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
        exc_log_level=None):
278
        '''
279
        @param exc_log_level The log_level if the query throws an exception.
280
            Defaults to the value of log_level.
281
        '''
282 2167 aaronmk
        assert query != None
283 2445 aaronmk
        if exc_log_level == None: exc_log_level = log_level
284 2167 aaronmk
285 2047 aaronmk
        if not self.caching: cacheable = False
286 1903 aaronmk
        used_cache = False
287 2445 aaronmk
        success = False
288 1903 aaronmk
        try:
289 1927 aaronmk
            # Get cursor
290
            if cacheable:
291
                query_lookup = _query_lookup(query, params)
292
                try:
293
                    cur = self.query_results[query_lookup]
294
                    used_cache = True
295
                except KeyError: cur = self.DbCursor(self)
296
            else: cur = self.db.cursor()
297
298
            # Run query
299 2148 aaronmk
            cur.execute(query, params)
300 2445 aaronmk
301
            success = True
302 1903 aaronmk
        finally:
303 2193 aaronmk
            if self.debug: # only compute msg if needed
304 2445 aaronmk
                if not success: log_level = exc_log_level
305 1903 aaronmk
                if used_cache: cache_status = 'Cache hit'
306
                elif cacheable: cache_status = 'Cache miss'
307
                else: cache_status = 'Non-cacheable'
308 2441 aaronmk
                self.log_debug(cache_status+': '+strings.one_line(
309
                    str(get_cur_query(cur, query, params))), log_level)
310 1903 aaronmk
311
        return cur
312 1914 aaronmk
313
    def is_cached(self, query, params=None):
314
        return _query_lookup(query, params) in self.query_results
315 2139 aaronmk
316
    def with_savepoint(self, func):
317 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
318 2443 aaronmk
        self.run_query('SAVEPOINT '+savepoint, log_level=4)
319 2139 aaronmk
        self._savepoint += 1
320
        try:
321
            try: return_val = func()
322
            finally:
323
                self._savepoint -= 1
324
                assert self._savepoint >= 0
325
        except:
326 2443 aaronmk
            self.run_query('ROLLBACK TO SAVEPOINT '+savepoint, log_level=4)
327 2139 aaronmk
            raise
328
        else:
329 2443 aaronmk
            self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
330 2191 aaronmk
            self.do_autocommit()
331 2139 aaronmk
            return return_val
332 2191 aaronmk
333
    def do_autocommit(self):
334
        '''Autocommits if outside savepoint'''
335
        assert self._savepoint >= 0
336
        if self.autocommit and self._savepoint == 0:
337
            self.log_debug('Autocommiting')
338
            self.db.commit()
339 1849 aaronmk
340 1869 aaronmk
connect = DbConn
341
342 832 aaronmk
##### Querying
343
344 1894 aaronmk
def run_raw_query(db, *args, **kw_args):
345 2085 aaronmk
    '''For params, see DbConn.run_query()'''
346 1894 aaronmk
    return db.run_query(*args, **kw_args)
347 11 aaronmk
348 2068 aaronmk
def mogrify(db, query, params):
349
    module = util.root_module(db.db)
350
    if module == 'psycopg2': return db.db.cursor().mogrify(query, params)
351
    else: raise NotImplementedError("Can't mogrify query for "+module+
352
        ' database')
353
354 832 aaronmk
##### Recoverable querying
355 15 aaronmk
356 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
357 11 aaronmk
358 2441 aaronmk
def run_query(db, query, params=None, recover=None, cacheable=False, **kw_args):
359
    '''For params, see run_raw_query()'''
360 830 aaronmk
    if recover == None: recover = False
361
362 2148 aaronmk
    try:
363 2441 aaronmk
        def run(): return run_raw_query(db, query, params, cacheable, **kw_args)
364 2148 aaronmk
        if recover and not db.is_cached(query, params):
365
            return with_savepoint(db, run)
366
        else: return run() # don't need savepoint if cached
367
    except Exception, e:
368
        if not recover: raise # need savepoint to run index_cols()
369 2247 aaronmk
        msg = exc.str_(e)
370 2240 aaronmk
371 2148 aaronmk
        match = re.search(r'duplicate key value violates unique constraint '
372 2368 aaronmk
            r'"((_?[^\W_]+)_[^"]+?)"', msg)
373 2148 aaronmk
        if match:
374
            constraint, table = match.groups()
375
            try: cols = index_cols(db, table, constraint)
376
            except NotImplementedError: raise e
377 2306 aaronmk
            else: raise DuplicateKeyException(constraint, cols, e)
378 2240 aaronmk
379 2368 aaronmk
        match = re.search(r'null value in column "(\w+?)" violates not-null '
380 2239 aaronmk
            r'constraint', msg)
381 2345 aaronmk
        if match: raise NullValueException('NOT NULL', [match.group(1)], e)
382 2240 aaronmk
383 2381 aaronmk
        match = re.search(r'\b(?:invalid input (?:syntax|value)\b.*?'
