Project

General

Profile

1 11 aaronmk
# Database access
2
3 1869 aaronmk
import copy
4 2127 aaronmk
import operator
5 11 aaronmk
import re
6 865 aaronmk
import warnings
7 11 aaronmk
8 300 aaronmk
import exc
9 1909 aaronmk
import dicts
10 1893 aaronmk
import iters
11 1960 aaronmk
import lists
12 1889 aaronmk
from Proxy import Proxy
13 1872 aaronmk
import rand
14 2217 aaronmk
import sql_gen
15 862 aaronmk
import strings
16 131 aaronmk
import util
17 11 aaronmk
18 832 aaronmk
##### Exceptions
19
20 2804 aaronmk
def get_cur_query(cur, input_query=None):
21 2168 aaronmk
    raw_query = None
22
    if hasattr(cur, 'query'): raw_query = cur.query
23
    elif hasattr(cur, '_last_executed'): raw_query = cur._last_executed
24 2170 aaronmk
25
    if raw_query != None: return raw_query
26 2804 aaronmk
    else: return '[input] '+strings.ustr(input_query)
27 14 aaronmk
28 2170 aaronmk
def _add_cursor_info(e, *args, **kw_args):
29
    '''For params, see get_cur_query()'''
30 2771 aaronmk
    exc.add_msg(e, 'query: '+strings.ustr(get_cur_query(*args, **kw_args)))
31 135 aaronmk
32 300 aaronmk
class DbException(exc.ExceptionWithCause):
33 14 aaronmk
    def __init__(self, msg, cause=None, cur=None):
34 2145 aaronmk
        exc.ExceptionWithCause.__init__(self, msg, cause, cause_newline=True)
35 14 aaronmk
        if cur != None: _add_cursor_info(self, cur)
36
37 2143 aaronmk
class ExceptionWithName(DbException):
38
    def __init__(self, name, cause=None):
39 2484 aaronmk
        DbException.__init__(self, 'for name: '+strings.as_tt(str(name)), cause)
40 2143 aaronmk
        self.name = name
41 360 aaronmk
42 2240 aaronmk
class ExceptionWithNameValue(DbException):
43
    def __init__(self, name, value, cause=None):
44 2484 aaronmk
        DbException.__init__(self, 'for name: '+strings.as_tt(str(name))
45
            +'; value: '+strings.as_tt(repr(value)), cause)
46 2240 aaronmk
        self.name = name
47
        self.value = value
48
49 2306 aaronmk
class ConstraintException(DbException):
50
    def __init__(self, name, cols, cause=None):
51 2484 aaronmk
        DbException.__init__(self, 'Violated '+strings.as_tt(name)
52
            +' constraint on columns: '+strings.as_tt(', '.join(cols)), cause)
53 2306 aaronmk
        self.name = name
54 468 aaronmk
        self.cols = cols
55 11 aaronmk
56 2523 aaronmk
class MissingCastException(DbException):
57
    def __init__(self, type_, col, cause=None):
58
        DbException.__init__(self, 'Missing cast to type '+strings.as_tt(type_)
59
            +' on column: '+strings.as_tt(col), cause)
60
        self.type = type_
61
        self.col = col
62
63 2143 aaronmk
class NameException(DbException): pass
64
65 2306 aaronmk
class DuplicateKeyException(ConstraintException): pass
66 13 aaronmk
67 2306 aaronmk
class NullValueException(ConstraintException): pass
68 13 aaronmk
69 2240 aaronmk
class FunctionValueException(ExceptionWithNameValue): pass
70 2239 aaronmk
71 2143 aaronmk
class DuplicateTableException(ExceptionWithName): pass
72
73 2188 aaronmk
class DuplicateFunctionException(ExceptionWithName): pass
74
75 89 aaronmk
class EmptyRowException(DbException): pass
76
77 865 aaronmk
##### Warnings
78
79
class DbWarning(UserWarning): pass
80
81 1930 aaronmk
##### Result retrieval
82
83
def col_names(cur): return (col[0] for col in cur.description)
84
85
def rows(cur): return iter(lambda: cur.fetchone(), None)
86
87
def consume_rows(cur):
88
    '''Used to fetch all rows so result will be cached'''
89
    iters.consume_iter(rows(cur))
90
91
def next_row(cur): return rows(cur).next()
92
93
def row(cur):
94
    row_ = next_row(cur)
95
    consume_rows(cur)
96
    return row_
97
98
def next_value(cur): return next_row(cur)[0]
99
100
def value(cur): return row(cur)[0]
101
102
def values(cur): return iters.func_iter(lambda: next_value(cur))
103
104
def value_or_none(cur):
105
    try: return value(cur)
106
    except StopIteration: return None
107
108 2762 aaronmk
##### Escaping
109 2101 aaronmk
110 2573 aaronmk
def esc_name_by_module(module, name):
111
    if module == 'psycopg2' or module == None: quote = '"'
112 2101 aaronmk
    elif module == 'MySQLdb': quote = '`'
113
    else: raise NotImplementedError("Can't escape name for "+module+' database')
114 2500 aaronmk
    return sql_gen.esc_name(name, quote)
115 2101 aaronmk
116
def esc_name_by_engine(engine, name, **kw_args):
117
    return esc_name_by_module(db_engines[engine][0], name, **kw_args)
118
119
def esc_name(db, name, **kw_args):
120
    return esc_name_by_module(util.root_module(db.db), name, **kw_args)
121
122
def qual_name(db, schema, table):
123
    def esc_name_(name): return esc_name(db, name)
124
    table = esc_name_(table)
125
    if schema != None: return esc_name_(schema)+'.'+table
126
    else: return table
127
128 1869 aaronmk
##### Database connections
129 1849 aaronmk
130 2097 aaronmk
db_config_names = ['engine', 'host', 'user', 'password', 'database', 'schemas']
131 1926 aaronmk
132 1869 aaronmk
db_engines = {
133
    'MySQL': ('MySQLdb', {'password': 'passwd', 'database': 'db'}),
134
    'PostgreSQL': ('psycopg2', {}),
135
}
136
137
DatabaseErrors_set = set([DbException])
138
DatabaseErrors = tuple(DatabaseErrors_set)
139
140
def _add_module(module):
141
    DatabaseErrors_set.add(module.DatabaseError)
142
    global DatabaseErrors
143
    DatabaseErrors = tuple(DatabaseErrors_set)
144
145
def db_config_str(db_config):
146
    return db_config['engine']+' database '+db_config['database']
147
148 2448 aaronmk
log_debug_none = lambda msg, level=2: None
149 1901 aaronmk
150 1849 aaronmk
class DbConn:
151 2190 aaronmk
    def __init__(self, db_config, serializable=True, autocommit=False,
152
        caching=True, log_debug=log_debug_none):
153 1869 aaronmk
        self.db_config = db_config
154
        self.serializable = serializable
155 2190 aaronmk
        self.autocommit = autocommit
156
        self.caching = caching
157 1901 aaronmk
        self.log_debug = log_debug
158 2193 aaronmk
        self.debug = log_debug != log_debug_none
159 1869 aaronmk
160
        self.__db = None
161 1889 aaronmk
        self.query_results = {}
162 2139 aaronmk
        self._savepoint = 0
163 2671 aaronmk
        self._notices_seen = set()
164 1869 aaronmk
165
    def __getattr__(self, name):
166
        if name == '__dict__': raise Exception('getting __dict__')
167
        if name == 'db': return self._db()
168
        else: raise AttributeError()
169
170
    def __getstate__(self):
171
        state = copy.copy(self.__dict__) # shallow copy
172 1915 aaronmk
        state['log_debug'] = None # don't pickle the debug callback
173 1869 aaronmk
        state['_DbConn__db'] = None # don't pickle the connection
174
        return state
175
176 2165 aaronmk
    def connected(self): return self.__db != None
177
178 1869 aaronmk
    def _db(self):
179
        if self.__db == None:
180
            # Process db_config
181
            db_config = self.db_config.copy() # don't modify input!
