Project

General

Profile

1 11 aaronmk
# Database access
2
3 1869 aaronmk
import copy
4 2127 aaronmk
import operator
5 11 aaronmk
import re
6 865 aaronmk
import warnings
7 11 aaronmk
8 300 aaronmk
import exc
9 1909 aaronmk
import dicts
10 1893 aaronmk
import iters
11 1960 aaronmk
import lists
12 1889 aaronmk
from Proxy import Proxy
13 1872 aaronmk
import rand
14 2217 aaronmk
import sql_gen
15 862 aaronmk
import strings
16 131 aaronmk
import util
17 11 aaronmk
18 832 aaronmk
##### Exceptions
19
20 2804 aaronmk
def get_cur_query(cur, input_query=None):
21 2168 aaronmk
    raw_query = None
22
    if hasattr(cur, 'query'): raw_query = cur.query
23
    elif hasattr(cur, '_last_executed'): raw_query = cur._last_executed
24 2170 aaronmk
25
    if raw_query != None: return raw_query
26 2804 aaronmk
    else: return '[input] '+strings.ustr(input_query)
27 14 aaronmk
28 2170 aaronmk
def _add_cursor_info(e, *args, **kw_args):
29
    '''For params, see get_cur_query()'''
30 2771 aaronmk
    exc.add_msg(e, 'query: '+strings.ustr(get_cur_query(*args, **kw_args)))
31 135 aaronmk
32 300 aaronmk
class DbException(exc.ExceptionWithCause):
33 14 aaronmk
    def __init__(self, msg, cause=None, cur=None):
34 2145 aaronmk
        exc.ExceptionWithCause.__init__(self, msg, cause, cause_newline=True)
35 14 aaronmk
        if cur != None: _add_cursor_info(self, cur)
36
37 2143 aaronmk
class ExceptionWithName(DbException):
38
    def __init__(self, name, cause=None):
39 2484 aaronmk
        DbException.__init__(self, 'for name: '+strings.as_tt(str(name)), cause)
40 2143 aaronmk
        self.name = name
41 360 aaronmk
42 2240 aaronmk
class ExceptionWithNameValue(DbException):
43
    def __init__(self, name, value, cause=None):
44 2484 aaronmk
        DbException.__init__(self, 'for name: '+strings.as_tt(str(name))
45
            +'; value: '+strings.as_tt(repr(value)), cause)
46 2240 aaronmk
        self.name = name
47
        self.value = value
48
49 2945 aaronmk
class ExceptionWithNameType(DbException):
50
    def __init__(self, type_, name, cause=None):
51
        DbException.__init__(self, 'for type: '+strings.as_tt(str(type_))
52
            +'; name: '+strings.as_tt(name), cause)
53
        self.type = type_
54
        self.name = name
55
56 2306 aaronmk
class ConstraintException(DbException):
57
    def __init__(self, name, cols, cause=None):
58 2484 aaronmk
        DbException.__init__(self, 'Violated '+strings.as_tt(name)
59
            +' constraint on columns: '+strings.as_tt(', '.join(cols)), cause)
60 2306 aaronmk
        self.name = name
61 468 aaronmk
        self.cols = cols
62 11 aaronmk
63 2523 aaronmk
class MissingCastException(DbException):
64
    def __init__(self, type_, col, cause=None):
65
        DbException.__init__(self, 'Missing cast to type '+strings.as_tt(type_)
66
            +' on column: '+strings.as_tt(col), cause)
67
        self.type = type_
68
        self.col = col
69
70 2143 aaronmk
class NameException(DbException): pass
71
72 2306 aaronmk
class DuplicateKeyException(ConstraintException): pass
73 13 aaronmk
74 2306 aaronmk
class NullValueException(ConstraintException): pass
75 13 aaronmk
76 2240 aaronmk
class FunctionValueException(ExceptionWithNameValue): pass
77 2239 aaronmk
78 2945 aaronmk
class DuplicateException(ExceptionWithNameType): pass
79 2143 aaronmk
80 89 aaronmk
class EmptyRowException(DbException): pass
81
82 865 aaronmk
##### Warnings
83
84
class DbWarning(UserWarning): pass
85
86 1930 aaronmk
##### Result retrieval
87
88
def col_names(cur): return (col[0] for col in cur.description)
89
90
def rows(cur): return iter(lambda: cur.fetchone(), None)
91
92
def consume_rows(cur):
93
    '''Used to fetch all rows so result will be cached'''
94
    iters.consume_iter(rows(cur))
95
96
def next_row(cur): return rows(cur).next()
97
98
def row(cur):
99
    row_ = next_row(cur)
100
    consume_rows(cur)
101
    return row_
102
103
def next_value(cur): return next_row(cur)[0]
104
105
def value(cur): return row(cur)[0]
106
107
def values(cur): return iters.func_iter(lambda: next_value(cur))
108
109
def value_or_none(cur):
110
    try: return value(cur)
111
    except StopIteration: return None
112
113 2762 aaronmk
##### Escaping
114 2101 aaronmk
115 2573 aaronmk
def esc_name_by_module(module, name):
116
    if module == 'psycopg2' or module == None: quote = '"'
117 2101 aaronmk
    elif module == 'MySQLdb': quote = '`'
118
    else: raise NotImplementedError("Can't escape name for "+module+' database')
119 2500 aaronmk
    return sql_gen.esc_name(name, quote)
120 2101 aaronmk
121
def esc_name_by_engine(engine, name, **kw_args):
122
    return esc_name_by_module(db_engines[engine][0], name, **kw_args)
123
124
def esc_name(db, name, **kw_args):
125
    return esc_name_by_module(util.root_module(db.db), name, **kw_args)
126
127
def qual_name(db, schema, table):
128
    def esc_name_(name): return esc_name(db, name)
129
    table = esc_name_(table)
130
    if schema != None: return esc_name_(schema)+'.'+table
131
    else: return table
132
133 1869 aaronmk
##### Database connections
134 1849 aaronmk
135 2097 aaronmk
db_config_names = ['engine', 'host', 'user', 'password', 'database', 'schemas']
136 1926 aaronmk
137 1869 aaronmk
db_engines = {
138
    'MySQL': ('MySQLdb', {'password': 'passwd', 'database': 'db'}),
139
    'PostgreSQL': ('psycopg2', {}),
140
}
141
142
DatabaseErrors_set = set([DbException])
143
DatabaseErrors = tuple(DatabaseErrors_set)
144
145
def _add_module(module):
146
    DatabaseErrors_set.add(module.DatabaseError)
147
    global DatabaseErrors
148
    DatabaseErrors = tuple(DatabaseErrors_set)
149
150
def db_config_str(db_config):
151
    return db_config['engine']+' database '+db_config['database']
152
153 2448 aaronmk
log_debug_none = lambda msg, level=2: None
154 1901 aaronmk
155 1849 aaronmk
class DbConn:
156 2923 aaronmk
    def __init__(self, db_config, autocommit=True, caching=True,
157 2915 aaronmk
        log_debug=log_debug_none, debug_temp=False):
158
        '''
159
        @param debug_temp Whether temporary objects should instead be permanent.
160
            This assists in debugging the internal objects used by the program.
161
        '''
162 1869 aaronmk
        self.db_config = db_config
163 2190 aaronmk
        self.autocommit = autocommit
164
        self.caching = caching
165 1901 aaronmk
        self.log_debug = log_debug
166 2193 aaronmk
        self.debug = log_debug != log_debug_none
167 2915 aaronmk
        self.debug_temp = debug_temp
168 1869 aaronmk
169
        self.__db = None
170 1889 aaronmk
        self.query_results = {}
171 2139 aaronmk
        self._savepoint = 0
172 2671 aaronmk
        self._notices_seen = set()
173 1869 aaronmk
174
    def __getattr__(self, name):
175
        if name == '__dict__': raise Exception('getting __dict__')
176
        if name == 'db': return self._db()
177
        else: raise AttributeError()
178
179
    def __getstate__(self):
180
        state = copy.copy(self.__dict__) # shallow copy
181 1915 aaronmk
        state['log_debug'] = None # don't pickle the debug callback
182 1869 aaronmk
        state['_DbConn__db'] = None # don't pickle the connection
183
        return state
184
185 2165 aaronmk
    def connected(self): return self.__db != None
186
187 1869 aaronmk
    def _db(self):
188
        if self.__db == None:
189
            # Process db_config
190
            db_config = self.db_config.copy() # don't modify input!