384
            r'|date/time field value out of range): "(.+?)"\n'
385 2368 aaronmk
            r'(?:(?s).*?)\bfunction "(\w+?)".*?\bat assignment', msg)
386 2240 aaronmk
        if match:
387 2373 aaronmk
            value, name = match.groups()
388
            raise FunctionValueException(name, strings.to_unicode(value), e)
389 2240 aaronmk
390 2368 aaronmk
        match = re.search(r'relation "(\w+?)" already exists', msg)
391 2148 aaronmk
        if match: raise DuplicateTableException(match.group(1), e)
392 2240 aaronmk
393 2368 aaronmk
        match = re.search(r'function "(\w+?)" already exists', msg)
394 2188 aaronmk
        if match: raise DuplicateFunctionException(match.group(1), e)
395 2240 aaronmk
396 2148 aaronmk
        raise # no specific exception raised
397 830 aaronmk
398 832 aaronmk
##### Basic queries
399
400 2153 aaronmk
def next_version(name):
401
    '''Prepends the version # so it won't be removed if the name is truncated'''
402 2163 aaronmk
    version = 1 # first existing name was version 0
403 2153 aaronmk
    match = re.match(r'^v(\d+)_(.*)$', name)
404
    if match:
405
        version = int(match.group(1))+1
406
        name = match.group(2)
407
    return 'v'+str(version)+'_'+name
408
409 2386 aaronmk
def run_query_into(db, query, params, into=None, *args, **kw_args):
410 2085 aaronmk
    '''Outputs a query to a temp table.
411
    For params, see run_query().
412
    '''
413 2386 aaronmk
    if into == None: return run_query(db, query, params, *args, **kw_args)
414 2085 aaronmk
    else: # place rows in temp table
415 2386 aaronmk
        assert isinstance(into, sql_gen.Table)
416 2385 aaronmk
417 2153 aaronmk
        kw_args['recover'] = True
418 2446 aaronmk
        kw_args.setdefault('exc_log_level', kw_args.get('log_level', 2) + 2)
419
            # by default, will have exc_log_level=4
420 2440 aaronmk
421
        temp = not db.debug # tables are created as permanent in debug mode
422
        # "temporary tables cannot specify a schema name", so remove schema
423
        if temp: into.schema = None
424
425 2153 aaronmk
        while True:
426
            try:
427 2194 aaronmk
                create_query = 'CREATE'
428 2440 aaronmk
                if temp: create_query += ' TEMP'
429 2386 aaronmk
                create_query += ' TABLE '+into.to_str(db)+' AS '+query
430 2194 aaronmk
431
                return run_query(db, create_query, params, *args, **kw_args)
432 2153 aaronmk
                    # CREATE TABLE AS sets rowcount to # rows in query
433
            except DuplicateTableException, e:
434 2386 aaronmk
                into.name = next_version(into.name)
435 2153 aaronmk
                # try again with next version of name
436 2085 aaronmk
437 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
438
439 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
440
441 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
442 2293 aaronmk
    start=None, order_by=order_by_pkey, default_table=None):
443 1981 aaronmk
    '''
444 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
445 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
446 1981 aaronmk
    @param fields Use None to select all fields in the table
447 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
448 2379 aaronmk
        * container can be any iterable type
449 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
450
        * compare_right_side: sql_gen.ValueCond|literal value
451 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
452
        use all columns
453 2054 aaronmk
    @return tuple(query, params)
