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 1901 aaronmk
log_debug_none = lambda msg: None
155
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 2441 aaronmk
    def run_query(self, query, params=None, cacheable=False, log_level=2):
277 2167 aaronmk
        assert query != None
278
279 2047 aaronmk
        if not self.caching: cacheable = False
280 1903 aaronmk
        used_cache = False
281
        try:
282 1927 aaronmk
            # Get cursor
283
            if cacheable:
284
                query_lookup = _query_lookup(query, params)
285
                try:
286
                    cur = self.query_results[query_lookup]
287
                    used_cache = True
288
                except KeyError: cur = self.DbCursor(self)
289
            else: cur = self.db.cursor()
290
291
            # Run query
292 2148 aaronmk
            cur.execute(query, params)
293 1903 aaronmk
        finally:
294 2193 aaronmk
            if self.debug: # only compute msg if needed
295 1903 aaronmk
                if used_cache: cache_status = 'Cache hit'
296
                elif cacheable: cache_status = 'Cache miss'
297
                else: cache_status = 'Non-cacheable'
298 2441 aaronmk
                self.log_debug(cache_status+': '+strings.one_line(
299
                    str(get_cur_query(cur, query, params))), log_level)
300 1903 aaronmk
301
        return cur
302 1914 aaronmk
303
    def is_cached(self, query, params=None):
304
        return _query_lookup(query, params) in self.query_results
305 2139 aaronmk
306
    def with_savepoint(self, func):
307 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
308 2443 aaronmk
        self.run_query('SAVEPOINT '+savepoint, log_level=4)
309 2139 aaronmk
        self._savepoint += 1
310
        try:
311
            try: return_val = func()
312
            finally:
313
                self._savepoint -= 1
314
                assert self._savepoint >= 0
315
        except:
316 2443 aaronmk
            self.run_query('ROLLBACK TO SAVEPOINT '+savepoint, log_level=4)
317 2139 aaronmk
            raise
318
        else:
319 2443 aaronmk
            self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
320 2191 aaronmk
            self.do_autocommit()
321 2139 aaronmk
            return return_val
322 2191 aaronmk
323
    def do_autocommit(self):
324
        '''Autocommits if outside savepoint'''
325
        assert self._savepoint >= 0
326
        if self.autocommit and self._savepoint == 0:
327
            self.log_debug('Autocommiting')
328
            self.db.commit()
329 1849 aaronmk
330 1869 aaronmk
connect = DbConn
331
332 832 aaronmk
##### Querying
333
334 1894 aaronmk
def run_raw_query(db, *args, **kw_args):
335 2085 aaronmk
    '''For params, see DbConn.run_query()'''
336 1894 aaronmk
    return db.run_query(*args, **kw_args)
337 11 aaronmk
338 2068 aaronmk
def mogrify(db, query, params):
339
    module = util.root_module(db.db)
340
    if module == 'psycopg2': return db.db.cursor().mogrify(query, params)
341
    else: raise NotImplementedError("Can't mogrify query for "+module+
342
        ' database')
343
344 832 aaronmk
##### Recoverable querying
345 15 aaronmk
346 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
347 11 aaronmk
348 2441 aaronmk
def run_query(db, query, params=None, recover=None, cacheable=False, **kw_args):
349
    '''For params, see run_raw_query()'''
350 830 aaronmk
    if recover == None: recover = False
351
352 2148 aaronmk
    try:
353 2441 aaronmk
        def run(): return run_raw_query(db, query, params, cacheable, **kw_args)
354 2148 aaronmk
        if recover and not db.is_cached(query, params):
355
            return with_savepoint(db, run)
356
        else: return run() # don't need savepoint if cached
357
    except Exception, e:
358
        if not recover: raise # need savepoint to run index_cols()
359 2247 aaronmk
        msg = exc.str_(e)
360 2240 aaronmk
361 2148 aaronmk
        match = re.search(r'duplicate key value violates unique constraint '
362 2368 aaronmk
            r'"((_?[^\W_]+)_[^"]+?)"', msg)
363 2148 aaronmk
        if match:
364
            constraint, table = match.groups()
365
            try: cols = index_cols(db, table, constraint)
366
            except NotImplementedError: raise e
367 2306 aaronmk
            else: raise DuplicateKeyException(constraint, cols, e)
368 2240 aaronmk
369 2368 aaronmk
        match = re.search(r'null value in column "(\w+?)" violates not-null '
370 2239 aaronmk
            r'constraint', msg)
371 2345 aaronmk
        if match: raise NullValueException('NOT NULL', [match.group(1)], e)
372 2240 aaronmk
373 2381 aaronmk
        match = re.search(r'\b(?:invalid input (?:syntax|value)\b.*?'