191 2097 aaronmk
            schemas = db_config.pop('schemas', None)
192 1869 aaronmk
            module_name, mappings = db_engines[db_config.pop('engine')]
193
            module = __import__(module_name)
194
            _add_module(module)
195
            for orig, new in mappings.iteritems():
196
                try: util.rename_key(db_config, orig, new)
197
                except KeyError: pass
198
199
            # Connect
200
            self.__db = module.connect(**db_config)
201
202
            # Configure connection
203 2906 aaronmk
            if hasattr(self.db, 'set_isolation_level'):
204
                import psycopg2.extensions
205
                self.db.set_isolation_level(
206
                    psycopg2.extensions.ISOLATION_LEVEL_READ_COMMITTED)
207 2101 aaronmk
            if schemas != None:
208 2893 aaronmk
                search_path = [self.esc_name(s) for s in schemas.split(',')]
209
                search_path.append(value(run_query(self, 'SHOW search_path',
210
                    log_level=4)))
211
                run_query(self, 'SET search_path TO '+(','.join(search_path)),
212
                    log_level=3)
213 1869 aaronmk
214
        return self.__db
215 1889 aaronmk
216 1891 aaronmk
    class DbCursor(Proxy):
217 1927 aaronmk
        def __init__(self, outer):
218 1891 aaronmk
            Proxy.__init__(self, outer.db.cursor())
219 2191 aaronmk
            self.outer = outer
220 1927 aaronmk
            self.query_results = outer.query_results
221 1894 aaronmk
            self.query_lookup = None
222 1891 aaronmk
            self.result = []
223 1889 aaronmk
224 2802 aaronmk
        def execute(self, query):
225 2764 aaronmk
            self._is_insert = query.startswith('INSERT')
226 2797 aaronmk
            self.query_lookup = query
227 2148 aaronmk
            try:
228 2191 aaronmk
                try:
229 2802 aaronmk
                    cur = self.inner.execute(query)
230 2191 aaronmk
                    self.outer.do_autocommit()
231 2802 aaronmk
                finally: self.query = get_cur_query(self.inner, query)
232 1904 aaronmk
            except Exception, e:
233 2802 aaronmk
                _add_cursor_info(e, self, query)
234 1904 aaronmk
                self.result = e # cache the exception as the result
235
                self._cache_result()
236
                raise
237 2800 aaronmk
            if self.rowcount == 0 and query.startswith('SELECT'): # empty SELECT
238
                consume_rows(self) # fetch all rows so result will be cached
239 2762 aaronmk
            return cur
240 1894 aaronmk
241 1891 aaronmk
        def fetchone(self):
242
            row = self.inner.fetchone()
243 1899 aaronmk
            if row != None: self.result.append(row)
244
            # otherwise, fetched all rows
245 1904 aaronmk
            else: self._cache_result()
246
            return row
247
248
        def _cache_result(self):
249 1906 aaronmk
            # For inserts, only cache exceptions since inserts are not
250
            # idempotent, but an invalid insert will always be invalid
251 1930 aaronmk
            if self.query_results != None and (not self._is_insert
252 1906 aaronmk
                or isinstance(self.result, Exception)):
253
254 1894 aaronmk
                assert self.query_lookup != None
255 1916 aaronmk
                self.query_results[self.query_lookup] = self.CacheCursor(
256
                    util.dict_subset(dicts.AttrsDictView(self),
257
                    ['query', 'result', 'rowcount', 'description']))
258 1906 aaronmk
259 1916 aaronmk
        class CacheCursor:
260
            def __init__(self, cached_result): self.__dict__ = cached_result
261
262 1927 aaronmk
            def execute(self, *args, **kw_args):
263 1916 aaronmk
                if isinstance(self.result, Exception): raise self.result
264
                # otherwise, result is a rows list
265
                self.iter = iter(self.result)
266
267
            def fetchone(self):
268
                try: return self.iter.next()
269
                except StopIteration: return None
270 1891 aaronmk
271 2212 aaronmk
    def esc_value(self, value):
272 2663 aaronmk
        try: str_ = self.mogrify('%s', [value])
273
        except NotImplementedError, e:
274
            module = util.root_module(self.db)
275
            if module == 'MySQLdb':
276
                import _mysql
277
                str_ = _mysql.escape_string(value)
278
            else: raise e
279 2374 aaronmk
        return strings.to_unicode(str_)
280 2212 aaronmk
281 2347 aaronmk
    def esc_name(self, name): return esc_name(self, name) # calls global func
282
283 2814 aaronmk
    def std_code(self, str_):
284
        '''Standardizes SQL code.
285
        * Ensures that string literals are prefixed by `E`
286
        '''
287
        if str_.startswith("'"): str_ = 'E'+str_
288
        return str_
289
290 2665 aaronmk
    def can_mogrify(self):
291 2663 aaronmk
        module = util.root_module(self.db)
292 2665 aaronmk
        return module == 'psycopg2'
293 2663 aaronmk
294 2665 aaronmk
    def mogrify(self, query, params=None):
295
        if self.can_mogrify(): return self.db.cursor().mogrify(query, params)
296
        else: raise NotImplementedError("Can't mogrify query")
297
298 2671 aaronmk
    def print_notices(self):
299 2725 aaronmk
        if hasattr(self.db, 'notices'):
300
            for msg in self.db.notices:
301
                if msg not in self._notices_seen:
302
                    self._notices_seen.add(msg)
303
                    self.log_debug(msg, level=2)
304 2671 aaronmk
305 2793 aaronmk
    def run_query(self, query, cacheable=False, log_level=2,
306 2464 aaronmk
        debug_msg_ref=None):
307 2445 aaronmk
        '''
308 2464 aaronmk
        @param log_ignore_excs The log_level will be increased by 2 if the query
309
            throws one of these exceptions.
310 2664 aaronmk
        @param debug_msg_ref If specified, the log message will be returned in
311
            this instead of being output. This allows you to filter log messages
312
            depending on the result of the query.
313 2445 aaronmk
        '''
314 2167 aaronmk
        assert query != None
315
316 2047 aaronmk
        if not self.caching: cacheable = False
317 1903 aaronmk
        used_cache = False
318 2664 aaronmk
319
        def log_msg(query):
320
            if used_cache: cache_status = 'cache hit'
321
            elif cacheable: cache_status = 'cache miss'
322
            else: cache_status = 'non-cacheable'
323
            return 'DB query: '+cache_status+':\n'+strings.as_code(query, 'SQL')
324
325 1903 aaronmk
        try:
326 1927 aaronmk
            # Get cursor
327
            if cacheable:
328
                try:
329 2797 aaronmk
                    cur = self.query_results[query]
330 1927 aaronmk
                    used_cache = True
331
                except KeyError: cur = self.DbCursor(self)
332
            else: cur = self.db.cursor()
333
334 2664 aaronmk
            # Log query
335
            if self.debug and debug_msg_ref == None: # log before running
336
                self.log_debug(log_msg(query), log_level)
337
338 1927 aaronmk
            # Run query
339 2793 aaronmk
            cur.execute(query)
340 1903 aaronmk
        finally:
341 2671 aaronmk
            self.print_notices()
342 2664 aaronmk
            if self.debug and debug_msg_ref != None: # return after running
343 2793 aaronmk
                debug_msg_ref[0] = log_msg(str(get_cur_query(cur, query)))
344 1903 aaronmk
345
        return cur
346 1914 aaronmk
347 2797 aaronmk
    def is_cached(self, query): return query in self.query_results
348 2139 aaronmk
349 2907 aaronmk
    def with_autocommit(self, func):
350 2801 aaronmk
        import psycopg2.extensions
351
352
        prev_isolation_level = self.db.isolation_level
353 2907 aaronmk
        self.db.set_isolation_level(
354
            psycopg2.extensions.ISOLATION_LEVEL_AUTOCOMMIT)
355 2683 aaronmk
        try: return func()
356 2801 aaronmk
        finally: self.db.set_isolation_level(prev_isolation_level)
357 2683 aaronmk
358 2139 aaronmk
    def with_savepoint(self, func):
359 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
360 2443 aaronmk
        self.run_query('SAVEPOINT '+savepoint, log_level=4)
361 2139 aaronmk
        self._savepoint += 1
362 2930 aaronmk
        try: return func()
363 2139 aaronmk
        except:
364 2443 aaronmk
            self.run_query('ROLLBACK TO SAVEPOINT '+savepoint, log_level=4)
365 2139 aaronmk
            raise
366 2930 aaronmk
        finally:
367
            # Always release savepoint, because after ROLLBACK TO SAVEPOINT,
368
            # "The savepoint remains valid and can be rolled back to again"
369
            # (http://www.postgresql.org/docs/8.3/static/sql-rollback-to.html).
370 2443 aaronmk
            self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
371 2930 aaronmk
372
            self._savepoint -= 1
373
            assert self._savepoint >= 0
374
375
            self.do_autocommit() # OK to do this after ROLLBACK TO SAVEPOINT
376 2191 aaronmk
377
    def do_autocommit(self):
378
        '''Autocommits if outside savepoint'''
379
        assert self._savepoint >= 0
380
        if self.autocommit and self._savepoint == 0:
381 2924 aaronmk
            self.log_debug('Autocommitting', level=4)
382 2191 aaronmk
            self.db.commit()
383 2643 aaronmk
384 2819 aaronmk
    def col_info(self, col):
385 2643 aaronmk
        table = sql_gen.Table('columns', 'information_schema')
386 2821 aaronmk
        cols = ['data_type', 'column_default',
387
            cast(self, 'boolean', 'is_nullable')]
388 2643 aaronmk
389
        conds = [('table_name', col.table.name), ('column_name', col.name)]
390
        schema = col.table.schema
391
        if schema != None: conds.append(('table_schema', schema))
392
393 2819 aaronmk
        type_, default, nullable = row(select(self, table, cols, conds,
394 2874 aaronmk
            order_by='table_schema', limit=1, log_level=4))
395 2643 aaronmk
            # TODO: order_by search_path schema order
396 2819 aaronmk
        default = sql_gen.as_Code(default, self)
397
398
        return sql_gen.TypedCol(col.name, type_, default, nullable)
399 2917 aaronmk
400
    def TempFunction(self, name):
401
        if self.debug_temp: schema = None
402
        else: schema = 'pg_temp'
403
        return sql_gen.Function(name, schema)
404 1849 aaronmk
405 1869 aaronmk
connect = DbConn
406
407 832 aaronmk
##### Recoverable querying
408 15 aaronmk
409 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
410 11 aaronmk
411 2791 aaronmk
def run_query(db, query, recover=None, cacheable=False, log_level=2,
412
    log_ignore_excs=None, **kw_args):
413 2794 aaronmk
    '''For params, see DbConn.run_query()'''
414 830 aaronmk
    if recover == None: recover = False
415 2464 aaronmk
    if log_ignore_excs == None: log_ignore_excs = ()
416
    log_ignore_excs = tuple(log_ignore_excs)
417 830 aaronmk
418 2666 aaronmk
    debug_msg_ref = None # usually, db.run_query() logs query before running it
419
    # But if filtering with log_ignore_excs, wait until after exception parsing
420
    if log_ignore_excs != () or not db.can_mogrify(): debug_msg_ref = [None]
421
422 2148 aaronmk
    try:
423 2464 aaronmk
        try:
424 2794 aaronmk
            def run(): return db.run_query(query, cacheable, log_level,
425 2793 aaronmk
                debug_msg_ref, **kw_args)
426 2796 aaronmk
            if recover and not db.is_cached(query):
427 2464 aaronmk
                return with_savepoint(db, run)
428
            else: return run() # don't need savepoint if cached
429
        except Exception, e:
430
            if not recover: raise # need savepoint to run index_cols()
431
            msg = exc.str_(e)
432
433
            match = re.search(r'duplicate key value violates unique constraint '
434 2493 aaronmk
                r'"((_?[^\W_]+)_.+?)"', msg)
435 2464 aaronmk
            if match:
436
                constraint, table = match.groups()
437
                try: cols = index_cols(db, table, constraint)
438
                except NotImplementedError: raise e
439
                else: raise DuplicateKeyException(constraint, cols, e)
440
441 2493 aaronmk
            match = re.search(r'null value in column "(.+?)" violates not-null'
442 2464 aaronmk
                r' constraint', msg)
443
            if match: raise NullValueException('NOT NULL', [match.group(1)], e)
444
445
            match = re.search(r'\b(?:invalid input (?:syntax|value)\b.*?'
446
                r'|date/time field value out of range): "(.+?)"\n'
447 2535 aaronmk
                r'(?:(?s).*?)\bfunction "(.+?)"', msg)
448 2464 aaronmk
            if match:
449
                value, name = match.groups()
450
                raise FunctionValueException(name, strings.to_unicode(value), e)
451
452 2526 aaronmk
            match = re.search(r'column "(.+?)" is of type (.+?) but expression '
453 2523 aaronmk
                r'is of type', msg)
454
            if match:
455
                col, type_ = match.groups()
456
                raise MissingCastException(type_, col, e)
457
458 2945 aaronmk
            match = re.search(r'\b(\S+) "(.+?)" already exists', msg)
459
            if match:
460
                type_, name = match.groups()
461
                raise DuplicateException(type_, name, e)
462 2464 aaronmk
463
            raise # no specific exception raised
464
    except log_ignore_excs:
465
        log_level += 2
466
        raise
467
    finally:
468 2666 aaronmk
        if debug_msg_ref != None and debug_msg_ref[0] != None:
469
            db.log_debug(debug_msg_ref[0], log_level)
470 830 aaronmk
471 832 aaronmk
##### Basic queries
472
473 2153 aaronmk
def next_version(name):
474 2163 aaronmk
    version = 1 # first existing name was version 0
475 2586 aaronmk
    match = re.match(r'^(.*)#(\d+)$', name)
476 2153 aaronmk
    if match:
477 2586 aaronmk
        name, version = match.groups()
478
        version = int(version)+1
479 2932 aaronmk
    return sql_gen.concat(name, '#'+str(version))
480 2153 aaronmk
481 2899 aaronmk
def lock_table(db, table, mode):
482
    table = sql_gen.as_Table(table)
483
    run_query(db, 'LOCK TABLE '+table.to_str(db)+' IN '+mode+' MODE')
484
485 2789 aaronmk
def run_query_into(db, query, into=None, add_indexes_=False, **kw_args):
486 2085 aaronmk
    '''Outputs a query to a temp table.
487
    For params, see run_query().
488
    '''
489 2789 aaronmk
    if into == None: return run_query(db, query, **kw_args)
490 2790 aaronmk
491
    assert isinstance(into, sql_gen.Table)
492
493
    kw_args['recover'] = True
494 2945 aaronmk
    kw_args.setdefault('log_ignore_excs', (DuplicateException,))
495 2790 aaronmk
496 2916 aaronmk
    temp = not db.debug_temp # tables are permanent in debug_temp mode
497 2790 aaronmk
    # "temporary tables cannot specify a schema name", so remove schema
498
    if temp: into.schema = None
499
500
    # Create table
501
    while True:
502
        create_query = 'CREATE'
503
        if temp: create_query += ' TEMP'
504
        create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
505 2385 aaronmk
506 2790 aaronmk
        try:
507
            cur = run_query(db, create_query, **kw_args)
508
                # CREATE TABLE AS sets rowcount to # rows in query
509
            break
510 2945 aaronmk
        except DuplicateException, e:
511 2790 aaronmk
            into.name = next_version(into.name)
512
            # try again with next version of name
513
514
    if add_indexes_: add_indexes(db, into)
515
516
    return cur
517 2085 aaronmk
518 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
519
520 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
521
522 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
523 2293 aaronmk
    start=None, order_by=order_by_pkey, default_table=None):
524 1981 aaronmk
    '''
525 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
526 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
527 1981 aaronmk
    @param fields Use None to select all fields in the table
528 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
529 2379 aaronmk
        * container can be any iterable type
530 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
531
        * compare_right_side: sql_gen.ValueCond|literal value
532 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
533
        use all columns
534 2786 aaronmk
    @return query
535 1981 aaronmk
    '''
536 2315 aaronmk
    # Parse tables param
537 2121 aaronmk
    if not lists.is_seq(tables): tables = [tables]
538 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
539 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
540 2121 aaronmk
541 2315 aaronmk
    # Parse other params
542 2376 aaronmk
    if conds == None: conds = []
543 2650 aaronmk
    elif dicts.is_dict(conds): conds = conds.items()
544 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
545 135 aaronmk
    assert limit == None or type(limit) == int
546 865 aaronmk
    assert start == None or type(start) == int
547 2315 aaronmk
    if order_by is order_by_pkey:
548
        if distinct_on != []: order_by = None
549
        else: order_by = pkey(db, table0, recover=True)
550 865 aaronmk
551 2315 aaronmk
    query = 'SELECT'
552 2056 aaronmk
553 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
554 2056 aaronmk
555 2200 aaronmk
    # DISTINCT ON columns
556 2233 aaronmk
    if distinct_on != []:
557 2467 aaronmk
        query += '\nDISTINCT'
558 2254 aaronmk
        if distinct_on is not distinct_on_all:
559 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
560
561
    # Columns
562 2467 aaronmk
    query += '\n'
563 1135 aaronmk
    if fields == None: query += '*'
564 2765 aaronmk
    else:
565
        assert fields != []
566
        query += '\n, '.join(map(parse_col, fields))
567 2200 aaronmk
568
    # Main table
569 2467 aaronmk
    query += '\nFROM '+table0.to_str(db)
570 865 aaronmk
571 2122 aaronmk
    # Add joins
572 2271 aaronmk
    left_table = table0
573 2263 aaronmk
    for join_ in tables:
574
        table = join_.table
575 2238 aaronmk
576 2343 aaronmk
        # Parse special values
577
        if join_.type_ is sql_gen.filter_out: # filter no match
578 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
579 2853 aaronmk
                sql_gen.CompareCond(None, '~=')))
580 2343 aaronmk
581 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
582 2122 aaronmk
583
        left_table = table
584
585 865 aaronmk
    missing = True
586 2376 aaronmk
    if conds != []:
587 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
588
        else: whitespace = '\n'
589 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
590
            .to_str(db) for l, r in conds], 'WHERE')
591 865 aaronmk
        missing = False
592 2227 aaronmk
    if order_by != None:
593 2467 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
594
    if limit != None: query += '\nLIMIT '+str(limit); missing = False
595 865 aaronmk
    if start != None:
596 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
597 865 aaronmk
        missing = False
598
    if missing: warnings.warn(DbWarning(
599
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
600
601 2786 aaronmk
    return query
602 11 aaronmk
603 2054 aaronmk
def select(db, *args, **kw_args):
604
    '''For params, see mk_select() and run_query()'''
605
    recover = kw_args.pop('recover', None)
606
    cacheable = kw_args.pop('cacheable', True)
607 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
608 2054 aaronmk
609 2791 aaronmk
    return run_query(db, mk_select(db, *args, **kw_args), recover, cacheable,
610
        log_level=log_level)
611 2054 aaronmk
612 2788 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, returning=None,
613
    embeddable=False):
614 1960 aaronmk
    '''
615
    @param returning str|None An inserted column (such as pkey) to return
616 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
617 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
618
        query will be fully cached, not just if it raises an exception.
