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 2676 aaronmk
            if self.serializable and not self.autocommit: self.run_query(
195 2678 aaronmk
                'SET TRANSACTION ISOLATION LEVEL SERIALIZABLE', log_level=4)
196 2101 aaronmk
            if schemas != None:
197
                schemas_ = ''.join((esc_name(self, s)+', '
198
                    for s in schemas.split(',')))
199 2792 aaronmk
                self.run_query("SELECT set_config('search_path', "
200
                    +self.esc_value(schemas_)
201
                    +" || current_setting('search_path'), false)", 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
            order_by='table_schema', limit=1, log_level=3))
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 2789 aaronmk
def run_query_into(db, query, into=None, add_indexes_=False, **kw_args):
467 2085 aaronmk
    '''Outputs a query to a temp table.
468
    For params, see run_query().
469
    '''
470 2789 aaronmk
    if into == None: return run_query(db, query, **kw_args)
471 2790 aaronmk
472
    assert isinstance(into, sql_gen.Table)
473
474
    kw_args['recover'] = True
475
    kw_args.setdefault('log_ignore_excs', (DuplicateTableException,))
476
477
    temp = not db.autocommit # tables are permanent in autocommit mode
478
    # "temporary tables cannot specify a schema name", so remove schema
479
    if temp: into.schema = None
480
481
    # Create table
482
    while True:
483
        create_query = 'CREATE'
484
        if temp: create_query += ' TEMP'
485
        create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
486 2385 aaronmk
487 2790 aaronmk
        try:
488
            cur = run_query(db, create_query, **kw_args)
489
                # CREATE TABLE AS sets rowcount to # rows in query
490
            break
491
        except DuplicateTableException, e:
492
            into.name = next_version(into.name)
493
            # try again with next version of name
494
495
    if add_indexes_: add_indexes(db, into)
496
497
    return cur
498 2085 aaronmk
499 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
500
501 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
502
503 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
504 2293 aaronmk
    start=None, order_by=order_by_pkey, default_table=None):
505 1981 aaronmk
    '''
506 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
507 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
508 1981 aaronmk
    @param fields Use None to select all fields in the table
509 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
510 2379 aaronmk
        * container can be any iterable type
511 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
512
        * compare_right_side: sql_gen.ValueCond|literal value
513 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
514
        use all columns
515 2786 aaronmk
    @return query
516 1981 aaronmk
    '''
517 2315 aaronmk
    # Parse tables param
518 2121 aaronmk
    if not lists.is_seq(tables): tables = [tables]
519 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
520 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
521 2121 aaronmk
522 2315 aaronmk
    # Parse other params
523 2376 aaronmk
    if conds == None: conds = []
524 2650 aaronmk
    elif dicts.is_dict(conds): conds = conds.items()
525 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
526 135 aaronmk
    assert limit == None or type(limit) == int
527 865 aaronmk
    assert start == None or type(start) == int
528 2315 aaronmk
    if order_by is order_by_pkey:
529
        if distinct_on != []: order_by = None
530
        else: order_by = pkey(db, table0, recover=True)
531 865 aaronmk
532 2315 aaronmk
    query = 'SELECT'
533 2056 aaronmk
534 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
535 2056 aaronmk
536 2200 aaronmk
    # DISTINCT ON columns
537 2233 aaronmk
    if distinct_on != []:
538 2467 aaronmk
        query += '\nDISTINCT'
539 2254 aaronmk
        if distinct_on is not distinct_on_all:
540 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
541
542
    # Columns
543 2467 aaronmk
    query += '\n'
544 1135 aaronmk
    if fields == None: query += '*'
545 2765 aaronmk
    else:
546
        assert fields != []
547
        query += '\n, '.join(map(parse_col, fields))
548 2200 aaronmk
549
    # Main table
550 2467 aaronmk
    query += '\nFROM '+table0.to_str(db)
551 865 aaronmk
552 2122 aaronmk
    # Add joins
553 2271 aaronmk
    left_table = table0
554 2263 aaronmk
    for join_ in tables:
555
        table = join_.table
556 2238 aaronmk
557 2343 aaronmk
        # Parse special values
558
        if join_.type_ is sql_gen.filter_out: # filter no match
559 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
560 2853 aaronmk
                sql_gen.CompareCond(None, '~=')))
561 2343 aaronmk
562 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
563 2122 aaronmk
564
        left_table = table
565
566 865 aaronmk
    missing = True
567 2376 aaronmk
    if conds != []:
568 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
569
        else: whitespace = '\n'
570 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
571
            .to_str(db) for l, r in conds], 'WHERE')
572 865 aaronmk
        missing = False
573 2227 aaronmk
    if order_by != None:
574 2467 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
575
    if limit != None: query += '\nLIMIT '+str(limit); missing = False
576 865 aaronmk
    if start != None:
577 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
578 865 aaronmk
        missing = False
579
    if missing: warnings.warn(DbWarning(
580
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
581
582 2786 aaronmk
    return query
583 11 aaronmk
584 2054 aaronmk
def select(db, *args, **kw_args):
585
    '''For params, see mk_select() and run_query()'''
586
    recover = kw_args.pop('recover', None)
587
    cacheable = kw_args.pop('cacheable', True)
588 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
589 2054 aaronmk
590 2791 aaronmk
    return run_query(db, mk_select(db, *args, **kw_args), recover, cacheable,
591
        log_level=log_level)
592 2054 aaronmk
593 2788 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, returning=None,
594
    embeddable=False):
595 1960 aaronmk
    '''
596
    @param returning str|None An inserted column (such as pkey) to return
597 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
598 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
599
        query will be fully cached, not just if it raises an exception.
