc_value(schemas_)
201
                    +" || current_setting('search_path'), false)", log_level=3)
202 1869 aaronmk
203
        return self.__db
204 1889 aaronmk
205 1891 aaronmk
    class DbCursor(Proxy):
206 1927 aaronmk
        def __init__(self, outer):
207 1891 aaronmk
            Proxy.__init__(self, outer.db.cursor())
208 2191 aaronmk
            self.outer = outer
209 1927 aaronmk
            self.query_results = outer.query_results
210 1894 aaronmk
            self.query_lookup = None
211 1891 aaronmk
            self.result = []
212 1889 aaronmk
213 2802 aaronmk
        def execute(self, query):
214 2764 aaronmk
            self._is_insert = query.startswith('INSERT')
215 2797 aaronmk
            self.query_lookup = query
216 2148 aaronmk
            try:
217 2191 aaronmk
                try:
218 2802 aaronmk
                    cur = self.inner.execute(query)
219 2191 aaronmk
                    self.outer.do_autocommit()
220 2802 aaronmk
                finally: self.query = get_cur_query(self.inner, query)
221 1904 aaronmk
            except Exception, e:
222 2802 aaronmk
                _add_cursor_info(e, self, query)
223 1904 aaronmk
                self.result = e # cache the exception as the result
224
                self._cache_result()
225
                raise
226 2800 aaronmk
            if self.rowcount == 0 and query.startswith('SELECT'): # empty SELECT
227
                consume_rows(self) # fetch all rows so result will be cached
228 2762 aaronmk
            return cur
229 1894 aaronmk
230 1891 aaronmk
        def fetchone(self):
231
            row = self.inner.fetchone()
232 1899 aaronmk
            if row != None: self.result.append(row)
233
            # otherwise, fetched all rows
234 1904 aaronmk
            else: self._cache_result()
235
            return row
236
237
        def _cache_result(self):
238 1906 aaronmk
            # For inserts, only cache exceptions since inserts are not
239
            # idempotent, but an invalid insert will always be invalid
240 1930 aaronmk
            if self.query_results != None and (not self._is_insert
241 1906 aaronmk
                or isinstance(self.result, Exception)):
242
243 1894 aaronmk
                assert self.query_lookup != None
244 1916 aaronmk
                self.query_results[self.query_lookup] = self.CacheCursor(
245
                    util.dict_subset(dicts.AttrsDictView(self),
246
                    ['query', 'result', 'rowcount', 'description']))
247 1906 aaronmk
248 1916 aaronmk
        class CacheCursor:
249
            def __init__(self, cached_result): self.__dict__ = cached_result
250
251 1927 aaronmk
            def execute(self, *args, **kw_args):
252 1916 aaronmk
                if isinstance(self.result, Exception): raise self.result
253
                # otherwise, result is a rows list
254
                self.iter = iter(self.result)
255
256
            def fetchone(self):
257
                try: return self.iter.next()
258
                except StopIteration: return None
259 1891 aaronmk
260 2212 aaronmk
    def esc_value(self, value):
261 2663 aaronmk
        try: str_ = self.mogrify('%s', [value])
262
        except NotImplementedError, e:
263
            module = util.root_module(self.db)
264
            if module == 'MySQLdb':
265
                import _mysql
266
                str_ = _mysql.escape_string(value)
267
            else: raise e
268 2374 aaronmk
        return strings.to_unicode(str_)
269 2212 aaronmk
270 2347 aaronmk
    def esc_name(self, name): return esc_name(self, name) # calls global func
271
272 2665 aaronmk
    def can_mogrify(self):
273 2663 aaronmk
        module = util.root_module(self.db)
274 2665 aaronmk
        return module == 'psycopg2'
275 2663 aaronmk
276 2665 aaronmk
    def mogrify(self, query, params=None):
277
        if self.can_mogrify(): return self.db.cursor().mogrify(query, params)
278
        else: raise NotImplementedError("Can't mogrify query")
279
280 2671 aaronmk
    def print_notices(self):
281 2725 aaronmk
        if hasattr(self.db, 'notices'):
282
            for msg in self.db.notices:
283
                if msg not in self._notices_seen:
284
                    self._notices_seen.add(msg)
285
                    self.log_debug(msg, level=2)
286 2671 aaronmk
287 2793 aaronmk
    def run_query(self, query, cacheable=False, log_level=2,
288 2464 aaronmk
        debug_msg_ref=None):
289 2445 aaronmk
        '''
290 2464 aaronmk
        @param log_ignore_excs The log_level will be increased by 2 if the query
291
            throws one of these exceptions.
292 2664 aaronmk
        @param debug_msg_ref If specified, the log message will be returned in
293
            this instead of being output. This allows you to filter log messages
294
            depending on the result of the query.
295 2445 aaronmk
        '''
296 2167 aaronmk
        assert query != None
297
298 2047 aaronmk
        if not self.caching: cacheable = False
299 1903 aaronmk
        used_cache = False
300 2664 aaronmk
301
        def log_msg(query):
302
            if used_cache: cache_status = 'cache hit'
303
            elif cacheable: cache_status = 'cache miss'
304
            else: cache_status = 'non-cacheable'
305
            return 'DB query: '+cache_status+':\n'+strings.as_code(query, 'SQL')
306
307 1903 aaronmk
        try:
308 1927 aaronmk
            # Get cursor
309
            if cacheable:
310
                try:
311 2797 aaronmk
                    cur = self.query_results[query]
312 1927 aaronmk
                    used_cache = True
313
                except KeyError: cur = self.DbCursor(self)
314
            else: cur = self.db.cursor()
315
316 2664 aaronmk
            # Log query
317
            if self.debug and debug_msg_ref == None: # log before running
318
                self.log_debug(log_msg(query), log_level)
319
320 1927 aaronmk
            # Run query
321 2793 aaronmk
            cur.execute(query)
322 1903 aaronmk
        finally:
323 2671 aaronmk
            self.print_notices()
324 2664 aaronmk
            if self.debug and debug_msg_ref != None: # return after running
325 2793 aaronmk
                debug_msg_ref[0] = log_msg(str(get_cur_query(cur, query)))
326 1903 aaronmk
327
        return cur
328 1914 aaronmk
329 2797 aaronmk
