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 2797 aaronmk
            self.query_lookup = query
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
                try:
317 2797 aaronmk
                    cur = self.query_results[query]
318 1927 aaronmk
                    used_cache = True
319
                except KeyError: cur = self.DbCursor(self)
320
            else: cur = self.db.cursor()
321
322 2664 aaronmk
            # Log query
323
            if self.debug and debug_msg_ref == None: # log before running
324
                self.log_debug(log_msg(query), log_level)
325
326 1927 aaronmk
            # Run query
327 2793 aaronmk
            cur.execute(query)
328 1903 aaronmk
        finally:
329 2671 aaronmk
            self.print_notices()
330 2664 aaronmk
            if self.debug and debug_msg_ref != None: # return after running
331 2793 aaronmk
                debug_msg_ref[0] = log_msg(str(get_cur_query(cur, query)))
332 1903 aaronmk
333
        return cur
334 1914 aaronmk
335 2797 aaronmk
    def is_cached(self, query): return query in self.query_results
336 2139 aaronmk
337 2683 aaronmk
    def with_autocommit(self, func, autocommit=True):
338
        prev_autocommit = self.db.autocommit
339
        self.db.autocommit = autocommit
340
        try: return func()
341
        finally: self.db.autocommit = prev_autocommit
342
343 2139 aaronmk
    def with_savepoint(self, func):
344 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
345 2443 aaronmk
        self.run_query('SAVEPOINT '+savepoint, log_level=4)
346 2139 aaronmk
        self._savepoint += 1
347 2683 aaronmk
        try:
348 2139 aaronmk
            try: return_val = func()
349
            finally:
350
                self._savepoint -= 1
351
                assert self._savepoint >= 0
352
        except:
353 2443 aaronmk
            self.run_query('ROLLBACK TO SAVEPOINT '+savepoint, log_level=4)
354 2139 aaronmk
            raise
355
        else:
356 2443 aaronmk
            self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
357 2191 aaronmk
            self.do_autocommit()
358 2139 aaronmk
            return return_val
359 2191 aaronmk
360
    def do_autocommit(self):
361
        '''Autocommits if outside savepoint'''
362
        assert self._savepoint >= 0
363
        if self.autocommit and self._savepoint == 0:
364 2668 aaronmk
            self.log_debug('Autocommitting')
365 2191 aaronmk
            self.db.commit()
366 2643 aaronmk
367
    def col_default(self, col):
368
        table = sql_gen.Table('columns', 'information_schema')
369
370
        conds = [('table_name', col.table.name), ('column_name', col.name)]
371
        schema = col.table.schema
372
        if schema != None: conds.append(('table_schema', schema))
373
374 2660 aaronmk
        return sql_gen.as_Code(value(select(self, table, ['column_default'],
375
            conds, order_by='table_schema', limit=1, log_level=3)))
376 2643 aaronmk
            # TODO: order_by search_path schema order
377 1849 aaronmk
378 1869 aaronmk
connect = DbConn
379
380 832 aaronmk
##### Querying
381
382 2068 aaronmk
def mogrify(db, query, params):
383
    module = util.root_module(db.db)
384
    if module == 'psycopg2': return db.db.cursor().mogrify(query, params)
385
    else: raise NotImplementedError("Can't mogrify query for "+module+
386
        ' database')
387
388 832 aaronmk
##### Recoverable querying
389 15 aaronmk
390 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
391 11 aaronmk
392 2791 aaronmk
def run_query(db, query, recover=None, cacheable=False, log_level=2,
393
    log_ignore_excs=None, **kw_args):
394 2794 aaronmk
    '''For params, see DbConn.run_query()'''
395 830 aaronmk
    if recover == None: recover = False
396 2464 aaronmk
    if log_ignore_excs == None: log_ignore_excs = ()
397
    log_ignore_excs = tuple(log_ignore_excs)
398 830 aaronmk
399 2666 aaronmk
    debug_msg_ref = None # usually, db.run_query() logs query before running it
400
    # But if filtering with log_ignore_excs, wait until after exception parsing
401
    if log_ignore_excs != () or not db.can_mogrify(): debug_msg_ref = [None]
402
403 2148 aaronmk
    try:
404 2464 aaronmk
        try:
405 2794 aaronmk
            def run(): return db.run_query(query, cacheable, log_level,
406 2793 aaronmk
                debug_msg_ref, **kw_args)
407 2796 aaronmk
            if recover and not db.is_cached(query):
408 2464 aaronmk
                return with_savepoint(db, run)
409
            else: return run() # don't need savepoint if cached
410
        except Exception, e:
411
            if not recover: raise # need savepoint to run index_cols()
412
            msg = exc.str_(e)
413
414
            match = re.search(r'duplicate key value violates unique constraint '
415 2493 aaronmk
                r'"((_?[^\W_]+)_.+?)"', msg)
416 2464 aaronmk
            if match:
417
                constraint, table = match.groups()
418
                try: cols = index_cols(db, table, constraint)
419
                except NotImplementedError: raise e
420
                else: raise DuplicateKeyException(constraint, cols, e)
421
422 2493 aaronmk
            match = re.search(r'null value in column "(.+?)" violates not-null'
423 2464 aaronmk
                r' constraint', msg)
424
            if match: raise NullValueException('NOT NULL', [match.group(1)], e)
425
426
            match = re.search(r'\b(?:invalid input (?:syntax|value)\b.*?'