374
            r'|date/time field value out of range): "(.+?)"\n'
375 2368 aaronmk
            r'(?:(?s).*?)\bfunction "(\w+?)".*?\bat assignment', msg)
376 2240 aaronmk
        if match:
377 2373 aaronmk
            value, name = match.groups()
378
            raise FunctionValueException(name, strings.to_unicode(value), e)
379 2240 aaronmk
380 2368 aaronmk
        match = re.search(r'relation "(\w+?)" already exists', msg)
381 2148 aaronmk
        if match: raise DuplicateTableException(match.group(1), e)
382 2240 aaronmk
383 2368 aaronmk
        match = re.search(r'function "(\w+?)" already exists', msg)
384 2188 aaronmk
        if match: raise DuplicateFunctionException(match.group(1), e)
385 2240 aaronmk
386 2148 aaronmk
        raise # no specific exception raised
387 830 aaronmk
388 832 aaronmk
##### Basic queries
389
390 2153 aaronmk
def next_version(name):
391
    '''Prepends the version # so it won't be removed if the name is truncated'''
392 2163 aaronmk
    version = 1 # first existing name was version 0
393 2153 aaronmk
    match = re.match(r'^v(\d+)_(.*)$', name)
394
    if match:
395
        version = int(match.group(1))+1
396
        name = match.group(2)
397
    return 'v'+str(version)+'_'+name
398
399 2386 aaronmk
def run_query_into(db, query, params, into=None, *args, **kw_args):
400 2085 aaronmk
    '''Outputs a query to a temp table.
401
    For params, see run_query().
402
    '''
403 2386 aaronmk
    if into == None: return run_query(db, query, params, *args, **kw_args)
404 2085 aaronmk
    else: # place rows in temp table
405 2386 aaronmk
        assert isinstance(into, sql_gen.Table)
406 2385 aaronmk
407 2153 aaronmk
        kw_args['recover'] = True
408 2440 aaronmk
409
        temp = not db.debug # tables are created as permanent in debug mode
410
        # "temporary tables cannot specify a schema name", so remove schema
411
        if temp: into.schema = None
412
413 2153 aaronmk
        while True:
414
            try:
415 2194 aaronmk
                create_query = 'CREATE'
416 2440 aaronmk
                if temp: create_query += ' TEMP'
417 2386 aaronmk
                create_query += ' TABLE '+into.to_str(db)+' AS '+query
418 2194 aaronmk
419
                return run_query(db, create_query, params, *args, **kw_args)
420 2153 aaronmk
                    # CREATE TABLE AS sets rowcount to # rows in query
421
            except DuplicateTableException, e:
422 2386 aaronmk
                into.name = next_version(into.name)
423 2153 aaronmk
                # try again with next version of name
424 2085 aaronmk
425 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
426
427 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
428
429 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
430 2293 aaronmk
    start=None, order_by=order_by_pkey, default_table=None):
431 1981 aaronmk
    '''
432 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
433 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
434 1981 aaronmk
    @param fields Use None to select all fields in the table
435 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
436 2379 aaronmk
        * container can be any iterable type
437 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
438
        * compare_right_side: sql_gen.ValueCond|literal value
439 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
440
        use all columns
441 2054 aaronmk
    @return tuple(query, params)
442 1981 aaronmk
    '''
443 2315 aaronmk
    # Parse tables param
444 2121 aaronmk
    if not lists.is_seq(tables): tables = [tables]
445 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
446 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
447 2121 aaronmk
448 2315 aaronmk
    # Parse other params
449 2376 aaronmk
    if conds == None: conds = []
450
    elif isinstance(conds, dict): conds = conds.items()
451 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
452 135 aaronmk
    assert limit == None or type(limit) == int
453 865 aaronmk
    assert start == None or type(start) == int
454 2315 aaronmk
    if order_by is order_by_pkey:
455
        if distinct_on != []: order_by = None
456
        else: order_by = pkey(db, table0, recover=True)
457 865 aaronmk
458 2315 aaronmk
    query = 'SELECT'
459 2056 aaronmk
460 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