619 1960 aaronmk
    '''
620 2754 aaronmk
    table = sql_gen.remove_table_rename(sql_gen.as_Table(table))
621 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
622 2646 aaronmk
    if cols != None:
623
        cols = [sql_gen.to_name_only_col(v, table).to_str(db) for v in cols]
624 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
625 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
626 2063 aaronmk
627
    # Build query
628 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
629
    query = first_line
630 2467 aaronmk
    if cols != None: query += '\n('+', '.join(cols)+')'
631
    query += '\n'+select_query
632 2063 aaronmk
633
    if returning != None:
634 2743 aaronmk
        query += '\nRETURNING '+sql_gen.to_name_only_col(returning).to_str(db)
635 2063 aaronmk
636 2070 aaronmk
    if embeddable:
637 2327 aaronmk
        assert returning != None
638
639 2070 aaronmk
        # Create function
640 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
641 2327 aaronmk
        return_type = 'SETOF '+returning.to_str(db)+'%TYPE'
642 2189 aaronmk
        while True:
643
            try:
644 2918 aaronmk
                function = db.TempFunction(function_name)
645 2194 aaronmk
646 2189 aaronmk
                function_query = '''\
647 2698 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''()
648 2467 aaronmk
RETURNS '''+return_type+'''
649
LANGUAGE sql
650
AS $$
651 2788 aaronmk
'''+query+''';
652 2467 aaronmk
$$;
653 2070 aaronmk
'''
654 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
655 2945 aaronmk
                    log_ignore_excs=(DuplicateException,))
656 2189 aaronmk
                break # this version was successful
657 2945 aaronmk
            except DuplicateException, e:
658 2189 aaronmk
                function_name = next_version(function_name)
659
                # try again with next version of name
660 2070 aaronmk
661 2337 aaronmk
        # Return query that uses function
662 2698 aaronmk
        func_table = sql_gen.NamedTable('f', sql_gen.FunctionCall(function),
663 2742 aaronmk
            [returning]) # AS clause requires function alias
664 2787 aaronmk
        return mk_select(db, func_table, start=0, order_by=None)
665 2070 aaronmk
666 2787 aaronmk
    return query
667 2066 aaronmk
668
def insert_select(db, *args, **kw_args):
669 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
670 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
671
        values in
672 2072 aaronmk
    '''
673 2386 aaronmk
    into = kw_args.pop('into', None)
674
    if into != None: kw_args['embeddable'] = True
675 2066 aaronmk
    recover = kw_args.pop('recover', None)
676
    cacheable = kw_args.pop('cacheable', True)
677 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
678 2066 aaronmk
679 2789 aaronmk
    return run_query_into(db, mk_insert_select(db, *args, **kw_args), into,
680 2787 aaronmk
        recover=recover, cacheable=cacheable, log_level=log_level)
681 2063 aaronmk
682 2738 aaronmk
default = sql_gen.default # tells insert() to use the default value for a column
683 2066 aaronmk
684 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
685 2085 aaronmk
    '''For params, see insert_select()'''
686 1960 aaronmk
    if lists.is_seq(row): cols = None
687
    else:
688
        cols = row.keys()
689
        row = row.values()
690 2738 aaronmk
    row = list(row) # ensure that "== []" works
691 1960 aaronmk
692 2738 aaronmk
    if row == []: query = None
693
    else: query = sql_gen.Values(row).to_str(db)
694 1961 aaronmk
695 2788 aaronmk
    return insert_select(db, table, cols, query, *args, **kw_args)
696 11 aaronmk
697 2402 aaronmk
def mk_update(db, table, changes=None, cond=None):
698
    '''
699
    @param changes [(col, new_value),...]
700
        * container can be any iterable type
701
        * col: sql_gen.Code|str (for col name)
702
        * new_value: sql_gen.Code|literal value
703
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
704
    @return str query
705
    '''
706
    query = 'UPDATE '+sql_gen.as_Table(table).to_str(db)+'\nSET\n'
707 2405 aaronmk
    query += ',\n'.join((sql_gen.to_name_only_col(col, table).to_str(db)+' = '
708 2402 aaronmk
        +sql_gen.as_Value(new_value).to_str(db) for col, new_value in changes))
709 2467 aaronmk
    if cond != None: query += '\nWHERE\n'+cond.to_str(db)
710 2402 aaronmk
711
    return query
712
713
def update(db, *args, **kw_args):
714
    '''For params, see mk_update() and run_query()'''
715
    recover = kw_args.pop('recover', None)
716
717 2791 aaronmk
    return run_query(db, mk_update(db, *args, **kw_args), recover)
718 2402 aaronmk
719 135 aaronmk
def last_insert_id(db):
720 1849 aaronmk
    module = util.root_module(db.db)
721 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
722
    elif module == 'MySQLdb': return db.insert_id()
723
    else: return None
724 13 aaronmk
725 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
726 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
727 2415 aaronmk
    to names that will be distinct among the columns' tables.
728 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
729 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
730
    @param into The table for the new columns.
731 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
732
        columns will be included in the mapping even if they are not in cols.
733
        The tables of the provided Col objects will be changed to into, so make
734
        copies of them if you want to keep the original tables.
735
    @param as_items Whether to return a list of dict items instead of a dict
736 2383 aaronmk
    @return dict(orig_col=new_col, ...)
737
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
738 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
739
        * All mappings use the into table so its name can easily be
740 2383 aaronmk
          changed for all columns at once
741
    '''
742 2415 aaronmk
    cols = lists.uniqify(cols)
743
744 2394 aaronmk
    items = []
745 2389 aaronmk
    for col in preserve:
746 2390 aaronmk
        orig_col = copy.copy(col)
747 2392 aaronmk
        col.table = into
748 2394 aaronmk
        items.append((orig_col, col))
749
    preserve = set(preserve)
750
    for col in cols:
751 2716 aaronmk
        if col not in preserve:
752
            items.append((col, sql_gen.Col(str(col), into, col.srcs)))
753 2394 aaronmk
754
    if not as_items: items = dict(items)
755
    return items
756 2383 aaronmk
757 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
758 2391 aaronmk
    '''For params, see mk_flatten_mapping()
759
    @return See return value of mk_flatten_mapping()
760
    '''
761 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
762
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
763 2786 aaronmk
    run_query_into(db, mk_select(db, joins, cols, limit=limit, start=start),
764 2846 aaronmk
        into=into, add_indexes_=True)
765 2394 aaronmk
    return dict(items)
766 2391 aaronmk
767 2751 aaronmk
def mk_track_data_error(db, errors_table, cols, value, error_code, error):
768 2810 aaronmk
    assert cols != ()
769
770 2751 aaronmk
    cols = map(sql_gen.to_name_only_col, cols)
771
772
    columns_cols = ['column']
773
    columns = sql_gen.NamedValues('columns', columns_cols,
774
        [[c.name] for c in cols])
775
    values_cols = ['value', 'error_code', 'error']
776
    values = sql_gen.NamedValues('values', values_cols,
777
        [value, error_code, error])
778
779
    select_cols = columns_cols+values_cols
780
    name_only_cols = map(sql_gen.to_name_only_col, select_cols)
781 2755 aaronmk
    errors_table = sql_gen.NamedTable('errors', errors_table)
782 2751 aaronmk
    joins = [columns, sql_gen.Join(values, type_='CROSS'),
783
        sql_gen.Join(errors_table, dict(zip(name_only_cols, select_cols)),
784
        sql_gen.filter_out)]
785
786
    return mk_insert_select(db, errors_table, name_only_cols,
787 2787 aaronmk
        mk_select(db, joins, select_cols, order_by=None))
788 2751 aaronmk
789 2810 aaronmk
def track_data_error(db, errors_table, cols, *args, **kw_args):
790 2758 aaronmk
    '''
791
    @param errors_table If None, does nothing.
