Project

General

Profile

1 11 aaronmk
# Database access
2
3 1869 aaronmk
import copy
4 2127 aaronmk
import operator
5 11 aaronmk
import re
6 865 aaronmk
import warnings
7 11 aaronmk
8 300 aaronmk
import exc
9 1909 aaronmk
import dicts
10 1893 aaronmk
import iters
11 1960 aaronmk
import lists
12 1889 aaronmk
from Proxy import Proxy
13 1872 aaronmk
import rand
14 2217 aaronmk
import sql_gen
15 862 aaronmk
import strings
16 131 aaronmk
import util
17 11 aaronmk
18 832 aaronmk
##### Exceptions
19
20 2170 aaronmk
def get_cur_query(cur, input_query=None, input_params=None):
21 2168 aaronmk
    raw_query = None
22
    if hasattr(cur, 'query'): raw_query = cur.query
23
    elif hasattr(cur, '_last_executed'): raw_query = cur._last_executed
24 2170 aaronmk
25
    if raw_query != None: return raw_query
26 2371 aaronmk
    else: return '[input] '+strings.ustr(input_query)+' % '+repr(input_params)
27 14 aaronmk
28 2170 aaronmk
def _add_cursor_info(e, *args, **kw_args):
29
    '''For params, see get_cur_query()'''
30
    exc.add_msg(e, 'query: '+str(get_cur_query(*args, **kw_args)))
31 135 aaronmk
32 300 aaronmk
class DbException(exc.ExceptionWithCause):
33 14 aaronmk
    def __init__(self, msg, cause=None, cur=None):
34 2145 aaronmk
        exc.ExceptionWithCause.__init__(self, msg, cause, cause_newline=True)
35 14 aaronmk
        if cur != None: _add_cursor_info(self, cur)
36
37 2143 aaronmk
class ExceptionWithName(DbException):
38
    def __init__(self, name, cause=None):
39 2145 aaronmk
        DbException.__init__(self, 'for name: '+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
        DbException.__init__(self,
45 2369 aaronmk
            'for name: '+str(name)+'; value: '+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 2345 aaronmk
        DbException.__init__(self, 'Violated '+name+ ' constraint on columns: '
52 2307 aaronmk
            +(', '.join(cols)), cause)
53 2306 aaronmk
        self.name = name
54 468 aaronmk
        self.cols = cols
55 11 aaronmk
56 2143 aaronmk
class NameException(DbException): pass
57
58 2306 aaronmk
class DuplicateKeyException(ConstraintException): pass
59 13 aaronmk
60 2306 aaronmk
class NullValueException(ConstraintException): pass
61 13 aaronmk
62 2240 aaronmk
class FunctionValueException(ExceptionWithNameValue): pass
63 2239 aaronmk
64 2143 aaronmk
class DuplicateTableException(ExceptionWithName): pass
65
66 2188 aaronmk
class DuplicateFunctionException(ExceptionWithName): pass
67
68 89 aaronmk
class EmptyRowException(DbException): pass
69
70 865 aaronmk
##### Warnings
71
72
class DbWarning(UserWarning): pass
73
74 1930 aaronmk
##### Result retrieval
75
76
def col_names(cur): return (col[0] for col in cur.description)
77
78
def rows(cur): return iter(lambda: cur.fetchone(), None)
79
80
def consume_rows(cur):
81
    '''Used to fetch all rows so result will be cached'''
82
    iters.consume_iter(rows(cur))
83
84
def next_row(cur): return rows(cur).next()
85
86
def row(cur):
87
    row_ = next_row(cur)
88
    consume_rows(cur)
89
    return row_
90
91
def next_value(cur): return next_row(cur)[0]
92
93
def value(cur): return row(cur)[0]
94
95
def values(cur): return iters.func_iter(lambda: next_value(cur))
96
97
def value_or_none(cur):
98
    try: return value(cur)
99
    except StopIteration: return None
100
101 2101 aaronmk
##### Input validation
102
103 2396 aaronmk
def clean_name(name): return re.sub(r'\W', r'', name.replace('.', '_'))
104
105 2101 aaronmk
def check_name(name):
106
    if re.search(r'\W', name) != None: raise NameException('Name "'+name
107
        +'" may contain only alphanumeric characters and _')
108
109
def esc_name_by_module(module, name, ignore_case=False):
110 2388 aaronmk
    if module == 'psycopg2' or module == None:
111 2101 aaronmk
        if ignore_case:
112
            # Don't enclose in quotes because this disables case-insensitivity
113
            check_name(name)
114
            return name
115
        else: quote = '"'
116
    elif module == 'MySQLdb': quote = '`'
117
    else: raise NotImplementedError("Can't escape name for "+module+' database')
118
    return quote + name.replace(quote, '') + quote
119
120
def esc_name_by_engine(engine, name, **kw_args):
121
    return esc_name_by_module(db_engines[engine][0], name, **kw_args)
122
123
def esc_name(db, name, **kw_args):
124
    return esc_name_by_module(util.root_module(db.db), name, **kw_args)
125
126
def qual_name(db, schema, table):
127
    def esc_name_(name): return esc_name(db, name)
128
    table = esc_name_(table)
129
    if schema != None: return esc_name_(schema)+'.'+table
130
    else: return table
131
132 1869 aaronmk
##### Database connections
133 1849 aaronmk
134 2097 aaronmk
db_config_names = ['engine', 'host', 'user', 'password', 'database', 'schemas']
135 1926 aaronmk
136 1869 aaronmk
db_engines = {
137
    'MySQL': ('MySQLdb', {'password': 'passwd', 'database': 'db'}),
138
    'PostgreSQL': ('psycopg2', {}),
139
}
140
141
DatabaseErrors_set = set([DbException])
142
DatabaseErrors = tuple(DatabaseErrors_set)
143
144
def _add_module(module):
145
    DatabaseErrors_set.add(module.DatabaseError)
146
    global DatabaseErrors
147
    DatabaseErrors = tuple(DatabaseErrors_set)
148
149
def db_config_str(db_config):
150
    return db_config['engine']+' database '+db_config['database']
151
152 1909 aaronmk
def _query_lookup(query, params): return (query, dicts.make_hashable(params))
153 1894 aaronmk
154 2448 aaronmk
log_debug_none = lambda msg, level=2: None
155 1901 aaronmk
156 1849 aaronmk
class DbConn:
157 2190 aaronmk
    def __init__(self, db_config, serializable=True, autocommit=False,
158
        caching=True, log_debug=log_debug_none):
159 1869 aaronmk
        self.db_config = db_config
160
        self.serializable = serializable
161 2190 aaronmk
        self.autocommit = autocommit
162
        self.caching = caching
163 1901 aaronmk
