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 2763 aaronmk
            if params == [] or params == (): params = None
216
                # None turns off the mogrifier
217 2762 aaronmk
218 2764 aaronmk
            self._is_insert = query.startswith('INSERT')
219 1894 aaronmk
            self.query_lookup = _query_lookup(query, params)
220 2148 aaronmk
            try:
221 2191 aaronmk
                try:
222 2763 aaronmk
                    cur = self.inner.execute(query, params)
223 2191 aaronmk
                    self.outer.do_autocommit()
224 2648 aaronmk
                finally: self.query = get_cur_query(self.inner, query, params)
225 1904 aaronmk
            except Exception, e:
226 2170 aaronmk
                _add_cursor_info(e, self, query, params)
227 1904 aaronmk
                self.result = e # cache the exception as the result
228
                self._cache_result()
229
                raise
230 1930 aaronmk
            # Fetch all rows so result will be cached
231
            if self.rowcount == 0 and not self._is_insert: consume_rows(self)
232 2762 aaronmk
            return cur
233 1894 aaronmk
234 1891 aaronmk
        def fetchone(self):
235
            row = self.inner.fetchone()
236 1899 aaronmk
            if row != None: self.result.append(row)
237
            # otherwise, fetched all rows
238 1904 aaronmk
            else: self._cache_result()
239
            return row
240
241
        def _cache_result(self):
242 1906 aaronmk
            # For inserts, only cache exceptions since inserts are not
243
            # idempotent, but an invalid insert will always be invalid
244 1930 aaronmk
            if self.query_results != None and (not self._is_insert
245 1906 aaronmk
                or isinstance(self.result, Exception)):
246
247 1894 aaronmk
                assert self.query_lookup != None
248 1916 aaronmk
                self.query_results[self.query_lookup] = self.CacheCursor(
249
                    util.dict_subset(dicts.AttrsDictView(self),
250
                    ['query', 'result', 'rowcount', 'description']))
251 1906 aaronmk
252 1916 aaronmk
        class CacheCursor:
253
            def __init__(self, cached_result): self.__dict__ = cached_result
254
255 1927 aaronmk
            def execute(self, *args, **kw_args):
256 1916 aaronmk
                if isinstance(self.result, Exception): raise self.result
257
                # otherwise, result is a rows list
258
                self.iter = iter(self.result)
259
260
            def fetchone(self):
261
                try: return self.iter.next()
262
                except StopIteration: return None
263 1891 aaronmk
264 2212 aaronmk
    def esc_value(self, value):
265 2663 aaronmk
        try: str_ = self.mogrify('%s', [value])
266
        except NotImplementedError, e:
267
            module = util.root_module(self.db)
268
            if module == 'MySQLdb':
269
                import _mysql
270
                str_ = _mysql.escape_string(value)
271
            else: raise e
272 2374 aaronmk
        return strings.to_unicode(str_)
273 2212 aaronmk
274 2347 aaronmk
    def esc_name(self, name): return esc_name(self, name) # calls global func
275
276 2665 aaronmk
    def can_mogrify(self):
277 2663 aaronmk
        module = util.root_module(self.db)
278 2665 aaronmk
        return module == 'psycopg2'
279 2663 aaronmk
280 2665 aaronmk
    def mogrify(self, query, params=None):
281
        if self.can_mogrify(): return self.db.cursor().mogrify(query, params)
282
        else: raise NotImplementedError("Can't mogrify query")
283
284 2671 aaronmk
    def print_notices(self):
285 2725 aaronmk
        if hasattr(self.db, 'notices'):
286
            for msg in self.db.notices:
287
                if msg not in self._notices_seen:
288
                    self._notices_seen.add(msg)
289
                    self.log_debug(msg, level=2)
290 2671 aaronmk
291 2445 aaronmk
    def run_query(self, query, params=None, cacheable=False, log_level=2,
292 2464 aaronmk
        debug_msg_ref=None):
293 2445 aaronmk
        '''
294 2464 aaronmk
        @param log_ignore_excs The log_level will be increased by 2 if the query
295
            throws one of these exceptions.
296 2664 aaronmk
        @param debug_msg_ref If specified, the log message will be returned in
297
            this instead of being output. This allows you to filter log messages
298
            depending on the result of the query.
299 2445 aaronmk
        '''
300 2167 aaronmk
        assert query != None
301
302 2047 aaronmk
        if not self.caching: cacheable = False
303 1903 aaronmk
        used_cache = False
304 2664 aaronmk
305
        def log_msg(query):
306
            if used_cache: cache_status = 'cache hit'
307
            elif cacheable: cache_status = 'cache miss'
308
            else: cache_status = 'non-cacheable'
309
            return 'DB query: '+cache_status+':\n'+strings.as_code(query, 'SQL')
310
311 1903 aaronmk
        try:
312 1927 aaronmk
            # Get cursor
313
            if cacheable:
314
                query_lookup = _query_lookup(query, params)
315
                try:
316
                    cur = self.query_results[query_lookup]
317
                    used_cache = True
318
                except KeyError: cur = self.DbCursor(self)
319
            else: cur = self.db.cursor()
320
321 2664 aaronmk
            # Log query
322
            if self.debug and debug_msg_ref == None: # log before running
323
                self.log_debug(log_msg(query), log_level)
324
325 1927 aaronmk
            # Run query
326 2148 aaronmk
            cur.execute(query, params)
327 1903 aaronmk
        finally:
328 2671 aaronmk
            self.print_notices()
329 2664 aaronmk
            if self.debug and debug_msg_ref != None: # return after running
330
                debug_msg_ref[0] = log_msg(str(get_cur_query(cur, query,
331
                    params)))
332 1903 aaronmk
333
        return cur
334 1914 aaronmk
335
    def is_cached(self, query, params=None):
336
        return _query_lookup(query, params) in self.query_results
337 2139 aaronmk
338 2683 aaronmk
    def with_autocommit(self, func, autocommit=True):
339
        prev_autocommit = self.db.autocommit
340
        self.db.autocommit = autocommit
