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 3004 aaronmk
238
            # Always cache certain queries
239
            if query.startswith('CREATE') or query.startswith('ALTER'):
240
                self._cache_result() # structural changes
241
            elif self.rowcount == 0 and query.startswith('SELECT'): # empty
242 2800 aaronmk
                consume_rows(self) # fetch all rows so result will be cached
243 3004 aaronmk
244 2762 aaronmk
            return cur
245 1894 aaronmk
246 1891 aaronmk
        def fetchone(self):
247
            row = self.inner.fetchone()
248 1899 aaronmk
            if row != None: self.result.append(row)
249
            # otherwise, fetched all rows
250 1904 aaronmk
            else: self._cache_result()
251
            return row
252
253
        def _cache_result(self):
254 2948 aaronmk
            # For inserts that return a result set, don't cache result set since
255
            # inserts are not idempotent. Other non-SELECT queries don't have
256
            # their result set read, so only exceptions will be cached (an
257
            # invalid query will always be invalid).
258 1930 aaronmk
            if self.query_results != None and (not self._is_insert
259 1906 aaronmk
                or isinstance(self.result, Exception)):
260
261 1894 aaronmk
                assert self.query_lookup != None
262 1916 aaronmk
                self.query_results[self.query_lookup] = self.CacheCursor(
263
                    util.dict_subset(dicts.AttrsDictView(self),
264
                    ['query', 'result', 'rowcount', 'description']))
265 1906 aaronmk
266 1916 aaronmk
        class CacheCursor:
267
            def __init__(self, cached_result): self.__dict__ = cached_result
268
269 1927 aaronmk
            def execute(self, *args, **kw_args):
270 1916 aaronmk
                if isinstance(self.result, Exception): raise self.result
271
                # otherwise, result is a rows list
272
                self.iter = iter(self.result)
273
274
            def fetchone(self):
275
                try: return self.iter.next()
276
                except StopIteration: return None
277 1891 aaronmk
278 2212 aaronmk
    def esc_value(self, value):
279 2663 aaronmk
        try: str_ = self.mogrify('%s', [value])
280
        except NotImplementedError, e:
281
            module = util.root_module(self.db)
282
            if module == 'MySQLdb':
283
                import _mysql
284
                str_ = _mysql.escape_string(value)
285
            else: raise e
286 2374 aaronmk
        return strings.to_unicode(str_)
287 2212 aaronmk
288 2347 aaronmk
    def esc_name(self, name): return esc_name(self, name) # calls global func
289
290 2814 aaronmk
    def std_code(self, str_):
291
        '''Standardizes SQL code.
292
        * Ensures that string literals are prefixed by `E`
293
        '''
294
        if str_.startswith("'"): str_ = 'E'+str_
295
        return str_
296
297 2665 aaronmk
    def can_mogrify(self):
298 2663 aaronmk
        module = util.root_module(self.db)
299 2665 aaronmk
        return module == 'psycopg2'
300 2663 aaronmk
301 2665 aaronmk
    def mogrify(self, query, params=None):
302
        if self.can_mogrify(): return self.db.cursor().mogrify(query, params)
303
        else: raise NotImplementedError("Can't mogrify query")
304
305 2671 aaronmk
    def print_notices(self):
306 2725 aaronmk
        if hasattr(self.db, 'notices'):
307
            for msg in self.db.notices:
308
                if msg not in self._notices_seen:
309
                    self._notices_seen.add(msg)
310
                    self.log_debug(msg, level=2)
311 2671 aaronmk
312 2793 aaronmk
    def run_query(self, query, cacheable=False, log_level=2,
313 2464 aaronmk
        debug_msg_ref=None):
314 2445 aaronmk
        '''
315 2464 aaronmk
        @param log_ignore_excs The log_level will be increased by 2 if the query
316
            throws one of these exceptions.
317 2664 aaronmk
        @param debug_msg_ref If specified, the log message will be returned in
318
            this instead of being output. This allows you to filter log messages
319
            depending on the result of the query.
320 2445 aaronmk
        '''
321 2167 aaronmk
        assert query != None
322
323 2047 aaronmk
        if not self.caching: cacheable = False
324 1903 aaronmk
        used_cache = False
325 2664 aaronmk
326
        def log_msg(query):
327
            if used_cache: cache_status = 'cache hit'
328
            elif cacheable: cache_status = 'cache miss'
329
            else: cache_status = 'non-cacheable'
330
            return 'DB query: '+cache_status+':\n'+strings.as_code(query, 'SQL')
331
332 1903 aaronmk
        try:
333 1927 aaronmk
            # Get cursor
334
            if cacheable:
335
                try:
336 2797 aaronmk
                    cur = self.query_results[query]
337 1927 aaronmk
                    used_cache = True
338
                except KeyError: cur = self.DbCursor(self)
339
            else: cur = self.db.cursor()
340
341 2664 aaronmk
            # Log query
342
            if self.debug and debug_msg_ref == None: # log before running
343
                self.log_debug(log_msg(query), log_level)
344
345 1927 aaronmk
            # Run query
346 2793 aaronmk
            cur.execute(query)
347 1903 aaronmk
        finally:
348 2671 aaronmk
            self.print_notices()
349 2664 aaronmk
            if self.debug and debug_msg_ref != None: # return after running
350 2793 aaronmk
                debug_msg_ref[0] = log_msg(str(get_cur_query(cur, query)))
351 1903 aaronmk
352
        return cur
353 1914 aaronmk
354 2797 aaronmk
    def is_cached(self, query): return query in self.query_results
355 2139 aaronmk
356 2907 aaronmk
    def with_autocommit(self, func):
357 2801 aaronmk
        import psycopg2.extensions
358
359
        prev_isolation_level = self.db.isolation_level
360 2907 aaronmk
        self.db.set_isolation_level(
361
            psycopg2.extensions.ISOLATION_LEVEL_AUTOCOMMIT)
362 2683 aaronmk
        try: return func()
363 2801 aaronmk
        finally: self.db.set_isolation_level(prev_isolation_level)
364 2683 aaronmk
365 2139 aaronmk
    def with_savepoint(self, func):
366 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
367 2443 aaronmk
        self.run_query('SAVEPOINT '+savepoint, log_level=4)
368 2139 aaronmk
        self._savepoint += 1
369 2930 aaronmk
        try: return func()
370 2139 aaronmk
        except:
371 2443 aaronmk
            self.run_query('ROLLBACK TO SAVEPOINT '+savepoint, log_level=4)
372 2139 aaronmk
            raise
373 2930 aaronmk
        finally:
374
            # Always release savepoint, because after ROLLBACK TO SAVEPOINT,
375
            # "The savepoint remains valid and can be rolled back to again"
376
            # (http://www.postgresql.org/docs/8.3/static/sql-rollback-to.html).
377 2443 aaronmk
            self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
378 2930 aaronmk
379
            self._savepoint -= 1
380
            assert self._savepoint >= 0
381
382
            self.do_autocommit() # OK to do this after ROLLBACK TO SAVEPOINT
383 2191 aaronmk
384
    def do_autocommit(self):
385
        '''Autocommits if outside savepoint'''
386
        assert self._savepoint >= 0
387
        if self.autocommit and self._savepoint == 0:
388 2924 aaronmk
            self.log_debug('Autocommitting', level=4)
389 2191 aaronmk
            self.db.commit()
390 2643 aaronmk
391 2819 aaronmk
    def col_info(self, col):
392 2643 aaronmk
        table = sql_gen.Table('columns', 'information_schema')
393 2821 aaronmk
        cols = ['data_type', 'column_default',
394
            cast(self, 'boolean', 'is_nullable')]
395 2643 aaronmk
396
        conds = [('table_name', col.table.name), ('column_name', col.name)]
397
        schema = col.table.schema
398
        if schema != None: conds.append(('table_schema', schema))
399
400 2819 aaronmk
        type_, default, nullable = row(select(self, table, cols, conds,
401 2874 aaronmk
            order_by='table_schema', limit=1, log_level=4))
402 2643 aaronmk
            # TODO: order_by search_path schema order
403 2819 aaronmk
        default = sql_gen.as_Code(default, self)
404
405
        return sql_gen.TypedCol(col.name, type_, default, nullable)
406 2917 aaronmk
407
    def TempFunction(self, name):
408
        if self.debug_temp: schema = None
409
        else: schema = 'pg_temp'
410
        return sql_gen.Function(name, schema)
411 1849 aaronmk
412 1869 aaronmk
connect = DbConn
413
414 832 aaronmk
##### Recoverable querying
415 15 aaronmk
416 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
417 11 aaronmk
418 2791 aaronmk
def run_query(db, query, recover=None, cacheable=False, log_level=2,
419
    log_ignore_excs=None, **kw_args):
420 2794 aaronmk
    '''For params, see DbConn.run_query()'''
421 830 aaronmk
    if recover == None: recover = False
422 2464 aaronmk
    if log_ignore_excs == None: log_ignore_excs = ()
423
    log_ignore_excs = tuple(log_ignore_excs)
424 830 aaronmk
425 2666 aaronmk
    debug_msg_ref = None # usually, db.run_query() logs query before running it
426
    # But if filtering with log_ignore_excs, wait until after exception parsing
427 2984 aaronmk
    if log_ignore_excs != () or not db.can_mogrify(): debug_msg_ref = [None]
428 2666 aaronmk
429 2148 aaronmk
    try:
430 2464 aaronmk
        try:
431 2794 aaronmk
            def run(): return db.run_query(query, cacheable, log_level,
432 2793 aaronmk
                debug_msg_ref, **kw_args)
433 2796 aaronmk
            if recover and not db.is_cached(query):
434 2464 aaronmk
                return with_savepoint(db, run)
435
            else: return run() # don't need savepoint if cached
436
        except Exception, e:
437
            if not recover: raise # need savepoint to run index_cols()
438
            msg = exc.str_(e)
439
440
            match = re.search(r'duplicate key value violates unique constraint '
441 2493 aaronmk
                r'"((_?[^\W_]+)_.+?)"', msg)
442 2464 aaronmk
            if match:
443
                constraint, table = match.groups()
444
                try: cols = index_cols(db, table, constraint)
445
                except NotImplementedError: raise e
446
                else: raise DuplicateKeyException(constraint, cols, e)
447
448 2493 aaronmk
            match = re.search(r'null value in column "(.+?)" violates not-null'
449 2464 aaronmk
                r' constraint', msg)
450
            if match: raise NullValueException('NOT NULL', [match.group(1)], e)
451
452
            match = re.search(r'\b(?:invalid input (?:syntax|value)\b.*?'
453
                r'|date/time field value out of range): "(.+?)"\n'
454 2535 aaronmk
                r'(?:(?s).*?)\bfunction "(.+?)"', msg)
455 2464 aaronmk
            if match:
456
                value, name = match.groups()
457
                raise FunctionValueException(name, strings.to_unicode(value), e)
458
459 2526 aaronmk
            match = re.search(r'column "(.+?)" is of type (.+?) but expression '
460 2523 aaronmk
                r'is of type', msg)
461
            if match:
462
                col, type_ = match.groups()
463
                raise MissingCastException(type_, col, e)
464
465 2946 aaronmk
            match = re.search(r'\b(\S+) "(.+?)".*? already exists', msg)
466 2945 aaronmk
            if match:
467
                type_, name = match.groups()
468
                raise DuplicateException(type_, name, e)
469 2464 aaronmk
470
            raise # no specific exception raised
471
    except log_ignore_excs:
472
        log_level += 2
473
        raise
474
    finally:
475 2666 aaronmk
        if debug_msg_ref != None and debug_msg_ref[0] != None:
476
            db.log_debug(debug_msg_ref[0], log_level)
477 830 aaronmk
478 832 aaronmk
##### Basic queries
479
480 2153 aaronmk
def next_version(name):
481 2163 aaronmk
    version = 1 # first existing name was version 0
482 2586 aaronmk
    match = re.match(r'^(.*)#(\d+)$', name)
483 2153 aaronmk
    if match:
484 2586 aaronmk
        name, version = match.groups()
485
        version = int(version)+1
486 2932 aaronmk
    return sql_gen.concat(name, '#'+str(version))
487 2153 aaronmk
488 2899 aaronmk
def lock_table(db, table, mode):
489
    table = sql_gen.as_Table(table)
490
    run_query(db, 'LOCK TABLE '+table.to_str(db)+' IN '+mode+' MODE')
491
492 2789 aaronmk
def run_query_into(db, query, into=None, add_indexes_=False, **kw_args):
493 2085 aaronmk
    '''Outputs a query to a temp table.
494
    For params, see run_query().
495
    '''
496 2789 aaronmk
    if into == None: return run_query(db, query, **kw_args)
497 2790 aaronmk
498
    assert isinstance(into, sql_gen.Table)
499
500 2992 aaronmk
    into.is_temp = True
501
502 2790 aaronmk
    kw_args['recover'] = True
503 2945 aaronmk
    kw_args.setdefault('log_ignore_excs', (DuplicateException,))
504 2790 aaronmk
505 2916 aaronmk
    temp = not db.debug_temp # tables are permanent in debug_temp mode
506 2790 aaronmk
    # "temporary tables cannot specify a schema name", so remove schema
507
    if temp: into.schema = None
508
509
    # Create table
510
    while True:
511
        create_query = 'CREATE'
512
        if temp: create_query += ' TEMP'
513
        create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
514 2385 aaronmk
515 2790 aaronmk
        try:
516
            cur = run_query(db, create_query, **kw_args)
517
                # CREATE TABLE AS sets rowcount to # rows in query
518
            break
519 2945 aaronmk
        except DuplicateException, e:
520 2790 aaronmk
            into.name = next_version(into.name)
521
            # try again with next version of name
522
523
    if add_indexes_: add_indexes(db, into)
524
525
    return cur
526 2085 aaronmk
527 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
528
529 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
530
531 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
532 2293 aaronmk
    start=None, order_by=order_by_pkey, default_table=None):
533 1981 aaronmk
    '''
534 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
535 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
536 1981 aaronmk
    @param fields Use None to select all fields in the table
537 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
538 2379 aaronmk
        * container can be any iterable type
539 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
540
        * compare_right_side: sql_gen.ValueCond|literal value
541 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
542
        use all columns
543 2786 aaronmk
    @return query
544 1981 aaronmk
    '''
545 2315 aaronmk
    # Parse tables param
546 2964 aaronmk
    tables = lists.mk_seq(tables)
547 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
548 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
549 2121 aaronmk
550 2315 aaronmk
    # Parse other params
551 2376 aaronmk
    if conds == None: conds = []
552 2650 aaronmk
    elif dicts.is_dict(conds): conds = conds.items()
553 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
554 135 aaronmk
    assert limit == None or type(limit) == int
555 865 aaronmk
    assert start == None or type(start) == int
556 2315 aaronmk
    if order_by is order_by_pkey:
557
        if distinct_on != []: order_by = None
558
        else: order_by = pkey(db, table0, recover=True)
559 865 aaronmk
560 2315 aaronmk
    query = 'SELECT'
561 2056 aaronmk
562 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
563 2056 aaronmk
564 2200 aaronmk
    # DISTINCT ON columns
565 2233 aaronmk
    if distinct_on != []:
566 2467 aaronmk
        query += '\nDISTINCT'
567 2254 aaronmk
        if distinct_on is not distinct_on_all:
568 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
569
570
    # Columns
571 2467 aaronmk
    query += '\n'
572 1135 aaronmk
    if fields == None: query += '*'
573 2765 aaronmk
    else:
574
        assert fields != []
575
        query += '\n, '.join(map(parse_col, fields))
576 2200 aaronmk
577
    # Main table
578 2467 aaronmk
    query += '\nFROM '+table0.to_str(db)
579 865 aaronmk
580 2122 aaronmk
    # Add joins
581 2271 aaronmk
    left_table = table0
582 2263 aaronmk
    for join_ in tables:
583
        table = join_.table
584 2238 aaronmk
585 2343 aaronmk
        # Parse special values
586
        if join_.type_ is sql_gen.filter_out: # filter no match
587 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
588 2853 aaronmk
                sql_gen.CompareCond(None, '~=')))
589 2343 aaronmk
590 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
591 2122 aaronmk
592
        left_table = table
593
594 865 aaronmk
    missing = True
595 2376 aaronmk
    if conds != []:
596 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
597
        else: whitespace = '\n'
598 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
599
            .to_str(db) for l, r in conds], 'WHERE')
600 865 aaronmk
        missing = False
601 2227 aaronmk
    if order_by != None:
602 2467 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
603
    if limit != None: query += '\nLIMIT '+str(limit); missing = False
604 865 aaronmk
    if start != None:
605 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
606 865 aaronmk
        missing = False
607
    if missing: warnings.warn(DbWarning(
608
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
609
610 2786 aaronmk
    return query
611 11 aaronmk
612 2054 aaronmk
def select(db, *args, **kw_args):
613
    '''For params, see mk_select() and run_query()'''
614
    recover = kw_args.pop('recover', None)
615
    cacheable = kw_args.pop('cacheable', True)
616 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
617 2054 aaronmk
618 2791 aaronmk
    return run_query(db, mk_select(db, *args, **kw_args), recover, cacheable,
619
        log_level=log_level)
620 2054 aaronmk
621 2788 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, returning=None,
622
    embeddable=False):
623 1960 aaronmk
    '''
624
    @param returning str|None An inserted column (such as pkey) to return
625 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
626 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
627
        query will be fully cached, not just if it raises an exception.