182 2097 aaronmk
            schemas = db_config.pop('schemas', None)
183 1869 aaronmk
            module_name, mappings = db_engines[db_config.pop('engine')]
184
            module = __import__(module_name)
185
            _add_module(module)
186
            for orig, new in mappings.iteritems():
187
                try: util.rename_key(db_config, orig, new)
188
                except KeyError: pass
189
190
            # Connect
191
            self.__db = module.connect(**db_config)
192
193
            # Configure connection
194 2894 aaronmk
            if self.serializable: self.run_query(
195 2678 aaronmk
                'SET TRANSACTION ISOLATION LEVEL SERIALIZABLE', log_level=4)
196 2101 aaronmk
            if schemas != None:
197 2893 aaronmk
                search_path = [self.esc_name(s) for s in schemas.split(',')]
198
                search_path.append(value(run_query(self, 'SHOW search_path',
199
                    log_level=4)))
200
                run_query(self, 'SET search_path TO '+(','.join(search_path)),
201
                    log_level=3)
202 1869 aaronmk
203
        return self.__db
204 1889 aaronmk
205 1891 aaronmk
    class DbCursor(Proxy):
206 1927 aaronmk
        def __init__(self, outer):
207 1891 aaronmk
            Proxy.__init__(self, outer.db.cursor())
208 2191 aaronmk
            self.outer = outer
209 1927 aaronmk
            self.query_results = outer.query_results
210 1894 aaronmk
            self.query_lookup = None
211 1891 aaronmk
            self.result = []
212 1889 aaronmk
213 2802 aaronmk
        def execute(self, query):
214 2764 aaronmk
            self._is_insert = query.startswith('INSERT')
215 2797 aaronmk
            self.query_lookup = query
216 2148 aaronmk
            try:
217 2191 aaronmk
                try:
218 2802 aaronmk
                    cur = self.inner.execute(query)
219 2191 aaronmk
                    self.outer.do_autocommit()
220 2802 aaronmk
                finally: self.query = get_cur_query(self.inner, query)
221 1904 aaronmk
            except Exception, e:
222 2802 aaronmk
                _add_cursor_info(e, self, query)
223 1904 aaronmk
                self.result = e # cache the exception as the result
224
                self._cache_result()
225
                raise
226 2800 aaronmk
            if self.rowcount == 0 and query.startswith('SELECT'): # empty SELECT
227
                consume_rows(self) # fetch all rows so result will be cached
228 2762 aaronmk
            return cur
229 1894 aaronmk
230 1891 aaronmk
        def fetchone(self):
231
            row = self.inner.fetchone()
232 1899 aaronmk
            if row != None: self.result.append(row)
233
            # otherwise, fetched all rows
234 1904 aaronmk
            else: self._cache_result()
235
            return row
236
237
        def _cache_result(self):
238 1906 aaronmk
            # For inserts, only cache exceptions since inserts are not
239
            # idempotent, but an invalid insert will always be invalid
240 1930 aaronmk
            if self.query_results != None and (not self._is_insert
241 1906 aaronmk
                or isinstance(self.result, Exception)):
242
243 1894 aaronmk
                assert self.query_lookup != None
244 1916 aaronmk
                self.query_results[self.query_lookup] = self.CacheCursor(
245
                    util.dict_subset(dicts.AttrsDictView(self),
246
                    ['query', 'result', 'rowcount', 'description']))
247 1906 aaronmk
248 1916 aaronmk
        class CacheCursor:
249
            def __init__(self, cached_result): self.__dict__ = cached_result
250
251 1927 aaronmk
            def execute(self, *args, **kw_args):
252 1916 aaronmk
                if isinstance(self.result, Exception): raise self.result
253
                # otherwise, result is a rows list
254
                self.iter = iter(self.result)
255
256
            def fetchone(self):
257
                try: return self.iter.next()
258
                except StopIteration: return None
259 1891 aaronmk
260 2212 aaronmk
    def esc_value(self, value):
261 2663 aaronmk
        try: str_ = self.mogrify('%s', [value])
262
        except NotImplementedError, e:
263
            module = util.root_module(self.db)
264
            if module == 'MySQLdb':
265
                import _mysql
266
                str_ = _mysql.escape_string(value)
267
            else: raise e
268 2374 aaronmk
        return strings.to_unicode(str_)
269 2212 aaronmk
270 2347 aaronmk
    def esc_name(self, name): return esc_name(self, name) # calls global func
271
272 2814 aaronmk
    def std_code(self, str_):
273
        '''Standardizes SQL code.
274
        * Ensures that string literals are prefixed by `E`
275
        '''
276
        if str_.startswith("'"): str_ = 'E'+str_
277
        return str_
278
279 2665 aaronmk
    def can_mogrify(self):
280 2663 aaronmk
        module = util.root_module(self.db)
281 2665 aaronmk
        return module == 'psycopg2'
282 2663 aaronmk
283 2665 aaronmk
    def mogrify(self, query, params=None):
284
        if self.can_mogrify(): return self.db.cursor().mogrify(query, params)
285
        else: raise NotImplementedError("Can't mogrify query")
286
287 2671 aaronmk
    def print_notices(self):
288 2725 aaronmk
        if hasattr(self.db, 'notices'):
289
            for msg in self.db.notices:
290
                if msg not in self._notices_seen:
291
                    self._notices_seen.add(msg)
292
                    self.log_debug(msg, level=2)
293 2671 aaronmk
294 2793 aaronmk
    def run_query(self, query, cacheable=False, log_level=2,
295 2464 aaronmk
        debug_msg_ref=None):
296 2445 aaronmk
        '''
297 2464 aaronmk
        @param log_ignore_excs The log_level will be increased by 2 if the query
298
            throws one of these exceptions.
299 2664 aaronmk
        @param debug_msg_ref If specified, the log message will be returned in
300
            this instead of being output. This allows you to filter log messages
301
            depending on the result of the query.
302 2445 aaronmk
        '''
303 2167 aaronmk
        assert query != None
304
305 2047 aaronmk
        if not self.caching: cacheable = False
306 1903 aaronmk
        used_cache = False
307 2664 aaronmk
308
        def log_msg(query):
309
            if used_cache: cache_status = 'cache hit'
310
            elif cacheable: cache_status = 'cache miss'
311
            else: cache_status = 'non-cacheable'
312
            return 'DB query: '+cache_status+':\n'+strings.as_code(query, 'SQL')
313
314 1903 aaronmk
        try:
315 1927 aaronmk
            # Get cursor
316
            if cacheable:
317
                try:
318 2797 aaronmk
                    cur = self.query_results[query]
319 1927 aaronmk
                    used_cache = True
320
                except KeyError: cur = self.DbCursor(self)
321
            else: cur = self.db.cursor()
322
323 2664 aaronmk
            # Log query
324
            if self.debug and debug_msg_ref == None: # log before running
325
                self.log_debug(log_msg(query), log_level)
326
327 1927 aaronmk
            # Run query
328 2793 aaronmk