341
        try: return func()
342
        finally: self.db.autocommit = prev_autocommit
343
344 2139 aaronmk
    def with_savepoint(self, func):
345 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
346 2443 aaronmk
        self.run_query('SAVEPOINT '+savepoint, log_level=4)
347 2139 aaronmk
        self._savepoint += 1
348 2683 aaronmk
        try:
349 2139 aaronmk
            try: return_val = func()
350
            finally:
351
                self._savepoint -= 1
352
                assert self._savepoint >= 0
353
        except:
354 2443 aaronmk
            self.run_query('ROLLBACK TO SAVEPOINT '+savepoint, log_level=4)
355 2139 aaronmk
            raise
356
        else:
357 2443 aaronmk
            self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
358 2191 aaronmk
            self.do_autocommit()
359 2139 aaronmk
            return return_val
360 2191 aaronmk
361
    def do_autocommit(self):
362
        '''Autocommits if outside savepoint'''
363
        assert self._savepoint >= 0
364
        if self.autocommit and self._savepoint == 0:
365 2668 aaronmk
            self.log_debug('Autocommitting')
366 2191 aaronmk
            self.db.commit()
367 2643 aaronmk
368
    def col_default(self, col):
369
        table = sql_gen.Table('columns', 'information_schema')
370
371
        conds = [('table_name', col.table.name), ('column_name', col.name)]
372
        schema = col.table.schema
373
        if schema != None: conds.append(('table_schema', schema))
374
375 2660 aaronmk
        return sql_gen.as_Code(value(select(self, table, ['column_default'],
376
            conds, order_by='table_schema', limit=1, log_level=3)))
377 2643 aaronmk
            # TODO: order_by search_path schema order
378 1849 aaronmk
379 1869 aaronmk
connect = DbConn
380
381 832 aaronmk
##### Querying
382
383 1894 aaronmk
def run_raw_query(db, *args, **kw_args):
384 2085 aaronmk
    '''For params, see DbConn.run_query()'''
385 1894 aaronmk
    return db.run_query(*args, **kw_args)
386 11 aaronmk
387 2068 aaronmk
def mogrify(db, query, params):
388
    module = util.root_module(db.db)
389
    if module == 'psycopg2': return db.db.cursor().mogrify(query, params)
390
    else: raise NotImplementedError("Can't mogrify query for "+module+
391
        ' database')
392
393 832 aaronmk
##### Recoverable querying
394 15 aaronmk
395 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
396 11 aaronmk
397 2464 aaronmk
def run_query(db, query, params=None, recover=None, cacheable=False,
398
    log_level=2, log_ignore_excs=None, **kw_args):
399 2441 aaronmk
    '''For params, see run_raw_query()'''
400 830 aaronmk
    if recover == None: recover = False
401 2464 aaronmk
    if log_ignore_excs == None: log_ignore_excs = ()
402
    log_ignore_excs = tuple(log_ignore_excs)
403 830 aaronmk
404 2666 aaronmk
    debug_msg_ref = None # usually, db.run_query() logs query before running it
405
    # But if filtering with log_ignore_excs, wait until after exception parsing
406
    if log_ignore_excs != () or not db.can_mogrify(): debug_msg_ref = [None]
407
408 2148 aaronmk
    try:
409 2464 aaronmk
        try:
410
            def run(): return run_raw_query(db, query, params, cacheable,
411
                log_level, debug_msg_ref, **kw_args)
412
            if recover and not db.is_cached(query, params):
413
                return with_savepoint(db, run)
414
            else: return run() # don't need savepoint if cached
415
        except Exception, e:
416
            if not recover: raise # need savepoint to run index_cols()
417
            msg = exc.str_(e)
418
419
            match = re.search(r'duplicate key value violates unique constraint '
420 2493 aaronmk
                r'"((_?[^\W_]+)_.+?)"', msg)
421 2464 aaronmk
            if match:
422
                constraint, table = match.groups()
423
                try: cols = index_cols(db, table, constraint)
424
                except NotImplementedError: raise e
425
                else: raise DuplicateKeyException(constraint, cols, e)
426
427 2493 aaronmk
            match = re.search(r'null value in column "(.+?)" violates not-null'
428 2464 aaronmk
                r' constraint', msg)
429
            if match: raise NullValueException('NOT NULL', [match.group(1)], e)
430
431
            match = re.search(r'\b(?:invalid input (?:syntax|value)\b.*?'
432
                r'|date/time field value out of range): "(.+?)"\n'
433 2535 aaronmk
                r'(?:(?s).*?)\bfunction "(.+?)"', msg)
434 2464 aaronmk
            if match:
435
                value, name = match.groups()
436
                raise FunctionValueException(name, strings.to_unicode(value), e)
437
438 2526 aaronmk
            match = re.search(r'column "(.+?)" is of type (.+?) but expression '
439 2523 aaronmk
                r'is of type', msg)
440
            if match:
441
                col, type_ = match.groups()
442
                raise MissingCastException(type_, col, e)
443
444 2493 aaronmk
            match = re.search(r'relation "(.+?)" already exists', msg)
445 2464 aaronmk
            if match: raise DuplicateTableException(match.group(1), e)
446
447 2493 aaronmk
            match = re.search(r'function "(.+?)" already exists', msg)
448 2464 aaronmk
            if match: raise DuplicateFunctionException(match.group(1), e)
449
450
            raise # no specific exception raised
451
    except log_ignore_excs:
452
        log_level += 2
453
        raise
454
    finally:
455 2666 aaronmk
        if debug_msg_ref != None and debug_msg_ref[0] != None:
456
            db.log_debug(debug_msg_ref[0], log_level)
457 830 aaronmk
458 832 aaronmk
##### Basic queries
459
460 2153 aaronmk
def next_version(name):
461 2163 aaronmk
    version = 1 # first existing name was version 0
462 2586 aaronmk
    match = re.match(r'^(.*)#(\d+)$', name)
463 2153 aaronmk
    if match:
464 2586 aaronmk
        name, version = match.groups()
465
        version = int(version)+1
466 2588 aaronmk
    return sql_gen.add_suffix(name, '#'+str(version))
467 2153 aaronmk
468 2774 aaronmk
def run_query_into(db, query, params, into=None, add_indexes_=False, **kw_args):
469 2085 aaronmk
    '''Outputs a query to a temp table.
470
    For params, see run_query().
