Project

General

Profile

1 11 aaronmk
# Database access
2
3 1869 aaronmk
import copy
4 2127 aaronmk
import operator
5 11 aaronmk
import re
6 865 aaronmk
import warnings
7 11 aaronmk
8 300 aaronmk
import exc
9 1909 aaronmk
import dicts
10 1893 aaronmk
import iters
11 1960 aaronmk
import lists
12 1889 aaronmk
from Proxy import Proxy
13 1872 aaronmk
import rand
14 2217 aaronmk
import sql_gen
15 862 aaronmk
import strings
16 131 aaronmk
import util
17 11 aaronmk
18 832 aaronmk
##### Exceptions
19
20 2804 aaronmk
def get_cur_query(cur, input_query=None):
21 2168 aaronmk
    raw_query = None
22
    if hasattr(cur, 'query'): raw_query = cur.query
23
    elif hasattr(cur, '_last_executed'): raw_query = cur._last_executed
24 2170 aaronmk
25
    if raw_query != None: return raw_query
26 2804 aaronmk
    else: return '[input] '+strings.ustr(input_query)
27 14 aaronmk
28 2170 aaronmk
def _add_cursor_info(e, *args, **kw_args):
29
    '''For params, see get_cur_query()'''
30 2771 aaronmk
    exc.add_msg(e, 'query: '+strings.ustr(get_cur_query(*args, **kw_args)))
31 135 aaronmk
32 300 aaronmk
class DbException(exc.ExceptionWithCause):
33 14 aaronmk
    def __init__(self, msg, cause=None, cur=None):
34 2145 aaronmk
        exc.ExceptionWithCause.__init__(self, msg, cause, cause_newline=True)
35 14 aaronmk
        if cur != None: _add_cursor_info(self, cur)
36
37 2143 aaronmk
class ExceptionWithName(DbException):
38
    def __init__(self, name, cause=None):
39 2484 aaronmk
        DbException.__init__(self, 'for name: '+strings.as_tt(str(name)), cause)
40 2143 aaronmk
        self.name = name
41 360 aaronmk
42 2240 aaronmk
class ExceptionWithNameValue(DbException):
43
    def __init__(self, name, value, cause=None):
44 2484 aaronmk
        DbException.__init__(self, 'for name: '+strings.as_tt(str(name))
45
            +'; value: '+strings.as_tt(repr(value)), cause)
46 2240 aaronmk
        self.name = name
47
        self.value = value
48
49 2306 aaronmk
class ConstraintException(DbException):
50
    def __init__(self, name, cols, cause=None):
51 2484 aaronmk
        DbException.__init__(self, 'Violated '+strings.as_tt(name)
52
            +' constraint on columns: '+strings.as_tt(', '.join(cols)), cause)
53 2306 aaronmk
        self.name = name
54 468 aaronmk
        self.cols = cols
55 11 aaronmk
56 2523 aaronmk
class MissingCastException(DbException):
57
    def __init__(self, type_, col, cause=None):
58
        DbException.__init__(self, 'Missing cast to type '+strings.as_tt(type_)
59
            +' on column: '+strings.as_tt(col), cause)
60
        self.type = type_
61
        self.col = col
62
63 2143 aaronmk
class NameException(DbException): pass
64
65 2306 aaronmk
class DuplicateKeyException(ConstraintException): pass
66 13 aaronmk
67 2306 aaronmk
class NullValueException(ConstraintException): pass
68 13 aaronmk
69 2240 aaronmk
class FunctionValueException(ExceptionWithNameValue): pass
70 2239 aaronmk
71 2143 aaronmk
class DuplicateTableException(ExceptionWithName): pass
72
73 2188 aaronmk
class DuplicateFunctionException(ExceptionWithName): pass
74
75 89 aaronmk
class EmptyRowException(DbException): pass
76
77 865 aaronmk
##### Warnings
78
79
class DbWarning(UserWarning): pass
80
81 1930 aaronmk
##### Result retrieval
82
83
def col_names(cur): return (col[0] for col in cur.description)
84
85
def rows(cur): return iter(lambda: cur.fetchone(), None)
86
87
def consume_rows(cur):
88
    '''Used to fetch all rows so result will be cached'''
89
    iters.consume_iter(rows(cur))
90
91
def next_row(cur): return rows(cur).next()
92
93
def row(cur):
94
    row_ = next_row(cur)
95
    consume_rows(cur)
96
    return row_
97
98
def next_value(cur): return next_row(cur)[0]
99
100
def value(cur): return row(cur)[0]
101
102
def values(cur): return iters.func_iter(lambda: next_value(cur))
103
104
def value_or_none(cur):
105
    try: return value(cur)
106
    except StopIteration: return None
107
108 2762 aaronmk
##### Escaping
109 2101 aaronmk
110 2573 aaronmk
def esc_name_by_module(module, name):
111
    if module == 'psycopg2' or module == None: quote = '"'
112 2101 aaronmk
    elif module == 'MySQLdb': quote = '`'
113
    else: raise NotImplementedError("Can't escape name for "+module+' database')
114 2500 aaronmk
    return sql_gen.esc_name(name, quote)
115 2101 aaronmk
116
def esc_name_by_engine(engine, name, **kw_args):
