Project

General

Profile

1 11 aaronmk
# Database access
2
3 1869 aaronmk
import copy
4 2127 aaronmk
import operator
5 11 aaronmk
import re
6 865 aaronmk
import warnings
7 11 aaronmk
8 300 aaronmk
import exc
9 1909 aaronmk
import dicts
10 1893 aaronmk
import iters
11 1960 aaronmk
import lists
12 1889 aaronmk
from Proxy import Proxy
13 1872 aaronmk
import rand
14 2217 aaronmk
import sql_gen
15 862 aaronmk
import strings
16 131 aaronmk
import util
17 11 aaronmk
18 832 aaronmk
##### Exceptions
19
20 2804 aaronmk
def get_cur_query(cur, input_query=None):
21 2168 aaronmk
    raw_query = None
22
    if hasattr(cur, 'query'): raw_query = cur.query
23
    elif hasattr(cur, '_last_executed'): raw_query = cur._last_executed
24 2170 aaronmk
25
    if raw_query != None: return raw_query
26 2804 aaronmk
    else: return '[input] '+strings.ustr(input_query)
27 14 aaronmk
28 2170 aaronmk
def _add_cursor_info(e, *args, **kw_args):
29
    '''For params, see get_cur_query()'''
30 2771 aaronmk
    exc.add_msg(e, 'query: '+strings.ustr(get_cur_query(*args, **kw_args)))
31 135 aaronmk
32 300 aaronmk
class DbException(exc.ExceptionWithCause):
33 14 aaronmk
    def __init__(self, msg, cause=None, cur=None):
34 2145 aaronmk
        exc.ExceptionWithCause.__init__(self, msg, cause, cause_newline=True)
35 14 aaronmk
        if cur != None: _add_cursor_info(self, cur)
36
37 2143 aaronmk
class ExceptionWithName(DbException):
38
    def __init__(self, name, cause=None):
39 2484 aaronmk
        DbException.__init__(self, 'for name: '+strings.as_tt(str(name)), cause)
40 2143 aaronmk
        self.name = name
41 360 aaronmk
42 2240 aaronmk
class ExceptionWithNameValue(DbException):
43
    def __init__(self, name, value, cause=None):
44 2484 aaronmk
        DbException.__init__(self, 'for name: '+strings.as_tt(str(name))
45
            +'; value: '+strings.as_tt(repr(value)), cause)
46 2240 aaronmk
        self.name = name
47
        self.value = value
48
49 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 2448 aaronmk
log_debug_none = lambda msg, level=2: None
149 1901 aaronmk
150 1849 aaronmk
class DbConn:
151 2190 aaronmk
    def __init__(self, db_config, serializable=True, autocommit=False,
152
        caching=True, log_debug=log_debug_none):
153 1869 aaronmk
        self.db_config = db_config
154
        self.serializable = serializable
155 2190 aaronmk
        self.autocommit = autocommit
156
        self.caching = caching
157 1901 aaronmk
        self.log_debug = log_debug
158 2193 aaronmk
        self.debug = log_debug != log_debug_none
159 1869 aaronmk
160
        self.__db = None
161 1889 aaronmk
        self.query_results = {}
162 2139 aaronmk
        self._savepoint = 0
163 2671 aaronmk
        self._notices_seen = set()
164 1869 aaronmk
165
    def __getattr__(self, name):
166
        if name == '__dict__': raise Exception('getting __dict__')
167
        if name == 'db': return self._db()
168
        else: raise AttributeError()
169
170
    def __getstate__(self):
171
        state = copy.copy(self.__dict__) # shallow copy
172 1915 aaronmk
        state['log_debug'] = None # don't pickle the debug callback
173 1869 aaronmk
        state['_DbConn__db'] = None # don't pickle the connection
174
        return state
175
176 2165 aaronmk
    def connected(self): return self.__db != None
177
178 1869 aaronmk
    def _db(self):
179
        if self.__db == None:
180
            # Process db_config
181
            db_config = self.db_config.copy() # don't modify input!
182 2097 aaronmk
            schemas = db_config.pop('schemas', None)
183 1869 aaronmk
            module_name, mappings = db_engines[db_config.pop('engine')]
184
            module = __import__(module_name)
185
            _add_module(module)
186
            for orig, new in mappings.iteritems():
187
                try: util.rename_key(db_config, orig, new)
188
                except KeyError: pass
189
190
            # Connect
191
            self.__db = module.connect(**db_config)
192
193
            # Configure connection
194 2894 aaronmk
            if self.serializable: self.run_query(
195 2678 aaronmk
                'SET TRANSACTION ISOLATION LEVEL SERIALIZABLE', log_level=4)
196 2101 aaronmk
            if schemas != None:
197 2893 aaronmk
                search_path = [self.esc_name(s) for s in schemas.split(',')]
198
                search_path.append(value(run_query(self, 'SHOW search_path',
199
                    log_level=4)))
200
                run_query(self, 'SET search_path TO '+(','.join(search_path)),
201
                    log_level=3)
202 1869 aaronmk
203
        return self.__db
204 1889 aaronmk
205 1891 aaronmk
    class DbCursor(Proxy):
206 1927 aaronmk
        def __init__(self, outer):
207 1891 aaronmk
            Proxy.__init__(self, outer.db.cursor())
208 2191 aaronmk
            self.outer = outer
209 1927 aaronmk
            self.query_results = outer.query_results
210 1894 aaronmk
            self.query_lookup = None
211 1891 aaronmk
            self.result = []
212 1889 aaronmk
213 2802 aaronmk
        def execute(self, query):
214 2764 aaronmk
            self._is_insert = query.startswith('INSERT')
215 2797 aaronmk
            self.query_lookup = query
216 2148 aaronmk
            try:
217 2191 aaronmk
                try:
218 2802 aaronmk
                    cur = self.inner.execute(query)
219 2191 aaronmk
                    self.outer.do_autocommit()
220 2802 aaronmk
                finally: self.query = get_cur_query(self.inner, query)
221 1904 aaronmk
            except Exception, e:
222 2802 aaronmk
                _add_cursor_info(e, self, query)
223 1904 aaronmk
                self.result = e # cache the exception as the result
224
                self._cache_result()
225
                raise
226 2800 aaronmk
            if self.rowcount == 0 and query.startswith('SELECT'): # empty SELECT
227
                consume_rows(self) # fetch all rows so result will be cached
228 2762 aaronmk
            return cur
229 1894 aaronmk
230 1891 aaronmk
        def fetchone(self):
231
            row = self.inner.fetchone()
232 1899 aaronmk
            if row != None: self.result.append(row)
233
            # otherwise, fetched all rows
234 1904 aaronmk
            else: self._cache_result()
235
            return row
236
237
        def _cache_result(self):
238 1906 aaronmk
            # For inserts, only cache exceptions since inserts are not
239
            # idempotent, but an invalid insert will always be invalid
240 1930 aaronmk
            if self.query_results != None and (not self._is_insert
241 1906 aaronmk
                or isinstance(self.result, Exception)):
242
243 1894 aaronmk
                assert self.query_lookup != None
244 1916 aaronmk
                self.query_results[self.query_lookup] = self.CacheCursor(
245
                    util.dict_subset(dicts.AttrsDictView(self),
246
                    ['query', 'result', 'rowcount', 'description']))
247 1906 aaronmk
248 1916 aaronmk
        class CacheCursor:
249
            def __init__(self, cached_result): self.__dict__ = cached_result
250
251 1927 aaronmk
            def execute(self, *args, **kw_args):
252 1916 aaronmk
                if isinstance(self.result, Exception): raise self.result
253
                # otherwise, result is a rows list
254
                self.iter = iter(self.result)
255
256
            def fetchone(self):
257
                try: return self.iter.next()
258
                except StopIteration: return None
259 1891 aaronmk
260 2212 aaronmk
    def esc_value(self, value):
261 2663 aaronmk
        try: str_ = self.mogrify('%s', [value])
262
        except NotImplementedError, e:
263
            module = util.root_module(self.db)
264
            if module == 'MySQLdb':
265
                import _mysql
266
                str_ = _mysql.escape_string(value)
267
            else: raise e
268 2374 aaronmk
        return strings.to_unicode(str_)
269 2212 aaronmk
270 2347 aaronmk
    def esc_name(self, name): return esc_name(self, name) # calls global func
271
272 2814 aaronmk
    def std_code(self, str_):
273
        '''Standardizes SQL code.