    def is_cached(self, query): return query in self.query_results
330 2139 aaronmk
331 2683 aaronmk
    def with_autocommit(self, func, autocommit=True):
332 2801 aaronmk
        import psycopg2.extensions
333
        if autocommit:
334
            isolation_level = psycopg2.extensions.ISOLATION_LEVEL_AUTOCOMMIT
335
        else: isolation_level = psycopg2.extensions.ISOLATION_LEVEL_SERIALIZABLE
336
337
        prev_isolation_level = self.db.isolation_level
338
        self.db.set_isolation_level(isolation_level)
339 2683 aaronmk
        try: return func()
340 2801 aaronmk
        finally: self.db.set_isolation_level(prev_isolation_level)
341 2683 aaronmk
342 2139 aaronmk
    def with_savepoint(self, func):
343 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
344 2443 aaronmk
        self.run_query('SAVEPOINT '+savepoint, log_level=4)
345 2139 aaronmk
        self._savepoint += 1
346 2683 aaronmk
        try:
347 2139 aaronmk
            try: return_val = func()
348
            finally:
349
                self._savepoint -= 1
350
                assert self._savepoint >= 0
351
        except:
352 2443 aaronmk
            self.run_query('ROLLBACK TO SAVEPOINT '+savepoint, log_level=4)
353 2139 aaronmk
            raise
354
        else:
355 2443 aaronmk
            self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
356 2191 aaronmk
            self.do_autocommit()
357 2139 aaronmk
            return return_val
358 2191 aaronmk
359
    def do_autocommit(self):
360
        '''Autocommits if outside savepoint'''
361
        assert self._savepoint >= 0
362
        if self.autocommit and self._savepoint == 0:
363 2668 aaronmk
            self.log_debug('Autocommitting')
364 2191 aaronmk
            self.db.commit()
365 2643 aaronmk
366
    def col_default(self, col):
367
        table = sql_gen.Table('columns', 'information_schema')
368
369
        conds = [('table_name', col.table.name), ('column_name', col.name)]
370
        schema = col.table.schema
371
        if schema != None: conds.append(('table_schema', schema))
372
373 2660 aaronmk
        return sql_gen.as_Code(value(select(self, table, ['column_default'],
374
            conds, order_by='table_schema', limit=1, log_level=3)))
375 2643 aaronmk
            # TODO: order_by search_path schema order
376 1849 aaronmk
377 1869 aaronmk
connect = DbConn
378
379 832 aaronmk
##### Recoverable querying
380 15 aaronmk
381 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
382 11 aaronmk
383 2791 aaronmk
def run_query(db, query, recover=None, cacheable=False, log_level=2,
384
    log_ignore_excs=None, **kw_args):
385 2794 aaronmk
    '''For params, see DbConn.run_query()'''
386 830 aaronmk
    if recover == None: recover = False
387 2464 aaronmk
    if log_ignore_excs == None: log_ignore_excs = ()
388
    log_ignore_excs = tuple(log_ignore_excs)
389 830 aaronmk
390 2666 aaronmk
    debug_msg_ref = None # usually, db.run_query() logs query before running it
391
    # But if filtering with log_ignore_excs, wait until after exception parsing
392
    if log_ignore_excs != () or not db.can_mogrify(): debug_msg_ref = [None]
393
394 2148 aaronmk
    try:
395 2464 aaronmk
        try:
396 2794 aaronmk
            def run(): return db.run_query(query, cacheable, log_level,
397 2793 aaronmk
                debug_msg_ref, **kw_args)
398 2796 aaronmk
            if recover and not db.is_cached(query):
399 2464 aaronmk
                return with_savepoint(db, run)
400
            else: return run() # don't need savepoint if cached
401
        except Exception, e:
402
            if not recover: raise # need savepoint to run index_cols()
403
            msg = exc.str_(e)
404
405
            match = re.search(r'duplicate key value violates unique constraint '
406 2493 aaronmk
                r'"((_?[^\W_]+)_.+?)"', msg)
407 2464 aaronmk
            if match:
408
                constraint, table = match.groups()
409
                try: cols = index_cols(db, table, constraint)
410
                except NotImplementedError: raise e
411
                else: raise DuplicateKeyException(constraint, cols, e)
412
413 2493 aaronmk
            match = re.search(r'null value in column "(.+?)" violates not-null'
414 2464 aaronmk
                r' constraint', msg)
415
            if match: raise NullValueException('NOT NULL', [match.group(1)], e)
416
417
            match = re.search(r'\b(?:invalid input (?:syntax|value)\b.*?'
418
                r'|date/time field value out of range): "(.+?)"\n'
419 2535 aaronmk
                r'(?:(?s).*?)\bfunction "(.+?)"', msg)
420 2464 aaronmk
            if match:
421
                value, name = match.groups()
422
                raise FunctionValueException(name, strings.to_unicode(value), e)
423
424 2526 aaronmk
            match = re.search(r'column "(.+?)" is of type (.+?) but expression '
425 2523 aaronmk
                r'is of type', msg)
426
            if match:
427
                col, type_ = match.groups()
428
                raise MissingCastException(type_, col, e)
429
430 2493 aaronmk
            match = re.search(r'relation "(.+?)" already exists', msg)
431 2464 aaronmk
            if match: raise DuplicateTableException(match.group(1), e)
432
433 2493 aaronmk
            match = re.search(r'function "(.+?)" already exists', msg)
434 2464 aaronmk
            if match: raise DuplicateFunctionException(match.group(1), e)
435
436
            raise # no specific exception raised
437
    except log_ignore_excs:
438
        log_level += 2
439
        raise
440
    finally:
441 2666 aaronmk
        if debug_msg_ref != None and debug_msg_ref[0] != None:
442
            db.log_debug(debug_msg_ref[0], log_level)
443 830 aaronmk
444 832 aaronmk
##### Basic queries
445
446 2153 aaronmk
def next_version(name):
447 2163 aaronmk
    version = 1 # first existing name was version 0
448 2586 aaronmk
    match = re.match(r'^(.*)#(\d+)$', name)
449 2153 aaronmk
    if match:
450 2586 aaronmk
        name, version = match.groups()
451
        version = int(version)+1
452 2588 aaronmk
    return sql_gen.add_suffix(name, '#'+str(version))
453 2153 aaronmk
454 2789 aaronmk
def run_query_into(db, query, into=None, add_indexes_=False, **kw_args):
455 2085 aaronmk
    '''Outputs a query to a temp table.
456
    For params, see run_query().