            cur.execute(query)
329 1903 aaronmk
        finally:
330 2671 aaronmk
            self.print_notices()
331 2664 aaronmk
            if self.debug and debug_msg_ref != None: # return after running
332 2793 aaronmk
                debug_msg_ref[0] = log_msg(str(get_cur_query(cur, query)))
333 1903 aaronmk
334
        return cur
335 1914 aaronmk
336 2797 aaronmk
    def is_cached(self, query): return query in self.query_results
337 2139 aaronmk
338 2683 aaronmk
    def with_autocommit(self, func, autocommit=True):
339 2801 aaronmk
        import psycopg2.extensions
340
        if autocommit:
341
            isolation_level = psycopg2.extensions.ISOLATION_LEVEL_AUTOCOMMIT
342
        else: isolation_level = psycopg2.extensions.ISOLATION_LEVEL_SERIALIZABLE
343
344
        prev_isolation_level = self.db.isolation_level
345
        self.db.set_isolation_level(isolation_level)
346 2683 aaronmk
        try: return func()
347 2801 aaronmk
        finally: self.db.set_isolation_level(prev_isolation_level)
348 2683 aaronmk
349 2139 aaronmk
    def with_savepoint(self, func):
350 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
351 2443 aaronmk
        self.run_query('SAVEPOINT '+savepoint, log_level=4)
352 2139 aaronmk
        self._savepoint += 1
353 2683 aaronmk
        try:
354 2139 aaronmk
            try: return_val = func()
355
            finally:
356
                self._savepoint -= 1
357
                assert self._savepoint >= 0
358
        except:
359 2443 aaronmk
            self.run_query('ROLLBACK TO SAVEPOINT '+savepoint, log_level=4)
360 2139 aaronmk
            raise
361
        else:
362 2443 aaronmk
            self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
363 2191 aaronmk
            self.do_autocommit()
364 2139 aaronmk
            return return_val
365 2191 aaronmk
366
    def do_autocommit(self):
367
        '''Autocommits if outside savepoint'''
368
        assert self._savepoint >= 0
369
        if self.autocommit and self._savepoint == 0:
370 2668 aaronmk
            self.log_debug('Autocommitting')
371 2191 aaronmk
            self.db.commit()
372 2643 aaronmk
373 2819 aaronmk
    def col_info(self, col):
374 2643 aaronmk
        table = sql_gen.Table('columns', 'information_schema')
375 2821 aaronmk
        cols = ['data_type', 'column_default',
376
            cast(self, 'boolean', 'is_nullable')]
377 2643 aaronmk
378
        conds = [('table_name', col.table.name), ('column_name', col.name)]
379
        schema = col.table.schema
380
        if schema != None: conds.append(('table_schema', schema))
381
382 2819 aaronmk
        type_, default, nullable = row(select(self, table, cols, conds,
383 2874 aaronmk
            order_by='table_schema', limit=1, log_level=4))
384 2643 aaronmk
            # TODO: order_by search_path schema order
385 2819 aaronmk
        default = sql_gen.as_Code(default, self)
386
387
        return sql_gen.TypedCol(col.name, type_, default, nullable)
388 1849 aaronmk
389 1869 aaronmk
connect = DbConn
390
391 832 aaronmk
##### Recoverable querying
392 15 aaronmk
393 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
394 11 aaronmk
395 2791 aaronmk
def run_query(db, query, recover=None, cacheable=False, log_level=2,
396
    log_ignore_excs=None, **kw_args):
397 2794 aaronmk
    '''For params, see DbConn.run_query()'''
398 830 aaronmk
    if recover == None: recover = False
399 2464 aaronmk
    if log_ignore_excs == None: log_ignore_excs = ()
400
    log_ignore_excs = tuple(log_ignore_excs)
401 830 aaronmk
402 2666 aaronmk
    debug_msg_ref = None # usually, db.run_query() logs query before running it
403
    # But if filtering with log_ignore_excs, wait until after exception parsing
404
    if log_ignore_excs != () or not db.can_mogrify(): debug_msg_ref = [None]
405
406 2148 aaronmk
    try:
407 2464 aaronmk
        try:
408 2794 aaronmk
            def run(): return db.run_query(query, cacheable, log_level,
409 2793 aaronmk
                debug_msg_ref, **kw_args)
410 2796 aaronmk
            if recover and not db.is_cached(query):
411 2464 aaronmk
                return with_savepoint(db, run)
412
            else: return run() # don't need savepoint if cached
413
        except Exception, e:
414
            if not recover: raise # need savepoint to run index_cols()
415
            msg = exc.str_(e)
416
417
            match = re.search(r'duplicate key value violates unique constraint '
418 2493 aaronmk
                r'"((_?[^\W_]+)_.+?)"', msg)
419 2464 aaronmk
            if match:
420
                constraint, table = match.groups()
421
                try: cols = index_cols(db, table, constraint)
422
                except NotImplementedError: raise e
423
                else: raise DuplicateKeyException(constraint, cols, e)
424
425 2493 aaronmk
            match = re.search(r'null value in column "(.+?)" violates not-null'
426 2464 aaronmk
                r' constraint', msg)
427
            if match: raise NullValueException('NOT NULL', [match.group(1)], e)
428
429
            match = re.search(r'\b(?:invalid input (?:syntax|value)\b.*?'
430
                r'|date/time field value out of range): "(.+?)"\n'
431 2535 aaronmk
                r'(?:(?s).*?)\bfunction "(.+?)"', msg)
432 2464 aaronmk
            if match:
433
                value, name = match.groups()
434
                raise FunctionValueException(name, strings.to_unicode(value), e)
435
436 2526 aaronmk
            match = re.search(r'column "(.+?)" is of type (.+?) but expression '
437 2523 aaronmk
                r'is of type', msg)
438
            if match:
439
                col, type_ = match.groups()
440
                raise MissingCastException(type_, col, e)
441
442 2493 aaronmk
            match = re.search(r'relation "(.+?)" already exists', msg)
443 2464 aaronmk
            if match: raise DuplicateTableException(match.group(1), e)
444
445 2493 aaronmk
            match = re.search(r'function "(.+?)" already exists', msg)
446 2464 aaronmk
            if match: raise DuplicateFunctionException(match.group(1), e)
447
448
            raise # no specific exception raised
449
    except log_ignore_excs:
450
        log_level += 2
451
        raise
452
    finally:
453 2666 aaronmk
        if debug_msg_ref != None and debug_msg_ref[0] != None:
454
            db.log_debug(debug_msg_ref[0], log_level)
455 830 aaronmk
456 832 aaronmk
##### Basic queries
457
458 2153 aaronmk
def next_version(name):
459 2163 aaronmk
    version = 1 # first existing name was version 0
460 2586 aaronmk
    match = re.match(r'^(.*)#(\d+)$', name)
461 2153 aaronmk
    if match:
462 2586 aaronmk
        name, version = match.groups()
463
        version = int(version)+1
464 2588 aaronmk
    return sql_gen.add_suffix(name, '#'+str(version))
465 2153 aaronmk
466 2899 aaronmk
def lock_table(db, table, mode):
467
    table = sql_gen.as_Table(table)
468
    run_query(db, 'LOCK TABLE '+table.to_str(db)+' IN '+mode+' MODE')
469
470 2789 aaronmk
def run_query_into(db, query, into=None, add_indexes_=False, **kw_args):
471 2085 aaronmk
    '''Outputs a query to a temp table.
472
    For params, see run_query().