274
        * Ensures that string literals are prefixed by `E`
275
        '''
276
        if str_.startswith("'"): str_ = 'E'+str_
277
        return str_
278
279 2665 aaronmk
    def can_mogrify(self):
280 2663 aaronmk
        module = util.root_module(self.db)
281 2665 aaronmk
        return module == 'psycopg2'
282 2663 aaronmk
283 2665 aaronmk
    def mogrify(self, query, params=None):
284
        if self.can_mogrify(): return self.db.cursor().mogrify(query, params)
285
        else: raise NotImplementedError("Can't mogrify query")
286
287 2671 aaronmk
    def print_notices(self):
288 2725 aaronmk
        if hasattr(self.db, 'notices'):
289
            for msg in self.db.notices:
290
                if msg not in self._notices_seen:
291
                    self._notices_seen.add(msg)
292
                    self.log_debug(msg, level=2)
293 2671 aaronmk
294 2793 aaronmk
    def run_query(self, query, cacheable=False, log_level=2,
295 2464 aaronmk
        debug_msg_ref=None):
296 2445 aaronmk
        '''
297 2464 aaronmk
        @param log_ignore_excs The log_level will be increased by 2 if the query
298
            throws one of these exceptions.
299 2664 aaronmk
        @param debug_msg_ref If specified, the log message will be returned in
300
            this instead of being output. This allows you to filter log messages
301
            depending on the result of the query.
302 2445 aaronmk
        '''
303 2167 aaronmk
        assert query != None
304
305 2047 aaronmk
        if not self.caching: cacheable = False
306 1903 aaronmk
        used_cache = False
307 2664 aaronmk
308
        def log_msg(query):
309
            if used_cache: cache_status = 'cache hit'
310
            elif cacheable: cache_status = 'cache miss'
311
            else: cache_status = 'non-cacheable'
312
            return 'DB query: '+cache_status+':\n'+strings.as_code(query, 'SQL')
313
314 1903 aaronmk
        try:
315 1927 aaronmk
            # Get cursor
316
            if cacheable:
317
                try:
318 2797 aaronmk
                    cur = self.query_results[query]
319 1927 aaronmk
                    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 2793 aaronmk
            cur.execute(query)
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 2793 aaronmk
                debug_msg_ref[0] = log_msg(str(get_cur_query(cur, query)))
333 1903 aaronmk
334
        return cur
335 1914 aaronmk
336 2797 aaronmk
    def is_cached(self, query): return query in self.query_results
337 2139 aaronmk
338 2683 aaronmk
    def with_autocommit(self, func, autocommit=True):
339 2801 aaronmk
        import psycopg2.extensions
340
        if autocommit:
341
            isolation_level = psycopg2.extensions.ISOLATION_LEVEL_AUTOCOMMIT
342
        else: isolation_level = psycopg2.extensions.ISOLATION_LEVEL_SERIALIZABLE
343
344
        prev_isolation_level = self.db.isolation_level
345
        self.db.set_isolation_level(isolation_level)
346 2683 aaronmk
        try: return func()
347 2801 aaronmk
        finally: self.db.set_isolation_level(prev_isolation_level)
348 2683 aaronmk
349 2139 aaronmk
    def with_savepoint(self, func):
350 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
351 2443 aaronmk
        self.run_query('SAVEPOINT '+savepoint, log_level=4)
352 2139 aaronmk
        self._savepoint += 1
353 2683 aaronmk
        try:
354 2139 aaronmk
            try: return_val = func()
355
            finally:
356
                self._savepoint -= 1
357
                assert self._savepoint >= 0
358
        except:
359 2443 aaronmk
            self.run_query('ROLLBACK TO SAVEPOINT '+savepoint, log_level=4)
360 2139 aaronmk
            raise
361
        else:
362 2443 aaronmk
            self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
363 2191 aaronmk
            self.do_autocommit()
364 2139 aaronmk
            return return_val
365 2191 aaronmk
366
    def do_autocommit(self):
367
        '''Autocommits if outside savepoint'''
368
        assert self._savepoint >= 0
369
        if self.autocommit and self._savepoint == 0:
370 2668 aaronmk
            self.log_debug('Autocommitting')
371 2191 aaronmk
            self.db.commit()
372 2643 aaronmk
373 2819 aaronmk
    def col_info(self, col):
374 2643 aaronmk
        table = sql_gen.Table('columns', 'information_schema')
375 2821 aaronmk
        cols = ['data_type', 'column_default',
376
            cast(self, 'boolean', 'is_nullable')]
377 2643 aaronmk
378
        conds = [('table_name', col.table.name), ('column_name', col.name)]
379
        schema = col.table.schema
380
        if schema != None: conds.append(('table_schema', schema))
381
382 2819 aaronmk
        type_, default, nullable = row(select(self, table, cols, conds,
383 2874 aaronmk
            order_by='table_schema', limit=1, log_level=4))
384 2643 aaronmk
            # TODO: order_by search_path schema order
385 2819 aaronmk
        default = sql_gen.as_Code(default, self)
386
387
        return sql_gen.TypedCol(col.name, type_, default, nullable)
388 1849 aaronmk
389 1869 aaronmk
connect = DbConn
390
391 832 aaronmk
##### Recoverable querying
392 15 aaronmk
393 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
394 11 aaronmk
395 2791 aaronmk
def run_query(db, query, recover=None, cacheable=False, log_level=2,
396
    log_ignore_excs=None, **kw_args):
397 2794 aaronmk
    '''For params, see DbConn.run_query()'''
398 830 aaronmk
    if recover == None: recover = False
399 2464 aaronmk
    if log_ignore_excs == None: log_ignore_excs = ()
400
    log_ignore_excs = tuple(log_ignore_excs)
401 830 aaronmk
402 2666 aaronmk
    debug_msg_ref = None # usually, db.run_query() logs query before running it
403
    # But if filtering with log_ignore_excs, wait until after exception parsing
404
    if log_ignore_excs != () or not db.can_mogrify(): debug_msg_ref = [None]
405
406 2148 aaronmk
    try:
407 2464 aaronmk
        try:
408 2794 aaronmk
            def run(): return db.run_query(query, cacheable, log_level,
409 2793 aaronmk
                debug_msg_ref, **kw_args)
410 2796 aaronmk
            if recover and not db.is_cached(query):
411 2464 aaronmk
                return with_savepoint(db, run)
412
            else: return run() # don't need savepoint if cached
413
        except Exception, e:
414
            if not recover: raise # need savepoint to run index_cols()
415
            msg = exc.str_(e)
416
417
            match = re.search(r'duplicate key value violates unique constraint '
418 2493 aaronmk
                r'"((_?[^\W_]+)_.+?)"', msg)
419 2464 aaronmk
            if match:
420
                constraint, table = match.groups()
421
                try: cols = index_cols(db, table, constraint)
422
                except NotImplementedError: raise e
423
                else: raise DuplicateKeyException(constraint, cols, e)
424
425 2493 aaronmk
            match = re.search(r'null value in column "(.+?)" violates not-null'
426 2464 aaronmk
                r' constraint', msg)
427
            if match: raise NullValueException('NOT NULL', [match.group(1)], e)
428
429
            match = re.search(r'\b(?:invalid input (?:syntax|value)\b.*?'