457
    '''
458 2789 aaronmk
    if into == None: return run_query(db, query, **kw_args)
459 2790 aaronmk
460
    assert isinstance(into, sql_gen.Table)
461
462
    kw_args['recover'] = True
463
    kw_args.setdefault('log_ignore_excs', (DuplicateTableException,))
464
465
    temp = not db.autocommit # tables are permanent in autocommit mode
466
    # "temporary tables cannot specify a schema name", so remove schema
467
    if temp: into.schema = None
468
469
    # Create table
470
    while True:
471
        create_query = 'CREATE'
472
        if temp: create_query += ' TEMP'
473
        create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
474 2385 aaronmk
475 2790 aaronmk
        try:
476
            cur = run_query(db, create_query, **kw_args)
477
                # CREATE TABLE AS sets rowcount to # rows in query
478
            break
479
        except DuplicateTableException, e:
480
            into.name = next_version(into.name)
481
            # try again with next version of name
482
483
    if add_indexes_: add_indexes(db, into)
484
485
    return cur
486 2085 aaronmk
487 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
488
489 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
490
491 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
492 2293 aaronmk
    start=None, order_by=order_by_pkey, default_table=None):
493 1981 aaronmk
    '''
494 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
495 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
496 1981 aaronmk
    @param fields Use None to select all fields in the table
497 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
498 2379 aaronmk
        * container can be any iterable type
499 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
500
        * compare_right_side: sql_gen.ValueCond|literal value
501 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
502
        use all columns
503 2786 aaronmk
    @return query
504 1981 aaronmk
    '''
505 2315 aaronmk
    # Parse tables param
506 2121 aaronmk
    if not lists.is_seq(tables): tables = [tables]
507 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
508 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
509 2121 aaronmk
510 2315 aaronmk
    # Parse other params
511 2376 aaronmk
    if conds == None: conds = []
512 2650 aaronmk
    elif dicts.is_dict(conds): conds = conds.items()
513 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
514 135 aaronmk
    assert limit == None or type(limit) == int
515 865 aaronmk
    assert start == None or type(start) == int
516 2315 aaronmk
    if order_by is order_by_pkey:
517
        if distinct_on != []: order_by = None
518
        else: order_by = pkey(db, table0, recover=True)
519 865 aaronmk
520 2315 aaronmk
    query = 'SELECT'
521 2056 aaronmk
522 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
523 2056 aaronmk
524 2200 aaronmk
    # DISTINCT ON columns
525 2233 aaronmk
    if distinct_on != []:
526 2467 aaronmk
        query += '\nDISTINCT'
527 2254 aaronmk
        if distinct_on is not distinct_on_all:
528 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
529
530
    # Columns
531 2467 aaronmk
    query += '\n'
532 1135 aaronmk
    if fields == None: query += '*'
533 2765 aaronmk
    else:
534
        assert fields != []
535
        query += '\n, '.join(map(parse_col, fields))
536 2200 aaronmk
537
    # Main table
538 2467 aaronmk
    query += '\nFROM '+table0.to_str(db)
539 865 aaronmk
540 2122 aaronmk
    # Add joins
541 2271 aaronmk
    left_table = table0
542 2263 aaronmk
    for join_ in tables:
543
        table = join_.table
544 2238 aaronmk
545 2343 aaronmk
        # Parse special values
546
        if join_.type_ is sql_gen.filter_out: # filter no match
547 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
548
                None))
549 2343 aaronmk
550 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
551 2122 aaronmk
552
        left_table = table
553
554 865 aaronmk
    missing = True
555 2376 aaronmk
    if conds != []:
556 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
557
        else: whitespace = '\n'
558 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
559
            .to_str(db) for l, r in conds], 'WHERE')
560 865 aaronmk
        missing = False
561 2227 aaronmk
    if order_by != None:
562 2467 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
563
    if limit != None: query += '\nLIMIT '+str(limit); missing = False
564 865 aaronmk
    if start != None:
565 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
566 865 aaronmk
        missing = False
567
    if missing: warnings.warn(DbWarning(
568
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
569
570 2786 aaronmk
    return query
571 11 aaronmk
572 2054 aaronmk
def select(db, *args, **kw_args):
573
    '''For params, see mk_select() and run_query()'''
574
    recover = kw_args.pop('recover', None)
575
    cacheable = kw_args.pop('cacheable', True)
576 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
577 2054 aaronmk
578 2791 aaronmk
    return run_query(db, mk_select(db, *args, **kw_args), recover, cacheable,
579
        log_level=log_level)
580 2054 aaronmk
581 2788 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, returning=None,
582
    embeddable=False):
583 1960 aaronmk
    '''
584
    @param returning str|None An inserted column (such as pkey) to return
585 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
586 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
587
        query will be fully cached, not just if it raises an exception.
588 1960 aaronmk
    '''
589 2754 aaronmk
    table = sql_gen.remove_table_rename(sql_gen.as_Table(table))
590 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
591 2646 aaronmk
    if cols != None:
592
        cols = [sql_gen.to_name_only_col(v, table).to_str(db) for v in cols]
593 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
594 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
595 2063 aaronmk
596
    # Build query
597 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
598
    query = first_line
599 2467 aaronmk
    if cols != None: query += '\n('+', '.join(cols)+')'
600
    query += '\n'+select_query
601 2063 aaronmk
602
    if returning != None:
603 2743 aaronmk
        query += '\nRETURNING '+sql_gen.to_name_only_col(returning).to_str(db)
604 2063 aaronmk
605 2070 aaronmk
    if embeddable:
606 2327 aaronmk
        assert returning != None
607
608 2070 aaronmk
        # Create function
609 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
610 2327 aaronmk
        return_type = 'SETOF '+returning.to_str(db)+'%TYPE'
611 2189 aaronmk
        while True:
612
            try:
613 2698 aaronmk
                function = sql_gen.TempFunction(function_name, db.autocommit)
614 2194 aaronmk
615 2189 aaronmk
                function_query = '''\
616 2698 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''()
617 2467 aaronmk
RETURNS '''+return_type+'''
618
LANGUAGE sql
619
AS $$
620 2788 aaronmk
'''+query+''';
621 2467 aaronmk
$$;
622 2070 aaronmk
'''
623 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
624 2464 aaronmk
                    log_ignore_excs=(DuplicateFunctionException,))
625 2189 aaronmk
                break # this version was successful
626
            except DuplicateFunctionException, e:
627
                function_name = next_version(function_name)
628
                # try again with next version of name
629 2070 aaronmk
630 2337 aaronmk
        # Return query that uses function
631 2698 aaronmk
        func_table = sql_gen.NamedTable('f', sql_gen.FunctionCall(function),
632 2742 aaronmk
            [returning]) # AS clause requires function alias
633 2787 aaronmk
        return mk_select(db, func_table, start=0, order_by=None)
634 2070 aaronmk
635 2787 aaronmk
    return query
636 2066 aaronmk
637
def insert_select(db, *args, **kw_args):
638 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
639 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
640
        values in
641 2072 aaronmk
    '''
642 2386 aaronmk
    into = kw_args.pop('into', None)
643
    if into != None: kw_args['embeddable'] = True
644 2066 aaronmk
    recover = kw_args.pop('recover', None)
645
    cacheable = kw_args.pop('cacheable', True)
646 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
647 2066 aaronmk
648 2789 aaronmk
    return run_query_into(db, mk_insert_select(db, *args, **kw_args), into,
649 2787 aaronmk
        recover=recover, cacheable=cacheable, log_level=log_level)
650 2063 aaronmk
651 2738 aaronmk
default = sql_gen.default # tells insert() to use the default value for a column
652 2066 aaronmk
653 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
654 2085 aaronmk
    '''For params, see insert_select()'''
655 1960 aaronmk
    if lists.is_seq(row): cols = None
656
    else:
657
        cols = row.keys()
658
        row = row.values()
659 2738 aaronmk
    row = list(row) # ensure that "== []" works
660 1960 aaronmk
661 2738 aaronmk
    if row == []: query = None
662
    else: query = sql_gen.Values(row).to_str(db)
663 1961 aaronmk
664 2788 aaronmk
    return insert_select(db, table, cols, query, *args, **kw_args)
665 11 aaronmk
666 2402 aaronmk
def mk_update(db, table, changes=None, cond=None):
667
    '''
668
    @param changes [(col, new_value),...]