471
    '''
472 2774 aaronmk
    if into == None: return run_query(db, query, params, **kw_args)
473 2085 aaronmk
    else: # place rows in temp table
474 2386 aaronmk
        assert isinstance(into, sql_gen.Table)
475 2385 aaronmk
476 2153 aaronmk
        kw_args['recover'] = True
477 2464 aaronmk
        kw_args.setdefault('log_ignore_excs', (DuplicateTableException,))
478 2440 aaronmk
479 2468 aaronmk
        temp = not db.autocommit # tables are permanent in autocommit mode
480 2440 aaronmk
        # "temporary tables cannot specify a schema name", so remove schema
481
        if temp: into.schema = None
482
483 2774 aaronmk
        # Create table
484 2153 aaronmk
        while True:
485 2774 aaronmk
            create_query = 'CREATE'
486
            if temp: create_query += ' TEMP'
487
            create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
488
489 2153 aaronmk
            try:
490 2774 aaronmk
                cur = run_query(db, create_query, params, **kw_args)
491 2153 aaronmk
                    # CREATE TABLE AS sets rowcount to # rows in query
492 2774 aaronmk
                break
493 2153 aaronmk
            except DuplicateTableException, e:
494 2386 aaronmk
                into.name = next_version(into.name)
495 2153 aaronmk
                # try again with next version of name
496 2774 aaronmk
497
        if add_indexes_: add_indexes(db, into)
498
499
        return cur
500 2085 aaronmk
501 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
502
503 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
504
505 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
506 2293 aaronmk
    start=None, order_by=order_by_pkey, default_table=None):
507 1981 aaronmk
    '''
508 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
509 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
510 1981 aaronmk
    @param fields Use None to select all fields in the table
511 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
512 2379 aaronmk
        * container can be any iterable type
513 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
514
        * compare_right_side: sql_gen.ValueCond|literal value
515 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
516
        use all columns
517 2054 aaronmk
    @return tuple(query, params)
518 1981 aaronmk
    '''
519 2315 aaronmk
    # Parse tables param
520 2121 aaronmk
    if not lists.is_seq(tables): tables = [tables]
521 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
522 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
523 2121 aaronmk
524 2315 aaronmk
    # Parse other params
525 2376 aaronmk
    if conds == None: conds = []
526 2650 aaronmk
    elif dicts.is_dict(conds): conds = conds.items()
527 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
528 135 aaronmk
    assert limit == None or type(limit) == int
529 865 aaronmk
    assert start == None or type(start) == int
530 2315 aaronmk
    if order_by is order_by_pkey:
531
        if distinct_on != []: order_by = None
532
        else: order_by = pkey(db, table0, recover=True)
533 865 aaronmk
534 2315 aaronmk
    query = 'SELECT'
535 2056 aaronmk
536 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
537 2056 aaronmk
538 2200 aaronmk
    # DISTINCT ON columns
539 2233 aaronmk
    if distinct_on != []:
540 2467 aaronmk
        query += '\nDISTINCT'
541 2254 aaronmk
        if distinct_on is not distinct_on_all:
542 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
543
544
    # Columns
545 2467 aaronmk
    query += '\n'
546 1135 aaronmk
    if fields == None: query += '*'
547 2765 aaronmk
    else:
548
        assert fields != []
549
        query += '\n, '.join(map(parse_col, fields))
550 2200 aaronmk
551
    # Main table
552 2467 aaronmk
    query += '\nFROM '+table0.to_str(db)
553 865 aaronmk
554 2122 aaronmk
    # Add joins
555 2271 aaronmk
    left_table = table0
556 2263 aaronmk
    for join_ in tables:
557
        table = join_.table
558 2238 aaronmk
559 2343 aaronmk
        # Parse special values
560
        if join_.type_ is sql_gen.filter_out: # filter no match
561 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
562
                None))
563 2343 aaronmk
564 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
565 2122 aaronmk
566
        left_table = table
567
568 865 aaronmk
    missing = True
569 2376 aaronmk
    if conds != []:
570 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
571
        else: whitespace = '\n'
572 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
573
            .to_str(db) for l, r in conds], 'WHERE')
574 865 aaronmk
        missing = False
575 2227 aaronmk
    if order_by != None:
576 2467 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
577
    if limit != None: query += '\nLIMIT '+str(limit); missing = False
578 865 aaronmk
    if start != None:
579 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
580 865 aaronmk
        missing = False
581
    if missing: warnings.warn(DbWarning(
582
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
583
584 2315 aaronmk
    return (query, [])
585 11 aaronmk
586 2054 aaronmk
def select(db, *args, **kw_args):
587
    '''For params, see mk_select() and run_query()'''
588
    recover = kw_args.pop('recover', None)
589
    cacheable = kw_args.pop('cacheable', True)
590 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
591 2054 aaronmk
592
    query, params = mk_select(db, *args, **kw_args)
593 2442 aaronmk
    return run_query(db, query, params, recover, cacheable, log_level=log_level)
594 2054 aaronmk
595 2066 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, params=None,
596 2292 aaronmk
    returning=None, embeddable=False):
597 1960 aaronmk
    '''
598
    @param returning str|None An inserted column (such as pkey) to return
599 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
600 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
601
        query will be fully cached, not just if it raises an exception.
602 1960 aaronmk
    '''
603 2754 aaronmk
    table = sql_gen.remove_table_rename(sql_gen.as_Table(table))
604 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
605 2646 aaronmk
    if cols != None:
606
        cols = [sql_gen.to_name_only_col(v, table).to_str(db) for v in cols]
607 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
608 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
609 2063 aaronmk
610
    # Build query
611 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
612
    query = first_line
613 2467 aaronmk
    if cols != None: query += '\n('+', '.join(cols)+')'
614
    query += '\n'+select_query
615 2063 aaronmk
616
    if returning != None:
617 2743 aaronmk
        query += '\nRETURNING '+sql_gen.to_name_only_col(returning).to_str(db)
618 2063 aaronmk
619 2070 aaronmk
    if embeddable:
620 2327 aaronmk
        assert returning != None
621
622 2070 aaronmk
        # Create function
623 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
624 2327 aaronmk
        return_type = 'SETOF '+returning.to_str(db)+'%TYPE'
625 2189 aaronmk
        while True:
626
            try:
627 2698 aaronmk
                function = sql_gen.TempFunction(function_name, db.autocommit)
628 2194 aaronmk
629 2189 aaronmk
                function_query = '''\
630 2698 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''()
631 2467 aaronmk
RETURNS '''+return_type+'''
632
LANGUAGE sql
633
AS $$
634
'''+mogrify(db, query, params)+''';
635
$$;
636 2070 aaronmk
'''
637 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
638 2464 aaronmk
                    log_ignore_excs=(DuplicateFunctionException,))
639 2189 aaronmk
                break # this version was successful
640
            except DuplicateFunctionException, e:
641
                function_name = next_version(function_name)
642
                # try again with next version of name
643 2070 aaronmk
644 2337 aaronmk
        # Return query that uses function
645 2698 aaronmk
        func_table = sql_gen.NamedTable('f', sql_gen.FunctionCall(function),
646 2742 aaronmk
            [returning]) # AS clause requires function alias
647 2337 aaronmk
        return mk_select(db, func_table, start=0, order_by=None)
648 2070 aaronmk
649 2066 aaronmk
    return (query, params)
650
651
def insert_select(db, *args, **kw_args):
652 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
653 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
654
        values in
655 2072 aaronmk
    '''
656 2386 aaronmk
    into = kw_args.pop('into', None)
657
    if into != None: kw_args['embeddable'] = True
658 2066 aaronmk
    recover = kw_args.pop('recover', None)
659
    cacheable = kw_args.pop('cacheable', True)
660 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
661 2066 aaronmk
662
    query, params = mk_insert_select(db, *args, **kw_args)
663 2386 aaronmk
    return run_query_into(db, query, params, into, recover=recover,
664 2673 aaronmk
        cacheable=cacheable, log_level=log_level)
665 2063 aaronmk
666 2738 aaronmk
default = sql_gen.default # tells insert() to use the default value for a column
667 2066 aaronmk
668 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
669 2085 aaronmk
    '''For params, see insert_select()'''
670 1960 aaronmk
    if lists.is_seq(row): cols = None
671
    else:
672
        cols = row.keys()
673
        row = row.values()
674 2738 aaronmk
    row = list(row) # ensure that "== []" works
675 1960 aaronmk
676 2738 aaronmk
    if row == []: query = None
677
    else: query = sql_gen.Values(row).to_str(db)
678 1961 aaronmk
679 2738 aaronmk
    return insert_select(db, table, cols, query, [], *args, **kw_args)
680 11 aaronmk
681 2402 aaronmk
def mk_update(db, table, changes=None, cond=None):
682
    '''
683
    @param changes [(col, new_value),...]