792
    '''
793 2810 aaronmk
    if errors_table == None or cols == (): return
794
    run_query(db, mk_track_data_error(db, errors_table, cols, *args, **kw_args),
795 2758 aaronmk
        cacheable=True, log_level=4)
796
797 2704 aaronmk
def cast(db, type_, col, errors_table=None):
798
    '''Casts an (unrenamed) column or value.
799 2769 aaronmk
    If errors_table set and col has srcs, saves errors in errors_table (using
800
    col's srcs attr as the source columns) and converts errors to warnings.
801 2820 aaronmk
    @param col str|sql_gen.Col|sql_gen.Literal
802 2704 aaronmk
    @param errors_table None|sql_gen.Table|str
803
    '''
804 2820 aaronmk
    col = sql_gen.as_Col(col)
805 2769 aaronmk
    save_errors = (errors_table != None and isinstance(col, sql_gen.Col)
806
        and col.srcs != ())
807
    if not save_errors: # can't save errors
808
        return sql_gen.CustomCode(col.to_str(db)+'::'+type_) # just cast
809 2704 aaronmk
810
    assert not isinstance(col, sql_gen.NamedCol)
811
812 2770 aaronmk
    errors_table = sql_gen.as_Table(errors_table)
813
    srcs = map(sql_gen.to_name_only_col, col.srcs)
814
    function_name = str(sql_gen.FunctionCall(type_, *srcs))
815 2918 aaronmk
    function = db.TempFunction(function_name)
816 2704 aaronmk
817 2718 aaronmk
    while True:
818
        # Create function definition
819
        query = '''\
820 2704 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''(value text)
821
RETURNS '''+type_+'''
822
LANGUAGE plpgsql
823
STRICT
824
AS $$
825
BEGIN
826
    /* The explicit cast to the return type is needed to make the cast happen
827
    inside the try block. (Implicit casts to the return type happen at the end
828
    of the function, outside any block.) */
829
    RETURN value::'''+type_+''';
830
EXCEPTION
831
    WHEN data_exception THEN
832
        -- Save error in errors table.
833 2751 aaronmk
        -- Insert the value and error for *each* source column.
834
'''+mk_track_data_error(db, errors_table, srcs,
835
    *map(sql_gen.CustomCode, ['value', 'SQLSTATE', 'SQLERRM']))+''';
836 2704 aaronmk
837
        RAISE WARNING '%', SQLERRM;
838
        RETURN NULL;
839
END;
840
$$;
841
'''
842 2718 aaronmk
843
        # Create function
844
        try:
845
            run_query(db, query, recover=True, cacheable=True,
846 2945 aaronmk
                log_ignore_excs=(DuplicateException,))
847 2718 aaronmk
            break # successful
848 2945 aaronmk
        except DuplicateException:
849 2770 aaronmk
            function.name = next_version(function.name)
850
            # try again with next version of name
851 2704 aaronmk
852
    return sql_gen.FunctionCall(function, col)
853
854 2414 aaronmk
##### Database structure queries
855
856 2426 aaronmk
def table_row_count(db, table, recover=None):
857 2786 aaronmk
    return value(run_query(db, mk_select(db, table, [sql_gen.row_count],
858 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
859 2426 aaronmk
860 2414 aaronmk
def table_cols(db, table, recover=None):
861
    return list(col_names(select(db, table, limit=0, order_by=None,
862 2443 aaronmk
        recover=recover, log_level=4)))
863 2414 aaronmk
864 2291 aaronmk
def pkey(db, table, recover=None):
865 832 aaronmk
    '''Assumed to be first column in table'''
866 2339 aaronmk
    return table_cols(db, table, recover)[0]
867 832 aaronmk
868 2559 aaronmk
not_null_col = 'not_null_col'
869 2340 aaronmk
870
def table_not_null_col(db, table, recover=None):
871
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
872
    if not_null_col in table_cols(db, table, recover): return not_null_col
873
    else: return pkey(db, table, recover)
874
875 853 aaronmk
def index_cols(db, table, index):
876
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
877
    automatically created. When you don't know whether something is a UNIQUE
878
    constraint or a UNIQUE index, use this function.'''
879 1909 aaronmk
    module = util.root_module(db.db)
880
    if module == 'psycopg2':
881
        return list(values(run_query(db, '''\
882 853 aaronmk
SELECT attname
883 866 aaronmk
FROM
884
(
885
        SELECT attnum, attname
886
        FROM pg_index
887
        JOIN pg_class index ON index.oid = indexrelid
888
        JOIN pg_class table_ ON table_.oid = indrelid
889
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
890
        WHERE
891 2782 aaronmk
            table_.relname = '''+db.esc_value(table)+'''
892
            AND index.relname = '''+db.esc_value(index)+'''
893 866 aaronmk
    UNION
894
        SELECT attnum, attname
895
        FROM
896
        (
897
            SELECT
898
                indrelid
899
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
900
                    AS indkey
901
            FROM pg_index
902
            JOIN pg_class index ON index.oid = indexrelid
903
            JOIN pg_class table_ ON table_.oid = indrelid
904
            WHERE
905 2782 aaronmk
                table_.relname = '''+db.esc_value(table)+'''
906
                AND index.relname = '''+db.esc_value(index)+'''
907 866 aaronmk
        ) s
908
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
909
) s
910 853 aaronmk
ORDER BY attnum
911 2782 aaronmk
'''
912
            , cacheable=True, log_level=4)))
913 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
914
        ' database')
915 853 aaronmk
916 464 aaronmk
def constraint_cols(db, table, constraint):
917 1849 aaronmk
    module = util.root_module(db.db)
918 464 aaronmk
    if module == 'psycopg2':
919
        return list(values(run_query(db, '''\
920
SELECT attname
921
FROM pg_constraint
922
JOIN pg_class ON pg_class.oid = conrelid
923
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
924
WHERE
925 2783 aaronmk
    relname = '''+db.esc_value(table)+'''
926
    AND conname = '''+db.esc_value(constraint)+'''
927 464 aaronmk
ORDER BY attnum
928 2783 aaronmk
'''
929
            )))
930 464 aaronmk
    else: raise NotImplementedError("Can't list constraint columns for "+module+
931
        ' database')
932
933 2096 aaronmk
row_num_col = '_row_num'
934
935 2847 aaronmk
def add_index(db, exprs, table=None, unique=False, ensure_not_null=True):
936 2688 aaronmk
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
937 2538 aaronmk
    Currently, only function calls are supported as expressions.
938 2847 aaronmk
    @param ensure_not_null If set, translates NULL values to sentinel values.
939
        This allows indexes to be used for comparisons where NULLs are equal.
940 2538 aaronmk
    '''
941 2688 aaronmk
    if not lists.is_seq(exprs): exprs = [exprs]
942 2538 aaronmk
943 2688 aaronmk
    # Parse exprs
944
    old_exprs = exprs[:]
945
    exprs = []
946
    cols = []
947
    for i, expr in enumerate(old_exprs):
948
        expr = copy.deepcopy(expr) # don't modify input!