669
        * container can be any iterable type
670
        * col: sql_gen.Code|str (for col name)
671
        * new_value: sql_gen.Code|literal value
672
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
673
    @return str query
674
    '''
675
    query = 'UPDATE '+sql_gen.as_Table(table).to_str(db)+'\nSET\n'
676 2405 aaronmk
    query += ',\n'.join((sql_gen.to_name_only_col(col, table).to_str(db)+' = '
677 2402 aaronmk
        +sql_gen.as_Value(new_value).to_str(db) for col, new_value in changes))
678 2467 aaronmk
    if cond != None: query += '\nWHERE\n'+cond.to_str(db)
679 2402 aaronmk
680
    return query
681
682
def update(db, *args, **kw_args):
683
    '''For params, see mk_update() and run_query()'''
684
    recover = kw_args.pop('recover', None)
685
686 2791 aaronmk
    return run_query(db, mk_update(db, *args, **kw_args), recover)
687 2402 aaronmk
688 135 aaronmk
def last_insert_id(db):
689 1849 aaronmk
    module = util.root_module(db.db)
690 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
691
    elif module == 'MySQLdb': return db.insert_id()
692
    else: return None
693 13 aaronmk
694 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
695 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
696 2415 aaronmk
    to names that will be distinct among the columns' tables.
697 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
698 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
699
    @param into The table for the new columns.
700 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
701
        columns will be included in the mapping even if they are not in cols.
702
        The tables of the provided Col objects will be changed to into, so make
703
        copies of them if you want to keep the original tables.
704
    @param as_items Whether to return a list of dict items instead of a dict
705 2383 aaronmk
    @return dict(orig_col=new_col, ...)
706
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
707 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
708
        * All mappings use the into table so its name can easily be
709 2383 aaronmk
          changed for all columns at once
710
    '''
711 2415 aaronmk
    cols = lists.uniqify(cols)
712
713 2394 aaronmk
    items = []
714 2389 aaronmk
    for col in preserve:
715 2390 aaronmk
        orig_col = copy.copy(col)
716 2392 aaronmk
        col.table = into
717 2394 aaronmk
        items.append((orig_col, col))
718
    preserve = set(preserve)
719
    for col in cols:
720 2716 aaronmk
        if col not in preserve:
721
            items.append((col, sql_gen.Col(str(col), into, col.srcs)))
722 2394 aaronmk
723
    if not as_items: items = dict(items)
724
    return items
725 2383 aaronmk
726 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
727 2391 aaronmk
    '''For params, see mk_flatten_mapping()
728
    @return See return value of mk_flatten_mapping()
729
    '''
730 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
731
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
732 2786 aaronmk
    run_query_into(db, mk_select(db, joins, cols, limit=limit, start=start),
733 2392 aaronmk
        into=into)
734 2394 aaronmk
    return dict(items)
735 2391 aaronmk
736 2751 aaronmk
def mk_track_data_error(db, errors_table, cols, value, error_code, error):
737 2810 aaronmk
    assert cols != ()
738
739 2751 aaronmk
    cols = map(sql_gen.to_name_only_col, cols)
740
741
    columns_cols = ['column']
742
    columns = sql_gen.NamedValues('columns', columns_cols,
743
        [[c.name] for c in cols])
744
    values_cols = ['value', 'error_code', 'error']
745
    values = sql_gen.NamedValues('values', values_cols,
746
        [value, error_code, error])
747
748
    select_cols = columns_cols+values_cols
749
    name_only_cols = map(sql_gen.to_name_only_col, select_cols)
750 2755 aaronmk
    errors_table = sql_gen.NamedTable('errors', errors_table)
751 2751 aaronmk
    joins = [columns, sql_gen.Join(values, type_='CROSS'),
752
        sql_gen.Join(errors_table, dict(zip(name_only_cols, select_cols)),
753
        sql_gen.filter_out)]
754
755
    return mk_insert_select(db, errors_table, name_only_cols,
756 2787 aaronmk
        mk_select(db, joins, select_cols, order_by=None))
757 2751 aaronmk
758 2810 aaronmk
def track_data_error(db, errors_table, cols, *args, **kw_args):
759 2758 aaronmk
    '''
760
    @param errors_table If None, does nothing.
761
    '''
762 2810 aaronmk
    if errors_table == None or cols == (): return
763
    run_query(db, mk_track_data_error(db, errors_table, cols, *args, **kw_args),
764 2758 aaronmk
        cacheable=True, log_level=4)
765
766 2704 aaronmk
def cast(db, type_, col, errors_table=None):
767
    '''Casts an (unrenamed) column or value.
768 2769 aaronmk
    If errors_table set and col has srcs, saves errors in errors_table (using
769
    col's srcs attr as the source columns) and converts errors to warnings.
770 2704 aaronmk
    @param col sql_gen.Col|sql_gen.Literal
771
    @param errors_table None|sql_gen.Table|str
772
    '''
773 2769 aaronmk
    save_errors = (errors_table != None and isinstance(col, sql_gen.Col)
774
        and col.srcs != ())
775
    if not save_errors: # can't save errors
776
        return sql_gen.CustomCode(col.to_str(db)+'::'+type_) # just cast
777 2704 aaronmk
778
    assert not isinstance(col, sql_gen.NamedCol)
779
780 2770 aaronmk
    errors_table = sql_gen.as_Table(errors_table)
781
    srcs = map(sql_gen.to_name_only_col, col.srcs)
782
    function_name = str(sql_gen.FunctionCall(type_, *srcs))
783 2704 aaronmk
    function = sql_gen.TempFunction(function_name, db.autocommit)
784
785 2718 aaronmk
    while True:
786
        # Create function definition
787
        query = '''\
788 2704 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''(value text)
789
RETURNS '''+type_+'''
790
LANGUAGE plpgsql
791
STRICT
792
AS $$
793
BEGIN
794
    /* The explicit cast to the return type is needed to make the cast happen
795
    inside the try block. (Implicit casts to the return type happen at the end
796
    of the function, outside any block.) */
797
    RETURN value::'''+type_+''';
798
EXCEPTION
799
    WHEN data_exception THEN
800
        -- Save error in errors table.
801 2751 aaronmk
        -- Insert the value and error for *each* source column.