427
                r'|date/time field value out of range): "(.+?)"\n'
428 2535 aaronmk
                r'(?:(?s).*?)\bfunction "(.+?)"', msg)
429 2464 aaronmk
            if match:
430
                value, name = match.groups()
431
                raise FunctionValueException(name, strings.to_unicode(value), e)
432
433 2526 aaronmk
            match = re.search(r'column "(.+?)" is of type (.+?) but expression '
434 2523 aaronmk
                r'is of type', msg)
435
            if match:
436
                col, type_ = match.groups()
437
                raise MissingCastException(type_, col, e)
438
439 2493 aaronmk
            match = re.search(r'relation "(.+?)" already exists', msg)
440 2464 aaronmk
            if match: raise DuplicateTableException(match.group(1), e)
441
442 2493 aaronmk
            match = re.search(r'function "(.+?)" already exists', msg)
443 2464 aaronmk
            if match: raise DuplicateFunctionException(match.group(1), e)
444
445
            raise # no specific exception raised
446
    except log_ignore_excs:
447
        log_level += 2
448
        raise
449
    finally:
450 2666 aaronmk
        if debug_msg_ref != None and debug_msg_ref[0] != None:
451
            db.log_debug(debug_msg_ref[0], log_level)
452 830 aaronmk
453 832 aaronmk
##### Basic queries
454
455 2153 aaronmk
def next_version(name):
456 2163 aaronmk
    version = 1 # first existing name was version 0
457 2586 aaronmk
    match = re.match(r'^(.*)#(\d+)$', name)
458 2153 aaronmk
    if match:
459 2586 aaronmk
        name, version = match.groups()
460
        version = int(version)+1
461 2588 aaronmk
    return sql_gen.add_suffix(name, '#'+str(version))
462 2153 aaronmk
463 2789 aaronmk
def run_query_into(db, query, into=None, add_indexes_=False, **kw_args):
464 2085 aaronmk
    '''Outputs a query to a temp table.
465
    For params, see run_query().
466
    '''
467 2789 aaronmk
    if into == None: return run_query(db, query, **kw_args)
468 2790 aaronmk
469
    assert isinstance(into, sql_gen.Table)
470
471
    kw_args['recover'] = True
472
    kw_args.setdefault('log_ignore_excs', (DuplicateTableException,))
473
474
    temp = not db.autocommit # tables are permanent in autocommit mode
475
    # "temporary tables cannot specify a schema name", so remove schema
476
    if temp: into.schema = None
477
478
    # Create table
479
    while True:
480
        create_query = 'CREATE'
481
        if temp: create_query += ' TEMP'
482
        create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
483 2385 aaronmk
484 2790 aaronmk
        try:
485
            cur = run_query(db, create_query, **kw_args)
486
                # CREATE TABLE AS sets rowcount to # rows in query
487
            break
488
        except DuplicateTableException, e:
489
            into.name = next_version(into.name)
490
            # try again with next version of name
491
492
    if add_indexes_: add_indexes(db, into)
493
494
    return cur
495 2085 aaronmk
496 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
497
498 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
499
500 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
501 2293 aaronmk
    start=None, order_by=order_by_pkey, default_table=None):
502 1981 aaronmk
    '''
503 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
504 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
505 1981 aaronmk
    @param fields Use None to select all fields in the table
506 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
507 2379 aaronmk
        * container can be any iterable type
508 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
509
        * compare_right_side: sql_gen.ValueCond|literal value
510 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
511
        use all columns
512 2786 aaronmk
    @return query
513 1981 aaronmk
    '''
514 2315 aaronmk
    # Parse tables param
515 2121 aaronmk
    if not lists.is_seq(tables): tables = [tables]
516 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
517 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
518 2121 aaronmk
519 2315 aaronmk
    # Parse other params
520 2376 aaronmk
    if conds == None: conds = []
521 2650 aaronmk
    elif dicts.is_dict(conds): conds = conds.items()
522 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
523 135 aaronmk
    assert limit == None or type(limit) == int
524 865 aaronmk
    assert start == None or type(start) == int
525 2315 aaronmk
    if order_by is order_by_pkey:
526
        if distinct_on != []: order_by = None
527
        else: order_by = pkey(db, table0, recover=True)
528 865 aaronmk
529 2315 aaronmk
    query = 'SELECT'
530 2056 aaronmk
531 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
532 2056 aaronmk
533 2200 aaronmk
    # DISTINCT ON columns
534 2233 aaronmk
    if distinct_on != []:
535 2467 aaronmk
        query += '\nDISTINCT'
536 2254 aaronmk
        if distinct_on is not distinct_on_all:
537 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
538
539
    # Columns
540 2467 aaronmk
    query += '\n'
541 1135 aaronmk
    if fields == None: query += '*'
542 2765 aaronmk
    else:
543
        assert fields != []
544
        query += '\n, '.join(map(parse_col, fields))
545 2200 aaronmk
546
    # Main table
547 2467 aaronmk
    query += '\nFROM '+table0.to_str(db)
548 865 aaronmk
549 2122 aaronmk
    # Add joins
550 2271 aaronmk
    left_table = table0
551 2263 aaronmk
    for join_ in tables:
552
        table = join_.table
553 2238 aaronmk
554 2343 aaronmk
        # Parse special values
555
        if join_.type_ is sql_gen.filter_out: # filter no match
556 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
557
                None))
558 2343 aaronmk
559 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
560 2122 aaronmk
561
        left_table = table
562
563 865 aaronmk
    missing = True
564 2376 aaronmk
    if conds != []:
565 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
566
        else: whitespace = '\n'
567 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
568
            .to_str(db) for l, r in conds], 'WHERE')
569 865 aaronmk
        missing = False
570 2227 aaronmk
    if order_by != None:
571 2467 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
572
    if limit != None: query += '\nLIMIT '+str(limit); missing = False
573 865 aaronmk
    if start != None:
574 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
575 865 aaronmk
        missing = False
576
    if missing: warnings.warn(DbWarning(
577
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
578
579 2786 aaronmk
    return query
580 11 aaronmk
581 2054 aaronmk
def select(db, *args, **kw_args):
582
    '''For params, see mk_select() and run_query()'''
583
    recover = kw_args.pop('recover', None)
584
    cacheable = kw_args.pop('cacheable', True)
585 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
586 2054 aaronmk
587 2791 aaronmk
    return run_query(db, mk_select(db, *args, **kw_args), recover, cacheable,
588
        log_level=log_level)
589 2054 aaronmk
590 2788 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, returning=None,
591
    embeddable=False):
592 1960 aaronmk
    '''
593
    @param returning str|None An inserted column (such as pkey) to return
594 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
595 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
596
        query will be fully cached, not just if it raises an exception.