430
                r'|date/time field value out of range): "(.+?)"\n'
431 2535 aaronmk
                r'(?:(?s).*?)\bfunction "(.+?)"', msg)
432 2464 aaronmk
            if match:
433
                value, name = match.groups()
434
                raise FunctionValueException(name, strings.to_unicode(value), e)
435
436 2526 aaronmk
            match = re.search(r'column "(.+?)" is of type (.+?) but expression '
437 2523 aaronmk
                r'is of type', msg)
438
            if match:
439
                col, type_ = match.groups()
440
                raise MissingCastException(type_, col, e)
441
442 2493 aaronmk
            match = re.search(r'relation "(.+?)" already exists', msg)
443 2464 aaronmk
            if match: raise DuplicateTableException(match.group(1), e)
444
445 2493 aaronmk
            match = re.search(r'function "(.+?)" already exists', msg)
446 2464 aaronmk
            if match: raise DuplicateFunctionException(match.group(1), e)
447
448
            raise # no specific exception raised
449
    except log_ignore_excs:
450
        log_level += 2
451
        raise
452
    finally:
453 2666 aaronmk
        if debug_msg_ref != None and debug_msg_ref[0] != None:
454
            db.log_debug(debug_msg_ref[0], log_level)
455 830 aaronmk
456 832 aaronmk
##### Basic queries
457
458 2153 aaronmk
def next_version(name):
459 2163 aaronmk
    version = 1 # first existing name was version 0
460 2586 aaronmk
    match = re.match(r'^(.*)#(\d+)$', name)
461 2153 aaronmk
    if match:
462 2586 aaronmk
        name, version = match.groups()
463
        version = int(version)+1
464 2588 aaronmk
    return sql_gen.add_suffix(name, '#'+str(version))
465 2153 aaronmk
466 2899 aaronmk
def lock_table(db, table, mode):
467
    table = sql_gen.as_Table(table)
468
    run_query(db, 'LOCK TABLE '+table.to_str(db)+' IN '+mode+' MODE')
469
470 2789 aaronmk
def run_query_into(db, query, into=None, add_indexes_=False, **kw_args):
471 2085 aaronmk
    '''Outputs a query to a temp table.
472
    For params, see run_query().
473
    '''
474 2789 aaronmk
    if into == None: return run_query(db, query, **kw_args)
475 2790 aaronmk
476
    assert isinstance(into, sql_gen.Table)
477
478
    kw_args['recover'] = True
479
    kw_args.setdefault('log_ignore_excs', (DuplicateTableException,))
480
481
    temp = not db.autocommit # tables are permanent in autocommit mode
482
    # "temporary tables cannot specify a schema name", so remove schema
483
    if temp: into.schema = None
484
485
    # Create table
486
    while True:
487
        create_query = 'CREATE'
488
        if temp: create_query += ' TEMP'
489
        create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
490 2385 aaronmk
491 2790 aaronmk
        try:
492
            cur = run_query(db, create_query, **kw_args)
493
                # CREATE TABLE AS sets rowcount to # rows in query
494
            break
495
        except DuplicateTableException, e:
496
            into.name = next_version(into.name)
497
            # try again with next version of name
498
499
    if add_indexes_: add_indexes(db, into)
500
501
    return cur
502 2085 aaronmk
503 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
504
505 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
506
507 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
508 2293 aaronmk
    start=None, order_by=order_by_pkey, default_table=None):
509 1981 aaronmk
    '''
510 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
511 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
512 1981 aaronmk
    @param fields Use None to select all fields in the table
513 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
514 2379 aaronmk
        * container can be any iterable type
515 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
516
        * compare_right_side: sql_gen.ValueCond|literal value
517 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
518
        use all columns
519 2786 aaronmk
    @return query
520 1981 aaronmk
    '''
521 2315 aaronmk
    # Parse tables param
522 2121 aaronmk
    if not lists.is_seq(tables): tables = [tables]
523 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
524 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
525 2121 aaronmk
526 2315 aaronmk
    # Parse other params
527 2376 aaronmk
    if conds == None: conds = []
528 2650 aaronmk
    elif dicts.is_dict(conds): conds = conds.items()
529 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
530 135 aaronmk
    assert limit == None or type(limit) == int
531 865 aaronmk
    assert start == None or type(start) == int
532 2315 aaronmk
    if order_by is order_by_pkey:
533
        if distinct_on != []: order_by = None
534
        else: order_by = pkey(db, table0, recover=True)
535 865 aaronmk
536 2315 aaronmk
    query = 'SELECT'
537 2056 aaronmk
538 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
539 2056 aaronmk
540 2200 aaronmk
    # DISTINCT ON columns
541 2233 aaronmk
    if distinct_on != []:
542 2467 aaronmk
        query += '\nDISTINCT'
543 2254 aaronmk
        if distinct_on is not distinct_on_all:
544 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
545
546
    # Columns
547 2467 aaronmk
    query += '\n'
548 1135 aaronmk
    if fields == None: query += '*'
549 2765 aaronmk
    else:
550
        assert fields != []
551
        query += '\n, '.join(map(parse_col, fields))
552 2200 aaronmk
553
    # Main table
554 2467 aaronmk
    query += '\nFROM '+table0.to_str(db)
555 865 aaronmk
556 2122 aaronmk
    # Add joins
557 2271 aaronmk
    left_table = table0
558 2263 aaronmk
    for join_ in tables:
559
        table = join_.table
560 2238 aaronmk
561 2343 aaronmk
        # Parse special values
562
        if join_.type_ is sql_gen.filter_out: # filter no match
563 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
564 2853 aaronmk
                sql_gen.CompareCond(None, '~=')))
565 2903 aaronmk
            # Prevent concurrent duplicate keys when used with INSERT SELECT
566
            lock_table(db, sql_gen.underlying_table(table), 'EXCLUSIVE')
567 2343 aaronmk
568 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
569 2122 aaronmk
570
        left_table = table
571
572 865 aaronmk
    missing = True
573 2376 aaronmk
    if conds != []:
574 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
575
        else: whitespace = '\n'
576 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
577
            .to_str(db) for l, r in conds], 'WHERE')
578 865 aaronmk
        missing = False
579 2227 aaronmk
    if order_by != None:
580 2467 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
581
    if limit != None: query += '\nLIMIT '+str(limit); missing = False
582 865 aaronmk
    if start != None:
583 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
584 865 aaronmk
        missing = False
585
    if missing: warnings.warn(DbWarning(
586
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
587
588 2786 aaronmk
    return query
589 11 aaronmk
590 2054 aaronmk
def select(db, *args, **kw_args):
591
    '''For params, see mk_select() and run_query()'''
592
    recover = kw_args.pop('recover', None)
593
    cacheable = kw_args.pop('cacheable', True)
594 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
595 2054 aaronmk
596 2791 aaronmk
    return run_query(db, mk_select(db, *args, **kw_args), recover, cacheable,
597
        log_level=log_level)
598 2054 aaronmk
599 2788 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, returning=None,
600
    embeddable=False):
601 1960 aaronmk
    '''
602
    @param returning str|None An inserted column (such as pkey) to return
603 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
604 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
605
        query will be fully cached, not just if it raises an exception.
