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 3007 aaronmk
                # structural changes
241
                if query.find('ADD COLUMN') < 0: # rest of query must be unique
242
                    self._cache_result()
243 3004 aaronmk
            elif self.rowcount == 0 and query.startswith('SELECT'): # empty
244 2800 aaronmk
                consume_rows(self) # fetch all rows so result will be cached
245 3004 aaronmk
246 2762 aaronmk
            return cur
247 1894 aaronmk
248 1891 aaronmk
        def fetchone(self):
249
            row = self.inner.fetchone()
250 1899 aaronmk
            if row != None: self.result.append(row)
251
            # otherwise, fetched all rows
252 1904 aaronmk
            else: self._cache_result()
253
            return row
254
255
        def _cache_result(self):
256 2948 aaronmk
            # For inserts that return a result set, don't cache result set since
257
            # inserts are not idempotent. Other non-SELECT queries don't have
258
            # their result set read, so only exceptions will be cached (an
259
            # invalid query will always be invalid).
260 1930 aaronmk
            if self.query_results != None and (not self._is_insert
261 1906 aaronmk
                or isinstance(self.result, Exception)):
262
263 1894 aaronmk
                assert self.query_lookup != None
264 1916 aaronmk
                self.query_results[self.query_lookup] = self.CacheCursor(
265
                    util.dict_subset(dicts.AttrsDictView(self),
266
                    ['query', 'result', 'rowcount', 'description']))
267 1906 aaronmk
268 1916 aaronmk
        class CacheCursor:
269
            def __init__(self, cached_result): self.__dict__ = cached_result
270
271 1927 aaronmk
            def execute(self, *args, **kw_args):
272 1916 aaronmk
                if isinstance(self.result, Exception): raise self.result
273
                # otherwise, result is a rows list
274
                self.iter = iter(self.result)
275
276
            def fetchone(self):
277
                try: return self.iter.next()
278
                except StopIteration: return None
279 1891 aaronmk
280 2212 aaronmk
    def esc_value(self, value):
281 2663 aaronmk
        try: str_ = self.mogrify('%s', [value])
282
        except NotImplementedError, e:
283
            module = util.root_module(self.db)
284
            if module == 'MySQLdb':
285
                import _mysql
286
                str_ = _mysql.escape_string(value)
287
            else: raise e
288 2374 aaronmk
        return strings.to_unicode(str_)
289 2212 aaronmk
290 2347 aaronmk
    def esc_name(self, name): return esc_name(self, name) # calls global func
291
292 2814 aaronmk
    def std_code(self, str_):
293
        '''Standardizes SQL code.
294
        * Ensures that string literals are prefixed by `E`
295
        '''
296
        if str_.startswith("'"): str_ = 'E'+str_
297
        return str_
298
299 2665 aaronmk
    def can_mogrify(self):
300 2663 aaronmk
        module = util.root_module(self.db)
301 2665 aaronmk
        return module == 'psycopg2'
302 2663 aaronmk
303 2665 aaronmk
    def mogrify(self, query, params=None):
304
        if self.can_mogrify(): return self.db.cursor().mogrify(query, params)
305
        else: raise NotImplementedError("Can't mogrify query")
306
307 2671 aaronmk
    def print_notices(self):
308 2725 aaronmk
        if hasattr(self.db, 'notices'):
309
            for msg in self.db.notices:
310
                if msg not in self._notices_seen:
311
                    self._notices_seen.add(msg)
312
                    self.log_debug(msg, level=2)
313 2671 aaronmk
314 2793 aaronmk
    def run_query(self, query, cacheable=False, log_level=2,
315 2464 aaronmk
        debug_msg_ref=None):
316 2445 aaronmk
        '''
317 2464 aaronmk
        @param log_ignore_excs The log_level will be increased by 2 if the query
318
            throws one of these exceptions.
319 2664 aaronmk
        @param debug_msg_ref If specified, the log message will be returned in
320
            this instead of being output. This allows you to filter log messages
321
            depending on the result of the query.
322 2445 aaronmk
        '''
323 2167 aaronmk
        assert query != None
324
325 2047 aaronmk
        if not self.caching: cacheable = False
326 1903 aaronmk
        used_cache = False
327 2664 aaronmk
328
        def log_msg(query):
329
            if used_cache: cache_status = 'cache hit'
330
            elif cacheable: cache_status = 'cache miss'
331
            else: cache_status = 'non-cacheable'
332
            return 'DB query: '+cache_status+':\n'+strings.as_code(query, 'SQL')
333
334 1903 aaronmk
        try:
335 1927 aaronmk
            # Get cursor
336
            if cacheable:
337
                try:
338 2797 aaronmk
                    cur = self.query_results[query]
339 1927 aaronmk
                    used_cache = True
340
                except KeyError: cur = self.DbCursor(self)
341
            else: cur = self.db.cursor()
342
343 2664 aaronmk
            # Log query
344
            if self.debug and debug_msg_ref == None: # log before running
345
                self.log_debug(log_msg(query), log_level)
346
347 1927 aaronmk
            # Run query
348 2793 aaronmk
            cur.execute(query)
349 1903 aaronmk
        finally:
350 2671 aaronmk
            self.print_notices()
351 2664 aaronmk
            if self.debug and debug_msg_ref != None: # return after running
352 2793 aaronmk
                debug_msg_ref[0] = log_msg(str(get_cur_query(cur, query)))
353 1903 aaronmk
354
        return cur
355 1914 aaronmk
356 2797 aaronmk
    def is_cached(self, query): return query in self.query_results
357 2139 aaronmk
358 2907 aaronmk
    def with_autocommit(self, func):
359 2801 aaronmk
        import psycopg2.extensions
360
361
        prev_isolation_level = self.db.isolation_level
362 2907 aaronmk
        self.db.set_isolation_level(
363
            psycopg2.extensions.ISOLATION_LEVEL_AUTOCOMMIT)
364 2683 aaronmk
        try: return func()
365 2801 aaronmk
        finally: self.db.set_isolation_level(prev_isolation_level)
366 2683 aaronmk
367 2139 aaronmk
    def with_savepoint(self, func):
368 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
369 2443 aaronmk
        self.run_query('SAVEPOINT '+savepoint, log_level=4)
370 2139 aaronmk
        self._savepoint += 1
371 2930 aaronmk
        try: return func()
372 2139 aaronmk
        except:
373 2443 aaronmk
            self.run_query('ROLLBACK TO SAVEPOINT '+savepoint, log_level=4)
374 2139 aaronmk
            raise
375 2930 aaronmk
        finally:
376
            # Always release savepoint, because after ROLLBACK TO SAVEPOINT,
377
            # "The savepoint remains valid and can be rolled back to again"
378
            # (http://www.postgresql.org/docs/8.3/static/sql-rollback-to.html).
379 2443 aaronmk
            self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
380 2930 aaronmk
381
            self._savepoint -= 1
382
            assert self._savepoint >= 0
383
384
            self.do_autocommit() # OK to do this after ROLLBACK TO SAVEPOINT
385 2191 aaronmk
386
    def do_autocommit(self):
387
        '''Autocommits if outside savepoint'''
388
        assert self._savepoint >= 0
389
        if self.autocommit and self._savepoint == 0:
390 2924 aaronmk
            self.log_debug('Autocommitting', level=4)
391 2191 aaronmk
            self.db.commit()
392 2643 aaronmk
393 2819 aaronmk
    def col_info(self, col):
394 2643 aaronmk
        table = sql_gen.Table('columns', 'information_schema')
395 2821 aaronmk
        cols = ['data_type', 'column_default',
396
            cast(self, 'boolean', 'is_nullable')]
397 2643 aaronmk
398
        conds = [('table_name', col.table.name), ('column_name', col.name)]
399
        schema = col.table.schema
400
        if schema != None: conds.append(('table_schema', schema))
401
402 2819 aaronmk
        type_, default, nullable = row(select(self, table, cols, conds,
403 2874 aaronmk
            order_by='table_schema', limit=1, log_level=4))
404 2643 aaronmk
            # TODO: order_by search_path schema order
405 2819 aaronmk
        default = sql_gen.as_Code(default, self)
406
407
        return sql_gen.TypedCol(col.name, type_, default, nullable)
408 2917 aaronmk
409
    def TempFunction(self, name):
410
        if self.debug_temp: schema = None
411
        else: schema = 'pg_temp'
412
        return sql_gen.Function(name, schema)
413 1849 aaronmk
414 1869 aaronmk
connect = DbConn
415
416 832 aaronmk
##### Recoverable querying
417 15 aaronmk
418 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
419 11 aaronmk
420 2791 aaronmk
def run_query(db, query, recover=None, cacheable=False, log_level=2,
421
    log_ignore_excs=None, **kw_args):
422 2794 aaronmk
    '''For params, see DbConn.run_query()'''
423 830 aaronmk
    if recover == None: recover = False
424 2464 aaronmk
    if log_ignore_excs == None: log_ignore_excs = ()
425
    log_ignore_excs = tuple(log_ignore_excs)
426 830 aaronmk
427 2666 aaronmk
    debug_msg_ref = None # usually, db.run_query() logs query before running it
428
    # But if filtering with log_ignore_excs, wait until after exception parsing
429 2984 aaronmk
    if log_ignore_excs != () or not db.can_mogrify(): debug_msg_ref = [None]
430 2666 aaronmk
431 2148 aaronmk
    try:
432 2464 aaronmk
        try:
433 2794 aaronmk
            def run(): return db.run_query(query, cacheable, log_level,
434 2793 aaronmk
                debug_msg_ref, **kw_args)
435 2796 aaronmk
            if recover and not db.is_cached(query):
436 2464 aaronmk
                return with_savepoint(db, run)
437
            else: return run() # don't need savepoint if cached
438
        except Exception, e:
439
            if not recover: raise # need savepoint to run index_cols()
440
            msg = exc.str_(e)
441
442
            match = re.search(r'duplicate key value violates unique constraint '
443 2493 aaronmk
                r'"((_?[^\W_]+)_.+?)"', msg)
444 2464 aaronmk
            if match:
445
                constraint, table = match.groups()
446
                try: cols = index_cols(db, table, constraint)
447
                except NotImplementedError: raise e
448
                else: raise DuplicateKeyException(constraint, cols, e)
449
450 2493 aaronmk
            match = re.search(r'null value in column "(.+?)" violates not-null'
451 2464 aaronmk
                r' constraint', msg)
452
            if match: raise NullValueException('NOT NULL', [match.group(1)], e)
453
454
            match = re.search(r'\b(?:invalid input (?:syntax|value)\b.*?'
455
                r'|date/time field value out of range): "(.+?)"\n'
456 2535 aaronmk
                r'(?:(?s).*?)\bfunction "(.+?)"', msg)
457 2464 aaronmk
            if match:
458
                value, name = match.groups()
459
                raise FunctionValueException(name, strings.to_unicode(value), e)
460
461 2526 aaronmk
            match = re.search(r'column "(.+?)" is of type (.+?) but expression '
462 2523 aaronmk
                r'is of type', msg)
463
            if match:
464
                col, type_ = match.groups()
465
                raise MissingCastException(type_, col, e)
466
467 2946 aaronmk
            match = re.search(r'\b(\S+) "(.+?)".*? already exists', msg)
468 2945 aaronmk
            if match:
469
                type_, name = match.groups()
470
                raise DuplicateException(type_, name, e)
471 2464 aaronmk
472
            raise # no specific exception raised
473
    except log_ignore_excs:
474
        log_level += 2
475
        raise
476
    finally:
477 2666 aaronmk
        if debug_msg_ref != None and debug_msg_ref[0] != None:
478
            db.log_debug(debug_msg_ref[0], log_level)
479 830 aaronmk
480 832 aaronmk
##### Basic queries
481
482 2153 aaronmk
def next_version(name):
483 2163 aaronmk
    version = 1 # first existing name was version 0
484 2586 aaronmk
    match = re.match(r'^(.*)#(\d+)$', name)
485 2153 aaronmk
    if match:
486 2586 aaronmk
        name, version = match.groups()
487
        version = int(version)+1
488 2932 aaronmk
    return sql_gen.concat(name, '#'+str(version))
489 2153 aaronmk
490 2899 aaronmk
def lock_table(db, table, mode):
491
    table = sql_gen.as_Table(table)
492
    run_query(db, 'LOCK TABLE '+table.to_str(db)+' IN '+mode+' MODE')
493
494 2789 aaronmk
def run_query_into(db, query, into=None, add_indexes_=False, **kw_args):
495 2085 aaronmk
    '''Outputs a query to a temp table.
496
    For params, see run_query().
497
    '''
498 2789 aaronmk
    if into == None: return run_query(db, query, **kw_args)
499 2790 aaronmk
500
    assert isinstance(into, sql_gen.Table)
501
502 2992 aaronmk
    into.is_temp = True
503 3008 aaronmk
    # "temporary tables cannot specify a schema name", so remove schema
504
    into.schema = None
505 2992 aaronmk
506 2790 aaronmk
    kw_args['recover'] = True
507 2945 aaronmk
    kw_args.setdefault('log_ignore_excs', (DuplicateException,))
508 2790 aaronmk
509 2916 aaronmk
    temp = not db.debug_temp # tables are permanent in debug_temp mode
510 2790 aaronmk
511
    # Create table
512
    while True:
513
        create_query = 'CREATE'
514
        if temp: create_query += ' TEMP'
515
        create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
516 2385 aaronmk
517 2790 aaronmk
        try:
518
            cur = run_query(db, create_query, **kw_args)
519
                # CREATE TABLE AS sets rowcount to # rows in query
520
            break
521 2945 aaronmk
        except DuplicateException, e:
522 2790 aaronmk
            into.name = next_version(into.name)
523
            # try again with next version of name
524
525
    if add_indexes_: add_indexes(db, into)
526
527
    return cur
528 2085 aaronmk
529 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
530
531 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
532
533 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
534 2293 aaronmk
    start=None, order_by=order_by_pkey, default_table=None):
535 1981 aaronmk
    '''
536 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
537 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
538 1981 aaronmk
    @param fields Use None to select all fields in the table
539 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
540 2379 aaronmk
        * container can be any iterable type
541 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
542
        * compare_right_side: sql_gen.ValueCond|literal value
543 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
544
        use all columns
545 2786 aaronmk
    @return query
546 1981 aaronmk
    '''
547 2315 aaronmk
    # Parse tables param
548 2964 aaronmk
    tables = lists.mk_seq(tables)
549 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
550 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
551 2121 aaronmk
552 2315 aaronmk
    # Parse other params
553 2376 aaronmk
    if conds == None: conds = []
554 2650 aaronmk
    elif dicts.is_dict(conds): conds = conds.items()
555 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
556 135 aaronmk
    assert limit == None or type(limit) == int
557 865 aaronmk
    assert start == None or type(start) == int
558 2315 aaronmk
    if order_by is order_by_pkey:
559
        if distinct_on != []: order_by = None
560
        else: order_by = pkey(db, table0, recover=True)
561 865 aaronmk
562 2315 aaronmk
    query = 'SELECT'
563 2056 aaronmk
564 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
565 2056 aaronmk
566 2200 aaronmk
    # DISTINCT ON columns
567 2233 aaronmk
    if distinct_on != []:
568 2467 aaronmk
        query += '\nDISTINCT'
569 2254 aaronmk
        if distinct_on is not distinct_on_all:
570 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
571
572
    # Columns
573 2467 aaronmk
    query += '\n'
574 1135 aaronmk
    if fields == None: query += '*'
575 2765 aaronmk
    else:
576
        assert fields != []
577
        query += '\n, '.join(map(parse_col, fields))
578 2200 aaronmk
579
    # Main table
580 2467 aaronmk
    query += '\nFROM '+table0.to_str(db)
581 865 aaronmk
582 2122 aaronmk
    # Add joins
583 2271 aaronmk
    left_table = table0
584 2263 aaronmk
    for join_ in tables:
585
        table = join_.table
586 2238 aaronmk
587 2343 aaronmk
        # Parse special values
588
        if join_.type_ is sql_gen.filter_out: # filter no match
589 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
590 2853 aaronmk
                sql_gen.CompareCond(None, '~=')))
591 2343 aaronmk
592 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
593 2122 aaronmk
594
        left_table = table
595
596 865 aaronmk
    missing = True
597 2376 aaronmk
    if conds != []:
598 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
599
        else: whitespace = '\n'
600 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
601
            .to_str(db) for l, r in conds], 'WHERE')
602 865 aaronmk
        missing = False
603 2227 aaronmk
    if order_by != None:
604 2467 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
605
    if limit != None: query += '\nLIMIT '+str(limit); missing = False
606 865 aaronmk
    if start != None:
607 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
608 865 aaronmk
        missing = False
609
    if missing: warnings.warn(DbWarning(
610
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
611
612 2786 aaronmk
    return query
613 11 aaronmk
614 2054 aaronmk
def select(db, *args, **kw_args):
615
    '''For params, see mk_select() and run_query()'''
616
    recover = kw_args.pop('recover', None)
617
    cacheable = kw_args.pop('cacheable', True)
618 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
619 2054 aaronmk
620 2791 aaronmk
    return run_query(db, mk_select(db, *args, **kw_args), recover, cacheable,
621
        log_level=log_level)
622 2054 aaronmk
623 2788 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, returning=None,
624 3009 aaronmk
    embeddable=False, ignore=False):
625 1960 aaronmk
    '''
626
    @param returning str|None An inserted column (such as pkey) to return
627 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
628 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
629
        query will be fully cached, not just if it raises an exception.
