Project

General

Profile

1 11 aaronmk
# Database access
2
3 1869 aaronmk
import copy
4 2127 aaronmk
import operator
5 11 aaronmk
import re
6 865 aaronmk
import warnings
7 11 aaronmk
8 300 aaronmk
import exc
9 1909 aaronmk
import dicts
10 1893 aaronmk
import iters
11 1960 aaronmk
import lists
12 1889 aaronmk
from Proxy import Proxy
13 1872 aaronmk
import rand
14 2217 aaronmk
import sql_gen
15 862 aaronmk
import strings
16 131 aaronmk
import util
17 11 aaronmk
18 832 aaronmk
##### Exceptions
19
20 2170 aaronmk
def get_cur_query(cur, input_query=None, input_params=None):
21 2168 aaronmk
    raw_query = None
22
    if hasattr(cur, 'query'): raw_query = cur.query
23
    elif hasattr(cur, '_last_executed'): raw_query = cur._last_executed
24 2170 aaronmk
25
    if raw_query != None: return raw_query
26 2371 aaronmk
    else: return '[input] '+strings.ustr(input_query)+' % '+repr(input_params)
27 14 aaronmk
28 2170 aaronmk
def _add_cursor_info(e, *args, **kw_args):
29
    '''For params, see get_cur_query()'''
30 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 2306 aaronmk
class ConstraintException(DbException):
50
    def __init__(self, name, cols, cause=None):
51 2484 aaronmk
        DbException.__init__(self, 'Violated '+strings.as_tt(name)
52
            +' constraint on columns: '+strings.as_tt(', '.join(cols)), cause)
53 2306 aaronmk
        self.name = name
54 468 aaronmk
        self.cols = cols
55 11 aaronmk
56 2523 aaronmk
class MissingCastException(DbException):
57
    def __init__(self, type_, col, cause=None):
58
        DbException.__init__(self, 'Missing cast to type '+strings.as_tt(type_)
59
            +' on column: '+strings.as_tt(col), cause)
60
        self.type = type_
61
        self.col = col
62
63 2143 aaronmk
class NameException(DbException): pass
64
65 2306 aaronmk
class DuplicateKeyException(ConstraintException): pass
66 13 aaronmk
67 2306 aaronmk
class NullValueException(ConstraintException): pass
68 13 aaronmk
69 2240 aaronmk
class FunctionValueException(ExceptionWithNameValue): pass
70 2239 aaronmk
71 2143 aaronmk
class DuplicateTableException(ExceptionWithName): pass
72
73 2188 aaronmk
class DuplicateFunctionException(ExceptionWithName): pass
74
75 89 aaronmk
class EmptyRowException(DbException): pass
76
77 865 aaronmk
##### Warnings
78
79
class DbWarning(UserWarning): pass
80
81 1930 aaronmk
##### Result retrieval
82
83
def col_names(cur): return (col[0] for col in cur.description)
84
85
def rows(cur): return iter(lambda: cur.fetchone(), None)
86
87
def consume_rows(cur):
88
    '''Used to fetch all rows so result will be cached'''
89
    iters.consume_iter(rows(cur))
90
91
def next_row(cur): return rows(cur).next()
92
93
def row(cur):
94
    row_ = next_row(cur)
95
    consume_rows(cur)
96
    return row_
97
98
def next_value(cur): return next_row(cur)[0]
99
100
def value(cur): return row(cur)[0]
101
102
def values(cur): return iters.func_iter(lambda: next_value(cur))
103
104
def value_or_none(cur):
105
    try: return value(cur)
106
    except StopIteration: return None
107
108 2762 aaronmk
##### Escaping
109 2101 aaronmk
110 2573 aaronmk
def esc_name_by_module(module, name):
111
    if module == 'psycopg2' or module == None: quote = '"'
112 2101 aaronmk
    elif module == 'MySQLdb': quote = '`'
113
    else: raise NotImplementedError("Can't escape name for "+module+' database')
114 2500 aaronmk
    return sql_gen.esc_name(name, quote)
115 2101 aaronmk
116
def esc_name_by_engine(engine, name, **kw_args):
117
    return esc_name_by_module(db_engines[engine][0], name, **kw_args)
118
119
def esc_name(db, name, **kw_args):
120
    return esc_name_by_module(util.root_module(db.db), name, **kw_args)
121
122
def qual_name(db, schema, table):
123
    def esc_name_(name): return esc_name(db, name)
124
    table = esc_name_(table)
125
    if schema != None: return esc_name_(schema)+'.'+table
126
    else: return table
127
128 1869 aaronmk
##### Database connections
129 1849 aaronmk
130 2097 aaronmk
db_config_names = ['engine', 'host', 'user', 'password', 'database', 'schemas']
131 1926 aaronmk
132 1869 aaronmk
db_engines = {
133
    'MySQL': ('MySQLdb', {'password': 'passwd', 'database': 'db'}),
134
    'PostgreSQL': ('psycopg2', {}),
135
}
136
137
DatabaseErrors_set = set([DbException])
138
DatabaseErrors = tuple(DatabaseErrors_set)
139
140
def _add_module(module):
141
    DatabaseErrors_set.add(module.DatabaseError)
142
    global DatabaseErrors
143
    DatabaseErrors = tuple(DatabaseErrors_set)
144
145
def db_config_str(db_config):
146
    return db_config['engine']+' database '+db_config['database']
147
148 1909 aaronmk
def _query_lookup(query, params): return (query, dicts.make_hashable(params))
149 1894 aaronmk
150 2448 aaronmk
log_debug_none = lambda msg, level=2: None
151 1901 aaronmk
152 1849 aaronmk
class DbConn:
153 2190 aaronmk
    def __init__(self, db_config, serializable=True, autocommit=False,
154
        caching=True, log_debug=log_debug_none):
155 1869 aaronmk
        self.db_config = db_config
156
        self.serializable = serializable
157 2190 aaronmk
        self.autocommit = autocommit
158
        self.caching = caching
159 1901 aaronmk
        self.log_debug = log_debug
160 2193 aaronmk
        self.debug = log_debug != log_debug_none
161 1869 aaronmk
162
        self.__db = None
163 1889 aaronmk
        self.query_results = {}
164 2139 aaronmk
        self._savepoint = 0
165 2671 aaronmk
        self._notices_seen = set()
166 1869 aaronmk
167
    def __getattr__(self, name):
168
        if name == '__dict__': raise Exception('getting __dict__')
169
        if name == 'db': return self._db()
170
        else: raise AttributeError()
171
172
    def __getstate__(self):
173
        state = copy.copy(self.__dict__) # shallow copy
174 1915 aaronmk
        state['log_debug'] = None # don't pickle the debug callback
175 1869 aaronmk
        state['_DbConn__db'] = None # don't pickle the connection
176
        return state
177
178 2165 aaronmk
    def connected(self): return self.__db != None
179
180 1869 aaronmk
    def _db(self):
181
        if self.__db == None:
182
            # Process db_config
183
            db_config = self.db_config.copy() # don't modify input!