628 1960 aaronmk
    '''
629 2754 aaronmk
    table = sql_gen.remove_table_rename(sql_gen.as_Table(table))
630 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
631 2646 aaronmk
    if cols != None:
632
        cols = [sql_gen.to_name_only_col(v, table).to_str(db) for v in cols]
633 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
634 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
635 2063 aaronmk
636
    # Build query
637 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
638
    query = first_line
639 2467 aaronmk
    if cols != None: query += '\n('+', '.join(cols)+')'
640
    query += '\n'+select_query
641 2063 aaronmk
642
    if returning != None:
643 2743 aaronmk
        query += '\nRETURNING '+sql_gen.to_name_only_col(returning).to_str(db)
644 2063 aaronmk
645 2070 aaronmk
    if embeddable:
646 2327 aaronmk
        assert returning != None
647
648 2070 aaronmk
        # Create function
649 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
650 2327 aaronmk
        return_type = 'SETOF '+returning.to_str(db)+'%TYPE'
651 2189 aaronmk
        while True:
652
            try:
653 2918 aaronmk
                function = db.TempFunction(function_name)
654 2194 aaronmk
655 2189 aaronmk
                function_query = '''\
656 2698 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''()
657 2467 aaronmk
RETURNS '''+return_type+'''
658
LANGUAGE sql
659
AS $$
660 2788 aaronmk
'''+query+''';
661 2467 aaronmk
$$;
662 2070 aaronmk
'''
663 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
664 2945 aaronmk
                    log_ignore_excs=(DuplicateException,))
665 2189 aaronmk
                break # this version was successful
666 2945 aaronmk
            except DuplicateException, e:
667 2189 aaronmk
                function_name = next_version(function_name)
668
                # try again with next version of name
669 2070 aaronmk
670 2337 aaronmk
        # Return query that uses function
671 2698 aaronmk
        func_table = sql_gen.NamedTable('f', sql_gen.FunctionCall(function),
672 2742 aaronmk
            [returning]) # AS clause requires function alias
673 2787 aaronmk
        return mk_select(db, func_table, start=0, order_by=None)
674 2070 aaronmk
675 2787 aaronmk
    return query
676 2066 aaronmk
677
def insert_select(db, *args, **kw_args):
678 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
679 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
680
        values in
681 2072 aaronmk
    '''
682 2386 aaronmk
    into = kw_args.pop('into', None)
683
    if into != None: kw_args['embeddable'] = True
684 2066 aaronmk
    recover = kw_args.pop('recover', None)
685
    cacheable = kw_args.pop('cacheable', True)
686 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
687 2066 aaronmk
688 2789 aaronmk
    return run_query_into(db, mk_insert_select(db, *args, **kw_args), into,
689 2787 aaronmk
        recover=recover, cacheable=cacheable, log_level=log_level)
690 2063 aaronmk
691 2738 aaronmk
default = sql_gen.default # tells insert() to use the default value for a column
692 2066 aaronmk
693 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
694 2085 aaronmk
    '''For params, see insert_select()'''
695 1960 aaronmk
    if lists.is_seq(row): cols = None
696
    else:
697
        cols = row.keys()
698
        row = row.values()
699 2738 aaronmk
    row = list(row) # ensure that "== []" works
700 1960 aaronmk
701 2738 aaronmk
    if row == []: query = None
702
    else: query = sql_gen.Values(row).to_str(db)
703 1961 aaronmk
704 2788 aaronmk
    return insert_select(db, table, cols, query, *args, **kw_args)
705 11 aaronmk
706 2402 aaronmk
def mk_update(db, table, changes=None, cond=None):
707
    '''
708
    @param changes [(col, new_value),...]
709
        * container can be any iterable type
710
        * col: sql_gen.Code|str (for col name)
711
        * new_value: sql_gen.Code|literal value
712
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
713
    @return str query
714
    '''
715
    query = 'UPDATE '+sql_gen.as_Table(table).to_str(db)+'\nSET\n'
716 2405 aaronmk
    query += ',\n'.join((sql_gen.to_name_only_col(col, table).to_str(db)+' = '
717 2402 aaronmk
        +sql_gen.as_Value(new_value).to_str(db) for col, new_value in changes))
718 2467 aaronmk
    if cond != None: query += '\nWHERE\n'+cond.to_str(db)
719 2402 aaronmk
720
    return query
721
722
def update(db, *args, **kw_args):
723
    '''For params, see mk_update() and run_query()'''
724
    recover = kw_args.pop('recover', None)
725
726 2791 aaronmk
    return run_query(db, mk_update(db, *args, **kw_args), recover)
727 2402 aaronmk
728 135 aaronmk
def last_insert_id(db):
729 1849 aaronmk
    module = util.root_module(db.db)
730 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
731
    elif module == 'MySQLdb': return db.insert_id()
732
    else: return None
733 13 aaronmk
734 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
735 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
736 2415 aaronmk
    to names that will be distinct among the columns' tables.
737 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
738 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
739
    @param into The table for the new columns.
740 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
741
        columns will be included in the mapping even if they are not in cols.
742
        The tables of the provided Col objects will be changed to into, so make
743
        copies of them if you want to keep the original tables.
744
    @param as_items Whether to return a list of dict items instead of a dict
745 2383 aaronmk
    @return dict(orig_col=new_col, ...)