606 1960 aaronmk
    '''
607 2754 aaronmk
    table = sql_gen.remove_table_rename(sql_gen.as_Table(table))
608 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
609 2646 aaronmk
    if cols != None:
610
        cols = [sql_gen.to_name_only_col(v, table).to_str(db) for v in cols]
611 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
612 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
613 2063 aaronmk
614
    # Build query
615 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
616
    query = first_line
617 2467 aaronmk
    if cols != None: query += '\n('+', '.join(cols)+')'
618
    query += '\n'+select_query
619 2063 aaronmk
620
    if returning != None:
621 2743 aaronmk
        query += '\nRETURNING '+sql_gen.to_name_only_col(returning).to_str(db)
622 2063 aaronmk
623 2070 aaronmk
    if embeddable:
624 2327 aaronmk
        assert returning != None
625
626 2070 aaronmk
        # Create function
627 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
628 2327 aaronmk
        return_type = 'SETOF '+returning.to_str(db)+'%TYPE'
629 2189 aaronmk
        while True:
630
            try:
631 2698 aaronmk
                function = sql_gen.TempFunction(function_name, db.autocommit)
632 2194 aaronmk
633 2189 aaronmk
                function_query = '''\
634 2698 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''()
635 2467 aaronmk
RETURNS '''+return_type+'''
636
LANGUAGE sql
637
AS $$
638 2788 aaronmk
'''+query+''';
639 2467 aaronmk
$$;
640 2070 aaronmk
'''
641 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
642 2464 aaronmk
                    log_ignore_excs=(DuplicateFunctionException,))
643 2189 aaronmk
                break # this version was successful
644
            except DuplicateFunctionException, e:
645
                function_name = next_version(function_name)
646
                # try again with next version of name
647 2070 aaronmk
648 2337 aaronmk
        # Return query that uses function
649 2698 aaronmk
        func_table = sql_gen.NamedTable('f', sql_gen.FunctionCall(function),
650 2742 aaronmk
            [returning]) # AS clause requires function alias
651 2787 aaronmk
        return mk_select(db, func_table, start=0, order_by=None)
652 2070 aaronmk
653 2787 aaronmk
    return query
654 2066 aaronmk
655
def insert_select(db, *args, **kw_args):
656 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
657 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
658
        values in
659 2072 aaronmk
    '''
660 2386 aaronmk
    into = kw_args.pop('into', None)
661
    if into != None: kw_args['embeddable'] = True
662 2066 aaronmk
    recover = kw_args.pop('recover', None)
663
    cacheable = kw_args.pop('cacheable', True)
664 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
665 2066 aaronmk
666 2789 aaronmk
    return run_query_into(db, mk_insert_select(db, *args, **kw_args), into,
667 2787 aaronmk
        recover=recover, 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 2788 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 2791 aaronmk
    return run_query(db, mk_update(db, *args, **kw_args), recover)
705 2402 aaronmk
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 2846 aaronmk
        into=into, add_indexes_=True)
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 2810 aaronmk
    assert cols != ()
756
757 2751 aaronmk
    cols = map(sql_gen.to_name_only_col, cols)
758
759
    columns_cols = ['column']
760
    columns = sql_gen.NamedValues('columns', columns_cols,
761
        [[c.name] for c in cols])
762
    values_cols = ['value', 'error_code', 'error']
763
    values = sql_gen.NamedValues('values', values_cols,
764
        [value, error_code, error])
765
766
    select_cols = columns_cols+values_cols
767
    name_only_cols = map(sql_gen.to_name_only_col, select_cols)
768 2755 aaronmk
    errors_table = sql_gen.NamedTable('errors', errors_table)
769 2751 aaronmk
    joins = [columns, sql_gen.Join(values, type_='CROSS'),
770
        sql_gen.Join(errors_table, dict(zip(name_only_cols, select_cols)),
771
        sql_gen.filter_out)]
772
773
    return mk_insert_select(db, errors_table, name_only_cols,
774 2787 aaronmk
        mk_select(db, joins, select_cols, order_by=None))
775 2751 aaronmk
776 2810 aaronmk
def track_data_error(db, errors_table, cols, *args, **kw_args):
777 2758 aaronmk
    '''
778
    @param errors_table If None, does nothing.
779
    '''
780 2810 aaronmk
    if errors_table == None or cols == (): return
781
    run_query(db, mk_track_data_error(db, errors_table, cols, *args, **kw_args),
782 2758 aaronmk
        cacheable=True, log_level=4)
783
784 2704 aaronmk
def cast(db, type_, col, errors_table=None):
785
    '''Casts an (unrenamed) column or value.
786 2769 aaronmk
    If errors_table set and col has srcs, saves errors in errors_table (using
787
    col's srcs attr as the source columns) and converts errors to warnings.
788 2820 aaronmk
    @param col str|sql_gen.Col|sql_gen.Literal
789 2704 aaronmk
    @param errors_table None|sql_gen.Table|str
790
    '''
791 2820 aaronmk
    col = sql_gen.as_Col(col)
792 2769 aaronmk
    save_errors = (errors_table != None and isinstance(col, sql_gen.Col)
793
        and col.srcs != ())
794
    if not save_errors: # can't save errors
795
        return sql_gen.CustomCode(col.to_str(db)+'::'+type_) # just cast
796 2704 aaronmk
797
    assert not isinstance(col, sql_gen.NamedCol)
798
799 2770 aaronmk
    errors_table = sql_gen.as_Table(errors_table)
800
    srcs = map(sql_gen.to_name_only_col, col.srcs)
801
    function_name = str(sql_gen.FunctionCall(type_, *srcs))
802 2704 aaronmk
    function = sql_gen.TempFunction(function_name, db.autocommit)
803
804 2718 aaronmk
    while True:
805
        # Create function definition
806
        query = '''\
807 2704 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''(value text)
808
RETURNS '''+type_+'''
809
LANGUAGE plpgsql
810
STRICT
811
AS $$
812
BEGIN
813
    /* The explicit cast to the return type is needed to make the cast happen
814
    inside the try block. (Implicit casts to the return type happen at the end
815
    of the function, outside any block.) */
816
    RETURN value::'''+type_+''';
817
EXCEPTION
818
    WHEN data_exception THEN
819
        -- Save error in errors table.
820 2751 aaronmk
        -- Insert the value and error for *each* source column.
