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 2905 aaronmk
    def __init__(self, db_config, autocommit=False, caching=True,
152
        log_debug=log_debug_none):
153 1869 aaronmk
        self.db_config = db_config
154 2190 aaronmk
        self.autocommit = autocommit
155
        self.caching = caching
156 1901 aaronmk
        self.log_debug = log_debug
157 2193 aaronmk
        self.debug = log_debug != log_debug_none
158 1869 aaronmk
159
        self.__db = None
160 1889 aaronmk
        self.query_results = {}
161 2139 aaronmk
        self._savepoint = 0
162 2671 aaronmk
        self._notices_seen = set()
163 1869 aaronmk
164
    def __getattr__(self, name):
165
        if name == '__dict__': raise Exception('getting __dict__')
166
        if name == 'db': return self._db()
167
        else: raise AttributeError()
168
169
    def __getstate__(self):
170
        state = copy.copy(self.__dict__) # shallow copy
171 1915 aaronmk
        state['log_debug'] = None # don't pickle the debug callback
172 1869 aaronmk
        state['_DbConn__db'] = None # don't pickle the connection
173
        return state
174
175 2165 aaronmk
    def connected(self): return self.__db != None
176
177 1869 aaronmk
    def _db(self):
178
        if self.__db == None:
179
            # Process db_config
180
            db_config = self.db_config.copy() # don't modify input!
181 2097 aaronmk
            schemas = db_config.pop('schemas', None)
182 1869 aaronmk
            module_name, mappings = db_engines[db_config.pop('engine')]
183
            module = __import__(module_name)
184
            _add_module(module)
185
            for orig, new in mappings.iteritems():
186
                try: util.rename_key(db_config, orig, new)
187
                except KeyError: pass
188
189
            # Connect
190
            self.__db = module.connect(**db_config)
191
192
            # Configure connection
193 2906 aaronmk
            if hasattr(self.db, 'set_isolation_level'):
194
                import psycopg2.extensions
195
                self.db.set_isolation_level(
196
                    psycopg2.extensions.ISOLATION_LEVEL_READ_COMMITTED)
197 2101 aaronmk
            if schemas != None:
198 2893 aaronmk
                search_path = [self.esc_name(s) for s in schemas.split(',')]
199
                search_path.append(value(run_query(self, 'SHOW search_path',
200
                    log_level=4)))
201
                run_query(self, 'SET search_path TO '+(','.join(search_path)),
202
                    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 2802 aaronmk
        def execute(self, query):
215 2764 aaronmk
            self._is_insert = query.startswith('INSERT')
216 2797 aaronmk
            self.query_lookup = query
217 2148 aaronmk
            try:
218 2191 aaronmk
                try:
219 2802 aaronmk
                    cur = self.inner.execute(query)
220 2191 aaronmk
                    self.outer.do_autocommit()
221 2802 aaronmk
                finally: self.query = get_cur_query(self.inner, query)
222 1904 aaronmk
            except Exception, e:
223 2802 aaronmk
                _add_cursor_info(e, self, query)
224 1904 aaronmk
                self.result = e # cache the exception as the result
225
                self._cache_result()
226
                raise
227 2800 aaronmk
            if self.rowcount == 0 and query.startswith('SELECT'): # empty SELECT
228
                consume_rows(self) # fetch all rows so result will be cached
229 2762 aaronmk
            return cur
230 1894 aaronmk
231 1891 aaronmk
        def fetchone(self):
232
            row = self.inner.fetchone()
233 1899 aaronmk
            if row != None: self.result.append(row)
234
            # otherwise, fetched all rows
235 1904 aaronmk
            else: self._cache_result()
236
            return row
237
238
        def _cache_result(self):
239 1906 aaronmk
            # For inserts, only cache exceptions since inserts are not
240
            # idempotent, but an invalid insert will always be invalid
241 1930 aaronmk
            if self.query_results != None and (not self._is_insert
242 1906 aaronmk
                or isinstance(self.result, Exception)):
243
244 1894 aaronmk
                assert self.query_lookup != None
245 1916 aaronmk
                self.query_results[self.query_lookup] = self.CacheCursor(
246
                    util.dict_subset(dicts.AttrsDictView(self),
247
                    ['query', 'result', 'rowcount', 'description']))
248 1906 aaronmk
249 1916 aaronmk
        class CacheCursor:
250
            def __init__(self, cached_result): self.__dict__ = cached_result
251
252 1927 aaronmk
            def execute(self, *args, **kw_args):
253 1916 aaronmk
                if isinstance(self.result, Exception): raise self.result
254
                # otherwise, result is a rows list
255
                self.iter = iter(self.result)
256
257
            def fetchone(self):
258
                try: return self.iter.next()
259
                except StopIteration: return None
260 1891 aaronmk
261 2212 aaronmk
    def esc_value(self, value):
262 2663 aaronmk
        try: str_ = self.mogrify('%s', [value])
263
        except NotImplementedError, e:
264
            module = util.root_module(self.db)
265
            if module == 'MySQLdb':
266
                import _mysql
267
                str_ = _mysql.escape_string(value)
268
            else: raise e
269 2374 aaronmk
        return strings.to_unicode(str_)
270 2212 aaronmk
271 2347 aaronmk
    def esc_name(self, name): return esc_name(self, name) # calls global func
272
273 2814 aaronmk
    def std_code(self, str_):
274
        '''Standardizes SQL code.
275
        * Ensures that string literals are prefixed by `E`
276
        '''
277
        if str_.startswith("'"): str_ = 'E'+str_
278
        return str_
279
280 2665 aaronmk
    def can_mogrify(self):
281 2663 aaronmk
        module = util.root_module(self.db)
282 2665 aaronmk
        return module == 'psycopg2'
283 2663 aaronmk
284 2665 aaronmk
    def mogrify(self, query, params=None):
285
        if self.can_mogrify(): return self.db.cursor().mogrify(query, params)
286
        else: raise NotImplementedError("Can't mogrify query")
287
288 2671 aaronmk
    def print_notices(self):
289 2725 aaronmk
        if hasattr(self.db, 'notices'):
290
            for msg in self.db.notices:
291
                if msg not in self._notices_seen:
292
                    self._notices_seen.add(msg)
293
                    self.log_debug(msg, level=2)
294 2671 aaronmk
295 2793 aaronmk
    def run_query(self, query, cacheable=False, log_level=2,
296 2464 aaronmk
        debug_msg_ref=None):
297 2445 aaronmk
        '''
298 2464 aaronmk
        @param log_ignore_excs The log_level will be increased by 2 if the query
299
            throws one of these exceptions.
300 2664 aaronmk
        @param debug_msg_ref If specified, the log message will be returned in
301
            this instead of being output. This allows you to filter log messages
302
            depending on the result of the query.