630 3009 aaronmk
    @param ignore Whether to ignore duplicate keys.
631 1960 aaronmk
    '''
632 2754 aaronmk
    table = sql_gen.remove_table_rename(sql_gen.as_Table(table))
633 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
634 3010 aaronmk
    if cols != None: cols = [sql_gen.to_name_only_col(c, table) for c in cols]
635 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
636 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
637 2063 aaronmk
638 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
639 2063 aaronmk
640 3009 aaronmk
    def mk_insert(select_query):
641
        query = first_line
642 3010 aaronmk
        if cols != None:
643
            query += '\n('+(', '.join((c.to_str(db) for c in cols)))+')'
644 3009 aaronmk
        query += '\n'+select_query
645
646
        if returning != None:
647
            returning_name_col = sql_gen.to_name_only_col(returning)
648
            query += '\nRETURNING '+returning_name_col.to_str(db)
649
650
        return query
651 2063 aaronmk
652 3009 aaronmk
    lang = 'sql'
653
    if ignore:
654
        embeddable = True # must use function
655
        lang = 'plpgsql'
656
657 3010 aaronmk
        assert cols != None
658
        row_cols = [sql_gen.Col(c.name, 'row') for c in cols]
659
        row_query = mk_insert(sql_gen.Values(row_cols).to_str(db))
660
661 3009 aaronmk
        query = '''\
662 3010 aaronmk
DECLARE
663
    row record;
664
    cur CURSOR FOR
665
'''+select_query+'''
666
;
667 3009 aaronmk
BEGIN
668 3010 aaronmk
    OPEN cur;
669
    FOUND := true; -- for initial check
670
    WHILE FOUND LOOP
671
        FETCH FROM cur INTO row;
672
'''+row_query+'''
673
;
674
    END LOOP;
675
    CLOSE cur;
676
END;\
677 3009 aaronmk
'''
678
    else: query = mk_insert(select_query)
679
680 2070 aaronmk
    if embeddable:
681 3009 aaronmk
        return_type = 'void'
682
        if returning != None:
683
            return_type = 'SETOF '+returning.to_str(db)+'%TYPE'
684 2327 aaronmk
685 2070 aaronmk
        # Create function
686 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
687 2189 aaronmk
        while True:
688
            try:
689 2918 aaronmk
                function = db.TempFunction(function_name)
690 2194 aaronmk
691 2189 aaronmk
                function_query = '''\
692 2698 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''()
693 2467 aaronmk
RETURNS '''+return_type+'''
694 3009 aaronmk
LANGUAGE '''+lang+'''
695 2467 aaronmk
AS $$
696 3009 aaronmk
'''+query+'''
697 2467 aaronmk
$$;
698 2070 aaronmk
'''
699 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
700 2945 aaronmk
                    log_ignore_excs=(DuplicateException,))
701 2189 aaronmk
                break # this version was successful
702 2945 aaronmk
            except DuplicateException, e:
703 2189 aaronmk
                function_name = next_version(function_name)
704
                # try again with next version of name
705 2070 aaronmk
706 2337 aaronmk
        # Return query that uses function
707 3009 aaronmk
        cols = None
708
        if returning != None: cols = [returning]
709 2698 aaronmk
        func_table = sql_gen.NamedTable('f', sql_gen.FunctionCall(function),
710 3009 aaronmk
            cols) # AS clause requires function alias
711 2787 aaronmk
        return mk_select(db, func_table, start=0, order_by=None)
712 2070 aaronmk
713 2787 aaronmk
    return query
714 2066 aaronmk
715
def insert_select(db, *args, **kw_args):
716 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
717 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
718
        values in
719 2072 aaronmk
    '''
720 2386 aaronmk
    into = kw_args.pop('into', None)
721
    if into != None: kw_args['embeddable'] = True
722 2066 aaronmk
    recover = kw_args.pop('recover', None)
723
    cacheable = kw_args.pop('cacheable', True)
724 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
725 2066 aaronmk
726 2789 aaronmk
    return run_query_into(db, mk_insert_select(db, *args, **kw_args), into,
727 2787 aaronmk
        recover=recover, cacheable=cacheable, log_level=log_level)
728 2063 aaronmk
729 2738 aaronmk
default = sql_gen.default # tells insert() to use the default value for a column
730 2066 aaronmk
731 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
732 2085 aaronmk
    '''For params, see insert_select()'''
733 1960 aaronmk
    if lists.is_seq(row): cols = None
734
    else:
735
        cols = row.keys()
736
        row = row.values()
737 2738 aaronmk
    row = list(row) # ensure that "== []" works
738 1960 aaronmk
739 2738 aaronmk
    if row == []: query = None
740
    else: query = sql_gen.Values(row).to_str(db)
741 1961 aaronmk
742 2788 aaronmk
    return insert_select(db, table, cols, query, *args, **kw_args)
743 11 aaronmk
744 2402 aaronmk
def mk_update(db, table, changes=None, cond=None):
745
    '''
746
    @param changes [(col, new_value),...]
747
        * container can be any iterable type
748
        * col: sql_gen.Code|str (for col name)
749
        * new_value: sql_gen.Code|literal value
750
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
751
    @return str query
752
    '''
753
    query = 'UPDATE '+sql_gen.as_Table(table).to_str(db)+'\nSET\n'
754 2405 aaronmk
    query += ',\n'.join((sql_gen.to_name_only_col(col, table).to_str(db)+' = '
755 2402 aaronmk
        +sql_gen.as_Value(new_value).to_str(db) for col, new_value in changes))
756 2467 aaronmk
    if cond != None: query += '\nWHERE\n'+cond.to_str(db)
757 2402 aaronmk
758
    return query
759
760
def update(db, *args, **kw_args):
761
    '''For params, see mk_update() and run_query()'''
762
    recover = kw_args.pop('recover', None)
763
764 2791 aaronmk
    return run_query(db, mk_update(db, *args, **kw_args), recover)
765 2402 aaronmk
766 135 aaronmk
def last_insert_id(db):
767 1849 aaronmk
    module = util.root_module(db.db)
768 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
769
    elif module == 'MySQLdb': return db.insert_id()
770
    else: return None
771 13 aaronmk
772 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
773 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
774 2415 aaronmk
    to names that will be distinct among the columns' tables.