454 1981 aaronmk
    '''
455 2315 aaronmk
    # Parse tables param
456 2121 aaronmk
    if not lists.is_seq(tables): tables = [tables]
457 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
458 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
459 2121 aaronmk
460 2315 aaronmk
    # Parse other params
461 2376 aaronmk
    if conds == None: conds = []
462
    elif isinstance(conds, dict): conds = conds.items()
463 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
464 135 aaronmk
    assert limit == None or type(limit) == int
465 865 aaronmk
    assert start == None or type(start) == int
466 2315 aaronmk
    if order_by is order_by_pkey:
467
        if distinct_on != []: order_by = None
468
        else: order_by = pkey(db, table0, recover=True)
469 865 aaronmk
470 2315 aaronmk
    query = 'SELECT'
471 2056 aaronmk
472 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
473 2056 aaronmk
474 2200 aaronmk
    # DISTINCT ON columns
475 2233 aaronmk
    if distinct_on != []:
476 2200 aaronmk
        query += ' DISTINCT'
477 2254 aaronmk
        if distinct_on is not distinct_on_all:
478 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
479
480
    # Columns
481
    query += ' '
482 1135 aaronmk
    if fields == None: query += '*'
483 2056 aaronmk
    else: query += ', '.join(map(parse_col, fields))
484 2200 aaronmk
485
    # Main table
486 2271 aaronmk
    query += ' FROM '+table0.to_str(db)
487 865 aaronmk
488 2122 aaronmk
    # Add joins
489 2271 aaronmk
    left_table = table0
490 2263 aaronmk
    for join_ in tables:
491
        table = join_.table
492 2238 aaronmk
493 2343 aaronmk
        # Parse special values
494
        if join_.type_ is sql_gen.filter_out: # filter no match
495 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
496
                None))
497 2343 aaronmk
498 2277 aaronmk
        query += ' '+join_.to_str(db, left_table)
499 2122 aaronmk
500
        left_table = table
501
502 865 aaronmk
    missing = True
503 2376 aaronmk
    if conds != []:
504 2410 aaronmk
        query += ' WHERE '+(' AND '.join(('('+sql_gen.ColValueCond(l, r)
505
            .to_str(db)+')' for l, r in conds)))
506 865 aaronmk
        missing = False
507 2227 aaronmk
    if order_by != None:
508 2285 aaronmk
        query += ' ORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
509 865 aaronmk
    if limit != None: query += ' LIMIT '+str(limit); missing = False
510
    if start != None:
511
        if start != 0: query += ' OFFSET '+str(start)
512
        missing = False
513
    if missing: warnings.warn(DbWarning(
514
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
515
516 2315 aaronmk
    return (query, [])
517 11 aaronmk
518 2054 aaronmk
def select(db, *args, **kw_args):
519
    '''For params, see mk_select() and run_query()'''
520
    recover = kw_args.pop('recover', None)
521
    cacheable = kw_args.pop('cacheable', True)
522 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
523 2054 aaronmk
524
    query, params = mk_select(db, *args, **kw_args)
525 2442 aaronmk
    return run_query(db, query, params, recover, cacheable, log_level=log_level)
526 2054 aaronmk
527 2066 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, params=None,
528 2292 aaronmk
    returning=None, embeddable=False):
529 1960 aaronmk
    '''
530
    @param returning str|None An inserted column (such as pkey) to return
531 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
532 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
533
        query will be fully cached, not just if it raises an exception.