117
    return esc_name_by_module(db_engines[engine][0], name, **kw_args)
118
119
def esc_name(db, name, **kw_args):
120
    return esc_name_by_module(util.root_module(db.db), name, **kw_args)
121
122
def qual_name(db, schema, table):
123
    def esc_name_(name): return esc_name(db, name)
124
    table = esc_name_(table)
125
    if schema != None: return esc_name_(schema)+'.'+table
126
    else: return table
127
128 1869 aaronmk
##### Database connections
129 1849 aaronmk
130 2097 aaronmk
db_config_names = ['engine', 'host', 'user', 'password', 'database', 'schemas']
131 1926 aaronmk
132 1869 aaronmk
db_engines = {
133
    'MySQL': ('MySQLdb', {'password': 'passwd', 'database': 'db'}),
134
    'PostgreSQL': ('psycopg2', {}),
135
}
136
137
DatabaseErrors_set = set([DbException])
138
DatabaseErrors = tuple(DatabaseErrors_set)
139
140
def _add_module(module):
141
    DatabaseErrors_set.add(module.DatabaseError)
142
    global DatabaseErrors
143
    DatabaseErrors = tuple(DatabaseErrors_set)
144
145
def db_config_str(db_config):
146
    return db_config['engine']+' database '+db_config['database']
147
148 2448 aaronmk
log_debug_none = lambda msg, level=2: None
149 1901 aaronmk
150 1849 aaronmk
class DbConn:
151 2905 aaronmk
    def __init__(self, db_config, autocommit=False, caching=True,
152
        log_debug=log_debug_none):
153 1869 aaronmk
        self.db_config = db_config
154 2190 aaronmk
        self.autocommit = autocommit
155
        self.caching = caching
156 1901 aaronmk
        self.log_debug = log_debug
157 2193 aaronmk
        self.debug = log_debug != log_debug_none
158 1869 aaronmk
159
        self.__db = None
160 1889 aaronmk
        self.query_results = {}
161 2139 aaronmk
        self._savepoint = 0
162 2671 aaronmk
        self._notices_seen = set()
163 1869 aaronmk
164
    def __getattr__(self, name):
165
        if name == '__dict__': raise Exception('getting __dict__')
166
        if name == 'db': return self._db()
167
        else: raise AttributeError()
168
169
    def __getstate__(self):
170
        state = copy.copy(self.__dict__) # shallow copy
171 1915 aaronmk
        state['log_debug'] = None # don't pickle the debug callback
172 1869 aaronmk
        state['_DbConn__db'] = None # don't pickle the connection
173
        return state
174
175 2165 aaronmk
    def connected(self): return self.__db != None
176
177 1869 aaronmk
    def _db(self):
178
        if self.__db == None:
179
            # Process db_config
180
            db_config = self.db_config.copy() # don't modify input!
181 2097 aaronmk
            schemas = db_config.pop('schemas', None)
182 1869 aaronmk
            module_name, mappings = db_engines[db_config.pop('engine')]
183
            module = __import__(module_name)
184
            _add_module(module)
185
            for orig, new in mappings.iteritems():
186
                try: util.rename_key(db_config, orig, new)
187
                except KeyError: pass
188
189
            # Connect
190
            self.__db = module.connect(**db_config)
191
192
            # Configure connection
193 2906 aaronmk
            if hasattr(self.db, 'set_isolation_level'):
194
                import psycopg2.extensions
195
                self.db.set_isolation_level(
196
                    psycopg2.extensions.ISOLATION_LEVEL_READ_COMMITTED)
197 2101 aaronmk
            if schemas != None:
198 2893 aaronmk
                search_path = [self.esc_name(s) for s in schemas.split(',')]
199
                search_path.append(value(run_query(self, 'SHOW search_path',
200
                    log_level=4)))
201
                run_query(self, 'SET search_path TO '+(','.join(search_path)),
202
                    log_level=3)
203 1869 aaronmk
204
        return self.__db
205 1889 aaronmk
206 1891 aaronmk
    class DbCursor(Proxy):
207 1927 aaronmk
        def __init__(self, outer):
208 1891 aaronmk
            Proxy.__init__(self, outer.db.cursor())
209 2191 aaronmk
            self.outer = outer
210 1927 aaronmk
            self.query_results = outer.query_results
211 1894 aaronmk
            self.query_lookup = None
212 1891 aaronmk
            self.result = []
213 1889 aaronmk
214 2802 aaronmk
        def execute(self, query):
215 2764 aaronmk
            self._is_insert = query.startswith('INSERT')
216 2797 aaronmk
            self.query_lookup = query