303 2445 aaronmk
        '''
304 2167 aaronmk
        assert query != None
305
306 2047 aaronmk
        if not self.caching: cacheable = False
307 1903 aaronmk
        used_cache = False
308 2664 aaronmk
309
        def log_msg(query):
310
            if used_cache: cache_status = 'cache hit'
311
            elif cacheable: cache_status = 'cache miss'
312
            else: cache_status = 'non-cacheable'
313
            return 'DB query: '+cache_status+':\n'+strings.as_code(query, 'SQL')
314
315 1903 aaronmk
        try:
316 1927 aaronmk
            # Get cursor
317
            if cacheable:
318
                try:
319 2797 aaronmk
                    cur = self.query_results[query]
320 1927 aaronmk
                    used_cache = True
321
                except KeyError: cur = self.DbCursor(self)
322
            else: cur = self.db.cursor()
323
324 2664 aaronmk
            # Log query
325
            if self.debug and debug_msg_ref == None: # log before running
326
                self.log_debug(log_msg(query), log_level)
327
328 1927 aaronmk
            # Run query
329 2793 aaronmk
            cur.execute(query)
330 1903 aaronmk
        finally:
331 2671 aaronmk
            self.print_notices()
332 2664 aaronmk
            if self.debug and debug_msg_ref != None: # return after running
333 2793 aaronmk
                debug_msg_ref[0] = log_msg(str(get_cur_query(cur, query)))
334 1903 aaronmk
335
        return cur
336 1914 aaronmk
337 2797 aaronmk
    def is_cached(self, query): return query in self.query_results
338 2139 aaronmk
339 2683 aaronmk
    def with_autocommit(self, func, autocommit=True):
340 2801 aaronmk
        import psycopg2.extensions
341
        if autocommit:
342
            isolation_level = psycopg2.extensions.ISOLATION_LEVEL_AUTOCOMMIT
343
        else: isolation_level = psycopg2.extensions.ISOLATION_LEVEL_SERIALIZABLE
344
345
        prev_isolation_level = self.db.isolation_level
346
        self.db.set_isolation_level(isolation_level)
347 2683 aaronmk
        try: return func()
348 2801 aaronmk
        finally: self.db.set_isolation_level(prev_isolation_level)
349 2683 aaronmk
350 2139 aaronmk
    def with_savepoint(self, func):
351 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
352 2443 aaronmk
        self.run_query('SAVEPOINT '+savepoint, log_level=4)
353 2139 aaronmk
        self._savepoint += 1
354 2683 aaronmk
        try:
355 2139 aaronmk
            try: return_val = func()
356
            finally:
357
                self._savepoint -= 1
358
                assert self._savepoint >= 0
359
        except:
360 2443 aaronmk
            self.run_query('ROLLBACK TO SAVEPOINT '+savepoint, log_level=4)
361 2139 aaronmk
            raise
362
        else:
363 2443 aaronmk
            self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
364 2191 aaronmk
            self.do_autocommit()
365 2139 aaronmk
            return return_val
366 2191 aaronmk
367
    def do_autocommit(self):
368
        '''Autocommits if outside savepoint'''
369
        assert self._savepoint >= 0
370
        if self.autocommit and self._savepoint == 0:
371 2668 aaronmk
            self.log_debug('Autocommitting')
372 2191 aaronmk
            self.db.commit()
373 2643 aaronmk
374 2819 aaronmk
    def col_info(self, col):
375 2643 aaronmk
        table = sql_gen.Table('columns', 'information_schema')
376 2821 aaronmk
        cols = ['data_type', 'column_default',
377
            cast(self, 'boolean', 'is_nullable')]
378 2643 aaronmk
379
        conds = [('table_name', col.table.name), ('column_name', col.name)]
380
        schema = col.table.schema
381
        if schema != None: conds.append(('table_schema', schema))
382
383 2819 aaronmk
        type_, default, nullable = row(select(self, table, cols, conds,
384 2874 aaronmk
            order_by='table_schema', limit=1, log_level=4))
385 2643 aaronmk
            # TODO: order_by search_path schema order
386 2819 aaronmk
        default = sql_gen.as_Code(default, self)
387
388
        return sql_gen.TypedCol(col.name, type_, default, nullable)
389 1849 aaronmk
390 1869 aaronmk
connect = DbConn
391
392 832 aaronmk
##### Recoverable querying
393 15 aaronmk
394 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
395 11 aaronmk
396 2791 aaronmk
def run_query(db, query, recover=None, cacheable=False, log_level=2,
397
    log_ignore_excs=None, **kw_args):
398 2794 aaronmk
    '''For params, see DbConn.run_query()'''
399 830 aaronmk
    if recover == None: recover = False
400 2464 aaronmk
    if log_ignore_excs == None: log_ignore_excs = ()
401
    log_ignore_excs = tuple(log_ignore_excs)
402 830 aaronmk
403 2666 aaronmk
    debug_msg_ref = None # usually, db.run_query() logs query before running it
404
    # But if filtering with log_ignore_excs, wait until after exception parsing
405
    if log_ignore_excs != () or not db.can_mogrify(): debug_msg_ref = [None]
406
407 2148 aaronmk
    try:
408 2464 aaronmk
        try:
409 2794 aaronmk
            def run(): return db.run_query(query, cacheable, log_level,
410 2793 aaronmk
                debug_msg_ref, **kw_args)
411 2796 aaronmk
            if recover and not db.is_cached(query):
412 2464 aaronmk
                return with_savepoint(db, run)
413
            else: return run() # don't need savepoint if cached
414
        except Exception, e:
415
            if not recover: raise # need savepoint to run index_cols()
416
            msg = exc.str_(e)
417
418
            match = re.search(r'duplicate key value violates unique constraint '
419 2493 aaronmk
                r'"((_?[^\W_]+)_.+?)"', msg)
420 2464 aaronmk
            if match:
421
                constraint, table = match.groups()
422
                try: cols = index_cols(db, table, constraint)
423
                except NotImplementedError: raise e
424
                else: raise DuplicateKeyException(constraint, cols, e)
425
426 2493 aaronmk
            match = re.search(r'null value in column "(.+?)" violates not-null'
427 2464 aaronmk
                r' constraint', msg)
428
            if match: raise NullValueException('NOT NULL', [match.group(1)], e)
429
430
            match = re.search(r'\b(?:invalid input (?:syntax|value)\b.*?'
431
                r'|date/time field value out of range): "(.+?)"\n'
432 2535 aaronmk
                r'(?:(?s).*?)\bfunction "(.+?)"', msg)
433 2464 aaronmk
            if match:
434
                value, name = match.groups()
435
                raise FunctionValueException(name, strings.to_unicode(value), e)
436
437 2526 aaronmk
            match = re.search(r'column "(.+?)" is of type (.+?) but expression '
438 2523 aaronmk
                r'is of type', msg)
439
            if match:
440
                col, type_ = match.groups()
441
                raise MissingCastException(type_, col, e)
442
443 2493 aaronmk
            match = re.search(r'relation "(.+?)" already exists', msg)
444 2464 aaronmk
            if match: raise DuplicateTableException(match.group(1), e)
445
446 2493 aaronmk
            match = re.search(r'function "(.+?)" already exists', msg)
447 2464 aaronmk
            if match: raise DuplicateFunctionException(match.group(1), e)
448
449
            raise # no specific exception raised
450
    except log_ignore_excs:
451
        log_level += 2
452
        raise
453
    finally:
454 2666 aaronmk
        if debug_msg_ref != None and debug_msg_ref[0] != None:
455
            db.log_debug(debug_msg_ref[0], log_level)
456 830 aaronmk
457 832 aaronmk
##### Basic queries
458
459 2153 aaronmk
def next_version(name):
460 2163 aaronmk
    version = 1 # first existing name was version 0
461 2586 aaronmk
    match = re.match(r'^(.*)#(\d+)$', name)
462 2153 aaronmk
    if match:
463 2586 aaronmk
        name, version = match.groups()
464
        version = int(version)+1
465 2588 aaronmk
    return sql_gen.add_suffix(name, '#'+str(version))
466 2153 aaronmk
467 2899 aaronmk
def lock_table(db, table, mode):
468
    table = sql_gen.as_Table(table)
469
    run_query(db, 'LOCK TABLE '+table.to_str(db)+' IN '+mode+' MODE')
470
471 2789 aaronmk
def run_query_into(db, query, into=None, add_indexes_=False, **kw_args):
472 2085 aaronmk
    '''Outputs a query to a temp table.