775 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
776 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
777
    @param into The table for the new columns.
778 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
779
        columns will be included in the mapping even if they are not in cols.
780
        The tables of the provided Col objects will be changed to into, so make
781
        copies of them if you want to keep the original tables.
782
    @param as_items Whether to return a list of dict items instead of a dict
783 2383 aaronmk
    @return dict(orig_col=new_col, ...)
784
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
785 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
786
        * All mappings use the into table so its name can easily be
787 2383 aaronmk
          changed for all columns at once
788
    '''
789 2415 aaronmk
    cols = lists.uniqify(cols)
790
791 2394 aaronmk
    items = []
792 2389 aaronmk
    for col in preserve:
793 2390 aaronmk
        orig_col = copy.copy(col)
794 2392 aaronmk
        col.table = into
795 2394 aaronmk
        items.append((orig_col, col))
796
    preserve = set(preserve)
797
    for col in cols:
798 2716 aaronmk
        if col not in preserve:
799
            items.append((col, sql_gen.Col(str(col), into, col.srcs)))
800 2394 aaronmk
801
    if not as_items: items = dict(items)
802
    return items
803 2383 aaronmk
804 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
805 2391 aaronmk
    '''For params, see mk_flatten_mapping()
806
    @return See return value of mk_flatten_mapping()
807
    '''
808 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
809
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
810 2786 aaronmk
    run_query_into(db, mk_select(db, joins, cols, limit=limit, start=start),
811 2846 aaronmk
        into=into, add_indexes_=True)
812 2394 aaronmk
    return dict(items)
813 2391 aaronmk
814 2751 aaronmk
def mk_track_data_error(db, errors_table, cols, value, error_code, error):
815 2810 aaronmk
    assert cols != ()
816
817 2751 aaronmk
    cols = map(sql_gen.to_name_only_col, cols)
818
819
    columns_cols = ['column']
820
    columns = sql_gen.NamedValues('columns', columns_cols,
821
        [[c.name] for c in cols])
822
    values_cols = ['value', 'error_code', 'error']
823
    values = sql_gen.NamedValues('values', values_cols,
824
        [value, error_code, error])
825
826
    select_cols = columns_cols+values_cols
827
    name_only_cols = map(sql_gen.to_name_only_col, select_cols)
828 2755 aaronmk
    errors_table = sql_gen.NamedTable('errors', errors_table)
829 2751 aaronmk
    joins = [columns, sql_gen.Join(values, type_='CROSS'),
830
        sql_gen.Join(errors_table, dict(zip(name_only_cols, select_cols)),
831
        sql_gen.filter_out)]
832
833
    return mk_insert_select(db, errors_table, name_only_cols,
834 2787 aaronmk
        mk_select(db, joins, select_cols, order_by=None))
835 2751 aaronmk
836 2810 aaronmk
def track_data_error(db, errors_table, cols, *args, **kw_args):
837 2758 aaronmk
    '''
838
    @param errors_table If None, does nothing.
839
    '''
840 2810 aaronmk
    if errors_table == None or cols == (): return
841
    run_query(db, mk_track_data_error(db, errors_table, cols, *args, **kw_args),
842 2758 aaronmk
        cacheable=True, log_level=4)
843
844 2704 aaronmk
def cast(db, type_, col, errors_table=None):
845
    '''Casts an (unrenamed) column or value.
846 2769 aaronmk
    If errors_table set and col has srcs, saves errors in errors_table (using
847
    col's srcs attr as the source columns) and converts errors to warnings.
848 2820 aaronmk
    @param col str|sql_gen.Col|sql_gen.Literal
849 2704 aaronmk
    @param errors_table None|sql_gen.Table|str
850
    '''
851 2820 aaronmk
    col = sql_gen.as_Col(col)
852 2769 aaronmk
    save_errors = (errors_table != None and isinstance(col, sql_gen.Col)
853
        and col.srcs != ())
854 2987 aaronmk
    if not save_errors: return sql_gen.Cast(type_, col) # can't save errors
855 2704 aaronmk
856
    assert not isinstance(col, sql_gen.NamedCol)
857
858 2770 aaronmk
    errors_table = sql_gen.as_Table(errors_table)
859
    srcs = map(sql_gen.to_name_only_col, col.srcs)
860
    function_name = str(sql_gen.FunctionCall(type_, *srcs))
861 2918 aaronmk
    function = db.TempFunction(function_name)
862 2704 aaronmk
863 2718 aaronmk
    while True:
864
        # Create function definition
865
        query = '''\
866 2704 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''(value text)
867
RETURNS '''+type_+'''
868
LANGUAGE plpgsql
869
STRICT
870
AS $$
871
BEGIN
872
    /* The explicit cast to the return type is needed to make the cast happen
873
    inside the try block. (Implicit casts to the return type happen at the end
874
    of the function, outside any block.) */
875
    RETURN value::'''+type_+''';
876
EXCEPTION
877
    WHEN data_exception THEN
878
        -- Save error in errors table.
879 2751 aaronmk
        -- Insert the value and error for *each* source column.
880
'''+mk_track_data_error(db, errors_table, srcs,
881
    *map(sql_gen.CustomCode, ['value', 'SQLSTATE', 'SQLERRM']))+''';
882 2704 aaronmk
883
        RAISE WARNING '%', SQLERRM;
884
        RETURN NULL;
885
END;
886
$$;
887
'''
888 2718 aaronmk
889
        # Create function
890
        try:
891
            run_query(db, query, recover=True, cacheable=True,
892 2945 aaronmk
                log_ignore_excs=(DuplicateException,))
893 2718 aaronmk
            break # successful
894 2945 aaronmk
        except DuplicateException:
895 2770 aaronmk
            function.name = next_version(function.name)
896
            # try again with next version of name
897 2704 aaronmk
898
    return sql_gen.FunctionCall(function, col)
899
900 2414 aaronmk
##### Database structure queries
901
902 2426 aaronmk
def table_row_count(db, table, recover=None):
903 2786 aaronmk
    return value(run_query(db, mk_select(db, table, [sql_gen.row_count],
904 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
905 2426 aaronmk
906 2414 aaronmk
def table_cols(db, table, recover=None):
907
    return list(col_names(select(db, table, limit=0, order_by=None,
908 2443 aaronmk
        recover=recover, log_level=4)))
909 2414 aaronmk
910 2291 aaronmk
def pkey(db, table, recover=None):
911 832 aaronmk
    '''Assumed to be first column in table'''
912 2339 aaronmk
    return table_cols(db, table, recover)[0]
913 832 aaronmk
914 2559 aaronmk
not_null_col = 'not_null_col'
915 2340 aaronmk
916
def table_not_null_col(db, table, recover=None):
917
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
918
    if not_null_col in table_cols(db, table, recover): return not_null_col
919
    else: return pkey(db, table, recover)
920
921 853 aaronmk
def index_cols(db, table, index):
922
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
923
    automatically created. When you don't know whether something is a UNIQUE
924
    constraint or a UNIQUE index, use this function.'''