        self.log_debug = log_debug
164 2193 aaronmk
        self.debug = log_debug != log_debug_none
165 1869 aaronmk
166
        self.__db = None
167 1889 aaronmk
        self.query_results = {}
168 2139 aaronmk
        self._savepoint = 0
169 1869 aaronmk
170
    def __getattr__(self, name):
171
        if name == '__dict__': raise Exception('getting __dict__')
172
        if name == 'db': return self._db()
173
        else: raise AttributeError()
174
175
    def __getstate__(self):
176
        state = copy.copy(self.__dict__) # shallow copy
177 1915 aaronmk
        state['log_debug'] = None # don't pickle the debug callback
178 1869 aaronmk
        state['_DbConn__db'] = None # don't pickle the connection
179
        return state
180
181 2165 aaronmk
    def connected(self): return self.__db != None
182
183 1869 aaronmk
    def _db(self):
184
        if self.__db == None:
185
            # Process db_config
186
            db_config = self.db_config.copy() # don't modify input!
187 2097 aaronmk
            schemas = db_config.pop('schemas', None)
188 1869 aaronmk
            module_name, mappings = db_engines[db_config.pop('engine')]
189
            module = __import__(module_name)
190
            _add_module(module)
191
            for orig, new in mappings.iteritems():
192
                try: util.rename_key(db_config, orig, new)
193
                except KeyError: pass
194
195
            # Connect
196
            self.__db = module.connect(**db_config)
197
198
            # Configure connection
199 2234 aaronmk
            if self.serializable and not self.autocommit: run_raw_query(self,
200
                'SET TRANSACTION ISOLATION LEVEL SERIALIZABLE')
201 2101 aaronmk
            if schemas != None:
202
                schemas_ = ''.join((esc_name(self, s)+', '
203
                    for s in schemas.split(',')))
204
                run_raw_query(self, "SELECT set_config('search_path', \
205
%s || current_setting('search_path'), false)", [schemas_])
206 1869 aaronmk
207
        return self.__db
208 1889 aaronmk
209 1891 aaronmk
    class DbCursor(Proxy):
210 1927 aaronmk
        def __init__(self, outer):
211 1891 aaronmk
            Proxy.__init__(self, outer.db.cursor())
212 2191 aaronmk
            self.outer = outer
213 1927 aaronmk
            self.query_results = outer.query_results
214 1894 aaronmk
            self.query_lookup = None
215 1891 aaronmk
            self.result = []
216 1889 aaronmk
217 1894 aaronmk
        def execute(self, query, params=None):
218 1930 aaronmk
            self._is_insert = query.upper().find('INSERT') >= 0
219 1894 aaronmk
            self.query_lookup = _query_lookup(query, params)
220 2148 aaronmk
            try:
221 2191 aaronmk
                try:
222
                    return_value = self.inner.execute(query, params)
223
                    self.outer.do_autocommit()
224 2148 aaronmk
                finally: self.query = get_cur_query(self.inner)
225 1904 aaronmk
            except Exception, e:
226 2170 aaronmk
                _add_cursor_info(e, self, query, params)
227 1904 aaronmk
                self.result = e # cache the exception as the result
228
                self._cache_result()
229
                raise
230 1930 aaronmk
            # Fetch all rows so result will be cached
231
            if self.rowcount == 0 and not self._is_insert: consume_rows(self)
232 1894 aaronmk
            return return_value
233
234 1891 aaronmk
        def fetchone(self):
235
            row = self.inner.fetchone()
236 1899 aaronmk
            if row != None: self.result.append(row)
237
            # otherwise, fetched all rows
238 1904 aaronmk
            else: self._cache_result()
239
            return row
240
241
        def _cache_result(self):
242 1906 aaronmk
            # For inserts, only cache exceptions since inserts are not
243
            # idempotent, but an invalid insert will always be invalid
244 1930 aaronmk
            if self.query_results != None and (not self._is_insert
245 1906 aaronmk
                or isinstance(self.result, Exception)):
246
247 1894 aaronmk
                assert self.query_lookup != None
248 1916 aaronmk
                self.query_results[self.query_lookup] = self.CacheCursor(
249
                    util.dict_subset(dicts.AttrsDictView(self),
250
                    ['query', 'result', 'rowcount', 'description']))
251 1906 aaronmk
252 1916 aaronmk
        class CacheCursor:
253
            def __init__(self, cached_result): self.__dict__ = cached_result
254
255 1927 aaronmk
            def execute(self, *args, **kw_args):
256 1916 aaronmk
                if isinstance(self.result, Exception): raise self.result
257
                # otherwise, result is a rows list
258
                self.iter = iter(self.result)
259
260
            def fetchone(self):
261
                try: return self.iter.next()
262
                except StopIteration: return None
263 1891 aaronmk
264 2212 aaronmk
    def esc_value(self, value):
265 2215 aaronmk
        module = util.root_module(self.db)
266 2374 aaronmk
        if module == 'psycopg2': str_ = self.db.cursor().mogrify('%s', [value])
267 2212 aaronmk
        elif module == 'MySQLdb':
268
            import _mysql
269 2374 aaronmk
            str_ = _mysql.escape_string(value)
270 2212 aaronmk
        else: raise NotImplementedError("Can't escape value for "+module
271
            +' database')
272 2374 aaronmk
        return strings.to_unicode(str_)
273 2212 aaronmk
274 2347 aaronmk
    def esc_name(self, name): return esc_name(self, name) # calls global func
275
276 2445 aaronmk
    def run_query(self, query, params=None, cacheable=False, log_level=2,
277 2464 aaronmk
        debug_msg_ref=None):
278 2445 aaronmk
        '''
279 2464 aaronmk
        @param log_ignore_excs The log_level will be increased by 2 if the query
280
            throws one of these exceptions.