534 1960 aaronmk
    '''
535 2328 aaronmk
    table = sql_gen.as_Table(table)
536 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
537
    if cols != None: cols = [sql_gen.as_Col(v).to_str(db) for v in cols]
538 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
539 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
540 2063 aaronmk
541
    # Build query
542 2328 aaronmk
    query = 'INSERT INTO '+table.to_str(db)
543 2318 aaronmk
    if cols != None: query += ' ('+', '.join(cols)+')'
544 2063 aaronmk
    query += ' '+select_query
545
546
    if returning != None:
547 2327 aaronmk
        returning_name = copy.copy(returning)
548
        returning_name.table = None
549
        returning_name = returning_name.to_str(db)
550
        query += ' RETURNING '+returning_name
551 2063 aaronmk
552 2070 aaronmk
    if embeddable:
553 2327 aaronmk
        assert returning != None
554
555 2070 aaronmk
        # Create function
556 2330 aaronmk
        function_name = '_'.join(['insert', table.name] + cols)
557 2327 aaronmk
        return_type = 'SETOF '+returning.to_str(db)+'%TYPE'
558 2189 aaronmk
        while True:
559
            try:
560 2327 aaronmk
                func_schema = None
561
                if not db.debug: func_schema = 'pg_temp'
562
                function = sql_gen.Table(function_name, func_schema).to_str(db)
563 2194 aaronmk
564 2189 aaronmk
                function_query = '''\
565
CREATE FUNCTION '''+function+'''() RETURNS '''+return_type+'''
566 2070 aaronmk
    LANGUAGE sql
567
    AS $$'''+mogrify(db, query, params)+''';$$;
568
'''
569 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
570
                    exc_log_level=4)
571 2189 aaronmk
                break # this version was successful
572
            except DuplicateFunctionException, e:
573
                function_name = next_version(function_name)
574
                # try again with next version of name
575 2070 aaronmk
576 2337 aaronmk
        # Return query that uses function
577
        func_table = sql_gen.NamedTable('f', sql_gen.CustomCode(function+'()'),
578
            [returning_name]) # AS clause requires function alias
579
        return mk_select(db, func_table, start=0, order_by=None)
580 2070 aaronmk
581 2066 aaronmk
    return (query, params)
582
583
def insert_select(db, *args, **kw_args):
584 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
585 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
586
        values in
587 2072 aaronmk
    '''
588 2386 aaronmk
    into = kw_args.pop('into', None)
589
    if into != None: kw_args['embeddable'] = True
590 2066 aaronmk
    recover = kw_args.pop('recover', None)
591
    cacheable = kw_args.pop('cacheable', True)
592
593
    query, params = mk_insert_select(db, *args, **kw_args)
594 2386 aaronmk
    return run_query_into(db, query, params, into, recover=recover,
595 2153 aaronmk
        cacheable=cacheable)
596 2063 aaronmk
597 2066 aaronmk
default = object() # tells insert() to use the default value for a column
598
599 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
600 2085 aaronmk
    '''For params, see insert_select()'''
601 1960 aaronmk
    if lists.is_seq(row): cols = None
602
    else:
603
        cols = row.keys()
604
        row = row.values()
605
    row = list(row) # ensure that "!= []" works
606
607 1961 aaronmk
    # Check for special values
608
    labels = []
609
    values = []
610
    for value in row:
611 2254 aaronmk
        if value is default: labels.append('DEFAULT')
612 1961 aaronmk
        else:
613
            labels.append('%s')
614
            values.append(value)
615
616
    # Build query
617 2063 aaronmk
    if values != []: query = ' VALUES ('+(', '.join(labels))+')'
618
    else: query = None
619 1554 aaronmk
620 2064 aaronmk
    return insert_select(db, table, cols, query, values, *args, **kw_args)
621 11 aaronmk
622 2402 aaronmk
def mk_update(db, table, changes=None, cond=None):
623
    '''
624
    @param changes [(col, new_value),...]
625
        * container can be any iterable type
626
        * col: sql_gen.Code|str (for col name)
627
        * new_value: sql_gen.Code|literal value
628
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
629
    @return str query
630
    '''
631
    query = 'UPDATE '+sql_gen.as_Table(table).to_str(db)+'\nSET\n'
632 2405 aaronmk
    query += ',\n'.join((sql_gen.to_name_only_col(col, table).to_str(db)+' = '
633 2402 aaronmk
        +sql_gen.as_Value(new_value).to_str(db) for col, new_value in changes))
634
    if cond != None: query += ' WHERE '+cond.to_str(db)
635
636
    return query
637
638
def update(db, *args, **kw_args):
639
    '''For params, see mk_update() and run_query()'''
640
    recover = kw_args.pop('recover', None)
641
642
    return run_query(db, mk_update(db, *args, **kw_args), [], recover)
643
644 135 aaronmk
def last_insert_id(db):
645 1849 aaronmk
    module = util.root_module(db.db)
646 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
647
    elif module == 'MySQLdb': return db.insert_id()
648
    else: return None
649 13 aaronmk
650 1968 aaronmk
def truncate(db, table, schema='public'):
651
    return run_query(db, 'TRUNCATE '+qual_name(db, schema, table)+' CASCADE')
652 832 aaronmk
653 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
654 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
655 2415 aaronmk
    to names that will be distinct among the columns' tables.
656 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
657 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
658
    @param into The table for the new columns.
659 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
660
        columns will be included in the mapping even if they are not in cols.