461 2056 aaronmk
462 2200 aaronmk
    # DISTINCT ON columns
463 2233 aaronmk
    if distinct_on != []:
464 2200 aaronmk
        query += ' DISTINCT'
465 2254 aaronmk
        if distinct_on is not distinct_on_all:
466 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
467
468
    # Columns
469
    query += ' '
470 1135 aaronmk
    if fields == None: query += '*'
471 2056 aaronmk
    else: query += ', '.join(map(parse_col, fields))
472 2200 aaronmk
473
    # Main table
474 2271 aaronmk
    query += ' FROM '+table0.to_str(db)
475 865 aaronmk
476 2122 aaronmk
    # Add joins
477 2271 aaronmk
    left_table = table0
478 2263 aaronmk
    for join_ in tables:
479
        table = join_.table
480 2238 aaronmk
481 2343 aaronmk
        # Parse special values
482
        if join_.type_ is sql_gen.filter_out: # filter no match
483 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
484
                None))
485 2343 aaronmk
486 2277 aaronmk
        query += ' '+join_.to_str(db, left_table)
487 2122 aaronmk
488
        left_table = table
489
490 865 aaronmk
    missing = True
491 2376 aaronmk
    if conds != []:
492 2410 aaronmk
        query += ' WHERE '+(' AND '.join(('('+sql_gen.ColValueCond(l, r)
493
            .to_str(db)+')' for l, r in conds)))
494 865 aaronmk
        missing = False
495 2227 aaronmk
    if order_by != None:
496 2285 aaronmk
        query += ' ORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
497 865 aaronmk
    if limit != None: query += ' LIMIT '+str(limit); missing = False
498
    if start != None:
499
        if start != 0: query += ' OFFSET '+str(start)
500
        missing = False
501
    if missing: warnings.warn(DbWarning(
502
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
503
504 2315 aaronmk
    return (query, [])
505 11 aaronmk
506 2054 aaronmk
def select(db, *args, **kw_args):
507
    '''For params, see mk_select() and run_query()'''
508
    recover = kw_args.pop('recover', None)
509
    cacheable = kw_args.pop('cacheable', True)
510 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
511 2054 aaronmk
512
    query, params = mk_select(db, *args, **kw_args)
513 2442 aaronmk
    return run_query(db, query, params, recover, cacheable, log_level=log_level)
514 2054 aaronmk
515 2066 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, params=None,
516 2292 aaronmk
    returning=None, embeddable=False):
517 1960 aaronmk
    '''
518
    @param returning str|None An inserted column (such as pkey) to return
519 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
520 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
521
        query will be fully cached, not just if it raises an exception.
522 1960 aaronmk
    '''
523 2328 aaronmk
    table = sql_gen.as_Table(table)
524 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
525
    if cols != None: cols = [sql_gen.as_Col(v).to_str(db) for v in cols]
526 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
527 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
528 2063 aaronmk
529
    # Build query
530 2328 aaronmk
    query = 'INSERT INTO '+table.to_str(db)
531 2318 aaronmk
    if cols != None: query += ' ('+', '.join(cols)+')'
532 2063 aaronmk
    query += ' '+select_query
533
534
    if returning != None:
535 2327 aaronmk
        returning_name = copy.copy(returning)
536
        returning_name.table = None
537
        returning_name = returning_name.to_str(db)
538
        query += ' RETURNING '+returning_name
539 2063 aaronmk
540 2070 aaronmk
    if embeddable:
541 2327 aaronmk
        assert returning != None
542
543 2070 aaronmk
        # Create function
544 2330 aaronmk
        function_name = '_'.join(['insert', table.name] + cols)
545 2327 aaronmk
        return_type = 'SETOF '+returning.to_str(db)+'%TYPE'
546 2189 aaronmk
        while True:
547
            try:
548 2327 aaronmk
                func_schema = None
549
                if not db.debug: func_schema = 'pg_temp'
550
                function = sql_gen.Table(function_name, func_schema).to_str(db)
551 2194 aaronmk
552 2189 aaronmk
                function_query = '''\
553
CREATE FUNCTION '''+function+'''() RETURNS '''+return_type+'''
554 2070 aaronmk
    LANGUAGE sql
555
    AS $$'''+mogrify(db, query, params)+''';$$;
556
'''
557 2189 aaronmk
                run_query(db, function_query, recover=True, cacheable=True)