473
    '''
474 2789 aaronmk
    if into == None: return run_query(db, query, **kw_args)
475 2790 aaronmk
476
    assert isinstance(into, sql_gen.Table)
477
478
    kw_args['recover'] = True
479
    kw_args.setdefault('log_ignore_excs', (DuplicateTableException,))
480
481
    temp = not db.autocommit # tables are permanent in autocommit mode
482
    # "temporary tables cannot specify a schema name", so remove schema
483
    if temp: into.schema = None
484
485
    # Create table
486
    while True:
487
        create_query = 'CREATE'
488
        if temp: create_query += ' TEMP'
489
        create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
490 2385 aaronmk
491 2790 aaronmk
        try:
492
            cur = run_query(db, create_query, **kw_args)
493
                # CREATE TABLE AS sets rowcount to # rows in query
494
            break
495
        except DuplicateTableException, e:
496
            into.name = next_version(into.name)
497
            # try again with next version of name
498
499
    if add_indexes_: add_indexes(db, into)
500
501
    return cur
502 2085 aaronmk
503 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
504
505 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
506
507 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
508 2293 aaronmk
    start=None, order_by=order_by_pkey, default_table=None):
509 1981 aaronmk
    '''
510 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
511 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
512 1981 aaronmk
    @param fields Use None to select all fields in the table
513 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
514 2379 aaronmk
        * container can be any iterable type
515 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
516
        * compare_right_side: sql_gen.ValueCond|literal value
517 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
518
        use all columns
519 2786 aaronmk
    @return query
520 1981 aaronmk
    '''
521 2315 aaronmk
    # Parse tables param
522 2121 aaronmk
    if not lists.is_seq(tables): tables = [tables]
523 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
524 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
525 2121 aaronmk
526 2315 aaronmk
    # Parse other params
527 2376 aaronmk
    if conds == None: conds = []
528 2650 aaronmk
    elif dicts.is_dict(conds): conds = conds.items()
529 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
530 135 aaronmk
    assert limit == None or type(limit) == int
531 865 aaronmk
    assert start == None or type(start) == int
532 2315 aaronmk
    if order_by is order_by_pkey:
533
        if distinct_on != []: order_by = None
534
        else: order_by = pkey(db, table0, recover=True)
535 865 aaronmk
536 2315 aaronmk
    query = 'SELECT'
537 2056 aaronmk
538 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
539 2056 aaronmk
540 2200 aaronmk
    # DISTINCT ON columns
541 2233 aaronmk
    if distinct_on != []:
542 2467 aaronmk
        query += '\nDISTINCT'
543 2254 aaronmk
        if distinct_on is not distinct_on_all:
544 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
545
546
    # Columns
547 2467 aaronmk
    query += '\n'
548 1135 aaronmk
    if fields == None: query += '*'
549 2765 aaronmk
    else:
550
        assert fields != []
551
        query += '\n, '.join(map(parse_col, fields))
552 2200 aaronmk
553
    # Main table
554 2467 aaronmk
    query += '\nFROM '+table0.to_str(db)
555 865 aaronmk
556 2122 aaronmk
    # Add joins
557 2271 aaronmk
    left_table = table0
558 2263 aaronmk
    for join_ in tables:
559
        table = join_.table
560 2238 aaronmk
561 2343 aaronmk
        # Parse special values
562
        if join_.type_ is sql_gen.filter_out: # filter no match
563 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
564 2853 aaronmk
                sql_gen.CompareCond(None, '~=')))
565 2343 aaronmk
566 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
567 2122 aaronmk
568
        left_table = table
569
570 865 aaronmk
    missing = True
571 2376 aaronmk
    if conds != []:
572 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
573
        else: whitespace = '\n'
574 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
575
            .to_str(db) for l, r in conds], 'WHERE')
576 865 aaronmk
        missing = False
577 2227 aaronmk
    if order_by != None:
578 2467 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
579
    if limit != None: query += '\nLIMIT '+str(limit); missing = False
580 865 aaronmk
    if start != None:
581 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
582 865 aaronmk
        missing = False
583
    if missing: warnings.warn(DbWarning(
584
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
585
586 2786 aaronmk
    return query
587 11 aaronmk
588 2054 aaronmk
def select(db, *args, **kw_args):
589
    '''For params, see mk_select() and run_query()'''
590
    recover = kw_args.pop('recover', None)
591
    cacheable = kw_args.pop('cacheable', True)
592 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
593 2054 aaronmk
594 2791 aaronmk
    return run_query(db, mk_select(db, *args, **kw_args), recover, cacheable,
595
        log_level=log_level)
596 2054 aaronmk
597 2788 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, returning=None,
598
    embeddable=False):
599 1960 aaronmk
    '''
600
    @param returning str|None An inserted column (such as pkey) to return
601 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
602 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
603
        query will be fully cached, not just if it raises an exception.
604 1960 aaronmk
    '''
605 2754 aaronmk
    table = sql_gen.remove_table_rename(sql_gen.as_Table(table))
606 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
607 2646 aaronmk
    if cols != None:
608
        cols = [sql_gen.to_name_only_col(v, table).to_str(db) for v in cols]
609 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
610 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
611 2063 aaronmk
612
    # Build query
613 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
614
    query = first_line
615 2467 aaronmk
    if cols != None: query += '\n('+', '.join(cols)+')'
616
    query += '\n'+select_query
617 2063 aaronmk
618
    if returning != None:
619 2743 aaronmk
        query += '\nRETURNING '+sql_gen.to_name_only_col(returning).to_str(db)
620 2063 aaronmk
621 2070 aaronmk
    if embeddable:
622 2327 aaronmk
        assert returning != None
623
624 2070 aaronmk
        # Create function
625 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
626 2327 aaronmk
        return_type = 'SETOF '+returning.to_str(db)+'%TYPE'
627 2189 aaronmk
        while True:
628
            try:
629 2698 aaronmk
                function = sql_gen.TempFunction(function_name, db.autocommit)
630 2194 aaronmk
631 2189 aaronmk
                function_query = '''\
632 2698 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''()
633 2467 aaronmk
RETURNS '''+return_type+'''
634
LANGUAGE sql
635
AS $$
636 2788 aaronmk
'''+query+''';
637 2467 aaronmk
$$;
638 2070 aaronmk
'''
639 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
640 2464 aaronmk
                    log_ignore_excs=(DuplicateFunctionException,))
641 2189 aaronmk
                break # this version was successful
642
            except DuplicateFunctionException, e:
643
                function_name = next_version(function_name)
644
                # try again with next version of name
645 2070 aaronmk
646 2337 aaronmk
        # Return query that uses function
647 2698 aaronmk
        func_table = sql_gen.NamedTable('f', sql_gen.FunctionCall(function),
648 2742 aaronmk
            [returning]) # AS clause requires function alias
649 2787 aaronmk
        return mk_select(db, func_table, start=0, order_by=None)
650 2070 aaronmk
651 2787 aaronmk
    return query
652 2066 aaronmk
653
def insert_select(db, *args, **kw_args):
654 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
655 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
656
        values in
657 2072 aaronmk
    '''
658 2386 aaronmk
    into = kw_args.pop('into', None)
659
    if into != None: kw_args['embeddable'] = True
660 2066 aaronmk
    recover = kw_args.pop('recover', None)
661
    cacheable = kw_args.pop('cacheable', True)
662 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
663 2066 aaronmk
664 2789 aaronmk
    return run_query_into(db, mk_insert_select(db, *args, **kw_args), into,
665 2787 aaronmk
        recover=recover, cacheable=cacheable, log_level=log_level)
666 2063 aaronmk
667 2738 aaronmk
default = sql_gen.default # tells insert() to use the default value for a column
668 2066 aaronmk
669 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
670 2085 aaronmk
    '''For params, see insert_select()'''
671 1960 aaronmk
    if lists.is_seq(row): cols = None
672
    else:
673
        cols = row.keys()
674
        row = row.values()
675 2738 aaronmk
    row = list(row) # ensure that "== []" works
676 1960 aaronmk
677 2738 aaronmk
    if row == []: query = None
678
    else: query = sql_gen.Values(row).to_str(db)
679 1961 aaronmk
680 2788 aaronmk
    return insert_select(db, table, cols, query, *args, **kw_args)
681 11 aaronmk
682 2402 aaronmk
def mk_update(db, table, changes=None, cond=None):
683
    '''
684
    @param changes [(col, new_value),...]
685
        * container can be any iterable type
686
        * col: sql_gen.Code|str (for col name)
687
        * new_value: sql_gen.Code|literal value
688
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
689
    @return str query
690
    '''
691
    query = 'UPDATE '+sql_gen.as_Table(table).to_str(db)+'\nSET\n'
692 2405 aaronmk
    query += ',\n'.join((sql_gen.to_name_only_col(col, table).to_str(db)+' = '
693 2402 aaronmk
        +sql_gen.as_Value(new_value).to_str(db) for col, new_value in changes))
694 2467 aaronmk
    if cond != None: query += '\nWHERE\n'+cond.to_str(db)
695 2402 aaronmk
696
    return query
697
698
def update(db, *args, **kw_args):
699
    '''For params, see mk_update() and run_query()'''
700
    recover = kw_args.pop('recover', None)
701
702 2791 aaronmk
    return run_query(db, mk_update(db, *args, **kw_args), recover)
703 2402 aaronmk
704 135 aaronmk
def last_insert_id(db):
705 1849 aaronmk
    module = util.root_module(db.db)
706 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
707
    elif module == 'MySQLdb': return db.insert_id()
708
    else: return None
709 13 aaronmk
710 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
711 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
712 2415 aaronmk
    to names that will be distinct among the columns' tables.
713 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
714 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
715
    @param into The table for the new columns.
716 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
717
        columns will be included in the mapping even if they are not in cols.
718
        The tables of the provided Col objects will be changed to into, so make
719
        copies of them if you want to keep the original tables.
720
    @param as_items Whether to return a list of dict items instead of a dict
721 2383 aaronmk
    @return dict(orig_col=new_col, ...)