661
        The tables of the provided Col objects will be changed to into, so make
662
        copies of them if you want to keep the original tables.
663
    @param as_items Whether to return a list of dict items instead of a dict
664 2383 aaronmk
    @return dict(orig_col=new_col, ...)
665
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
666 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
667
        * All mappings use the into table so its name can easily be
668 2383 aaronmk
          changed for all columns at once
669
    '''
670 2415 aaronmk
    cols = lists.uniqify(cols)
671
672 2394 aaronmk
    items = []
673 2389 aaronmk
    for col in preserve:
674 2390 aaronmk
        orig_col = copy.copy(col)
675 2392 aaronmk
        col.table = into
676 2394 aaronmk
        items.append((orig_col, col))
677
    preserve = set(preserve)
678
    for col in cols:
679 2397 aaronmk
        if col not in preserve:
680
            items.append((col, sql_gen.Col(clean_name(str(col)), into)))
681 2394 aaronmk
682
    if not as_items: items = dict(items)
683
    return items
684 2383 aaronmk
685 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
686 2391 aaronmk
    '''For params, see mk_flatten_mapping()
687
    @return See return value of mk_flatten_mapping()
688
    '''
689 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
690
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
691 2391 aaronmk
    run_query_into(db, *mk_select(db, joins, cols, limit=limit, start=start),
692 2392 aaronmk
        into=into)
693 2394 aaronmk
    return dict(items)
694 2391 aaronmk
695 2414 aaronmk
##### Database structure queries
696
697 2426 aaronmk
def table_row_count(db, table, recover=None):
698
    return value(run_query(db, *mk_select(db, table, [sql_gen.row_count],
699 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
700 2426 aaronmk
701 2414 aaronmk
def table_cols(db, table, recover=None):
702
    return list(col_names(select(db, table, limit=0, order_by=None,
703 2443 aaronmk
        recover=recover, log_level=4)))
704 2414 aaronmk
705 2291 aaronmk
def pkey(db, table, recover=None):
706 832 aaronmk
    '''Assumed to be first column in table'''
707 2339 aaronmk
    return table_cols(db, table, recover)[0]
708 832 aaronmk
709 2340 aaronmk
not_null_col = 'not_null'
710
711
def table_not_null_col(db, table, recover=None):
712
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
713
    if not_null_col in table_cols(db, table, recover): return not_null_col
714
    else: return pkey(db, table, recover)
715
716 853 aaronmk
def index_cols(db, table, index):
717
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
718
    automatically created. When you don't know whether something is a UNIQUE
719
    constraint or a UNIQUE index, use this function.'''
720 1909 aaronmk
    module = util.root_module(db.db)
721
    if module == 'psycopg2':
722
        return list(values(run_query(db, '''\
723 853 aaronmk
SELECT attname
724 866 aaronmk
FROM
725
(
726
        SELECT attnum, attname
727
        FROM pg_index
728
        JOIN pg_class index ON index.oid = indexrelid
729
        JOIN pg_class table_ ON table_.oid = indrelid
730
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
731
        WHERE
732
            table_.relname = %(table)s
733
            AND index.relname = %(index)s
734
    UNION
735
        SELECT attnum, attname
736
        FROM
737
        (
738
            SELECT
739
                indrelid
740
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
741
                    AS indkey
742
            FROM pg_index
743
            JOIN pg_class index ON index.oid = indexrelid
744
            JOIN pg_class table_ ON table_.oid = indrelid
745
            WHERE
746
                table_.relname = %(table)s
747
                AND index.relname = %(index)s
748
        ) s
749
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
750
) s
751 853 aaronmk
ORDER BY attnum
752
''',
753 2443 aaronmk
            {'table': table, 'index': index}, cacheable=True, log_level=4)))
754 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
755
        ' database')
756 853 aaronmk
757 464 aaronmk
def constraint_cols(db, table, constraint):
758 1849 aaronmk
    module = util.root_module(db.db)
759 464 aaronmk
    if module == 'psycopg2':
760
        return list(values(run_query(db, '''\
761
SELECT attname
762
FROM pg_constraint
763
JOIN pg_class ON pg_class.oid = conrelid
764
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
765
WHERE
766
    relname = %(table)s
767
    AND conname = %(constraint)s
768
ORDER BY attnum
769
''',
770
            {'table': table, 'constraint': constraint})))
771
    else: raise NotImplementedError("Can't list constraint columns for "+module+
772
        ' database')
773
774 2096 aaronmk
row_num_col = '_row_num'
775
776 2408 aaronmk
def index_col(db, col):
777
    '''Adds an index on a column if it doesn't already exist.'''