473
    For params, see run_query().
474
    '''
475 2789 aaronmk
    if into == None: return run_query(db, query, **kw_args)
476 2790 aaronmk
477
    assert isinstance(into, sql_gen.Table)
478
479
    kw_args['recover'] = True
480
    kw_args.setdefault('log_ignore_excs', (DuplicateTableException,))
481
482
    temp = not db.autocommit # tables are permanent in autocommit mode
483
    # "temporary tables cannot specify a schema name", so remove schema
484
    if temp: into.schema = None
485
486
    # Create table
487
    while True:
488
        create_query = 'CREATE'
489
        if temp: create_query += ' TEMP'
490
        create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
491 2385 aaronmk
492 2790 aaronmk
        try:
493
            cur = run_query(db, create_query, **kw_args)
494
                # CREATE TABLE AS sets rowcount to # rows in query
495
            break
496
        except DuplicateTableException, e:
497
            into.name = next_version(into.name)
498
            # try again with next version of name
499
500
    if add_indexes_: add_indexes(db, into)
501
502
    return cur
503 2085 aaronmk
504 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
505
506 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
507
508 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
509 2293 aaronmk
    start=None, order_by=order_by_pkey, default_table=None):
510 1981 aaronmk
    '''
511 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
512 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
513 1981 aaronmk
    @param fields Use None to select all fields in the table
514 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
515 2379 aaronmk
        * container can be any iterable type
516 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
517
        * compare_right_side: sql_gen.ValueCond|literal value
518 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
519
        use all columns
520 2786 aaronmk
    @return query
521 1981 aaronmk
    '''
522 2315 aaronmk
    # Parse tables param
523 2121 aaronmk
    if not lists.is_seq(tables): tables = [tables]
524 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
525 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
526 2121 aaronmk
527 2315 aaronmk
    # Parse other params
528 2376 aaronmk
    if conds == None: conds = []
529 2650 aaronmk
    elif dicts.is_dict(conds): conds = conds.items()
530 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
531 135 aaronmk
    assert limit == None or type(limit) == int
532 865 aaronmk
    assert start == None or type(start) == int
533 2315 aaronmk
    if order_by is order_by_pkey:
534
        if distinct_on != []: order_by = None
535
        else: order_by = pkey(db, table0, recover=True)
536 865 aaronmk
537 2315 aaronmk
    query = 'SELECT'
538 2056 aaronmk
539 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
540 2056 aaronmk
541 2200 aaronmk
    # DISTINCT ON columns
542 2233 aaronmk
    if distinct_on != []:
543 2467 aaronmk
        query += '\nDISTINCT'
544 2254 aaronmk
        if distinct_on is not distinct_on_all:
545 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
546
547
    # Columns
548 2467 aaronmk
    query += '\n'
549 1135 aaronmk
    if fields == None: query += '*'
550 2765 aaronmk
    else:
551
        assert fields != []
552
        query += '\n, '.join(map(parse_col, fields))
553 2200 aaronmk
554
    # Main table
555 2467 aaronmk
    query += '\nFROM '+table0.to_str(db)
556 865 aaronmk
557 2122 aaronmk
    # Add joins
558 2271 aaronmk
    left_table = table0
559 2263 aaronmk
    for join_ in tables:
560
        table = join_.table
561 2238 aaronmk
562 2343 aaronmk
        # Parse special values
563
        if join_.type_ is sql_gen.filter_out: # filter no match
564 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
565 2853 aaronmk
                sql_gen.CompareCond(None, '~=')))
566 2903 aaronmk
            # Prevent concurrent duplicate keys when used with INSERT SELECT
567
            lock_table(db, sql_gen.underlying_table(table), 'EXCLUSIVE')
568 2343 aaronmk
569 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
570 2122 aaronmk
571
        left_table = table
572
573 865 aaronmk
    missing = True
574 2376 aaronmk
    if conds != []:
575 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
576
        else: whitespace = '\n'
577 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
578
            .to_str(db) for l, r in conds], 'WHERE')
579 865 aaronmk
        missing = False
580 2227 aaronmk
    if order_by != None:
581 2467 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
582
    if limit != None: query += '\nLIMIT '+str(limit); missing = False
583 865 aaronmk
    if start != None:
584 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
585 865 aaronmk
        missing = False
586
    if missing: warnings.warn(DbWarning(
587
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
588
589 2786 aaronmk
    return query
590 11 aaronmk
591 2054 aaronmk
def select(db, *args, **kw_args):
592
    '''For params, see mk_select() and run_query()'''
593
    recover = kw_args.pop('recover', None)
594
    cacheable = kw_args.pop('cacheable', True)
595 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
596 2054 aaronmk
597 2791 aaronmk
    return run_query(db, mk_select(db, *args, **kw_args), recover, cacheable,
598
        log_level=log_level)
599 2054 aaronmk
600 2788 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, returning=None,
601
    embeddable=False):
602 1960 aaronmk
    '''
603
    @param returning str|None An inserted column (such as pkey) to return
604 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
605 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
606
        query will be fully cached, not just if it raises an exception.
607 1960 aaronmk
    '''
608 2754 aaronmk
    table = sql_gen.remove_table_rename(sql_gen.as_Table(table))
609 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
610 2646 aaronmk
    if cols != None:
611
        cols = [sql_gen.to_name_only_col(v, table).to_str(db) for v in cols]
612 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
613 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
614 2063 aaronmk
615
    # Build query
616 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
617
    query = first_line
618 2467 aaronmk
    if cols != None: query += '\n('+', '.join(cols)+')'
619
    query += '\n'+select_query
620 2063 aaronmk
621
    if returning != None:
622 2743 aaronmk
        query += '\nRETURNING '+sql_gen.to_name_only_col(returning).to_str(db)
623 2063 aaronmk
624 2070 aaronmk
    if embeddable:
625 2327 aaronmk
        assert returning != None
626
627 2070 aaronmk
        # Create function
628 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
629 2327 aaronmk
        return_type = 'SETOF '+returning.to_str(db)+'%TYPE'
630 2189 aaronmk
        while True:
631
            try:
632 2698 aaronmk
                function = sql_gen.TempFunction(function_name, db.autocommit)
633 2194 aaronmk
634 2189 aaronmk
                function_query = '''\
635 2698 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''()
636 2467 aaronmk
RETURNS '''+return_type+'''
637
LANGUAGE sql
638
AS $$
639 2788 aaronmk
'''+query+''';
640 2467 aaronmk
$$;
641 2070 aaronmk
'''
642 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
643 2464 aaronmk
                    log_ignore_excs=(DuplicateFunctionException,))
644 2189 aaronmk
                break # this version was successful
645
            except DuplicateFunctionException, e:
646
                function_name = next_version(function_name)
647
                # try again with next version of name
648 2070 aaronmk
649 2337 aaronmk
        # Return query that uses function
650 2698 aaronmk
        func_table = sql_gen.NamedTable('f', sql_gen.FunctionCall(function),
651 2742 aaronmk
            [returning]) # AS clause requires function alias
652 2787 aaronmk
        return mk_select(db, func_table, start=0, order_by=None)
653 2070 aaronmk
654 2787 aaronmk
    return query
655 2066 aaronmk
656
def insert_select(db, *args, **kw_args):
657 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
658 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
659
        values in
660 2072 aaronmk
    '''
661 2386 aaronmk
    into = kw_args.pop('into', None)
662
    if into != None: kw_args['embeddable'] = True
663 2066 aaronmk
    recover = kw_args.pop('recover', None)
664
    cacheable = kw_args.pop('cacheable', True)
665 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
666 2066 aaronmk
667 2789 aaronmk
    return run_query_into(db, mk_insert_select(db, *args, **kw_args), into,
668 2787 aaronmk
        recover=recover, cacheable=cacheable, log_level=log_level)
669 2063 aaronmk
670 2738 aaronmk
default = sql_gen.default # tells insert() to use the default value for a column
671 2066 aaronmk
672 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
673 2085 aaronmk
    '''For params, see insert_select()'''
674 1960 aaronmk
    if lists.is_seq(row): cols = None
675
    else:
676
        cols = row.keys()
677
        row = row.values()
678 2738 aaronmk
    row = list(row) # ensure that "== []" works
679 1960 aaronmk
680 2738 aaronmk
    if row == []: query = None
681
    else: query = sql_gen.Values(row).to_str(db)
682 1961 aaronmk
683 2788 aaronmk
    return insert_select(db, table, cols, query, *args, **kw_args)
684 11 aaronmk
685 2402 aaronmk
def mk_update(db, table, changes=None, cond=None):
686
    '''
687
    @param changes [(col, new_value),...]