684
        * container can be any iterable type
685
        * col: sql_gen.Code|str (for col name)
686
        * new_value: sql_gen.Code|literal value
687
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
688
    @return str query
689
    '''
690
    query = 'UPDATE '+sql_gen.as_Table(table).to_str(db)+'\nSET\n'
691 2405 aaronmk
    query += ',\n'.join((sql_gen.to_name_only_col(col, table).to_str(db)+' = '
692 2402 aaronmk
        +sql_gen.as_Value(new_value).to_str(db) for col, new_value in changes))
693 2467 aaronmk
    if cond != None: query += '\nWHERE\n'+cond.to_str(db)
694 2402 aaronmk
695
    return query
696
697
def update(db, *args, **kw_args):
698
    '''For params, see mk_update() and run_query()'''
699
    recover = kw_args.pop('recover', None)
700
701
    return run_query(db, mk_update(db, *args, **kw_args), [], recover)
702
703 135 aaronmk
def last_insert_id(db):
704 1849 aaronmk
    module = util.root_module(db.db)
705 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
706
    elif module == 'MySQLdb': return db.insert_id()
707
    else: return None
708 13 aaronmk
709 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
710 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
711 2415 aaronmk
    to names that will be distinct among the columns' tables.
712 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
713 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
714
    @param into The table for the new columns.
715 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
716
        columns will be included in the mapping even if they are not in cols.
717
        The tables of the provided Col objects will be changed to into, so make
718
        copies of them if you want to keep the original tables.
719
    @param as_items Whether to return a list of dict items instead of a dict
720 2383 aaronmk
    @return dict(orig_col=new_col, ...)
721
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
722 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
723
        * All mappings use the into table so its name can easily be
724 2383 aaronmk
          changed for all columns at once
725
    '''
726 2415 aaronmk
    cols = lists.uniqify(cols)
727
728 2394 aaronmk
    items = []
729 2389 aaronmk
    for col in preserve:
730 2390 aaronmk
        orig_col = copy.copy(col)
731 2392 aaronmk
        col.table = into
732 2394 aaronmk
        items.append((orig_col, col))
733
    preserve = set(preserve)
734
    for col in cols:
735 2716 aaronmk
        if col not in preserve:
736
            items.append((col, sql_gen.Col(str(col), into, col.srcs)))
737 2394 aaronmk
738
    if not as_items: items = dict(items)
739
    return items
740 2383 aaronmk
741 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
742 2391 aaronmk
    '''For params, see mk_flatten_mapping()
743
    @return See return value of mk_flatten_mapping()
744
    '''
745 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
746
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
747 2391 aaronmk
    run_query_into(db, *mk_select(db, joins, cols, limit=limit, start=start),
748 2392 aaronmk
        into=into)
749 2394 aaronmk
    return dict(items)
750 2391 aaronmk
751 2751 aaronmk
def mk_track_data_error(db, errors_table, cols, value, error_code, error):
752
    cols = map(sql_gen.to_name_only_col, cols)
753
754
    columns_cols = ['column']
755
    columns = sql_gen.NamedValues('columns', columns_cols,
756
        [[c.name] for c in cols])
757
    values_cols = ['value', 'error_code', 'error']
758
    values = sql_gen.NamedValues('values', values_cols,
759
        [value, error_code, error])
760
761
    select_cols = columns_cols+values_cols
762
    name_only_cols = map(sql_gen.to_name_only_col, select_cols)
763 2755 aaronmk
    errors_table = sql_gen.NamedTable('errors', errors_table)
764 2751 aaronmk
    joins = [columns, sql_gen.Join(values, type_='CROSS'),
765
        sql_gen.Join(errors_table, dict(zip(name_only_cols, select_cols)),
766
        sql_gen.filter_out)]
767
768
    return mk_insert_select(db, errors_table, name_only_cols,
769
        *mk_select(db, joins, select_cols, order_by=None))[0]
770
771 2758 aaronmk
def track_data_error(db, errors_table, *args, **kw_args):
772
    '''
773
    @param errors_table If None, does nothing.
774
    '''
775
    if errors_table == None: return
776
    run_query(db, mk_track_data_error(db, errors_table, *args, **kw_args),
777
        cacheable=True, log_level=4)
778
779 2704 aaronmk
def cast(db, type_, col, errors_table=None):
780
    '''Casts an (unrenamed) column or value.
781 2769 aaronmk
    If errors_table set and col has srcs, saves errors in errors_table (using
782
    col's srcs attr as the source columns) and converts errors to warnings.
783 2704 aaronmk
    @param col sql_gen.Col|sql_gen.Literal
784
    @param errors_table None|sql_gen.Table|str
785
    '''
786 2769 aaronmk
    save_errors = (errors_table != None and isinstance(col, sql_gen.Col)
787
        and col.srcs != ())
788
    if not save_errors: # can't save errors
789
        return sql_gen.CustomCode(col.to_str(db)+'::'+type_) # just cast
790 2704 aaronmk
791
    assert not isinstance(col, sql_gen.NamedCol)
792
793 2770 aaronmk
    errors_table = sql_gen.as_Table(errors_table)
794
    srcs = map(sql_gen.to_name_only_col, col.srcs)
795
    function_name = str(sql_gen.FunctionCall(type_, *srcs))
796 2704 aaronmk
    function = sql_gen.TempFunction(function_name, db.autocommit)
797
798 2718 aaronmk
    while True:
799
        # Create function definition
800
        query = '''\
801 2704 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''(value text)
802
RETURNS '''+type_+'''
803
LANGUAGE plpgsql
804
STRICT
805
AS $$
806
BEGIN
807
    /* The explicit cast to the return type is needed to make the cast happen
808
    inside the try block. (Implicit casts to the return type happen at the end
809
    of the function, outside any block.) */
810
    RETURN value::'''+type_+''';
811
EXCEPTION
812
    WHEN data_exception THEN
813
        -- Save error in errors table.
814 2751 aaronmk
        -- Insert the value and error for *each* source column.