949 2823 aaronmk
        expr = sql_gen.as_Col(expr, table)
950 2688 aaronmk
951 2823 aaronmk
        # Handle nullable columns
952 2860 aaronmk
        if ensure_not_null:
953
            try: expr = sql_gen.ensure_not_null(db, expr)
954
            except KeyError: pass # unknown type, so just create plain index
955 2823 aaronmk
956 2688 aaronmk
        # Extract col
957
        if isinstance(expr, sql_gen.FunctionCall):
958
            col = expr.args[0]
959
            expr = sql_gen.Expr(expr)
960
        else: col = expr
961 2823 aaronmk
        assert isinstance(col, sql_gen.Col)
962 2688 aaronmk
963
        # Extract table
964
        if table == None:
965
            assert sql_gen.is_table_col(col)
966
            table = col.table
967
968
        col.table = None
969
970
        exprs.append(expr)
971
        cols.append(col)
972 2408 aaronmk
973 2688 aaronmk
    table = sql_gen.as_Table(table)
974
    index = sql_gen.Table(str(sql_gen.Col(','.join(map(str, cols)), table)))
975
976
    str_ = 'CREATE'
977
    if unique: str_ += ' UNIQUE'
978
    str_ += ' INDEX '+index.to_str(db)+' ON '+table.to_str(db)+' ('+(
979
        ', '.join((v.to_str(db) for v in exprs)))+')'
980
981
    try: run_query(db, str_, recover=True, cacheable=True, log_level=3)
982 2945 aaronmk
    except DuplicateException: pass # index already existed
983 2408 aaronmk
984 2686 aaronmk
def add_pkey(db, table, cols=None, recover=None):
985
    '''Adds a primary key.
986
    @param cols [sql_gen.Col,...] The columns in the primary key.
987
        Defaults to the first column in the table.
988 2406 aaronmk
    @pre The table must not already have a primary key.
989
    '''
990
    table = sql_gen.as_Table(table)
991 2686 aaronmk
    if cols == None: cols = [pkey(db, table, recover)]
992
    col_strs = [sql_gen.to_name_only_col(v).to_str(db) for v in cols]
993 2406 aaronmk
994 2781 aaronmk
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD PRIMARY KEY ('
995
        +(', '.join(col_strs))+')', recover=True, cacheable=True, log_level=3,
996 2945 aaronmk
        log_ignore_excs=(DuplicateException,))
997 2406 aaronmk
998 2772 aaronmk
already_indexed = object() # tells add_indexes() the pkey has already been added
999
1000
def add_indexes(db, table, has_pkey=True):
1001
    '''Adds an index on all columns in a table.
1002
    @param has_pkey bool|already_indexed Whether a pkey instead of a regular
1003
        index should be added on the first column.
1004
        * If already_indexed, the pkey is assumed to have already been added
1005
    '''
1006
    cols = table_cols(db, table)
1007
    if has_pkey:
1008
        if has_pkey is not already_indexed: add_pkey(db, table)
1009
        cols = cols[1:]
1010
    for col in cols: add_index(db, col, table)
1011
1012 2935 aaronmk
def add_col(db, table, col, **kw_args):
1013 2934 aaronmk
    assert isinstance(col, sql_gen.TypedCol)
1014 2935 aaronmk
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD COLUMN '+col.to_str(db),
1015
        **kw_args)
1016 2934 aaronmk
1017 2935 aaronmk
row_num_typed_col = sql_gen.TypedCol(row_num_col, 'serial', nullable=False,
1018
    constraints='PRIMARY KEY')
1019
1020 2086 aaronmk
def add_row_num(db, table):
1021 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
1022
    be the primary key.'''
1023 2935 aaronmk
    add_col(db, table, row_num_typed_col, log_level=3)
1024 2086 aaronmk
1025 2936 aaronmk
def cast_temp_col(db, type_, col, errors_table=None):
1026
    '''Like cast(), but creates a new column with the cast values if the input
1027
    is a column.
1028
    @return The new column or cast value
1029
    '''
1030
    def cast_(col): return cast(db, type_, col, errors_table)
1031
1032
    try: col = sql_gen.underlying_col(col)
1033
    except sql_gen.NoUnderlyingTableException: return sql_gen.wrap(cast_, col)
1034
1035
    table = col.table
1036
    new_col = sql_gen.Col(sql_gen.concat(col.name, '::'+type_), table, col.srcs)
1037
    expr = cast_(col)
1038
    add_col(db, table, sql_gen.TypedCol(new_col.name, type_))
1039
    update(db, table, [(new_col, expr)])
1040
1041
    return new_col
1042
1043 2889 aaronmk
def drop_table(db, table):
1044
    table = sql_gen.as_Table(table)
1045
    return run_query(db, 'DROP TABLE IF EXISTS '+table.to_str(db)+' CASCADE')
1046
1047 2760 aaronmk
def create_table(db, table, cols, has_pkey=True, col_indexes=True):
1048 2675 aaronmk
    '''Creates a table.
1049 2681 aaronmk
    @param cols [sql_gen.TypedCol,...] The column names and types
1050
    @param has_pkey If set, the first column becomes the primary key.
1051 2760 aaronmk
    @param col_indexes bool|[ref]
1052
        * If True, indexes will be added on all non-pkey columns.
1053
        * If a list reference, [0] will be set to a function to do this.
1054
          This can be used to delay index creation until the table is populated.
1055 2675 aaronmk
    '''
1056
    table = sql_gen.as_Table(table)
1057
1058 2681 aaronmk
    if has_pkey:
1059
        cols[0] = pkey = copy.copy(cols[0]) # don't modify input!
1060 2872 aaronmk
        pkey.constraints = 'PRIMARY KEY'
1061 2681 aaronmk
1062 2675 aaronmk
    str_ = 'CREATE TABLE '+table.to_str(db)+' (\n'
1063 2681 aaronmk
    str_ += '\n, '.join(v.to_str(db) for v in cols)
1064
    str_ += '\n);\n'
1065 2675 aaronmk
    run_query(db, str_, cacheable=True, log_level=2)
1066 2760 aaronmk
1067
    # Add indexes
1068 2773 aaronmk
    if has_pkey: has_pkey = already_indexed
1069
    def add_indexes_(): add_indexes(db, table, has_pkey)
1070
    if isinstance(col_indexes, list): col_indexes[0] = add_indexes_ # defer
1071
    elif col_indexes: add_indexes_() # add now
1072 2675 aaronmk
1073 2684 aaronmk
def vacuum(db, table):
1074
    table = sql_gen.as_Table(table)
1075
    db.with_autocommit(lambda: run_query(db, 'VACUUM ANALYZE '+table.to_str(db),
1076
        log_level=3))
1077
1078 2732 aaronmk
def truncate(db, table, schema='public'):
1079 2777 aaronmk
    table = sql_gen.as_Table(table, schema)
1080 2733 aaronmk
    return run_query(db, 'TRUNCATE '+table.to_str(db)+' CASCADE')
1081 2732 aaronmk
1082 2729 aaronmk
def tables(db, schema_like='public', table_like='%', exact=False):
1083
    if exact: compare = '='
1084
    else: compare = 'LIKE'
1085
1086 1849 aaronmk
    module = util.root_module(db.db)
1087 832 aaronmk
    if module == 'psycopg2':
1088 2779 aaronmk
        conds = [('schemaname', sql_gen.CompareCond(schema_like, compare)),
1089
            ('tablename', sql_gen.CompareCond(table_like, compare))]
1090
        return values(select(db, 'pg_tables', ['tablename'], conds,
1091
            order_by='tablename', log_level=4))
1092 1968 aaronmk
    elif module == 'MySQLdb':
1093 2779 aaronmk
        return values(run_query(db, 'SHOW TABLES LIKE '+db.esc_value(table_like)
1094
            , cacheable=True, log_level=4))
1095 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
1096 830 aaronmk
1097 2726 aaronmk
def table_exists(db, table):
1098
    table = sql_gen.as_Table(table)
1099 2730 aaronmk
    return list(tables(db, table.schema, table.name, exact=True)) != []
1100 2726 aaronmk
1101 2913 aaronmk
def function_exists(db, function):
1102
    function = sql_gen.as_Function(function)
1103
1104
    info_table = sql_gen.Table('routines', 'information_schema')
1105
    conds = [('routine_name', function.name)]
1106
    schema = function.schema
1107
    if schema != None: conds.append(('routine_schema', schema))
1108 2914 aaronmk
    # Exclude trigger functions, since they cannot be called directly
1109
    conds.append(('data_type', sql_gen.CompareCond('trigger', '!=')))
1110 2913 aaronmk
1111
    return list(values(select(db, info_table, ['routine_name'], conds,
1112
        order_by='routine_schema', limit=1, log_level=4))) != []
1113
        # TODO: order_by search_path schema order
1114
1115 2726 aaronmk
def errors_table(db, table, if_exists=True):
1116
    '''
1117
    @param if_exists If set, returns None if the errors table doesn't exist
1118
    @return None|sql_gen.Table
1119
    '''
1120
    table = sql_gen.as_Table(table)
1121
    if table.srcs != (): table = table.srcs[0]
1122
1123
    errors_table = sql_gen.suffixed_table(table, '.errors')
1124
    if if_exists and not table_exists(db, errors_table): return None
1125
    return errors_table
1126
1127 833 aaronmk
##### Database management
1128
1129 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
1130
    '''For kw_args, see tables()'''
1131
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
1132 833 aaronmk
1133 832 aaronmk
##### Heuristic queries
1134
1135 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
1136 1554 aaronmk
    '''Recovers from errors.