821
'''+mk_track_data_error(db, errors_table, srcs,
822
    *map(sql_gen.CustomCode, ['value', 'SQLSTATE', 'SQLERRM']))+''';
823 2704 aaronmk
824
        RAISE WARNING '%', SQLERRM;
825
        RETURN NULL;
826
END;
827
$$;
828
'''
829 2718 aaronmk
830
        # Create function
831
        try:
832
            run_query(db, query, recover=True, cacheable=True,
833
                log_ignore_excs=(DuplicateFunctionException,))
834
            break # successful
835
        except DuplicateFunctionException:
836 2770 aaronmk
            function.name = next_version(function.name)
837
            # try again with next version of name
838 2704 aaronmk
839
    return sql_gen.FunctionCall(function, col)
840
841 2414 aaronmk
##### Database structure queries
842
843 2426 aaronmk
def table_row_count(db, table, recover=None):
844 2786 aaronmk
    return value(run_query(db, mk_select(db, table, [sql_gen.row_count],
845 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
846 2426 aaronmk
847 2414 aaronmk
def table_cols(db, table, recover=None):
848
    return list(col_names(select(db, table, limit=0, order_by=None,
849 2443 aaronmk
        recover=recover, log_level=4)))
850 2414 aaronmk
851 2291 aaronmk
def pkey(db, table, recover=None):
852 832 aaronmk
    '''Assumed to be first column in table'''
853 2339 aaronmk
    return table_cols(db, table, recover)[0]
854 832 aaronmk
855 2559 aaronmk
not_null_col = 'not_null_col'
856 2340 aaronmk
857
def table_not_null_col(db, table, recover=None):
858
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
859
    if not_null_col in table_cols(db, table, recover): return not_null_col
860
    else: return pkey(db, table, recover)
861
862 853 aaronmk
def index_cols(db, table, index):
863
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
864
    automatically created. When you don't know whether something is a UNIQUE
865
    constraint or a UNIQUE index, use this function.'''
866 1909 aaronmk
    module = util.root_module(db.db)
867
    if module == 'psycopg2':
868
        return list(values(run_query(db, '''\
869 853 aaronmk
SELECT attname
870 866 aaronmk
FROM
871
(
872
        SELECT attnum, attname
873
        FROM pg_index
874
        JOIN pg_class index ON index.oid = indexrelid
875
        JOIN pg_class table_ ON table_.oid = indrelid
876
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
877
        WHERE
878 2782 aaronmk
            table_.relname = '''+db.esc_value(table)+'''
879
            AND index.relname = '''+db.esc_value(index)+'''
880 866 aaronmk
    UNION
881
        SELECT attnum, attname
882
        FROM
883
        (
884
            SELECT
885
                indrelid
886
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
887
                    AS indkey
888
            FROM pg_index
889
            JOIN pg_class index ON index.oid = indexrelid
890
            JOIN pg_class table_ ON table_.oid = indrelid
891
            WHERE
892 2782 aaronmk
                table_.relname = '''+db.esc_value(table)+'''
893
                AND index.relname = '''+db.esc_value(index)+'''
894 866 aaronmk
        ) s
895
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
896
) s
897 853 aaronmk
ORDER BY attnum
898 2782 aaronmk
'''
899
            , cacheable=True, log_level=4)))
900 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
901
        ' database')
902 853 aaronmk
903 464 aaronmk
def constraint_cols(db, table, constraint):
904 1849 aaronmk
    module = util.root_module(db.db)
905 464 aaronmk
    if module == 'psycopg2':
906
        return list(values(run_query(db, '''\
907
SELECT attname
908
FROM pg_constraint
909
JOIN pg_class ON pg_class.oid = conrelid
910
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
911
WHERE
912 2783 aaronmk
    relname = '''+db.esc_value(table)+'''
913
    AND conname = '''+db.esc_value(constraint)+'''
914 464 aaronmk
ORDER BY attnum
915 2783 aaronmk
'''
916
            )))
917 464 aaronmk
    else: raise NotImplementedError("Can't list constraint columns for "+module+
918
        ' database')
919
920 2096 aaronmk
row_num_col = '_row_num'
921
922 2847 aaronmk
def add_index(db, exprs, table=None, unique=False, ensure_not_null=True):
923 2688 aaronmk
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
924 2538 aaronmk
    Currently, only function calls are supported as expressions.
925 2847 aaronmk
    @param ensure_not_null If set, translates NULL values to sentinel values.
926
        This allows indexes to be used for comparisons where NULLs are equal.
927 2538 aaronmk
    '''
928 2688 aaronmk
    if not lists.is_seq(exprs): exprs = [exprs]
929 2538 aaronmk
930 2688 aaronmk
    # Parse exprs
931
    old_exprs = exprs[:]
932
    exprs = []
933
    cols = []
934
    for i, expr in enumerate(old_exprs):
935
        expr = copy.deepcopy(expr) # don't modify input!
936 2823 aaronmk
        expr = sql_gen.as_Col(expr, table)
937 2688 aaronmk
938 2823 aaronmk
        # Handle nullable columns
939 2860 aaronmk
        if ensure_not_null:
940
            try: expr = sql_gen.ensure_not_null(db, expr)
941
            except KeyError: pass # unknown type, so just create plain index
942 2823 aaronmk
943 2688 aaronmk
        # Extract col
944
        if isinstance(expr, sql_gen.FunctionCall):
945
            col = expr.args[0]
946
            expr = sql_gen.Expr(expr)
947
        else: col = expr
948 2823 aaronmk
        assert isinstance(col, sql_gen.Col)
949 2688 aaronmk
950
        # Extract table
951
        if table == None:
952
            assert sql_gen.is_table_col(col)
953
            table = col.table
954
955
        col.table = None
956
957
        exprs.append(expr)
958
        cols.append(col)
959 2408 aaronmk
960 2688 aaronmk
    table = sql_gen.as_Table(table)
961
    index = sql_gen.Table(str(sql_gen.Col(','.join(map(str, cols)), table)))
962
963
    str_ = 'CREATE'
964
    if unique: str_ += ' UNIQUE'
965
    str_ += ' INDEX '+index.to_str(db)+' ON '+table.to_str(db)+' ('+(
966
        ', '.join((v.to_str(db) for v in exprs)))+')'
967
968
    try: run_query(db, str_, recover=True, cacheable=True, log_level=3)
969 2408 aaronmk
    except DuplicateTableException: pass # index already existed
970
971 2686 aaronmk
def add_pkey(db, table, cols=None, recover=None):
972
    '''Adds a primary key.
973
    @param cols [sql_gen.Col,...] The columns in the primary key.
974
        Defaults to the first column in the table.
975 2406 aaronmk
    @pre The table must not already have a primary key.
976
    '''
977
    table = sql_gen.as_Table(table)
978 2686 aaronmk
    if cols == None: cols = [pkey(db, table, recover)]
979
    col_strs = [sql_gen.to_name_only_col(v).to_str(db) for v in cols]
980 2406 aaronmk
981 2781 aaronmk
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD PRIMARY KEY ('
982
        +(', '.join(col_strs))+')', recover=True, cacheable=True, log_level=3,
983
        log_ignore_excs=(DuplicateTableException,))
984 2406 aaronmk
985 2772 aaronmk
already_indexed = object() # tells add_indexes() the pkey has already been added
986
987
def add_indexes(db, table, has_pkey=True):
988
    '''Adds an index on all columns in a table.