184 2097 aaronmk
            schemas = db_config.pop('schemas', None)
185 1869 aaronmk
            module_name, mappings = db_engines[db_config.pop('engine')]
186
            module = __import__(module_name)
187
            _add_module(module)
188
            for orig, new in mappings.iteritems():
189
                try: util.rename_key(db_config, orig, new)
190
                except KeyError: pass
191
192
            # Connect
193
            self.__db = module.connect(**db_config)
194
195
            # Configure connection
196 2676 aaronmk
            if self.serializable and not self.autocommit: self.run_query(
197 2678 aaronmk
                'SET TRANSACTION ISOLATION LEVEL SERIALIZABLE', log_level=4)
198 2101 aaronmk
            if schemas != None:
199
                schemas_ = ''.join((esc_name(self, s)+', '
200
                    for s in schemas.split(',')))
201 2676 aaronmk
                self.run_query("SELECT set_config('search_path', \
202
%s || current_setting('search_path'), false)", [schemas_], log_level=3)
203 1869 aaronmk
204
        return self.__db
205 1889 aaronmk
206 1891 aaronmk
    class DbCursor(Proxy):
207 1927 aaronmk
        def __init__(self, outer):
208 1891 aaronmk
            Proxy.__init__(self, outer.db.cursor())
209 2191 aaronmk
            self.outer = outer
210 1927 aaronmk
            self.query_results = outer.query_results
211 1894 aaronmk
            self.query_lookup = None
212 1891 aaronmk
            self.result = []
213 1889 aaronmk
214 1894 aaronmk
        def execute(self, query, params=None):
215 2784 aaronmk
            assert params == None or params == [] or params == ()
216 2763 aaronmk
            if params == [] or params == (): params = None
217
                # None turns off the mogrifier
218 2762 aaronmk
219 2764 aaronmk
            self._is_insert = query.startswith('INSERT')
220 1894 aaronmk
            self.query_lookup = _query_lookup(query, params)
221 2148 aaronmk
            try:
222 2191 aaronmk
                try:
223 2763 aaronmk
                    cur = self.inner.execute(query, params)
224 2191 aaronmk
                    self.outer.do_autocommit()
225 2648 aaronmk
                finally: self.query = get_cur_query(self.inner, query, params)
226 1904 aaronmk
            except Exception, e:
227 2170 aaronmk
                _add_cursor_info(e, self, query, params)
228 1904 aaronmk
                self.result = e # cache the exception as the result
229
                self._cache_result()
230
                raise
231 1930 aaronmk
            # Fetch all rows so result will be cached
232
            if self.rowcount == 0 and not self._is_insert: consume_rows(self)
233 2762 aaronmk
            return cur
234 1894 aaronmk
235 1891 aaronmk
        def fetchone(self):
236
            row = self.inner.fetchone()
237 1899 aaronmk
            if row != None: self.result.append(row)
238
            # otherwise, fetched all rows
239 1904 aaronmk
            else: self._cache_result()
240
            return row
241
242
        def _cache_result(self):
243 1906 aaronmk
            # For inserts, only cache exceptions since inserts are not
244
            # idempotent, but an invalid insert will always be invalid
245 1930 aaronmk
            if self.query_results != None and (not self._is_insert
246 1906 aaronmk
                or isinstance(self.result, Exception)):
247
248 1894 aaronmk
                assert self.query_lookup != None
249 1916 aaronmk
                self.query_results[self.query_lookup] = self.CacheCursor(
250
                    util.dict_subset(dicts.AttrsDictView(self),
251
                    ['query', 'result', 'rowcount', 'description']))
252 1906 aaronmk
253 1916 aaronmk
        class CacheCursor:
254
            def __init__(self, cached_result): self.__dict__ = cached_result
255
256 1927 aaronmk
            def execute(self, *args, **kw_args):
257 1916 aaronmk
                if isinstance(self.result, Exception): raise self.result
258
                # otherwise, result is a rows list
259
                self.iter = iter(self.result)
260
261
            def fetchone(self):
262
                try: return self.iter.next()
263
                except StopIteration: return None
264 1891 aaronmk
265 2212 aaronmk
    def esc_value(self, value):
266 2663 aaronmk
        try: str_ = self.mogrify('%s', [value])
267
        except NotImplementedError, e:
268
            module = util.root_module(self.db)
269
            if module == 'MySQLdb':
270
                import _mysql
271
                str_ = _mysql.escape_string(value)
272
            else: raise e
273 2374 aaronmk
        return strings.to_unicode(str_)
274 2212 aaronmk
275 2347 aaronmk
    def esc_name(self, name): return esc_name(self, name) # calls global func
276
277 2665 aaronmk
    def can_mogrify(self):
278 2663 aaronmk
        module = util.root_module(self.db)
279 2665 aaronmk
        return module == 'psycopg2'
280 2663 aaronmk
281 2665 aaronmk
    def mogrify(self, query, params=None):
282
        if self.can_mogrify(): return self.db.cursor().mogrify(query, params)
283
        else: raise NotImplementedError("Can't mogrify query")
284
285 2671 aaronmk
    def print_notices(self):
286 2725 aaronmk
        if hasattr(self.db, 'notices'):
287
            for msg in self.db.notices:
288
                if msg not in self._notices_seen:
289
                    self._notices_seen.add(msg)
290
                    self.log_debug(msg, level=2)
291 2671 aaronmk
292 2445 aaronmk
    def run_query(self, query, params=None, cacheable=False, log_level=2,
293 2464 aaronmk
        debug_msg_ref=None):
294 2445 aaronmk
        '''
295 2784 aaronmk
        @param params Not supported. Use self.esc_value() instead.
296 2464 aaronmk
        @param log_ignore_excs The log_level will be increased by 2 if the query
297
            throws one of these exceptions.
298 2664 aaronmk
        @param debug_msg_ref If specified, the log message will be returned in
299
            this instead of being output. This allows you to filter log messages
300
            depending on the result of the query.
301 2445 aaronmk
        '''
302 2167 aaronmk
        assert query != None
303
304 2047 aaronmk
        if not self.caching: cacheable = False
305 1903 aaronmk
        used_cache = False
306 2664 aaronmk
307
        def log_msg(query):
308
            if used_cache: cache_status = 'cache hit'
309
            elif cacheable: cache_status = 'cache miss'
310
            else: cache_status = 'non-cacheable'
311
            return 'DB query: '+cache_status+':\n'+strings.as_code(query, 'SQL')
312
313 1903 aaronmk
        try:
314 1927 aaronmk
            # Get cursor
315
            if cacheable:
316
                query_lookup = _query_lookup(query, params)
317
                try:
318
                    cur = self.query_results[query_lookup]
319
                    used_cache = True
320
                except KeyError: cur = self.DbCursor(self)
321
            else: cur = self.db.cursor()
322
323 2664 aaronmk
            # Log query
324
            if self.debug and debug_msg_ref == None: # log before running
325
                self.log_debug(log_msg(query), log_level)
326
327 1927 aaronmk
            # Run query
328 2148 aaronmk
            cur.execute(query, params)
329 1903 aaronmk
        finally:
330 2671 aaronmk
            self.print_notices()
331 2664 aaronmk
            if self.debug and debug_msg_ref != None: # return after running
332
                debug_msg_ref[0] = log_msg(str(get_cur_query(cur, query,
333
                    params)))
334 1903 aaronmk
335
        return cur
336 1914 aaronmk
337
    def is_cached(self, query, params=None):
338
        return _query_lookup(query, params) in self.query_results
339 2139 aaronmk
340 2683 aaronmk
    def with_autocommit(self, func, autocommit=True):
341
        prev_autocommit = self.db.autocommit
342
        self.db.autocommit = autocommit
343
        try: return func()
344
        finally: self.db.autocommit = prev_autocommit
345
346 2139 aaronmk
    def with_savepoint(self, func):
347 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
348 2443 aaronmk
        self.run_query('SAVEPOINT '+savepoint, log_level=4)
349 2139 aaronmk
        self._savepoint += 1
350 2683 aaronmk
        try:
351 2139 aaronmk
            try: return_val = func()
352
            finally:
353
                self._savepoint -= 1
354
                assert self._savepoint >= 0
355
        except:
356 2443 aaronmk
            self.run_query('ROLLBACK TO SAVEPOINT '+savepoint, log_level=4)
357 2139 aaronmk
            raise
358
        else:
359 2443 aaronmk
            self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
360 2191 aaronmk
            self.do_autocommit()
361 2139 aaronmk
            return return_val
362 2191 aaronmk
363
    def do_autocommit(self):
364
        '''Autocommits if outside savepoint'''
365
        assert self._savepoint >= 0
366
        if self.autocommit and self._savepoint == 0:
367 2668 aaronmk
            self.log_debug('Autocommitting')
368 2191 aaronmk
            self.db.commit()
369 2643 aaronmk
370
    def col_default(self, col):
371
        table = sql_gen.Table('columns', 'information_schema')
372
373
        conds = [('table_name', col.table.name), ('column_name', col.name)]
374
        schema = col.table.schema
375
        if schema != None: conds.append(('table_schema', schema))
376
377 2660 aaronmk
        return sql_gen.as_Code(value(select(self, table, ['column_default'],
378
            conds, order_by='table_schema', limit=1, log_level=3)))
379 2643 aaronmk
            # TODO: order_by search_path schema order
380 1849 aaronmk
381 1869 aaronmk
connect = DbConn
382
383 832 aaronmk
##### Querying
384
385 1894 aaronmk
def run_raw_query(db, *args, **kw_args):
386 2085 aaronmk
    '''For params, see DbConn.run_query()'''
387 1894 aaronmk
    return db.run_query(*args, **kw_args)
388 11 aaronmk
389 2068 aaronmk
def mogrify(db, query, params):
390
    module = util.root_module(db.db)
391