746
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
747 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
748
        * All mappings use the into table so its name can easily be
749 2383 aaronmk
          changed for all columns at once
750
    '''
751 2415 aaronmk
    cols = lists.uniqify(cols)
752
753 2394 aaronmk
    items = []
754 2389 aaronmk
    for col in preserve:
755 2390 aaronmk
        orig_col = copy.copy(col)
756 2392 aaronmk
        col.table = into
757 2394 aaronmk
        items.append((orig_col, col))
758
    preserve = set(preserve)
759
    for col in cols:
760 2716 aaronmk
        if col not in preserve:
761
            items.append((col, sql_gen.Col(str(col), into, col.srcs)))
762 2394 aaronmk
763
    if not as_items: items = dict(items)
764
    return items
765 2383 aaronmk
766 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
767 2391 aaronmk
    '''For params, see mk_flatten_mapping()
768
    @return See return value of mk_flatten_mapping()
769
    '''
770 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
771
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
772 2786 aaronmk
    run_query_into(db, mk_select(db, joins, cols, limit=limit, start=start),
773 2846 aaronmk
        into=into, add_indexes_=True)
774 2394 aaronmk
    return dict(items)
775 2391 aaronmk
776 2751 aaronmk
def mk_track_data_error(db, errors_table, cols, value, error_code, error):
777 2810 aaronmk
    assert cols != ()
778
779 2751 aaronmk
    cols = map(sql_gen.to_name_only_col, cols)
780
781
    columns_cols = ['column']
782
    columns = sql_gen.NamedValues('columns', columns_cols,
783
        [[c.name] for c in cols])
784
    values_cols = ['value', 'error_code', 'error']
785
    values = sql_gen.NamedValues('values', values_cols,
786
        [value, error_code, error])
787
788
    select_cols = columns_cols+values_cols
789
    name_only_cols = map(sql_gen.to_name_only_col, select_cols)
790 2755 aaronmk
    errors_table = sql_gen.NamedTable('errors', errors_table)
791 2751 aaronmk
    joins = [columns, sql_gen.Join(values, type_='CROSS'),
792
        sql_gen.Join(errors_table, dict(zip(name_only_cols, select_cols)),
793
        sql_gen.filter_out)]
794
795
    return mk_insert_select(db, errors_table, name_only_cols,
796 2787 aaronmk
        mk_select(db, joins, select_cols, order_by=None))
797 2751 aaronmk
798 2810 aaronmk
def track_data_error(db, errors_table, cols, *args, **kw_args):
799 2758 aaronmk
    '''
800
    @param errors_table If None, does nothing.
801
    '''
802 2810 aaronmk
    if errors_table == None or cols == (): return
803
    run_query(db, mk_track_data_error(db, errors_table, cols, *args, **kw_args),
804 2758 aaronmk
        cacheable=True, log_level=4)
805
806 2704 aaronmk
def cast(db, type_, col, errors_table=None):
807
    '''Casts an (unrenamed) column or value.
808 2769 aaronmk
    If errors_table set and col has srcs, saves errors in errors_table (using
809
    col's srcs attr as the source columns) and converts errors to warnings.
810 2820 aaronmk
    @param col str|sql_gen.Col|sql_gen.Literal
811 2704 aaronmk
    @param errors_table None|sql_gen.Table|str
812
    '''
813 2820 aaronmk
    col = sql_gen.as_Col(col)
814 2769 aaronmk
    save_errors = (errors_table != None and isinstance(col, sql_gen.Col)
815
        and col.srcs != ())
816 2987 aaronmk
    if not save_errors: return sql_gen.Cast(type_, col) # can't save errors
817 2704 aaronmk
818
    assert not isinstance(col, sql_gen.NamedCol)
819
820 2770 aaronmk
    errors_table = sql_gen.as_Table(errors_table)
821
    srcs = map(sql_gen.to_name_only_col, col.srcs)
822
    function_name = str(sql_gen.FunctionCall(type_, *srcs))
823 2918 aaronmk
    function = db.TempFunction(function_name)
824 2704 aaronmk
825 2718 aaronmk
    while True:
826
        # Create function definition
827
        query = '''\
828 2704 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''(value text)
829
RETURNS '''+type_+'''
830
LANGUAGE plpgsql
831
STRICT
832
AS $$
833
BEGIN
834
    /* The explicit cast to the return type is needed to make the cast happen
835
    inside the try block. (Implicit casts to the return type happen at the end
836
    of the function, outside any block.) */
837
    RETURN value::'''+type_+''';
838
EXCEPTION
839
    WHEN data_exception THEN
840
        -- Save error in errors table.
841 2751 aaronmk
        -- Insert the value and error for *each* source column.
842
'''+mk_track_data_error(db, errors_table, srcs,
843
    *map(sql_gen.CustomCode, ['value', 'SQLSTATE', 'SQLERRM']))+''';
844 2704 aaronmk
845
        RAISE WARNING '%', SQLERRM;
846
        RETURN NULL;
847
END;
848
$$;
849
'''
850 2718 aaronmk
851
        # Create function
852
        try:
853
            run_query(db, query, recover=True, cacheable=True,
854 2945 aaronmk
                log_ignore_excs=(DuplicateException,))
855 2718 aaronmk
            break # successful
856 2945 aaronmk
        except DuplicateException:
857 2770 aaronmk
            function.name = next_version(function.name)
858
            # try again with next version of name
859 2704 aaronmk
860
    return sql_gen.FunctionCall(function, col)
861
862 2414 aaronmk
##### Database structure queries
863
864 2426 aaronmk
def table_row_count(db, table, recover=None):
865 2786 aaronmk
    return value(run_query(db, mk_select(db, table, [sql_gen.row_count],
866 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
867 2426 aaronmk
868 2414 aaronmk
def table_cols(db, table, recover=None):
869
    return list(col_names(select(db, table, limit=0, order_by=None,
870 2443 aaronmk
        recover=recover, log_level=4)))
871 2414 aaronmk
872 2291 aaronmk
def pkey(db, table, recover=None):
873 832 aaronmk
    '''Assumed to be first column in table'''
874 2339 aaronmk
    return table_cols(db, table, recover)[0]
875 832 aaronmk
876 2559 aaronmk
not_null_col = 'not_null_col'
877 2340 aaronmk
878
def table_not_null_col(db, table, recover=None):
879
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
880
    if not_null_col in table_cols(db, table, recover): return not_null_col
881
    else: return pkey(db, table, recover)
882
883 853 aaronmk
def index_cols(db, table, index):
884
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
885
    automatically created. When you don't know whether something is a UNIQUE
886
    constraint or a UNIQUE index, use this function.'''
887 1909 aaronmk
    module = util.root_module(db.db)
888
    if module == 'psycopg2':
889
        return list(values(run_query(db, '''\
890 853 aaronmk
SELECT attname
891 866 aaronmk
FROM
892
(
893
        SELECT attnum, attname
894
        FROM pg_index
895
        JOIN pg_class index ON index.oid = indexrelid
896
        JOIN pg_class table_ ON table_.oid = indrelid
897
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
898
        WHERE
899 2782 aaronmk
            table_.relname = '''+db.esc_value(table)+'''
900
            AND index.relname = '''+db.esc_value(index)+'''
901 866 aaronmk
    UNION
902
        SELECT attnum, attname
903
        FROM
904
        (
905
            SELECT
906
                indrelid
907
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
908
                    AS indkey
909
            FROM pg_index
910
            JOIN pg_class index ON index.oid = indexrelid
911
            JOIN pg_class table_ ON table_.oid = indrelid
912
            WHERE
913 2782 aaronmk
                table_.relname = '''+db.esc_value(table)+'''
914
                AND index.relname = '''+db.esc_value(index)+'''
915 866 aaronmk
        ) s
916
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
917
) s
918 853 aaronmk
ORDER BY attnum
919 2782 aaronmk
'''
920
            , cacheable=True, log_level=4)))
921 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
922
        ' database')
923 853 aaronmk
924 464 aaronmk
def constraint_cols(db, table, constraint):
925 1849 aaronmk
    module = util.root_module(db.db)
926 464 aaronmk
    if module == 'psycopg2':
927
        return list(values(run_query(db, '''\
928
SELECT attname
929
FROM pg_constraint
930
JOIN pg_class ON pg_class.oid = conrelid
931
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
932
WHERE
933 2783 aaronmk
    relname = '''+db.esc_value(table)+'''
934
    AND conname = '''+db.esc_value(constraint)+'''
935 464 aaronmk
ORDER BY attnum
936 2783 aaronmk
'''
937
            )))
938 464 aaronmk
    else: raise NotImplementedError("Can't list constraint columns for "+module+
939
        ' database')
940
941 2096 aaronmk
row_num_col = '_row_num'
942
943 2998 aaronmk
def add_index(db, exprs, table=None, unique=False, ensure_not_null_=True):
944 2688 aaronmk
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
945 2538 aaronmk
    Currently, only function calls are supported as expressions.