989
    @param has_pkey bool|already_indexed Whether a pkey instead of a regular
990
        index should be added on the first column.
991
        * If already_indexed, the pkey is assumed to have already been added
992
    '''
993
    cols = table_cols(db, table)
994
    if has_pkey:
995
        if has_pkey is not already_indexed: add_pkey(db, table)
996
        cols = cols[1:]
997
    for col in cols: add_index(db, col, table)
998
999 2086 aaronmk
def add_row_num(db, table):
1000 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
1001
    be the primary key.'''
1002 2320 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
1003 2096 aaronmk
    run_query(db, 'ALTER TABLE '+table+' ADD COLUMN '+row_num_col
1004 2443 aaronmk
        +' serial NOT NULL PRIMARY KEY', log_level=3)
1005 2086 aaronmk
1006 2889 aaronmk
def drop_table(db, table):
1007
    table = sql_gen.as_Table(table)
1008
    return run_query(db, 'DROP TABLE IF EXISTS '+table.to_str(db)+' CASCADE')
1009
1010 2760 aaronmk
def create_table(db, table, cols, has_pkey=True, col_indexes=True):
1011 2675 aaronmk
    '''Creates a table.
1012 2681 aaronmk
    @param cols [sql_gen.TypedCol,...] The column names and types
1013
    @param has_pkey If set, the first column becomes the primary key.
1014 2760 aaronmk
    @param col_indexes bool|[ref]
1015
        * If True, indexes will be added on all non-pkey columns.
1016
        * If a list reference, [0] will be set to a function to do this.
1017
          This can be used to delay index creation until the table is populated.
1018 2675 aaronmk
    '''
1019
    table = sql_gen.as_Table(table)
1020
1021 2681 aaronmk
    if has_pkey:
1022
        cols[0] = pkey = copy.copy(cols[0]) # don't modify input!
1023 2872 aaronmk
        pkey.constraints = 'PRIMARY KEY'
1024 2681 aaronmk
1025 2675 aaronmk
    str_ = 'CREATE TABLE '+table.to_str(db)+' (\n'
1026 2681 aaronmk
    str_ += '\n, '.join(v.to_str(db) for v in cols)
1027
    str_ += '\n);\n'
1028 2675 aaronmk
    run_query(db, str_, cacheable=True, log_level=2)
1029 2760 aaronmk
1030
    # Add indexes
1031 2773 aaronmk
    if has_pkey: has_pkey = already_indexed
1032
    def add_indexes_(): add_indexes(db, table, has_pkey)
1033
    if isinstance(col_indexes, list): col_indexes[0] = add_indexes_ # defer
1034
    elif col_indexes: add_indexes_() # add now
1035 2675 aaronmk
1036 2684 aaronmk
def vacuum(db, table):
1037
    table = sql_gen.as_Table(table)
1038
    db.with_autocommit(lambda: run_query(db, 'VACUUM ANALYZE '+table.to_str(db),
1039
        log_level=3))
1040
1041 2732 aaronmk
def truncate(db, table, schema='public'):
1042 2777 aaronmk
    table = sql_gen.as_Table(table, schema)
1043 2733 aaronmk
    return run_query(db, 'TRUNCATE '+table.to_str(db)+' CASCADE')
1044 2732 aaronmk
1045 2729 aaronmk
def tables(db, schema_like='public', table_like='%', exact=False):
1046
    if exact: compare = '='
1047
    else: compare = 'LIKE'
1048
1049 1849 aaronmk
    module = util.root_module(db.db)
1050 832 aaronmk
    if module == 'psycopg2':
1051 2779 aaronmk
        conds = [('schemaname', sql_gen.CompareCond(schema_like, compare)),
1052
            ('tablename', sql_gen.CompareCond(table_like, compare))]
1053
        return values(select(db, 'pg_tables', ['tablename'], conds,
1054
            order_by='tablename', log_level=4))
1055 1968 aaronmk
    elif module == 'MySQLdb':
1056 2779 aaronmk
        return values(run_query(db, 'SHOW TABLES LIKE '+db.esc_value(table_like)
1057
            , cacheable=True, log_level=4))
1058 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
1059 830 aaronmk
1060 2726 aaronmk
def table_exists(db, table):
1061
    table = sql_gen.as_Table(table)
1062 2730 aaronmk
    return list(tables(db, table.schema, table.name, exact=True)) != []
1063 2726 aaronmk
1064
def errors_table(db, table, if_exists=True):
1065
    '''
1066
    @param if_exists If set, returns None if the errors table doesn't exist
1067
    @return None|sql_gen.Table
1068
    '''
1069
    table = sql_gen.as_Table(table)
1070
    if table.srcs != (): table = table.srcs[0]
1071
1072
    errors_table = sql_gen.suffixed_table(table, '.errors')
1073
    if if_exists and not table_exists(db, errors_table): return None
1074
    return errors_table
1075
1076 833 aaronmk
##### Database management
1077
1078 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
1079
    '''For kw_args, see tables()'''
1080
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
1081 833 aaronmk
1082 832 aaronmk
##### Heuristic queries
1083
1084 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
1085 1554 aaronmk
    '''Recovers from errors.
1086 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
1087
    '''
1088 2642 aaronmk
    row = sql_gen.ColDict(db, table, row)
1089 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
1090
1091 471 aaronmk
    try:
1092 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
1093 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
1094
            row_ct_ref[0] += cur.rowcount
1095
        return value(cur)
1096 471 aaronmk
    except DuplicateKeyException, e:
1097 2653 aaronmk
        row = sql_gen.ColDict(db, table,
1098
            util.dict_subset_right_join(row, e.cols))
1099
        return value(select(db, table, [pkey_], row, recover=True))
1100 471 aaronmk
1101 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
1102 830 aaronmk
    '''Recovers from errors'''
1103 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
1104 14 aaronmk
    except StopIteration:
1105 40 aaronmk
        if not create: raise
1106 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
1107 2078 aaronmk
1108 2593 aaronmk
def is_func_result(col):
1109
    return col.table.name.find('(') >= 0 and col.name == 'result'
1110
1111 2592 aaronmk
def into_table_name(out_table, in_tables0, mapping, is_func):
1112 2615 aaronmk
    def in_col_str(in_col):
1113
        in_col = sql_gen.remove_col_rename(in_col)
1114
        if isinstance(in_col, sql_gen.Col):
1115
            table = in_col.table
1116
            if table == in_tables0:
1117
                in_col = sql_gen.to_name_only_col(in_col)
1118
            elif is_func_result(in_col): in_col = table # omit col name
1119
        return str(in_col)
1120
1121 2580 aaronmk
    str_ = str(out_table)
1122
    if is_func:
1123 2615 aaronmk
        str_ += '('
1124 2580 aaronmk
1125 2615 aaronmk
        try: value_in_col = mapping['value']
1126
        except KeyError:
1127 2654 aaronmk
            str_ += ', '.join((str(k)+'='+in_col_str(v)
1128 2615 aaronmk
                for k, v in mapping.iteritems()))
1129
        else: str_ += in_col_str(value_in_col)
1130
1131
        str_ += ')'
1132 2812 aaronmk
    else:
1133
        out_col = 'rank'
1134
        try: in_col = mapping[out_col]
1135
        except KeyError: str_ += '_pkeys'
1136
        else: # has a rank column, so hierarchical
1137
            str_ += '['+str(out_col)+'='+in_col_str(in_col)+']'
1138 2580 aaronmk
    return str_
1139
1140 2508 aaronmk
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None, into=None,
1141 2805 aaronmk
    default=None, is_func=False, on_error=exc.raise_):
1142 2078 aaronmk
    '''Recovers from errors.