558
                break # this version was successful
559
            except DuplicateFunctionException, e:
560
                function_name = next_version(function_name)
561
                # try again with next version of name
562 2070 aaronmk
563 2337 aaronmk
        # Return query that uses function
564
        func_table = sql_gen.NamedTable('f', sql_gen.CustomCode(function+'()'),
565
            [returning_name]) # AS clause requires function alias
566
        return mk_select(db, func_table, start=0, order_by=None)
567 2070 aaronmk
568 2066 aaronmk
    return (query, params)
569
570
def insert_select(db, *args, **kw_args):
571 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
572 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
573
        values in
574 2072 aaronmk
    '''
575 2386 aaronmk
    into = kw_args.pop('into', None)
576
    if into != None: kw_args['embeddable'] = True
577 2066 aaronmk
    recover = kw_args.pop('recover', None)
578
    cacheable = kw_args.pop('cacheable', True)
579
580
    query, params = mk_insert_select(db, *args, **kw_args)
581 2386 aaronmk
    return run_query_into(db, query, params, into, recover=recover,
582 2153 aaronmk
        cacheable=cacheable)
583 2063 aaronmk
584 2066 aaronmk
default = object() # tells insert() to use the default value for a column
585
586 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
587 2085 aaronmk
    '''For params, see insert_select()'''
588 1960 aaronmk
    if lists.is_seq(row): cols = None
589
    else:
590
        cols = row.keys()
591
        row = row.values()
592
    row = list(row) # ensure that "!= []" works
593
594 1961 aaronmk
    # Check for special values
595
    labels = []
596
    values = []
597
    for value in row:
598 2254 aaronmk
        if value is default: labels.append('DEFAULT')
599 1961 aaronmk
        else:
600
            labels.append('%s')
601
            values.append(value)
602
603
    # Build query
604 2063 aaronmk
    if values != []: query = ' VALUES ('+(', '.join(labels))+')'
605
    else: query = None
606 1554 aaronmk
607 2064 aaronmk
    return insert_select(db, table, cols, query, values, *args, **kw_args)
608 11 aaronmk
609 2402 aaronmk
def mk_update(db, table, changes=None, cond=None):
610
    '''
611
    @param changes [(col, new_value),...]
612
        * container can be any iterable type
613
        * col: sql_gen.Code|str (for col name)
614
        * new_value: sql_gen.Code|literal value
615
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
616
    @return str query
617
    '''
618
    query = 'UPDATE '+sql_gen.as_Table(table).to_str(db)+'\nSET\n'
619 2405 aaronmk
    query += ',\n'.join((sql_gen.to_name_only_col(col, table).to_str(db)+' = '
620 2402 aaronmk
        +sql_gen.as_Value(new_value).to_str(db) for col, new_value in changes))
621
    if cond != None: query += ' WHERE '+cond.to_str(db)
622
623
    return query
624
625
def update(db, *args, **kw_args):
626
    '''For params, see mk_update() and run_query()'''
627
    recover = kw_args.pop('recover', None)
628
629
    return run_query(db, mk_update(db, *args, **kw_args), [], recover)
630
631 135 aaronmk
def last_insert_id(db):
632 1849 aaronmk
    module = util.root_module(db.db)
633 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
634
    elif module == 'MySQLdb': return db.insert_id()
635
    else: return None
636 13 aaronmk
637 1968 aaronmk
def truncate(db, table, schema='public'):
638
    return run_query(db, 'TRUNCATE '+qual_name(db, schema, table)+' CASCADE')
639 832 aaronmk
640 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
641 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
642 2415 aaronmk
    to names that will be distinct among the columns' tables.
643 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
644 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
645
    @param into The table for the new columns.
646 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
647
        columns will be included in the mapping even if they are not in cols.
648
        The tables of the provided Col objects will be changed to into, so make
649
        copies of them if you want to keep the original tables.
650
    @param as_items Whether to return a list of dict items instead of a dict
651 2383 aaronmk
    @return dict(orig_col=new_col, ...)