    if module == 'psycopg2': return db.db.cursor().mogrify(query, params)
392
    else: raise NotImplementedError("Can't mogrify query for "+module+
393
        ' database')
394
395 832 aaronmk
##### Recoverable querying
396 15 aaronmk
397 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
398 11 aaronmk
399 2464 aaronmk
def run_query(db, query, params=None, recover=None, cacheable=False,
400
    log_level=2, log_ignore_excs=None, **kw_args):
401 2441 aaronmk
    '''For params, see run_raw_query()'''
402 830 aaronmk
    if recover == None: recover = False
403 2464 aaronmk
    if log_ignore_excs == None: log_ignore_excs = ()
404
    log_ignore_excs = tuple(log_ignore_excs)
405 830 aaronmk
406 2666 aaronmk
    debug_msg_ref = None # usually, db.run_query() logs query before running it
407
    # But if filtering with log_ignore_excs, wait until after exception parsing
408
    if log_ignore_excs != () or not db.can_mogrify(): debug_msg_ref = [None]
409
410 2148 aaronmk
    try:
411 2464 aaronmk
        try:
412
            def run(): return run_raw_query(db, query, params, cacheable,
413
                log_level, debug_msg_ref, **kw_args)
414
            if recover and not db.is_cached(query, params):
415
                return with_savepoint(db, run)
416
            else: return run() # don't need savepoint if cached
417
        except Exception, e:
418
            if not recover: raise # need savepoint to run index_cols()
419
            msg = exc.str_(e)
420
421
            match = re.search(r'duplicate key value violates unique constraint '
422 2493 aaronmk
                r'"((_?[^\W_]+)_.+?)"', msg)
423 2464 aaronmk
            if match:
424
                constraint, table = match.groups()
425
                try: cols = index_cols(db, table, constraint)
426
                except NotImplementedError: raise e
427
                else: raise DuplicateKeyException(constraint, cols, e)
428
429 2493 aaronmk
            match = re.search(r'null value in column "(.+?)" violates not-null'
430 2464 aaronmk
                r' constraint', msg)
431
            if match: raise NullValueException('NOT NULL', [match.group(1)], e)
432
433
            match = re.search(r'\b(?:invalid input (?:syntax|value)\b.*?'
434
                r'|date/time field value out of range): "(.+?)"\n'
435 2535 aaronmk
                r'(?:(?s).*?)\bfunction "(.+?)"', msg)
436 2464 aaronmk
            if match:
437
                value, name = match.groups()
438
                raise FunctionValueException(name, strings.to_unicode(value), e)
439
440 2526 aaronmk
            match = re.search(r'column "(.+?)" is of type (.+?) but expression '
441 2523 aaronmk
                r'is of type', msg)
442
            if match:
443
                col, type_ = match.groups()
444
                raise MissingCastException(type_, col, e)
445
446 2493 aaronmk
            match = re.search(r'relation "(.+?)" already exists', msg)
447 2464 aaronmk
            if match: raise DuplicateTableException(match.group(1), e)
448
449 2493 aaronmk
            match = re.search(r'function "(.+?)" already exists', msg)
450 2464 aaronmk
            if match: raise DuplicateFunctionException(match.group(1), e)
451
452
            raise # no specific exception raised
453
    except log_ignore_excs:
454
        log_level += 2
455
        raise
456
    finally:
457 2666 aaronmk
        if debug_msg_ref != None and debug_msg_ref[0] != None:
458
            db.log_debug(debug_msg_ref[0], log_level)
459 830 aaronmk
460 832 aaronmk
##### Basic queries
461
462 2153 aaronmk
def next_version(name):
463 2163 aaronmk
    version = 1 # first existing name was version 0
464 2586 aaronmk
    match = re.match(r'^(.*)#(\d+)$', name)
465 2153 aaronmk
    if match:
466 2586 aaronmk
        name, version = match.groups()
467
        version = int(version)+1
468 2588 aaronmk
    return sql_gen.add_suffix(name, '#'+str(version))
469 2153 aaronmk
470 2786 aaronmk
def run_query_into(db, query, params=None, into=None, add_indexes_=False,
471
    **kw_args):
472 2085 aaronmk
    '''Outputs a query to a temp table.
473
    For params, see run_query().
474
    '''
475 2774 aaronmk
    if into == None: return run_query(db, query, params, **kw_args)
476 2085 aaronmk
    else: # place rows in temp table
477 2386 aaronmk
        assert isinstance(into, sql_gen.Table)
478 2385 aaronmk
479 2153 aaronmk
        kw_args['recover'] = True
480 2464 aaronmk
        kw_args.setdefault('log_ignore_excs', (DuplicateTableException,))
481 2440 aaronmk
482 2468 aaronmk
        temp = not db.autocommit # tables are permanent in autocommit mode
483 2440 aaronmk
        # "temporary tables cannot specify a schema name", so remove schema
484
        if temp: into.schema = None
485
486 2774 aaronmk
        # Create table
487 2153 aaronmk
        while True:
488 2774 aaronmk
            create_query = 'CREATE'
489
            if temp: create_query += ' TEMP'
490
            create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
491
492 2153 aaronmk
            try:
493 2774 aaronmk
                cur = run_query(db, create_query, params, **kw_args)
494 2153 aaronmk
                    # CREATE TABLE AS sets rowcount to # rows in query
495 2774 aaronmk
                break
496 2153 aaronmk
            except DuplicateTableException, e:
497 2386 aaronmk
                into.name = next_version(into.name)
498 2153 aaronmk
                # try again with next version of name
499 2774 aaronmk
500
        if add_indexes_: add_indexes(db, into)
501
502
        return cur
503 2085 aaronmk
504 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
505
506 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
507
508 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
509 2293 aaronmk
    start=None, order_by=order_by_pkey, default_table=None):
510 1981 aaronmk
    '''
511 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
512 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
513 1981 aaronmk
    @param fields Use None to select all fields in the table
514 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
515 2379 aaronmk
        * container can be any iterable type
516 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
517
        * compare_right_side: sql_gen.ValueCond|literal value
518 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
519
        use all columns
520 2786 aaronmk
    @return query
521 1981 aaronmk
    '''
522 2315 aaronmk
    # Parse tables param
523 2121 aaronmk
    if not lists.is_seq(tables): tables = [tables]
524 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
525 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
526 2121 aaronmk
527 2315 aaronmk
    # Parse other params
528 2376 aaronmk
    if conds == None: conds = []
529 2650 aaronmk
    elif dicts.is_dict(conds): conds = conds.items()
530 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
531 135 aaronmk
    assert limit == None or type(limit) == int
532 865 aaronmk
    assert start == None or type(start) == int
533 2315 aaronmk
    if order_by is order_by_pkey:
534
        if distinct_on != []: order_by = None
535
        else: order_by = pkey(db, table0, recover=True)
536 865 aaronmk
537 2315 aaronmk
    query = 'SELECT'
538 2056 aaronmk
539 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
540 2056 aaronmk
541 2200 aaronmk
    # DISTINCT ON columns
542 2233 aaronmk
    if distinct_on != []:
543 2467 aaronmk
        query += '\nDISTINCT'
544 2254 aaronmk
        if distinct_on is not distinct_on_all:
545 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
546
547
    # Columns
548 2467 aaronmk
    query += '\n'
549 1135 aaronmk
    if fields == None: query += '*'
550 2765 aaronmk
    else:
551
        assert fields != []
552
        query += '\n, '.join(map(parse_col, fields))
553 2200 aaronmk
554
    # Main table
555 2467 aaronmk
    query += '\nFROM '+table0.to_str(db)
556 865 aaronmk
557 2122 aaronmk
    # Add joins
558 2271 aaronmk
    left_table = table0
559 2263 aaronmk
    for join_ in tables:
560
        table = join_.table
561 2238 aaronmk
562 2343 aaronmk
        # Parse special values
563
        if join_.type_ is sql_gen.filter_out: # filter no match
564 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
565
                None))
566 2343 aaronmk
567 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
568 2122 aaronmk
569
        left_table = table
570
571 865 aaronmk
    missing = True
572 2376 aaronmk
    if conds != []:
573 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
574
        else: whitespace = '\n'
575 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
576
            .to_str(db) for l, r in conds], 'WHERE')
577 865 aaronmk
        missing = False
578 2227 aaronmk
    if order_by != None:
579 2467 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
580
    if limit != None: query += '\nLIMIT '+str(limit); missing = False
581 865 aaronmk
    if start != None:
582 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
583 865 aaronmk
        missing = False
584
    if missing: warnings.warn(DbWarning(
585
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
586
587 2786 aaronmk
    return query
588 11 aaronmk
589 2054 aaronmk
def select(db, *args, **kw_args):
590
    '''For params, see mk_select() and run_query()'''
591
    recover = kw_args.pop('recover', None)
592
    cacheable = kw_args.pop('cacheable', True)
593 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
594 2054 aaronmk
595 2786 aaronmk
    return run_query(db, mk_select(db, *args, **kw_args), [], recover,
596
        cacheable, log_level=log_level)
597 2054 aaronmk
598 2066 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, params=None,
599 2292 aaronmk
    returning=None, embeddable=False):
600 1960 aaronmk
    '''
601
    @param returning str|None An inserted column (such as pkey) to return
602 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
603 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
604
        query will be fully cached, not just if it raises an exception.