722
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
723 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
724
        * All mappings use the into table so its name can easily be
725 2383 aaronmk
          changed for all columns at once
726
    '''
727 2415 aaronmk
    cols = lists.uniqify(cols)
728
729 2394 aaronmk
    items = []
730 2389 aaronmk
    for col in preserve:
731 2390 aaronmk
        orig_col = copy.copy(col)
732 2392 aaronmk
        col.table = into
733 2394 aaronmk
        items.append((orig_col, col))
734
    preserve = set(preserve)
735
    for col in cols:
736 2716 aaronmk
        if col not in preserve:
737
            items.append((col, sql_gen.Col(str(col), into, col.srcs)))
738 2394 aaronmk
739
    if not as_items: items = dict(items)
740
    return items
741 2383 aaronmk
742 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
743 2391 aaronmk
    '''For params, see mk_flatten_mapping()
744
    @return See return value of mk_flatten_mapping()
745
    '''
746 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
747
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
748 2786 aaronmk
    run_query_into(db, mk_select(db, joins, cols, limit=limit, start=start),
749 2846 aaronmk
        into=into, add_indexes_=True)
750 2394 aaronmk
    return dict(items)
751 2391 aaronmk
752 2751 aaronmk
def mk_track_data_error(db, errors_table, cols, value, error_code, error):
753 2810 aaronmk
    assert cols != ()
754
755 2751 aaronmk
    cols = map(sql_gen.to_name_only_col, cols)
756
757
    columns_cols = ['column']
758
    columns = sql_gen.NamedValues('columns', columns_cols,
759
        [[c.name] for c in cols])
760
    values_cols = ['value', 'error_code', 'error']
761
    values = sql_gen.NamedValues('values', values_cols,
762
        [value, error_code, error])
763
764
    select_cols = columns_cols+values_cols
765
    name_only_cols = map(sql_gen.to_name_only_col, select_cols)
766 2755 aaronmk
    errors_table = sql_gen.NamedTable('errors', errors_table)
767 2751 aaronmk
    joins = [columns, sql_gen.Join(values, type_='CROSS'),
768
        sql_gen.Join(errors_table, dict(zip(name_only_cols, select_cols)),
769
        sql_gen.filter_out)]
770
771
    return mk_insert_select(db, errors_table, name_only_cols,
772 2787 aaronmk
        mk_select(db, joins, select_cols, order_by=None))
773 2751 aaronmk
774 2810 aaronmk
def track_data_error(db, errors_table, cols, *args, **kw_args):
775 2758 aaronmk
    '''
776
    @param errors_table If None, does nothing.
777
    '''
778 2810 aaronmk
    if errors_table == None or cols == (): return
779
    run_query(db, mk_track_data_error(db, errors_table, cols, *args, **kw_args),
780 2758 aaronmk
        cacheable=True, log_level=4)
781
782 2704 aaronmk
def cast(db, type_, col, errors_table=None):
783
    '''Casts an (unrenamed) column or value.
784 2769 aaronmk
    If errors_table set and col has srcs, saves errors in errors_table (using
785
    col's srcs attr as the source columns) and converts errors to warnings.
786 2820 aaronmk
    @param col str|sql_gen.Col|sql_gen.Literal
787 2704 aaronmk
    @param errors_table None|sql_gen.Table|str
788
    '''
789 2820 aaronmk
    col = sql_gen.as_Col(col)
790 2769 aaronmk
    save_errors = (errors_table != None and isinstance(col, sql_gen.Col)
791
        and col.srcs != ())
792
    if not save_errors: # can't save errors
793
        return sql_gen.CustomCode(col.to_str(db)+'::'+type_) # just cast
794 2704 aaronmk
795
    assert not isinstance(col, sql_gen.NamedCol)
796
797 2770 aaronmk
    errors_table = sql_gen.as_Table(errors_table)
798
    srcs = map(sql_gen.to_name_only_col, col.srcs)
799
    function_name = str(sql_gen.FunctionCall(type_, *srcs))
800 2704 aaronmk
    function = sql_gen.TempFunction(function_name, db.autocommit)
801
802 2718 aaronmk
    while True:
803
        # Create function definition
804
        query = '''\
805 2704 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''(value text)
806
RETURNS '''+type_+'''
807
LANGUAGE plpgsql
808
STRICT
809
AS $$
810
BEGIN
811
    /* The explicit cast to the return type is needed to make the cast happen
812
    inside the try block. (Implicit casts to the return type happen at the end
813
    of the function, outside any block.) */
814
    RETURN value::'''+type_+''';
815
EXCEPTION
816
    WHEN data_exception THEN
817
        -- Save error in errors table.
818 2751 aaronmk
        -- Insert the value and error for *each* source column.
819
'''+mk_track_data_error(db, errors_table, srcs,
820
    *map(sql_gen.CustomCode, ['value', 'SQLSTATE', 'SQLERRM']))+''';
821 2704 aaronmk
822
        RAISE WARNING '%', SQLERRM;
823
        RETURN NULL;
824
END;
825
$$;
826
'''
827 2718 aaronmk
828
        # Create function
829
        try:
830
            run_query(db, query, recover=True, cacheable=True,
831
                log_ignore_excs=(DuplicateFunctionException,))
832
            break # successful
833
        except DuplicateFunctionException:
834 2770 aaronmk
            function.name = next_version(function.name)
835
            # try again with next version of name
836 2704 aaronmk
837
    return sql_gen.FunctionCall(function, col)
838
839 2414 aaronmk
##### Database structure queries
840
841 2426 aaronmk
def table_row_count(db, table, recover=None):
842 2786 aaronmk
    return value(run_query(db, mk_select(db, table, [sql_gen.row_count],
843 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
844 2426 aaronmk
845 2414 aaronmk
def table_cols(db, table, recover=None):
846
    return list(col_names(select(db, table, limit=0, order_by=None,
847 2443 aaronmk
        recover=recover, log_level=4)))
848 2414 aaronmk
849 2291 aaronmk
def pkey(db, table, recover=None):
850 832 aaronmk
    '''Assumed to be first column in table'''
851 2339 aaronmk
    return table_cols(db, table, recover)[0]
852 832 aaronmk
853 2559 aaronmk
not_null_col = 'not_null_col'
854 2340 aaronmk
855
def table_not_null_col(db, table, recover=None):
856
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
857
    if not_null_col in table_cols(db, table, recover): return not_null_col
858
    else: return pkey(db, table, recover)
859
860 853 aaronmk
def index_cols(db, table, index):
861
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
862
    automatically created. When you don't know whether something is a UNIQUE
863
    constraint or a UNIQUE index, use this function.'''
864 1909 aaronmk
    module = util.root_module(db.db)
865
    if module == 'psycopg2':
866
        return list(values(run_query(db, '''\
867 853 aaronmk
SELECT attname
868 866 aaronmk
FROM
869
(
870
        SELECT attnum, attname
871
        FROM pg_index
872
        JOIN pg_class index ON index.oid = indexrelid
873
        JOIN pg_class table_ ON table_.oid = indrelid
874
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
875
        WHERE
876 2782 aaronmk
            table_.relname = '''+db.esc_value(table)+'''
877
            AND index.relname = '''+db.esc_value(index)+'''
878 866 aaronmk
    UNION
879
        SELECT attnum, attname
880
        FROM
881
        (
882
            SELECT
883
                indrelid
884
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
885
                    AS indkey
886
            FROM pg_index
887
            JOIN pg_class index ON index.oid = indexrelid
888
            JOIN pg_class table_ ON table_.oid = indrelid
889
            WHERE
890 2782 aaronmk
                table_.relname = '''+db.esc_value(table)+'''
891
                AND index.relname = '''+db.esc_value(index)+'''
892 866 aaronmk
        ) s
893
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
894
) s
895 853 aaronmk
ORDER BY attnum
896 2782 aaronmk
'''
897
            , cacheable=True, log_level=4)))
898 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
899
        ' database')
900 853 aaronmk
901 464 aaronmk
def constraint_cols(db, table, constraint):
902 1849 aaronmk
    module = util.root_module(db.db)
903 464 aaronmk
    if module == 'psycopg2':
904
        return list(values(run_query(db, '''\
905
SELECT attname
906
FROM pg_constraint
907
JOIN pg_class ON pg_class.oid = conrelid
908
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
909
WHERE
910 2783 aaronmk
    relname = '''+db.esc_value(table)+'''
911
    AND conname = '''+db.esc_value(constraint)+'''
912 464 aaronmk
ORDER BY attnum
913 2783 aaronmk
'''
914
            )))
915 464 aaronmk
    else: raise NotImplementedError("Can't list constraint columns for "+module+
916
        ' database')
917
918 2096 aaronmk
row_num_col = '_row_num'
919
920 2847 aaronmk
def add_index(db, exprs, table=None, unique=False, ensure_not_null=True):
921 2688 aaronmk
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
922 2538 aaronmk
    Currently, only function calls are supported as expressions.