652
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
653 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
654
        * All mappings use the into table so its name can easily be
655 2383 aaronmk
          changed for all columns at once
656
    '''
657 2415 aaronmk
    cols = lists.uniqify(cols)
658
659 2394 aaronmk
    items = []
660 2389 aaronmk
    for col in preserve:
661 2390 aaronmk
        orig_col = copy.copy(col)
662 2392 aaronmk
        col.table = into
663 2394 aaronmk
        items.append((orig_col, col))
664
    preserve = set(preserve)
665
    for col in cols:
666 2397 aaronmk
        if col not in preserve:
667
            items.append((col, sql_gen.Col(clean_name(str(col)), into)))
668 2394 aaronmk
669
    if not as_items: items = dict(items)
670
    return items
671 2383 aaronmk
672 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
673 2391 aaronmk
    '''For params, see mk_flatten_mapping()
674
    @return See return value of mk_flatten_mapping()
675
    '''
676 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
677
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
678 2391 aaronmk
    run_query_into(db, *mk_select(db, joins, cols, limit=limit, start=start),
679 2392 aaronmk
        into=into)
680 2394 aaronmk
    return dict(items)
681 2391 aaronmk
682 2414 aaronmk
##### Database structure queries
683
684 2426 aaronmk
def table_row_count(db, table, recover=None):
685
    return value(run_query(db, *mk_select(db, table, [sql_gen.row_count],
686 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
687 2426 aaronmk
688 2414 aaronmk
def table_cols(db, table, recover=None):
689
    return list(col_names(select(db, table, limit=0, order_by=None,
690 2443 aaronmk
        recover=recover, log_level=4)))
691 2414 aaronmk
692 2291 aaronmk
def pkey(db, table, recover=None):
693 832 aaronmk
    '''Assumed to be first column in table'''
694 2339 aaronmk
    return table_cols(db, table, recover)[0]
695 832 aaronmk
696 2340 aaronmk
not_null_col = 'not_null'
697
698
def table_not_null_col(db, table, recover=None):
699
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
700
    if not_null_col in table_cols(db, table, recover): return not_null_col
701
    else: return pkey(db, table, recover)
702
703 853 aaronmk
def index_cols(db, table, index):
704
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
705
    automatically created. When you don't know whether something is a UNIQUE
706
    constraint or a UNIQUE index, use this function.'''
707 1909 aaronmk
    module = util.root_module(db.db)
708
    if module == 'psycopg2':
709
        return list(values(run_query(db, '''\
710 853 aaronmk
SELECT attname
711 866 aaronmk
FROM
712
(
713
        SELECT attnum, attname
714
        FROM pg_index
715
        JOIN pg_class index ON index.oid = indexrelid
716
        JOIN pg_class table_ ON table_.oid = indrelid
717
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
718
        WHERE
719
            table_.relname = %(table)s
720
            AND index.relname = %(index)s
721
    UNION
722
        SELECT attnum, attname
723
        FROM
724
        (
725
            SELECT
726
                indrelid
727
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
728
                    AS indkey
729
            FROM pg_index
730
            JOIN pg_class index ON index.oid = indexrelid
731
            JOIN pg_class table_ ON table_.oid = indrelid
732
            WHERE
733
                table_.relname = %(table)s
734
                AND index.relname = %(index)s
735
        ) s
736
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
737
) s
738 853 aaronmk
ORDER BY attnum
739
''',
740 2443 aaronmk
            {'table': table, 'index': index}, cacheable=True, log_level=4)))
741 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
742
        ' database')
743 853 aaronmk
744 464 aaronmk
def constraint_cols(db, table, constraint):
745 1849 aaronmk
    module = util.root_module(db.db)
746 464 aaronmk
    if module == 'psycopg2':
747
        return list(values(run_query(db, '''\
748
SELECT attname
749
FROM pg_constraint
750
JOIN pg_class ON pg_class.oid = conrelid
751
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
752
WHERE
753
    relname = %(table)s
754
    AND conname = %(constraint)s
755
ORDER BY attnum
756
''',
757
            {'table': table, 'constraint': constraint})))
758
    else: raise NotImplementedError("Can't list constraint columns for "+module+
759
        ' database')
760
761 2096 aaronmk
row_num_col = '_row_num'
762
763 2408 aaronmk
def index_col(db, col):
764
    '''Adds an index on a column if it doesn't already exist.'''