281 2445 aaronmk
        '''
282 2167 aaronmk
        assert query != None
283
284 2047 aaronmk
        if not self.caching: cacheable = False
285 1903 aaronmk
        used_cache = False
286
        try:
287 1927 aaronmk
            # Get cursor
288
            if cacheable:
289
                query_lookup = _query_lookup(query, params)
290
                try:
291
                    cur = self.query_results[query_lookup]
292
                    used_cache = True
293
                except KeyError: cur = self.DbCursor(self)
294
            else: cur = self.db.cursor()
295
296
            # Run query
297 2148 aaronmk
            cur.execute(query, params)
298 1903 aaronmk
        finally:
299 2464 aaronmk
            if self.debug and debug_msg_ref != None:# only compute msg if needed
300 1903 aaronmk
                if used_cache: cache_status = 'Cache hit'
301
                elif cacheable: cache_status = 'Cache miss'
302
                else: cache_status = 'Non-cacheable'
303 2466 aaronmk
                debug_msg_ref[0] = cache_status+':\n'+str(get_cur_query(cur,
304 2465 aaronmk
                    query, params)).rstrip()
305 1903 aaronmk
306
        return cur
307 1914 aaronmk
308
    def is_cached(self, query, params=None):
309
        return _query_lookup(query, params) in self.query_results
310 2139 aaronmk
311
    def with_savepoint(self, func):
312 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
313 2443 aaronmk
        self.run_query('SAVEPOINT '+savepoint, log_level=4)
314 2139 aaronmk
        self._savepoint += 1
315
        try:
316
            try: return_val = func()
317
            finally:
318
                self._savepoint -= 1
319
                assert self._savepoint >= 0
320
        except:
321 2443 aaronmk
            self.run_query('ROLLBACK TO SAVEPOINT '+savepoint, log_level=4)
322 2139 aaronmk
            raise
323
        else:
324 2443 aaronmk
            self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
325 2191 aaronmk
            self.do_autocommit()
326 2139 aaronmk
            return return_val
327 2191 aaronmk
328
    def do_autocommit(self):
329
        '''Autocommits if outside savepoint'''
330
        assert self._savepoint >= 0
331
        if self.autocommit and self._savepoint == 0:
332
            self.log_debug('Autocommiting')
333
            self.db.commit()
334 1849 aaronmk
335 1869 aaronmk
connect = DbConn
336
337 832 aaronmk
##### Querying
338
339 1894 aaronmk
def run_raw_query(db, *args, **kw_args):
340 2085 aaronmk
    '''For params, see DbConn.run_query()'''
341 1894 aaronmk
    return db.run_query(*args, **kw_args)
342 11 aaronmk
343 2068 aaronmk
def mogrify(db, query, params):
344
    module = util.root_module(db.db)
345
    if module == 'psycopg2': return db.db.cursor().mogrify(query, params)
346
    else: raise NotImplementedError("Can't mogrify query for "+module+
347
        ' database')
348
349 832 aaronmk
##### Recoverable querying
350 15 aaronmk
351 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
352 11 aaronmk
353 2464 aaronmk
def run_query(db, query, params=None, recover=None, cacheable=False,
354
    log_level=2, log_ignore_excs=None, **kw_args):
355 2441 aaronmk
    '''For params, see run_raw_query()'''
356 830 aaronmk
    if recover == None: recover = False
357 2464 aaronmk
    if log_ignore_excs == None: log_ignore_excs = ()
358
    log_ignore_excs = tuple(log_ignore_excs)
359 830 aaronmk
360 2464 aaronmk
    debug_msg_ref = [None]
361 2148 aaronmk
    try:
362 2464 aaronmk
        try:
363
            def run(): return run_raw_query(db, query, params, cacheable,
364
                log_level, debug_msg_ref, **kw_args)
365
            if recover and not db.is_cached(query, params):
366
                return with_savepoint(db, run)
367
            else: return run() # don't need savepoint if cached
368
        except Exception, e:
369
            if not recover: raise # need savepoint to run index_cols()
370
            msg = exc.str_(e)
371
372
            match = re.search(r'duplicate key value violates unique constraint '
373
                r'"((_?[^\W_]+)_[^"]+?)"', msg)
374
            if match:
375
                constraint, table = match.groups()
376
                try: cols = index_cols(db, table, constraint)
377
                except NotImplementedError: raise e
378
                else: raise DuplicateKeyException(constraint, cols, e)
379
380
            match = re.search(r'null value in column "(\w+?)" violates not-null'
381
                r' constraint', msg)
382
            if match: raise NullValueException('NOT NULL', [match.group(1)], e)
383
384
            match = re.search(r'\b(?:invalid input (?:syntax|value)\b.*?'