217 2148 aaronmk
            try:
218 2191 aaronmk
                try:
219 2802 aaronmk
                    cur = self.inner.execute(query)
220 2191 aaronmk
                    self.outer.do_autocommit()
221 2802 aaronmk
                finally: self.query = get_cur_query(self.inner, query)
222 1904 aaronmk
            except Exception, e:
223 2802 aaronmk
                _add_cursor_info(e, self, query)
224 1904 aaronmk
                self.result = e # cache the exception as the result
225
                self._cache_result()
226
                raise
227 2800 aaronmk
            if self.rowcount == 0 and query.startswith('SELECT'): # empty SELECT
228
                consume_rows(self) # fetch all rows so result will be cached
229 2762 aaronmk
            return cur
230 1894 aaronmk
231 1891 aaronmk
        def fetchone(self):
232
            row = self.inner.fetchone()
233 1899 aaronmk
            if row != None: self.result.append(row)
234
            # otherwise, fetched all rows
235 1904 aaronmk
            else: self._cache_result()
236
            return row
237
238
        def _cache_result(self):
239 1906 aaronmk
            # For inserts, only cache exceptions since inserts are not
240
            # idempotent, but an invalid insert will always be invalid
241 1930 aaronmk
            if self.query_results != None and (not self._is_insert
242 1906 aaronmk
                or isinstance(self.result, Exception)):
243
244 1894 aaronmk
                assert self.query_lookup != None
245 1916 aaronmk
                self.query_results[self.query_lookup] = self.CacheCursor(
246
                    util.dict_subset(dicts.AttrsDictView(self),
247
                    ['query', 'result', 'rowcount', 'description']))
248 1906 aaronmk
249 1916 aaronmk
        class CacheCursor:
250
            def __init__(self, cached_result): self.__dict__ = cached_result
251
252 1927 aaronmk
            def execute(self, *args, **kw_args):
253 1916 aaronmk
                if isinstance(self.result, Exception): raise self.result
254
                # otherwise, result is a rows list
255
                self.iter = iter(self.result)
256
257
            def fetchone(self):
258
                try: return self.iter.next()
259
                except StopIteration: return None
260 1891 aaronmk
261 2212 aaronmk
    def esc_value(self, value):
262 2663 aaronmk
        try: str_ = self.mogrify('%s', [value])
263
        except NotImplementedError, e:
264
            module = util.root_module(self.db)
265
            if module == 'MySQLdb':
266
                import _mysql
267
                str_ = _mysql.escape_string(value)
268
            else: raise e
269 2374 aaronmk
        return strings.to_unicode(str_)
270 2212 aaronmk
271 2347 aaronmk
    def esc_name(self, name): return esc_name(self, name) # calls global func
272
273 2814 aaronmk
    def std_code(self, str_):
274
        '''Standardizes SQL code.
275
        * Ensures that string literals are prefixed by `E`
276
        '''
277
        if str_.startswith("'"): str_ = 'E'+str_
278
        return str_
279
280 2665 aaronmk
    def can_mogrify(self):
281 2663 aaronmk
        module = util.root_module(self.db)
282 2665 aaronmk
        return module == 'psycopg2'
283 2663 aaronmk
284 2665 aaronmk
    def mogrify(self, query, params=None):
285
        if self.can_mogrify(): return self.db.cursor().mogrify(query, params)
286
        else: raise NotImplementedError("Can't mogrify query")
287
288 2671 aaronmk
    def print_notices(self):
289 2725 aaronmk
        if hasattr(self.db, 'notices'):
290
            for msg in self.db.notices:
291
                if msg not in self._notices_seen:
292
                    self._notices_seen.add(msg)
293
                    self.log_debug(msg, level=2)
294 2671 aaronmk
295 2793 aaronmk
    def run_query(self, query, cacheable=False, log_level=2,
296 2464 aaronmk
        debug_msg_ref=None):
297 2445 aaronmk
        '''
298 2464 aaronmk
        @param log_ignore_excs The log_level will be increased by 2 if the query
299
            throws one of these exceptions.
300 2664 aaronmk
        @param debug_msg_ref If specified, the log message will be returned in
301
            this instead of being output. This allows you to filter log messages
302
            depending on the result of the query.