925 1909 aaronmk
    module = util.root_module(db.db)
926
    if module == 'psycopg2':
927
        return list(values(run_query(db, '''\
928 853 aaronmk
SELECT attname
929 866 aaronmk
FROM
930
(
931
        SELECT attnum, attname
932
        FROM pg_index
933
        JOIN pg_class index ON index.oid = indexrelid
934
        JOIN pg_class table_ ON table_.oid = indrelid
935
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
936
        WHERE
937 2782 aaronmk
            table_.relname = '''+db.esc_value(table)+'''
938
            AND index.relname = '''+db.esc_value(index)+'''
939 866 aaronmk
    UNION
940
        SELECT attnum, attname
941
        FROM
942
        (
943
            SELECT
944
                indrelid
945
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
946
                    AS indkey
947
            FROM pg_index
948
            JOIN pg_class index ON index.oid = indexrelid
949
            JOIN pg_class table_ ON table_.oid = indrelid
950
            WHERE
951 2782 aaronmk
                table_.relname = '''+db.esc_value(table)+'''
952
                AND index.relname = '''+db.esc_value(index)+'''
953 866 aaronmk
        ) s
954
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
955
) s
956 853 aaronmk
ORDER BY attnum
957 2782 aaronmk
'''
958
            , cacheable=True, log_level=4)))
959 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
960
        ' database')
961 853 aaronmk
962 464 aaronmk
def constraint_cols(db, table, constraint):
963 1849 aaronmk
    module = util.root_module(db.db)
964 464 aaronmk
    if module == 'psycopg2':
965
        return list(values(run_query(db, '''\
966
SELECT attname
967
FROM pg_constraint
968
JOIN pg_class ON pg_class.oid = conrelid
969
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
970
WHERE
971 2783 aaronmk
    relname = '''+db.esc_value(table)+'''
972
    AND conname = '''+db.esc_value(constraint)+'''
973 464 aaronmk
ORDER BY attnum
974 2783 aaronmk
'''
975
            )))
976 464 aaronmk
    else: raise NotImplementedError("Can't list constraint columns for "+module+
977
        ' database')
978
979 2096 aaronmk
row_num_col = '_row_num'
980
981 2998 aaronmk
def add_index(db, exprs, table=None, unique=False, ensure_not_null_=True):
982 2688 aaronmk
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
983 2538 aaronmk
    Currently, only function calls are supported as expressions.
984 2998 aaronmk
    @param ensure_not_null_ If set, translates NULL values to sentinel values.
985 2847 aaronmk
        This allows indexes to be used for comparisons where NULLs are equal.
986 2538 aaronmk
    '''
987 2964 aaronmk
    exprs = lists.mk_seq(exprs)
988 2538 aaronmk
989 2688 aaronmk
    # Parse exprs
990
    old_exprs = exprs[:]
991
    exprs = []
992
    cols = []
993
    for i, expr in enumerate(old_exprs):
994 2823 aaronmk
        expr = sql_gen.as_Col(expr, table)
995 2688 aaronmk
996 2823 aaronmk
        # Handle nullable columns
997 2998 aaronmk
        if ensure_not_null_:
998
            try: expr = ensure_not_null(db, expr)
999 2860 aaronmk
            except KeyError: pass # unknown type, so just create plain index
1000 2823 aaronmk
1001 2688 aaronmk
        # Extract col
1002 3002 aaronmk
        expr = copy.deepcopy(expr) # don't modify input!
1003 2688 aaronmk
        if isinstance(expr, sql_gen.FunctionCall):
1004
            col = expr.args[0]
1005
            expr = sql_gen.Expr(expr)
1006
        else: col = expr
1007 2823 aaronmk
        assert isinstance(col, sql_gen.Col)
1008 2688 aaronmk
1009
        # Extract table
1010
        if table == None:
1011
            assert sql_gen.is_table_col(col)
1012
            table = col.table
1013
1014
        col.table = None
1015
1016
        exprs.append(expr)
1017
        cols.append(col)
1018 2408 aaronmk
1019 2688 aaronmk
    table = sql_gen.as_Table(table)
1020
    index = sql_gen.Table(str(sql_gen.Col(','.join(map(str, cols)), table)))
1021
1022 3005 aaronmk
    # Add index
1023
    while True:
1024
        str_ = 'CREATE'
1025
        if unique: str_ += ' UNIQUE'
1026
        str_ += ' INDEX '+index.to_str(db)+' ON '+table.to_str(db)+' ('+(
1027
            ', '.join((v.to_str(db) for v in exprs)))+')'
1028
1029
        try:
1030
            run_query(db, str_, recover=True, cacheable=True, log_level=3,
1031
                log_ignore_excs=(DuplicateException,))
1032
            break
1033
        except DuplicateException:
1034
            index.name = next_version(index.name)
1035
            # try again with next version of name
1036 2408 aaronmk
1037 2686 aaronmk
def add_pkey(db, table, cols=None, recover=None):
1038
    '''Adds a primary key.
1039
    @param cols [sql_gen.Col,...] The columns in the primary key.
1040
        Defaults to the first column in the table.
1041 2406 aaronmk
    @pre The table must not already have a primary key.
1042
    '''
1043
    table = sql_gen.as_Table(table)
1044 2686 aaronmk
    if cols == None: cols = [pkey(db, table, recover)]
1045
    col_strs = [sql_gen.to_name_only_col(v).to_str(db) for v in cols]
1046 2406 aaronmk
1047 2781 aaronmk
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD PRIMARY KEY ('
1048
        +(', '.join(col_strs))+')', recover=True, cacheable=True, log_level=3,
1049 2945 aaronmk
        log_ignore_excs=(DuplicateException,))
1050 2406 aaronmk
1051 2997 aaronmk
def add_not_null(db, col):
1052
    table = col.table
1053
    col = sql_gen.to_name_only_col(col)
1054
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ALTER COLUMN '
1055
        +col.to_str(db)+' SET NOT NULL', cacheable=True)
1056
1057
def add_index_col(db, col, suffix, expr, nullable=True):
1058 3000 aaronmk
    if sql_gen.index_col(col) != None: return # already has index col
1059 2997 aaronmk
1060
    new_col = sql_gen.suffixed_col(col, suffix)
1061
1062 3006 aaronmk
    # Add column
1063
    while True:
1064
        new_typed_col = sql_gen.TypedCol(new_col.name, db.col_info(col).type)
1065
        try:
1066
            add_col(db, col.table, new_typed_col,
1067
                log_ignore_excs=(DuplicateException,))
1068
            break
1069
        except DuplicateException:
1070
            new_col.name = next_version(new_col.name)
1071
            # try again with next version of name
1072
1073 2997 aaronmk
    update(db, col.table, [(new_col, expr)])
1074
    if not nullable: add_not_null(db, new_col)
1075
    add_index(db, new_col)
1076
1077 3000 aaronmk
    col.table.index_cols[col.name] = new_col
1078 2997 aaronmk
1079
def ensure_not_null(db, col):
1080
    '''For params, see sql_gen.ensure_not_null()'''
1081
    expr = sql_gen.ensure_not_null(db, col)
1082
1083
    # If nullable column in a temp table, add separate column instead
1084
    if sql_gen.is_temp_col(col) and isinstance(expr, sql_gen.EnsureNotNull):
1085
        add_index_col(db, col, '::NOT NULL', expr, nullable=False)
1086 3000 aaronmk
        expr = sql_gen.index_col(col)
1087 2997 aaronmk
1088
    return expr
1089
1090 2772 aaronmk
already_indexed = object() # tells add_indexes() the pkey has already been added
1091
1092
def add_indexes(db, table, has_pkey=True):
1093
    '''Adds an index on all columns in a table.
1094
    @param has_pkey bool|already_indexed Whether a pkey instead of a regular
1095
        index should be added on the first column.
1096
        * If already_indexed, the pkey is assumed to have already been added
1097
    '''
1098
    cols = table_cols(db, table)
1099
    if has_pkey:
1100
        if has_pkey is not already_indexed: add_pkey(db, table)
1101
        cols = cols[1:]
1102
    for col in cols: add_index(db, col, table)
1103
1104 2935 aaronmk
def add_col(db, table, col, **kw_args):
1105 2934 aaronmk
    assert isinstance(col, sql_gen.TypedCol)
1106 3006 aaronmk
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD COLUMN '
1107 2947 aaronmk
        +col.to_str(db), recover=True, cacheable=True, **kw_args)
1108 2934 aaronmk
1109 2935 aaronmk
row_num_typed_col = sql_gen.TypedCol(row_num_col, 'serial', nullable=False,
1110
    constraints='PRIMARY KEY')
1111
1112 2086 aaronmk
def add_row_num(db, table):
1113 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
1114
    be the primary key.'''
1115 2935 aaronmk
    add_col(db, table, row_num_typed_col, log_level=3)
1116 2086 aaronmk
1117 2936 aaronmk
def cast_temp_col(db, type_, col, errors_table=None):
1118
    '''Like cast(), but creates a new column with the cast values if the input
1119
    is a column.