385
                r'|date/time field value out of range): "(.+?)"\n'
386
                r'(?:(?s).*?)\bfunction "(\w+?)".*?\bat assignment', msg)
387
            if match:
388
                value, name = match.groups()
389
                raise FunctionValueException(name, strings.to_unicode(value), e)
390
391
            match = re.search(r'relation "(\w+?)" already exists', msg)
392
            if match: raise DuplicateTableException(match.group(1), e)
393
394
            match = re.search(r'function "(\w+?)" already exists', msg)
395
            if match: raise DuplicateFunctionException(match.group(1), e)
396
397
            raise # no specific exception raised
398
    except log_ignore_excs:
399
        log_level += 2
400
        raise
401
    finally:
402
        if debug_msg_ref[0] != None: db.log_debug(debug_msg_ref[0], log_level)
403 830 aaronmk
404 832 aaronmk
##### Basic queries
405
406 2153 aaronmk
def next_version(name):
407
    '''Prepends the version # so it won't be removed if the name is truncated'''
408 2163 aaronmk
    version = 1 # first existing name was version 0
409 2153 aaronmk
    match = re.match(r'^v(\d+)_(.*)$', name)
410
    if match:
411
        version = int(match.group(1))+1
412
        name = match.group(2)
413
    return 'v'+str(version)+'_'+name
414
415 2386 aaronmk
def run_query_into(db, query, params, into=None, *args, **kw_args):
416 2085 aaronmk
    '''Outputs a query to a temp table.
417
    For params, see run_query().
418
    '''
419 2386 aaronmk
    if into == None: return run_query(db, query, params, *args, **kw_args)
420 2085 aaronmk
    else: # place rows in temp table
421 2386 aaronmk
        assert isinstance(into, sql_gen.Table)
422 2385 aaronmk
423 2153 aaronmk
        kw_args['recover'] = True
424 2464 aaronmk
        kw_args.setdefault('log_ignore_excs', (DuplicateTableException,))
425 2440 aaronmk
426
        temp = not db.debug # tables are created as permanent in debug mode
427
        # "temporary tables cannot specify a schema name", so remove schema
428
        if temp: into.schema = None
429
430 2153 aaronmk
        while True:
431
            try:
432 2194 aaronmk
                create_query = 'CREATE'
433 2440 aaronmk
                if temp: create_query += ' TEMP'
434 2386 aaronmk
                create_query += ' TABLE '+into.to_str(db)+' AS '+query
435 2194 aaronmk
436
                return run_query(db, create_query, params, *args, **kw_args)
437 2153 aaronmk
                    # CREATE TABLE AS sets rowcount to # rows in query
438
            except DuplicateTableException, e:
439 2386 aaronmk
                into.name = next_version(into.name)
440 2153 aaronmk
                # try again with next version of name
441 2085 aaronmk
442 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
443
444 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
445
446 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
447 2293 aaronmk
    start=None, order_by=order_by_pkey, default_table=None):
448 1981 aaronmk
    '''
449 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
450 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
451 1981 aaronmk
    @param fields Use None to select all fields in the table
452 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
453 2379 aaronmk
        * container can be any iterable type
454 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
455
        * compare_right_side: sql_gen.ValueCond|literal value
456 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
457
        use all columns
458 2054 aaronmk
    @return tuple(query, params)
459 1981 aaronmk
    '''
460 2315 aaronmk
    # Parse tables param
461 2121 aaronmk
    if not lists.is_seq(tables): tables = [tables]
462 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
463 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
464 2121 aaronmk
465 2315 aaronmk
    # Parse other params
466 2376 aaronmk
    if conds == None: conds = []
467
    elif isinstance(conds, dict): conds = conds.items()
468 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
469 135 aaronmk
    assert limit == None or type(limit) == int
470 865 aaronmk
    assert start == None or type(start) == int
471 2315 aaronmk
    if order_by is order_by_pkey:
472
        if distinct_on != []: order_by = None
473
        else: order_by = pkey(db, table0, recover=True)
474 865 aaronmk
475 2315 aaronmk
    query = 'SELECT'
476 2056 aaronmk
477 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
478 2056 aaronmk
479 2200 aaronmk
    # DISTINCT ON columns
480 2233 aaronmk
    if distinct_on != []:
481 2200 aaronmk
        query += ' DISTINCT'
482 2254 aaronmk
        if distinct_on is not distinct_on_all:
483 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
484
485
    # Columns
486
    query += ' '
487 1135 aaronmk
    if fields == None: query += '*'
488 2056 aaronmk
    else: query += ', '.join(map(parse_col, fields))
489 2200 aaronmk
490
    # Main table
491 2271 aaronmk
    query += ' FROM '+table0.to_str(db)
492 865 aaronmk
493 2122 aaronmk
    # Add joins
494 2271 aaronmk
    left_table = table0
495 2263 aaronmk
    for join_ in tables:
496
        table = join_.table
497 2238 aaronmk
498 2343 aaronmk
        # Parse special values
499
        if join_.type_ is sql_gen.filter_out: # filter no match
500 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
501
                None))
502 2343 aaronmk
503 2277 aaronmk
        query += ' '+join_.to_str(db, left_table)
504 2122 aaronmk
505
        left_table = table
506
507 865 aaronmk
    missing = True
508 2376 aaronmk
    if conds != []:
509 2410 aaronmk
        query += ' WHERE '+(' AND '.join(('('+sql_gen.ColValueCond(l, r)
510
            .to_str(db)+')' for l, r in conds)))
511 865 aaronmk
        missing = False
512 2227 aaronmk
    if order_by != None:
513 2285 aaronmk
        query += ' ORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
514 865 aaronmk
    if limit != None: query += ' LIMIT '+str(limit); missing = False
515
    if start != None:
516
        if start != 0: query += ' OFFSET '+str(start)
517
        missing = False
518
    if missing: warnings.warn(DbWarning(
519
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
520
521 2315 aaronmk
    return (query, [])
522 11 aaronmk
523 2054 aaronmk
def select(db, *args, **kw_args):
524
    '''For params, see mk_select() and run_query()'''
525
    recover = kw_args.pop('recover', None)
526
    cacheable = kw_args.pop('cacheable', True)
527 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
528 2054 aaronmk
529
    query, params = mk_select(db, *args, **kw_args)
530 2442 aaronmk
    return run_query(db, query, params, recover, cacheable, log_level=log_level)
531 2054 aaronmk
532 2066 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, params=None,
533 2292 aaronmk
    returning=None, embeddable=False):
534 1960 aaronmk
    '''
535
    @param returning str|None An inserted column (such as pkey) to return
536 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
537 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
538
        query will be fully cached, not just if it raises an exception.