605 1960 aaronmk
    '''
606 2754 aaronmk
    table = sql_gen.remove_table_rename(sql_gen.as_Table(table))
607 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
608 2646 aaronmk
    if cols != None:
609
        cols = [sql_gen.to_name_only_col(v, table).to_str(db) for v in cols]
610 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
611 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
612 2063 aaronmk
613
    # Build query
614 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
615
    query = first_line
616 2467 aaronmk
    if cols != None: query += '\n('+', '.join(cols)+')'
617
    query += '\n'+select_query
618 2063 aaronmk
619
    if returning != None:
620 2743 aaronmk
        query += '\nRETURNING '+sql_gen.to_name_only_col(returning).to_str(db)
621 2063 aaronmk
622 2070 aaronmk
    if embeddable:
623 2327 aaronmk
        assert returning != None
624
625 2070 aaronmk
        # Create function
626 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
627 2327 aaronmk
        return_type = 'SETOF '+returning.to_str(db)+'%TYPE'
628 2189 aaronmk
        while True:
629
            try:
630 2698 aaronmk
                function = sql_gen.TempFunction(function_name, db.autocommit)
631 2194 aaronmk
632 2189 aaronmk
                function_query = '''\
633 2698 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''()
634 2467 aaronmk
RETURNS '''+return_type+'''
635
LANGUAGE sql
636
AS $$
637
'''+mogrify(db, query, params)+''';
638
$$;
639 2070 aaronmk
'''
640 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
641 2464 aaronmk
                    log_ignore_excs=(DuplicateFunctionException,))
642 2189 aaronmk
                break # this version was successful
643
            except DuplicateFunctionException, e:
644
                function_name = next_version(function_name)
645
                # try again with next version of name
646 2070 aaronmk
647 2337 aaronmk
        # Return query that uses function
648 2698 aaronmk
        func_table = sql_gen.NamedTable('f', sql_gen.FunctionCall(function),
649 2742 aaronmk
            [returning]) # AS clause requires function alias
650 2786 aaronmk
        return (mk_select(db, func_table, start=0, order_by=None), [])
651 2070 aaronmk
652 2066 aaronmk
    return (query, params)
653
654
def insert_select(db, *args, **kw_args):
655 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
656 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
657
        values in
658 2072 aaronmk
    '''
659 2386 aaronmk
    into = kw_args.pop('into', None)
660
    if into != None: kw_args['embeddable'] = True
661 2066 aaronmk
    recover = kw_args.pop('recover', None)
662
    cacheable = kw_args.pop('cacheable', True)
663 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
664 2066 aaronmk
665
    query, params = mk_insert_select(db, *args, **kw_args)
666 2386 aaronmk
    return run_query_into(db, query, params, into, recover=recover,
667 2673 aaronmk
        cacheable=cacheable, log_level=log_level)
668 2063 aaronmk
669 2738 aaronmk
default = sql_gen.default # tells insert() to use the default value for a column
670 2066 aaronmk
671 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
672 2085 aaronmk
    '''For params, see insert_select()'''
673 1960 aaronmk
    if lists.is_seq(row): cols = None
674
    else:
675
        cols = row.keys()
676
        row = row.values()
677 2738 aaronmk
    row = list(row) # ensure that "== []" works
678 1960 aaronmk
679 2738 aaronmk
    if row == []: query = None
680
    else: query = sql_gen.Values(row).to_str(db)
681 1961 aaronmk
682 2738 aaronmk
    return insert_select(db, table, cols, query, [], *args, **kw_args)
683 11 aaronmk
684 2402 aaronmk
def mk_update(db, table, changes=None, cond=None):
685
    '''
686
    @param changes [(col, new_value),...]
687
        * container can be any iterable type
688
        * col: sql_gen.Code|str (for col name)
689
        * new_value: sql_gen.Code|literal value
690
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
691
    @return str query
692
    '''
693
    query = 'UPDATE '+sql_gen.as_Table(table).to_str(db)+'\nSET\n'
694 2405 aaronmk
    query += ',\n'.join((sql_gen.to_name_only_col(col, table).to_str(db)+' = '
695 2402 aaronmk
        +sql_gen.as_Value(new_value).to_str(db) for col, new_value in changes))
696 2467 aaronmk
    if cond != None: query += '\nWHERE\n'+cond.to_str(db)
697 2402 aaronmk
698
    return query
699
700
def update(db, *args, **kw_args):
701
    '''For params, see mk_update() and run_query()'''
702
    recover = kw_args.pop('recover', None)
703
704
    return run_query(db, mk_update(db, *args, **kw_args), [], recover)
705
706 135 aaronmk
def last_insert_id(db):
707 1849 aaronmk
    module = util.root_module(db.db)
708 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
709
    elif module == 'MySQLdb': return db.insert_id()
710
    else: return None
711 13 aaronmk
712 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
713 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
714 2415 aaronmk
    to names that will be distinct among the columns' tables.
715 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
716 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
717
    @param into The table for the new columns.
718 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
719
        columns will be included in the mapping even if they are not in cols.
720
        The tables of the provided Col objects will be changed to into, so make
721
        copies of them if you want to keep the original tables.
722
    @param as_items Whether to return a list of dict items instead of a dict
723 2383 aaronmk
    @return dict(orig_col=new_col, ...)