1143
    Only works under PostgreSQL (uses INSERT RETURNING).
1144 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
1145
        tables to join with it using the main input table's pkey
1146 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
1147 2612 aaronmk
        * out_table_col: str (*not* sql_gen.Col)
1148
        * in_table_col: sql_gen.Col|literal-value
1149 2489 aaronmk
    @param into The table to contain the output and input pkeys.
1150 2574 aaronmk
        Defaults to `out_table.name+'_pkeys'`.
1151 2509 aaronmk
    @param default The *output* column to use as the pkey for missing rows.
1152
        If this output column does not exist in the mapping, uses None.
1153 2552 aaronmk
    @param is_func Whether out_table is the name of a SQL function, not a table
1154 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
1155 2078 aaronmk
    '''
1156 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
1157 2312 aaronmk
1158 2450 aaronmk
    def log_debug(msg): db.log_debug(msg, level=1.5)
1159 2505 aaronmk
    def col_ustr(str_):
1160 2567 aaronmk
        return strings.repr_no_u(sql_gen.remove_col_rename(str_))
1161 2450 aaronmk
1162 2768 aaronmk
    out_pkey = pkey(db, out_table, recover=True)
1163
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
1164
1165
    if mapping == {}: # need at least one column for INSERT SELECT
1166
        mapping = {out_pkey: None} # ColDict will replace with default value
1167
1168 2486 aaronmk
    log_debug('********** New iteration **********')
1169 2505 aaronmk
    log_debug('Inserting these input columns into '+strings.as_tt(
1170
        out_table.to_str(db))+':\n'+strings.as_table(mapping, ustr=col_ustr))
1171 2463 aaronmk
1172 2382 aaronmk
    # Create input joins from list of input tables
1173
    in_tables_ = in_tables[:] # don't modify input!
1174
    in_tables0 = in_tables_.pop(0) # first table is separate
1175 2735 aaronmk
    errors_table_ = errors_table(db, in_tables0)
1176 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
1177 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
1178 2460 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v,
1179
        {in_pkey: sql_gen.join_same_not_null}) for v in in_tables_]
1180 2131 aaronmk
1181 2592 aaronmk
    if into == None:
1182
        into = into_table_name(out_table, in_tables0, mapping, is_func)
1183
    into = sql_gen.as_Table(into)
1184
1185 2702 aaronmk
    # Set column sources
1186
    in_cols = filter(sql_gen.is_table_col, mapping.values())
1187
    for col in in_cols:
1188
        if col.table == in_tables0: col.set_srcs(sql_gen.src_self)
1189
1190 2486 aaronmk
    log_debug('Joining together input tables into temp table')
1191 2395 aaronmk
    # Place in new table for speed and so don't modify input if values edited
1192 2584 aaronmk
    in_table = sql_gen.Table('in')
1193 2702 aaronmk
    mapping = dicts.join(mapping, flatten(db, in_table, input_joins, in_cols,
1194
        preserve=[in_pkey_col], start=0))
1195 2395 aaronmk
    input_joins = [in_table]
1196 2486 aaronmk
    db.log_debug('Temp table: '+strings.as_tt(in_table.to_str(db)), level=2)
1197 2395 aaronmk
1198 2656 aaronmk
    mapping = sql_gen.ColDict(db, out_table, mapping)
1199
        # after applying dicts.join() because that returns a plain dict
1200
1201 2509 aaronmk
    # Resolve default value column
1202 2841 aaronmk
    if default != None:
1203
        try: default = mapping[default]
1204
        except KeyError:
1205 2509 aaronmk
            db.log_debug('Default value column '
1206
                +strings.as_tt(strings.repr_no_u(default))
1207 2511 aaronmk
                +' does not exist in mapping, falling back to None', level=2.1)
1208 2509 aaronmk
            default = None
1209
1210 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
1211 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
1212
1213 2201 aaronmk
    pkeys_table_exists_ref = [False]
1214 2420 aaronmk
    def insert_into_pkeys(joins, cols):
1215 2786 aaronmk
        query = mk_select(db, joins, cols, order_by=None, start=0)
1216 2201 aaronmk
        if pkeys_table_exists_ref[0]:
1217 2786 aaronmk
            insert_select(db, into, pkeys_names, query)
1218 2201 aaronmk
        else:
1219 2786 aaronmk
            run_query_into(db, query, into=into)
1220 2201 aaronmk
            pkeys_table_exists_ref[0] = True
1221
1222 2429 aaronmk
    limit_ref = [None]
1223 2380 aaronmk
    conds = set()
1224 2811 aaronmk
    distinct_on = sql_gen.ColDict(db, out_table)
1225 2325 aaronmk
    def mk_main_select(joins, cols):
1226 2811 aaronmk
        distinct_on_cols = [c.to_Col() for c in distinct_on.values()]
1227
        return mk_select(db, joins, cols, conds, distinct_on_cols,
1228 2429 aaronmk
            limit=limit_ref[0], start=0)
1229 2132 aaronmk
1230 2519 aaronmk
    exc_strs = set()
1231 2309 aaronmk
    def log_exc(e):
1232 2519 aaronmk
        e_str = exc.str_(e, first_line_only=True)
1233
        log_debug('Caught exception: '+e_str)
1234
        assert e_str not in exc_strs # avoid infinite loops
1235
        exc_strs.add(e_str)
1236 2735 aaronmk
1237 2451 aaronmk
    def remove_all_rows():
1238 2896 aaronmk
        log_debug('Ignoring all rows')
1239 2429 aaronmk
        limit_ref[0] = 0 # just create an empty pkeys table
1240 2735 aaronmk
1241 2758 aaronmk
    def ignore(in_col, value, e):
1242
        track_data_error(db, errors_table_, in_col.srcs, value, e.cause.pgcode,
1243
            e.cause.pgerror)
1244 2895 aaronmk
        log_debug('Ignoring rows with '+strings.as_tt(repr(in_col))+' = '
1245 2545 aaronmk
            +strings.as_tt(repr(value)))
1246 2895 aaronmk
1247 2758 aaronmk
    def remove_rows(in_col, value, e):
1248
        ignore(in_col, value, e)
1249 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
1250
        assert cond not in conds # avoid infinite loops
1251 2380 aaronmk
        conds.add(cond)
1252 2895 aaronmk
1253 2758 aaronmk
    def invalid2null(in_col, value, e):
1254
        ignore(in_col, value, e)
1255 2403 aaronmk
        update(db, in_table, [(in_col, None)],
1256
            sql_gen.ColValueCond(in_col, value))
1257 2245 aaronmk
1258 2589 aaronmk
    def insert_pkeys_table(which):
1259
        return sql_gen.Table(sql_gen.add_suffix(in_table.name,
1260
            '_insert_'+which+'_pkeys'))
1261
    insert_out_pkeys = insert_pkeys_table('out')
1262
    insert_in_pkeys = insert_pkeys_table('in')
1263
1264 2900 aaronmk
    log_debug('Locking output tables to prevent concurrent duplicate keys')
1265
    for table in [out_table, errors_table_]: lock_table(db, table, 'EXCLUSIVE')
1266
1267 2206 aaronmk
    # Do inserts and selects
1268 2642 aaronmk
    join_cols = sql_gen.ColDict(db, out_table)
1269 2206 aaronmk
    while True:
1270 2521 aaronmk
        if limit_ref[0] == 0: # special case
1271
            log_debug('Creating an empty pkeys table')
1272 2786 aaronmk
            cur = run_query_into(db, mk_select(db, out_table, [out_pkey],
1273 2521 aaronmk
                limit=limit_ref[0]), into=insert_out_pkeys)
1274
            break # don't do main case
1275
1276 2303 aaronmk
        has_joins = join_cols != {}
1277
1278 2903 aaronmk
        log_debug('Trying to insert new rows')
1279
1280 2305 aaronmk
        # Prepare to insert new rows
1281 2325 aaronmk
        insert_joins = input_joins[:] # don't modify original!