539 1960 aaronmk
    '''
540 2328 aaronmk
    table = sql_gen.as_Table(table)
541 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
542
    if cols != None: cols = [sql_gen.as_Col(v).to_str(db) for v in cols]
543 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
544 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
545 2063 aaronmk
546
    # Build query
547 2328 aaronmk
    query = 'INSERT INTO '+table.to_str(db)
548 2318 aaronmk
    if cols != None: query += ' ('+', '.join(cols)+')'
549 2063 aaronmk
    query += ' '+select_query
550
551
    if returning != None:
552 2327 aaronmk
        returning_name = copy.copy(returning)
553
        returning_name.table = None
554
        returning_name = returning_name.to_str(db)
555
        query += ' RETURNING '+returning_name
556 2063 aaronmk
557 2070 aaronmk
    if embeddable:
558 2327 aaronmk
        assert returning != None
559
560 2070 aaronmk
        # Create function
561 2330 aaronmk
        function_name = '_'.join(['insert', table.name] + cols)
562 2327 aaronmk
        return_type = 'SETOF '+returning.to_str(db)+'%TYPE'
563 2189 aaronmk
        while True:
564
            try:
565 2327 aaronmk
                func_schema = None
566
                if not db.debug: func_schema = 'pg_temp'
567
                function = sql_gen.Table(function_name, func_schema).to_str(db)
568 2194 aaronmk
569 2189 aaronmk
                function_query = '''\
570
CREATE FUNCTION '''+function+'''() RETURNS '''+return_type+'''
571 2070 aaronmk
    LANGUAGE sql
572
    AS $$'''+mogrify(db, query, params)+''';$$;
573
'''
574 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
575 2464 aaronmk
                    log_ignore_excs=(DuplicateFunctionException,))
576 2189 aaronmk
                break # this version was successful
577
            except DuplicateFunctionException, e:
578
                function_name = next_version(function_name)
579
                # try again with next version of name
580 2070 aaronmk
581 2337 aaronmk
        # Return query that uses function
582
        func_table = sql_gen.NamedTable('f', sql_gen.CustomCode(function+'()'),
583
            [returning_name]) # AS clause requires function alias
584
        return mk_select(db, func_table, start=0, order_by=None)
585 2070 aaronmk
586 2066 aaronmk
    return (query, params)
587
588
def insert_select(db, *args, **kw_args):
589 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
590 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
591
        values in
592 2072 aaronmk
    '''
593 2386 aaronmk
    into = kw_args.pop('into', None)
594
    if into != None: kw_args['embeddable'] = True
595 2066 aaronmk
    recover = kw_args.pop('recover', None)
596
    cacheable = kw_args.pop('cacheable', True)
597
598
    query, params = mk_insert_select(db, *args, **kw_args)
599 2386 aaronmk
    return run_query_into(db, query, params, into, recover=recover,
600 2153 aaronmk
        cacheable=cacheable)
601 2063 aaronmk
602 2066 aaronmk
default = object() # tells insert() to use the default value for a column
603
604 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
605 2085 aaronmk
    '''For params, see insert_select()'''
606 1960 aaronmk
    if lists.is_seq(row): cols = None
607
    else:
608
        cols = row.keys()
609
        row = row.values()
610
    row = list(row) # ensure that "!= []" works
611
612 1961 aaronmk
    # Check for special values
613
    labels = []
614
    values = []
615
    for value in row:
616 2254 aaronmk
        if value is default: labels.append('DEFAULT')
617 1961 aaronmk
        else:
618
            labels.append('%s')
619
            values.append(value)
620
621
    # Build query
622 2063 aaronmk
    if values != []: query = ' VALUES ('+(', '.join(labels))+')'
623
    else: query = None
624 1554 aaronmk
625 2064 aaronmk
    return insert_select(db, table, cols, query, values, *args, **kw_args)
626 11 aaronmk
627 2402 aaronmk
def mk_update(db, table, changes=None, cond=None):
628
    '''
629
    @param changes [(col, new_value),...]
630
        * container can be any iterable type
631
        * col: sql_gen.Code|str (for col name)
632
        * new_value: sql_gen.Code|literal value
633
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
634
    @return str query
635
    '''
636
    query = 'UPDATE '+sql_gen.as_Table(table).to_str(db)+'\nSET\n'
637 2405 aaronmk
    query += ',\n'.join((sql_gen.to_name_only_col(col, table).to_str(db)+' = '
638 2402 aaronmk
        +sql_gen.as_Value(new_value).to_str(db) for col, new_value in changes))
639
    if cond != None: query += ' WHERE '+cond.to_str(db)
640
641
    return query
642
643
def update(db, *args, **kw_args):
644
    '''For params, see mk_update() and run_query()'''
645
    recover = kw_args.pop('recover', None)
646
647
    return run_query(db, mk_update(db, *args, **kw_args), [], recover)
648
649 135 aaronmk
def last_insert_id(db):
650 1849 aaronmk
    module = util.root_module(db.db)
651 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
652
    elif module == 'MySQLdb': return db.insert_id()
653
    else: return None
654 13 aaronmk
655 1968 aaronmk
def truncate(db, table, schema='public'):
656
    return run_query(db, 'TRUNCATE '+qual_name(db, schema, table)+' CASCADE')
657 832 aaronmk
658 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
659 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
660 2415 aaronmk
    to names that will be distinct among the columns' tables.
661 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
662 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
663
    @param into The table for the new columns.
664 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
665
        columns will be included in the mapping even if they are not in cols.