724
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
725 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
726
        * All mappings use the into table so its name can easily be
727 2383 aaronmk
          changed for all columns at once
728
    '''
729 2415 aaronmk
    cols = lists.uniqify(cols)
730
731 2394 aaronmk
    items = []
732 2389 aaronmk
    for col in preserve:
733 2390 aaronmk
        orig_col = copy.copy(col)
734 2392 aaronmk
        col.table = into
735 2394 aaronmk
        items.append((orig_col, col))
736
    preserve = set(preserve)
737
    for col in cols:
738 2716 aaronmk
        if col not in preserve:
739
            items.append((col, sql_gen.Col(str(col), into, col.srcs)))
740 2394 aaronmk
741
    if not as_items: items = dict(items)
742
    return items
743 2383 aaronmk
744 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
745 2391 aaronmk
    '''For params, see mk_flatten_mapping()
746
    @return See return value of mk_flatten_mapping()
747
    '''
748 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
749
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
750 2786 aaronmk
    run_query_into(db, mk_select(db, joins, cols, limit=limit, start=start),
751 2392 aaronmk
        into=into)
752 2394 aaronmk
    return dict(items)
753 2391 aaronmk
754 2751 aaronmk
def mk_track_data_error(db, errors_table, cols, value, error_code, error):
755
    cols = map(sql_gen.to_name_only_col, cols)
756
757
    columns_cols = ['column']
758
    columns = sql_gen.NamedValues('columns', columns_cols,
759
        [[c.name] for c in cols])
760
    values_cols = ['value', 'error_code', 'error']
761
    values = sql_gen.NamedValues('values', values_cols,
762
        [value, error_code, error])
763
764
    select_cols = columns_cols+values_cols
765
    name_only_cols = map(sql_gen.to_name_only_col, select_cols)
766 2755 aaronmk
    errors_table = sql_gen.NamedTable('errors', errors_table)
767 2751 aaronmk
    joins = [columns, sql_gen.Join(values, type_='CROSS'),
768
        sql_gen.Join(errors_table, dict(zip(name_only_cols, select_cols)),
769
        sql_gen.filter_out)]
770
771
    return mk_insert_select(db, errors_table, name_only_cols,
772 2786 aaronmk
        mk_select(db, joins, select_cols, order_by=None))[0]
773 2751 aaronmk
774 2758 aaronmk
def track_data_error(db, errors_table, *args, **kw_args):
775
    '''
776
    @param errors_table If None, does nothing.
777
    '''
778
    if errors_table == None: return
779
    run_query(db, mk_track_data_error(db, errors_table, *args, **kw_args),
780
        cacheable=True, log_level=4)
781
782 2704 aaronmk
def cast(db, type_, col, errors_table=None):
783
    '''Casts an (unrenamed) column or value.
784 2769 aaronmk
    If errors_table set and col has srcs, saves errors in errors_table (using
785
    col's srcs attr as the source columns) and converts errors to warnings.
786 2704 aaronmk
    @param col sql_gen.Col|sql_gen.Literal
787
    @param errors_table None|sql_gen.Table|str
788
    '''
789 2769 aaronmk
    save_errors = (errors_table != None and isinstance(col, sql_gen.Col)
790
        and col.srcs != ())
791
    if not save_errors: # can't save errors
792
        return sql_gen.CustomCode(col.to_str(db)+'::'+type_) # just cast
793 2704 aaronmk
794
    assert not isinstance(col, sql_gen.NamedCol)
795
796 2770 aaronmk
    errors_table = sql_gen.as_Table(errors_table)
797
    srcs = map(sql_gen.to_name_only_col, col.srcs)
798
    function_name = str(sql_gen.FunctionCall(type_, *srcs))
799 2704 aaronmk
    function = sql_gen.TempFunction(function_name, db.autocommit)
800
801 2718 aaronmk
    while True:
802
        # Create function definition
803
        query = '''\
804 2704 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''(value text)
805
RETURNS '''+type_+'''
806
LANGUAGE plpgsql
807
STRICT
808
AS $$
809
BEGIN
810
    /* The explicit cast to the return type is needed to make the cast happen
811
    inside the try block. (Implicit casts to the return type happen at the end
812
    of the function, outside any block.) */
813
    RETURN value::'''+type_+''';
814
EXCEPTION
815
    WHEN data_exception THEN
816
        -- Save error in errors table.
817 2751 aaronmk
        -- Insert the value and error for *each* source column.
818
'''+mk_track_data_error(db, errors_table, srcs,
819
    *map(sql_gen.CustomCode, ['value', 'SQLSTATE', 'SQLERRM']))+''';
820 2704 aaronmk
821
        RAISE WARNING '%', SQLERRM;
822
        RETURN NULL;
823
END;
824
$$;
825
'''
826 2718 aaronmk
827
        # Create function
828
        try:
829
            run_query(db, query, recover=True, cacheable=True,
830
                log_ignore_excs=(DuplicateFunctionException,))
831
            break # successful
832
        except DuplicateFunctionException:
833 2770 aaronmk
            function.name = next_version(function.name)
834
            # try again with next version of name
835 2704 aaronmk
836
    return sql_gen.FunctionCall(function, col)
837
838 2414 aaronmk
##### Database structure queries
839
840 2426 aaronmk
def table_row_count(db, table, recover=None):
841 2786 aaronmk
    return value(run_query(db, mk_select(db, table, [sql_gen.row_count],
842 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
843 2426 aaronmk
844 2414 aaronmk
def table_cols(db, table, recover=None):
845
    return list(col_names(select(db, table, limit=0, order_by=None,
846 2443 aaronmk
        recover=recover, log_level=4)))
847 2414 aaronmk
848 2291 aaronmk
def pkey(db, table, recover=None):
849 832 aaronmk
    '''Assumed to be first column in table'''
850 2339 aaronmk
    return table_cols(db, table, recover)[0]
851 832 aaronmk
852 2559 aaronmk
not_null_col = 'not_null_col'
853 2340 aaronmk
854
def table_not_null_col(db, table, recover=None):
855
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
856
    if not_null_col in table_cols(db, table, recover): return not_null_col
857
    else: return pkey(db, table, recover)
858
859 853 aaronmk
def index_cols(db, table, index):
860
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
861
    automatically created. When you don't know whether something is a UNIQUE
862
    constraint or a UNIQUE index, use this function.'''
863 1909 aaronmk
    module = util.root_module(db.db)
864
    if module == 'psycopg2':
865
        return list(values(run_query(db, '''\
866 853 aaronmk
SELECT attname
867 866 aaronmk
FROM
868
(
869
        SELECT attnum, attname
870
        FROM pg_index
871
        JOIN pg_class index ON index.oid = indexrelid
872
        JOIN pg_class table_ ON table_.oid = indrelid
873
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
874
        WHERE
875 2782 aaronmk
            table_.relname = '''+db.esc_value(table)+'''
876
            AND index.relname = '''+db.esc_value(index)+'''
877 866 aaronmk
    UNION
878
        SELECT attnum, attname
879
        FROM
880
        (
881
            SELECT
882
                indrelid
883
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
884
                    AS indkey
885
            FROM pg_index
886
            JOIN pg_class index ON index.oid = indexrelid
887
            JOIN pg_class table_ ON table_.oid = indrelid
888
            WHERE
889 2782 aaronmk
                table_.relname = '''+db.esc_value(table)+'''
890
                AND index.relname = '''+db.esc_value(index)+'''
891 866 aaronmk
        ) s
892
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
893
) s
894 853 aaronmk
ORDER BY attnum
895 2782 aaronmk
'''
896
            , cacheable=True, log_level=4)))
897 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
898
        ' database')
899 853 aaronmk
900 464 aaronmk
def constraint_cols(db, table, constraint):
901 1849 aaronmk
    module = util.root_module(db.db)
902 464 aaronmk
    if module == 'psycopg2':
903
        return list(values(run_query(db, '''\
904
SELECT attname
905
FROM pg_constraint
906
JOIN pg_class ON pg_class.oid = conrelid
907
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
908
WHERE
909 2783 aaronmk
    relname = '''+db.esc_value(table)+'''
910
    AND conname = '''+db.esc_value(constraint)+'''
911 464 aaronmk
ORDER BY attnum
912 2783 aaronmk
'''
913
            )))
914 464 aaronmk
    else: raise NotImplementedError("Can't list constraint columns for "+module+
915
        ' database')
916
917 2096 aaronmk
row_num_col = '_row_num'
918
919 2688 aaronmk
def add_index(db, exprs, table=None, unique=False):
920
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
921 2538 aaronmk
    Currently, only function calls are supported as expressions.