688
        * container can be any iterable type
689
        * col: sql_gen.Code|str (for col name)
690
        * new_value: sql_gen.Code|literal value
691
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
692
    @return str query
693
    '''
694
    query = 'UPDATE '+sql_gen.as_Table(table).to_str(db)+'\nSET\n'
695 2405 aaronmk
    query += ',\n'.join((sql_gen.to_name_only_col(col, table).to_str(db)+' = '
696 2402 aaronmk
        +sql_gen.as_Value(new_value).to_str(db) for col, new_value in changes))
697 2467 aaronmk
    if cond != None: query += '\nWHERE\n'+cond.to_str(db)
698 2402 aaronmk
699
    return query
700
701
def update(db, *args, **kw_args):
702
    '''For params, see mk_update() and run_query()'''
703
    recover = kw_args.pop('recover', None)
704
705 2791 aaronmk
    return run_query(db, mk_update(db, *args, **kw_args), recover)
706 2402 aaronmk
707 135 aaronmk
def last_insert_id(db):
708 1849 aaronmk
    module = util.root_module(db.db)
709 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
710
    elif module == 'MySQLdb': return db.insert_id()
711
    else: return None
712 13 aaronmk
713 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
714 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
715 2415 aaronmk
    to names that will be distinct among the columns' tables.
716 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
717 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
718
    @param into The table for the new columns.
719 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
720
        columns will be included in the mapping even if they are not in cols.
721
        The tables of the provided Col objects will be changed to into, so make
722
        copies of them if you want to keep the original tables.
723
    @param as_items Whether to return a list of dict items instead of a dict
724 2383 aaronmk
    @return dict(orig_col=new_col, ...)
725
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
726 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
727
        * All mappings use the into table so its name can easily be
728 2383 aaronmk
          changed for all columns at once
729
    '''
730 2415 aaronmk
    cols = lists.uniqify(cols)
731
732 2394 aaronmk
    items = []
733 2389 aaronmk
    for col in preserve:
734 2390 aaronmk
        orig_col = copy.copy(col)
735 2392 aaronmk
        col.table = into
736 2394 aaronmk
        items.append((orig_col, col))
737
    preserve = set(preserve)
738
    for col in cols:
739 2716 aaronmk
        if col not in preserve:
740
            items.append((col, sql_gen.Col(str(col), into, col.srcs)))
741 2394 aaronmk
742
    if not as_items: items = dict(items)
743
    return items
744 2383 aaronmk
745 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
746 2391 aaronmk
    '''For params, see mk_flatten_mapping()
747
    @return See return value of mk_flatten_mapping()
748
    '''
749 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
750
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
751 2786 aaronmk
    run_query_into(db, mk_select(db, joins, cols, limit=limit, start=start),
752 2846 aaronmk
        into=into, add_indexes_=True)
753 2394 aaronmk
    return dict(items)
754 2391 aaronmk
755 2751 aaronmk
def mk_track_data_error(db, errors_table, cols, value, error_code, error):
756 2810 aaronmk
    assert cols != ()
757
758 2751 aaronmk
    cols = map(sql_gen.to_name_only_col, cols)
759
760
    columns_cols = ['column']
761
    columns = sql_gen.NamedValues('columns', columns_cols,
762
        [[c.name] for c in cols])
763
    values_cols = ['value', 'error_code', 'error']
764
    values = sql_gen.NamedValues('values', values_cols,
765
        [value, error_code, error])
766
767
    select_cols = columns_cols+values_cols
768
    name_only_cols = map(sql_gen.to_name_only_col, select_cols)
769 2755 aaronmk
    errors_table = sql_gen.NamedTable('errors', errors_table)
770 2751 aaronmk
    joins = [columns, sql_gen.Join(values, type_='CROSS'),
771
        sql_gen.Join(errors_table, dict(zip(name_only_cols, select_cols)),
772
        sql_gen.filter_out)]
773
774
    return mk_insert_select(db, errors_table, name_only_cols,
775 2787 aaronmk
        mk_select(db, joins, select_cols, order_by=None))
776 2751 aaronmk
777 2810 aaronmk
def track_data_error(db, errors_table, cols, *args, **kw_args):
778 2758 aaronmk
    '''
779
    @param errors_table If None, does nothing.
780
    '''
781 2810 aaronmk
    if errors_table == None or cols == (): return
782
    run_query(db, mk_track_data_error(db, errors_table, cols, *args, **kw_args),
783 2758 aaronmk
        cacheable=True, log_level=4)
784
785 2704 aaronmk
def cast(db, type_, col, errors_table=None):
786
    '''Casts an (unrenamed) column or value.
787 2769 aaronmk
    If errors_table set and col has srcs, saves errors in errors_table (using
788
    col's srcs attr as the source columns) and converts errors to warnings.
789 2820 aaronmk
    @param col str|sql_gen.Col|sql_gen.Literal
790 2704 aaronmk
    @param errors_table None|sql_gen.Table|str
791
    '''
792 2820 aaronmk
    col = sql_gen.as_Col(col)
793 2769 aaronmk
    save_errors = (errors_table != None and isinstance(col, sql_gen.Col)
794
        and col.srcs != ())
795
    if not save_errors: # can't save errors
796
        return sql_gen.CustomCode(col.to_str(db)+'::'+type_) # just cast
797 2704 aaronmk
798
    assert not isinstance(col, sql_gen.NamedCol)
799
800 2770 aaronmk
    errors_table = sql_gen.as_Table(errors_table)
801
    srcs = map(sql_gen.to_name_only_col, col.srcs)
802
    function_name = str(sql_gen.FunctionCall(type_, *srcs))
803 2704 aaronmk
    function = sql_gen.TempFunction(function_name, db.autocommit)
804
805 2718 aaronmk
    while True:
806
        # Create function definition
807
        query = '''\
808 2704 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''(value text)
809
RETURNS '''+type_+'''
810
LANGUAGE plpgsql
811
STRICT
812
AS $$
813
BEGIN
814
    /* The explicit cast to the return type is needed to make the cast happen
815
    inside the try block. (Implicit casts to the return type happen at the end
816
    of the function, outside any block.) */
817
    RETURN value::'''+type_+''';
818
EXCEPTION
819
    WHEN data_exception THEN
820
        -- Save error in errors table.