600 1960 aaronmk
    '''
601 2754 aaronmk
    table = sql_gen.remove_table_rename(sql_gen.as_Table(table))
602 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
603 2646 aaronmk
    if cols != None:
604
        cols = [sql_gen.to_name_only_col(v, table).to_str(db) for v in cols]
605 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
606 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
607 2063 aaronmk
608
    # Build query
609 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
610
    query = first_line
611 2467 aaronmk
    if cols != None: query += '\n('+', '.join(cols)+')'
612
    query += '\n'+select_query
613 2063 aaronmk
614
    if returning != None:
615 2743 aaronmk
        query += '\nRETURNING '+sql_gen.to_name_only_col(returning).to_str(db)
616 2063 aaronmk
617 2070 aaronmk
    if embeddable:
618 2327 aaronmk
        assert returning != None
619
620 2070 aaronmk
        # Create function
621 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
622 2327 aaronmk
        return_type = 'SETOF '+returning.to_str(db)+'%TYPE'
623 2189 aaronmk
        while True:
624
            try:
625 2698 aaronmk
                function = sql_gen.TempFunction(function_name, db.autocommit)
626 2194 aaronmk
627 2189 aaronmk
                function_query = '''\
628 2698 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''()
629 2467 aaronmk
RETURNS '''+return_type+'''
630
LANGUAGE sql
631
AS $$
632 2788 aaronmk
'''+query+''';
633 2467 aaronmk
$$;
634 2070 aaronmk
'''
635 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
636 2464 aaronmk
                    log_ignore_excs=(DuplicateFunctionException,))
637 2189 aaronmk
                break # this version was successful
638
            except DuplicateFunctionException, e:
639
                function_name = next_version(function_name)
640
                # try again with next version of name
641 2070 aaronmk
642 2337 aaronmk
        # Return query that uses function
643 2698 aaronmk
        func_table = sql_gen.NamedTable('f', sql_gen.FunctionCall(function),
644 2742 aaronmk
            [returning]) # AS clause requires function alias
645 2787 aaronmk
        return mk_select(db, func_table, start=0, order_by=None)
646 2070 aaronmk
647 2787 aaronmk
    return query
648 2066 aaronmk
649
def insert_select(db, *args, **kw_args):
650 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
651 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
652
        values in
653 2072 aaronmk
    '''
654 2386 aaronmk
    into = kw_args.pop('into', None)
655
    if into != None: kw_args['embeddable'] = True
656 2066 aaronmk
    recover = kw_args.pop('recover', None)
657
    cacheable = kw_args.pop('cacheable', True)
658 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
659 2066 aaronmk
660 2789 aaronmk
    return run_query_into(db, mk_insert_select(db, *args, **kw_args), into,
661 2787 aaronmk
        recover=recover, cacheable=cacheable, log_level=log_level)
662 2063 aaronmk
663 2738 aaronmk
default = sql_gen.default # tells insert() to use the default value for a column
664 2066 aaronmk
665 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
666 2085 aaronmk
    '''For params, see insert_select()'''
667 1960 aaronmk
    if lists.is_seq(row): cols = None
668
    else:
669
        cols = row.keys()
670
        row = row.values()
671 2738 aaronmk
    row = list(row) # ensure that "== []" works
672 1960 aaronmk
673 2738 aaronmk
    if row == []: query = None
674
    else: query = sql_gen.Values(row).to_str(db)
675 1961 aaronmk
676 2788 aaronmk
    return insert_select(db, table, cols, query, *args, **kw_args)
677 11 aaronmk
678 2402 aaronmk
def mk_update(db, table, changes=None, cond=None):
679
    '''
680
    @param changes [(col, new_value),...]
681
        * container can be any iterable type
682
        * col: sql_gen.Code|str (for col name)
683
        * new_value: sql_gen.Code|literal value
684
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
685
    @return str query
686
    '''
687
    query = 'UPDATE '+sql_gen.as_Table(table).to_str(db)+'\nSET\n'
688 2405 aaronmk
    query += ',\n'.join((sql_gen.to_name_only_col(col, table).to_str(db)+' = '
689 2402 aaronmk
        +sql_gen.as_Value(new_value).to_str(db) for col, new_value in changes))
690 2467 aaronmk
    if cond != None: query += '\nWHERE\n'+cond.to_str(db)
691 2402 aaronmk
692
    return query
693
694
def update(db, *args, **kw_args):
695
    '''For params, see mk_update() and run_query()'''
696
    recover = kw_args.pop('recover', None)
697
698 2791 aaronmk
    return run_query(db, mk_update(db, *args, **kw_args), recover)
699 2402 aaronmk
700 135 aaronmk
def last_insert_id(db):
701 1849 aaronmk
    module = util.root_module(db.db)
702 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
703
    elif module == 'MySQLdb': return db.insert_id()
704
    else: return None
705 13 aaronmk
706 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
707 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
708 2415 aaronmk
    to names that will be distinct among the columns' tables.
709 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
710 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
711
    @param into The table for the new columns.
712 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
713
        columns will be included in the mapping even if they are not in cols.
714
        The tables of the provided Col objects will be changed to into, so make
715
        copies of them if you want to keep the original tables.
716
    @param as_items Whether to return a list of dict items instead of a dict
717 2383 aaronmk
    @return dict(orig_col=new_col, ...)