597 1960 aaronmk
    '''
598 2754 aaronmk
    table = sql_gen.remove_table_rename(sql_gen.as_Table(table))
599 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
600 2646 aaronmk
    if cols != None:
601
        cols = [sql_gen.to_name_only_col(v, table).to_str(db) for v in cols]
602 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
603 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
604 2063 aaronmk
605
    # Build query
606 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
607
    query = first_line
608 2467 aaronmk
    if cols != None: query += '\n('+', '.join(cols)+')'
609
    query += '\n'+select_query
610 2063 aaronmk
611
    if returning != None:
612 2743 aaronmk
        query += '\nRETURNING '+sql_gen.to_name_only_col(returning).to_str(db)
613 2063 aaronmk
614 2070 aaronmk
    if embeddable:
615 2327 aaronmk
        assert returning != None
616
617 2070 aaronmk
        # Create function
618 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
619 2327 aaronmk
        return_type = 'SETOF '+returning.to_str(db)+'%TYPE'
620 2189 aaronmk
        while True:
621
            try:
622 2698 aaronmk
                function = sql_gen.TempFunction(function_name, db.autocommit)
623 2194 aaronmk
624 2189 aaronmk
                function_query = '''\
625 2698 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''()
626 2467 aaronmk
RETURNS '''+return_type+'''
627
LANGUAGE sql
628
AS $$
629 2788 aaronmk
'''+query+''';
630 2467 aaronmk
$$;
631 2070 aaronmk
'''
632 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
633 2464 aaronmk
                    log_ignore_excs=(DuplicateFunctionException,))
634 2189 aaronmk
                break # this version was successful
635
            except DuplicateFunctionException, e:
636
                function_name = next_version(function_name)
637
                # try again with next version of name
638 2070 aaronmk
639 2337 aaronmk
        # Return query that uses function
640 2698 aaronmk
        func_table = sql_gen.NamedTable('f', sql_gen.FunctionCall(function),
641 2742 aaronmk
            [returning]) # AS clause requires function alias
642 2787 aaronmk
        return mk_select(db, func_table, start=0, order_by=None)
643 2070 aaronmk
644 2787 aaronmk
    return query
645 2066 aaronmk
646
def insert_select(db, *args, **kw_args):
647 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
648 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
649
        values in
650 2072 aaronmk
    '''
651 2386 aaronmk
    into = kw_args.pop('into', None)
652
    if into != None: kw_args['embeddable'] = True
653 2066 aaronmk
    recover = kw_args.pop('recover', None)
654
    cacheable = kw_args.pop('cacheable', True)
655 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
656 2066 aaronmk
657 2789 aaronmk
    return run_query_into(db, mk_insert_select(db, *args, **kw_args), into,
658 2787 aaronmk
        recover=recover, cacheable=cacheable, log_level=log_level)
659 2063 aaronmk
660 2738 aaronmk
default = sql_gen.default # tells insert() to use the default value for a column
661 2066 aaronmk
662 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
663 2085 aaronmk
    '''For params, see insert_select()'''
664 1960 aaronmk
    if lists.is_seq(row): cols = None
665
    else:
666
        cols = row.keys()
667
        row = row.values()
668 2738 aaronmk
    row = list(row) # ensure that "== []" works
669 1960 aaronmk
670 2738 aaronmk
    if row == []: query = None
671
    else: query = sql_gen.Values(row).to_str(db)
672 1961 aaronmk
673 2788 aaronmk
    return insert_select(db, table, cols, query, *args, **kw_args)
674 11 aaronmk
675 2402 aaronmk
def mk_update(db, table, changes=None, cond=None):
676
    '''
677
    @param changes [(col, new_value),...]
678
        * container can be any iterable type
679
        * col: sql_gen.Code|str (for col name)
680
        * new_value: sql_gen.Code|literal value
681
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
682
    @return str query
683
    '''
684
    query = 'UPDATE '+sql_gen.as_Table(table).to_str(db)+'\nSET\n'
685 2405 aaronmk
    query += ',\n'.join((sql_gen.to_name_only_col(col, table).to_str(db)+' = '
686 2402 aaronmk
        +sql_gen.as_Value(new_value).to_str(db) for col, new_value in changes))
687 2467 aaronmk
    if cond != None: query += '\nWHERE\n'+cond.to_str(db)
688 2402 aaronmk
689
    return query
690
691
def update(db, *args, **kw_args):
692
    '''For params, see mk_update() and run_query()'''
693
    recover = kw_args.pop('recover', None)
694
695 2791 aaronmk
    return run_query(db, mk_update(db, *args, **kw_args), recover)
696 2402 aaronmk
697 135 aaronmk
def last_insert_id(db):
698 1849 aaronmk
    module = util.root_module(db.db)
699 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
700
    elif module == 'MySQLdb': return db.insert_id()
701
    else: return None
702 13 aaronmk
703 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
704 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
705 2415 aaronmk
    to names that will be distinct among the columns' tables.
706 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
707 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
708
    @param into The table for the new columns.
709 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
710
        columns will be included in the mapping even if they are not in cols.
711
        The tables of the provided Col objects will be changed to into, so make
712
        copies of them if you want to keep the original tables.
713
    @param as_items Whether to return a list of dict items instead of a dict
714 2383 aaronmk
    @return dict(orig_col=new_col, ...)