815
'''+mk_track_data_error(db, errors_table, srcs,
816
    *map(sql_gen.CustomCode, ['value', 'SQLSTATE', 'SQLERRM']))+''';
817 2704 aaronmk
818
        RAISE WARNING '%', SQLERRM;
819
        RETURN NULL;
820
END;
821
$$;
822
'''
823 2718 aaronmk
824
        # Create function
825
        try:
826
            run_query(db, query, recover=True, cacheable=True,
827
                log_ignore_excs=(DuplicateFunctionException,))
828
            break # successful
829
        except DuplicateFunctionException:
830 2770 aaronmk
            function.name = next_version(function.name)
831
            # try again with next version of name
832 2704 aaronmk
833
    return sql_gen.FunctionCall(function, col)
834
835 2414 aaronmk
##### Database structure queries
836
837 2426 aaronmk
def table_row_count(db, table, recover=None):
838
    return value(run_query(db, *mk_select(db, table, [sql_gen.row_count],
839 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
840 2426 aaronmk
841 2414 aaronmk
def table_cols(db, table, recover=None):
842
    return list(col_names(select(db, table, limit=0, order_by=None,
843 2443 aaronmk
        recover=recover, log_level=4)))
844 2414 aaronmk
845 2291 aaronmk
def pkey(db, table, recover=None):
846 832 aaronmk
    '''Assumed to be first column in table'''
847 2339 aaronmk
    return table_cols(db, table, recover)[0]
848 832 aaronmk
849 2559 aaronmk
not_null_col = 'not_null_col'
850 2340 aaronmk
851
def table_not_null_col(db, table, recover=None):
852
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
853
    if not_null_col in table_cols(db, table, recover): return not_null_col
854
    else: return pkey(db, table, recover)
855
856 853 aaronmk
def index_cols(db, table, index):
857
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
858
    automatically created. When you don't know whether something is a UNIQUE
859
    constraint or a UNIQUE index, use this function.'''
860 1909 aaronmk
    module = util.root_module(db.db)
861
    if module == 'psycopg2':
862
        return list(values(run_query(db, '''\
863 853 aaronmk
SELECT attname
864 866 aaronmk
FROM
865
(
866
        SELECT attnum, attname
867
        FROM pg_index
868
        JOIN pg_class index ON index.oid = indexrelid
869
        JOIN pg_class table_ ON table_.oid = indrelid
870
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
871
        WHERE
872
            table_.relname = %(table)s
873
            AND index.relname = %(index)s
874
    UNION
875
        SELECT attnum, attname
876
        FROM
877
        (
878
            SELECT
879
                indrelid
880
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
881
                    AS indkey
882
            FROM pg_index
883
            JOIN pg_class index ON index.oid = indexrelid
884
            JOIN pg_class table_ ON table_.oid = indrelid
885
            WHERE
886
                table_.relname = %(table)s
887
                AND index.relname = %(index)s
888
        ) s
889
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
890
) s
891 853 aaronmk
ORDER BY attnum
892
''',
893 2443 aaronmk
            {'table': table, 'index': index}, cacheable=True, log_level=4)))
894 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
895
        ' database')
896 853 aaronmk
897 464 aaronmk
def constraint_cols(db, table, constraint):
898 1849 aaronmk
    module = util.root_module(db.db)
899 464 aaronmk
    if module == 'psycopg2':
900
        return list(values(run_query(db, '''\
901
SELECT attname
902
FROM pg_constraint
903
JOIN pg_class ON pg_class.oid = conrelid
904
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
905
WHERE
906
    relname = %(table)s
907
    AND conname = %(constraint)s
908
ORDER BY attnum
909
''',
910
            {'table': table, 'constraint': constraint})))
911
    else: raise NotImplementedError("Can't list constraint columns for "+module+
912
        ' database')
913
914 2096 aaronmk
row_num_col = '_row_num'
915
916 2688 aaronmk
def add_index(db, exprs, table=None, unique=False):
917
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
918 2538 aaronmk
    Currently, only function calls are supported as expressions.
919
    '''
920 2688 aaronmk
    if not lists.is_seq(exprs): exprs = [exprs]
921 2538 aaronmk
922 2688 aaronmk
    # Parse exprs
923
    old_exprs = exprs[:]
924
    exprs = []
925
    cols = []
926
    for i, expr in enumerate(old_exprs):
927
        expr = copy.deepcopy(expr) # don't modify input!
928
        expr = sql_gen.as_Col(expr)
929
930
        # Extract col
931
        if isinstance(expr, sql_gen.FunctionCall):
932
            col = expr.args[0]
933
            expr = sql_gen.Expr(expr)
934
        else: col = expr
935
936
        # Extract table
937
        if table == None:
938
            assert sql_gen.is_table_col(col)
939
            table = col.table
940
941
        col.table = None
942
943
        exprs.append(expr)
944
        cols.append(col)
945 2408 aaronmk
946 2688 aaronmk
    table = sql_gen.as_Table(table)
947
    index = sql_gen.Table(str(sql_gen.Col(','.join(map(str, cols)), table)))
948
949
    str_ = 'CREATE'
950
    if unique: str_ += ' UNIQUE'
951
    str_ += ' INDEX '+index.to_str(db)+' ON '+table.to_str(db)+' ('+(
952
        ', '.join((v.to_str(db) for v in exprs)))+')'
953
954
    try: run_query(db, str_, recover=True, cacheable=True, log_level=3)
955 2408 aaronmk
    except DuplicateTableException: pass # index already existed
956
957 2686 aaronmk
def add_pkey(db, table, cols=None, recover=None):
958
    '''Adds a primary key.
959
    @param cols [sql_gen.Col,...] The columns in the primary key.
960
        Defaults to the first column in the table.
961 2406 aaronmk
    @pre The table must not already have a primary key.
962
    '''
963
    table = sql_gen.as_Table(table)
964 2686 aaronmk
    if cols == None: cols = [pkey(db, table, recover)]
965
    col_strs = [sql_gen.to_name_only_col(v).to_str(db) for v in cols]
966 2406 aaronmk
967 2590 aaronmk
    index = sql_gen.as_Table(sql_gen.add_suffix(table.name, '_pkey'))
968 2614 aaronmk
    try:
969
        run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD CONSTRAINT '
970 2686 aaronmk
            +index.to_str(db)+' PRIMARY KEY ('+(', '.join(col_strs))+')',
971 2614 aaronmk
            recover=True, cacheable=True, log_level=3,
972
            log_ignore_excs=(DuplicateTableException,))
973
    except DuplicateTableException, e:
974
        index.name = next_version(index.name)
975
        # try again with next version of name
976 2406 aaronmk
977 2772 aaronmk
already_indexed = object() # tells add_indexes() the pkey has already been added
978
979
def add_indexes(db, table, has_pkey=True):
980
    '''Adds an index on all columns in a table.
981
    @param has_pkey bool|already_indexed Whether a pkey instead of a regular
982
        index should be added on the first column.
983
        * If already_indexed, the pkey is assumed to have already been added
984
    '''
985
    cols = table_cols(db, table)
986
    if has_pkey:
987
        if has_pkey is not already_indexed: add_pkey(db, table)
988
        cols = cols[1:]
989
    for col in cols: add_index(db, col, table)
990
991 2086 aaronmk
def add_row_num(db, table):
992 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
993
    be the primary key.'''