718
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
719 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
720
        * All mappings use the into table so its name can easily be
721 2383 aaronmk
          changed for all columns at once
722
    '''
723 2415 aaronmk
    cols = lists.uniqify(cols)
724
725 2394 aaronmk
    items = []
726 2389 aaronmk
    for col in preserve:
727 2390 aaronmk
        orig_col = copy.copy(col)
728 2392 aaronmk
        col.table = into
729 2394 aaronmk
        items.append((orig_col, col))
730
    preserve = set(preserve)
731
    for col in cols:
732 2716 aaronmk
        if col not in preserve:
733
            items.append((col, sql_gen.Col(str(col), into, col.srcs)))
734 2394 aaronmk
735
    if not as_items: items = dict(items)
736
    return items
737 2383 aaronmk
738 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
739 2391 aaronmk
    '''For params, see mk_flatten_mapping()
740
    @return See return value of mk_flatten_mapping()
741
    '''
742 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
743
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
744 2786 aaronmk
    run_query_into(db, mk_select(db, joins, cols, limit=limit, start=start),
745 2846 aaronmk
        into=into, add_indexes_=True)
746 2394 aaronmk
    return dict(items)
747 2391 aaronmk
748 2751 aaronmk
def mk_track_data_error(db, errors_table, cols, value, error_code, error):
749 2810 aaronmk
    assert cols != ()
750
751 2751 aaronmk
    cols = map(sql_gen.to_name_only_col, cols)
752
753
    columns_cols = ['column']
754
    columns = sql_gen.NamedValues('columns', columns_cols,
755
        [[c.name] for c in cols])
756
    values_cols = ['value', 'error_code', 'error']
757
    values = sql_gen.NamedValues('values', values_cols,
758
        [value, error_code, error])
759
760
    select_cols = columns_cols+values_cols
761
    name_only_cols = map(sql_gen.to_name_only_col, select_cols)
762 2755 aaronmk
    errors_table = sql_gen.NamedTable('errors', errors_table)
763 2751 aaronmk
    joins = [columns, sql_gen.Join(values, type_='CROSS'),
764
        sql_gen.Join(errors_table, dict(zip(name_only_cols, select_cols)),
765
        sql_gen.filter_out)]
766
767
    return mk_insert_select(db, errors_table, name_only_cols,
768 2787 aaronmk
        mk_select(db, joins, select_cols, order_by=None))
769 2751 aaronmk
770 2810 aaronmk
def track_data_error(db, errors_table, cols, *args, **kw_args):
771 2758 aaronmk
    '''
772
    @param errors_table If None, does nothing.
773
    '''
774 2810 aaronmk
    if errors_table == None or cols == (): return
775
    run_query(db, mk_track_data_error(db, errors_table, cols, *args, **kw_args),
776 2758 aaronmk
        cacheable=True, log_level=4)
777
778 2704 aaronmk
def cast(db, type_, col, errors_table=None):
779
    '''Casts an (unrenamed) column or value.
780 2769 aaronmk
    If errors_table set and col has srcs, saves errors in errors_table (using
781
    col's srcs attr as the source columns) and converts errors to warnings.
782 2820 aaronmk
    @param col str|sql_gen.Col|sql_gen.Literal
783 2704 aaronmk
    @param errors_table None|sql_gen.Table|str
784
    '''
785 2820 aaronmk
    col = sql_gen.as_Col(col)
786 2769 aaronmk
    save_errors = (errors_table != None and isinstance(col, sql_gen.Col)
787
        and col.srcs != ())
788
    if not save_errors: # can't save errors
789
        return sql_gen.CustomCode(col.to_str(db)+'::'+type_) # just cast
790 2704 aaronmk
791
    assert not isinstance(col, sql_gen.NamedCol)
792
793 2770 aaronmk
    errors_table = sql_gen.as_Table(errors_table)
794
    srcs = map(sql_gen.to_name_only_col, col.srcs)
795
    function_name = str(sql_gen.FunctionCall(type_, *srcs))
796 2704 aaronmk
    function = sql_gen.TempFunction(function_name, db.autocommit)
797
798 2718 aaronmk
    while True:
799
        # Create function definition
800
        query = '''\
801 2704 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''(value text)
802
RETURNS '''+type_+'''
803
LANGUAGE plpgsql
804
STRICT
805
AS $$
806
BEGIN
807
    /* The explicit cast to the return type is needed to make the cast happen
808
    inside the try block. (Implicit casts to the return type happen at the end
809
    of the function, outside any block.) */
810
    RETURN value::'''+type_+''';
811
EXCEPTION
812
    WHEN data_exception THEN
813
        -- Save error in errors table.
814 2751 aaronmk
        -- Insert the value and error for *each* source column.
815
'''+mk_track_data_error(db, errors_table, srcs,
816
    *map(sql_gen.CustomCode, ['value', 'SQLSTATE', 'SQLERRM']))+''';
817 2704 aaronmk
818
        RAISE WARNING '%', SQLERRM;
819
        RETURN NULL;
820
END;
821
$$;
822
'''
823 2718 aaronmk
824
        # Create function
825
        try:
826
            run_query(db, query, recover=True, cacheable=True,
827
                log_ignore_excs=(DuplicateFunctionException,))
828
            break # successful
829
        except DuplicateFunctionException:
830 2770 aaronmk
            function.name = next_version(function.name)
831
            # try again with next version of name
832 2704 aaronmk
833
    return sql_gen.FunctionCall(function, col)
834
835 2414 aaronmk
##### Database structure queries
836
837 2426 aaronmk
def table_row_count(db, table, recover=None):
838 2786 aaronmk
    return value(run_query(db, mk_select(db, table, [sql_gen.row_count],
839 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
840 2426 aaronmk
841 2414 aaronmk
def table_cols(db, table, recover=None):
842
    return list(col_names(select(db, table, limit=0, order_by=None,
843 2443 aaronmk
        recover=recover, log_level=4)))
844 2414 aaronmk
845 2291 aaronmk
def pkey(db, table, recover=None):
846 832 aaronmk
    '''Assumed to be first column in table'''
847 2339 aaronmk
    return table_cols(db, table, recover)[0]
848 832 aaronmk
849 2559 aaronmk
not_null_col = 'not_null_col'
850 2340 aaronmk
851
def table_not_null_col(db, table, recover=None):
852
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
853
    if not_null_col in table_cols(db, table, recover): return not_null_col
854
    else: return pkey(db, table, recover)
855
856 853 aaronmk
def index_cols(db, table, index):
857
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
858
    automatically created. When you don't know whether something is a UNIQUE
859
    constraint or a UNIQUE index, use this function.'''