715
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
716 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
717
        * All mappings use the into table so its name can easily be
718 2383 aaronmk
          changed for all columns at once
719
    '''
720 2415 aaronmk
    cols = lists.uniqify(cols)
721
722 2394 aaronmk
    items = []
723 2389 aaronmk
    for col in preserve:
724 2390 aaronmk
        orig_col = copy.copy(col)
725 2392 aaronmk
        col.table = into
726 2394 aaronmk
        items.append((orig_col, col))
727
    preserve = set(preserve)
728
    for col in cols:
729 2716 aaronmk
        if col not in preserve:
730
            items.append((col, sql_gen.Col(str(col), into, col.srcs)))
731 2394 aaronmk
732
    if not as_items: items = dict(items)
733
    return items
734 2383 aaronmk
735 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
736 2391 aaronmk
    '''For params, see mk_flatten_mapping()
737
    @return See return value of mk_flatten_mapping()
738
    '''
739 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
740
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
741 2786 aaronmk
    run_query_into(db, mk_select(db, joins, cols, limit=limit, start=start),
742 2392 aaronmk
        into=into)
743 2394 aaronmk
    return dict(items)
744 2391 aaronmk
745 2751 aaronmk
def mk_track_data_error(db, errors_table, cols, value, error_code, error):
746
    cols = map(sql_gen.to_name_only_col, cols)
747
748
    columns_cols = ['column']
749
    columns = sql_gen.NamedValues('columns', columns_cols,
750
        [[c.name] for c in cols])
751
    values_cols = ['value', 'error_code', 'error']
752
    values = sql_gen.NamedValues('values', values_cols,
753
        [value, error_code, error])
754
755
    select_cols = columns_cols+values_cols
756
    name_only_cols = map(sql_gen.to_name_only_col, select_cols)
757 2755 aaronmk
    errors_table = sql_gen.NamedTable('errors', errors_table)
758 2751 aaronmk
    joins = [columns, sql_gen.Join(values, type_='CROSS'),
759
        sql_gen.Join(errors_table, dict(zip(name_only_cols, select_cols)),
760
        sql_gen.filter_out)]
761
762
    return mk_insert_select(db, errors_table, name_only_cols,
763 2787 aaronmk
        mk_select(db, joins, select_cols, order_by=None))
764 2751 aaronmk
765 2758 aaronmk
def track_data_error(db, errors_table, *args, **kw_args):
766
    '''
767
    @param errors_table If None, does nothing.
768
    '''
769
    if errors_table == None: return
770
    run_query(db, mk_track_data_error(db, errors_table, *args, **kw_args),
771
        cacheable=True, log_level=4)
772
773 2704 aaronmk
def cast(db, type_, col, errors_table=None):
774
    '''Casts an (unrenamed) column or value.
775 2769 aaronmk
    If errors_table set and col has srcs, saves errors in errors_table (using
776
    col's srcs attr as the source columns) and converts errors to warnings.
777 2704 aaronmk
    @param col sql_gen.Col|sql_gen.Literal
778
    @param errors_table None|sql_gen.Table|str
779
    '''
780 2769 aaronmk
    save_errors = (errors_table != None and isinstance(col, sql_gen.Col)
781
        and col.srcs != ())
782
    if not save_errors: # can't save errors
783
        return sql_gen.CustomCode(col.to_str(db)+'::'+type_) # just cast
784 2704 aaronmk
785
    assert not isinstance(col, sql_gen.NamedCol)
786
787 2770 aaronmk
    errors_table = sql_gen.as_Table(errors_table)
788
    srcs = map(sql_gen.to_name_only_col, col.srcs)
789
    function_name = str(sql_gen.FunctionCall(type_, *srcs))
790 2704 aaronmk
    function = sql_gen.TempFunction(function_name, db.autocommit)
791
792 2718 aaronmk
    while True:
793
        # Create function definition
794
        query = '''\
795 2704 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''(value text)
796
RETURNS '''+type_+'''
797
LANGUAGE plpgsql
798
STRICT
799
AS $$
800
BEGIN
801
    /* The explicit cast to the return type is needed to make the cast happen
802
    inside the try block. (Implicit casts to the return type happen at the end
803
    of the function, outside any block.) */
804
    RETURN value::'''+type_+''';
805
EXCEPTION
806
    WHEN data_exception THEN
807
        -- Save error in errors table.
808 2751 aaronmk
        -- Insert the value and error for *each* source column.
809
'''+mk_track_data_error(db, errors_table, srcs,
810
    *map(sql_gen.CustomCode, ['value', 'SQLSTATE', 'SQLERRM']))+''';
811 2704 aaronmk
812
        RAISE WARNING '%', SQLERRM;
813
        RETURN NULL;
814
END;
815
$$;
816
'''
817 2718 aaronmk
818
        # Create function
819
        try:
820
            run_query(db, query, recover=True, cacheable=True,
821
                log_ignore_excs=(DuplicateFunctionException,))
822
            break # successful
823
        except DuplicateFunctionException:
824 2770 aaronmk
            function.name = next_version(function.name)
825
            # try again with next version of name
826 2704 aaronmk
827
    return sql_gen.FunctionCall(function, col)
828
829 2414 aaronmk
##### Database structure queries
830
831 2426 aaronmk
def table_row_count(db, table, recover=None):
832 2786 aaronmk
    return value(run_query(db, mk_select(db, table, [sql_gen.row_count],
833 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
834 2426 aaronmk
835 2414 aaronmk
def table_cols(db, table, recover=None):
836
    return list(col_names(select(db, table, limit=0, order_by=None,
837 2443 aaronmk
        recover=recover, log_level=4)))
838 2414 aaronmk
839 2291 aaronmk
def pkey(db, table, recover=None):
840 832 aaronmk
    '''Assumed to be first column in table'''
841 2339 aaronmk
    return table_cols(db, table, recover)[0]
842 832 aaronmk
843 2559 aaronmk
not_null_col = 'not_null_col'
844 2340 aaronmk
845
def table_not_null_col(db, table, recover=None):
846
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
847
    if not_null_col in table_cols(db, table, recover): return not_null_col
848
    else: return pkey(db, table, recover)
849
850 853 aaronmk
def index_cols(db, table, index):
851
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
852
    automatically created. When you don't know whether something is a UNIQUE
853
    constraint or a UNIQUE index, use this function.'''