1120
    @return The new column or cast value
1121
    '''
1122
    def cast_(col): return cast(db, type_, col, errors_table)
1123
1124
    try: col = sql_gen.underlying_col(col)
1125
    except sql_gen.NoUnderlyingTableException: return sql_gen.wrap(cast_, col)
1126
1127
    table = col.table
1128
    new_col = sql_gen.Col(sql_gen.concat(col.name, '::'+type_), table, col.srcs)
1129
    expr = cast_(col)
1130
    add_col(db, table, sql_gen.TypedCol(new_col.name, type_))
1131
    update(db, table, [(new_col, expr)])
1132 2955 aaronmk
    add_index(db, new_col)
1133 2936 aaronmk
1134
    return new_col
1135
1136 2889 aaronmk
def drop_table(db, table):
1137
    table = sql_gen.as_Table(table)
1138
    return run_query(db, 'DROP TABLE IF EXISTS '+table.to_str(db)+' CASCADE')
1139
1140 2760 aaronmk
def create_table(db, table, cols, has_pkey=True, col_indexes=True):
1141 2675 aaronmk
    '''Creates a table.
1142 2681 aaronmk
    @param cols [sql_gen.TypedCol,...] The column names and types
1143
    @param has_pkey If set, the first column becomes the primary key.
1144 2760 aaronmk
    @param col_indexes bool|[ref]
1145
        * If True, indexes will be added on all non-pkey columns.
1146
        * If a list reference, [0] will be set to a function to do this.
1147
          This can be used to delay index creation until the table is populated.
1148 2675 aaronmk
    '''
1149
    table = sql_gen.as_Table(table)
1150
1151 2681 aaronmk
    if has_pkey:
1152
        cols[0] = pkey = copy.copy(cols[0]) # don't modify input!
1153 2872 aaronmk
        pkey.constraints = 'PRIMARY KEY'
1154 2681 aaronmk
1155 2675 aaronmk
    str_ = 'CREATE TABLE '+table.to_str(db)+' (\n'
1156 2681 aaronmk
    str_ += '\n, '.join(v.to_str(db) for v in cols)
1157
    str_ += '\n);\n'
1158 2675 aaronmk
    run_query(db, str_, cacheable=True, log_level=2)
1159 2760 aaronmk
1160
    # Add indexes
1161 2773 aaronmk
    if has_pkey: has_pkey = already_indexed
1162
    def add_indexes_(): add_indexes(db, table, has_pkey)
1163
    if isinstance(col_indexes, list): col_indexes[0] = add_indexes_ # defer
1164
    elif col_indexes: add_indexes_() # add now
1165 2675 aaronmk
1166 2684 aaronmk
def vacuum(db, table):
1167
    table = sql_gen.as_Table(table)
1168
    db.with_autocommit(lambda: run_query(db, 'VACUUM ANALYZE '+table.to_str(db),
1169
        log_level=3))
1170
1171 2970 aaronmk
def truncate(db, table, schema='public', **kw_args):
1172
    '''For params, see run_query()'''
1173 2777 aaronmk
    table = sql_gen.as_Table(table, schema)
1174 2970 aaronmk
    return run_query(db, 'TRUNCATE '+table.to_str(db)+' CASCADE', **kw_args)
1175 2732 aaronmk
1176 2965 aaronmk
def empty_temp(db, tables):
1177 2972 aaronmk
    if db.debug_temp: return # leave temp tables there for debugging
1178 2965 aaronmk
    tables = lists.mk_seq(tables)
1179 2971 aaronmk
    for table in tables: truncate(db, table, log_level=3)
1180 2965 aaronmk
1181 2729 aaronmk
def tables(db, schema_like='public', table_like='%', exact=False):
1182
    if exact: compare = '='
1183
    else: compare = 'LIKE'
1184
1185 1849 aaronmk
    module = util.root_module(db.db)
1186 832 aaronmk
    if module == 'psycopg2':
1187 2779 aaronmk
        conds = [('schemaname', sql_gen.CompareCond(schema_like, compare)),
1188
            ('tablename', sql_gen.CompareCond(table_like, compare))]
1189
        return values(select(db, 'pg_tables', ['tablename'], conds,
1190
            order_by='tablename', log_level=4))
1191 1968 aaronmk
    elif module == 'MySQLdb':
1192 2779 aaronmk
        return values(run_query(db, 'SHOW TABLES LIKE '+db.esc_value(table_like)
1193
            , cacheable=True, log_level=4))
1194 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
1195 830 aaronmk
1196 2726 aaronmk
def table_exists(db, table):
1197
    table = sql_gen.as_Table(table)
1198 2730 aaronmk
    return list(tables(db, table.schema, table.name, exact=True)) != []
1199 2726 aaronmk
1200 2913 aaronmk
def function_exists(db, function):
1201
    function = sql_gen.as_Function(function)
1202
1203
    info_table = sql_gen.Table('routines', 'information_schema')
1204
    conds = [('routine_name', function.name)]
1205
    schema = function.schema
1206
    if schema != None: conds.append(('routine_schema', schema))
1207 2914 aaronmk
    # Exclude trigger functions, since they cannot be called directly
1208
    conds.append(('data_type', sql_gen.CompareCond('trigger', '!=')))
1209 2913 aaronmk
1210
    return list(values(select(db, info_table, ['routine_name'], conds,
1211
        order_by='routine_schema', limit=1, log_level=4))) != []
1212
        # TODO: order_by search_path schema order
1213
1214 2726 aaronmk
def errors_table(db, table, if_exists=True):
1215
    '''
1216
    @param if_exists If set, returns None if the errors table doesn't exist
1217
    @return None|sql_gen.Table
1218
    '''
1219
    table = sql_gen.as_Table(table)
1220
    if table.srcs != (): table = table.srcs[0]
1221
1222
    errors_table = sql_gen.suffixed_table(table, '.errors')
1223
    if if_exists and not table_exists(db, errors_table): return None
1224
    return errors_table
1225
1226 833 aaronmk
##### Database management
1227
1228 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
1229
    '''For kw_args, see tables()'''
1230
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
1231 833 aaronmk
1232 832 aaronmk
##### Heuristic queries
1233
1234 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
1235 1554 aaronmk
    '''Recovers from errors.
1236 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
1237
    '''
1238 2642 aaronmk
    row = sql_gen.ColDict(db, table, row)
1239 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
1240
1241 471 aaronmk
    try:
1242 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
1243 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
1244
            row_ct_ref[0] += cur.rowcount
1245
        return value(cur)
1246 471 aaronmk
    except DuplicateKeyException, e:
1247 2653 aaronmk
        row = sql_gen.ColDict(db, table,
1248
            util.dict_subset_right_join(row, e.cols))
1249
        return value(select(db, table, [pkey_], row, recover=True))
1250 471 aaronmk
1251 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
1252 830 aaronmk
    '''Recovers from errors'''
1253 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
1254 14 aaronmk
    except StopIteration:
1255 40 aaronmk
        if not create: raise
1256 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
1257 2078 aaronmk
1258 2593 aaronmk
def is_func_result(col):
1259
    return col.table.name.find('(') >= 0 and col.name == 'result'
1260
1261 2592 aaronmk
def into_table_name(out_table, in_tables0, mapping, is_func):
1262 2615 aaronmk
    def in_col_str(in_col):
1263
        in_col = sql_gen.remove_col_rename(in_col)
1264
        if isinstance(in_col, sql_gen.Col):
1265
            table = in_col.table
1266
            if table == in_tables0:
1267
                in_col = sql_gen.to_name_only_col(in_col)
1268
            elif is_func_result(in_col): in_col = table # omit col name
1269
        return str(in_col)
1270
1271 2580 aaronmk
    str_ = str(out_table)
1272
    if is_func:
1273 2615 aaronmk
        str_ += '('
1274 2580 aaronmk
1275 2615 aaronmk
        try: value_in_col = mapping['value']
1276
        except KeyError:
1277 2654 aaronmk
            str_ += ', '.join((str(k)+'='+in_col_str(v)
1278 2615 aaronmk
                for k, v in mapping.iteritems()))
1279
        else: str_ += in_col_str(value_in_col)
1280
1281
        str_ += ')'
1282 2812 aaronmk
    else:
1283
        out_col = 'rank'
1284
        try: in_col = mapping[out_col]
1285
        except KeyError: str_ += '_pkeys'
1286
        else: # has a rank column, so hierarchical
1287
            str_ += '['+str(out_col)+'='+in_col_str(in_col)+']'
1288 2580 aaronmk
    return str_
1289
1290 2508 aaronmk
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None, into=None,
1291 2805 aaronmk
    default=None, is_func=False, on_error=exc.raise_):
1292 2078 aaronmk
    '''Recovers from errors.