946 2998 aaronmk
    @param ensure_not_null_ If set, translates NULL values to sentinel values.
947 2847 aaronmk
        This allows indexes to be used for comparisons where NULLs are equal.
948 2538 aaronmk
    '''
949 2964 aaronmk
    exprs = lists.mk_seq(exprs)
950 2538 aaronmk
951 2688 aaronmk
    # Parse exprs
952
    old_exprs = exprs[:]
953
    exprs = []
954
    cols = []
955
    for i, expr in enumerate(old_exprs):
956 2823 aaronmk
        expr = sql_gen.as_Col(expr, table)
957 2688 aaronmk
958 2823 aaronmk
        # Handle nullable columns
959 2998 aaronmk
        if ensure_not_null_:
960
            try: expr = ensure_not_null(db, expr)
961 2860 aaronmk
            except KeyError: pass # unknown type, so just create plain index
962 2823 aaronmk
963 2688 aaronmk
        # Extract col
964 3002 aaronmk
        expr = copy.deepcopy(expr) # don't modify input!
965 2688 aaronmk
        if isinstance(expr, sql_gen.FunctionCall):
966
            col = expr.args[0]
967
            expr = sql_gen.Expr(expr)
968
        else: col = expr
969 2823 aaronmk
        assert isinstance(col, sql_gen.Col)
970 2688 aaronmk
971
        # Extract table
972
        if table == None:
973
            assert sql_gen.is_table_col(col)
974
            table = col.table
975
976
        col.table = None
977
978
        exprs.append(expr)
979
        cols.append(col)
980 2408 aaronmk
981 2688 aaronmk
    table = sql_gen.as_Table(table)
982
    index = sql_gen.Table(str(sql_gen.Col(','.join(map(str, cols)), table)))
983
984 3005 aaronmk
    # Add index
985
    while True:
986
        str_ = 'CREATE'
987
        if unique: str_ += ' UNIQUE'
988
        str_ += ' INDEX '+index.to_str(db)+' ON '+table.to_str(db)+' ('+(
989
            ', '.join((v.to_str(db) for v in exprs)))+')'
990
991
        try:
992
            run_query(db, str_, recover=True, cacheable=True, log_level=3,
993
                log_ignore_excs=(DuplicateException,))
994
            break
995
        except DuplicateException:
996
            index.name = next_version(index.name)
997
            # try again with next version of name
998 2408 aaronmk
999 2686 aaronmk
def add_pkey(db, table, cols=None, recover=None):
1000
    '''Adds a primary key.
1001
    @param cols [sql_gen.Col,...] The columns in the primary key.
1002
        Defaults to the first column in the table.
1003 2406 aaronmk
    @pre The table must not already have a primary key.
1004
    '''
1005
    table = sql_gen.as_Table(table)
1006 2686 aaronmk
    if cols == None: cols = [pkey(db, table, recover)]
1007
    col_strs = [sql_gen.to_name_only_col(v).to_str(db) for v in cols]
1008 2406 aaronmk
1009 2781 aaronmk
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD PRIMARY KEY ('
1010
        +(', '.join(col_strs))+')', recover=True, cacheable=True, log_level=3,
1011 2945 aaronmk
        log_ignore_excs=(DuplicateException,))
1012 2406 aaronmk
1013 2997 aaronmk
def add_not_null(db, col):
1014
    table = col.table
1015
    col = sql_gen.to_name_only_col(col)
1016
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ALTER COLUMN '
1017
        +col.to_str(db)+' SET NOT NULL', cacheable=True)
1018
1019
def add_index_col(db, col, suffix, expr, nullable=True):
1020 3000 aaronmk
    if sql_gen.index_col(col) != None: return # already has index col
1021 2997 aaronmk
1022
    new_col = sql_gen.suffixed_col(col, suffix)
1023
    new_typed_col = sql_gen.TypedCol(new_col.name, db.col_info(col).type)
1024
1025
    add_col(db, col.table, new_typed_col)
1026
    update(db, col.table, [(new_col, expr)])
1027
    if not nullable: add_not_null(db, new_col)
1028
    add_index(db, new_col)
1029
1030 3000 aaronmk
    col.table.index_cols[col.name] = new_col
1031 2997 aaronmk
1032
def ensure_not_null(db, col):
1033
    '''For params, see sql_gen.ensure_not_null()'''
1034
    expr = sql_gen.ensure_not_null(db, col)
1035
1036
    # If nullable column in a temp table, add separate column instead
1037
    if sql_gen.is_temp_col(col) and isinstance(expr, sql_gen.EnsureNotNull):
1038
        add_index_col(db, col, '::NOT NULL', expr, nullable=False)
1039 3000 aaronmk
        expr = sql_gen.index_col(col)
1040 2997 aaronmk
1041
    return expr
1042
1043 2772 aaronmk
already_indexed = object() # tells add_indexes() the pkey has already been added
1044
1045
def add_indexes(db, table, has_pkey=True):
1046
    '''Adds an index on all columns in a table.
1047
    @param has_pkey bool|already_indexed Whether a pkey instead of a regular
1048
        index should be added on the first column.
1049
        * If already_indexed, the pkey is assumed to have already been added
1050
    '''
1051
    cols = table_cols(db, table)
1052
    if has_pkey:
1053
        if has_pkey is not already_indexed: add_pkey(db, table)
1054
        cols = cols[1:]
1055
    for col in cols: add_index(db, col, table)
1056
1057 2935 aaronmk
def add_col(db, table, col, **kw_args):
1058 2934 aaronmk
    assert isinstance(col, sql_gen.TypedCol)
1059 2947 aaronmk
    try: run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD COLUMN '
1060
        +col.to_str(db), recover=True, cacheable=True, **kw_args)
1061
    except DuplicateException: pass # column already existed
1062 2934 aaronmk
1063 2935 aaronmk
row_num_typed_col = sql_gen.TypedCol(row_num_col, 'serial', nullable=False,
1064
    constraints='PRIMARY KEY')
1065
1066 2086 aaronmk
def add_row_num(db, table):
1067 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
1068
    be the primary key.'''
1069 2935 aaronmk
    add_col(db, table, row_num_typed_col, log_level=3)
1070 2086 aaronmk
1071 2936 aaronmk
def cast_temp_col(db, type_, col, errors_table=None):
1072
    '''Like cast(), but creates a new column with the cast values if the input
1073
    is a column.