922
    '''
923 2688 aaronmk
    if not lists.is_seq(exprs): exprs = [exprs]
924 2538 aaronmk
925 2688 aaronmk
    # Parse exprs
926
    old_exprs = exprs[:]
927
    exprs = []
928
    cols = []
929
    for i, expr in enumerate(old_exprs):
930
        expr = copy.deepcopy(expr) # don't modify input!
931
        expr = sql_gen.as_Col(expr)
932
933
        # Extract col
934
        if isinstance(expr, sql_gen.FunctionCall):
935
            col = expr.args[0]
936
            expr = sql_gen.Expr(expr)
937
        else: col = expr
938
939
        # Extract table
940
        if table == None:
941
            assert sql_gen.is_table_col(col)
942
            table = col.table
943
944
        col.table = None
945
946
        exprs.append(expr)
947
        cols.append(col)
948 2408 aaronmk
949 2688 aaronmk
    table = sql_gen.as_Table(table)
950
    index = sql_gen.Table(str(sql_gen.Col(','.join(map(str, cols)), table)))
951
952
    str_ = 'CREATE'
953
    if unique: str_ += ' UNIQUE'
954
    str_ += ' INDEX '+index.to_str(db)+' ON '+table.to_str(db)+' ('+(
955
        ', '.join((v.to_str(db) for v in exprs)))+')'
956
957
    try: run_query(db, str_, recover=True, cacheable=True, log_level=3)
958 2408 aaronmk
    except DuplicateTableException: pass # index already existed
959
960 2686 aaronmk
def add_pkey(db, table, cols=None, recover=None):
961
    '''Adds a primary key.
962
    @param cols [sql_gen.Col,...] The columns in the primary key.
963
        Defaults to the first column in the table.
964 2406 aaronmk
    @pre The table must not already have a primary key.
965
    '''
966
    table = sql_gen.as_Table(table)
967 2686 aaronmk
    if cols == None: cols = [pkey(db, table, recover)]
968
    col_strs = [sql_gen.to_name_only_col(v).to_str(db) for v in cols]
969 2406 aaronmk
970 2781 aaronmk
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD PRIMARY KEY ('
971
        +(', '.join(col_strs))+')', recover=True, cacheable=True, log_level=3,
972
        log_ignore_excs=(DuplicateTableException,))
973 2406 aaronmk
974 2772 aaronmk
already_indexed = object() # tells add_indexes() the pkey has already been added
975
976
def add_indexes(db, table, has_pkey=True):
977
    '''Adds an index on all columns in a table.
978
    @param has_pkey bool|already_indexed Whether a pkey instead of a regular
979
        index should be added on the first column.
980
        * If already_indexed, the pkey is assumed to have already been added
981
    '''
982
    cols = table_cols(db, table)
983
    if has_pkey:
984
        if has_pkey is not already_indexed: add_pkey(db, table)
985
        cols = cols[1:]
986
    for col in cols: add_index(db, col, table)
987
988 2086 aaronmk
def add_row_num(db, table):
989 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
990
    be the primary key.'''
991 2320 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
992 2096 aaronmk
    run_query(db, 'ALTER TABLE '+table+' ADD COLUMN '+row_num_col
993 2443 aaronmk
        +' serial NOT NULL PRIMARY KEY', log_level=3)
994 2086 aaronmk
995 2760 aaronmk
def create_table(db, table, cols, has_pkey=True, col_indexes=True):
996 2675 aaronmk
    '''Creates a table.
997 2681 aaronmk
    @param cols [sql_gen.TypedCol,...] The column names and types
998
    @param has_pkey If set, the first column becomes the primary key.
999 2760 aaronmk
    @param col_indexes bool|[ref]
1000
        * If True, indexes will be added on all non-pkey columns.
1001
        * If a list reference, [0] will be set to a function to do this.
1002
          This can be used to delay index creation until the table is populated.
1003 2675 aaronmk
    '''
1004
    table = sql_gen.as_Table(table)
1005
1006 2681 aaronmk
    if has_pkey:
1007
        cols[0] = pkey = copy.copy(cols[0]) # don't modify input!
1008
        pkey.type += ' NOT NULL PRIMARY KEY'
1009
1010 2675 aaronmk
    str_ = 'CREATE TABLE '+table.to_str(db)+' (\n'
1011 2681 aaronmk
    str_ += '\n, '.join(v.to_str(db) for v in cols)
1012
    str_ += '\n);\n'
1013 2675 aaronmk
    run_query(db, str_, cacheable=True, log_level=2)
1014 2760 aaronmk
1015
    # Add indexes
1016 2773 aaronmk
    if has_pkey: has_pkey = already_indexed
1017
    def add_indexes_(): add_indexes(db, table, has_pkey)
1018
    if isinstance(col_indexes, list): col_indexes[0] = add_indexes_ # defer
1019
    elif col_indexes: add_indexes_() # add now
1020 2675 aaronmk
1021 2684 aaronmk
def vacuum(db, table):
1022
    table = sql_gen.as_Table(table)
1023
    db.with_autocommit(lambda: run_query(db, 'VACUUM ANALYZE '+table.to_str(db),
1024
        log_level=3))
1025
1026 2732 aaronmk
def truncate(db, table, schema='public'):
1027 2777 aaronmk
    table = sql_gen.as_Table(table, schema)
1028 2733 aaronmk
    return run_query(db, 'TRUNCATE '+table.to_str(db)+' CASCADE')
1029 2732 aaronmk
1030 2729 aaronmk
def tables(db, schema_like='public', table_like='%', exact=False):
1031
    if exact: compare = '='
1032
    else: compare = 'LIKE'
1033
1034 1849 aaronmk
    module = util.root_module(db.db)
1035 832 aaronmk
    if module == 'psycopg2':
1036 2779 aaronmk
        conds = [('schemaname', sql_gen.CompareCond(schema_like, compare)),
1037
            ('tablename', sql_gen.CompareCond(table_like, compare))]
1038
        return values(select(db, 'pg_tables', ['tablename'], conds,
1039
            order_by='tablename', log_level=4))
1040 1968 aaronmk
    elif module == 'MySQLdb':
1041 2779 aaronmk
        return values(run_query(db, 'SHOW TABLES LIKE '+db.esc_value(table_like)
1042
            , cacheable=True, log_level=4))
1043 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
1044 830 aaronmk
1045 2726 aaronmk
def table_exists(db, table):
1046
    table = sql_gen.as_Table(table)
1047 2730 aaronmk
    return list(tables(db, table.schema, table.name, exact=True)) != []
1048 2726 aaronmk
1049
def errors_table(db, table, if_exists=True):
1050
    '''
1051
    @param if_exists If set, returns None if the errors table doesn't exist
1052
    @return None|sql_gen.Table
1053
    '''
1054
    table = sql_gen.as_Table(table)
1055
    if table.srcs != (): table = table.srcs[0]
1056
1057
    errors_table = sql_gen.suffixed_table(table, '.errors')
1058
    if if_exists and not table_exists(db, errors_table): return None
1059
    return errors_table
1060
1061 833 aaronmk
##### Database management
1062
1063 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
1064
    '''For kw_args, see tables()'''
1065
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
1066 833 aaronmk
1067 832 aaronmk
##### Heuristic queries
1068
1069 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
1070 1554 aaronmk
    '''Recovers from errors.