854 1909 aaronmk
    module = util.root_module(db.db)
855
    if module == 'psycopg2':
856
        return list(values(run_query(db, '''\
857 853 aaronmk
SELECT attname
858 866 aaronmk
FROM
859
(
860
        SELECT attnum, attname
861
        FROM pg_index
862
        JOIN pg_class index ON index.oid = indexrelid
863
        JOIN pg_class table_ ON table_.oid = indrelid
864
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
865
        WHERE
866 2782 aaronmk
            table_.relname = '''+db.esc_value(table)+'''
867
            AND index.relname = '''+db.esc_value(index)+'''
868 866 aaronmk
    UNION
869
        SELECT attnum, attname
870
        FROM
871
        (
872
            SELECT
873
                indrelid
874
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
875
                    AS indkey
876
            FROM pg_index
877
            JOIN pg_class index ON index.oid = indexrelid
878
            JOIN pg_class table_ ON table_.oid = indrelid
879
            WHERE
880 2782 aaronmk
                table_.relname = '''+db.esc_value(table)+'''
881
                AND index.relname = '''+db.esc_value(index)+'''
882 866 aaronmk
        ) s
883
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
884
) s
885 853 aaronmk
ORDER BY attnum
886 2782 aaronmk
'''
887
            , cacheable=True, log_level=4)))
888 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
889
        ' database')
890 853 aaronmk
891 464 aaronmk
def constraint_cols(db, table, constraint):
892 1849 aaronmk
    module = util.root_module(db.db)
893 464 aaronmk
    if module == 'psycopg2':
894
        return list(values(run_query(db, '''\
895
SELECT attname
896
FROM pg_constraint
897
JOIN pg_class ON pg_class.oid = conrelid
898
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
899
WHERE
900 2783 aaronmk
    relname = '''+db.esc_value(table)+'''
901
    AND conname = '''+db.esc_value(constraint)+'''
902 464 aaronmk
ORDER BY attnum
903 2783 aaronmk
'''
904
            )))
905 464 aaronmk
    else: raise NotImplementedError("Can't list constraint columns for "+module+
906
        ' database')
907
908 2096 aaronmk
row_num_col = '_row_num'
909
910 2688 aaronmk
def add_index(db, exprs, table=None, unique=False):
911
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
912 2538 aaronmk
    Currently, only function calls are supported as expressions.
913
    '''
914 2688 aaronmk
    if not lists.is_seq(exprs): exprs = [exprs]
915 2538 aaronmk
916 2688 aaronmk
    # Parse exprs
917
    old_exprs = exprs[:]
918
    exprs = []
919
    cols = []
920
    for i, expr in enumerate(old_exprs):
921
        expr = copy.deepcopy(expr) # don't modify input!
922
        expr = sql_gen.as_Col(expr)
923
924
        # Extract col
925
        if isinstance(expr, sql_gen.FunctionCall):
926
            col = expr.args[0]
927
            expr = sql_gen.Expr(expr)
928
        else: col = expr
929
930
        # Extract table
931
        if table == None:
932
            assert sql_gen.is_table_col(col)
933
            table = col.table
934
935
        col.table = None
936
937
        exprs.append(expr)
938
        cols.append(col)
939 2408 aaronmk
940 2688 aaronmk
    table = sql_gen.as_Table(table)
941
    index = sql_gen.Table(str(sql_gen.Col(','.join(map(str, cols)), table)))
942
943
    str_ = 'CREATE'
944
    if unique: str_ += ' UNIQUE'
945
    str_ += ' INDEX '+index.to_str(db)+' ON '+table.to_str(db)+' ('+(
946
        ', '.join((v.to_str(db) for v in exprs)))+')'
947
948
    try: run_query(db, str_, recover=True, cacheable=True, log_level=3)
949 2408 aaronmk
    except DuplicateTableException: pass # index already existed
950
951 2686 aaronmk
def add_pkey(db, table, cols=None, recover=None):
952
    '''Adds a primary key.
953
    @param cols [sql_gen.Col,...] The columns in the primary key.
954
        Defaults to the first column in the table.
955 2406 aaronmk
    @pre The table must not already have a primary key.
956
    '''
957
    table = sql_gen.as_Table(table)
958 2686 aaronmk
    if cols == None: cols = [pkey(db, table, recover)]
959
    col_strs = [sql_gen.to_name_only_col(v).to_str(db) for v in cols]
960 2406 aaronmk
961 2781 aaronmk
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD PRIMARY KEY ('
962
        +(', '.join(col_strs))+')', recover=True, cacheable=True, log_level=3,
963
        log_ignore_excs=(DuplicateTableException,))
964 2406 aaronmk
965 2772 aaronmk
already_indexed = object() # tells add_indexes() the pkey has already been added
966
967
def add_indexes(db, table, has_pkey=True):
968
    '''Adds an index on all columns in a table.
969
    @param has_pkey bool|already_indexed Whether a pkey instead of a regular
970
        index should be added on the first column.
971
        * If already_indexed, the pkey is assumed to have already been added
972
    '''
973
    cols = table_cols(db, table)
974
    if has_pkey:
975
        if has_pkey is not already_indexed: add_pkey(db, table)
976
        cols = cols[1:]
977
    for col in cols: add_index(db, col, table)
978
979 2086 aaronmk
def add_row_num(db, table):
980 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
981
    be the primary key.'''
982 2320 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
983 2096 aaronmk
    run_query(db, 'ALTER TABLE '+table+' ADD COLUMN '+row_num_col
984 2443 aaronmk
        +' serial NOT NULL PRIMARY KEY', log_level=3)
985 2086 aaronmk
986 2760 aaronmk
def create_table(db, table, cols, has_pkey=True, col_indexes=True):
987 2675 aaronmk
    '''Creates a table.
988 2681 aaronmk
    @param cols [sql_gen.TypedCol,...] The column names and types
989
    @param has_pkey If set, the first column becomes the primary key.
990 2760 aaronmk
    @param col_indexes bool|[ref]
991
        * If True, indexes will be added on all non-pkey columns.
992
        * If a list reference, [0] will be set to a function to do this.
993
          This can be used to delay index creation until the table is populated.
994 2675 aaronmk
    '''
995
    table = sql_gen.as_Table(table)
996
997 2681 aaronmk
    if has_pkey:
998
        cols[0] = pkey = copy.copy(cols[0]) # don't modify input!