765
    assert sql_gen.is_table_col(col)
766
767
    table = col.table
768
    index = sql_gen.as_Table(clean_name(str(col)))
769
    col = sql_gen.to_name_only_col(col)
770
    try: run_query(db, 'CREATE INDEX '+index.to_str(db)+' ON '+table.to_str(db)
771 2443 aaronmk
        +' ('+col.to_str(db)+')', recover=True, cacheable=True, log_level=3)
772 2408 aaronmk
    except DuplicateTableException: pass # index already existed
773
774 2406 aaronmk
def index_pkey(db, table, recover=None):
775
    '''Makes the first column in a table the primary key.
776
    @pre The table must not already have a primary key.
777
    '''
778
    table = sql_gen.as_Table(table)
779
780
    index = sql_gen.as_Table(table.name+'_pkey')
781
    col = sql_gen.to_name_only_col(pkey(db, table, recover))
782
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD CONSTRAINT '
783 2443 aaronmk
        +index.to_str(db)+' PRIMARY KEY('+col.to_str(db)+')', recover=recover,
784
        log_level=3)
785 2406 aaronmk
786 2086 aaronmk
def add_row_num(db, table):
787 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
788
    be the primary key.'''
789 2320 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
790 2096 aaronmk
    run_query(db, 'ALTER TABLE '+table+' ADD COLUMN '+row_num_col
791 2443 aaronmk
        +' serial NOT NULL PRIMARY KEY', log_level=3)
792 2086 aaronmk
793 1968 aaronmk
def tables(db, schema='public', table_like='%'):
794 1849 aaronmk
    module = util.root_module(db.db)
795 1968 aaronmk
    params = {'schema': schema, 'table_like': table_like}
796 832 aaronmk
    if module == 'psycopg2':
797 1968 aaronmk
        return values(run_query(db, '''\
798
SELECT tablename
799
FROM pg_tables
800
WHERE
801
    schemaname = %(schema)s
802
    AND tablename LIKE %(table_like)s
803
ORDER BY tablename
804
''',
805
            params, cacheable=True))
806
    elif module == 'MySQLdb':
807
        return values(run_query(db, 'SHOW TABLES LIKE %(table_like)s', params,
808
            cacheable=True))
809 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
810 830 aaronmk
811 833 aaronmk
##### Database management
812
813 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
814
    '''For kw_args, see tables()'''
815
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
816 833 aaronmk
817 832 aaronmk
##### Heuristic queries
818
819 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
820 1554 aaronmk
    '''Recovers from errors.
821 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
822
    '''
823 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
824
825 471 aaronmk
    try:
826 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
827 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
828
            row_ct_ref[0] += cur.rowcount
829
        return value(cur)
830 471 aaronmk
    except DuplicateKeyException, e:
831 2104 aaronmk
        return value(select(db, table, [pkey_],
832 1069 aaronmk
            util.dict_subset_right_join(row, e.cols), recover=True))
833 471 aaronmk
834 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
835 830 aaronmk
    '''Recovers from errors'''
836 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
837 14 aaronmk
    except StopIteration:
838 40 aaronmk
        if not create: raise
839 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
840 2078 aaronmk
841 2416 aaronmk
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None):
842 2078 aaronmk
    '''Recovers from errors.
843
    Only works under PostgreSQL (uses INSERT RETURNING).
844 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
845
        tables to join with it using the main input table's pkey
846 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
847
        * out_table_col: sql_gen.Col|str
848 2323 aaronmk
        * in_table_col: sql_gen.Col Wrap literal values in a sql_gen.NamedCol
849 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
850 2078 aaronmk
    '''
851 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
852 2312 aaronmk
    for in_table_col in mapping.itervalues():
853
        assert isinstance(in_table_col, sql_gen.Col)
854
855 2329 aaronmk
    temp_prefix = out_table.name
856 2387 aaronmk
    pkeys = sql_gen.Table(temp_prefix+'_pkeys')
857 2131 aaronmk
858 2382 aaronmk
    # Create input joins from list of input tables
859
    in_tables_ = in_tables[:] # don't modify input!