666
        The tables of the provided Col objects will be changed to into, so make
667
        copies of them if you want to keep the original tables.
668
    @param as_items Whether to return a list of dict items instead of a dict
669 2383 aaronmk
    @return dict(orig_col=new_col, ...)
670
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
671 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
672
        * All mappings use the into table so its name can easily be
673 2383 aaronmk
          changed for all columns at once
674
    '''
675 2415 aaronmk
    cols = lists.uniqify(cols)
676
677 2394 aaronmk
    items = []
678 2389 aaronmk
    for col in preserve:
679 2390 aaronmk
        orig_col = copy.copy(col)
680 2392 aaronmk
        col.table = into
681 2394 aaronmk
        items.append((orig_col, col))
682
    preserve = set(preserve)
683
    for col in cols:
684 2397 aaronmk
        if col not in preserve:
685
            items.append((col, sql_gen.Col(clean_name(str(col)), into)))
686 2394 aaronmk
687
    if not as_items: items = dict(items)
688
    return items
689 2383 aaronmk
690 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
691 2391 aaronmk
    '''For params, see mk_flatten_mapping()
692
    @return See return value of mk_flatten_mapping()
693
    '''
694 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
695
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
696 2391 aaronmk
    run_query_into(db, *mk_select(db, joins, cols, limit=limit, start=start),
697 2392 aaronmk
        into=into)
698 2394 aaronmk
    return dict(items)
699 2391 aaronmk
700 2414 aaronmk
##### Database structure queries
701
702 2426 aaronmk
def table_row_count(db, table, recover=None):
703
    return value(run_query(db, *mk_select(db, table, [sql_gen.row_count],
704 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
705 2426 aaronmk
706 2414 aaronmk
def table_cols(db, table, recover=None):
707
    return list(col_names(select(db, table, limit=0, order_by=None,
708 2443 aaronmk
        recover=recover, log_level=4)))
709 2414 aaronmk
710 2291 aaronmk
def pkey(db, table, recover=None):
711 832 aaronmk
    '''Assumed to be first column in table'''
712 2339 aaronmk
    return table_cols(db, table, recover)[0]
713 832 aaronmk
714 2340 aaronmk
not_null_col = 'not_null'
715
716
def table_not_null_col(db, table, recover=None):
717
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
718
    if not_null_col in table_cols(db, table, recover): return not_null_col
719
    else: return pkey(db, table, recover)
720
721 853 aaronmk
def index_cols(db, table, index):
722
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
723
    automatically created. When you don't know whether something is a UNIQUE
724
    constraint or a UNIQUE index, use this function.'''
725 1909 aaronmk
    module = util.root_module(db.db)
726
    if module == 'psycopg2':
727
        return list(values(run_query(db, '''\
728 853 aaronmk
SELECT attname
729 866 aaronmk
FROM
730
(
731
        SELECT attnum, attname
732
        FROM pg_index
733
        JOIN pg_class index ON index.oid = indexrelid
734
        JOIN pg_class table_ ON table_.oid = indrelid
735
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
736
        WHERE
737
            table_.relname = %(table)s
738
            AND index.relname = %(index)s
739
    UNION
740
        SELECT attnum, attname
741
        FROM
742
        (
743
            SELECT
744
                indrelid
745
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
746
                    AS indkey
747
            FROM pg_index
748
            JOIN pg_class index ON index.oid = indexrelid
749
            JOIN pg_class table_ ON table_.oid = indrelid
750
            WHERE
751
                table_.relname = %(table)s
752
                AND index.relname = %(index)s
753
        ) s
754
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
755
) s
756 853 aaronmk
ORDER BY attnum
757
''',
758 2443 aaronmk
            {'table': table, 'index': index}, cacheable=True, log_level=4)))
759 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
760
        ' database')
761 853 aaronmk
762 464 aaronmk
def constraint_cols(db, table, constraint):
763 1849 aaronmk
    module = util.root_module(db.db)
764 464 aaronmk
    if module == 'psycopg2':
765
        return list(values(run_query(db, '''\
766
SELECT attname
767
FROM pg_constraint
768
JOIN pg_class ON pg_class.oid = conrelid
769
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
770
WHERE
771
    relname = %(table)s
772
    AND conname = %(constraint)s
773
ORDER BY attnum
774
''',
775
            {'table': table, 'constraint': constraint})))
776
    else: raise NotImplementedError("Can't list constraint columns for "+module+
777
        ' database')
778
779 2096 aaronmk
row_num_col = '_row_num'
780
781 2408 aaronmk
def index_col(db, col):
782
    '''Adds an index on a column if it doesn't already exist.'''
783
    assert sql_gen.is_table_col(col)
784
785
    table = col.table
786
    index = sql_gen.as_Table(clean_name(str(col)))
787
    col = sql_gen.to_name_only_col(col)
788
    try: run_query(db, 'CREATE INDEX '+index.to_str(db)+' ON '+table.to_str(db)
789 2443 aaronmk
        +' ('+col.to_str(db)+')', recover=True, cacheable=True, log_level=3)
790 2408 aaronmk
    except DuplicateTableException: pass # index already existed
791
792 2406 aaronmk
def index_pkey(db, table, recover=None):
793
    '''Makes the first column in a table the primary key.
794
    @pre The table must not already have a primary key.
795
    '''
796
    table = sql_gen.as_Table(table)
797
798
    index = sql_gen.as_Table(table.name+'_pkey')
799
    col = sql_gen.to_name_only_col(pkey(db, table, recover))
800
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD CONSTRAINT '
801 2443 aaronmk
        +index.to_str(db)+' PRIMARY KEY('+col.to_str(db)+')', recover=recover,
802
        log_level=3)
803 2406 aaronmk
804 2086 aaronmk
def add_row_num(db, table):
805 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
806
    be the primary key.'''