860 1909 aaronmk
    module = util.root_module(db.db)
861
    if module == 'psycopg2':
862
        return list(values(run_query(db, '''\
863 853 aaronmk
SELECT attname
864 866 aaronmk
FROM
865
(
866
        SELECT attnum, attname
867
        FROM pg_index
868
        JOIN pg_class index ON index.oid = indexrelid
869
        JOIN pg_class table_ ON table_.oid = indrelid
870
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
871
        WHERE
872 2782 aaronmk
            table_.relname = '''+db.esc_value(table)+'''
873
            AND index.relname = '''+db.esc_value(index)+'''
874 866 aaronmk
    UNION
875
        SELECT attnum, attname
876
        FROM
877
        (
878
            SELECT
879
                indrelid
880
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
881
                    AS indkey
882
            FROM pg_index
883
            JOIN pg_class index ON index.oid = indexrelid
884
            JOIN pg_class table_ ON table_.oid = indrelid
885
            WHERE
886 2782 aaronmk
                table_.relname = '''+db.esc_value(table)+'''
887
                AND index.relname = '''+db.esc_value(index)+'''
888 866 aaronmk
        ) s
889
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
890
) s
891 853 aaronmk
ORDER BY attnum
892 2782 aaronmk
'''
893
            , cacheable=True, log_level=4)))
894 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
895
        ' database')
896 853 aaronmk
897 464 aaronmk
def constraint_cols(db, table, constraint):
898 1849 aaronmk
    module = util.root_module(db.db)
899 464 aaronmk
    if module == 'psycopg2':
900
        return list(values(run_query(db, '''\
901
SELECT attname
902
FROM pg_constraint
903
JOIN pg_class ON pg_class.oid = conrelid
904
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
905
WHERE
906 2783 aaronmk
    relname = '''+db.esc_value(table)+'''
907
    AND conname = '''+db.esc_value(constraint)+'''
908 464 aaronmk
ORDER BY attnum
909 2783 aaronmk
'''
910
            )))
911 464 aaronmk
    else: raise NotImplementedError("Can't list constraint columns for "+module+
912
        ' database')
913
914 2096 aaronmk
row_num_col = '_row_num'
915
916 2847 aaronmk
def add_index(db, exprs, table=None, unique=False, ensure_not_null=True):
917 2688 aaronmk
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
918 2538 aaronmk
    Currently, only function calls are supported as expressions.
919 2847 aaronmk
    @param ensure_not_null If set, translates NULL values to sentinel values.
920
        This allows indexes to be used for comparisons where NULLs are equal.
921 2538 aaronmk
    '''
922 2688 aaronmk
    if not lists.is_seq(exprs): exprs = [exprs]
923 2538 aaronmk
924 2688 aaronmk
    # Parse exprs
925
    old_exprs = exprs[:]
926
    exprs = []
927
    cols = []
928
    for i, expr in enumerate(old_exprs):
929
        expr = copy.deepcopy(expr) # don't modify input!
930 2823 aaronmk
        expr = sql_gen.as_Col(expr, table)
931 2688 aaronmk
932 2823 aaronmk
        # Handle nullable columns
933 2860 aaronmk
        if ensure_not_null:
934
            try: expr = sql_gen.ensure_not_null(db, expr)
935
            except KeyError: pass # unknown type, so just create plain index
936 2823 aaronmk
937 2688 aaronmk
        # Extract col
938
        if isinstance(expr, sql_gen.FunctionCall):
939
            col = expr.args[0]
940
            expr = sql_gen.Expr(expr)
941
        else: col = expr
942 2823 aaronmk
        assert isinstance(col, sql_gen.Col)
943 2688 aaronmk
944
        # Extract table
945
        if table == None:
946
            assert sql_gen.is_table_col(col)
947
            table = col.table
948
949
        col.table = None
950
951
        exprs.append(expr)
952
        cols.append(col)
953 2408 aaronmk
954 2688 aaronmk
    table = sql_gen.as_Table(table)
955
    index = sql_gen.Table(str(sql_gen.Col(','.join(map(str, cols)), table)))
956
957
    str_ = 'CREATE'
958
    if unique: str_ += ' UNIQUE'
959
    str_ += ' INDEX '+index.to_str(db)+' ON '+table.to_str(db)+' ('+(
960
        ', '.join((v.to_str(db) for v in exprs)))+')'
961
962
    try: run_query(db, str_, recover=True, cacheable=True, log_level=3)
963 2408 aaronmk
    except DuplicateTableException: pass # index already existed
964
965 2686 aaronmk
def add_pkey(db, table, cols=None, recover=None):
966
    '''Adds a primary key.
967
    @param cols [sql_gen.Col,...] The columns in the primary key.
968
        Defaults to the first column in the table.