923 2847 aaronmk
    @param ensure_not_null If set, translates NULL values to sentinel values.
924
        This allows indexes to be used for comparisons where NULLs are equal.
925 2538 aaronmk
    '''
926 2688 aaronmk
    if not lists.is_seq(exprs): exprs = [exprs]
927 2538 aaronmk
928 2688 aaronmk
    # Parse exprs
929
    old_exprs = exprs[:]
930
    exprs = []
931
    cols = []
932
    for i, expr in enumerate(old_exprs):
933
        expr = copy.deepcopy(expr) # don't modify input!
934 2823 aaronmk
        expr = sql_gen.as_Col(expr, table)
935 2688 aaronmk
936 2823 aaronmk
        # Handle nullable columns
937 2860 aaronmk
        if ensure_not_null:
938
            try: expr = sql_gen.ensure_not_null(db, expr)
939
            except KeyError: pass # unknown type, so just create plain index
940 2823 aaronmk
941 2688 aaronmk
        # Extract col
942
        if isinstance(expr, sql_gen.FunctionCall):
943
            col = expr.args[0]
944
            expr = sql_gen.Expr(expr)
945
        else: col = expr
946 2823 aaronmk
        assert isinstance(col, sql_gen.Col)
947 2688 aaronmk
948
        # Extract table
949
        if table == None:
950
            assert sql_gen.is_table_col(col)
951
            table = col.table
952
953
        col.table = None
954
955
        exprs.append(expr)
956
        cols.append(col)
957 2408 aaronmk
958 2688 aaronmk
    table = sql_gen.as_Table(table)
959
    index = sql_gen.Table(str(sql_gen.Col(','.join(map(str, cols)), table)))
960
961
    str_ = 'CREATE'
962
    if unique: str_ += ' UNIQUE'
963
    str_ += ' INDEX '+index.to_str(db)+' ON '+table.to_str(db)+' ('+(
964
        ', '.join((v.to_str(db) for v in exprs)))+')'
965
966
    try: run_query(db, str_, recover=True, cacheable=True, log_level=3)
967 2408 aaronmk
    except DuplicateTableException: pass # index already existed
968
969 2686 aaronmk
def add_pkey(db, table, cols=None, recover=None):
970
    '''Adds a primary key.
971
    @param cols [sql_gen.Col,...] The columns in the primary key.
972
        Defaults to the first column in the table.
973 2406 aaronmk
    @pre The table must not already have a primary key.
974
    '''
975
    table = sql_gen.as_Table(table)
976 2686 aaronmk
    if cols == None: cols = [pkey(db, table, recover)]
977
    col_strs = [sql_gen.to_name_only_col(v).to_str(db) for v in cols]
978 2406 aaronmk
979 2781 aaronmk
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD PRIMARY KEY ('
980
        +(', '.join(col_strs))+')', recover=True, cacheable=True, log_level=3,
981
        log_ignore_excs=(DuplicateTableException,))
982 2406 aaronmk
983 2772 aaronmk
already_indexed = object() # tells add_indexes() the pkey has already been added
984
985
def add_indexes(db, table, has_pkey=True):
986
    '''Adds an index on all columns in a table.
987
    @param has_pkey bool|already_indexed Whether a pkey instead of a regular
988
        index should be added on the first column.
989
        * If already_indexed, the pkey is assumed to have already been added
990
    '''
991
    cols = table_cols(db, table)
992
    if has_pkey:
993
        if has_pkey is not already_indexed: add_pkey(db, table)
994
        cols = cols[1:]
995
    for col in cols: add_index(db, col, table)
996
997 2086 aaronmk
def add_row_num(db, table):
998 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
999
    be the primary key.'''
1000 2320 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
1001 2096 aaronmk
    run_query(db, 'ALTER TABLE '+table+' ADD COLUMN '+row_num_col
1002 2443 aaronmk
        +' serial NOT NULL PRIMARY KEY', log_level=3)
1003 2086 aaronmk
1004 2889 aaronmk
def drop_table(db, table):
1005
    table = sql_gen.as_Table(table)
1006
    return run_query(db, 'DROP TABLE IF EXISTS '+table.to_str(db)+' CASCADE')
1007
1008 2760 aaronmk
def create_table(db, table, cols, has_pkey=True, col_indexes=True):
1009 2675 aaronmk
    '''Creates a table.
1010 2681 aaronmk
    @param cols [sql_gen.TypedCol,...] The column names and types
1011
    @param has_pkey If set, the first column becomes the primary key.
1012 2760 aaronmk
    @param col_indexes bool|[ref]
1013
        * If True, indexes will be added on all non-pkey columns.
1014
        * If a list reference, [0] will be set to a function to do this.
1015
          This can be used to delay index creation until the table is populated.
1016 2675 aaronmk
    '''
1017
    table = sql_gen.as_Table(table)
1018
1019 2681 aaronmk
    if has_pkey:
1020
        cols[0] = pkey = copy.copy(cols[0]) # don't modify input!
1021 2872 aaronmk
        pkey.constraints = 'PRIMARY KEY'
1022 2681 aaronmk
1023 2675 aaronmk
    str_ = 'CREATE TABLE '+table.to_str(db)+' (\n'
1024 2681 aaronmk
    str_ += '\n, '.join(v.to_str(db) for v in cols)
1025
    str_ += '\n);\n'
1026 2675 aaronmk
    run_query(db, str_, cacheable=True, log_level=2)
1027 2760 aaronmk
1028
    # Add indexes
1029 2773 aaronmk
    if has_pkey: has_pkey = already_indexed
1030
    def add_indexes_(): add_indexes(db, table, has_pkey)
1031
    if isinstance(col_indexes, list): col_indexes[0] = add_indexes_ # defer
1032
    elif col_indexes: add_indexes_() # add now
1033 2675 aaronmk
1034 2684 aaronmk
def vacuum(db, table):
1035
    table = sql_gen.as_Table(table)
1036
    db.with_autocommit(lambda: run_query(db, 'VACUUM ANALYZE '+table.to_str(db),
1037
        log_level=3))
1038
1039 2732 aaronmk
def truncate(db, table, schema='public'):
1040 2777 aaronmk
    table = sql_gen.as_Table(table, schema)
1041 2733 aaronmk
    return run_query(db, 'TRUNCATE '+table.to_str(db)+' CASCADE')
1042 2732 aaronmk
1043 2729 aaronmk
def tables(db, schema_like='public', table_like='%', exact=False):
1044
    if exact: compare = '='
1045
    else: compare = 'LIKE'
1046
1047 1849 aaronmk
    module = util.root_module(db.db)
1048 832 aaronmk
    if module == 'psycopg2':
1049 2779 aaronmk
        conds = [('schemaname', sql_gen.CompareCond(schema_like, compare)),
1050
            ('tablename', sql_gen.CompareCond(table_like, compare))]
1051
        return values(select(db, 'pg_tables', ['tablename'], conds,
1052
            order_by='tablename', log_level=4))
1053 1968 aaronmk
    elif module == 'MySQLdb':
1054 2779 aaronmk
        return values(run_query(db, 'SHOW TABLES LIKE '+db.esc_value(table_like)
1055
            , cacheable=True, log_level=4))
1056 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
1057 830 aaronmk
1058 2726 aaronmk
def table_exists(db, table):
1059
    table = sql_gen.as_Table(table)
1060 2730 aaronmk
    return list(tables(db, table.schema, table.name, exact=True)) != []
1061 2726 aaronmk
1062
def errors_table(db, table, if_exists=True):
1063
    '''
1064
    @param if_exists If set, returns None if the errors table doesn't exist
1065
    @return None|sql_gen.Table
1066
    '''
1067
    table = sql_gen.as_Table(table)
1068
    if table.srcs != (): table = table.srcs[0]
1069
1070
    errors_table = sql_gen.suffixed_table(table, '.errors')
1071
    if if_exists and not table_exists(db, errors_table): return None
1072
    return errors_table
1073
1074 833 aaronmk
##### Database management
1075
1076 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
1077
    '''For kw_args, see tables()'''
1078
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
1079 833 aaronmk
1080 832 aaronmk
##### Heuristic queries
1081
1082 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
1083 1554 aaronmk
    '''Recovers from errors.