999
        pkey.type += ' NOT NULL PRIMARY KEY'
1000
1001 2675 aaronmk
    str_ = 'CREATE TABLE '+table.to_str(db)+' (\n'
1002 2681 aaronmk
    str_ += '\n, '.join(v.to_str(db) for v in cols)
1003
    str_ += '\n);\n'
1004 2675 aaronmk
    run_query(db, str_, cacheable=True, log_level=2)
1005 2760 aaronmk
1006
    # Add indexes
1007 2773 aaronmk
    if has_pkey: has_pkey = already_indexed
1008
    def add_indexes_(): add_indexes(db, table, has_pkey)
1009
    if isinstance(col_indexes, list): col_indexes[0] = add_indexes_ # defer
1010
    elif col_indexes: add_indexes_() # add now
1011 2675 aaronmk
1012 2684 aaronmk
def vacuum(db, table):
1013
    table = sql_gen.as_Table(table)
1014
    db.with_autocommit(lambda: run_query(db, 'VACUUM ANALYZE '+table.to_str(db),
1015
        log_level=3))
1016
1017 2732 aaronmk
def truncate(db, table, schema='public'):
1018 2777 aaronmk
    table = sql_gen.as_Table(table, schema)
1019 2733 aaronmk
    return run_query(db, 'TRUNCATE '+table.to_str(db)+' CASCADE')
1020 2732 aaronmk
1021 2729 aaronmk
def tables(db, schema_like='public', table_like='%', exact=False):
1022
    if exact: compare = '='
1023
    else: compare = 'LIKE'
1024
1025 1849 aaronmk
    module = util.root_module(db.db)
1026 832 aaronmk
    if module == 'psycopg2':
1027 2779 aaronmk
        conds = [('schemaname', sql_gen.CompareCond(schema_like, compare)),
1028
            ('tablename', sql_gen.CompareCond(table_like, compare))]
1029
        return values(select(db, 'pg_tables', ['tablename'], conds,
1030
            order_by='tablename', log_level=4))
1031 1968 aaronmk
    elif module == 'MySQLdb':
1032 2779 aaronmk
        return values(run_query(db, 'SHOW TABLES LIKE '+db.esc_value(table_like)
1033
            , cacheable=True, log_level=4))
1034 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
1035 830 aaronmk
1036 2726 aaronmk
def table_exists(db, table):
1037
    table = sql_gen.as_Table(table)
1038 2730 aaronmk
    return list(tables(db, table.schema, table.name, exact=True)) != []
1039 2726 aaronmk
1040
def errors_table(db, table, if_exists=True):
1041
    '''
1042
    @param if_exists If set, returns None if the errors table doesn't exist
1043
    @return None|sql_gen.Table
1044
    '''
1045
    table = sql_gen.as_Table(table)
1046
    if table.srcs != (): table = table.srcs[0]
1047
1048
    errors_table = sql_gen.suffixed_table(table, '.errors')
1049
    if if_exists and not table_exists(db, errors_table): return None
1050
    return errors_table
1051
1052 833 aaronmk
##### Database management
1053
1054 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
1055
    '''For kw_args, see tables()'''
1056
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
1057 833 aaronmk
1058 832 aaronmk
##### Heuristic queries
1059
1060 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
1061 1554 aaronmk
    '''Recovers from errors.
1062 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
1063
    '''
1064 2642 aaronmk
    row = sql_gen.ColDict(db, table, row)
1065 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
1066
1067 471 aaronmk
    try:
1068 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
1069 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
1070
            row_ct_ref[0] += cur.rowcount
1071
        return value(cur)
1072 471 aaronmk
    except DuplicateKeyException, e:
1073 2653 aaronmk
        row = sql_gen.ColDict(db, table,
1074
            util.dict_subset_right_join(row, e.cols))
1075
        return value(select(db, table, [pkey_], row, recover=True))
1076 471 aaronmk
1077 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
1078 830 aaronmk
    '''Recovers from errors'''
1079 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
1080 14 aaronmk
    except StopIteration:
1081 40 aaronmk
        if not create: raise
1082 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
1083 2078 aaronmk
1084 2593 aaronmk
def is_func_result(col):
1085
    return col.table.name.find('(') >= 0 and col.name == 'result'
1086
1087 2592 aaronmk
def into_table_name(out_table, in_tables0, mapping, is_func):
1088 2615 aaronmk
    def in_col_str(in_col):
1089
        in_col = sql_gen.remove_col_rename(in_col)
1090
        if isinstance(in_col, sql_gen.Col):
1091
            table = in_col.table
1092
            if table == in_tables0:
1093
                in_col = sql_gen.to_name_only_col(in_col)
1094
            elif is_func_result(in_col): in_col = table # omit col name
1095
        return str(in_col)
1096
1097 2580 aaronmk
    str_ = str(out_table)
1098
    if is_func:
1099 2615 aaronmk
        str_ += '('
1100 2580 aaronmk
1101 2615 aaronmk
        try: value_in_col = mapping['value']
1102
        except KeyError:
1103 2654 aaronmk
            str_ += ', '.join((str(k)+'='+in_col_str(v)
1104 2615 aaronmk
                for k, v in mapping.iteritems()))
1105
        else: str_ += in_col_str(value_in_col)
1106
1107
        str_ += ')'
1108 2580 aaronmk
    else: str_ += '_pkeys'
1109
    return str_
1110
1111 2508 aaronmk
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None, into=None,
1112 2552 aaronmk
    default=None, is_func=False):
1113 2078 aaronmk
    '''Recovers from errors.
1114
    Only works under PostgreSQL (uses INSERT RETURNING).
1115 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
1116
        tables to join with it using the main input table's pkey
1117 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
1118 2612 aaronmk
        * out_table_col: str (*not* sql_gen.Col)
1119
        * in_table_col: sql_gen.Col|literal-value
1120 2489 aaronmk
    @param into The table to contain the output and input pkeys.
1121 2574 aaronmk
        Defaults to `out_table.name+'_pkeys'`.
1122 2509 aaronmk
    @param default The *output* column to use as the pkey for missing rows.
1123
        If this output column does not exist in the mapping, uses None.
1124 2552 aaronmk
    @param is_func Whether out_table is the name of a SQL function, not a table
1125 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
1126 2078 aaronmk
    '''
1127 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
1128 2312 aaronmk
1129 2450 aaronmk
    def log_debug(msg): db.log_debug(msg, level=1.5)
1130 2505 aaronmk
    def col_ustr(str_):
1131 2567 aaronmk
        return strings.repr_no_u(sql_gen.remove_col_rename(str_))
1132 2450 aaronmk
1133 2768 aaronmk
    out_pkey = pkey(db, out_table, recover=True)
1134
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
1135
1136
    if mapping == {}: # need at least one column for INSERT SELECT
1137
        mapping = {out_pkey: None} # ColDict will replace with default value
1138
1139 2486 aaronmk
    log_debug('********** New iteration **********')
1140 2505 aaronmk
    log_debug('Inserting these input columns into '+strings.as_tt(
1141
        out_table.to_str(db))+':\n'+strings.as_table(mapping, ustr=col_ustr))
1142 2463 aaronmk
1143 2382 aaronmk
    # Create input joins from list of input tables
1144
    in_tables_ = in_tables[:] # don't modify input!