807 2320 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
808 2096 aaronmk
    run_query(db, 'ALTER TABLE '+table+' ADD COLUMN '+row_num_col
809 2443 aaronmk
        +' serial NOT NULL PRIMARY KEY', log_level=3)
810 2086 aaronmk
811 1968 aaronmk
def tables(db, schema='public', table_like='%'):
812 1849 aaronmk
    module = util.root_module(db.db)
813 1968 aaronmk
    params = {'schema': schema, 'table_like': table_like}
814 832 aaronmk
    if module == 'psycopg2':
815 1968 aaronmk
        return values(run_query(db, '''\
816
SELECT tablename
817
FROM pg_tables
818
WHERE
819
    schemaname = %(schema)s
820
    AND tablename LIKE %(table_like)s
821
ORDER BY tablename
822
''',
823
            params, cacheable=True))
824
    elif module == 'MySQLdb':
825
        return values(run_query(db, 'SHOW TABLES LIKE %(table_like)s', params,
826
            cacheable=True))
827 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
828 830 aaronmk
829 833 aaronmk
##### Database management
830
831 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
832
    '''For kw_args, see tables()'''
833
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
834 833 aaronmk
835 832 aaronmk
##### Heuristic queries
836
837 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
838 1554 aaronmk
    '''Recovers from errors.
839 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
840
    '''
841 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
842
843 471 aaronmk
    try:
844 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
845 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
846
            row_ct_ref[0] += cur.rowcount
847
        return value(cur)
848 471 aaronmk
    except DuplicateKeyException, e:
849 2104 aaronmk
        return value(select(db, table, [pkey_],
850 1069 aaronmk
            util.dict_subset_right_join(row, e.cols), recover=True))
851 471 aaronmk
852 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
853 830 aaronmk
    '''Recovers from errors'''
854 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
855 14 aaronmk
    except StopIteration:
856 40 aaronmk
        if not create: raise
857 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
858 2078 aaronmk
859 2416 aaronmk
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None):
860 2078 aaronmk
    '''Recovers from errors.
861
    Only works under PostgreSQL (uses INSERT RETURNING).
862 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
863
        tables to join with it using the main input table's pkey
864 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
865
        * out_table_col: sql_gen.Col|str
866 2323 aaronmk
        * in_table_col: sql_gen.Col Wrap literal values in a sql_gen.NamedCol
867 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
868 2078 aaronmk
    '''
869 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
870 2312 aaronmk
    for in_table_col in mapping.itervalues():
871
        assert isinstance(in_table_col, sql_gen.Col)
872
873 2450 aaronmk
    def log_debug(msg): db.log_debug(msg, level=1.5)
874
875 2463 aaronmk
    log_debug('Putting columns: '+str(mapping))
876
877 2329 aaronmk
    temp_prefix = out_table.name
878 2387 aaronmk
    pkeys = sql_gen.Table(temp_prefix+'_pkeys')
879 2131 aaronmk
880 2382 aaronmk
    # Create input joins from list of input tables
881
    in_tables_ = in_tables[:] # don't modify input!
882
    in_tables0 = in_tables_.pop(0) # first table is separate
883 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
884 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
885 2460 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v,
886
        {in_pkey: sql_gen.join_same_not_null}) for v in in_tables_]
887 2131 aaronmk
888 2450 aaronmk
    log_debug('Joining together input tables')
889 2395 aaronmk
    # Place in new table for speed and so don't modify input if values edited
890
    in_table = sql_gen.Table(temp_prefix+'_in')
891
    flatten_cols = filter(sql_gen.is_table_col, mapping.values())
892
    mapping = dicts.join(mapping, flatten(db, in_table, input_joins,
893
        flatten_cols, preserve=[in_pkey_col], start=0))
894
    input_joins = [in_table]
895
896 2279 aaronmk
    out_pkey = pkey(db, out_table, recover=True)
897 2285 aaronmk
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
898 2142 aaronmk
899 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
900 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
901
902 2201 aaronmk
    pkeys_table_exists_ref = [False]
903 2420 aaronmk
    def insert_into_pkeys(joins, cols):
904
        query, params = mk_select(db, joins, cols, order_by=None, start=0)
905 2201 aaronmk
        if pkeys_table_exists_ref[0]:
906 2387 aaronmk
            insert_select(db, pkeys, pkeys_names, query, params)
907 2201 aaronmk
        else:
908 2387 aaronmk
            run_query_into(db, query, params, into=pkeys)
909 2201 aaronmk
            pkeys_table_exists_ref[0] = True
910
911 2429 aaronmk
    limit_ref = [None]
912 2380 aaronmk
    conds = set()
913 2233 aaronmk
    distinct_on = []
914 2325 aaronmk
    def mk_main_select(joins, cols):
915 2429 aaronmk
        return mk_select(db, joins, cols, conds, distinct_on,
916
            limit=limit_ref[0], start=0)
917 2132 aaronmk
918 2309 aaronmk
    def log_exc(e):
919 2450 aaronmk
        log_debug('Caught exception: '+exc.str_(e, first_line_only=True))
920 2451 aaronmk
    def remove_all_rows():
921 2450 aaronmk
        log_debug('Returning NULL for all rows')
922 2429 aaronmk
        limit_ref[0] = 0 # just create an empty pkeys table
923 2409 aaronmk
    def ignore(in_col, value):
924
        in_col_str = str(in_col)
925 2450 aaronmk
        log_debug('Adding index on '+in_col_str+' to enable fast filtering')
926 2409 aaronmk
        index_col(db, in_col)
927 2450 aaronmk
        log_debug('Ignoring rows with '+in_col_str+' = '+repr(value))
928 2403 aaronmk
    def remove_rows(in_col, value):
929 2409 aaronmk
        ignore(in_col, value)
930 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
931
        assert cond not in conds # avoid infinite loops
932 2380 aaronmk
        conds.add(cond)
933 2403 aaronmk
    def invalid2null(in_col, value):
934 2409 aaronmk
        ignore(in_col, value)
935 2403 aaronmk
        update(db, in_table, [(in_col, None)],
936
            sql_gen.ColValueCond(in_col, value))
937 2245 aaronmk
938 2206 aaronmk
    # Do inserts and selects
939 2257 aaronmk
    join_cols = {}
940 2404 aaronmk
    insert_out_pkeys = sql_gen.Table(temp_prefix+'_insert_out_pkeys')
941
    insert_in_pkeys = sql_gen.Table(temp_prefix+'_insert_in_pkeys')
942 2206 aaronmk
    while True:
943 2303 aaronmk
        has_joins = join_cols != {}
944
945 2305 aaronmk
        # Prepare to insert new rows
946 2325 aaronmk
        insert_joins = input_joins[:] # don't modify original!