1071 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
1072
    '''
1073 2642 aaronmk
    row = sql_gen.ColDict(db, table, row)
1074 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
1075
1076 471 aaronmk
    try:
1077 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
1078 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
1079
            row_ct_ref[0] += cur.rowcount
1080
        return value(cur)
1081 471 aaronmk
    except DuplicateKeyException, e:
1082 2653 aaronmk
        row = sql_gen.ColDict(db, table,
1083
            util.dict_subset_right_join(row, e.cols))
1084
        return value(select(db, table, [pkey_], row, recover=True))
1085 471 aaronmk
1086 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
1087 830 aaronmk
    '''Recovers from errors'''
1088 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
1089 14 aaronmk
    except StopIteration:
1090 40 aaronmk
        if not create: raise
1091 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
1092 2078 aaronmk
1093 2593 aaronmk
def is_func_result(col):
1094
    return col.table.name.find('(') >= 0 and col.name == 'result'
1095
1096 2592 aaronmk
def into_table_name(out_table, in_tables0, mapping, is_func):
1097 2615 aaronmk
    def in_col_str(in_col):
1098
        in_col = sql_gen.remove_col_rename(in_col)
1099
        if isinstance(in_col, sql_gen.Col):
1100
            table = in_col.table
1101
            if table == in_tables0:
1102
                in_col = sql_gen.to_name_only_col(in_col)
1103
            elif is_func_result(in_col): in_col = table # omit col name
1104
        return str(in_col)
1105
1106 2580 aaronmk
    str_ = str(out_table)
1107
    if is_func:
1108 2615 aaronmk
        str_ += '('
1109 2580 aaronmk
1110 2615 aaronmk
        try: value_in_col = mapping['value']
1111
        except KeyError:
1112 2654 aaronmk
            str_ += ', '.join((str(k)+'='+in_col_str(v)
1113 2615 aaronmk
                for k, v in mapping.iteritems()))
1114
        else: str_ += in_col_str(value_in_col)
1115
1116
        str_ += ')'
1117 2580 aaronmk
    else: str_ += '_pkeys'
1118
    return str_
1119
1120 2508 aaronmk
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None, into=None,
1121 2552 aaronmk
    default=None, is_func=False):
1122 2078 aaronmk
    '''Recovers from errors.
1123
    Only works under PostgreSQL (uses INSERT RETURNING).
1124 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
1125
        tables to join with it using the main input table's pkey
1126 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
1127 2612 aaronmk
        * out_table_col: str (*not* sql_gen.Col)
1128
        * in_table_col: sql_gen.Col|literal-value
1129 2489 aaronmk
    @param into The table to contain the output and input pkeys.
1130 2574 aaronmk
        Defaults to `out_table.name+'_pkeys'`.
1131 2509 aaronmk
    @param default The *output* column to use as the pkey for missing rows.
1132
        If this output column does not exist in the mapping, uses None.
1133 2552 aaronmk
    @param is_func Whether out_table is the name of a SQL function, not a table
1134 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
1135 2078 aaronmk
    '''
1136 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
1137 2312 aaronmk
1138 2450 aaronmk
    def log_debug(msg): db.log_debug(msg, level=1.5)
1139 2505 aaronmk
    def col_ustr(str_):
1140 2567 aaronmk
        return strings.repr_no_u(sql_gen.remove_col_rename(str_))
1141 2450 aaronmk
1142 2768 aaronmk
    out_pkey = pkey(db, out_table, recover=True)
1143
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
1144
1145
    if mapping == {}: # need at least one column for INSERT SELECT
1146
        mapping = {out_pkey: None} # ColDict will replace with default value
1147
1148 2486 aaronmk
    log_debug('********** New iteration **********')
1149 2505 aaronmk
    log_debug('Inserting these input columns into '+strings.as_tt(
1150
        out_table.to_str(db))+':\n'+strings.as_table(mapping, ustr=col_ustr))
1151 2463 aaronmk
1152 2382 aaronmk
    # Create input joins from list of input tables
1153
    in_tables_ = in_tables[:] # don't modify input!
1154
    in_tables0 = in_tables_.pop(0) # first table is separate
1155 2735 aaronmk
    errors_table_ = errors_table(db, in_tables0)
1156 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
1157 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
1158 2460 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v,
1159
        {in_pkey: sql_gen.join_same_not_null}) for v in in_tables_]
1160 2131 aaronmk
1161 2592 aaronmk
    if into == None:
1162
        into = into_table_name(out_table, in_tables0, mapping, is_func)
1163
    into = sql_gen.as_Table(into)
1164
1165 2702 aaronmk
    # Set column sources
1166
    in_cols = filter(sql_gen.is_table_col, mapping.values())
1167
    for col in in_cols:
1168
        if col.table == in_tables0: col.set_srcs(sql_gen.src_self)
1169
1170 2486 aaronmk
    log_debug('Joining together input tables into temp table')
1171 2395 aaronmk
    # Place in new table for speed and so don't modify input if values edited
1172 2584 aaronmk
    in_table = sql_gen.Table('in')
1173 2702 aaronmk
    mapping = dicts.join(mapping, flatten(db, in_table, input_joins, in_cols,
1174
        preserve=[in_pkey_col], start=0))
1175 2395 aaronmk
    input_joins = [in_table]
1176 2486 aaronmk
    db.log_debug('Temp table: '+strings.as_tt(in_table.to_str(db)), level=2)
1177 2395 aaronmk
1178 2656 aaronmk
    mapping = sql_gen.ColDict(db, out_table, mapping)
1179
        # after applying dicts.join() because that returns a plain dict
1180
1181 2509 aaronmk
    # Resolve default value column
1182
    try: default = mapping[default]
1183
    except KeyError:
1184
        if default != None:
1185
            db.log_debug('Default value column '
1186
                +strings.as_tt(strings.repr_no_u(default))
1187 2511 aaronmk
                +' does not exist in mapping, falling back to None', level=2.1)
1188 2509 aaronmk
            default = None
1189
1190 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
1191 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
1192
1193 2201 aaronmk
    pkeys_table_exists_ref = [False]
1194 2420 aaronmk
    def insert_into_pkeys(joins, cols):
1195 2786 aaronmk
        query = mk_select(db, joins, cols, order_by=None, start=0)
1196 2201 aaronmk
        if pkeys_table_exists_ref[0]:
1197 2786 aaronmk
            insert_select(db, into, pkeys_names, query)
1198 2201 aaronmk
        else:
1199 2786 aaronmk
            run_query_into(db, query, into=into)
1200 2201 aaronmk
            pkeys_table_exists_ref[0] = True
1201
1202 2429 aaronmk
    limit_ref = [None]
1203 2380 aaronmk
    conds = set()
1204 2233 aaronmk
    distinct_on = []
1205 2325 aaronmk
    def mk_main_select(joins, cols):
1206 2429 aaronmk
        return mk_select(db, joins, cols, conds, distinct_on,
1207
            limit=limit_ref[0], start=0)
1208 2132 aaronmk
1209 2519 aaronmk
    exc_strs = set()
1210 2309 aaronmk
    def log_exc(e):
1211 2519 aaronmk
        e_str = exc.str_(e, first_line_only=True)
1212
        log_debug('Caught exception: '+e_str)
1213
        assert e_str not in exc_strs # avoid infinite loops
1214
        exc_strs.add(e_str)
1215 2735 aaronmk
1216 2451 aaronmk
    def remove_all_rows():
1217 2450 aaronmk
        log_debug('Returning NULL for all rows')
1218 2429 aaronmk
        limit_ref[0] = 0 # just create an empty pkeys table
1219 2735 aaronmk
1220 2758 aaronmk
    def ignore(in_col, value, e):
1221
        track_data_error(db, errors_table_, in_col.srcs, value, e.cause.pgcode,
1222
            e.cause.pgerror)
1223
1224 2545 aaronmk
        in_col_str = strings.as_tt(repr(in_col))
1225 2544 aaronmk
        db.log_debug('Adding index on '+in_col_str+' to enable fast filtering',
1226
            level=2.5)
1227 2537 aaronmk
        add_index(db, in_col)
1228 2735 aaronmk
1229 2545 aaronmk
        log_debug('Ignoring rows with '+in_col_str+' = '
1230
            +strings.as_tt(repr(value)))
1231 2758 aaronmk
    def remove_rows(in_col, value, e):
1232
        ignore(in_col, value, e)
1233 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
1234
        assert cond not in conds # avoid infinite loops
1235 2380 aaronmk
        conds.add(cond)
1236 2758 aaronmk
    def invalid2null(in_col, value, e):
1237
        ignore(in_col, value, e)
1238 2403 aaronmk
        update(db, in_table, [(in_col, None)],
1239
            sql_gen.ColValueCond(in_col, value))
1240 2245 aaronmk
1241 2589 aaronmk
    def insert_pkeys_table(which):
1242
        return sql_gen.Table(sql_gen.add_suffix(in_table.name,
1243
            '_insert_'+which+'_pkeys'))
1244
    insert_out_pkeys = insert_pkeys_table('out')
1245
    insert_in_pkeys = insert_pkeys_table('in')
1246
1247 2206 aaronmk
    # Do inserts and selects
1248 2642 aaronmk
    join_cols = sql_gen.ColDict(db, out_table)
1249 2206 aaronmk
    while True:
1250 2521 aaronmk
        if limit_ref[0] == 0: # special case
1251
            log_debug('Creating an empty pkeys table')
1252 2786 aaronmk
            cur = run_query_into(db, mk_select(db, out_table, [out_pkey],
1253 2521 aaronmk
                limit=limit_ref[0]), into=insert_out_pkeys)
1254
            break # don't do main case
1255
1256 2303 aaronmk
        has_joins = join_cols != {}
1257
1258 2305 aaronmk
        # Prepare to insert new rows
1259 2325 aaronmk
        insert_joins = input_joins[:] # don't modify original!