969 2406 aaronmk
    @pre The table must not already have a primary key.
970
    '''
971
    table = sql_gen.as_Table(table)
972 2686 aaronmk
    if cols == None: cols = [pkey(db, table, recover)]
973
    col_strs = [sql_gen.to_name_only_col(v).to_str(db) for v in cols]
974 2406 aaronmk
975 2781 aaronmk
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD PRIMARY KEY ('
976
        +(', '.join(col_strs))+')', recover=True, cacheable=True, log_level=3,
977
        log_ignore_excs=(DuplicateTableException,))
978 2406 aaronmk
979 2772 aaronmk
already_indexed = object() # tells add_indexes() the pkey has already been added
980
981
def add_indexes(db, table, has_pkey=True):
982
    '''Adds an index on all columns in a table.
983
    @param has_pkey bool|already_indexed Whether a pkey instead of a regular
984
        index should be added on the first column.
985
        * If already_indexed, the pkey is assumed to have already been added
986
    '''
987
    cols = table_cols(db, table)
988
    if has_pkey:
989
        if has_pkey is not already_indexed: add_pkey(db, table)
990
        cols = cols[1:]
991
    for col in cols: add_index(db, col, table)
992
993 2086 aaronmk
def add_row_num(db, table):
994 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
995
    be the primary key.'''
996 2320 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
997 2096 aaronmk
    run_query(db, 'ALTER TABLE '+table+' ADD COLUMN '+row_num_col
998 2443 aaronmk
        +' serial NOT NULL PRIMARY KEY', log_level=3)
999 2086 aaronmk
1000 2760 aaronmk
def create_table(db, table, cols, has_pkey=True, col_indexes=True):
1001 2675 aaronmk
    '''Creates a table.
1002 2681 aaronmk
    @param cols [sql_gen.TypedCol,...] The column names and types
1003
    @param has_pkey If set, the first column becomes the primary key.
1004 2760 aaronmk
    @param col_indexes bool|[ref]
1005
        * If True, indexes will be added on all non-pkey columns.
1006
        * If a list reference, [0] will be set to a function to do this.
1007
          This can be used to delay index creation until the table is populated.
1008 2675 aaronmk
    '''
1009
    table = sql_gen.as_Table(table)
1010
1011 2681 aaronmk
    if has_pkey:
1012
        cols[0] = pkey = copy.copy(cols[0]) # don't modify input!
1013
        pkey.type += ' NOT NULL PRIMARY KEY'
1014
1015 2675 aaronmk
    str_ = 'CREATE TABLE '+table.to_str(db)+' (\n'
1016 2681 aaronmk
    str_ += '\n, '.join(v.to_str(db) for v in cols)
1017
    str_ += '\n);\n'
1018 2675 aaronmk
    run_query(db, str_, cacheable=True, log_level=2)
1019 2760 aaronmk
1020
    # Add indexes
1021 2773 aaronmk
    if has_pkey: has_pkey = already_indexed
1022
    def add_indexes_(): add_indexes(db, table, has_pkey)
1023
    if isinstance(col_indexes, list): col_indexes[0] = add_indexes_ # defer
1024
    elif col_indexes: add_indexes_() # add now
1025 2675 aaronmk
1026 2684 aaronmk
def vacuum(db, table):
1027
    table = sql_gen.as_Table(table)
1028
    db.with_autocommit(lambda: run_query(db, 'VACUUM ANALYZE '+table.to_str(db),
1029
        log_level=3))
1030
1031 2732 aaronmk
def truncate(db, table, schema='public'):
1032 2777 aaronmk
    table = sql_gen.as_Table(table, schema)
1033 2733 aaronmk
    return run_query(db, 'TRUNCATE '+table.to_str(db)+' CASCADE')
1034 2732 aaronmk
1035 2729 aaronmk
def tables(db, schema_like='public', table_like='%', exact=False):
1036
    if exact: compare = '='
1037
    else: compare = 'LIKE'
1038
1039 1849 aaronmk
    module = util.root_module(db.db)
1040 832 aaronmk
    if module == 'psycopg2':
1041 2779 aaronmk
        conds = [('schemaname', sql_gen.CompareCond(schema_like, compare)),
1042
            ('tablename', sql_gen.CompareCond(table_like, compare))]
1043
        return values(select(db, 'pg_tables', ['tablename'], conds,
1044
            order_by='tablename', log_level=4))
1045 1968 aaronmk
    elif module == 'MySQLdb':
1046 2779 aaronmk
        return values(run_query(db, 'SHOW TABLES LIKE '+db.esc_value(table_like)
1047
            , cacheable=True, log_level=4))
1048 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
1049 830 aaronmk
1050 2726 aaronmk
def table_exists(db, table):
1051
    table = sql_gen.as_Table(table)
1052 2730 aaronmk
    return list(tables(db, table.schema, table.name, exact=True)) != []
1053 2726 aaronmk
1054
def errors_table(db, table, if_exists=True):
1055
    '''
1056
    @param if_exists If set, returns None if the errors table doesn't exist
1057
    @return None|sql_gen.Table
1058
    '''
1059
    table = sql_gen.as_Table(table)
1060
    if table.srcs != (): table = table.srcs[0]
1061
1062
    errors_table = sql_gen.suffixed_table(table, '.errors')
1063
    if if_exists and not table_exists(db, errors_table): return None
1064
    return errors_table
1065
1066 833 aaronmk
##### Database management
1067
1068 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
1069
    '''For kw_args, see tables()'''
1070
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
1071 833 aaronmk
1072 832 aaronmk
##### Heuristic queries
1073
1074 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
1075 1554 aaronmk
    '''Recovers from errors.