778
    assert sql_gen.is_table_col(col)
779
780
    table = col.table
781
    index = sql_gen.as_Table(clean_name(str(col)))
782
    col = sql_gen.to_name_only_col(col)
783
    try: run_query(db, 'CREATE INDEX '+index.to_str(db)+' ON '+table.to_str(db)
784 2443 aaronmk
        +' ('+col.to_str(db)+')', recover=True, cacheable=True, log_level=3)
785 2408 aaronmk
    except DuplicateTableException: pass # index already existed
786
787 2406 aaronmk
def index_pkey(db, table, recover=None):
788
    '''Makes the first column in a table the primary key.
789
    @pre The table must not already have a primary key.
790
    '''
791
    table = sql_gen.as_Table(table)
792
793
    index = sql_gen.as_Table(table.name+'_pkey')
794
    col = sql_gen.to_name_only_col(pkey(db, table, recover))
795
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD CONSTRAINT '
796 2443 aaronmk
        +index.to_str(db)+' PRIMARY KEY('+col.to_str(db)+')', recover=recover,
797
        log_level=3)
798 2406 aaronmk
799 2086 aaronmk
def add_row_num(db, table):
800 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
801
    be the primary key.'''
802 2320 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
803 2096 aaronmk
    run_query(db, 'ALTER TABLE '+table+' ADD COLUMN '+row_num_col
804 2443 aaronmk
        +' serial NOT NULL PRIMARY KEY', log_level=3)
805 2086 aaronmk
806 1968 aaronmk
def tables(db, schema='public', table_like='%'):
807 1849 aaronmk
    module = util.root_module(db.db)
808 1968 aaronmk
    params = {'schema': schema, 'table_like': table_like}
809 832 aaronmk
    if module == 'psycopg2':
810 1968 aaronmk
        return values(run_query(db, '''\
811
SELECT tablename
812
FROM pg_tables
813
WHERE
814
    schemaname = %(schema)s
815
    AND tablename LIKE %(table_like)s
816
ORDER BY tablename
817
''',
818
            params, cacheable=True))
819
    elif module == 'MySQLdb':
820
        return values(run_query(db, 'SHOW TABLES LIKE %(table_like)s', params,
821
            cacheable=True))
822 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
823 830 aaronmk
824 833 aaronmk
##### Database management
825
826 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
827
    '''For kw_args, see tables()'''
828
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
829 833 aaronmk
830 832 aaronmk
##### Heuristic queries
831
832 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
833 1554 aaronmk
    '''Recovers from errors.
834 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
835
    '''
836 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
837
838 471 aaronmk
    try:
839 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
840 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
841
            row_ct_ref[0] += cur.rowcount
842
        return value(cur)
843 471 aaronmk
    except DuplicateKeyException, e:
844 2104 aaronmk
        return value(select(db, table, [pkey_],
845 1069 aaronmk
            util.dict_subset_right_join(row, e.cols), recover=True))
846 471 aaronmk
847 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
848 830 aaronmk
    '''Recovers from errors'''
849 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
850 14 aaronmk
    except StopIteration:
851 40 aaronmk
        if not create: raise
852 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
853 2078 aaronmk
854 2416 aaronmk
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None):
855 2078 aaronmk
    '''Recovers from errors.
856
    Only works under PostgreSQL (uses INSERT RETURNING).
857 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
858
        tables to join with it using the main input table's pkey
859 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
860
        * out_table_col: sql_gen.Col|str
861 2323 aaronmk
        * in_table_col: sql_gen.Col Wrap literal values in a sql_gen.NamedCol
862 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
863 2078 aaronmk
    '''
864 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
865 2312 aaronmk
    for in_table_col in mapping.itervalues():
866
        assert isinstance(in_table_col, sql_gen.Col)
867
868 2329 aaronmk
    temp_prefix = out_table.name
869 2387 aaronmk
    pkeys = sql_gen.Table(temp_prefix+'_pkeys')
870 2131 aaronmk
871 2382 aaronmk
    # Create input joins from list of input tables
872
    in_tables_ = in_tables[:] # don't modify input!