303 2445 aaronmk
        '''
304 2167 aaronmk
        assert query != None
305
306 2047 aaronmk
        if not self.caching: cacheable = False
307 1903 aaronmk
        used_cache = False
308 2664 aaronmk
309
        def log_msg(query):
310
            if used_cache: cache_status = 'cache hit'
311
            elif cacheable: cache_status = 'cache miss'
312
            else: cache_status = 'non-cacheable'
313
            return 'DB query: '+cache_status+':\n'+strings.as_code(query, 'SQL')
314
315 1903 aaronmk
        try:
316 1927 aaronmk
            # Get cursor
317
            if cacheable:
318
                try:
319 2797 aaronmk
                    cur = self.query_results[query]
320 1927 aaronmk
                    used_cache = True
321
                except KeyError: cur = self.DbCursor(self)
322
            else: cur = self.db.cursor()
323
324 2664 aaronmk
            # Log query
325
            if self.debug and debug_msg_ref == None: # log before running
326
                self.log_debug(log_msg(query), log_level)
327
328 1927 aaronmk
            # Run query
329 2793 aaronmk
            cur.execute(query)
330 1903 aaronmk
        finally:
331 2671 aaronmk
            self.print_notices()
332 2664 aaronmk
            if self.debug and debug_msg_ref != None: # return after running
333 2793 aaronmk
                debug_msg_ref[0] = log_msg(str(get_cur_query(cur, query)))
334 1903 aaronmk
335
        return cur
336 1914 aaronmk
337 2797 aaronmk
    def is_cached(self, query): return query in self.query_results
338 2139 aaronmk
339 2907 aaronmk
    def with_autocommit(self, func):
340 2801 aaronmk
        import psycopg2.extensions
341
342
        prev_isolation_level = self.db.isolation_level
343 2907 aaronmk
        self.db.set_isolation_level(
344
            psycopg2.extensions.ISOLATION_LEVEL_AUTOCOMMIT)
345 2683 aaronmk
        try: return func()
346 2801 aaronmk
        finally: self.db.set_isolation_level(prev_isolation_level)
347 2683 aaronmk
348 2139 aaronmk
    def with_savepoint(self, func):
349 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
350 2443 aaronmk
        self.run_query('SAVEPOINT '+savepoint, log_level=4)
351 2139 aaronmk
        self._savepoint += 1
352 2683 aaronmk
        try:
353 2139 aaronmk
            try: return_val = func()
354
            finally:
355
                self._savepoint -= 1
356
                assert self._savepoint >= 0
357
        except:
358 2443 aaronmk
            self.run_query('ROLLBACK TO SAVEPOINT '+savepoint, log_level=4)
359 2139 aaronmk
            raise
360
        else:
361 2443 aaronmk
            self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
362 2191 aaronmk
            self.do_autocommit()
363 2139 aaronmk
            return return_val
364 2191 aaronmk
365
    def do_autocommit(self):
366
        '''Autocommits if outside savepoint'''
367
        assert self._savepoint >= 0
368
        if self.autocommit and self._savepoint == 0:
369 2668 aaronmk
            self.log_debug('Autocommitting')
370 2191 aaronmk
            self.db.commit()
371 2643 aaronmk
372 2819 aaronmk
    def col_info(self, col):
373 2643 aaronmk
        table = sql_gen.Table('columns', 'information_schema')
374 2821 aaronmk
        cols = ['data_type', 'column_default',
375
            cast(self, 'boolean', 'is_nullable')]
376 2643 aaronmk
377
        conds = [('table_name', col.table.name), ('column_name', col.name)]
378
        schema = col.table.schema
379
        if schema != None: conds.append(('table_schema', schema))
380
381 2819 aaronmk
        type_, default, nullable = row(select(self, table, cols, conds,
382 2874 aaronmk
            order_by='table_schema', limit=1, log_level=4))
383 2643 aaronmk
            # TODO: order_by search_path schema order
384 2819 aaronmk
        default = sql_gen.as_Code(default, self)
385
386
        return sql_gen.TypedCol(col.name, type_, default, nullable)
387 1849 aaronmk
388 1869 aaronmk
connect = DbConn
389
390 832 aaronmk
##### Recoverable querying
391 15 aaronmk
392 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
393 11 aaronmk
394 2791 aaronmk
def run_query(db, query, recover=None, cacheable=False, log_level=2,
395
    log_ignore_excs=None, **kw_args):
396 2794 aaronmk
    '''For params, see DbConn.run_query()'''
397 830 aaronmk
    if recover == None: recover = False
398 2464 aaronmk
    if log_ignore_excs == None: log_ignore_excs = ()
399
    log_ignore_excs = tuple(log_ignore_excs)
400 830 aaronmk
401 2666 aaronmk
    debug_msg_ref = None # usually, db.run_query() logs query before running it
402
    # But if filtering with log_ignore_excs, wait until after exception parsing
403
    if log_ignore_excs != () or not db.can_mogrify(): debug_msg_ref = [None]
404
405 2148 aaronmk
    try:
406 2464 aaronmk
        try:
407 2794 aaronmk
            def run(): return db.run_query(query, cacheable, log_level,
408 2793 aaronmk
                debug_msg_ref, **kw_args)
409 2796 aaronmk
            if recover and not db.is_cached(query):
410 2464 aaronmk
                return with_savepoint(db, run)
411
            else: return run() # don't need savepoint if cached
412
        except Exception, e:
413
            if not recover: raise # need savepoint to run index_cols()
414
            msg = exc.str_(e)
415
416
            match = re.search(r'duplicate key value violates unique constraint '
417 2493 aaronmk
                r'"((_?[^\W_]+)_.+?)"', msg)
418 2464 aaronmk
            if match:
419
                constraint, table = match.groups()
420
                try: cols = index_cols(db, table, constraint)
421
                except NotImplementedError: raise e
422
                else: raise DuplicateKeyException(constraint, cols, e)
423
424 2493 aaronmk
            match = re.search(r'null value in column "(.+?)" violates not-null'
425 2464 aaronmk
                r' constraint', msg)
426
            if match: raise NullValueException('NOT NULL', [match.group(1)], e)
427
428
            match = re.search(r'\b(?:invalid input (?:syntax|value)\b.*?'