802
'''+mk_track_data_error(db, errors_table, srcs,
803
    *map(sql_gen.CustomCode, ['value', 'SQLSTATE', 'SQLERRM']))+''';
804 2704 aaronmk
805
        RAISE WARNING '%', SQLERRM;
806
        RETURN NULL;
807
END;
808
$$;
809
'''
810 2718 aaronmk
811
        # Create function
812
        try:
813
            run_query(db, query, recover=True, cacheable=True,
814
                log_ignore_excs=(DuplicateFunctionException,))
815
            break # successful
816
        except DuplicateFunctionException:
817 2770 aaronmk
            function.name = next_version(function.name)
818
            # try again with next version of name
819 2704 aaronmk
820
    return sql_gen.FunctionCall(function, col)
821
822 2414 aaronmk
##### Database structure queries
823
824 2426 aaronmk
def table_row_count(db, table, recover=None):
825 2786 aaronmk
    return value(run_query(db, mk_select(db, table, [sql_gen.row_count],
826 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
827 2426 aaronmk
828 2414 aaronmk
def table_cols(db, table, recover=None):
829
    return list(col_names(select(db, table, limit=0, order_by=None,
830 2443 aaronmk
        recover=recover, log_level=4)))
831 2414 aaronmk
832 2291 aaronmk
def pkey(db, table, recover=None):
833 832 aaronmk
    '''Assumed to be first column in table'''
834 2339 aaronmk
    return table_cols(db, table, recover)[0]
835 832 aaronmk
836 2559 aaronmk
not_null_col = 'not_null_col'
837 2340 aaronmk
838
def table_not_null_col(db, table, recover=None):
839
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
840
    if not_null_col in table_cols(db, table, recover): return not_null_col
841
    else: return pkey(db, table, recover)
842
843 853 aaronmk
def index_cols(db, table, index):
844
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
845
    automatically created. When you don't know whether something is a UNIQUE
846
    constraint or a UNIQUE index, use this function.'''
847 1909 aaronmk
    module = util.root_module(db.db)
848
    if module == 'psycopg2':
849
        return list(values(run_query(db, '''\
850 853 aaronmk
SELECT attname
851 866 aaronmk
FROM
852
(
853
        SELECT attnum, attname
854
        FROM pg_index
855
        JOIN pg_class index ON index.oid = indexrelid
856
        JOIN pg_class table_ ON table_.oid = indrelid
857
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
858
        WHERE
859 2782 aaronmk
            table_.relname = '''+db.esc_value(table)+'''
860
            AND index.relname = '''+db.esc_value(index)+'''
861 866 aaronmk
    UNION
862
        SELECT attnum, attname
863
        FROM
864
        (
865
            SELECT
866
                indrelid
867
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
868
                    AS indkey
869
            FROM pg_index
870
            JOIN pg_class index ON index.oid = indexrelid
871
            JOIN pg_class table_ ON table_.oid = indrelid
872
            WHERE
873 2782 aaronmk
                table_.relname = '''+db.esc_value(table)+'''
874
                AND index.relname = '''+db.esc_value(index)+'''
875 866 aaronmk
        ) s
876
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
877
) s
878 853 aaronmk
ORDER BY attnum
879 2782 aaronmk
'''
880
            , cacheable=True, log_level=4)))
881 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
882
        ' database')
883 853 aaronmk
884 464 aaronmk
def constraint_cols(db, table, constraint):
885 1849 aaronmk
    module = util.root_module(db.db)
886 464 aaronmk
    if module == 'psycopg2':
887
        return list(values(run_query(db, '''\
888
SELECT attname
889
FROM pg_constraint
890
JOIN pg_class ON pg_class.oid = conrelid
891
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
892
WHERE
893 2783 aaronmk
    relname = '''+db.esc_value(table)+'''
894
    AND conname = '''+db.esc_value(constraint)+'''
895 464 aaronmk
ORDER BY attnum
896 2783 aaronmk
'''
897
            )))
898 464 aaronmk
    else: raise NotImplementedError("Can't list constraint columns for "+module+
899
        ' database')
900
901 2096 aaronmk
row_num_col = '_row_num'
902
903 2688 aaronmk
def add_index(db, exprs, table=None, unique=False):
904
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
905 2538 aaronmk
    Currently, only function calls are supported as expressions.
906
    '''
907 2688 aaronmk
    if not lists.is_seq(exprs): exprs = [exprs]
908 2538 aaronmk
909 2688 aaronmk
    # Parse exprs
910
    old_exprs = exprs[:]
911
    exprs = []
912
    cols = []
913
    for i, expr in enumerate(old_exprs):
914
        expr = copy.deepcopy(expr) # don't modify input!
915
        expr = sql_gen.as_Col(expr)
916
917
        # Extract col
918
        if isinstance(expr, sql_gen.FunctionCall):
919
            col = expr.args[0]
920
            expr = sql_gen.Expr(expr)
921
        else: col = expr
922
923
        # Extract table
924
        if table == None:
925
            assert sql_gen.is_table_col(col)
926
            table = col.table
927
928
        col.table = None
929
930
        exprs.append(expr)
931
        cols.append(col)
932 2408 aaronmk
933 2688 aaronmk
    table = sql_gen.as_Table(table)
934
    index = sql_gen.Table(str(sql_gen.Col(','.join(map(str, cols)), table)))
935
936
    str_ = 'CREATE'
937
    if unique: str_ += ' UNIQUE'
938
    str_ += ' INDEX '+index.to_str(db)+' ON '+table.to_str(db)+' ('+(
939
        ', '.join((v.to_str(db) for v in exprs)))+')'
940
941
    try: run_query(db, str_, recover=True, cacheable=True, log_level=3)
942 2408 aaronmk
    except DuplicateTableException: pass # index already existed
943
944 2686 aaronmk
def add_pkey(db, table, cols=None, recover=None):
945
    '''Adds a primary key.
946
    @param cols [sql_gen.Col,...] The columns in the primary key.
947
        Defaults to the first column in the table.
948 2406 aaronmk
    @pre The table must not already have a primary key.
949
    '''
950
    table = sql_gen.as_Table(table)
951 2686 aaronmk
    if cols == None: cols = [pkey(db, table, recover)]
952
    col_strs = [sql_gen.to_name_only_col(v).to_str(db) for v in cols]
953 2406 aaronmk
954 2781 aaronmk
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD PRIMARY KEY ('
955
        +(', '.join(col_strs))+')', recover=True, cacheable=True, log_level=3,
956
        log_ignore_excs=(DuplicateTableException,))
957 2406 aaronmk
958 2772 aaronmk
already_indexed = object() # tells add_indexes() the pkey has already been added
959
960
def add_indexes(db, table, has_pkey=True):
961
    '''Adds an index on all columns in a table.
962
    @param has_pkey bool|already_indexed Whether a pkey instead of a regular
963
        index should be added on the first column.
964
        * If already_indexed, the pkey is assumed to have already been added
965
    '''
966
    cols = table_cols(db, table)
967
    if has_pkey:
968
        if has_pkey is not already_indexed: add_pkey(db, table)
969
        cols = cols[1:]
970
    for col in cols: add_index(db, col, table)
971
972 2086 aaronmk
def add_row_num(db, table):
973 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
974
    be the primary key.'''