1074
    @return The new column or cast value
1075
    '''
1076
    def cast_(col): return cast(db, type_, col, errors_table)
1077
1078
    try: col = sql_gen.underlying_col(col)
1079
    except sql_gen.NoUnderlyingTableException: return sql_gen.wrap(cast_, col)
1080
1081
    table = col.table
1082
    new_col = sql_gen.Col(sql_gen.concat(col.name, '::'+type_), table, col.srcs)
1083
    expr = cast_(col)
1084
    add_col(db, table, sql_gen.TypedCol(new_col.name, type_))
1085
    update(db, table, [(new_col, expr)])
1086 2955 aaronmk
    add_index(db, new_col)
1087 2936 aaronmk
1088
    return new_col
1089
1090 2889 aaronmk
def drop_table(db, table):
1091
    table = sql_gen.as_Table(table)
1092
    return run_query(db, 'DROP TABLE IF EXISTS '+table.to_str(db)+' CASCADE')
1093
1094 2760 aaronmk
def create_table(db, table, cols, has_pkey=True, col_indexes=True):
1095 2675 aaronmk
    '''Creates a table.
1096 2681 aaronmk
    @param cols [sql_gen.TypedCol,...] The column names and types
1097
    @param has_pkey If set, the first column becomes the primary key.
1098 2760 aaronmk
    @param col_indexes bool|[ref]
1099
        * If True, indexes will be added on all non-pkey columns.
1100
        * If a list reference, [0] will be set to a function to do this.
1101
          This can be used to delay index creation until the table is populated.
1102 2675 aaronmk
    '''
1103
    table = sql_gen.as_Table(table)
1104
1105 2681 aaronmk
    if has_pkey:
1106
        cols[0] = pkey = copy.copy(cols[0]) # don't modify input!
1107 2872 aaronmk
        pkey.constraints = 'PRIMARY KEY'
1108 2681 aaronmk
1109 2675 aaronmk
    str_ = 'CREATE TABLE '+table.to_str(db)+' (\n'
1110 2681 aaronmk
    str_ += '\n, '.join(v.to_str(db) for v in cols)
1111
    str_ += '\n);\n'
1112 2675 aaronmk
    run_query(db, str_, cacheable=True, log_level=2)
1113 2760 aaronmk
1114
    # Add indexes
1115 2773 aaronmk
    if has_pkey: has_pkey = already_indexed
1116
    def add_indexes_(): add_indexes(db, table, has_pkey)
1117
    if isinstance(col_indexes, list): col_indexes[0] = add_indexes_ # defer
1118
    elif col_indexes: add_indexes_() # add now
1119 2675 aaronmk
1120 2684 aaronmk
def vacuum(db, table):
1121
    table = sql_gen.as_Table(table)
1122
    db.with_autocommit(lambda: run_query(db, 'VACUUM ANALYZE '+table.to_str(db),
1123
        log_level=3))
1124
1125 2970 aaronmk
def truncate(db, table, schema='public', **kw_args):
1126
    '''For params, see run_query()'''
1127 2777 aaronmk
    table = sql_gen.as_Table(table, schema)
1128 2970 aaronmk
    return run_query(db, 'TRUNCATE '+table.to_str(db)+' CASCADE', **kw_args)
1129 2732 aaronmk
1130 2965 aaronmk
def empty_temp(db, tables):
1131 2972 aaronmk
    if db.debug_temp: return # leave temp tables there for debugging
1132 2965 aaronmk
    tables = lists.mk_seq(tables)
1133 2971 aaronmk
    for table in tables: truncate(db, table, log_level=3)
1134 2965 aaronmk
1135 2729 aaronmk
def tables(db, schema_like='public', table_like='%', exact=False):
1136
    if exact: compare = '='
1137
    else: compare = 'LIKE'
1138
1139 1849 aaronmk
    module = util.root_module(db.db)
1140 832 aaronmk
    if module == 'psycopg2':
1141 2779 aaronmk
        conds = [('schemaname', sql_gen.CompareCond(schema_like, compare)),
1142
            ('tablename', sql_gen.CompareCond(table_like, compare))]
1143
        return values(select(db, 'pg_tables', ['tablename'], conds,
1144
            order_by='tablename', log_level=4))
1145 1968 aaronmk
    elif module == 'MySQLdb':
1146 2779 aaronmk
        return values(run_query(db, 'SHOW TABLES LIKE '+db.esc_value(table_like)
1147
            , cacheable=True, log_level=4))
1148 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
1149 830 aaronmk
1150 2726 aaronmk
def table_exists(db, table):
1151
    table = sql_gen.as_Table(table)
1152 2730 aaronmk
    return list(tables(db, table.schema, table.name, exact=True)) != []
1153 2726 aaronmk
1154 2913 aaronmk
def function_exists(db, function):
1155
    function = sql_gen.as_Function(function)
1156
1157
    info_table = sql_gen.Table('routines', 'information_schema')
1158
    conds = [('routine_name', function.name)]
1159
    schema = function.schema
1160
    if schema != None: conds.append(('routine_schema', schema))
1161 2914 aaronmk
    # Exclude trigger functions, since they cannot be called directly
1162
    conds.append(('data_type', sql_gen.CompareCond('trigger', '!=')))
1163 2913 aaronmk
1164
    return list(values(select(db, info_table, ['routine_name'], conds,
1165
        order_by='routine_schema', limit=1, log_level=4))) != []
1166
        # TODO: order_by search_path schema order
1167
1168 2726 aaronmk
def errors_table(db, table, if_exists=True):
1169
    '''
1170
    @param if_exists If set, returns None if the errors table doesn't exist
1171
    @return None|sql_gen.Table
1172
    '''
1173
    table = sql_gen.as_Table(table)
1174
    if table.srcs != (): table = table.srcs[0]
1175
1176
    errors_table = sql_gen.suffixed_table(table, '.errors')
1177
    if if_exists and not table_exists(db, errors_table): return None
1178
    return errors_table
1179
1180 833 aaronmk
##### Database management
1181
1182 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
1183
    '''For kw_args, see tables()'''
1184
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
1185 833 aaronmk
1186 832 aaronmk
##### Heuristic queries
1187
1188 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
1189 1554 aaronmk
    '''Recovers from errors.
1190 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
1191
    '''
1192 2642 aaronmk
    row = sql_gen.ColDict(db, table, row)
1193 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
1194
1195 471 aaronmk
    try:
1196 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
1197 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
1198
            row_ct_ref[0] += cur.rowcount
1199
        return value(cur)
1200 471 aaronmk
    except DuplicateKeyException, e:
1201 2653 aaronmk
        row = sql_gen.ColDict(db, table,
1202
            util.dict_subset_right_join(row, e.cols))
1203
        return value(select(db, table, [pkey_], row, recover=True))
1204 471 aaronmk
1205 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
1206 830 aaronmk
    '''Recovers from errors'''
1207 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
1208 14 aaronmk
    except StopIteration:
1209 40 aaronmk
        if not create: raise
1210 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
1211 2078 aaronmk
1212 2593 aaronmk
def is_func_result(col):
1213
    return col.table.name.find('(') >= 0 and col.name == 'result'
1214
1215 2592 aaronmk
def into_table_name(out_table, in_tables0, mapping, is_func):
1216 2615 aaronmk
    def in_col_str(in_col):
1217
        in_col = sql_gen.remove_col_rename(in_col)
1218
        if isinstance(in_col, sql_gen.Col):
1219
            table = in_col.table
1220
            if table == in_tables0:
1221
                in_col = sql_gen.to_name_only_col(in_col)
1222
            elif is_func_result(in_col): in_col = table # omit col name
1223
        return str(in_col)
1224
1225 2580 aaronmk
    str_ = str(out_table)
1226
    if is_func:
1227 2615 aaronmk
        str_ += '('
1228 2580 aaronmk
1229 2615 aaronmk
        try: value_in_col = mapping['value']
1230
        except KeyError:
1231 2654 aaronmk
            str_ += ', '.join((str(k)+'='+in_col_str(v)
1232 2615 aaronmk
                for k, v in mapping.iteritems()))
1233
        else: str_ += in_col_str(value_in_col)
1234
1235
        str_ += ')'
1236 2812 aaronmk
    else:
1237
        out_col = 'rank'
1238
        try: in_col = mapping[out_col]
1239
        except KeyError: str_ += '_pkeys'
1240
        else: # has a rank column, so hierarchical
1241
            str_ += '['+str(out_col)+'='+in_col_str(in_col)+']'
1242 2580 aaronmk
    return str_
1243
1244 2508 aaronmk
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None, into=None,
1245 2805 aaronmk
    default=None, is_func=False, on_error=exc.raise_):
1246 2078 aaronmk
    '''Recovers from errors.