947 2403 aaronmk
        insert_args = dict(recover=True, cacheable=False)
948 2303 aaronmk
        if has_joins:
949 2317 aaronmk
            distinct_on = [v.to_Col() for v in join_cols.values()]
950 2325 aaronmk
            insert_joins.append(sql_gen.Join(out_table, join_cols,
951
                sql_gen.filter_out))
952
        else:
953 2404 aaronmk
            insert_args.update(dict(returning=out_pkey, into=insert_out_pkeys))
954 2303 aaronmk
955 2450 aaronmk
        log_debug('Inserting new rows')
956 2206 aaronmk
        try:
957
            cur = insert_select(db, out_table, mapping.keys(),
958 2325 aaronmk
                *mk_main_select(insert_joins, mapping.values()), **insert_args)
959 2357 aaronmk
            break # insert successful
960 2206 aaronmk
        except DuplicateKeyException, e:
961 2309 aaronmk
            log_exc(e)
962
963 2258 aaronmk
            old_join_cols = join_cols.copy()
964 2334 aaronmk
            join_cols.update(util.dict_subset(mapping, e.cols))
965 2450 aaronmk
            log_debug('Ignoring existing rows, comparing on '+str(join_cols))
966 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
967 2230 aaronmk
        except NullValueException, e:
968 2309 aaronmk
            log_exc(e)
969
970 2230 aaronmk
            out_col, = e.cols
971
            try: in_col = mapping[out_col]
972 2356 aaronmk
            except KeyError:
973 2451 aaronmk
                log_debug('Missing mapping for NOT NULL '+out_col)
974
                remove_all_rows()
975 2403 aaronmk
            else: remove_rows(in_col, None)
976 2243 aaronmk
        except FunctionValueException, e:
977 2309 aaronmk
            log_exc(e)
978
979 2344 aaronmk
            assert e.name == out_table.name
980 2243 aaronmk
            out_col = 'value' # assume function param was named "value"
981 2403 aaronmk
            invalid2null(mapping[out_col], e.value)
982 2429 aaronmk
        except DatabaseErrors, e:
983
            log_exc(e)
984
985 2451 aaronmk
            msg = 'No handler for exception: '+exc.str_(e, first_line_only=True)
986
            warnings.warn(DbWarning(msg))
987
            log_debug(msg)
988
            remove_all_rows()
989 2358 aaronmk
        # after exception handled, rerun loop with additional constraints
990 2132 aaronmk
991 2357 aaronmk
    if row_ct_ref != None and cur.rowcount >= 0:
992
        row_ct_ref[0] += cur.rowcount
993
994
    if has_joins:
995
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
996 2450 aaronmk
        log_debug('Getting output pkeys of existing/inserted rows')
997 2420 aaronmk
        insert_into_pkeys(select_joins, pkeys_cols)
998 2357 aaronmk
    else:
999 2404 aaronmk
        add_row_num(db, insert_out_pkeys) # for joining with input pkeys
1000 2357 aaronmk
1001 2450 aaronmk
        log_debug('Getting input pkeys for rows in insert')
1002 2357 aaronmk
        run_query_into(db, *mk_main_select(input_joins, [in_pkey]),
1003 2404 aaronmk
            into=insert_in_pkeys)
1004
        add_row_num(db, insert_in_pkeys) # for joining with output pkeys
1005 2357 aaronmk
1006 2428 aaronmk
        assert table_row_count(db, insert_out_pkeys) == table_row_count(db,
1007
            insert_in_pkeys)
1008
1009 2450 aaronmk
        log_debug('Joining together output and input pkeys')
1010 2404 aaronmk
        pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
1011 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
1012 2420 aaronmk
        insert_into_pkeys(pkey_joins, pkeys_names)
1013 2357 aaronmk
1014 2450 aaronmk
    log_debug('Adding pkey on returned pkeys table to enable fast joins')
1015 2407 aaronmk
    index_pkey(db, pkeys)
1016
1017 2450 aaronmk
    log_debug("Setting missing rows' pkeys to NULL")
1018 2387 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(pkeys,
1019 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
1020
        # must use join_same_not_null or query will take forever
1021 2420 aaronmk
    insert_into_pkeys(missing_rows_joins,
1022
        [in_pkey_col, sql_gen.NamedCol(out_pkey, None)])
1023 2357 aaronmk
1024 2428 aaronmk
    assert table_row_count(db, pkeys) == table_row_count(db, in_table)
1025
1026 2387 aaronmk
    return sql_gen.Col(out_pkey, pkeys)
1027 2115 aaronmk
1028
##### Data cleanup
1029
1030 2290 aaronmk
def cleanup_table(db, table, cols):
1031 2115 aaronmk
    def esc_name_(name): return esc_name(db, name)
1032
1033 2290 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
1034 2115 aaronmk
    cols = map(esc_name_, cols)
1035
1036
    run_query(db, 'UPDATE '+table+' SET\n'+(',\n'.join(('\n'+col
1037
        +' = nullif(nullif(trim(both from '+col+"), %(null0)s), %(null1)s)"
1038
            for col in cols))),
1039
        dict(null0='', null1=r'\N'))