1076 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
1077
    '''
1078 2642 aaronmk
    row = sql_gen.ColDict(db, table, row)
1079 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
1080
1081 471 aaronmk
    try:
1082 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
1083 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
1084
            row_ct_ref[0] += cur.rowcount
1085
        return value(cur)
1086 471 aaronmk
    except DuplicateKeyException, e:
1087 2653 aaronmk
        row = sql_gen.ColDict(db, table,
1088
            util.dict_subset_right_join(row, e.cols))
1089
        return value(select(db, table, [pkey_], row, recover=True))
1090 471 aaronmk
1091 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
1092 830 aaronmk
    '''Recovers from errors'''
1093 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
1094 14 aaronmk
    except StopIteration:
1095 40 aaronmk
        if not create: raise
1096 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
1097 2078 aaronmk
1098 2593 aaronmk
def is_func_result(col):
1099
    return col.table.name.find('(') >= 0 and col.name == 'result'
1100
1101 2592 aaronmk
def into_table_name(out_table, in_tables0, mapping, is_func):
1102 2615 aaronmk
    def in_col_str(in_col):
1103
        in_col = sql_gen.remove_col_rename(in_col)
1104
        if isinstance(in_col, sql_gen.Col):
1105
            table = in_col.table
1106
            if table == in_tables0:
1107
                in_col = sql_gen.to_name_only_col(in_col)
1108
            elif is_func_result(in_col): in_col = table # omit col name
1109
        return str(in_col)
1110
1111 2580 aaronmk
    str_ = str(out_table)
1112
    if is_func:
1113 2615 aaronmk
        str_ += '('
1114 2580 aaronmk
1115 2615 aaronmk
        try: value_in_col = mapping['value']
1116
        except KeyError:
1117 2654 aaronmk
            str_ += ', '.join((str(k)+'='+in_col_str(v)
1118 2615 aaronmk
                for k, v in mapping.iteritems()))
1119
        else: str_ += in_col_str(value_in_col)
1120
1121
        str_ += ')'
1122 2812 aaronmk
    else:
1123
        out_col = 'rank'
1124
        try: in_col = mapping[out_col]
1125
        except KeyError: str_ += '_pkeys'
1126
        else: # has a rank column, so hierarchical
1127
            str_ += '['+str(out_col)+'='+in_col_str(in_col)+']'
1128 2580 aaronmk
    return str_
1129
1130 2508 aaronmk
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None, into=None,
1131 2805 aaronmk
    default=None, is_func=False, on_error=exc.raise_):
1132 2078 aaronmk
    '''Recovers from errors.
1133
    Only works under PostgreSQL (uses INSERT RETURNING).
1134 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
1135
        tables to join with it using the main input table's pkey
1136 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
1137 2612 aaronmk
        * out_table_col: str (*not* sql_gen.Col)
1138
        * in_table_col: sql_gen.Col|literal-value
1139 2489 aaronmk
    @param into The table to contain the output and input pkeys.
1140 2574 aaronmk
        Defaults to `out_table.name+'_pkeys'`.
1141 2509 aaronmk
    @param default The *output* column to use as the pkey for missing rows.
1142
        If this output column does not exist in the mapping, uses None.
1143 2552 aaronmk
    @param is_func Whether out_table is the name of a SQL function, not a table
1144 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
1145 2078 aaronmk
    '''
1146 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
1147 2312 aaronmk
1148 2450 aaronmk
    def log_debug(msg): db.log_debug(msg, level=1.5)
1149 2505 aaronmk
    def col_ustr(str_):
1150 2567 aaronmk
        return strings.repr_no_u(sql_gen.remove_col_rename(str_))
1151 2450 aaronmk
1152 2768 aaronmk
    out_pkey = pkey(db, out_table, recover=True)
1153
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
1154
1155
    if mapping == {}: # need at least one column for INSERT SELECT
1156
        mapping = {out_pkey: None} # ColDict will replace with default value
1157
1158 2486 aaronmk
    log_debug('********** New iteration **********')
1159 2505 aaronmk
    log_debug('Inserting these input columns into '+strings.as_tt(
1160
        out_table.to_str(db))+':\n'+strings.as_table(mapping, ustr=col_ustr))
1161 2463 aaronmk
1162 2382 aaronmk
    # Create input joins from list of input tables
1163
    in_tables_ = in_tables[:] # don't modify input!