1247
    Only works under PostgreSQL (uses INSERT RETURNING).
1248 2909 aaronmk
    IMPORTANT: Must be run at the *beginning* of a transaction.
1249 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
1250
        tables to join with it using the main input table's pkey
1251 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
1252 2612 aaronmk
        * out_table_col: str (*not* sql_gen.Col)
1253
        * in_table_col: sql_gen.Col|literal-value
1254 2489 aaronmk
    @param into The table to contain the output and input pkeys.
1255 2574 aaronmk
        Defaults to `out_table.name+'_pkeys'`.
1256 2509 aaronmk
    @param default The *output* column to use as the pkey for missing rows.
1257
        If this output column does not exist in the mapping, uses None.
1258 2552 aaronmk
    @param is_func Whether out_table is the name of a SQL function, not a table
1259 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
1260 2078 aaronmk
    '''
1261 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
1262 2312 aaronmk
1263 2450 aaronmk
    def log_debug(msg): db.log_debug(msg, level=1.5)
1264 2505 aaronmk
    def col_ustr(str_):
1265 2567 aaronmk
        return strings.repr_no_u(sql_gen.remove_col_rename(str_))
1266 2450 aaronmk
1267 2910 aaronmk
    log_debug('********** New iteration **********')
1268
    log_debug('Inserting these input columns into '+strings.as_tt(
1269
        out_table.to_str(db))+':\n'+strings.as_table(mapping, ustr=col_ustr))
1270
1271 2942 aaronmk
    is_function = function_exists(db, out_table)
1272
1273
    if is_function: out_pkey = 'result'
1274
    else: out_pkey = pkey(db, out_table, recover=True)
1275 2768 aaronmk
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
1276
1277
    if mapping == {}: # need at least one column for INSERT SELECT
1278
        mapping = {out_pkey: None} # ColDict will replace with default value
1279
1280 2382 aaronmk
    # Create input joins from list of input tables
1281
    in_tables_ = in_tables[:] # don't modify input!
1282
    in_tables0 = in_tables_.pop(0) # first table is separate
1283 2735 aaronmk
    errors_table_ = errors_table(db, in_tables0)
1284 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
1285 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
1286 2460 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v,
1287
        {in_pkey: sql_gen.join_same_not_null}) for v in in_tables_]
1288 2131 aaronmk
1289 2592 aaronmk
    if into == None:
1290
        into = into_table_name(out_table, in_tables0, mapping, is_func)
1291
    into = sql_gen.as_Table(into)
1292
1293 2702 aaronmk
    # Set column sources
1294
    in_cols = filter(sql_gen.is_table_col, mapping.values())
1295
    for col in in_cols:
1296
        if col.table == in_tables0: col.set_srcs(sql_gen.src_self)
1297
1298 2486 aaronmk
    log_debug('Joining together input tables into temp table')
1299 2395 aaronmk
    # Place in new table for speed and so don't modify input if values edited
1300 2584 aaronmk
    in_table = sql_gen.Table('in')
1301 2702 aaronmk
    mapping = dicts.join(mapping, flatten(db, in_table, input_joins, in_cols,
1302
        preserve=[in_pkey_col], start=0))
1303 2395 aaronmk
    input_joins = [in_table]
1304 2486 aaronmk
    db.log_debug('Temp table: '+strings.as_tt(in_table.to_str(db)), level=2)
1305 2395 aaronmk
1306 2656 aaronmk
    mapping = sql_gen.ColDict(db, out_table, mapping)
1307
        # after applying dicts.join() because that returns a plain dict
1308
1309 2509 aaronmk
    # Resolve default value column
1310 2841 aaronmk
    if default != None:
1311
        try: default = mapping[default]
1312
        except KeyError:
1313 2509 aaronmk
            db.log_debug('Default value column '
1314
                +strings.as_tt(strings.repr_no_u(default))
1315 2511 aaronmk
                +' does not exist in mapping, falling back to None', level=2.1)
1316 2509 aaronmk
            default = None
1317
1318 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
1319 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
1320
1321 2201 aaronmk
    pkeys_table_exists_ref = [False]
1322 2960 aaronmk
    def insert_into_pkeys(joins, cols, distinct=False):
1323
        kw_args = {}
1324
        if distinct: kw_args.update(dict(distinct_on=[in_pkey_col]))
1325
        query = mk_select(db, joins, cols, order_by=None, start=0, **kw_args)
1326
1327 2201 aaronmk
        if pkeys_table_exists_ref[0]:
1328 2786 aaronmk
            insert_select(db, into, pkeys_names, query)
1329 2201 aaronmk
        else:
1330 2786 aaronmk
            run_query_into(db, query, into=into)
1331 2201 aaronmk
            pkeys_table_exists_ref[0] = True
1332
1333 2429 aaronmk
    limit_ref = [None]
1334 2380 aaronmk
    conds = set()
1335 2811 aaronmk
    distinct_on = sql_gen.ColDict(db, out_table)
1336 2325 aaronmk
    def mk_main_select(joins, cols):
1337 2811 aaronmk
        distinct_on_cols = [c.to_Col() for c in distinct_on.values()]
1338
        return mk_select(db, joins, cols, conds, distinct_on_cols,
1339 2429 aaronmk
            limit=limit_ref[0], start=0)
1340 2132 aaronmk
1341 2519 aaronmk
    exc_strs = set()
1342 2309 aaronmk
    def log_exc(e):
1343 2519 aaronmk
        e_str = exc.str_(e, first_line_only=True)
1344
        log_debug('Caught exception: '+e_str)
1345
        assert e_str not in exc_strs # avoid infinite loops
1346
        exc_strs.add(e_str)
1347 2735 aaronmk
1348 2451 aaronmk
    def remove_all_rows():
1349 2896 aaronmk
        log_debug('Ignoring all rows')
1350 2429 aaronmk
        limit_ref[0] = 0 # just create an empty pkeys table
1351 2735 aaronmk
1352 2758 aaronmk
    def ignore(in_col, value, e):
1353
        track_data_error(db, errors_table_, in_col.srcs, value, e.cause.pgcode,
1354
            e.cause.pgerror)
1355 2895 aaronmk
        log_debug('Ignoring rows with '+strings.as_tt(repr(in_col))+' = '
1356 2545 aaronmk
            +strings.as_tt(repr(value)))
1357 2895 aaronmk
1358 2758 aaronmk
    def remove_rows(in_col, value, e):
1359
        ignore(in_col, value, e)
1360 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
1361
        assert cond not in conds # avoid infinite loops
1362 2380 aaronmk
        conds.add(cond)
1363 2895 aaronmk
1364 2758 aaronmk
    def invalid2null(in_col, value, e):
1365
        ignore(in_col, value, e)
1366 2403 aaronmk
        update(db, in_table, [(in_col, None)],
1367
            sql_gen.ColValueCond(in_col, value))
1368 2245 aaronmk
1369 2589 aaronmk
    def insert_pkeys_table(which):
1370 2932 aaronmk
        return sql_gen.Table(sql_gen.concat(in_table.name,
1371 2589 aaronmk
            '_insert_'+which+'_pkeys'))
1372
    insert_out_pkeys = insert_pkeys_table('out')
1373
    insert_in_pkeys = insert_pkeys_table('in')
1374
1375 2206 aaronmk
    # Do inserts and selects
1376 2642 aaronmk
    join_cols = sql_gen.ColDict(db, out_table)
1377 2206 aaronmk
    while True:
1378 2521 aaronmk
        if limit_ref[0] == 0: # special case
1379
            log_debug('Creating an empty pkeys table')
1380 2786 aaronmk
            cur = run_query_into(db, mk_select(db, out_table, [out_pkey],
1381 2521 aaronmk
                limit=limit_ref[0]), into=insert_out_pkeys)
1382
            break # don't do main case
1383
1384 2303 aaronmk
        has_joins = join_cols != {}
1385
1386 2903 aaronmk
        log_debug('Trying to insert new rows')
1387
1388 2305 aaronmk
        # Prepare to insert new rows
1389 2325 aaronmk
        insert_joins = input_joins[:] # don't modify original!