860
    in_tables0 = in_tables_.pop(0) # first table is separate
861 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
862 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
863 2352 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v, {in_pkey: sql_gen.join_same})
864 2382 aaronmk
        for v in in_tables_]
865 2131 aaronmk
866 2395 aaronmk
    db.log_debug('Joining together input tables')
867
    # Place in new table for speed and so don't modify input if values edited
868
    in_table = sql_gen.Table(temp_prefix+'_in')
869
    flatten_cols = filter(sql_gen.is_table_col, mapping.values())
870
    mapping = dicts.join(mapping, flatten(db, in_table, input_joins,
871
        flatten_cols, preserve=[in_pkey_col], start=0))
872
    input_joins = [in_table]
873
874 2279 aaronmk
    out_pkey = pkey(db, out_table, recover=True)
875 2285 aaronmk
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
876 2142 aaronmk
877 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
878 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
879
880 2201 aaronmk
    pkeys_table_exists_ref = [False]
881 2420 aaronmk
    def insert_into_pkeys(joins, cols):
882
        query, params = mk_select(db, joins, cols, order_by=None, start=0)
883 2201 aaronmk
        if pkeys_table_exists_ref[0]:
884 2387 aaronmk
            insert_select(db, pkeys, pkeys_names, query, params)
885 2201 aaronmk
        else:
886 2387 aaronmk
            run_query_into(db, query, params, into=pkeys)
887 2201 aaronmk
            pkeys_table_exists_ref[0] = True
888
889 2429 aaronmk
    limit_ref = [None]
890 2380 aaronmk
    conds = set()
891 2233 aaronmk
    distinct_on = []
892 2325 aaronmk
    def mk_main_select(joins, cols):
893 2429 aaronmk
        return mk_select(db, joins, cols, conds, distinct_on,
894
            limit=limit_ref[0], start=0)
895 2132 aaronmk
896 2309 aaronmk
    def log_exc(e):
897
        db.log_debug('Caught exception: '+exc.str_(e, first_line_only=True))
898 2429 aaronmk
    def remove_all_rows(msg):
899
        warnings.warn(DbWarning(msg))
900
        db.log_debug(msg.partition('\n')[0])
901
        db.log_debug('Returning NULL for all rows')
902
        limit_ref[0] = 0 # just create an empty pkeys table
903 2409 aaronmk
    def ignore(in_col, value):
904
        in_col_str = str(in_col)
905
        db.log_debug('Adding index on '+in_col_str+' to enable fast filtering')
906
        index_col(db, in_col)
907
        db.log_debug('Ignoring rows with '+in_col_str+' = '+repr(value))
908 2403 aaronmk
    def remove_rows(in_col, value):
909 2409 aaronmk
        ignore(in_col, value)
910 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
911
        assert cond not in conds # avoid infinite loops
912 2380 aaronmk
        conds.add(cond)
913 2403 aaronmk
    def invalid2null(in_col, value):
914 2409 aaronmk
        ignore(in_col, value)
915 2403 aaronmk
        update(db, in_table, [(in_col, None)],
916
            sql_gen.ColValueCond(in_col, value))
917 2245 aaronmk
918 2206 aaronmk
    # Do inserts and selects
919 2257 aaronmk
    join_cols = {}
920 2404 aaronmk
    insert_out_pkeys = sql_gen.Table(temp_prefix+'_insert_out_pkeys')
921
    insert_in_pkeys = sql_gen.Table(temp_prefix+'_insert_in_pkeys')
922 2206 aaronmk
    while True:
923 2303 aaronmk
        has_joins = join_cols != {}
924
925 2305 aaronmk
        # Prepare to insert new rows
926 2325 aaronmk
        insert_joins = input_joins[:] # don't modify original!