821 2751 aaronmk
        -- Insert the value and error for *each* source column.
822
'''+mk_track_data_error(db, errors_table, srcs,
823
    *map(sql_gen.CustomCode, ['value', 'SQLSTATE', 'SQLERRM']))+''';
824 2704 aaronmk
825
        RAISE WARNING '%', SQLERRM;
826
        RETURN NULL;
827
END;
828
$$;
829
'''
830 2718 aaronmk
831
        # Create function
832
        try:
833
            run_query(db, query, recover=True, cacheable=True,
834
                log_ignore_excs=(DuplicateFunctionException,))
835
            break # successful
836
        except DuplicateFunctionException:
837 2770 aaronmk
            function.name = next_version(function.name)
838
            # try again with next version of name
839 2704 aaronmk
840
    return sql_gen.FunctionCall(function, col)
841
842 2414 aaronmk
##### Database structure queries
843
844 2426 aaronmk
def table_row_count(db, table, recover=None):
845 2786 aaronmk
    return value(run_query(db, mk_select(db, table, [sql_gen.row_count],
846 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
847 2426 aaronmk
848 2414 aaronmk
def table_cols(db, table, recover=None):
849
    return list(col_names(select(db, table, limit=0, order_by=None,
850 2443 aaronmk
        recover=recover, log_level=4)))
851 2414 aaronmk
852 2291 aaronmk
def pkey(db, table, recover=None):
853 832 aaronmk
    '''Assumed to be first column in table'''
854 2339 aaronmk
    return table_cols(db, table, recover)[0]
855 832 aaronmk
856 2559 aaronmk
not_null_col = 'not_null_col'
857 2340 aaronmk
858
def table_not_null_col(db, table, recover=None):
859
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
860
    if not_null_col in table_cols(db, table, recover): return not_null_col
861
    else: return pkey(db, table, recover)
862
863 853 aaronmk
def index_cols(db, table, index):
864
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
865
    automatically created. When you don't know whether something is a UNIQUE
866
    constraint or a UNIQUE index, use this function.'''
867 1909 aaronmk
    module = util.root_module(db.db)
868
    if module == 'psycopg2':
869
        return list(values(run_query(db, '''\
870 853 aaronmk
SELECT attname
871 866 aaronmk
FROM
872
(
873
        SELECT attnum, attname
874
        FROM pg_index
875
        JOIN pg_class index ON index.oid = indexrelid
876
        JOIN pg_class table_ ON table_.oid = indrelid
877
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
878
        WHERE
879 2782 aaronmk
            table_.relname = '''+db.esc_value(table)+'''
880
            AND index.relname = '''+db.esc_value(index)+'''
881 866 aaronmk
    UNION
882
        SELECT attnum, attname
883
        FROM
884
        (
885
            SELECT
886
                indrelid
887
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
888
                    AS indkey
889
            FROM pg_index
890
            JOIN pg_class index ON index.oid = indexrelid
891
            JOIN pg_class table_ ON table_.oid = indrelid
892
            WHERE
893 2782 aaronmk
                table_.relname = '''+db.esc_value(table)+'''
894
                AND index.relname = '''+db.esc_value(index)+'''
895 866 aaronmk
        ) s
896
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
897
) s
898 853 aaronmk
ORDER BY attnum
899 2782 aaronmk
'''
900
            , cacheable=True, log_level=4)))
901 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
902
        ' database')
903 853 aaronmk
904 464 aaronmk
def constraint_cols(db, table, constraint):
905 1849 aaronmk
    module = util.root_module(db.db)
906 464 aaronmk
    if module == 'psycopg2':
907
        return list(values(run_query(db, '''\
908
SELECT attname
909
FROM pg_constraint
910
JOIN pg_class ON pg_class.oid = conrelid
911
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
912
WHERE
913 2783 aaronmk
    relname = '''+db.esc_value(table)+'''
914
    AND conname = '''+db.esc_value(constraint)+'''
915 464 aaronmk
ORDER BY attnum
916 2783 aaronmk
'''
917
            )))
918 464 aaronmk
    else: raise NotImplementedError("Can't list constraint columns for "+module+
919
        ' database')
920
921 2096 aaronmk
row_num_col = '_row_num'
922
923 2847 aaronmk
def add_index(db, exprs, table=None, unique=False, ensure_not_null=True):
924 2688 aaronmk
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
925 2538 aaronmk
    Currently, only function calls are supported as expressions.
926 2847 aaronmk
    @param ensure_not_null If set, translates NULL values to sentinel values.
927
        This allows indexes to be used for comparisons where NULLs are equal.
928 2538 aaronmk
    '''
929 2688 aaronmk
    if not lists.is_seq(exprs): exprs = [exprs]
930 2538 aaronmk
931 2688 aaronmk
    # Parse exprs
932
    old_exprs = exprs[:]
933
    exprs = []
934
    cols = []
935
    for i, expr in enumerate(old_exprs):
936
        expr = copy.deepcopy(expr) # don't modify input!
937 2823 aaronmk
        expr = sql_gen.as_Col(expr, table)
938 2688 aaronmk
939 2823 aaronmk
        # Handle nullable columns
940 2860 aaronmk
        if ensure_not_null:
941
            try: expr = sql_gen.ensure_not_null(db, expr)
942
            except KeyError: pass # unknown type, so just create plain index
943 2823 aaronmk
944 2688 aaronmk
        # Extract col
945
        if isinstance(expr, sql_gen.FunctionCall):
946
            col = expr.args[0]
947
            expr = sql_gen.Expr(expr)
948
        else: col = expr
949 2823 aaronmk
        assert isinstance(col, sql_gen.Col)
950 2688 aaronmk
951
        # Extract table
952
        if table == None:
953
            assert sql_gen.is_table_col(col)
954
            table = col.table
955
956
        col.table = None
957
958
        exprs.append(expr)
959
        cols.append(col)
960 2408 aaronmk
961 2688 aaronmk
    table = sql_gen.as_Table(table)
962
    index = sql_gen.Table(str(sql_gen.Col(','.join(map(str, cols)), table)))
963
964
    str_ = 'CREATE'
965
    if unique: str_ += ' UNIQUE'
966
    str_ += ' INDEX '+index.to_str(db)+' ON '+table.to_str(db)+' ('+(
967
        ', '.join((v.to_str(db) for v in exprs)))+')'
968
969
    try: run_query(db, str_, recover=True, cacheable=True, log_level=3)
970 2408 aaronmk
    except DuplicateTableException: pass # index already existed
971
972 2686 aaronmk
def add_pkey(db, table, cols=None, recover=None):
973
    '''Adds a primary key.
974
    @param cols [sql_gen.Col,...] The columns in the primary key.
975
        Defaults to the first column in the table.
976 2406 aaronmk
    @pre The table must not already have a primary key.