1084 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
1085
    '''
1086 2642 aaronmk
    row = sql_gen.ColDict(db, table, row)
1087 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
1088
1089 471 aaronmk
    try:
1090 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
1091 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
1092
            row_ct_ref[0] += cur.rowcount
1093
        return value(cur)
1094 471 aaronmk
    except DuplicateKeyException, e:
1095 2653 aaronmk
        row = sql_gen.ColDict(db, table,
1096
            util.dict_subset_right_join(row, e.cols))
1097
        return value(select(db, table, [pkey_], row, recover=True))
1098 471 aaronmk
1099 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
1100 830 aaronmk
    '''Recovers from errors'''
1101 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
1102 14 aaronmk
    except StopIteration:
1103 40 aaronmk
        if not create: raise
1104 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
1105 2078 aaronmk
1106 2593 aaronmk
def is_func_result(col):
1107
    return col.table.name.find('(') >= 0 and col.name == 'result'
1108
1109 2592 aaronmk
def into_table_name(out_table, in_tables0, mapping, is_func):
1110 2615 aaronmk
    def in_col_str(in_col):
1111
        in_col = sql_gen.remove_col_rename(in_col)
1112
        if isinstance(in_col, sql_gen.Col):
1113
            table = in_col.table
1114
            if table == in_tables0:
1115
                in_col = sql_gen.to_name_only_col(in_col)
1116
            elif is_func_result(in_col): in_col = table # omit col name
1117
        return str(in_col)
1118
1119 2580 aaronmk
    str_ = str(out_table)
1120
    if is_func:
1121 2615 aaronmk
        str_ += '('
1122 2580 aaronmk
1123 2615 aaronmk
        try: value_in_col = mapping['value']
1124
        except KeyError:
1125 2654 aaronmk
            str_ += ', '.join((str(k)+'='+in_col_str(v)
1126 2615 aaronmk
                for k, v in mapping.iteritems()))
1127
        else: str_ += in_col_str(value_in_col)
1128
1129
        str_ += ')'
1130 2812 aaronmk
    else:
1131
        out_col = 'rank'
1132
        try: in_col = mapping[out_col]
1133
        except KeyError: str_ += '_pkeys'
1134
        else: # has a rank column, so hierarchical
1135
            str_ += '['+str(out_col)+'='+in_col_str(in_col)+']'
1136 2580 aaronmk
    return str_
1137
1138 2508 aaronmk
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None, into=None,
1139 2805 aaronmk
    default=None, is_func=False, on_error=exc.raise_):
1140 2078 aaronmk
    '''Recovers from errors.
1141
    Only works under PostgreSQL (uses INSERT RETURNING).
1142 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
1143
        tables to join with it using the main input table's pkey
1144 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
1145 2612 aaronmk
        * out_table_col: str (*not* sql_gen.Col)
1146
        * in_table_col: sql_gen.Col|literal-value
1147 2489 aaronmk
    @param into The table to contain the output and input pkeys.
1148 2574 aaronmk
        Defaults to `out_table.name+'_pkeys'`.
1149 2509 aaronmk
    @param default The *output* column to use as the pkey for missing rows.
1150
        If this output column does not exist in the mapping, uses None.
1151 2552 aaronmk
    @param is_func Whether out_table is the name of a SQL function, not a table
1152 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
1153 2078 aaronmk
    '''
1154 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
1155 2312 aaronmk
1156 2450 aaronmk
    def log_debug(msg): db.log_debug(msg, level=1.5)
1157 2505 aaronmk
    def col_ustr(str_):
1158 2567 aaronmk
        return strings.repr_no_u(sql_gen.remove_col_rename(str_))
1159 2450 aaronmk
1160 2768 aaronmk
    out_pkey = pkey(db, out_table, recover=True)
1161
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
1162
1163
    if mapping == {}: # need at least one column for INSERT SELECT
1164
        mapping = {out_pkey: None} # ColDict will replace with default value
1165
1166 2486 aaronmk
    log_debug('********** New iteration **********')
1167 2505 aaronmk
    log_debug('Inserting these input columns into '+strings.as_tt(
1168
        out_table.to_str(db))+':\n'+strings.as_table(mapping, ustr=col_ustr))
1169 2463 aaronmk
1170 2382 aaronmk
    # Create input joins from list of input tables
1171
    in_tables_ = in_tables[:] # don't modify input!
1172
    in_tables0 = in_tables_.pop(0) # first table is separate
1173 2735 aaronmk
    errors_table_ = errors_table(db, in_tables0)
1174 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
1175 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
1176 2460 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v,
1177
        {in_pkey: sql_gen.join_same_not_null}) for v in in_tables_]
1178 2131 aaronmk
1179 2592 aaronmk
    if into == None:
1180
        into = into_table_name(out_table, in_tables0, mapping, is_func)
1181
    into = sql_gen.as_Table(into)
1182
1183 2702 aaronmk
    # Set column sources
1184
    in_cols = filter(sql_gen.is_table_col, mapping.values())
1185
    for col in in_cols:
1186
        if col.table == in_tables0: col.set_srcs(sql_gen.src_self)
1187
1188 2486 aaronmk
    log_debug('Joining together input tables into temp table')
1189 2395 aaronmk
    # Place in new table for speed and so don't modify input if values edited
1190 2584 aaronmk
    in_table = sql_gen.Table('in')
1191 2702 aaronmk
    mapping = dicts.join(mapping, flatten(db, in_table, input_joins, in_cols,
1192
        preserve=[in_pkey_col], start=0))
1193 2395 aaronmk
    input_joins = [in_table]
1194 2486 aaronmk
    db.log_debug('Temp table: '+strings.as_tt(in_table.to_str(db)), level=2)
1195 2395 aaronmk
1196 2656 aaronmk
    mapping = sql_gen.ColDict(db, out_table, mapping)
1197
        # after applying dicts.join() because that returns a plain dict
1198
1199 2509 aaronmk
    # Resolve default value column
1200 2841 aaronmk
    if default != None:
1201
        try: default = mapping[default]
1202
        except KeyError:
1203 2509 aaronmk
            db.log_debug('Default value column '
1204
                +strings.as_tt(strings.repr_no_u(default))
1205 2511 aaronmk
                +' does not exist in mapping, falling back to None', level=2.1)
1206 2509 aaronmk
            default = None
1207
1208 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
1209 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
1210
1211 2201 aaronmk
    pkeys_table_exists_ref = [False]
1212 2420 aaronmk
    def insert_into_pkeys(joins, cols):
1213 2786 aaronmk
        query = mk_select(db, joins, cols, order_by=None, start=0)
1214 2201 aaronmk
        if pkeys_table_exists_ref[0]:
1215 2786 aaronmk
            insert_select(db, into, pkeys_names, query)
1216 2201 aaronmk
        else:
1217 2786 aaronmk
            run_query_into(db, query, into=into)
1218 2201 aaronmk
            pkeys_table_exists_ref[0] = True
1219
1220 2429 aaronmk
    limit_ref = [None]
1221 2380 aaronmk
    conds = set()
1222 2811 aaronmk
    distinct_on = sql_gen.ColDict(db, out_table)
1223 2325 aaronmk
    def mk_main_select(joins, cols):
1224 2811 aaronmk
        distinct_on_cols = [c.to_Col() for c in distinct_on.values()]
1225
        return mk_select(db, joins, cols, conds, distinct_on_cols,
1226 2429 aaronmk
            limit=limit_ref[0], start=0)
1227 2132 aaronmk
1228 2519 aaronmk
    exc_strs = set()
1229 2309 aaronmk
    def log_exc(e):
1230 2519 aaronmk
        e_str = exc.str_(e, first_line_only=True)
1231
        log_debug('Caught exception: '+e_str)
1232
        assert e_str not in exc_strs # avoid infinite loops
1233
        exc_strs.add(e_str)
1234 2735 aaronmk
1235 2451 aaronmk
    def remove_all_rows():
1236 2896 aaronmk
        log_debug('Ignoring all rows')
1237 2429 aaronmk
        limit_ref[0] = 0 # just create an empty pkeys table
1238 2735 aaronmk
1239 2758 aaronmk
    def ignore(in_col, value, e):
1240
        track_data_error(db, errors_table_, in_col.srcs, value, e.cause.pgcode,
1241
            e.cause.pgerror)
1242 2895 aaronmk
        log_debug('Ignoring rows with '+strings.as_tt(repr(in_col))+' = '
1243 2545 aaronmk
            +strings.as_tt(repr(value)))
1244 2895 aaronmk
1245 2758 aaronmk
    def remove_rows(in_col, value, e):
1246
        ignore(in_col, value, e)
1247 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
1248
        assert cond not in conds # avoid infinite loops
1249 2380 aaronmk
        conds.add(cond)
1250 2895 aaronmk
1251 2758 aaronmk
    def invalid2null(in_col, value, e):
1252
        ignore(in_col, value, e)
1253 2403 aaronmk
        update(db, in_table, [(in_col, None)],
1254
            sql_gen.ColValueCond(in_col, value))
1255 2245 aaronmk
1256 2589 aaronmk
    def insert_pkeys_table(which):
1257
        return sql_gen.Table(sql_gen.add_suffix(in_table.name,
1258
            '_insert_'+which+'_pkeys'))
1259
    insert_out_pkeys = insert_pkeys_table('out')
1260
    insert_in_pkeys = insert_pkeys_table('in')
1261
1262 2900 aaronmk
    log_debug('Locking output tables to prevent concurrent duplicate keys')
1263
    for table in [out_table, errors_table_]: lock_table(db, table, 'EXCLUSIVE')
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 2305 aaronmk
        # Prepare to insert new rows
1277 2325 aaronmk
        insert_joins = input_joins[:] # don't modify original!