994 2320 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
995 2096 aaronmk
    run_query(db, 'ALTER TABLE '+table+' ADD COLUMN '+row_num_col
996 2443 aaronmk
        +' serial NOT NULL PRIMARY KEY', log_level=3)
997 2086 aaronmk
998 2760 aaronmk
def create_table(db, table, cols, has_pkey=True, col_indexes=True):
999 2675 aaronmk
    '''Creates a table.
1000 2681 aaronmk
    @param cols [sql_gen.TypedCol,...] The column names and types
1001
    @param has_pkey If set, the first column becomes the primary key.
1002 2760 aaronmk
    @param col_indexes bool|[ref]
1003
        * If True, indexes will be added on all non-pkey columns.
1004
        * If a list reference, [0] will be set to a function to do this.
1005
          This can be used to delay index creation until the table is populated.
1006 2675 aaronmk
    '''
1007
    table = sql_gen.as_Table(table)
1008
1009 2681 aaronmk
    if has_pkey:
1010
        cols[0] = pkey = copy.copy(cols[0]) # don't modify input!
1011
        pkey.type += ' NOT NULL PRIMARY KEY'
1012
1013 2675 aaronmk
    str_ = 'CREATE TABLE '+table.to_str(db)+' (\n'
1014 2681 aaronmk
    str_ += '\n, '.join(v.to_str(db) for v in cols)
1015
    str_ += '\n);\n'
1016 2675 aaronmk
    run_query(db, str_, cacheable=True, log_level=2)
1017 2760 aaronmk
1018
    # Add indexes
1019 2773 aaronmk
    if has_pkey: has_pkey = already_indexed
1020
    def add_indexes_(): add_indexes(db, table, has_pkey)
1021
    if isinstance(col_indexes, list): col_indexes[0] = add_indexes_ # defer
1022
    elif col_indexes: add_indexes_() # add now
1023 2675 aaronmk
1024 2684 aaronmk
def vacuum(db, table):
1025
    table = sql_gen.as_Table(table)
1026
    db.with_autocommit(lambda: run_query(db, 'VACUUM ANALYZE '+table.to_str(db),
1027
        log_level=3))
1028
1029 2732 aaronmk
def truncate(db, table, schema='public'):
1030 2777 aaronmk
    table = sql_gen.as_Table(table, schema)
1031 2733 aaronmk
    return run_query(db, 'TRUNCATE '+table.to_str(db)+' CASCADE')
1032 2732 aaronmk
1033 2729 aaronmk
def tables(db, schema_like='public', table_like='%', exact=False):
1034
    if exact: compare = '='
1035
    else: compare = 'LIKE'
1036
1037 1849 aaronmk
    module = util.root_module(db.db)
1038 2548 aaronmk
    params = {'schema_like': schema_like, 'table_like': table_like}
1039 832 aaronmk
    if module == 'psycopg2':
1040 1968 aaronmk
        return values(run_query(db, '''\
1041
SELECT tablename
1042
FROM pg_tables
1043
WHERE
1044 2729 aaronmk
    schemaname '''+compare+''' %(schema_like)s
1045
    AND tablename '''+compare+''' %(table_like)s
1046 1968 aaronmk
ORDER BY tablename
1047
''',
1048 2731 aaronmk
            params, cacheable=True, log_level=4))
1049 1968 aaronmk
    elif module == 'MySQLdb':
1050
        return values(run_query(db, 'SHOW TABLES LIKE %(table_like)s', params,
1051 2731 aaronmk
            cacheable=True, log_level=4))
1052 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
1053 830 aaronmk
1054 2726 aaronmk
def table_exists(db, table):
1055
    table = sql_gen.as_Table(table)
1056 2730 aaronmk
    return list(tables(db, table.schema, table.name, exact=True)) != []
1057 2726 aaronmk
1058
def errors_table(db, table, if_exists=True):
1059
    '''
1060
    @param if_exists If set, returns None if the errors table doesn't exist
1061
    @return None|sql_gen.Table
1062
    '''
1063
    table = sql_gen.as_Table(table)
1064
    if table.srcs != (): table = table.srcs[0]
1065
1066
    errors_table = sql_gen.suffixed_table(table, '.errors')
1067
    if if_exists and not table_exists(db, errors_table): return None
1068
    return errors_table
1069
1070 833 aaronmk
##### Database management
1071
1072 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
1073
    '''For kw_args, see tables()'''
1074
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
1075 833 aaronmk
1076 832 aaronmk
##### Heuristic queries
1077
1078 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
1079 1554 aaronmk
    '''Recovers from errors.
1080 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
1081
    '''
1082 2642 aaronmk
    row = sql_gen.ColDict(db, table, row)
1083 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
1084
1085 471 aaronmk
    try:
1086 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
1087 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
1088
            row_ct_ref[0] += cur.rowcount
1089
        return value(cur)
1090 471 aaronmk
    except DuplicateKeyException, e:
1091 2653 aaronmk
        row = sql_gen.ColDict(db, table,
1092
            util.dict_subset_right_join(row, e.cols))
1093
        return value(select(db, table, [pkey_], row, recover=True))
1094 471 aaronmk
1095 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
1096 830 aaronmk
    '''Recovers from errors'''
1097 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
1098 14 aaronmk
    except StopIteration:
1099 40 aaronmk
        if not create: raise
1100 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
1101 2078 aaronmk
1102 2593 aaronmk
def is_func_result(col):
1103
    return col.table.name.find('(') >= 0 and col.name == 'result'
1104
1105 2592 aaronmk
def into_table_name(out_table, in_tables0, mapping, is_func):
1106 2615 aaronmk
    def in_col_str(in_col):
1107
        in_col = sql_gen.remove_col_rename(in_col)
1108
        if isinstance(in_col, sql_gen.Col):
1109
            table = in_col.table
1110
            if table == in_tables0:
1111
                in_col = sql_gen.to_name_only_col(in_col)
1112
            elif is_func_result(in_col): in_col = table # omit col name
1113
        return str(in_col)
1114
1115 2580 aaronmk
    str_ = str(out_table)
1116
    if is_func:
1117 2615 aaronmk
        str_ += '('
1118 2580 aaronmk
1119 2615 aaronmk
        try: value_in_col = mapping['value']
1120
        except KeyError:
1121 2654 aaronmk
            str_ += ', '.join((str(k)+'='+in_col_str(v)
1122 2615 aaronmk
                for k, v in mapping.iteritems()))
1123
        else: str_ += in_col_str(value_in_col)
1124
1125
        str_ += ')'
1126 2580 aaronmk
    else: str_ += '_pkeys'
1127
    return str_
1128
1129 2508 aaronmk
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None, into=None,
1130 2552 aaronmk
    default=None, is_func=False):
1131 2078 aaronmk
    '''Recovers from errors.
1132
    Only works under PostgreSQL (uses INSERT RETURNING).