975 2320 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
976 2096 aaronmk
    run_query(db, 'ALTER TABLE '+table+' ADD COLUMN '+row_num_col
977 2443 aaronmk
        +' serial NOT NULL PRIMARY KEY', log_level=3)
978 2086 aaronmk
979 2760 aaronmk
def create_table(db, table, cols, has_pkey=True, col_indexes=True):
980 2675 aaronmk
    '''Creates a table.
981 2681 aaronmk
    @param cols [sql_gen.TypedCol,...] The column names and types
982
    @param has_pkey If set, the first column becomes the primary key.
983 2760 aaronmk
    @param col_indexes bool|[ref]
984
        * If True, indexes will be added on all non-pkey columns.
985
        * If a list reference, [0] will be set to a function to do this.
986
          This can be used to delay index creation until the table is populated.
987 2675 aaronmk
    '''
988
    table = sql_gen.as_Table(table)
989
990 2681 aaronmk
    if has_pkey:
991
        cols[0] = pkey = copy.copy(cols[0]) # don't modify input!
992
        pkey.type += ' NOT NULL PRIMARY KEY'
993
994 2675 aaronmk
    str_ = 'CREATE TABLE '+table.to_str(db)+' (\n'
995 2681 aaronmk
    str_ += '\n, '.join(v.to_str(db) for v in cols)
996
    str_ += '\n);\n'
997 2675 aaronmk
    run_query(db, str_, cacheable=True, log_level=2)
998 2760 aaronmk
999
    # Add indexes
1000 2773 aaronmk
    if has_pkey: has_pkey = already_indexed
1001
    def add_indexes_(): add_indexes(db, table, has_pkey)
1002
    if isinstance(col_indexes, list): col_indexes[0] = add_indexes_ # defer
1003
    elif col_indexes: add_indexes_() # add now
1004 2675 aaronmk
1005 2684 aaronmk
def vacuum(db, table):
1006
    table = sql_gen.as_Table(table)
1007
    db.with_autocommit(lambda: run_query(db, 'VACUUM ANALYZE '+table.to_str(db),
1008
        log_level=3))
1009
1010 2732 aaronmk
def truncate(db, table, schema='public'):
1011 2777 aaronmk
    table = sql_gen.as_Table(table, schema)
1012 2733 aaronmk
    return run_query(db, 'TRUNCATE '+table.to_str(db)+' CASCADE')
1013 2732 aaronmk
1014 2729 aaronmk
def tables(db, schema_like='public', table_like='%', exact=False):
1015
    if exact: compare = '='
1016
    else: compare = 'LIKE'
1017
1018 1849 aaronmk
    module = util.root_module(db.db)
1019 832 aaronmk
    if module == 'psycopg2':
1020 2779 aaronmk
        conds = [('schemaname', sql_gen.CompareCond(schema_like, compare)),
1021
            ('tablename', sql_gen.CompareCond(table_like, compare))]
1022
        return values(select(db, 'pg_tables', ['tablename'], conds,
1023
            order_by='tablename', log_level=4))
1024 1968 aaronmk
    elif module == 'MySQLdb':
1025 2779 aaronmk
        return values(run_query(db, 'SHOW TABLES LIKE '+db.esc_value(table_like)
1026
            , cacheable=True, log_level=4))
1027 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
1028 830 aaronmk
1029 2726 aaronmk
def table_exists(db, table):
1030
    table = sql_gen.as_Table(table)
1031 2730 aaronmk
    return list(tables(db, table.schema, table.name, exact=True)) != []
1032 2726 aaronmk
1033
def errors_table(db, table, if_exists=True):
1034
    '''
1035
    @param if_exists If set, returns None if the errors table doesn't exist
1036
    @return None|sql_gen.Table
1037
    '''
1038
    table = sql_gen.as_Table(table)
1039
    if table.srcs != (): table = table.srcs[0]
1040
1041
    errors_table = sql_gen.suffixed_table(table, '.errors')
1042
    if if_exists and not table_exists(db, errors_table): return None
1043
    return errors_table
1044
1045 833 aaronmk
##### Database management
1046
1047 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
1048
    '''For kw_args, see tables()'''
1049
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
1050 833 aaronmk
1051 832 aaronmk
##### Heuristic queries
1052
1053 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
1054 1554 aaronmk
    '''Recovers from errors.
1055 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
1056
    '''
1057 2642 aaronmk
    row = sql_gen.ColDict(db, table, row)
1058 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
1059
1060 471 aaronmk
    try:
1061 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
1062 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
1063
            row_ct_ref[0] += cur.rowcount
1064
        return value(cur)
1065 471 aaronmk
    except DuplicateKeyException, e:
1066 2653 aaronmk
        row = sql_gen.ColDict(db, table,
1067
            util.dict_subset_right_join(row, e.cols))
1068
        return value(select(db, table, [pkey_], row, recover=True))
1069 471 aaronmk
1070 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
1071 830 aaronmk
    '''Recovers from errors'''
1072 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
1073 14 aaronmk
    except StopIteration:
1074 40 aaronmk
        if not create: raise
1075 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
1076 2078 aaronmk
1077 2593 aaronmk
def is_func_result(col):
1078
    return col.table.name.find('(') >= 0 and col.name == 'result'
1079
1080 2592 aaronmk
def into_table_name(out_table, in_tables0, mapping, is_func):
1081 2615 aaronmk
    def in_col_str(in_col):
1082
        in_col = sql_gen.remove_col_rename(in_col)
1083
        if isinstance(in_col, sql_gen.Col):
1084
            table = in_col.table
1085
            if table == in_tables0:
1086
                in_col = sql_gen.to_name_only_col(in_col)
1087
            elif is_func_result(in_col): in_col = table # omit col name
1088
        return str(in_col)
1089
1090 2580 aaronmk
    str_ = str(out_table)
1091
    if is_func:
1092 2615 aaronmk
        str_ += '('
1093 2580 aaronmk
1094 2615 aaronmk
        try: value_in_col = mapping['value']
1095
        except KeyError:
1096 2654 aaronmk
            str_ += ', '.join((str(k)+'='+in_col_str(v)
1097 2615 aaronmk
                for k, v in mapping.iteritems()))
1098
        else: str_ += in_col_str(value_in_col)
1099
1100
        str_ += ')'
1101 2580 aaronmk
    else: str_ += '_pkeys'
1102
    return str_
1103
1104 2508 aaronmk
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None, into=None,
1105 2805 aaronmk
    default=None, is_func=False, on_error=exc.raise_):
1106 2078 aaronmk
    '''Recovers from errors.
1107
    Only works under PostgreSQL (uses INSERT RETURNING).
1108 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
1109
        tables to join with it using the main input table's pkey
1110 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
1111 2612 aaronmk
        * out_table_col: str (*not* sql_gen.Col)
1112
        * in_table_col: sql_gen.Col|literal-value
1113 2489 aaronmk
    @param into The table to contain the output and input pkeys.