1137 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
1138
    '''
1139 2642 aaronmk
    row = sql_gen.ColDict(db, table, row)
1140 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
1141
1142 471 aaronmk
    try:
1143 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
1144 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
1145
            row_ct_ref[0] += cur.rowcount
1146
        return value(cur)
1147 471 aaronmk
    except DuplicateKeyException, e:
1148 2653 aaronmk
        row = sql_gen.ColDict(db, table,
1149
            util.dict_subset_right_join(row, e.cols))
1150
        return value(select(db, table, [pkey_], row, recover=True))
1151 471 aaronmk
1152 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
1153 830 aaronmk
    '''Recovers from errors'''
1154 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
1155 14 aaronmk
    except StopIteration:
1156 40 aaronmk
        if not create: raise
1157 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
1158 2078 aaronmk
1159 2593 aaronmk
def is_func_result(col):
1160
    return col.table.name.find('(') >= 0 and col.name == 'result'
1161
1162 2592 aaronmk
def into_table_name(out_table, in_tables0, mapping, is_func):
1163 2615 aaronmk
    def in_col_str(in_col):
1164
        in_col = sql_gen.remove_col_rename(in_col)
1165
        if isinstance(in_col, sql_gen.Col):
1166
            table = in_col.table
1167
            if table == in_tables0:
1168
                in_col = sql_gen.to_name_only_col(in_col)
1169
            elif is_func_result(in_col): in_col = table # omit col name
1170
        return str(in_col)
1171
1172 2580 aaronmk
    str_ = str(out_table)
1173
    if is_func:
1174 2615 aaronmk
        str_ += '('
1175 2580 aaronmk
1176 2615 aaronmk
        try: value_in_col = mapping['value']
1177
        except KeyError:
1178 2654 aaronmk
            str_ += ', '.join((str(k)+'='+in_col_str(v)
1179 2615 aaronmk
                for k, v in mapping.iteritems()))
1180
        else: str_ += in_col_str(value_in_col)
1181
1182
        str_ += ')'
1183 2812 aaronmk
    else:
1184
        out_col = 'rank'
1185
        try: in_col = mapping[out_col]
1186
        except KeyError: str_ += '_pkeys'
1187
        else: # has a rank column, so hierarchical
1188
            str_ += '['+str(out_col)+'='+in_col_str(in_col)+']'
1189 2580 aaronmk
    return str_
1190
1191 2508 aaronmk
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None, into=None,
1192 2805 aaronmk
    default=None, is_func=False, on_error=exc.raise_):
1193 2078 aaronmk
    '''Recovers from errors.
1194
    Only works under PostgreSQL (uses INSERT RETURNING).
1195 2909 aaronmk
    IMPORTANT: Must be run at the *beginning* of a transaction.
1196 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
1197
        tables to join with it using the main input table's pkey
1198 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
1199 2612 aaronmk
        * out_table_col: str (*not* sql_gen.Col)
1200
        * in_table_col: sql_gen.Col|literal-value
1201 2489 aaronmk
    @param into The table to contain the output and input pkeys.
1202 2574 aaronmk
        Defaults to `out_table.name+'_pkeys'`.
1203 2509 aaronmk
    @param default The *output* column to use as the pkey for missing rows.
1204
        If this output column does not exist in the mapping, uses None.
1205 2552 aaronmk
    @param is_func Whether out_table is the name of a SQL function, not a table
1206 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
1207 2078 aaronmk
    '''
1208 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
1209 2312 aaronmk
1210 2450 aaronmk
    def log_debug(msg): db.log_debug(msg, level=1.5)
1211 2505 aaronmk
    def col_ustr(str_):
1212 2567 aaronmk
        return strings.repr_no_u(sql_gen.remove_col_rename(str_))
1213 2450 aaronmk
1214 2910 aaronmk
    log_debug('********** New iteration **********')
1215
    log_debug('Inserting these input columns into '+strings.as_tt(
1216
        out_table.to_str(db))+':\n'+strings.as_table(mapping, ustr=col_ustr))
1217
1218 2942 aaronmk
    is_function = function_exists(db, out_table)
1219
1220
    if is_function: out_pkey = 'result'
1221
    else: out_pkey = pkey(db, out_table, recover=True)
1222 2768 aaronmk
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
1223
1224
    if mapping == {}: # need at least one column for INSERT SELECT
1225
        mapping = {out_pkey: None} # ColDict will replace with default value
1226
1227 2382 aaronmk
    # Create input joins from list of input tables
1228
    in_tables_ = in_tables[:] # don't modify input!
1229
    in_tables0 = in_tables_.pop(0) # first table is separate
1230 2735 aaronmk
    errors_table_ = errors_table(db, in_tables0)
1231 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
1232 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
1233 2460 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v,
1234
        {in_pkey: sql_gen.join_same_not_null}) for v in in_tables_]
1235 2131 aaronmk
1236 2592 aaronmk
    if into == None:
1237
        into = into_table_name(out_table, in_tables0, mapping, is_func)
1238
    into = sql_gen.as_Table(into)
1239
1240 2702 aaronmk
    # Set column sources
1241
    in_cols = filter(sql_gen.is_table_col, mapping.values())
1242
    for col in in_cols:
1243
        if col.table == in_tables0: col.set_srcs(sql_gen.src_self)
1244
1245 2486 aaronmk
    log_debug('Joining together input tables into temp table')
1246 2395 aaronmk
    # Place in new table for speed and so don't modify input if values edited
1247 2584 aaronmk
    in_table = sql_gen.Table('in')
1248 2702 aaronmk
    mapping = dicts.join(mapping, flatten(db, in_table, input_joins, in_cols,
1249
        preserve=[in_pkey_col], start=0))
1250 2395 aaronmk
    input_joins = [in_table]
1251 2486 aaronmk
    db.log_debug('Temp table: '+strings.as_tt(in_table.to_str(db)), level=2)
1252 2395 aaronmk
1253 2656 aaronmk
    mapping = sql_gen.ColDict(db, out_table, mapping)
1254
        # after applying dicts.join() because that returns a plain dict
1255
1256 2509 aaronmk
    # Resolve default value column
1257 2841 aaronmk
    if default != None:
1258
        try: default = mapping[default]
1259
        except KeyError:
1260 2509 aaronmk
            db.log_debug('Default value column '
1261
                +strings.as_tt(strings.repr_no_u(default))
1262 2511 aaronmk
                +' does not exist in mapping, falling back to None', level=2.1)
1263 2509 aaronmk
            default = None
1264
1265 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
1266 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
1267
1268 2201 aaronmk
    pkeys_table_exists_ref = [False]
1269 2420 aaronmk
    def insert_into_pkeys(joins, cols):
1270 2786 aaronmk
        query = mk_select(db, joins, cols, order_by=None, start=0)
1271 2201 aaronmk
        if pkeys_table_exists_ref[0]:
1272 2786 aaronmk
            insert_select(db, into, pkeys_names, query)
1273 2201 aaronmk
        else:
1274 2786 aaronmk
            run_query_into(db, query, into=into)
1275 2201 aaronmk
            pkeys_table_exists_ref[0] = True
1276
1277 2429 aaronmk
    limit_ref = [None]
1278 2380 aaronmk
    conds = set()
1279 2811 aaronmk
    distinct_on = sql_gen.ColDict(db, out_table)
1280 2325 aaronmk
    def mk_main_select(joins, cols):
1281 2811 aaronmk
        distinct_on_cols = [c.to_Col() for c in distinct_on.values()]
1282
        return mk_select(db, joins, cols, conds, distinct_on_cols,
1283 2429 aaronmk
            limit=limit_ref[0], start=0)
1284 2132 aaronmk
1285 2519 aaronmk
    exc_strs = set()
1286 2309 aaronmk
    def log_exc(e):
1287 2519 aaronmk
        e_str = exc.str_(e, first_line_only=True)
1288
        log_debug('Caught exception: '+e_str)
1289
        assert e_str not in exc_strs # avoid infinite loops
1290
        exc_strs.add(e_str)
1291 2735 aaronmk
1292 2451 aaronmk
    def remove_all_rows():
1293 2896 aaronmk
        log_debug('Ignoring all rows')
1294 2429 aaronmk
        limit_ref[0] = 0 # just create an empty pkeys table
1295 2735 aaronmk
1296 2758 aaronmk
    def ignore(in_col, value, e):
1297
        track_data_error(db, errors_table_, in_col.srcs, value, e.cause.pgcode,
1298
            e.cause.pgerror)
1299 2895 aaronmk
        log_debug('Ignoring rows with '+strings.as_tt(repr(in_col))+' = '
1300 2545 aaronmk
            +strings.as_tt(repr(value)))
1301 2895 aaronmk
1302 2758 aaronmk
    def remove_rows(in_col, value, e):
1303
        ignore(in_col, value, e)
1304 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
1305
        assert cond not in conds # avoid infinite loops
1306 2380 aaronmk
        conds.add(cond)
1307 2895 aaronmk
1308 2758 aaronmk
    def invalid2null(in_col, value, e):
1309
        ignore(in_col, value, e)
1310 2403 aaronmk
        update(db, in_table, [(in_col, None)],
1311
            sql_gen.ColValueCond(in_col, value))
1312 2245 aaronmk
1313 2589 aaronmk
    def insert_pkeys_table(which):
1314 2932 aaronmk
        return sql_gen.Table(sql_gen.concat(in_table.name,
1315 2589 aaronmk
            '_insert_'+which+'_pkeys'))
1316
    insert_out_pkeys = insert_pkeys_table('out')
1317
    insert_in_pkeys = insert_pkeys_table('in')
1318
1319 2206 aaronmk
    # Do inserts and selects
1320 2642 aaronmk
    join_cols = sql_gen.ColDict(db, out_table)
1321 2206 aaronmk
    while True:
1322 2521 aaronmk
        if limit_ref[0] == 0: # special case
1323
            log_debug('Creating an empty pkeys table')
1324 2786 aaronmk
            cur = run_query_into(db, mk_select(db, out_table, [out_pkey],
1325 2521 aaronmk
                limit=limit_ref[0]), into=insert_out_pkeys)
1326
            break # don't do main case
1327
1328 2303 aaronmk
        has_joins = join_cols != {}
1329
1330 2903 aaronmk
        log_debug('Trying to insert new rows')
1331
1332 2305 aaronmk
        # Prepare to insert new rows
1333 2325 aaronmk
        insert_joins = input_joins[:] # don't modify original!