1133 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
1134
        tables to join with it using the main input table's pkey
1135 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
1136 2612 aaronmk
        * out_table_col: str (*not* sql_gen.Col)
1137
        * in_table_col: sql_gen.Col|literal-value
1138 2489 aaronmk
    @param into The table to contain the output and input pkeys.
1139 2574 aaronmk
        Defaults to `out_table.name+'_pkeys'`.
1140 2509 aaronmk
    @param default The *output* column to use as the pkey for missing rows.
1141
        If this output column does not exist in the mapping, uses None.
1142 2552 aaronmk
    @param is_func Whether out_table is the name of a SQL function, not a table
1143 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
1144 2078 aaronmk
    '''
1145 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
1146 2312 aaronmk
1147 2450 aaronmk
    def log_debug(msg): db.log_debug(msg, level=1.5)
1148 2505 aaronmk
    def col_ustr(str_):
1149 2567 aaronmk
        return strings.repr_no_u(sql_gen.remove_col_rename(str_))
1150 2450 aaronmk
1151 2768 aaronmk
    out_pkey = pkey(db, out_table, recover=True)
1152
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
1153
1154
    if mapping == {}: # need at least one column for INSERT SELECT
1155
        mapping = {out_pkey: None} # ColDict will replace with default value
1156
1157 2486 aaronmk
    log_debug('********** New iteration **********')
1158 2505 aaronmk
    log_debug('Inserting these input columns into '+strings.as_tt(
1159
        out_table.to_str(db))+':\n'+strings.as_table(mapping, ustr=col_ustr))
1160 2463 aaronmk
1161 2382 aaronmk
    # Create input joins from list of input tables
1162
    in_tables_ = in_tables[:] # don't modify input!
1163
    in_tables0 = in_tables_.pop(0) # first table is separate
1164 2735 aaronmk
    errors_table_ = errors_table(db, in_tables0)
1165 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
1166 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
1167 2460 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v,
1168
        {in_pkey: sql_gen.join_same_not_null}) for v in in_tables_]
1169 2131 aaronmk
1170 2592 aaronmk
    if into == None:
1171
        into = into_table_name(out_table, in_tables0, mapping, is_func)
1172
    into = sql_gen.as_Table(into)
1173
1174 2702 aaronmk
    # Set column sources
1175
    in_cols = filter(sql_gen.is_table_col, mapping.values())
1176
    for col in in_cols:
1177
        if col.table == in_tables0: col.set_srcs(sql_gen.src_self)
1178
1179 2486 aaronmk
    log_debug('Joining together input tables into temp table')
1180 2395 aaronmk
    # Place in new table for speed and so don't modify input if values edited
1181 2584 aaronmk
    in_table = sql_gen.Table('in')
1182 2702 aaronmk
    mapping = dicts.join(mapping, flatten(db, in_table, input_joins, in_cols,
1183
        preserve=[in_pkey_col], start=0))
1184 2395 aaronmk
    input_joins = [in_table]
1185 2486 aaronmk
    db.log_debug('Temp table: '+strings.as_tt(in_table.to_str(db)), level=2)
1186 2395 aaronmk
1187 2656 aaronmk
    mapping = sql_gen.ColDict(db, out_table, mapping)
1188
        # after applying dicts.join() because that returns a plain dict
1189
1190 2509 aaronmk
    # Resolve default value column
1191
    try: default = mapping[default]
1192
    except KeyError:
1193
        if default != None:
1194
            db.log_debug('Default value column '
1195
                +strings.as_tt(strings.repr_no_u(default))
1196 2511 aaronmk
                +' does not exist in mapping, falling back to None', level=2.1)
1197 2509 aaronmk
            default = None
1198
1199 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
1200 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
1201
1202 2201 aaronmk
    pkeys_table_exists_ref = [False]
1203 2420 aaronmk
    def insert_into_pkeys(joins, cols):
1204
        query, params = mk_select(db, joins, cols, order_by=None, start=0)
1205 2201 aaronmk
        if pkeys_table_exists_ref[0]:
1206 2489 aaronmk
            insert_select(db, into, pkeys_names, query, params)
1207 2201 aaronmk
        else:
1208 2489 aaronmk
            run_query_into(db, query, params, into=into)
1209 2201 aaronmk
            pkeys_table_exists_ref[0] = True
1210
1211 2429 aaronmk
    limit_ref = [None]
1212 2380 aaronmk
    conds = set()
1213 2233 aaronmk
    distinct_on = []
1214 2325 aaronmk
    def mk_main_select(joins, cols):
1215 2429 aaronmk
        return mk_select(db, joins, cols, conds, distinct_on,
1216
            limit=limit_ref[0], start=0)
1217 2132 aaronmk
1218 2519 aaronmk
    exc_strs = set()
1219 2309 aaronmk
    def log_exc(e):
1220 2519 aaronmk
        e_str = exc.str_(e, first_line_only=True)
1221
        log_debug('Caught exception: '+e_str)
1222
        assert e_str not in exc_strs # avoid infinite loops
1223
        exc_strs.add(e_str)
1224 2735 aaronmk
1225 2451 aaronmk
    def remove_all_rows():
1226 2450 aaronmk
        log_debug('Returning NULL for all rows')
1227 2429 aaronmk
        limit_ref[0] = 0 # just create an empty pkeys table
1228 2735 aaronmk
1229 2758 aaronmk
    def ignore(in_col, value, e):
1230
        track_data_error(db, errors_table_, in_col.srcs, value, e.cause.pgcode,
1231
            e.cause.pgerror)
1232
1233 2545 aaronmk
        in_col_str = strings.as_tt(repr(in_col))
1234 2544 aaronmk
        db.log_debug('Adding index on '+in_col_str+' to enable fast filtering',
1235
            level=2.5)
1236 2537 aaronmk
        add_index(db, in_col)
1237 2735 aaronmk
1238 2545 aaronmk
        log_debug('Ignoring rows with '+in_col_str+' = '
1239
            +strings.as_tt(repr(value)))
1240 2758 aaronmk
    def remove_rows(in_col, value, e):
1241
        ignore(in_col, value, e)
1242 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
1243
        assert cond not in conds # avoid infinite loops
1244 2380 aaronmk
        conds.add(cond)
1245 2758 aaronmk
    def invalid2null(in_col, value, e):
1246
        ignore(in_col, value, e)
1247 2403 aaronmk
        update(db, in_table, [(in_col, None)],
1248
            sql_gen.ColValueCond(in_col, value))
1249 2245 aaronmk
1250 2589 aaronmk
    def insert_pkeys_table(which):
1251
        return sql_gen.Table(sql_gen.add_suffix(in_table.name,
1252
            '_insert_'+which+'_pkeys'))
1253
    insert_out_pkeys = insert_pkeys_table('out')
1254
    insert_in_pkeys = insert_pkeys_table('in')
1255
1256 2206 aaronmk
    # Do inserts and selects
1257 2642 aaronmk
    join_cols = sql_gen.ColDict(db, out_table)
1258 2206 aaronmk
    while True:
1259 2521 aaronmk
        if limit_ref[0] == 0: # special case
1260
            log_debug('Creating an empty pkeys table')
1261
            cur = run_query_into(db, *mk_select(db, out_table, [out_pkey],
1262
                limit=limit_ref[0]), into=insert_out_pkeys)
1263
            break # don't do main case
1264
1265 2303 aaronmk
        has_joins = join_cols != {}
1266
1267 2305 aaronmk
        # Prepare to insert new rows
1268 2325 aaronmk
        insert_joins = input_joins[:] # don't modify original!