1260 2403 aaronmk
        insert_args = dict(recover=True, cacheable=False)
1261 2303 aaronmk
        if has_joins:
1262 2317 aaronmk
            distinct_on = [v.to_Col() for v in join_cols.values()]
1263 2325 aaronmk
            insert_joins.append(sql_gen.Join(out_table, join_cols,
1264
                sql_gen.filter_out))
1265
        else:
1266 2404 aaronmk
            insert_args.update(dict(returning=out_pkey, into=insert_out_pkeys))
1267 2786 aaronmk
        main_select = mk_main_select(insert_joins, mapping.values())
1268 2303 aaronmk
1269 2486 aaronmk
        log_debug('Trying to insert new rows')
1270 2206 aaronmk
        try:
1271 2518 aaronmk
            cur = insert_select(db, out_table, mapping.keys(), main_select,
1272
                **insert_args)
1273 2357 aaronmk
            break # insert successful
1274 2206 aaronmk
        except DuplicateKeyException, e:
1275 2309 aaronmk
            log_exc(e)
1276
1277 2258 aaronmk
            old_join_cols = join_cols.copy()
1278 2565 aaronmk
            join_cols.update(util.dict_subset_right_join(mapping, e.cols))
1279 2486 aaronmk
            log_debug('Ignoring existing rows, comparing on these columns:\n'
1280 2505 aaronmk
                +strings.as_inline_table(join_cols, ustr=col_ustr))
1281 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
1282 2230 aaronmk
        except NullValueException, e:
1283 2309 aaronmk
            log_exc(e)
1284
1285 2230 aaronmk
            out_col, = e.cols
1286
            try: in_col = mapping[out_col]
1287 2356 aaronmk
            except KeyError:
1288 2486 aaronmk
                log_debug('Missing mapping for NOT NULL column '+out_col)
1289 2451 aaronmk
                remove_all_rows()
1290 2758 aaronmk
            else: remove_rows(in_col, None, e)
1291 2542 aaronmk
        except FunctionValueException, e:
1292
            log_exc(e)
1293
1294
            func_name = e.name
1295
            value = e.value
1296
            for out_col, in_col in mapping.iteritems():
1297 2758 aaronmk
                in_col = sql_gen.unwrap_func_call(in_col, func_name)
1298
                invalid2null(in_col, value, e)
1299 2525 aaronmk
        except MissingCastException, e:
1300
            log_exc(e)
1301
1302
            out_col = e.col
1303 2706 aaronmk
            type_ = e.type
1304
1305 2724 aaronmk
            log_debug('Casting '+strings.as_tt(out_col)+' input to '
1306
                +strings.as_tt(type_))
1307 2735 aaronmk
            def wrap_func(col): return cast(db, type_, col, errors_table_)
1308 2705 aaronmk
            mapping[out_col] = sql_gen.wrap(wrap_func, mapping[out_col])
1309 2429 aaronmk
        except DatabaseErrors, e:
1310
            log_exc(e)
1311
1312 2531 aaronmk
            msg = 'No handler for exception: '+exc.str_(e)
1313 2451 aaronmk
            warnings.warn(DbWarning(msg))
1314
            log_debug(msg)
1315
            remove_all_rows()
1316 2358 aaronmk
        # after exception handled, rerun loop with additional constraints
1317 2132 aaronmk
1318 2357 aaronmk
    if row_ct_ref != None and cur.rowcount >= 0:
1319
        row_ct_ref[0] += cur.rowcount
1320
1321
    if has_joins:
1322
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
1323 2486 aaronmk
        log_debug('Getting output table pkeys of existing/inserted rows')
1324 2420 aaronmk
        insert_into_pkeys(select_joins, pkeys_cols)
1325 2357 aaronmk
    else:
1326 2404 aaronmk
        add_row_num(db, insert_out_pkeys) # for joining with input pkeys
1327 2357 aaronmk
1328 2486 aaronmk
        log_debug('Getting input table pkeys of inserted rows')
1329 2786 aaronmk
        run_query_into(db, mk_main_select(input_joins, [in_pkey]),
1330 2404 aaronmk
            into=insert_in_pkeys)
1331
        add_row_num(db, insert_in_pkeys) # for joining with output pkeys
1332 2357 aaronmk
1333 2428 aaronmk
        assert table_row_count(db, insert_out_pkeys) == table_row_count(db,
1334
            insert_in_pkeys)
1335
1336 2486 aaronmk
        log_debug('Combining output and input pkeys in inserted order')
1337 2404 aaronmk
        pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
1338 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
1339 2420 aaronmk
        insert_into_pkeys(pkey_joins, pkeys_names)
1340 2357 aaronmk
1341 2486 aaronmk
    db.log_debug('Adding pkey on pkeys table to enable fast joins', level=2.5)
1342 2594 aaronmk
    add_pkey(db, into)
1343 2407 aaronmk
1344 2508 aaronmk
    log_debug('Setting pkeys of missing rows to '+strings.as_tt(repr(default)))
1345 2489 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(into,
1346 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
1347
        # must use join_same_not_null or query will take forever
1348 2420 aaronmk
    insert_into_pkeys(missing_rows_joins,
1349 2508 aaronmk
        [in_pkey_col, sql_gen.NamedCol(out_pkey, default)])
1350 2357 aaronmk
1351 2489 aaronmk
    assert table_row_count(db, into) == table_row_count(db, in_table)
1352 2428 aaronmk
1353 2702 aaronmk
    srcs = []
1354
    if is_func: srcs = sql_gen.cols_srcs(in_cols)
1355
    return sql_gen.Col(out_pkey, into, srcs)
1356 2115 aaronmk
1357
##### Data cleanup
1358
1359 2290 aaronmk
def cleanup_table(db, table, cols):
1360 2677 aaronmk
    table = sql_gen.as_Table(table)
1361
    cols = map(sql_gen.as_Col, cols)
1362 2115 aaronmk
1363 2677 aaronmk
    expr = ('nullif(nullif(trim(both from %s), '+db.esc_value('')+'), '
1364
        +db.esc_value(r'\N')+')')
1365
    changes = [(v, sql_gen.CustomCode(expr % v.to_str(db)))
1366
        for v in cols]
1367 2115 aaronmk
1368 2677 aaronmk
    update(db, table, changes)