429
                r'|date/time field value out of range): "(.+?)"\n'
430 2535 aaronmk
                r'(?:(?s).*?)\bfunction "(.+?)"', msg)
431 2464 aaronmk
            if match:
432
                value, name = match.groups()
433
                raise FunctionValueException(name, strings.to_unicode(value), e)
434
435 2526 aaronmk
            match = re.search(r'column "(.+?)" is of type (.+?) but expression '
436 2523 aaronmk
                r'is of type', msg)
437
            if match:
438
                col, type_ = match.groups()
439
                raise MissingCastException(type_, col, e)
440
441 2493 aaronmk
            match = re.search(r'relation "(.+?)" already exists', msg)
442 2464 aaronmk
            if match: raise DuplicateTableException(match.group(1), e)
443
444 2493 aaronmk
            match = re.search(r'function "(.+?)" already exists', msg)
445 2464 aaronmk
            if match: raise DuplicateFunctionException(match.group(1), e)
446
447
            raise # no specific exception raised
448
    except log_ignore_excs:
449
        log_level += 2
450
        raise
451
    finally:
452 2666 aaronmk
        if debug_msg_ref != None and debug_msg_ref[0] != None:
453
            db.log_debug(debug_msg_ref[0], log_level)
454 830 aaronmk
455 832 aaronmk
##### Basic queries
456
457 2153 aaronmk
def next_version(name):
458 2163 aaronmk
    version = 1 # first existing name was version 0
459 2586 aaronmk
    match = re.match(r'^(.*)#(\d+)$', name)
460 2153 aaronmk
    if match:
461 2586 aaronmk
        name, version = match.groups()
462
        version = int(version)+1
463 2588 aaronmk
    return sql_gen.add_suffix(name, '#'+str(version))
464 2153 aaronmk
465 2899 aaronmk
def lock_table(db, table, mode):
466
    table = sql_gen.as_Table(table)
467
    run_query(db, 'LOCK TABLE '+table.to_str(db)+' IN '+mode+' MODE')
468
469 2789 aaronmk
def run_query_into(db, query, into=None, add_indexes_=False, **kw_args):
470 2085 aaronmk
    '''Outputs a query to a temp table.
471
    For params, see run_query().
472
    '''
473 2789 aaronmk
    if into == None: return run_query(db, query, **kw_args)
474 2790 aaronmk
475
    assert isinstance(into, sql_gen.Table)
476
477
    kw_args['recover'] = True
478
    kw_args.setdefault('log_ignore_excs', (DuplicateTableException,))
479
480
    temp = not db.autocommit # tables are permanent in autocommit mode
481
    # "temporary tables cannot specify a schema name", so remove schema
482
    if temp: into.schema = None
483
484
    # Create table
485
    while True:
486
        create_query = 'CREATE'
487
        if temp: create_query += ' TEMP'
488
        create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
489 2385 aaronmk
490 2790 aaronmk
        try:
491
            cur = run_query(db, create_query, **kw_args)
492
                # CREATE TABLE AS sets rowcount to # rows in query
493
            break
494
        except DuplicateTableException, e:
495
            into.name = next_version(into.name)
496
            # try again with next version of name
497
498
    if add_indexes_: add_indexes(db, into)
499
500
    return cur
501 2085 aaronmk
502 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
503
504 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
505
506 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
507 2293 aaronmk
    start=None, order_by=order_by_pkey, default_table=None):
508 1981 aaronmk
    '''
509 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
510 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
511 1981 aaronmk
    @param fields Use None to select all fields in the table
512 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
513 2379 aaronmk
        * container can be any iterable type
514 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
515
        * compare_right_side: sql_gen.ValueCond|literal value
516 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
517
        use all columns
518 2786 aaronmk
    @return query
519 1981 aaronmk
    '''
520 2315 aaronmk
    # Parse tables param
521 2121 aaronmk
    if not lists.is_seq(tables): tables = [tables]
522 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
523 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
524 2121 aaronmk
525 2315 aaronmk
    # Parse other params
526 2376 aaronmk
    if conds == None: conds = []
527 2650 aaronmk
    elif dicts.is_dict(conds): conds = conds.items()
528 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
529 135 aaronmk
    assert limit == None or type(limit) == int
530 865 aaronmk
    assert start == None or type(start) == int
531 2315 aaronmk
    if order_by is order_by_pkey:
532
        if distinct_on != []: order_by = None
533
        else: order_by = pkey(db, table0, recover=True)
534 865 aaronmk
535 2315 aaronmk
    query = 'SELECT'
536 2056 aaronmk
537 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
538 2056 aaronmk
539 2200 aaronmk
    # DISTINCT ON columns
540 2233 aaronmk
    if distinct_on != []:
541 2467 aaronmk
        query += '\nDISTINCT'
542 2254 aaronmk
        if distinct_on is not distinct_on_all:
543 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
544
545
    # Columns
546 2467 aaronmk
    query += '\n'
547 1135 aaronmk
    if fields == None: query += '*'
548 2765 aaronmk
    else:
549
        assert fields != []
550
        query += '\n, '.join(map(parse_col, fields))
551 2200 aaronmk
552
    # Main table
553 2467 aaronmk
    query += '\nFROM '+table0.to_str(db)
554 865 aaronmk
555 2122 aaronmk
    # Add joins
556 2271 aaronmk
    left_table = table0
557 2263 aaronmk
    for join_ in tables:
558
        table = join_.table
559 2238 aaronmk
560 2343 aaronmk
        # Parse special values