873
    in_tables0 = in_tables_.pop(0) # first table is separate
874 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
875 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
876 2352 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v, {in_pkey: sql_gen.join_same})
877 2382 aaronmk
        for v in in_tables_]
878 2131 aaronmk
879 2395 aaronmk
    db.log_debug('Joining together input tables')
880
    # Place in new table for speed and so don't modify input if values edited
881
    in_table = sql_gen.Table(temp_prefix+'_in')
882
    flatten_cols = filter(sql_gen.is_table_col, mapping.values())
883
    mapping = dicts.join(mapping, flatten(db, in_table, input_joins,
884
        flatten_cols, preserve=[in_pkey_col], start=0))
885
    input_joins = [in_table]
886
887 2279 aaronmk
    out_pkey = pkey(db, out_table, recover=True)
888 2285 aaronmk
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
889 2142 aaronmk
890 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
891 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
892
893 2201 aaronmk
    pkeys_table_exists_ref = [False]
894 2420 aaronmk
    def insert_into_pkeys(joins, cols):
895
        query, params = mk_select(db, joins, cols, order_by=None, start=0)
896 2201 aaronmk
        if pkeys_table_exists_ref[0]:
897 2387 aaronmk
            insert_select(db, pkeys, pkeys_names, query, params)
898 2201 aaronmk
        else:
899 2387 aaronmk
            run_query_into(db, query, params, into=pkeys)
900 2201 aaronmk
            pkeys_table_exists_ref[0] = True
901
902 2429 aaronmk
    limit_ref = [None]
903 2380 aaronmk
    conds = set()
904 2233 aaronmk
    distinct_on = []
905 2325 aaronmk
    def mk_main_select(joins, cols):
906 2429 aaronmk
        return mk_select(db, joins, cols, conds, distinct_on,
907
            limit=limit_ref[0], start=0)
908 2132 aaronmk
909 2309 aaronmk
    def log_exc(e):
910
        db.log_debug('Caught exception: '+exc.str_(e, first_line_only=True))
911 2429 aaronmk
    def remove_all_rows(msg):
912
        warnings.warn(DbWarning(msg))
913
        db.log_debug(msg.partition('\n')[0])
914
        db.log_debug('Returning NULL for all rows')
915
        limit_ref[0] = 0 # just create an empty pkeys table
916 2409 aaronmk
    def ignore(in_col, value):
917
        in_col_str = str(in_col)
918
        db.log_debug('Adding index on '+in_col_str+' to enable fast filtering')
919
        index_col(db, in_col)
920
        db.log_debug('Ignoring rows with '+in_col_str+' = '+repr(value))
921 2403 aaronmk
    def remove_rows(in_col, value):
922 2409 aaronmk
        ignore(in_col, value)
923 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
924
        assert cond not in conds # avoid infinite loops
925 2380 aaronmk
        conds.add(cond)
926 2403 aaronmk
    def invalid2null(in_col, value):
927 2409 aaronmk
        ignore(in_col, value)
928 2403 aaronmk
        update(db, in_table, [(in_col, None)],
929
            sql_gen.ColValueCond(in_col, value))
930 2245 aaronmk
931 2206 aaronmk
    # Do inserts and selects
932 2257 aaronmk
    join_cols = {}
933 2404 aaronmk
    insert_out_pkeys = sql_gen.Table(temp_prefix+'_insert_out_pkeys')
934
    insert_in_pkeys = sql_gen.Table(temp_prefix+'_insert_in_pkeys')
935 2206 aaronmk
    while True:
936 2303 aaronmk
        has_joins = join_cols != {}
937
938 2305 aaronmk
        # Prepare to insert new rows
939 2325 aaronmk
        insert_joins = input_joins[:] # don't modify original!