1293
    Only works under PostgreSQL (uses INSERT RETURNING).
1294 2909 aaronmk
    IMPORTANT: Must be run at the *beginning* of a transaction.
1295 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
1296
        tables to join with it using the main input table's pkey
1297 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
1298 2612 aaronmk
        * out_table_col: str (*not* sql_gen.Col)
1299
        * in_table_col: sql_gen.Col|literal-value
1300 2489 aaronmk
    @param into The table to contain the output and input pkeys.
1301 2574 aaronmk
        Defaults to `out_table.name+'_pkeys'`.
1302 2509 aaronmk
    @param default The *output* column to use as the pkey for missing rows.
1303
        If this output column does not exist in the mapping, uses None.
1304 2552 aaronmk
    @param is_func Whether out_table is the name of a SQL function, not a table
1305 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
1306 2078 aaronmk
    '''
1307 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
1308 2312 aaronmk
1309 2450 aaronmk
    def log_debug(msg): db.log_debug(msg, level=1.5)
1310 2505 aaronmk
    def col_ustr(str_):
1311 2567 aaronmk
        return strings.repr_no_u(sql_gen.remove_col_rename(str_))
1312 2450 aaronmk
1313 2910 aaronmk
    log_debug('********** New iteration **********')
1314
    log_debug('Inserting these input columns into '+strings.as_tt(
1315
        out_table.to_str(db))+':\n'+strings.as_table(mapping, ustr=col_ustr))
1316
1317 2942 aaronmk
    is_function = function_exists(db, out_table)
1318
1319
    if is_function: out_pkey = 'result'
1320
    else: out_pkey = pkey(db, out_table, recover=True)
1321 2768 aaronmk
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
1322
1323
    if mapping == {}: # need at least one column for INSERT SELECT
1324
        mapping = {out_pkey: None} # ColDict will replace with default value
1325
1326 2382 aaronmk
    # Create input joins from list of input tables
1327
    in_tables_ = in_tables[:] # don't modify input!
1328
    in_tables0 = in_tables_.pop(0) # first table is separate
1329 2735 aaronmk
    errors_table_ = errors_table(db, in_tables0)
1330 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
1331 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
1332 2460 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v,
1333
        {in_pkey: sql_gen.join_same_not_null}) for v in in_tables_]
1334 2131 aaronmk
1335 2592 aaronmk
    if into == None:
1336
        into = into_table_name(out_table, in_tables0, mapping, is_func)
1337
    into = sql_gen.as_Table(into)
1338
1339 2702 aaronmk
    # Set column sources
1340
    in_cols = filter(sql_gen.is_table_col, mapping.values())
1341
    for col in in_cols:
1342
        if col.table == in_tables0: col.set_srcs(sql_gen.src_self)
1343
1344 2486 aaronmk
    log_debug('Joining together input tables into temp table')
1345 2395 aaronmk
    # Place in new table for speed and so don't modify input if values edited
1346 2584 aaronmk
    in_table = sql_gen.Table('in')
1347 2702 aaronmk
    mapping = dicts.join(mapping, flatten(db, in_table, input_joins, in_cols,
1348
        preserve=[in_pkey_col], start=0))
1349 2395 aaronmk
    input_joins = [in_table]
1350 2486 aaronmk
    db.log_debug('Temp table: '+strings.as_tt(in_table.to_str(db)), level=2)
1351 2395 aaronmk
1352 2656 aaronmk
    mapping = sql_gen.ColDict(db, out_table, mapping)
1353
        # after applying dicts.join() because that returns a plain dict
1354
1355 2509 aaronmk
    # Resolve default value column
1356 2841 aaronmk
    if default != None:
1357
        try: default = mapping[default]
1358
        except KeyError:
1359 2509 aaronmk
            db.log_debug('Default value column '
1360
                +strings.as_tt(strings.repr_no_u(default))
1361 2511 aaronmk
                +' does not exist in mapping, falling back to None', level=2.1)
1362 2509 aaronmk
            default = None
1363
1364 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
1365 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
1366
1367 2201 aaronmk
    pkeys_table_exists_ref = [False]
1368 2960 aaronmk
    def insert_into_pkeys(joins, cols, distinct=False):
1369
        kw_args = {}
1370
        if distinct: kw_args.update(dict(distinct_on=[in_pkey_col]))
1371
        query = mk_select(db, joins, cols, order_by=None, start=0, **kw_args)
1372
1373 2201 aaronmk
        if pkeys_table_exists_ref[0]:
1374 2786 aaronmk
            insert_select(db, into, pkeys_names, query)
1375 2201 aaronmk
        else:
1376 2786 aaronmk
            run_query_into(db, query, into=into)
1377 2201 aaronmk
            pkeys_table_exists_ref[0] = True
1378
1379 2429 aaronmk
    limit_ref = [None]
1380 2380 aaronmk
    conds = set()
1381 2811 aaronmk
    distinct_on = sql_gen.ColDict(db, out_table)
1382 2325 aaronmk
    def mk_main_select(joins, cols):
1383 2811 aaronmk
        distinct_on_cols = [c.to_Col() for c in distinct_on.values()]
1384
        return mk_select(db, joins, cols, conds, distinct_on_cols,
1385 2429 aaronmk
            limit=limit_ref[0], start=0)
1386 2132 aaronmk
1387 2519 aaronmk
    exc_strs = set()
1388 2309 aaronmk
    def log_exc(e):
1389 2519 aaronmk
        e_str = exc.str_(e, first_line_only=True)
1390
        log_debug('Caught exception: '+e_str)
1391
        assert e_str not in exc_strs # avoid infinite loops
1392
        exc_strs.add(e_str)
1393 2735 aaronmk
1394 2451 aaronmk
    def remove_all_rows():
1395 2896 aaronmk
        log_debug('Ignoring all rows')
1396 2429 aaronmk
        limit_ref[0] = 0 # just create an empty pkeys table
1397 2735 aaronmk
1398 2758 aaronmk
    def ignore(in_col, value, e):
1399
        track_data_error(db, errors_table_, in_col.srcs, value, e.cause.pgcode,
1400
            e.cause.pgerror)
1401 2895 aaronmk
        log_debug('Ignoring rows with '+strings.as_tt(repr(in_col))+' = '
1402 2545 aaronmk
            +strings.as_tt(repr(value)))
1403 2895 aaronmk
1404 2758 aaronmk
    def remove_rows(in_col, value, e):
1405
        ignore(in_col, value, e)
1406 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
1407
        assert cond not in conds # avoid infinite loops
1408 2380 aaronmk
        conds.add(cond)
1409 2895 aaronmk
1410 2758 aaronmk
    def invalid2null(in_col, value, e):
1411
        ignore(in_col, value, e)
1412 2403 aaronmk
        update(db, in_table, [(in_col, None)],
1413
            sql_gen.ColValueCond(in_col, value))
1414 2245 aaronmk
1415 2589 aaronmk
    def insert_pkeys_table(which):
1416 2932 aaronmk
        return sql_gen.Table(sql_gen.concat(in_table.name,
1417 2589 aaronmk
            '_insert_'+which+'_pkeys'))
1418
    insert_out_pkeys = insert_pkeys_table('out')
1419
    insert_in_pkeys = insert_pkeys_table('in')
1420
1421 2206 aaronmk
    # Do inserts and selects
1422 2642 aaronmk
    join_cols = sql_gen.ColDict(db, out_table)
1423 2206 aaronmk
    while True:
1424 2521 aaronmk
        if limit_ref[0] == 0: # special case
1425
            log_debug('Creating an empty pkeys table')
1426 2786 aaronmk
            cur = run_query_into(db, mk_select(db, out_table, [out_pkey],
1427 2521 aaronmk
                limit=limit_ref[0]), into=insert_out_pkeys)
1428
            break # don't do main case
1429
1430 2303 aaronmk
        has_joins = join_cols != {}
1431
1432 2903 aaronmk
        log_debug('Trying to insert new rows')
1433
1434 2305 aaronmk
        # Prepare to insert new rows
1435 2325 aaronmk
        insert_joins = input_joins[:] # don't modify original!