1145
    in_tables0 = in_tables_.pop(0) # first table is separate
1146 2735 aaronmk
    errors_table_ = errors_table(db, in_tables0)
1147 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
1148 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
1149 2460 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v,
1150
        {in_pkey: sql_gen.join_same_not_null}) for v in in_tables_]
1151 2131 aaronmk
1152 2592 aaronmk
    if into == None:
1153
        into = into_table_name(out_table, in_tables0, mapping, is_func)
1154
    into = sql_gen.as_Table(into)
1155
1156 2702 aaronmk
    # Set column sources
1157
    in_cols = filter(sql_gen.is_table_col, mapping.values())
1158
    for col in in_cols:
1159
        if col.table == in_tables0: col.set_srcs(sql_gen.src_self)
1160
1161 2486 aaronmk
    log_debug('Joining together input tables into temp table')
1162 2395 aaronmk
    # Place in new table for speed and so don't modify input if values edited
1163 2584 aaronmk
    in_table = sql_gen.Table('in')
1164 2702 aaronmk
    mapping = dicts.join(mapping, flatten(db, in_table, input_joins, in_cols,
1165
        preserve=[in_pkey_col], start=0))
1166 2395 aaronmk
    input_joins = [in_table]
1167 2486 aaronmk
    db.log_debug('Temp table: '+strings.as_tt(in_table.to_str(db)), level=2)
1168 2395 aaronmk
1169 2656 aaronmk
    mapping = sql_gen.ColDict(db, out_table, mapping)
1170
        # after applying dicts.join() because that returns a plain dict
1171
1172 2509 aaronmk
    # Resolve default value column
1173
    try: default = mapping[default]
1174
    except KeyError:
1175
        if default != None:
1176
            db.log_debug('Default value column '
1177
                +strings.as_tt(strings.repr_no_u(default))
1178 2511 aaronmk
                +' does not exist in mapping, falling back to None', level=2.1)
1179 2509 aaronmk
            default = None
1180
1181 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
1182 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
1183
1184 2201 aaronmk
    pkeys_table_exists_ref = [False]
1185 2420 aaronmk
    def insert_into_pkeys(joins, cols):
1186 2786 aaronmk
        query = mk_select(db, joins, cols, order_by=None, start=0)
1187 2201 aaronmk
        if pkeys_table_exists_ref[0]:
1188 2786 aaronmk
            insert_select(db, into, pkeys_names, query)
1189 2201 aaronmk
        else:
1190 2786 aaronmk
            run_query_into(db, query, into=into)
1191 2201 aaronmk
            pkeys_table_exists_ref[0] = True
1192
1193 2429 aaronmk
    limit_ref = [None]
1194 2380 aaronmk
    conds = set()
1195 2233 aaronmk
    distinct_on = []
1196 2325 aaronmk
    def mk_main_select(joins, cols):
1197 2429 aaronmk
        return mk_select(db, joins, cols, conds, distinct_on,
1198
            limit=limit_ref[0], start=0)
1199 2132 aaronmk
1200 2519 aaronmk
    exc_strs = set()
1201 2309 aaronmk
    def log_exc(e):
1202 2519 aaronmk
        e_str = exc.str_(e, first_line_only=True)
1203
        log_debug('Caught exception: '+e_str)
1204
        assert e_str not in exc_strs # avoid infinite loops
1205
        exc_strs.add(e_str)
1206 2735 aaronmk
1207 2451 aaronmk
    def remove_all_rows():
1208 2450 aaronmk
        log_debug('Returning NULL for all rows')
1209 2429 aaronmk
        limit_ref[0] = 0 # just create an empty pkeys table
1210 2735 aaronmk
1211 2758 aaronmk
    def ignore(in_col, value, e):
1212
        track_data_error(db, errors_table_, in_col.srcs, value, e.cause.pgcode,
1213
            e.cause.pgerror)
1214
1215 2545 aaronmk
        in_col_str = strings.as_tt(repr(in_col))
1216 2544 aaronmk
        db.log_debug('Adding index on '+in_col_str+' to enable fast filtering',
1217
            level=2.5)
1218 2537 aaronmk
        add_index(db, in_col)
1219 2735 aaronmk
1220 2545 aaronmk
        log_debug('Ignoring rows with '+in_col_str+' = '
1221
            +strings.as_tt(repr(value)))
1222 2758 aaronmk
    def remove_rows(in_col, value, e):
1223
        ignore(in_col, value, e)
1224 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
1225
        assert cond not in conds # avoid infinite loops
1226 2380 aaronmk
        conds.add(cond)
1227 2758 aaronmk
    def invalid2null(in_col, value, e):
1228
        ignore(in_col, value, e)
1229 2403 aaronmk
        update(db, in_table, [(in_col, None)],
1230
            sql_gen.ColValueCond(in_col, value))
1231 2245 aaronmk
1232 2589 aaronmk
    def insert_pkeys_table(which):
1233
        return sql_gen.Table(sql_gen.add_suffix(in_table.name,
1234
            '_insert_'+which+'_pkeys'))
1235
    insert_out_pkeys = insert_pkeys_table('out')
1236
    insert_in_pkeys = insert_pkeys_table('in')
1237
1238 2206 aaronmk
    # Do inserts and selects
1239 2642 aaronmk
    join_cols = sql_gen.ColDict(db, out_table)
1240 2206 aaronmk
    while True:
1241 2521 aaronmk
        if limit_ref[0] == 0: # special case
1242
            log_debug('Creating an empty pkeys table')
1243 2786 aaronmk
            cur = run_query_into(db, mk_select(db, out_table, [out_pkey],
1244 2521 aaronmk
                limit=limit_ref[0]), into=insert_out_pkeys)
1245
            break # don't do main case
1246
1247 2303 aaronmk
        has_joins = join_cols != {}
1248
1249 2305 aaronmk
        # Prepare to insert new rows
1250 2325 aaronmk
        insert_joins = input_joins[:] # don't modify original!