561
        if join_.type_ is sql_gen.filter_out: # filter no match
562 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
563 2853 aaronmk
                sql_gen.CompareCond(None, '~=')))
564 2343 aaronmk
565 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
566 2122 aaronmk
567
        left_table = table
568
569 865 aaronmk
    missing = True
570 2376 aaronmk
    if conds != []:
571 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
572
        else: whitespace = '\n'
573 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
574
            .to_str(db) for l, r in conds], 'WHERE')
575 865 aaronmk
        missing = False
576 2227 aaronmk
    if order_by != None:
577 2467 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
578
    if limit != None: query += '\nLIMIT '+str(limit); missing = False
579 865 aaronmk
    if start != None:
580 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
581 865 aaronmk
        missing = False
582
    if missing: warnings.warn(DbWarning(
583
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
584
585 2786 aaronmk
    return query
586 11 aaronmk
587 2054 aaronmk
def select(db, *args, **kw_args):
588
    '''For params, see mk_select() and run_query()'''
589
    recover = kw_args.pop('recover', None)
590
    cacheable = kw_args.pop('cacheable', True)
591 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
592 2054 aaronmk
593 2791 aaronmk
    return run_query(db, mk_select(db, *args, **kw_args), recover, cacheable,
594
        log_level=log_level)
595 2054 aaronmk
596 2788 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, returning=None,
597
    embeddable=False):
598 1960 aaronmk
    '''
599
    @param returning str|None An inserted column (such as pkey) to return
600 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
601 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
602
        query will be fully cached, not just if it raises an exception.
603 1960 aaronmk
    '''
604 2754 aaronmk
    table = sql_gen.remove_table_rename(sql_gen.as_Table(table))
605 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
606 2646 aaronmk
    if cols != None:
607
        cols = [sql_gen.to_name_only_col(v, table).to_str(db) for v in cols]
608 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
609 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
610 2063 aaronmk
611
    # Build query
612 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
613
    query = first_line
614 2467 aaronmk
    if cols != None: query += '\n('+', '.join(cols)+')'
615
    query += '\n'+select_query
616 2063 aaronmk
617
    if returning != None:
618 2743 aaronmk
        query += '\nRETURNING '+sql_gen.to_name_only_col(returning).to_str(db)
619 2063 aaronmk
620 2070 aaronmk
    if embeddable:
621 2327 aaronmk
        assert returning != None
622
623 2070 aaronmk
        # Create function
624 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
625 2327 aaronmk
        return_type = 'SETOF '+returning.to_str(db)+'%TYPE'
626 2189 aaronmk
        while True:
627
            try:
628 2698 aaronmk
                function = sql_gen.TempFunction(function_name, db.autocommit)
629 2194 aaronmk
630 2189 aaronmk
                function_query = '''\
631 2698 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''()
632 2467 aaronmk
RETURNS '''+return_type+'''
633
LANGUAGE sql
634
AS $$
635 2788 aaronmk
'''+query+''';
636 2467 aaronmk
$$;
637 2070 aaronmk
'''
638 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
639 2464 aaronmk
                    log_ignore_excs=(DuplicateFunctionException,))
640 2189 aaronmk
                break # this version was successful
641
            except DuplicateFunctionException, e:
642
                function_name = next_version(function_name)
643
                # try again with next version of name
644 2070 aaronmk
645 2337 aaronmk
        # Return query that uses function
646 2698 aaronmk
        func_table = sql_gen.NamedTable('f', sql_gen.FunctionCall(function),
647 2742 aaronmk
            [returning]) # AS clause requires function alias
648 2787 aaronmk
        return mk_select(db, func_table, start=0, order_by=None)
649 2070 aaronmk
650 2787 aaronmk
    return query
651 2066 aaronmk
652
def insert_select(db, *args, **kw_args):
653 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
654 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
655
        values in
656 2072 aaronmk
    '''
657 2386 aaronmk
    into = kw_args.pop('into', None)
658
    if into != None: kw_args['embeddable'] = True
659 2066 aaronmk
    recover = kw_args.pop('recover', None)
660
    cacheable = kw_args.pop('cacheable', True)
661 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
662 2066 aaronmk
663 2789 aaronmk
    return run_query_into(db, mk_insert_select(db, *args, **kw_args), into,
664 2787 aaronmk
        recover=recover, cacheable=cacheable, log_level=log_level)
665 2063 aaronmk
666 2738 aaronmk
default = sql_gen.default # tells insert() to use the default value for a column
667 2066 aaronmk
668 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
669 2085 aaronmk
    '''For params, see insert_select()'''
670 1960 aaronmk
    if lists.is_seq(row): cols = None
671
    else:
672
        cols = row.keys()
673
        row = row.values()
674 2738 aaronmk
    row = list(row) # ensure that "== []" works
675 1960 aaronmk
676 2738 aaronmk
    if row == []: query = None
677
    else: query = sql_gen.Values(row).to_str(db)
678 1961 aaronmk
679 2788 aaronmk
    return insert_select(db, table, cols, query, *args, **kw_args)
680 11 aaronmk
681 2402 aaronmk
def mk_update(db, table, changes=None, cond=None):
682
    '''
683
    @param changes [(col, new_value),...]