1164
    in_tables0 = in_tables_.pop(0) # first table is separate
1165 2735 aaronmk
    errors_table_ = errors_table(db, in_tables0)
1166 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
1167 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
1168 2460 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v,
1169
        {in_pkey: sql_gen.join_same_not_null}) for v in in_tables_]
1170 2131 aaronmk
1171 2592 aaronmk
    if into == None:
1172
        into = into_table_name(out_table, in_tables0, mapping, is_func)
1173
    into = sql_gen.as_Table(into)
1174
1175 2702 aaronmk
    # Set column sources
1176
    in_cols = filter(sql_gen.is_table_col, mapping.values())
1177
    for col in in_cols:
1178
        if col.table == in_tables0: col.set_srcs(sql_gen.src_self)
1179
1180 2486 aaronmk
    log_debug('Joining together input tables into temp table')
1181 2395 aaronmk
    # Place in new table for speed and so don't modify input if values edited
1182 2584 aaronmk
    in_table = sql_gen.Table('in')
1183 2702 aaronmk
    mapping = dicts.join(mapping, flatten(db, in_table, input_joins, in_cols,
1184
        preserve=[in_pkey_col], start=0))
1185 2395 aaronmk
    input_joins = [in_table]
1186 2486 aaronmk
    db.log_debug('Temp table: '+strings.as_tt(in_table.to_str(db)), level=2)
1187 2395 aaronmk
1188 2656 aaronmk
    mapping = sql_gen.ColDict(db, out_table, mapping)
1189
        # after applying dicts.join() because that returns a plain dict
1190
1191 2509 aaronmk
    # Resolve default value column
1192 2841 aaronmk
    if default != None:
1193
        try: default = mapping[default]
1194
        except KeyError:
1195 2509 aaronmk
            db.log_debug('Default value column '
1196
                +strings.as_tt(strings.repr_no_u(default))
1197 2511 aaronmk
                +' does not exist in mapping, falling back to None', level=2.1)
1198 2509 aaronmk
            default = None
1199
1200 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
1201 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
1202
1203 2201 aaronmk
    pkeys_table_exists_ref = [False]
1204 2420 aaronmk
    def insert_into_pkeys(joins, cols):
1205 2786 aaronmk
        query = mk_select(db, joins, cols, order_by=None, start=0)
1206 2201 aaronmk
        if pkeys_table_exists_ref[0]:
1207 2786 aaronmk
            insert_select(db, into, pkeys_names, query)
1208 2201 aaronmk
        else:
1209 2786 aaronmk
            run_query_into(db, query, into=into)
1210 2201 aaronmk
            pkeys_table_exists_ref[0] = True
1211
1212 2429 aaronmk
    limit_ref = [None]
1213 2380 aaronmk
    conds = set()
1214 2811 aaronmk
    distinct_on = sql_gen.ColDict(db, out_table)
1215 2325 aaronmk
    def mk_main_select(joins, cols):
1216 2811 aaronmk
        distinct_on_cols = [c.to_Col() for c in distinct_on.values()]
1217
        return mk_select(db, joins, cols, conds, distinct_on_cols,
1218 2429 aaronmk
            limit=limit_ref[0], start=0)
1219 2132 aaronmk
1220 2519 aaronmk
    exc_strs = set()
1221 2309 aaronmk
    def log_exc(e):
1222 2519 aaronmk
        e_str = exc.str_(e, first_line_only=True)
1223
        log_debug('Caught exception: '+e_str)
1224
        assert e_str not in exc_strs # avoid infinite loops
1225
        exc_strs.add(e_str)
1226 2735 aaronmk
1227 2451 aaronmk
    def remove_all_rows():
1228 2450 aaronmk
        log_debug('Returning NULL for all rows')
1229 2429 aaronmk
        limit_ref[0] = 0 # just create an empty pkeys table
1230 2735 aaronmk
1231 2758 aaronmk
    def ignore(in_col, value, e):
1232
        track_data_error(db, errors_table_, in_col.srcs, value, e.cause.pgcode,
1233
            e.cause.pgerror)
1234
1235 2545 aaronmk
        in_col_str = strings.as_tt(repr(in_col))
1236 2544 aaronmk
        db.log_debug('Adding index on '+in_col_str+' to enable fast filtering',
1237
            level=2.5)
1238 2537 aaronmk
        add_index(db, in_col)
1239 2735 aaronmk
1240 2545 aaronmk
        log_debug('Ignoring rows with '+in_col_str+' = '
1241
            +strings.as_tt(repr(value)))
1242 2758 aaronmk
    def remove_rows(in_col, value, e):
1243
        ignore(in_col, value, e)
1244 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
1245
        assert cond not in conds # avoid infinite loops
1246 2380 aaronmk
        conds.add(cond)
1247 2758 aaronmk
    def invalid2null(in_col, value, e):
1248
        ignore(in_col, value, e)
1249 2403 aaronmk
        update(db, in_table, [(in_col, None)],
1250
            sql_gen.ColValueCond(in_col, value))
1251 2245 aaronmk
1252 2589 aaronmk
    def insert_pkeys_table(which):
1253
        return sql_gen.Table(sql_gen.add_suffix(in_table.name,
1254
            '_insert_'+which+'_pkeys'))
1255
    insert_out_pkeys = insert_pkeys_table('out')
1256
    insert_in_pkeys = insert_pkeys_table('in')
1257
1258 2206 aaronmk
    # Do inserts and selects
1259 2642 aaronmk
    join_cols = sql_gen.ColDict(db, out_table)
1260 2206 aaronmk
    while True:
1261 2521 aaronmk
        if limit_ref[0] == 0: # special case
1262
            log_debug('Creating an empty pkeys table')
1263 2786 aaronmk
            cur = run_query_into(db, mk_select(db, out_table, [out_pkey],
1264 2521 aaronmk
                limit=limit_ref[0]), into=insert_out_pkeys)
1265
            break # don't do main case
1266
1267 2303 aaronmk
        has_joins = join_cols != {}
1268
1269 2305 aaronmk
        # Prepare to insert new rows
1270 2325 aaronmk
        insert_joins = input_joins[:] # don't modify original!