1278 2403 aaronmk
        insert_args = dict(recover=True, cacheable=False)
1279 2303 aaronmk
        if has_joins:
1280 2325 aaronmk
            insert_joins.append(sql_gen.Join(out_table, join_cols,
1281
                sql_gen.filter_out))
1282
        else:
1283 2404 aaronmk
            insert_args.update(dict(returning=out_pkey, into=insert_out_pkeys))
1284 2786 aaronmk
        main_select = mk_main_select(insert_joins, mapping.values())
1285 2303 aaronmk
1286 2486 aaronmk
        log_debug('Trying to insert new rows')
1287 2206 aaronmk
        try:
1288 2518 aaronmk
            cur = insert_select(db, out_table, mapping.keys(), main_select,
1289
                **insert_args)
1290 2357 aaronmk
            break # insert successful
1291 2206 aaronmk
        except DuplicateKeyException, e:
1292 2309 aaronmk
            log_exc(e)
1293
1294 2258 aaronmk
            old_join_cols = join_cols.copy()
1295 2811 aaronmk
            distinct_on.update(util.dict_subset(mapping, e.cols))
1296 2565 aaronmk
            join_cols.update(util.dict_subset_right_join(mapping, e.cols))
1297 2486 aaronmk
            log_debug('Ignoring existing rows, comparing on these columns:\n'
1298 2505 aaronmk
                +strings.as_inline_table(join_cols, ustr=col_ustr))
1299 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
1300 2230 aaronmk
        except NullValueException, e:
1301 2309 aaronmk
            log_exc(e)
1302
1303 2230 aaronmk
            out_col, = e.cols
1304
            try: in_col = mapping[out_col]
1305 2356 aaronmk
            except KeyError:
1306 2486 aaronmk
                log_debug('Missing mapping for NOT NULL column '+out_col)
1307 2451 aaronmk
                remove_all_rows()
1308 2758 aaronmk
            else: remove_rows(in_col, None, e)
1309 2542 aaronmk
        except FunctionValueException, e:
1310
            log_exc(e)
1311
1312
            func_name = e.name
1313
            value = e.value
1314
            for out_col, in_col in mapping.iteritems():
1315 2758 aaronmk
                in_col = sql_gen.unwrap_func_call(in_col, func_name)
1316
                invalid2null(in_col, value, e)
1317 2525 aaronmk
        except MissingCastException, e:
1318
            log_exc(e)
1319
1320
            out_col = e.col
1321 2706 aaronmk
            type_ = e.type
1322
1323 2724 aaronmk
            log_debug('Casting '+strings.as_tt(out_col)+' input to '
1324
                +strings.as_tt(type_))
1325 2735 aaronmk
            def wrap_func(col): return cast(db, type_, col, errors_table_)
1326 2705 aaronmk
            mapping[out_col] = sql_gen.wrap(wrap_func, mapping[out_col])
1327 2429 aaronmk
        except DatabaseErrors, e:
1328
            log_exc(e)
1329
1330 2808 aaronmk
            log_debug('No handler for exception')
1331
            on_error(e)
1332 2451 aaronmk
            remove_all_rows()
1333 2358 aaronmk
        # after exception handled, rerun loop with additional constraints
1334 2132 aaronmk
1335 2357 aaronmk
    if row_ct_ref != None and cur.rowcount >= 0:
1336
        row_ct_ref[0] += cur.rowcount
1337
1338
    if has_joins:
1339
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
1340 2486 aaronmk
        log_debug('Getting output table pkeys of existing/inserted rows')
1341 2420 aaronmk
        insert_into_pkeys(select_joins, pkeys_cols)
1342 2357 aaronmk
    else:
1343 2404 aaronmk
        add_row_num(db, insert_out_pkeys) # for joining with input pkeys
1344 2357 aaronmk
1345 2486 aaronmk
        log_debug('Getting input table pkeys of inserted rows')
1346 2786 aaronmk
        run_query_into(db, mk_main_select(input_joins, [in_pkey]),
1347 2404 aaronmk
            into=insert_in_pkeys)
1348
        add_row_num(db, insert_in_pkeys) # for joining with output pkeys
1349 2357 aaronmk
1350 2428 aaronmk
        assert table_row_count(db, insert_out_pkeys) == table_row_count(db,
1351
            insert_in_pkeys)
1352
1353 2486 aaronmk
        log_debug('Combining output and input pkeys in inserted order')
1354 2404 aaronmk
        pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
1355 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
1356 2420 aaronmk
        insert_into_pkeys(pkey_joins, pkeys_names)
1357 2357 aaronmk
1358 2486 aaronmk
    db.log_debug('Adding pkey on pkeys table to enable fast joins', level=2.5)
1359 2594 aaronmk
    add_pkey(db, into)
1360 2407 aaronmk
1361 2508 aaronmk
    log_debug('Setting pkeys of missing rows to '+strings.as_tt(repr(default)))
1362 2489 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(into,
1363 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
1364
        # must use join_same_not_null or query will take forever
1365 2420 aaronmk
    insert_into_pkeys(missing_rows_joins,
1366 2508 aaronmk
        [in_pkey_col, sql_gen.NamedCol(out_pkey, default)])
1367 2357 aaronmk
1368 2489 aaronmk
    assert table_row_count(db, into) == table_row_count(db, in_table)
1369 2428 aaronmk
1370 2702 aaronmk
    srcs = []
1371
    if is_func: srcs = sql_gen.cols_srcs(in_cols)
1372
    return sql_gen.Col(out_pkey, into, srcs)
1373 2115 aaronmk
1374
##### Data cleanup
1375
1376 2290 aaronmk
def cleanup_table(db, table, cols):
1377 2677 aaronmk
    table = sql_gen.as_Table(table)
1378
    cols = map(sql_gen.as_Col, cols)
1379 2115 aaronmk
1380 2677 aaronmk
    expr = ('nullif(nullif(trim(both from %s), '+db.esc_value('')+'), '
1381
        +db.esc_value(r'\N')+')')
1382
    changes = [(v, sql_gen.CustomCode(expr % v.to_str(db)))
1383
        for v in cols]
1384 2115 aaronmk
1385 2677 aaronmk
    update(db, table, changes)