1114 2574 aaronmk
        Defaults to `out_table.name+'_pkeys'`.
1115 2509 aaronmk
    @param default The *output* column to use as the pkey for missing rows.
1116
        If this output column does not exist in the mapping, uses None.
1117 2552 aaronmk
    @param is_func Whether out_table is the name of a SQL function, not a table
1118 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
1119 2078 aaronmk
    '''
1120 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
1121 2312 aaronmk
1122 2450 aaronmk
    def log_debug(msg): db.log_debug(msg, level=1.5)
1123 2505 aaronmk
    def col_ustr(str_):
1124 2567 aaronmk
        return strings.repr_no_u(sql_gen.remove_col_rename(str_))
1125 2450 aaronmk
1126 2768 aaronmk
    out_pkey = pkey(db, out_table, recover=True)
1127
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
1128
1129
    if mapping == {}: # need at least one column for INSERT SELECT
1130
        mapping = {out_pkey: None} # ColDict will replace with default value
1131
1132 2486 aaronmk
    log_debug('********** New iteration **********')
1133 2505 aaronmk
    log_debug('Inserting these input columns into '+strings.as_tt(
1134
        out_table.to_str(db))+':\n'+strings.as_table(mapping, ustr=col_ustr))
1135 2463 aaronmk
1136 2382 aaronmk
    # Create input joins from list of input tables
1137
    in_tables_ = in_tables[:] # don't modify input!
1138
    in_tables0 = in_tables_.pop(0) # first table is separate
1139 2735 aaronmk
    errors_table_ = errors_table(db, in_tables0)
1140 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
1141 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
1142 2460 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v,
1143
        {in_pkey: sql_gen.join_same_not_null}) for v in in_tables_]
1144 2131 aaronmk
1145 2592 aaronmk
    if into == None:
1146
        into = into_table_name(out_table, in_tables0, mapping, is_func)
1147
    into = sql_gen.as_Table(into)
1148
1149 2702 aaronmk
    # Set column sources
1150
    in_cols = filter(sql_gen.is_table_col, mapping.values())
1151
    for col in in_cols:
1152
        if col.table == in_tables0: col.set_srcs(sql_gen.src_self)
1153
1154 2486 aaronmk
    log_debug('Joining together input tables into temp table')
1155 2395 aaronmk
    # Place in new table for speed and so don't modify input if values edited
1156 2584 aaronmk
    in_table = sql_gen.Table('in')
1157 2702 aaronmk
    mapping = dicts.join(mapping, flatten(db, in_table, input_joins, in_cols,
1158
        preserve=[in_pkey_col], start=0))
1159 2395 aaronmk
    input_joins = [in_table]
1160 2486 aaronmk
    db.log_debug('Temp table: '+strings.as_tt(in_table.to_str(db)), level=2)
1161 2395 aaronmk
1162 2656 aaronmk
    mapping = sql_gen.ColDict(db, out_table, mapping)
1163
        # after applying dicts.join() because that returns a plain dict
1164
1165 2509 aaronmk
    # Resolve default value column
1166
    try: default = mapping[default]
1167
    except KeyError:
1168
        if default != None:
1169
            db.log_debug('Default value column '
1170
                +strings.as_tt(strings.repr_no_u(default))
1171 2511 aaronmk
                +' does not exist in mapping, falling back to None', level=2.1)
1172 2509 aaronmk
            default = None
1173
1174 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
1175 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
1176
1177 2201 aaronmk
    pkeys_table_exists_ref = [False]
1178 2420 aaronmk
    def insert_into_pkeys(joins, cols):
1179 2786 aaronmk
        query = mk_select(db, joins, cols, order_by=None, start=0)
1180 2201 aaronmk
        if pkeys_table_exists_ref[0]:
1181 2786 aaronmk
            insert_select(db, into, pkeys_names, query)
1182 2201 aaronmk
        else:
1183 2786 aaronmk
            run_query_into(db, query, into=into)
1184 2201 aaronmk
            pkeys_table_exists_ref[0] = True
1185
1186 2429 aaronmk
    limit_ref = [None]
1187 2380 aaronmk
    conds = set()
1188 2811 aaronmk
    distinct_on = sql_gen.ColDict(db, out_table)
1189 2325 aaronmk
    def mk_main_select(joins, cols):
1190 2811 aaronmk
        distinct_on_cols = [c.to_Col() for c in distinct_on.values()]
1191
        return mk_select(db, joins, cols, conds, distinct_on_cols,
1192 2429 aaronmk
            limit=limit_ref[0], start=0)
1193 2132 aaronmk
1194 2519 aaronmk
    exc_strs = set()
1195 2309 aaronmk
    def log_exc(e):
1196 2519 aaronmk
        e_str = exc.str_(e, first_line_only=True)
1197
        log_debug('Caught exception: '+e_str)
1198
        assert e_str not in exc_strs # avoid infinite loops
1199
        exc_strs.add(e_str)
1200 2735 aaronmk
1201 2451 aaronmk
    def remove_all_rows():
1202 2450 aaronmk
        log_debug('Returning NULL for all rows')
1203 2429 aaronmk
        limit_ref[0] = 0 # just create an empty pkeys table
1204 2735 aaronmk
1205 2758 aaronmk
    def ignore(in_col, value, e):
1206
        track_data_error(db, errors_table_, in_col.srcs, value, e.cause.pgcode,
1207
            e.cause.pgerror)
1208
1209 2545 aaronmk
        in_col_str = strings.as_tt(repr(in_col))
1210 2544 aaronmk
        db.log_debug('Adding index on '+in_col_str+' to enable fast filtering',
1211
            level=2.5)
1212 2537 aaronmk
        add_index(db, in_col)
1213 2735 aaronmk
1214 2545 aaronmk
        log_debug('Ignoring rows with '+in_col_str+' = '
1215
            +strings.as_tt(repr(value)))
1216 2758 aaronmk
    def remove_rows(in_col, value, e):
1217
        ignore(in_col, value, e)
1218 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
1219
        assert cond not in conds # avoid infinite loops
1220 2380 aaronmk
        conds.add(cond)
1221 2758 aaronmk
    def invalid2null(in_col, value, e):
1222
        ignore(in_col, value, e)
1223 2403 aaronmk
        update(db, in_table, [(in_col, None)],
1224
            sql_gen.ColValueCond(in_col, value))
1225 2245 aaronmk
1226 2589 aaronmk
    def insert_pkeys_table(which):
1227
        return sql_gen.Table(sql_gen.add_suffix(in_table.name,
1228
            '_insert_'+which+'_pkeys'))
1229
    insert_out_pkeys = insert_pkeys_table('out')
1230
    insert_in_pkeys = insert_pkeys_table('in')
1231
1232 2206 aaronmk
    # Do inserts and selects
1233 2642 aaronmk
    join_cols = sql_gen.ColDict(db, out_table)
1234 2206 aaronmk
    while True:
1235 2521 aaronmk
        if limit_ref[0] == 0: # special case
1236
            log_debug('Creating an empty pkeys table')
1237 2786 aaronmk
            cur = run_query_into(db, mk_select(db, out_table, [out_pkey],
1238 2521 aaronmk
                limit=limit_ref[0]), into=insert_out_pkeys)
1239
            break # don't do main case
1240
1241 2303 aaronmk
        has_joins = join_cols != {}
1242
1243 2305 aaronmk
        # Prepare to insert new rows
1244 2325 aaronmk
        insert_joins = input_joins[:] # don't modify original!