684
        * container can be any iterable type
685
        * col: sql_gen.Code|str (for col name)
686
        * new_value: sql_gen.Code|literal value
687
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
688
    @return str query
689
    '''
690
    query = 'UPDATE '+sql_gen.as_Table(table).to_str(db)+'\nSET\n'
691 2405 aaronmk
    query += ',\n'.join((sql_gen.to_name_only_col(col, table).to_str(db)+' = '
692 2402 aaronmk
        +sql_gen.as_Value(new_value).to_str(db) for col, new_value in changes))
693 2467 aaronmk
    if cond != None: query += '\nWHERE\n'+cond.to_str(db)
694 2402 aaronmk
695
    return query
696
697
def update(db, *args, **kw_args):
698
    '''For params, see mk_update() and run_query()'''
699
    recover = kw_args.pop('recover', None)
700
701 2791 aaronmk
    return run_query(db, mk_update(db, *args, **kw_args), recover)
702 2402 aaronmk
703 135 aaronmk
def last_insert_id(db):
704 1849 aaronmk
    module = util.root_module(db.db)
705 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
706
    elif module == 'MySQLdb': return db.insert_id()
707
    else: return None
708 13 aaronmk
709 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
710 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
711 2415 aaronmk
    to names that will be distinct among the columns' tables.
712 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
713 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
714
    @param into The table for the new columns.
715 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
716
        columns will be included in the mapping even if they are not in cols.
717
        The tables of the provided Col objects will be changed to into, so make
718
        copies of them if you want to keep the original tables.
719
    @param as_items Whether to return a list of dict items instead of a dict
720 2383 aaronmk
    @return dict(orig_col=new_col, ...)
721
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
722 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
723
        * All mappings use the into table so its name can easily be
724 2383 aaronmk
          changed for all columns at once
725
    '''
726 2415 aaronmk
    cols = lists.uniqify(cols)
727
728 2394 aaronmk
    items = []
729 2389 aaronmk
    for col in preserve:
730 2390 aaronmk
        orig_col = copy.copy(col)
731 2392 aaronmk
        col.table = into
732 2394 aaronmk
        items.append((orig_col, col))
733
    preserve = set(preserve)
734
    for col in cols:
735 2716 aaronmk
        if col not in preserve:
736
            items.append((col, sql_gen.Col(str(col), into, col.srcs)))
737 2394 aaronmk
738
    if not as_items: items = dict(items)
739
    return items
740 2383 aaronmk
741 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
742 2391 aaronmk
    '''For params, see mk_flatten_mapping()
743
    @return See return value of mk_flatten_mapping()
744
    '''
745 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
746
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
747 2786 aaronmk
    run_query_into(db, mk_select(db, joins, cols, limit=limit, start=start),
748 2846 aaronmk
        into=into, add_indexes_=True)
749 2394 aaronmk
    return dict(items)
750 2391 aaronmk
751 2751 aaronmk
def mk_track_data_error(db, errors_table, cols, value, error_code, error):
752 2810 aaronmk
    assert cols != ()
753
754 2751 aaronmk
    cols = map(sql_gen.to_name_only_col, cols)
755
756
    columns_cols = ['column']
757
    columns = sql_gen.NamedValues('columns', columns_cols,
758
        [[c.name] for c in cols])
759
    values_cols = ['value', 'error_code', 'error']
760
    values = sql_gen.NamedValues('values', values_cols,
761
        [value, error_code, error])
762
763
    select_cols = columns_cols+values_cols
764
    name_only_cols = map(sql_gen.to_name_only_col, select_cols)
765 2755 aaronmk
    errors_table = sql_gen.NamedTable('errors', errors_table)
766 2751 aaronmk
    joins = [columns, sql_gen.Join(values, type_='CROSS'),
767
        sql_gen.Join(errors_table, dict(zip(name_only_cols, select_cols)),
768
        sql_gen.filter_out)]
769
770
    return mk_insert_select(db, errors_table, name_only_cols,
771 2787 aaronmk
        mk_select(db, joins, select_cols, order_by=None))
772 2751 aaronmk
773 2810 aaronmk
def track_data_error(db, errors_table, cols, *args, **kw_args):
774 2758 aaronmk
    '''
775
    @param errors_table If None, does nothing.
776
    '''
777 2810 aaronmk
    if errors_table == None or cols == (): return
778
    run_query(db, mk_track_data_error(db, errors_table, cols, *args, **kw_args),
779 2758 aaronmk
        cacheable=True, log_level=4)
780
781 2704 aaronmk
def cast(db, type_, col, errors_table=None):
782
    '''Casts an (unrenamed) column or value.
783 2769 aaronmk
    If errors_table set and col has srcs, saves errors in errors_table (using
784
    col's srcs attr as the source columns) and converts errors to warnings.