927 2403 aaronmk
        insert_args = dict(recover=True, cacheable=False)
928 2303 aaronmk
        if has_joins:
929 2317 aaronmk
            distinct_on = [v.to_Col() for v in join_cols.values()]
930 2325 aaronmk
            insert_joins.append(sql_gen.Join(out_table, join_cols,
931
                sql_gen.filter_out))
932
        else:
933 2404 aaronmk
            insert_args.update(dict(returning=out_pkey, into=insert_out_pkeys))
934 2303 aaronmk
935 2305 aaronmk
        db.log_debug('Inserting new rows')
936 2206 aaronmk
        try:
937
            cur = insert_select(db, out_table, mapping.keys(),
938 2325 aaronmk
                *mk_main_select(insert_joins, mapping.values()), **insert_args)
939 2357 aaronmk
            break # insert successful
940 2206 aaronmk
        except DuplicateKeyException, e:
941 2309 aaronmk
            log_exc(e)
942
943 2258 aaronmk
            old_join_cols = join_cols.copy()
944 2334 aaronmk
            join_cols.update(util.dict_subset(mapping, e.cols))
945 2245 aaronmk
            db.log_debug('Ignoring existing rows, comparing on '+str(join_cols))
946 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
947 2230 aaronmk
        except NullValueException, e:
948 2309 aaronmk
            log_exc(e)
949
950 2230 aaronmk
            out_col, = e.cols
951
            try: in_col = mapping[out_col]
952 2356 aaronmk
            except KeyError:
953 2429 aaronmk
                remove_all_rows('Missing mapping for NOT NULL '+out_col)
954 2403 aaronmk
            else: remove_rows(in_col, None)
955 2243 aaronmk
        except FunctionValueException, e:
956 2309 aaronmk
            log_exc(e)
957
958 2344 aaronmk
            assert e.name == out_table.name
959 2243 aaronmk
            out_col = 'value' # assume function param was named "value"
960 2403 aaronmk
            invalid2null(mapping[out_col], e.value)
961 2429 aaronmk
        except DatabaseErrors, e:
962
            log_exc(e)
963
964
            remove_all_rows('No handler for exception: '+exc.str_(e))
965 2358 aaronmk
        # after exception handled, rerun loop with additional constraints
966 2132 aaronmk
967 2357 aaronmk
    if row_ct_ref != None and cur.rowcount >= 0:
968
        row_ct_ref[0] += cur.rowcount
969
970
    if has_joins:
971
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
972
        db.log_debug('Getting output pkeys of existing/inserted rows')
973 2420 aaronmk
        insert_into_pkeys(select_joins, pkeys_cols)
974 2357 aaronmk
    else:
975 2404 aaronmk
        add_row_num(db, insert_out_pkeys) # for joining with input pkeys
976 2357 aaronmk
977
        db.log_debug('Getting input pkeys for rows in insert')
978
        run_query_into(db, *mk_main_select(input_joins, [in_pkey]),
979 2404 aaronmk
            into=insert_in_pkeys)
980
        add_row_num(db, insert_in_pkeys) # for joining with output pkeys
981 2357 aaronmk
982 2428 aaronmk
        assert table_row_count(db, insert_out_pkeys) == table_row_count(db,
983
            insert_in_pkeys)
984
985 2357 aaronmk
        db.log_debug('Joining together output and input pkeys')
986 2404 aaronmk
        pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
987 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
988 2420 aaronmk
        insert_into_pkeys(pkey_joins, pkeys_names)
989 2357 aaronmk
990 2413 aaronmk
    db.log_debug('Adding pkey on returned pkeys table to enable fast joins')
991 2407 aaronmk
    index_pkey(db, pkeys)
992
993 2357 aaronmk
    db.log_debug("Setting missing rows' pkeys to NULL")
994 2387 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(pkeys,
995 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
996
        # must use join_same_not_null or query will take forever
997 2420 aaronmk
    insert_into_pkeys(missing_rows_joins,
998
        [in_pkey_col, sql_gen.NamedCol(out_pkey, None)])
999 2357 aaronmk
1000 2428 aaronmk
    assert table_row_count(db, pkeys) == table_row_count(db, in_table)
1001
1002 2387 aaronmk
    return sql_gen.Col(out_pkey, pkeys)
1003 2115 aaronmk
1004
##### Data cleanup
1005
1006 2290 aaronmk
def cleanup_table(db, table, cols):
1007 2115 aaronmk
    def esc_name_(name): return esc_name(db, name)
1008
1009 2290 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
1010 2115 aaronmk
    cols = map(esc_name_, cols)
1011
1012
    run_query(db, 'UPDATE '+table+' SET\n'+(',\n'.join(('\n'+col
1013
        +' = nullif(nullif(trim(both from '+col+"), %(null0)s), %(null1)s)"
1014
            for col in cols))),
1015
        dict(null0='', null1=r'\N'))