1245 2403 aaronmk
        insert_args = dict(recover=True, cacheable=False)
1246 2303 aaronmk
        if has_joins:
1247 2325 aaronmk
            insert_joins.append(sql_gen.Join(out_table, join_cols,
1248
                sql_gen.filter_out))
1249
        else:
1250 2404 aaronmk
            insert_args.update(dict(returning=out_pkey, into=insert_out_pkeys))
1251 2786 aaronmk
        main_select = mk_main_select(insert_joins, mapping.values())
1252 2303 aaronmk
1253 2486 aaronmk
        log_debug('Trying to insert new rows')
1254 2206 aaronmk
        try:
1255 2518 aaronmk
            cur = insert_select(db, out_table, mapping.keys(), main_select,
1256
                **insert_args)
1257 2357 aaronmk
            break # insert successful
1258 2206 aaronmk
        except DuplicateKeyException, e:
1259 2309 aaronmk
            log_exc(e)
1260
1261 2258 aaronmk
            old_join_cols = join_cols.copy()
1262 2811 aaronmk
            distinct_on.update(util.dict_subset(mapping, e.cols))
1263 2565 aaronmk
            join_cols.update(util.dict_subset_right_join(mapping, e.cols))
1264 2486 aaronmk
            log_debug('Ignoring existing rows, comparing on these columns:\n'
1265 2505 aaronmk
                +strings.as_inline_table(join_cols, ustr=col_ustr))
1266 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
1267 2230 aaronmk
        except NullValueException, e:
1268 2309 aaronmk
            log_exc(e)
1269
1270 2230 aaronmk
            out_col, = e.cols
1271
            try: in_col = mapping[out_col]
1272 2356 aaronmk
            except KeyError:
1273 2486 aaronmk
                log_debug('Missing mapping for NOT NULL column '+out_col)
1274 2451 aaronmk
                remove_all_rows()
1275 2758 aaronmk
            else: remove_rows(in_col, None, e)
1276 2542 aaronmk
        except FunctionValueException, e:
1277
            log_exc(e)
1278
1279
            func_name = e.name
1280
            value = e.value
1281
            for out_col, in_col in mapping.iteritems():
1282 2758 aaronmk
                in_col = sql_gen.unwrap_func_call(in_col, func_name)
1283
                invalid2null(in_col, value, e)
1284 2525 aaronmk
        except MissingCastException, e:
1285
            log_exc(e)
1286
1287
            out_col = e.col
1288 2706 aaronmk
            type_ = e.type
1289
1290 2724 aaronmk
            log_debug('Casting '+strings.as_tt(out_col)+' input to '
1291
                +strings.as_tt(type_))
1292 2735 aaronmk
            def wrap_func(col): return cast(db, type_, col, errors_table_)
1293 2705 aaronmk
            mapping[out_col] = sql_gen.wrap(wrap_func, mapping[out_col])
1294 2429 aaronmk
        except DatabaseErrors, e:
1295
            log_exc(e)
1296
1297 2808 aaronmk
            log_debug('No handler for exception')
1298
            on_error(e)
1299 2451 aaronmk
            remove_all_rows()
1300 2358 aaronmk
        # after exception handled, rerun loop with additional constraints
1301 2132 aaronmk
1302 2357 aaronmk
    if row_ct_ref != None and cur.rowcount >= 0:
1303
        row_ct_ref[0] += cur.rowcount
1304
1305
    if has_joins:
1306
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
1307 2486 aaronmk
        log_debug('Getting output table pkeys of existing/inserted rows')
1308 2420 aaronmk
        insert_into_pkeys(select_joins, pkeys_cols)
1309 2357 aaronmk
    else:
1310 2404 aaronmk
        add_row_num(db, insert_out_pkeys) # for joining with input pkeys
1311 2357 aaronmk
1312 2486 aaronmk
        log_debug('Getting input table pkeys of inserted rows')
1313 2786 aaronmk
        run_query_into(db, mk_main_select(input_joins, [in_pkey]),
1314 2404 aaronmk
            into=insert_in_pkeys)
1315
        add_row_num(db, insert_in_pkeys) # for joining with output pkeys
1316 2357 aaronmk
1317 2428 aaronmk
        assert table_row_count(db, insert_out_pkeys) == table_row_count(db,
1318
            insert_in_pkeys)
1319
1320 2486 aaronmk
        log_debug('Combining output and input pkeys in inserted order')
1321 2404 aaronmk
        pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
1322 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
1323 2420 aaronmk
        insert_into_pkeys(pkey_joins, pkeys_names)
1324 2357 aaronmk
1325 2486 aaronmk
    db.log_debug('Adding pkey on pkeys table to enable fast joins', level=2.5)
1326 2594 aaronmk
    add_pkey(db, into)
1327 2407 aaronmk
1328 2508 aaronmk
    log_debug('Setting pkeys of missing rows to '+strings.as_tt(repr(default)))
1329 2489 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(into,
1330 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
1331
        # must use join_same_not_null or query will take forever
1332 2420 aaronmk
    insert_into_pkeys(missing_rows_joins,
1333 2508 aaronmk
        [in_pkey_col, sql_gen.NamedCol(out_pkey, default)])
1334 2357 aaronmk
1335 2489 aaronmk
    assert table_row_count(db, into) == table_row_count(db, in_table)
1336 2428 aaronmk
1337 2702 aaronmk
    srcs = []
1338
    if is_func: srcs = sql_gen.cols_srcs(in_cols)
1339
    return sql_gen.Col(out_pkey, into, srcs)
1340 2115 aaronmk
1341
##### Data cleanup
1342
1343 2290 aaronmk
def cleanup_table(db, table, cols):
1344 2677 aaronmk
    table = sql_gen.as_Table(table)
1345
    cols = map(sql_gen.as_Col, cols)
1346 2115 aaronmk
1347 2677 aaronmk
    expr = ('nullif(nullif(trim(both from %s), '+db.esc_value('')+'), '
1348
        +db.esc_value(r'\N')+')')
1349
    changes = [(v, sql_gen.CustomCode(expr % v.to_str(db)))
1350
        for v in cols]
1351 2115 aaronmk
1352 2677 aaronmk
    update(db, table, changes)