1271 2403 aaronmk
        insert_args = dict(recover=True, cacheable=False)
1272 2303 aaronmk
        if has_joins:
1273 2325 aaronmk
            insert_joins.append(sql_gen.Join(out_table, join_cols,
1274
                sql_gen.filter_out))
1275
        else:
1276 2404 aaronmk
            insert_args.update(dict(returning=out_pkey, into=insert_out_pkeys))
1277 2786 aaronmk
        main_select = mk_main_select(insert_joins, mapping.values())
1278 2303 aaronmk
1279 2486 aaronmk
        log_debug('Trying to insert new rows')
1280 2206 aaronmk
        try:
1281 2518 aaronmk
            cur = insert_select(db, out_table, mapping.keys(), main_select,
1282
                **insert_args)
1283 2357 aaronmk
            break # insert successful
1284 2206 aaronmk
        except DuplicateKeyException, e:
1285 2309 aaronmk
            log_exc(e)
1286
1287 2258 aaronmk
            old_join_cols = join_cols.copy()
1288 2811 aaronmk
            distinct_on.update(util.dict_subset(mapping, e.cols))
1289 2565 aaronmk
            join_cols.update(util.dict_subset_right_join(mapping, e.cols))
1290 2486 aaronmk
            log_debug('Ignoring existing rows, comparing on these columns:\n'
1291 2505 aaronmk
                +strings.as_inline_table(join_cols, ustr=col_ustr))
1292 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
1293 2230 aaronmk
        except NullValueException, e:
1294 2309 aaronmk
            log_exc(e)
1295
1296 2230 aaronmk
            out_col, = e.cols
1297
            try: in_col = mapping[out_col]
1298 2356 aaronmk
            except KeyError:
1299 2486 aaronmk
                log_debug('Missing mapping for NOT NULL column '+out_col)
1300 2451 aaronmk
                remove_all_rows()
1301 2758 aaronmk
            else: remove_rows(in_col, None, e)
1302 2542 aaronmk
        except FunctionValueException, e:
1303
            log_exc(e)
1304
1305
            func_name = e.name
1306
            value = e.value
1307
            for out_col, in_col in mapping.iteritems():
1308 2758 aaronmk
                in_col = sql_gen.unwrap_func_call(in_col, func_name)
1309
                invalid2null(in_col, value, e)
1310 2525 aaronmk
        except MissingCastException, e:
1311
            log_exc(e)
1312
1313
            out_col = e.col
1314 2706 aaronmk
            type_ = e.type
1315
1316 2724 aaronmk
            log_debug('Casting '+strings.as_tt(out_col)+' input to '
1317
                +strings.as_tt(type_))
1318 2735 aaronmk
            def wrap_func(col): return cast(db, type_, col, errors_table_)
1319 2705 aaronmk
            mapping[out_col] = sql_gen.wrap(wrap_func, mapping[out_col])
1320 2429 aaronmk
        except DatabaseErrors, e:
1321
            log_exc(e)
1322
1323 2808 aaronmk
            log_debug('No handler for exception')
1324
            on_error(e)
1325 2451 aaronmk
            remove_all_rows()
1326 2358 aaronmk
        # after exception handled, rerun loop with additional constraints
1327 2132 aaronmk
1328 2357 aaronmk
    if row_ct_ref != None and cur.rowcount >= 0:
1329
        row_ct_ref[0] += cur.rowcount
1330
1331
    if has_joins:
1332
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
1333 2486 aaronmk
        log_debug('Getting output table pkeys of existing/inserted rows')
1334 2420 aaronmk
        insert_into_pkeys(select_joins, pkeys_cols)
1335 2357 aaronmk
    else:
1336 2404 aaronmk
        add_row_num(db, insert_out_pkeys) # for joining with input pkeys
1337 2357 aaronmk
1338 2486 aaronmk
        log_debug('Getting input table pkeys of inserted rows')
1339 2786 aaronmk
        run_query_into(db, mk_main_select(input_joins, [in_pkey]),
1340 2404 aaronmk
            into=insert_in_pkeys)
1341
        add_row_num(db, insert_in_pkeys) # for joining with output pkeys
1342 2357 aaronmk
1343 2428 aaronmk
        assert table_row_count(db, insert_out_pkeys) == table_row_count(db,
1344
            insert_in_pkeys)
1345
1346 2486 aaronmk
        log_debug('Combining output and input pkeys in inserted order')
1347 2404 aaronmk
        pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
1348 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
1349 2420 aaronmk
        insert_into_pkeys(pkey_joins, pkeys_names)
1350 2357 aaronmk
1351 2486 aaronmk
    db.log_debug('Adding pkey on pkeys table to enable fast joins', level=2.5)
1352 2594 aaronmk
    add_pkey(db, into)
1353 2407 aaronmk
1354 2508 aaronmk
    log_debug('Setting pkeys of missing rows to '+strings.as_tt(repr(default)))
1355 2489 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(into,
1356 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
1357
        # must use join_same_not_null or query will take forever
1358 2420 aaronmk
    insert_into_pkeys(missing_rows_joins,
1359 2508 aaronmk
        [in_pkey_col, sql_gen.NamedCol(out_pkey, default)])
1360 2357 aaronmk
1361 2489 aaronmk
    assert table_row_count(db, into) == table_row_count(db, in_table)
1362 2428 aaronmk
1363 2702 aaronmk
    srcs = []
1364
    if is_func: srcs = sql_gen.cols_srcs(in_cols)
1365
    return sql_gen.Col(out_pkey, into, srcs)
1366 2115 aaronmk
1367
##### Data cleanup
1368
1369 2290 aaronmk
def cleanup_table(db, table, cols):
1370 2677 aaronmk
    table = sql_gen.as_Table(table)
1371
    cols = map(sql_gen.as_Col, cols)
1372 2115 aaronmk
1373 2677 aaronmk
    expr = ('nullif(nullif(trim(both from %s), '+db.esc_value('')+'), '
1374
        +db.esc_value(r'\N')+')')
1375
    changes = [(v, sql_gen.CustomCode(expr % v.to_str(db)))
1376
        for v in cols]
1377 2115 aaronmk
1378 2677 aaronmk
    update(db, table, changes)