1269 2403 aaronmk
        insert_args = dict(recover=True, cacheable=False)
1270 2303 aaronmk
        if has_joins:
1271 2317 aaronmk
            distinct_on = [v.to_Col() for v in join_cols.values()]
1272 2325 aaronmk
            insert_joins.append(sql_gen.Join(out_table, join_cols,
1273
                sql_gen.filter_out))
1274
        else:
1275 2404 aaronmk
            insert_args.update(dict(returning=out_pkey, into=insert_out_pkeys))
1276 2520 aaronmk
        main_select = mk_main_select(insert_joins, mapping.values())[0]
1277 2303 aaronmk
1278 2486 aaronmk
        log_debug('Trying to insert new rows')
1279 2206 aaronmk
        try:
1280 2518 aaronmk
            cur = insert_select(db, out_table, mapping.keys(), main_select,
1281
                **insert_args)
1282 2357 aaronmk
            break # insert successful
1283 2206 aaronmk
        except DuplicateKeyException, e:
1284 2309 aaronmk
            log_exc(e)
1285
1286 2258 aaronmk
            old_join_cols = join_cols.copy()
1287 2565 aaronmk
            join_cols.update(util.dict_subset_right_join(mapping, e.cols))
1288 2486 aaronmk
            log_debug('Ignoring existing rows, comparing on these columns:\n'
1289 2505 aaronmk
                +strings.as_inline_table(join_cols, ustr=col_ustr))
1290 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
1291 2230 aaronmk
        except NullValueException, e:
1292 2309 aaronmk
            log_exc(e)
1293
1294 2230 aaronmk
            out_col, = e.cols
1295
            try: in_col = mapping[out_col]
1296 2356 aaronmk
            except KeyError:
1297 2486 aaronmk
                log_debug('Missing mapping for NOT NULL column '+out_col)
1298 2451 aaronmk
                remove_all_rows()
1299 2758 aaronmk
            else: remove_rows(in_col, None, e)
1300 2542 aaronmk
        except FunctionValueException, e:
1301
            log_exc(e)
1302
1303
            func_name = e.name
1304
            value = e.value
1305
            for out_col, in_col in mapping.iteritems():
1306 2758 aaronmk
                in_col = sql_gen.unwrap_func_call(in_col, func_name)
1307
                invalid2null(in_col, value, e)
1308 2525 aaronmk
        except MissingCastException, e:
1309
            log_exc(e)
1310
1311
            out_col = e.col
1312 2706 aaronmk
            type_ = e.type
1313
1314 2724 aaronmk
            log_debug('Casting '+strings.as_tt(out_col)+' input to '
1315
                +strings.as_tt(type_))
1316 2735 aaronmk
            def wrap_func(col): return cast(db, type_, col, errors_table_)
1317 2705 aaronmk
            mapping[out_col] = sql_gen.wrap(wrap_func, mapping[out_col])
1318 2429 aaronmk
        except DatabaseErrors, e:
1319
            log_exc(e)
1320
1321 2531 aaronmk
            msg = 'No handler for exception: '+exc.str_(e)
1322 2451 aaronmk
            warnings.warn(DbWarning(msg))
1323
            log_debug(msg)
1324
            remove_all_rows()
1325 2358 aaronmk
        # after exception handled, rerun loop with additional constraints
1326 2132 aaronmk
1327 2357 aaronmk
    if row_ct_ref != None and cur.rowcount >= 0:
1328
        row_ct_ref[0] += cur.rowcount
1329
1330
    if has_joins:
1331
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
1332 2486 aaronmk
        log_debug('Getting output table pkeys of existing/inserted rows')
1333 2420 aaronmk
        insert_into_pkeys(select_joins, pkeys_cols)
1334 2357 aaronmk
    else:
1335 2404 aaronmk
        add_row_num(db, insert_out_pkeys) # for joining with input pkeys
1336 2357 aaronmk
1337 2486 aaronmk
        log_debug('Getting input table pkeys of inserted rows')
1338 2357 aaronmk
        run_query_into(db, *mk_main_select(input_joins, [in_pkey]),
1339 2404 aaronmk
            into=insert_in_pkeys)
1340
        add_row_num(db, insert_in_pkeys) # for joining with output pkeys
1341 2357 aaronmk
1342 2428 aaronmk
        assert table_row_count(db, insert_out_pkeys) == table_row_count(db,
1343
            insert_in_pkeys)
1344
1345 2486 aaronmk
        log_debug('Combining output and input pkeys in inserted order')
1346 2404 aaronmk
        pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
1347 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
1348 2420 aaronmk
        insert_into_pkeys(pkey_joins, pkeys_names)
1349 2357 aaronmk
1350 2486 aaronmk
    db.log_debug('Adding pkey on pkeys table to enable fast joins', level=2.5)
1351 2594 aaronmk
    add_pkey(db, into)
1352 2407 aaronmk
1353 2508 aaronmk
    log_debug('Setting pkeys of missing rows to '+strings.as_tt(repr(default)))
1354 2489 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(into,
1355 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
1356
        # must use join_same_not_null or query will take forever
1357 2420 aaronmk
    insert_into_pkeys(missing_rows_joins,
1358 2508 aaronmk
        [in_pkey_col, sql_gen.NamedCol(out_pkey, default)])
1359 2357 aaronmk
1360 2489 aaronmk
    assert table_row_count(db, into) == table_row_count(db, in_table)
1361 2428 aaronmk
1362 2702 aaronmk
    srcs = []
1363
    if is_func: srcs = sql_gen.cols_srcs(in_cols)
1364
    return sql_gen.Col(out_pkey, into, srcs)
1365 2115 aaronmk
1366
##### Data cleanup
1367
1368 2290 aaronmk
def cleanup_table(db, table, cols):
1369 2677 aaronmk
    table = sql_gen.as_Table(table)
1370
    cols = map(sql_gen.as_Col, cols)
1371 2115 aaronmk
1372 2677 aaronmk
    expr = ('nullif(nullif(trim(both from %s), '+db.esc_value('')+'), '
1373
        +db.esc_value(r'\N')+')')
1374
    changes = [(v, sql_gen.CustomCode(expr % v.to_str(db)))
1375
        for v in cols]
1376 2115 aaronmk
1377 2677 aaronmk
    update(db, table, changes)