1334 2403 aaronmk
        insert_args = dict(recover=True, cacheable=False)
1335 2303 aaronmk
        if has_joins:
1336 2325 aaronmk
            insert_joins.append(sql_gen.Join(out_table, join_cols,
1337
                sql_gen.filter_out))
1338
        else:
1339 2404 aaronmk
            insert_args.update(dict(returning=out_pkey, into=insert_out_pkeys))
1340 2786 aaronmk
        main_select = mk_main_select(insert_joins, mapping.values())
1341 2303 aaronmk
1342 2929 aaronmk
        def main_insert():
1343 2942 aaronmk
            if is_function:
1344
                log_debug('Calling function on input rows')
1345
                args = dict(((k.name, v) for k, v in mapping.iteritems()))
1346
                func_call = sql_gen.NamedCol(out_pkey,
1347
                    sql_gen.FunctionCall(out_table, **args))
1348
                insert_into_pkeys(input_joins, [in_pkey_col, func_call])
1349
                return None
1350
            else:
1351
                lock_table(db, out_table, 'EXCLUSIVE')
1352
                return insert_select(db, out_table, mapping.keys(), main_select,
1353
                    **insert_args)
1354 2929 aaronmk
1355 2206 aaronmk
        try:
1356 2929 aaronmk
            cur = with_savepoint(db, main_insert)
1357 2357 aaronmk
            break # insert successful
1358 2931 aaronmk
        except MissingCastException, e:
1359
            log_exc(e)
1360
1361
            out_col = e.col
1362
            type_ = e.type
1363
1364
            log_debug('Casting '+strings.as_tt(out_col)+' input to '
1365
                +strings.as_tt(type_))
1366 2937 aaronmk
            mapping[out_col] = cast_temp_col(db, type_, mapping[out_col],
1367
                errors_table_)
1368 2206 aaronmk
        except DuplicateKeyException, e:
1369 2309 aaronmk
            log_exc(e)
1370
1371 2258 aaronmk
            old_join_cols = join_cols.copy()
1372 2811 aaronmk
            distinct_on.update(util.dict_subset(mapping, e.cols))
1373 2565 aaronmk
            join_cols.update(util.dict_subset_right_join(mapping, e.cols))
1374 2486 aaronmk
            log_debug('Ignoring existing rows, comparing on these columns:\n'
1375 2505 aaronmk
                +strings.as_inline_table(join_cols, ustr=col_ustr))
1376 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
1377 2230 aaronmk
        except NullValueException, e:
1378 2309 aaronmk
            log_exc(e)
1379
1380 2230 aaronmk
            out_col, = e.cols
1381
            try: in_col = mapping[out_col]
1382 2356 aaronmk
            except KeyError:
1383 2486 aaronmk
                log_debug('Missing mapping for NOT NULL column '+out_col)
1384 2451 aaronmk
                remove_all_rows()
1385 2758 aaronmk
            else: remove_rows(in_col, None, e)
1386 2542 aaronmk
        except FunctionValueException, e:
1387
            log_exc(e)
1388
1389
            func_name = e.name
1390
            value = e.value
1391
            for out_col, in_col in mapping.iteritems():
1392 2758 aaronmk
                in_col = sql_gen.unwrap_func_call(in_col, func_name)
1393
                invalid2null(in_col, value, e)
1394 2429 aaronmk
        except DatabaseErrors, e:
1395
            log_exc(e)
1396
1397 2808 aaronmk
            log_debug('No handler for exception')
1398
            on_error(e)
1399 2451 aaronmk
            remove_all_rows()
1400 2358 aaronmk
        # after exception handled, rerun loop with additional constraints
1401 2132 aaronmk
1402 2942 aaronmk
    if cur != None and row_ct_ref != None and cur.rowcount >= 0:
1403 2357 aaronmk
        row_ct_ref[0] += cur.rowcount
1404
1405 2942 aaronmk
    if is_function: pass # pkeys table already created
1406
    elif has_joins:
1407 2357 aaronmk
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
1408 2486 aaronmk
        log_debug('Getting output table pkeys of existing/inserted rows')
1409 2420 aaronmk
        insert_into_pkeys(select_joins, pkeys_cols)
1410 2357 aaronmk
    else:
1411 2404 aaronmk
        add_row_num(db, insert_out_pkeys) # for joining with input pkeys
1412 2357 aaronmk
1413 2486 aaronmk
        log_debug('Getting input table pkeys of inserted rows')
1414 2786 aaronmk
        run_query_into(db, mk_main_select(input_joins, [in_pkey]),
1415 2404 aaronmk
            into=insert_in_pkeys)
1416
        add_row_num(db, insert_in_pkeys) # for joining with output pkeys
1417 2357 aaronmk
1418 2428 aaronmk
        assert table_row_count(db, insert_out_pkeys) == table_row_count(db,
1419
            insert_in_pkeys)
1420
1421 2486 aaronmk
        log_debug('Combining output and input pkeys in inserted order')
1422 2404 aaronmk
        pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
1423 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
1424 2420 aaronmk
        insert_into_pkeys(pkey_joins, pkeys_names)
1425 2357 aaronmk
1426 2486 aaronmk
    db.log_debug('Adding pkey on pkeys table to enable fast joins', level=2.5)
1427 2594 aaronmk
    add_pkey(db, into)
1428 2407 aaronmk
1429 2508 aaronmk
    log_debug('Setting pkeys of missing rows to '+strings.as_tt(repr(default)))
1430 2489 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(into,
1431 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
1432
        # must use join_same_not_null or query will take forever
1433 2420 aaronmk
    insert_into_pkeys(missing_rows_joins,
1434 2508 aaronmk
        [in_pkey_col, sql_gen.NamedCol(out_pkey, default)])
1435 2357 aaronmk
1436 2489 aaronmk
    assert table_row_count(db, into) == table_row_count(db, in_table)
1437 2428 aaronmk
1438 2702 aaronmk
    srcs = []
1439
    if is_func: srcs = sql_gen.cols_srcs(in_cols)
1440
    return sql_gen.Col(out_pkey, into, srcs)
1441 2115 aaronmk
1442
##### Data cleanup
1443
1444 2290 aaronmk
def cleanup_table(db, table, cols):
1445 2677 aaronmk
    table = sql_gen.as_Table(table)
1446
    cols = map(sql_gen.as_Col, cols)
1447 2115 aaronmk
1448 2677 aaronmk
    expr = ('nullif(nullif(trim(both from %s), '+db.esc_value('')+'), '
1449
        +db.esc_value(r'\N')+')')
1450
    changes = [(v, sql_gen.CustomCode(expr % v.to_str(db)))
1451
        for v in cols]
1452 2115 aaronmk
1453 2677 aaronmk
    update(db, table, changes)