977
    '''
978
    table = sql_gen.as_Table(table)
979 2686 aaronmk
    if cols == None: cols = [pkey(db, table, recover)]
980
    col_strs = [sql_gen.to_name_only_col(v).to_str(db) for v in cols]
981 2406 aaronmk
982 2781 aaronmk
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD PRIMARY KEY ('
983
        +(', '.join(col_strs))+')', recover=True, cacheable=True, log_level=3,
984
        log_ignore_excs=(DuplicateTableException,))
985 2406 aaronmk
986 2772 aaronmk
already_indexed = object() # tells add_indexes() the pkey has already been added
987
988
def add_indexes(db, table, has_pkey=True):
989
    '''Adds an index on all columns in a table.
990
    @param has_pkey bool|already_indexed Whether a pkey instead of a regular
991
        index should be added on the first column.
992
        * If already_indexed, the pkey is assumed to have already been added
993
    '''
994
    cols = table_cols(db, table)
995
    if has_pkey:
996
        if has_pkey is not already_indexed: add_pkey(db, table)
997
        cols = cols[1:]
998
    for col in cols: add_index(db, col, table)
999
1000 2086 aaronmk
def add_row_num(db, table):
1001 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
1002
    be the primary key.'''
1003 2320 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
1004 2096 aaronmk
    run_query(db, 'ALTER TABLE '+table+' ADD COLUMN '+row_num_col
1005 2443 aaronmk
        +' serial NOT NULL PRIMARY KEY', log_level=3)
1006 2086 aaronmk
1007 2889 aaronmk
def drop_table(db, table):
1008
    table = sql_gen.as_Table(table)
1009
    return run_query(db, 'DROP TABLE IF EXISTS '+table.to_str(db)+' CASCADE')
1010
1011 2760 aaronmk
def create_table(db, table, cols, has_pkey=True, col_indexes=True):
1012 2675 aaronmk
    '''Creates a table.
1013 2681 aaronmk
    @param cols [sql_gen.TypedCol,...] The column names and types
1014
    @param has_pkey If set, the first column becomes the primary key.
1015 2760 aaronmk
    @param col_indexes bool|[ref]
1016
        * If True, indexes will be added on all non-pkey columns.
1017
        * If a list reference, [0] will be set to a function to do this.
1018
          This can be used to delay index creation until the table is populated.
1019 2675 aaronmk
    '''
1020
    table = sql_gen.as_Table(table)
1021
1022 2681 aaronmk
    if has_pkey:
1023
        cols[0] = pkey = copy.copy(cols[0]) # don't modify input!
1024 2872 aaronmk
        pkey.constraints = 'PRIMARY KEY'
1025 2681 aaronmk
1026 2675 aaronmk
    str_ = 'CREATE TABLE '+table.to_str(db)+' (\n'
1027 2681 aaronmk
    str_ += '\n, '.join(v.to_str(db) for v in cols)
1028
    str_ += '\n);\n'
1029 2675 aaronmk
    run_query(db, str_, cacheable=True, log_level=2)
1030 2760 aaronmk
1031
    # Add indexes
1032 2773 aaronmk
    if has_pkey: has_pkey = already_indexed
1033
    def add_indexes_(): add_indexes(db, table, has_pkey)
1034
    if isinstance(col_indexes, list): col_indexes[0] = add_indexes_ # defer
1035
    elif col_indexes: add_indexes_() # add now
1036 2675 aaronmk
1037 2684 aaronmk
def vacuum(db, table):
1038
    table = sql_gen.as_Table(table)
1039
    db.with_autocommit(lambda: run_query(db, 'VACUUM ANALYZE '+table.to_str(db),
1040
        log_level=3))
1041
1042 2732 aaronmk
def truncate(db, table, schema='public'):
1043 2777 aaronmk
    table = sql_gen.as_Table(table, schema)
1044 2733 aaronmk
    return run_query(db, 'TRUNCATE '+table.to_str(db)+' CASCADE')
1045 2732 aaronmk
1046 2729 aaronmk
def tables(db, schema_like='public', table_like='%', exact=False):
1047
    if exact: compare = '='
1048
    else: compare = 'LIKE'
1049
1050 1849 aaronmk
    module = util.root_module(db.db)
1051 832 aaronmk
    if module == 'psycopg2':
1052 2779 aaronmk
        conds = [('schemaname', sql_gen.CompareCond(schema_like, compare)),
1053
            ('tablename', sql_gen.CompareCond(table_like, compare))]
1054
        return values(select(db, 'pg_tables', ['tablename'], conds,
1055
            order_by='tablename', log_level=4))
1056 1968 aaronmk
    elif module == 'MySQLdb':
1057 2779 aaronmk
        return values(run_query(db, 'SHOW TABLES LIKE '+db.esc_value(table_like)
1058
            , cacheable=True, log_level=4))
1059 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
1060 830 aaronmk
1061 2726 aaronmk
def table_exists(db, table):
1062
    table = sql_gen.as_Table(table)
1063 2730 aaronmk
    return list(tables(db, table.schema, table.name, exact=True)) != []
1064 2726 aaronmk
1065
def errors_table(db, table, if_exists=True):
1066
    '''
1067
    @param if_exists If set, returns None if the errors table doesn't exist
1068
    @return None|sql_gen.Table
1069
    '''
1070
    table = sql_gen.as_Table(table)
1071
    if table.srcs != (): table = table.srcs[0]
1072
1073
    errors_table = sql_gen.suffixed_table(table, '.errors')
1074
    if if_exists and not table_exists(db, errors_table): return None
1075
    return errors_table
1076
1077 833 aaronmk
##### Database management
1078
1079 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
1080
    '''For kw_args, see tables()'''
1081
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
1082 833 aaronmk
1083 832 aaronmk
##### Heuristic queries
1084
1085 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
1086 1554 aaronmk
    '''Recovers from errors.
1087 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
1088
    '''
1089 2642 aaronmk
    row = sql_gen.ColDict(db, table, row)
1090 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
1091
1092 471 aaronmk
    try:
1093 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
1094 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
1095
            row_ct_ref[0] += cur.rowcount
1096
        return value(cur)
1097 471 aaronmk
    except DuplicateKeyException, e:
1098 2653 aaronmk
        row = sql_gen.ColDict(db, table,
1099
            util.dict_subset_right_join(row, e.cols))
1100
        return value(select(db, table, [pkey_], row, recover=True))
1101 471 aaronmk
1102 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
1103 830 aaronmk
    '''Recovers from errors'''
1104 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
1105 14 aaronmk
    except StopIteration:
1106 40 aaronmk
        if not create: raise
1107 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
1108 2078 aaronmk
1109 2593 aaronmk
def is_func_result(col):
1110
    return col.table.name.find('(') >= 0 and col.name == 'result'
1111
1112 2592 aaronmk
def into_table_name(out_table, in_tables0, mapping, is_func):
1113 2615 aaronmk
    def in_col_str(in_col):
1114
        in_col = sql_gen.remove_col_rename(in_col)
1115
        if isinstance(in_col, sql_gen.Col):
1116
            table = in_col.table
1117
            if table == in_tables0:
1118
                in_col = sql_gen.to_name_only_col(in_col)
1119
            elif is_func_result(in_col): in_col = table # omit col name
1120
        return str(in_col)
1121
1122 2580 aaronmk
    str_ = str(out_table)
1123
    if is_func:
1124 2615 aaronmk
        str_ += '('
1125 2580 aaronmk
1126 2615 aaronmk
        try: value_in_col = mapping['value']
1127
        except KeyError:
1128 2654 aaronmk
            str_ += ', '.join((str(k)+'='+in_col_str(v)
1129 2615 aaronmk
                for k, v in mapping.iteritems()))
1130
        else: str_ += in_col_str(value_in_col)
1131
1132
        str_ += ')'
1133 2812 aaronmk
    else:
1134
        out_col = 'rank'
1135
        try: in_col = mapping[out_col]
1136
        except KeyError: str_ += '_pkeys'
1137
        else: # has a rank column, so hierarchical
1138
            str_ += '['+str(out_col)+'='+in_col_str(in_col)+']'
1139 2580 aaronmk
    return str_
1140
1141 2508 aaronmk
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None, into=None,
1142 2805 aaronmk
    default=None, is_func=False, on_error=exc.raise_):
1143 2078 aaronmk
    '''Recovers from errors.