940 2403 aaronmk
        insert_args = dict(recover=True, cacheable=False)
941 2303 aaronmk
        if has_joins:
942 2317 aaronmk
            distinct_on = [v.to_Col() for v in join_cols.values()]
943 2325 aaronmk
            insert_joins.append(sql_gen.Join(out_table, join_cols,
944
                sql_gen.filter_out))
945
        else:
946 2404 aaronmk
            insert_args.update(dict(returning=out_pkey, into=insert_out_pkeys))
947 2303 aaronmk
948 2305 aaronmk
        db.log_debug('Inserting new rows')
949 2206 aaronmk
        try:
950
            cur = insert_select(db, out_table, mapping.keys(),
951 2325 aaronmk
                *mk_main_select(insert_joins, mapping.values()), **insert_args)
952 2357 aaronmk
            break # insert successful
953 2206 aaronmk
        except DuplicateKeyException, e:
954 2309 aaronmk
            log_exc(e)
955
956 2258 aaronmk
            old_join_cols = join_cols.copy()
957 2334 aaronmk
            join_cols.update(util.dict_subset(mapping, e.cols))
958 2245 aaronmk
            db.log_debug('Ignoring existing rows, comparing on '+str(join_cols))
959 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
960 2230 aaronmk
        except NullValueException, e:
961 2309 aaronmk
            log_exc(e)
962
963 2230 aaronmk
            out_col, = e.cols
964
            try: in_col = mapping[out_col]
965 2356 aaronmk
            except KeyError:
966 2429 aaronmk
                remove_all_rows('Missing mapping for NOT NULL '+out_col)
967 2403 aaronmk
            else: remove_rows(in_col, None)
968 2243 aaronmk
        except FunctionValueException, e:
969 2309 aaronmk
            log_exc(e)
970
971 2344 aaronmk
            assert e.name == out_table.name
972 2243 aaronmk
            out_col = 'value' # assume function param was named "value"
973 2403 aaronmk
            invalid2null(mapping[out_col], e.value)
974 2429 aaronmk
        except DatabaseErrors, e:
975
            log_exc(e)
976
977
            remove_all_rows('No handler for exception: '+exc.str_(e))
978 2358 aaronmk
        # after exception handled, rerun loop with additional constraints
979 2132 aaronmk
980 2357 aaronmk
    if row_ct_ref != None and cur.rowcount >= 0:
981
        row_ct_ref[0] += cur.rowcount
982
983
    if has_joins:
984
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
985
        db.log_debug('Getting output pkeys of existing/inserted rows')
986 2420 aaronmk
        insert_into_pkeys(select_joins, pkeys_cols)
987 2357 aaronmk
    else:
988 2404 aaronmk
        add_row_num(db, insert_out_pkeys) # for joining with input pkeys
989 2357 aaronmk
990
        db.log_debug('Getting input pkeys for rows in insert')
991
        run_query_into(db, *mk_main_select(input_joins, [in_pkey]),
992 2404 aaronmk
            into=insert_in_pkeys)
993
        add_row_num(db, insert_in_pkeys) # for joining with output pkeys
994 2357 aaronmk
995 2428 aaronmk
        assert table_row_count(db, insert_out_pkeys) == table_row_count(db,
996
            insert_in_pkeys)
997
998 2357 aaronmk
        db.log_debug('Joining together output and input pkeys')
999 2404 aaronmk
        pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
1000 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
1001 2420 aaronmk
        insert_into_pkeys(pkey_joins, pkeys_names)
1002 2357 aaronmk
1003 2413 aaronmk
    db.log_debug('Adding pkey on returned pkeys table to enable fast joins')
1004 2407 aaronmk
    index_pkey(db, pkeys)
1005
1006 2357 aaronmk
    db.log_debug("Setting missing rows' pkeys to NULL")
1007 2387 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(pkeys,
1008 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
1009
        # must use join_same_not_null or query will take forever
1010 2420 aaronmk
    insert_into_pkeys(missing_rows_joins,
1011
        [in_pkey_col, sql_gen.NamedCol(out_pkey, None)])
1012 2357 aaronmk
1013 2428 aaronmk
    assert table_row_count(db, pkeys) == table_row_count(db, in_table)
1014
1015 2387 aaronmk
    return sql_gen.Col(out_pkey, pkeys)
1016 2115 aaronmk
1017
##### Data cleanup
1018
1019 2290 aaronmk
def cleanup_table(db, table, cols):
1020 2115 aaronmk
    def esc_name_(name): return esc_name(db, name)
1021
1022 2290 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
1023 2115 aaronmk
    cols = map(esc_name_, cols)
1024
1025
    run_query(db, 'UPDATE '+table+' SET\n'+(',\n'.join(('\n'+col
1026
        +' = nullif(nullif(trim(both from '+col+"), %(null0)s), %(null1)s)"
1027
            for col in cols))),
1028
        dict(null0='', null1=r'\N'))