1390 2403 aaronmk
        insert_args = dict(recover=True, cacheable=False)
1391 2303 aaronmk
        if has_joins:
1392 2325 aaronmk
            insert_joins.append(sql_gen.Join(out_table, join_cols,
1393
                sql_gen.filter_out))
1394
        else:
1395 2404 aaronmk
            insert_args.update(dict(returning=out_pkey, into=insert_out_pkeys))
1396 2786 aaronmk
        main_select = mk_main_select(insert_joins, mapping.values())
1397 2303 aaronmk
1398 2929 aaronmk
        def main_insert():
1399 2942 aaronmk
            if is_function:
1400
                log_debug('Calling function on input rows')
1401
                args = dict(((k.name, v) for k, v in mapping.iteritems()))
1402
                func_call = sql_gen.NamedCol(out_pkey,
1403
                    sql_gen.FunctionCall(out_table, **args))
1404
                insert_into_pkeys(input_joins, [in_pkey_col, func_call])
1405
                return None
1406
            else:
1407
                lock_table(db, out_table, 'EXCLUSIVE')
1408
                return insert_select(db, out_table, mapping.keys(), main_select,
1409
                    **insert_args)
1410 2929 aaronmk
1411 2206 aaronmk
        try:
1412 2929 aaronmk
            cur = with_savepoint(db, main_insert)
1413 2357 aaronmk
            break # insert successful
1414 2931 aaronmk
        except MissingCastException, e:
1415
            log_exc(e)
1416
1417
            out_col = e.col
1418
            type_ = e.type
1419
1420
            log_debug('Casting '+strings.as_tt(out_col)+' input to '
1421
                +strings.as_tt(type_))
1422 2937 aaronmk
            mapping[out_col] = cast_temp_col(db, type_, mapping[out_col],
1423
                errors_table_)
1424 2206 aaronmk
        except DuplicateKeyException, e:
1425 2309 aaronmk
            log_exc(e)
1426
1427 2258 aaronmk
            old_join_cols = join_cols.copy()
1428 2811 aaronmk
            distinct_on.update(util.dict_subset(mapping, e.cols))
1429 2565 aaronmk
            join_cols.update(util.dict_subset_right_join(mapping, e.cols))
1430 2486 aaronmk
            log_debug('Ignoring existing rows, comparing on these columns:\n'
1431 2505 aaronmk
                +strings.as_inline_table(join_cols, ustr=col_ustr))
1432 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
1433 2230 aaronmk
        except NullValueException, e:
1434 2309 aaronmk
            log_exc(e)
1435
1436 2230 aaronmk
            out_col, = e.cols
1437
            try: in_col = mapping[out_col]
1438 2356 aaronmk
            except KeyError:
1439 2486 aaronmk
                log_debug('Missing mapping for NOT NULL column '+out_col)
1440 2451 aaronmk
                remove_all_rows()
1441 2758 aaronmk
            else: remove_rows(in_col, None, e)
1442 2542 aaronmk
        except FunctionValueException, e:
1443
            log_exc(e)
1444
1445
            func_name = e.name
1446
            value = e.value
1447
            for out_col, in_col in mapping.iteritems():
1448 2758 aaronmk
                in_col = sql_gen.unwrap_func_call(in_col, func_name)
1449
                invalid2null(in_col, value, e)
1450 2429 aaronmk
        except DatabaseErrors, e:
1451
            log_exc(e)
1452
1453 2808 aaronmk
            log_debug('No handler for exception')
1454
            on_error(e)
1455 2451 aaronmk
            remove_all_rows()
1456 2358 aaronmk
        # after exception handled, rerun loop with additional constraints
1457 2132 aaronmk
1458 2942 aaronmk
    if cur != None and row_ct_ref != None and cur.rowcount >= 0:
1459 2357 aaronmk
        row_ct_ref[0] += cur.rowcount
1460
1461 2942 aaronmk
    if is_function: pass # pkeys table already created
1462
    elif has_joins:
1463 2357 aaronmk
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
1464 2486 aaronmk
        log_debug('Getting output table pkeys of existing/inserted rows')
1465 2960 aaronmk
        insert_into_pkeys(select_joins, pkeys_cols, distinct=True)
1466 2357 aaronmk
    else:
1467 2404 aaronmk
        add_row_num(db, insert_out_pkeys) # for joining with input pkeys
1468 2357 aaronmk
1469 2486 aaronmk
        log_debug('Getting input table pkeys of inserted rows')
1470 2786 aaronmk
        run_query_into(db, mk_main_select(input_joins, [in_pkey]),
1471 2404 aaronmk
            into=insert_in_pkeys)
1472
        add_row_num(db, insert_in_pkeys) # for joining with output pkeys
1473 2357 aaronmk
1474 2428 aaronmk
        assert table_row_count(db, insert_out_pkeys) == table_row_count(db,
1475
            insert_in_pkeys)
1476
1477 2486 aaronmk
        log_debug('Combining output and input pkeys in inserted order')
1478 2404 aaronmk
        pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
1479 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
1480 2420 aaronmk
        insert_into_pkeys(pkey_joins, pkeys_names)
1481 2959 aaronmk
1482 2974 aaronmk
        empty_temp(db, [insert_out_pkeys, insert_in_pkeys])
1483 2357 aaronmk
1484 2486 aaronmk
    db.log_debug('Adding pkey on pkeys table to enable fast joins', level=2.5)
1485 2594 aaronmk
    add_pkey(db, into)
1486 2407 aaronmk
1487 2508 aaronmk
    log_debug('Setting pkeys of missing rows to '+strings.as_tt(repr(default)))
1488 2489 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(into,
1489 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
1490
        # must use join_same_not_null or query will take forever
1491 2420 aaronmk
    insert_into_pkeys(missing_rows_joins,
1492 2508 aaronmk
        [in_pkey_col, sql_gen.NamedCol(out_pkey, default)])
1493 2357 aaronmk
1494 2489 aaronmk
    assert table_row_count(db, into) == table_row_count(db, in_table)
1495 2428 aaronmk
1496 2974 aaronmk
    empty_temp(db, in_table)
1497 2959 aaronmk
1498 2702 aaronmk
    srcs = []
1499
    if is_func: srcs = sql_gen.cols_srcs(in_cols)
1500
    return sql_gen.Col(out_pkey, into, srcs)
1501 2115 aaronmk
1502
##### Data cleanup
1503
1504 2290 aaronmk
def cleanup_table(db, table, cols):
1505 2677 aaronmk
    table = sql_gen.as_Table(table)
1506
    cols = map(sql_gen.as_Col, cols)
1507 2115 aaronmk
1508 2677 aaronmk
    expr = ('nullif(nullif(trim(both from %s), '+db.esc_value('')+'), '
1509
        +db.esc_value(r'\N')+')')
1510
    changes = [(v, sql_gen.CustomCode(expr % v.to_str(db)))
1511
        for v in cols]
1512 2115 aaronmk
1513 2677 aaronmk
    update(db, table, changes)