1144
    Only works under PostgreSQL (uses INSERT RETURNING).
1145 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
1146
        tables to join with it using the main input table's pkey
1147 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
1148 2612 aaronmk
        * out_table_col: str (*not* sql_gen.Col)
1149
        * in_table_col: sql_gen.Col|literal-value
1150 2489 aaronmk
    @param into The table to contain the output and input pkeys.
1151 2574 aaronmk
        Defaults to `out_table.name+'_pkeys'`.
1152 2509 aaronmk
    @param default The *output* column to use as the pkey for missing rows.
1153
        If this output column does not exist in the mapping, uses None.
1154 2552 aaronmk
    @param is_func Whether out_table is the name of a SQL function, not a table
1155 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
1156 2078 aaronmk
    '''
1157 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
1158 2312 aaronmk
1159 2450 aaronmk
    def log_debug(msg): db.log_debug(msg, level=1.5)
1160 2505 aaronmk
    def col_ustr(str_):
1161 2567 aaronmk
        return strings.repr_no_u(sql_gen.remove_col_rename(str_))
1162 2450 aaronmk
1163 2768 aaronmk
    out_pkey = pkey(db, out_table, recover=True)
1164
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
1165
1166
    if mapping == {}: # need at least one column for INSERT SELECT
1167
        mapping = {out_pkey: None} # ColDict will replace with default value
1168
1169 2486 aaronmk
    log_debug('********** New iteration **********')
1170 2505 aaronmk
    log_debug('Inserting these input columns into '+strings.as_tt(
1171
        out_table.to_str(db))+':\n'+strings.as_table(mapping, ustr=col_ustr))
1172 2463 aaronmk
1173 2382 aaronmk
    # Create input joins from list of input tables
1174
    in_tables_ = in_tables[:] # don't modify input!
1175
    in_tables0 = in_tables_.pop(0) # first table is separate
1176 2735 aaronmk
    errors_table_ = errors_table(db, in_tables0)
1177 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
1178 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
1179 2460 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v,
1180
        {in_pkey: sql_gen.join_same_not_null}) for v in in_tables_]
1181 2131 aaronmk
1182 2592 aaronmk
    if into == None:
1183
        into = into_table_name(out_table, in_tables0, mapping, is_func)
1184
    into = sql_gen.as_Table(into)
1185
1186 2702 aaronmk
    # Set column sources
1187
    in_cols = filter(sql_gen.is_table_col, mapping.values())
1188
    for col in in_cols:
1189
        if col.table == in_tables0: col.set_srcs(sql_gen.src_self)
1190
1191 2486 aaronmk
    log_debug('Joining together input tables into temp table')
1192 2395 aaronmk
    # Place in new table for speed and so don't modify input if values edited
1193 2584 aaronmk
    in_table = sql_gen.Table('in')
1194 2702 aaronmk
    mapping = dicts.join(mapping, flatten(db, in_table, input_joins, in_cols,
1195
        preserve=[in_pkey_col], start=0))
1196 2395 aaronmk
    input_joins = [in_table]
1197 2486 aaronmk
    db.log_debug('Temp table: '+strings.as_tt(in_table.to_str(db)), level=2)
1198 2395 aaronmk
1199 2656 aaronmk
    mapping = sql_gen.ColDict(db, out_table, mapping)
1200
        # after applying dicts.join() because that returns a plain dict
1201
1202 2509 aaronmk
    # Resolve default value column
1203 2841 aaronmk
    if default != None:
1204
        try: default = mapping[default]
1205
        except KeyError:
1206 2509 aaronmk
            db.log_debug('Default value column '
1207
                +strings.as_tt(strings.repr_no_u(default))
1208 2511 aaronmk
                +' does not exist in mapping, falling back to None', level=2.1)
1209 2509 aaronmk
            default = None
1210
1211 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
1212 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
1213
1214 2201 aaronmk
    pkeys_table_exists_ref = [False]
1215 2420 aaronmk
    def insert_into_pkeys(joins, cols):
1216 2786 aaronmk
        query = mk_select(db, joins, cols, order_by=None, start=0)
1217 2201 aaronmk
        if pkeys_table_exists_ref[0]:
1218 2786 aaronmk
            insert_select(db, into, pkeys_names, query)
1219 2201 aaronmk
        else:
1220 2786 aaronmk
            run_query_into(db, query, into=into)
1221 2201 aaronmk
            pkeys_table_exists_ref[0] = True
1222
1223 2429 aaronmk
    limit_ref = [None]
1224 2380 aaronmk
    conds = set()
1225 2811 aaronmk
    distinct_on = sql_gen.ColDict(db, out_table)
1226 2325 aaronmk
    def mk_main_select(joins, cols):
1227 2811 aaronmk
        distinct_on_cols = [c.to_Col() for c in distinct_on.values()]
1228
        return mk_select(db, joins, cols, conds, distinct_on_cols,
1229 2429 aaronmk
            limit=limit_ref[0], start=0)
1230 2132 aaronmk
1231 2519 aaronmk
    exc_strs = set()
1232 2309 aaronmk
    def log_exc(e):
1233 2519 aaronmk
        e_str = exc.str_(e, first_line_only=True)
1234
        log_debug('Caught exception: '+e_str)
1235
        assert e_str not in exc_strs # avoid infinite loops
1236
        exc_strs.add(e_str)
1237 2735 aaronmk
1238 2451 aaronmk
    def remove_all_rows():
1239 2896 aaronmk
        log_debug('Ignoring all rows')
1240 2429 aaronmk
        limit_ref[0] = 0 # just create an empty pkeys table
1241 2735 aaronmk
1242 2758 aaronmk
    def ignore(in_col, value, e):
1243
        track_data_error(db, errors_table_, in_col.srcs, value, e.cause.pgcode,
1244
            e.cause.pgerror)
1245 2895 aaronmk
        log_debug('Ignoring rows with '+strings.as_tt(repr(in_col))+' = '
1246 2545 aaronmk
            +strings.as_tt(repr(value)))
1247 2895 aaronmk
1248 2758 aaronmk
    def remove_rows(in_col, value, e):
1249
        ignore(in_col, value, e)
1250 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
1251
        assert cond not in conds # avoid infinite loops
1252 2380 aaronmk
        conds.add(cond)
1253 2895 aaronmk
1254 2758 aaronmk
    def invalid2null(in_col, value, e):
1255
        ignore(in_col, value, e)
1256 2403 aaronmk
        update(db, in_table, [(in_col, None)],
1257
            sql_gen.ColValueCond(in_col, value))
1258 2245 aaronmk
1259 2589 aaronmk
    def insert_pkeys_table(which):
1260
        return sql_gen.Table(sql_gen.add_suffix(in_table.name,
1261
            '_insert_'+which+'_pkeys'))
1262
    insert_out_pkeys = insert_pkeys_table('out')
1263
    insert_in_pkeys = insert_pkeys_table('in')
1264
1265 2206 aaronmk
    # Do inserts and selects
1266 2642 aaronmk
    join_cols = sql_gen.ColDict(db, out_table)
1267 2206 aaronmk
    while True:
1268 2521 aaronmk
        if limit_ref[0] == 0: # special case
1269
            log_debug('Creating an empty pkeys table')
1270 2786 aaronmk
            cur = run_query_into(db, mk_select(db, out_table, [out_pkey],
1271 2521 aaronmk
                limit=limit_ref[0]), into=insert_out_pkeys)
1272
            break # don't do main case
1273
1274 2303 aaronmk
        has_joins = join_cols != {}
1275
1276 2903 aaronmk
        log_debug('Trying to insert new rows')
1277
1278 2305 aaronmk
        # Prepare to insert new rows
1279 2325 aaronmk
        insert_joins = input_joins[:] # don't modify original!