1251 2403 aaronmk
        insert_args = dict(recover=True, cacheable=False)
1252 2303 aaronmk
        if has_joins:
1253 2317 aaronmk
            distinct_on = [v.to_Col() for v in join_cols.values()]
1254 2325 aaronmk
            insert_joins.append(sql_gen.Join(out_table, join_cols,
1255
                sql_gen.filter_out))
1256
        else:
1257 2404 aaronmk
            insert_args.update(dict(returning=out_pkey, into=insert_out_pkeys))
1258 2786 aaronmk
        main_select = mk_main_select(insert_joins, mapping.values())
1259 2303 aaronmk
1260 2486 aaronmk
        log_debug('Trying to insert new rows')
1261 2206 aaronmk
        try:
1262 2518 aaronmk
            cur = insert_select(db, out_table, mapping.keys(), main_select,
1263
                **insert_args)
1264 2357 aaronmk
            break # insert successful
1265 2206 aaronmk
        except DuplicateKeyException, e:
1266 2309 aaronmk
            log_exc(e)
1267
1268 2258 aaronmk
            old_join_cols = join_cols.copy()
1269 2565 aaronmk
            join_cols.update(util.dict_subset_right_join(mapping, e.cols))
1270 2486 aaronmk
            log_debug('Ignoring existing rows, comparing on these columns:\n'
1271 2505 aaronmk
                +strings.as_inline_table(join_cols, ustr=col_ustr))
1272 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
1273 2230 aaronmk
        except NullValueException, e:
1274 2309 aaronmk
            log_exc(e)
1275
1276 2230 aaronmk
            out_col, = e.cols
1277
            try: in_col = mapping[out_col]
1278 2356 aaronmk
            except KeyError:
1279 2486 aaronmk
                log_debug('Missing mapping for NOT NULL column '+out_col)
1280 2451 aaronmk
                remove_all_rows()
1281 2758 aaronmk
            else: remove_rows(in_col, None, e)
1282 2542 aaronmk
        except FunctionValueException, e:
1283
            log_exc(e)
1284
1285
            func_name = e.name
1286
            value = e.value
1287
            for out_col, in_col in mapping.iteritems():
1288 2758 aaronmk
                in_col = sql_gen.unwrap_func_call(in_col, func_name)
1289
                invalid2null(in_col, value, e)
1290 2525 aaronmk
        except MissingCastException, e:
1291
            log_exc(e)
1292
1293
            out_col = e.col
1294 2706 aaronmk
            type_ = e.type
1295
1296 2724 aaronmk
            log_debug('Casting '+strings.as_tt(out_col)+' input to '
1297
                +strings.as_tt(type_))
1298 2735 aaronmk
            def wrap_func(col): return cast(db, type_, col, errors_table_)
1299 2705 aaronmk
            mapping[out_col] = sql_gen.wrap(wrap_func, mapping[out_col])
1300 2429 aaronmk
        except DatabaseErrors, e:
1301
            log_exc(e)
1302
1303 2531 aaronmk
            msg = 'No handler for exception: '+exc.str_(e)
1304 2451 aaronmk
            warnings.warn(DbWarning(msg))
1305
            log_debug(msg)
1306
            remove_all_rows()
1307 2358 aaronmk
        # after exception handled, rerun loop with additional constraints
1308 2132 aaronmk
1309 2357 aaronmk
    if row_ct_ref != None and cur.rowcount >= 0:
1310
        row_ct_ref[0] += cur.rowcount
1311
1312
    if has_joins:
1313
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
1314 2486 aaronmk
        log_debug('Getting output table pkeys of existing/inserted rows')
1315 2420 aaronmk
        insert_into_pkeys(select_joins, pkeys_cols)
1316 2357 aaronmk
    else:
1317 2404 aaronmk
        add_row_num(db, insert_out_pkeys) # for joining with input pkeys
1318 2357 aaronmk
1319 2486 aaronmk
        log_debug('Getting input table pkeys of inserted rows')
1320 2786 aaronmk
        run_query_into(db, mk_main_select(input_joins, [in_pkey]),
1321 2404 aaronmk
            into=insert_in_pkeys)
1322
        add_row_num(db, insert_in_pkeys) # for joining with output pkeys
1323 2357 aaronmk
1324 2428 aaronmk
        assert table_row_count(db, insert_out_pkeys) == table_row_count(db,
1325
            insert_in_pkeys)
1326
1327 2486 aaronmk
        log_debug('Combining output and input pkeys in inserted order')
1328 2404 aaronmk
        pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
1329 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
1330 2420 aaronmk
        insert_into_pkeys(pkey_joins, pkeys_names)
1331 2357 aaronmk
1332 2486 aaronmk
    db.log_debug('Adding pkey on pkeys table to enable fast joins', level=2.5)
1333 2594 aaronmk
    add_pkey(db, into)
1334 2407 aaronmk
1335 2508 aaronmk
    log_debug('Setting pkeys of missing rows to '+strings.as_tt(repr(default)))
1336 2489 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(into,
1337 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
1338
        # must use join_same_not_null or query will take forever
1339 2420 aaronmk
    insert_into_pkeys(missing_rows_joins,
1340 2508 aaronmk
        [in_pkey_col, sql_gen.NamedCol(out_pkey, default)])
1341 2357 aaronmk
1342 2489 aaronmk
    assert table_row_count(db, into) == table_row_count(db, in_table)
1343 2428 aaronmk
1344 2702 aaronmk
    srcs = []
1345
    if is_func: srcs = sql_gen.cols_srcs(in_cols)
1346
    return sql_gen.Col(out_pkey, into, srcs)
1347 2115 aaronmk
1348
##### Data cleanup
1349
1350 2290 aaronmk
def cleanup_table(db, table, cols):
1351 2677 aaronmk
    table = sql_gen.as_Table(table)
1352
    cols = map(sql_gen.as_Col, cols)
1353 2115 aaronmk
1354 2677 aaronmk
    expr = ('nullif(nullif(trim(both from %s), '+db.esc_value('')+'), '
1355
        +db.esc_value(r'\N')+')')
1356
    changes = [(v, sql_gen.CustomCode(expr % v.to_str(db)))
1357
        for v in cols]
1358 2115 aaronmk
1359 2677 aaronmk
    update(db, table, changes)