1436 2403 aaronmk
        insert_args = dict(recover=True, cacheable=False)
1437 2303 aaronmk
        if has_joins:
1438 2325 aaronmk
            insert_joins.append(sql_gen.Join(out_table, join_cols,
1439
                sql_gen.filter_out))
1440
        else:
1441 2404 aaronmk
            insert_args.update(dict(returning=out_pkey, into=insert_out_pkeys))
1442 2786 aaronmk
        main_select = mk_main_select(insert_joins, mapping.values())
1443 2303 aaronmk
1444 2929 aaronmk
        def main_insert():
1445 2942 aaronmk
            if is_function:
1446
                log_debug('Calling function on input rows')
1447
                args = dict(((k.name, v) for k, v in mapping.iteritems()))
1448
                func_call = sql_gen.NamedCol(out_pkey,
1449
                    sql_gen.FunctionCall(out_table, **args))
1450
                insert_into_pkeys(input_joins, [in_pkey_col, func_call])
1451
                return None
1452
            else:
1453
                lock_table(db, out_table, 'EXCLUSIVE')
1454
                return insert_select(db, out_table, mapping.keys(), main_select,
1455
                    **insert_args)
1456 2929 aaronmk
1457 2206 aaronmk
        try:
1458 2929 aaronmk
            cur = with_savepoint(db, main_insert)
1459 2357 aaronmk
            break # insert successful
1460 2931 aaronmk
        except MissingCastException, e:
1461
            log_exc(e)
1462
1463
            out_col = e.col
1464
            type_ = e.type
1465
1466
            log_debug('Casting '+strings.as_tt(out_col)+' input to '
1467
                +strings.as_tt(type_))
1468 2937 aaronmk
            mapping[out_col] = cast_temp_col(db, type_, mapping[out_col],
1469
                errors_table_)
1470 2206 aaronmk
        except DuplicateKeyException, e:
1471 2309 aaronmk
            log_exc(e)
1472
1473 2258 aaronmk
            old_join_cols = join_cols.copy()
1474 2811 aaronmk
            distinct_on.update(util.dict_subset(mapping, e.cols))
1475 2565 aaronmk
            join_cols.update(util.dict_subset_right_join(mapping, e.cols))
1476 2486 aaronmk
            log_debug('Ignoring existing rows, comparing on these columns:\n'
1477 2505 aaronmk
                +strings.as_inline_table(join_cols, ustr=col_ustr))
1478 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
1479 2230 aaronmk
        except NullValueException, e:
1480 2309 aaronmk
            log_exc(e)
1481
1482 2230 aaronmk
            out_col, = e.cols
1483
            try: in_col = mapping[out_col]
1484 2356 aaronmk
            except KeyError:
1485 2486 aaronmk
                log_debug('Missing mapping for NOT NULL column '+out_col)
1486 2451 aaronmk
                remove_all_rows()
1487 2758 aaronmk
            else: remove_rows(in_col, None, e)
1488 2542 aaronmk
        except FunctionValueException, e:
1489
            log_exc(e)
1490
1491
            func_name = e.name
1492
            value = e.value
1493
            for out_col, in_col in mapping.iteritems():
1494 2758 aaronmk
                in_col = sql_gen.unwrap_func_call(in_col, func_name)
1495
                invalid2null(in_col, value, e)
1496 2429 aaronmk
        except DatabaseErrors, e:
1497
            log_exc(e)
1498
1499 2808 aaronmk
            log_debug('No handler for exception')
1500
            on_error(e)
1501 2451 aaronmk
            remove_all_rows()
1502 2358 aaronmk
        # after exception handled, rerun loop with additional constraints
1503 2132 aaronmk
1504 2942 aaronmk
    if cur != None and row_ct_ref != None and cur.rowcount >= 0:
1505 2357 aaronmk
        row_ct_ref[0] += cur.rowcount
1506
1507 2942 aaronmk
    if is_function: pass # pkeys table already created
1508
    elif has_joins:
1509 2357 aaronmk
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
1510 2486 aaronmk
        log_debug('Getting output table pkeys of existing/inserted rows')
1511 2960 aaronmk
        insert_into_pkeys(select_joins, pkeys_cols, distinct=True)
1512 2357 aaronmk
    else:
1513 2404 aaronmk
        add_row_num(db, insert_out_pkeys) # for joining with input pkeys
1514 2357 aaronmk
1515 2486 aaronmk
        log_debug('Getting input table pkeys of inserted rows')
1516 2786 aaronmk
        run_query_into(db, mk_main_select(input_joins, [in_pkey]),
1517 2404 aaronmk
            into=insert_in_pkeys)
1518
        add_row_num(db, insert_in_pkeys) # for joining with output pkeys
1519 2357 aaronmk
1520 2428 aaronmk
        assert table_row_count(db, insert_out_pkeys) == table_row_count(db,
1521
            insert_in_pkeys)
1522
1523 2486 aaronmk
        log_debug('Combining output and input pkeys in inserted order')
1524 2404 aaronmk
        pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
1525 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
1526 2420 aaronmk
        insert_into_pkeys(pkey_joins, pkeys_names)
1527 2959 aaronmk
1528 2974 aaronmk
        empty_temp(db, [insert_out_pkeys, insert_in_pkeys])
1529 2357 aaronmk
1530 2486 aaronmk
    db.log_debug('Adding pkey on pkeys table to enable fast joins', level=2.5)
1531 2594 aaronmk
    add_pkey(db, into)
1532 2407 aaronmk
1533 2508 aaronmk
    log_debug('Setting pkeys of missing rows to '+strings.as_tt(repr(default)))
1534 2489 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(into,
1535 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
1536
        # must use join_same_not_null or query will take forever
1537 2420 aaronmk
    insert_into_pkeys(missing_rows_joins,
1538 2508 aaronmk
        [in_pkey_col, sql_gen.NamedCol(out_pkey, default)])
1539 2357 aaronmk
1540 2489 aaronmk
    assert table_row_count(db, into) == table_row_count(db, in_table)
1541 2428 aaronmk
1542 2974 aaronmk
    empty_temp(db, in_table)
1543 2959 aaronmk
1544 2702 aaronmk
    srcs = []
1545
    if is_func: srcs = sql_gen.cols_srcs(in_cols)
1546
    return sql_gen.Col(out_pkey, into, srcs)
1547 2115 aaronmk
1548
##### Data cleanup
1549
1550 2290 aaronmk
def cleanup_table(db, table, cols):
1551 2677 aaronmk
    table = sql_gen.as_Table(table)
1552
    cols = map(sql_gen.as_Col, cols)
1553 2115 aaronmk
1554 2677 aaronmk
    expr = ('nullif(nullif(trim(both from %s), '+db.esc_value('')+'), '
1555
        +db.esc_value(r'\N')+')')
1556
    changes = [(v, sql_gen.CustomCode(expr % v.to_str(db)))
1557
        for v in cols]
1558 2115 aaronmk
1559 2677 aaronmk
    update(db, table, changes)