1280 2403 aaronmk
        insert_args = dict(recover=True, cacheable=False)
1281 2303 aaronmk
        if has_joins:
1282 2325 aaronmk
            insert_joins.append(sql_gen.Join(out_table, join_cols,
1283
                sql_gen.filter_out))
1284
        else:
1285 2404 aaronmk
            insert_args.update(dict(returning=out_pkey, into=insert_out_pkeys))
1286 2786 aaronmk
        main_select = mk_main_select(insert_joins, mapping.values())
1287 2303 aaronmk
1288 2206 aaronmk
        try:
1289 2518 aaronmk
            cur = insert_select(db, out_table, mapping.keys(), main_select,
1290
                **insert_args)
1291 2357 aaronmk
            break # insert successful
1292 2206 aaronmk
        except DuplicateKeyException, e:
1293 2309 aaronmk
            log_exc(e)
1294
1295 2258 aaronmk
            old_join_cols = join_cols.copy()
1296 2811 aaronmk
            distinct_on.update(util.dict_subset(mapping, e.cols))
1297 2565 aaronmk
            join_cols.update(util.dict_subset_right_join(mapping, e.cols))
1298 2486 aaronmk
            log_debug('Ignoring existing rows, comparing on these columns:\n'
1299 2505 aaronmk
                +strings.as_inline_table(join_cols, ustr=col_ustr))
1300 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
1301 2230 aaronmk
        except NullValueException, e:
1302 2309 aaronmk
            log_exc(e)
1303
1304 2230 aaronmk
            out_col, = e.cols
1305
            try: in_col = mapping[out_col]
1306 2356 aaronmk
            except KeyError:
1307 2486 aaronmk
                log_debug('Missing mapping for NOT NULL column '+out_col)
1308 2451 aaronmk
                remove_all_rows()
1309 2758 aaronmk
            else: remove_rows(in_col, None, e)
1310 2542 aaronmk
        except FunctionValueException, e:
1311
            log_exc(e)
1312
1313
            func_name = e.name
1314
            value = e.value
1315
            for out_col, in_col in mapping.iteritems():
1316 2758 aaronmk
                in_col = sql_gen.unwrap_func_call(in_col, func_name)
1317
                invalid2null(in_col, value, e)
1318 2525 aaronmk
        except MissingCastException, e:
1319
            log_exc(e)
1320
1321
            out_col = e.col
1322 2706 aaronmk
            type_ = e.type
1323
1324 2724 aaronmk
            log_debug('Casting '+strings.as_tt(out_col)+' input to '
1325
                +strings.as_tt(type_))
1326 2735 aaronmk
            def wrap_func(col): return cast(db, type_, col, errors_table_)
1327 2705 aaronmk
            mapping[out_col] = sql_gen.wrap(wrap_func, mapping[out_col])
1328 2429 aaronmk
        except DatabaseErrors, e:
1329
            log_exc(e)
1330
1331 2808 aaronmk
            log_debug('No handler for exception')
1332
            on_error(e)
1333 2451 aaronmk
            remove_all_rows()
1334 2358 aaronmk
        # after exception handled, rerun loop with additional constraints
1335 2132 aaronmk
1336 2357 aaronmk
    if row_ct_ref != None and cur.rowcount >= 0:
1337
        row_ct_ref[0] += cur.rowcount
1338
1339
    if has_joins:
1340
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
1341 2486 aaronmk
        log_debug('Getting output table pkeys of existing/inserted rows')
1342 2420 aaronmk
        insert_into_pkeys(select_joins, pkeys_cols)
1343 2357 aaronmk
    else:
1344 2404 aaronmk
        add_row_num(db, insert_out_pkeys) # for joining with input pkeys
1345 2357 aaronmk
1346 2486 aaronmk
        log_debug('Getting input table pkeys of inserted rows')
1347 2786 aaronmk
        run_query_into(db, mk_main_select(input_joins, [in_pkey]),
1348 2404 aaronmk
            into=insert_in_pkeys)
1349
        add_row_num(db, insert_in_pkeys) # for joining with output pkeys
1350 2357 aaronmk
1351 2428 aaronmk
        assert table_row_count(db, insert_out_pkeys) == table_row_count(db,
1352
            insert_in_pkeys)
1353
1354 2486 aaronmk
        log_debug('Combining output and input pkeys in inserted order')
1355 2404 aaronmk
        pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
1356 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
1357 2420 aaronmk
        insert_into_pkeys(pkey_joins, pkeys_names)
1358 2357 aaronmk
1359 2486 aaronmk
    db.log_debug('Adding pkey on pkeys table to enable fast joins', level=2.5)
1360 2594 aaronmk
    add_pkey(db, into)
1361 2407 aaronmk
1362 2508 aaronmk
    log_debug('Setting pkeys of missing rows to '+strings.as_tt(repr(default)))
1363 2489 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(into,
1364 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
1365
        # must use join_same_not_null or query will take forever
1366 2420 aaronmk
    insert_into_pkeys(missing_rows_joins,
1367 2508 aaronmk
        [in_pkey_col, sql_gen.NamedCol(out_pkey, default)])
1368 2357 aaronmk
1369 2489 aaronmk
    assert table_row_count(db, into) == table_row_count(db, in_table)
1370 2428 aaronmk
1371 2702 aaronmk
    srcs = []
1372
    if is_func: srcs = sql_gen.cols_srcs(in_cols)
1373
    return sql_gen.Col(out_pkey, into, srcs)
1374 2115 aaronmk
1375
##### Data cleanup
1376
1377 2290 aaronmk
def cleanup_table(db, table, cols):
1378 2677 aaronmk
    table = sql_gen.as_Table(table)
1379
    cols = map(sql_gen.as_Col, cols)
1380 2115 aaronmk
1381 2677 aaronmk
    expr = ('nullif(nullif(trim(both from %s), '+db.esc_value('')+'), '
1382
        +db.esc_value(r'\N')+')')
1383
    changes = [(v, sql_gen.CustomCode(expr % v.to_str(db)))
1384
        for v in cols]
1385 2115 aaronmk
1386 2677 aaronmk
    update(db, table, changes)