1282 2403 aaronmk
        insert_args = dict(recover=True, cacheable=False)
1283 2303 aaronmk
        if has_joins:
1284 2325 aaronmk
            insert_joins.append(sql_gen.Join(out_table, join_cols,
1285
                sql_gen.filter_out))
1286
        else:
1287 2404 aaronmk
            insert_args.update(dict(returning=out_pkey, into=insert_out_pkeys))
1288 2786 aaronmk
        main_select = mk_main_select(insert_joins, mapping.values())
1289 2303 aaronmk
1290 2206 aaronmk
        try:
1291 2518 aaronmk
            cur = insert_select(db, out_table, mapping.keys(), main_select,
1292
                **insert_args)
1293 2357 aaronmk
            break # insert successful
1294 2206 aaronmk
        except DuplicateKeyException, e:
1295 2309 aaronmk
            log_exc(e)
1296
1297 2258 aaronmk
            old_join_cols = join_cols.copy()
1298 2811 aaronmk
            distinct_on.update(util.dict_subset(mapping, e.cols))
1299 2565 aaronmk
            join_cols.update(util.dict_subset_right_join(mapping, e.cols))
1300 2486 aaronmk
            log_debug('Ignoring existing rows, comparing on these columns:\n'
1301 2505 aaronmk
                +strings.as_inline_table(join_cols, ustr=col_ustr))
1302 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
1303 2230 aaronmk
        except NullValueException, e:
1304 2309 aaronmk
            log_exc(e)
1305
1306 2230 aaronmk
            out_col, = e.cols
1307
            try: in_col = mapping[out_col]
1308 2356 aaronmk
            except KeyError:
1309 2486 aaronmk
                log_debug('Missing mapping for NOT NULL column '+out_col)
1310 2451 aaronmk
                remove_all_rows()
1311 2758 aaronmk
            else: remove_rows(in_col, None, e)
1312 2542 aaronmk
        except FunctionValueException, e:
1313
            log_exc(e)
1314
1315
            func_name = e.name
1316
            value = e.value
1317
            for out_col, in_col in mapping.iteritems():
1318 2758 aaronmk
                in_col = sql_gen.unwrap_func_call(in_col, func_name)
1319
                invalid2null(in_col, value, e)
1320 2525 aaronmk
        except MissingCastException, e:
1321
            log_exc(e)
1322
1323
            out_col = e.col
1324 2706 aaronmk
            type_ = e.type
1325
1326 2724 aaronmk
            log_debug('Casting '+strings.as_tt(out_col)+' input to '
1327
                +strings.as_tt(type_))
1328 2735 aaronmk
            def wrap_func(col): return cast(db, type_, col, errors_table_)
1329 2705 aaronmk
            mapping[out_col] = sql_gen.wrap(wrap_func, mapping[out_col])
1330 2429 aaronmk
        except DatabaseErrors, e:
1331
            log_exc(e)
1332
1333 2808 aaronmk
            log_debug('No handler for exception')
1334
            on_error(e)
1335 2451 aaronmk
            remove_all_rows()
1336 2358 aaronmk
        # after exception handled, rerun loop with additional constraints
1337 2132 aaronmk
1338 2357 aaronmk
    if row_ct_ref != None and cur.rowcount >= 0:
1339
        row_ct_ref[0] += cur.rowcount
1340
1341
    if has_joins:
1342
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
1343 2486 aaronmk
        log_debug('Getting output table pkeys of existing/inserted rows')
1344 2420 aaronmk
        insert_into_pkeys(select_joins, pkeys_cols)
1345 2357 aaronmk
    else:
1346 2404 aaronmk
        add_row_num(db, insert_out_pkeys) # for joining with input pkeys
1347 2357 aaronmk
1348 2486 aaronmk
        log_debug('Getting input table pkeys of inserted rows')
1349 2786 aaronmk
        run_query_into(db, mk_main_select(input_joins, [in_pkey]),
1350 2404 aaronmk
            into=insert_in_pkeys)
1351
        add_row_num(db, insert_in_pkeys) # for joining with output pkeys
1352 2357 aaronmk
1353 2428 aaronmk
        assert table_row_count(db, insert_out_pkeys) == table_row_count(db,
1354
            insert_in_pkeys)
1355
1356 2486 aaronmk
        log_debug('Combining output and input pkeys in inserted order')
1357 2404 aaronmk
        pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
1358 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
1359 2420 aaronmk
        insert_into_pkeys(pkey_joins, pkeys_names)
1360 2357 aaronmk
1361 2486 aaronmk
    db.log_debug('Adding pkey on pkeys table to enable fast joins', level=2.5)
1362 2594 aaronmk
    add_pkey(db, into)
1363 2407 aaronmk
1364 2508 aaronmk
    log_debug('Setting pkeys of missing rows to '+strings.as_tt(repr(default)))
1365 2489 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(into,
1366 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
1367
        # must use join_same_not_null or query will take forever
1368 2420 aaronmk
    insert_into_pkeys(missing_rows_joins,
1369 2508 aaronmk
        [in_pkey_col, sql_gen.NamedCol(out_pkey, default)])
1370 2357 aaronmk
1371 2489 aaronmk
    assert table_row_count(db, into) == table_row_count(db, in_table)
1372 2428 aaronmk
1373 2702 aaronmk
    srcs = []
1374
    if is_func: srcs = sql_gen.cols_srcs(in_cols)
1375
    return sql_gen.Col(out_pkey, into, srcs)
1376 2115 aaronmk
1377
##### Data cleanup
1378
1379 2290 aaronmk
def cleanup_table(db, table, cols):
1380 2677 aaronmk
    table = sql_gen.as_Table(table)
1381
    cols = map(sql_gen.as_Col, cols)
1382 2115 aaronmk
1383 2677 aaronmk
    expr = ('nullif(nullif(trim(both from %s), '+db.esc_value('')+'), '
1384
        +db.esc_value(r'\N')+')')
1385
    changes = [(v, sql_gen.CustomCode(expr % v.to_str(db)))
1386
        for v in cols]
1387 2115 aaronmk
1388 2677 aaronmk
    update(db, table, changes)