785 2820 aaronmk
    @param col str|sql_gen.Col|sql_gen.Literal
786 2704 aaronmk
    @param errors_table None|sql_gen.Table|str
787
    '''
788 2820 aaronmk
    col = sql_gen.as_Col(col)
789 2769 aaronmk
    save_errors = (errors_table != None and isinstance(col, sql_gen.Col)
790
        and col.srcs != ())
791
    if not save_errors: # can't save errors
792
        return sql_gen.CustomCode(col.to_str(db)+'::'+type_) # just cast
793 2704 aaronmk
794
    assert not isinstance(col, sql_gen.NamedCol)
795
796 2770 aaronmk
    errors_table = sql_gen.as_Table(errors_table)
797
    srcs = map(sql_gen.to_name_only_col, col.srcs)
798
    function_name = str(sql_gen.FunctionCall(type_, *srcs))
799 2704 aaronmk
    function = sql_gen.TempFunction(function_name, db.autocommit)
800
801 2718 aaronmk
    while True:
802
        # Create function definition
803
        query = '''\
804 2704 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''(value text)
805
RETURNS '''+type_+'''
806
LANGUAGE plpgsql
807
STRICT
808
AS $$
809
BEGIN
810
    /* The explicit cast to the return type is needed to make the cast happen
811
    inside the try block. (Implicit casts to the return type happen at the end
812
    of the function, outside any block.) */
813
    RETURN value::'''+type_+''';
814
EXCEPTION
815
    WHEN data_exception THEN
816
        -- Save error in errors table.
817 2751 aaronmk
        -- Insert the value and error for *each* source column.
818
'''+mk_track_data_error(db, errors_table, srcs,
819
    *map(sql_gen.CustomCode, ['value', 'SQLSTATE', 'SQLERRM']))+''';
820 2704 aaronmk
821
        RAISE WARNING '%', SQLERRM;
822
        RETURN NULL;
823
END;
824
$$;
825
'''
826 2718 aaronmk
827
        # Create function
828
        try:
829
            run_query(db, query, recover=True, cacheable=True,
830
                log_ignore_excs=(DuplicateFunctionException,))
831
            break # successful
832
        except DuplicateFunctionException:
833 2770 aaronmk
            function.name = next_version(function.name)
834
            # try again with next version of name
835 2704 aaronmk
836
    return sql_gen.FunctionCall(function, col)
837
838 2414 aaronmk
##### Database structure queries
839
840 2426 aaronmk
def table_row_count(db, table, recover=None):
841 2786 aaronmk
    return value(run_query(db, mk_select(db, table, [sql_gen.row_count],
842 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
843 2426 aaronmk
844 2414 aaronmk
def table_cols(db, table, recover=None):
845
    return list(col_names(select(db, table, limit=0, order_by=None,
846 2443 aaronmk
        recover=recover, log_level=4)))
847 2414 aaronmk
848 2291 aaronmk
def pkey(db, table, recover=None):
849 832 aaronmk
    '''Assumed to be first column in table'''
850 2339 aaronmk
    return table_cols(db, table, recover)[0]
851 832 aaronmk
852 2559 aaronmk
not_null_col = 'not_null_col'
853 2340 aaronmk
854
def table_not_null_col(db, table, recover=None):
855
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
856
    if not_null_col in table_cols(db, table, recover): return not_null_col
857
    else: return pkey(db, table, recover)
858
859 853 aaronmk
def index_cols(db, table, index):
860
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
861
    automatically created. When you don't know whether something is a UNIQUE
862
    constraint or a UNIQUE index, use this function.'''
863 1909 aaronmk
    module = util.root_module(db.db)
864
    if module == 'psycopg2':
865
        return list(values(run_query(db, '''\
866 853 aaronmk
SELECT attname
867 866 aaronmk
FROM
868
(
869
        SELECT attnum, attname
870
        FROM pg_index
871
        JOIN pg_class index ON index.oid = indexrelid
872
        JOIN pg_class table_ ON table_.oid = indrelid
873
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
874
        WHERE
875 2782 aaronmk
            table_.relname = '''+db.esc_value(table)+'''
876
            AND index.relname = '''+db.esc_value(index)+'''
877 866 aaronmk
    UNION
878
        SELECT attnum, attname
879
        FROM
880
        (
881
            SELECT
882
                indrelid
883
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
884
                    AS indkey
885
            FROM pg_index
886
            JOIN pg_class index ON index.oid = indexrelid
887
            JOIN pg_class table_ ON table_.oid = indrelid
888
            WHERE
889 2782 aaronmk
                table_.relname = '''+db.esc_value(table)+'''
890
                AND index.relname = '''+db.esc_value(index)+'''
891 866 aaronmk
        ) s
892
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
893
) s
894 853 aaronmk
ORDER BY attnum
895 2782 aaronmk
'''
896
            , cacheable=True, log_level=4)))
897 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
898
        ' database')
899 853 aaronmk
900 464 aaronmk
def constraint_cols(db, table, constraint):
901 1849 aaronmk
    module = util.root_module(db.db)
902 464 aaronmk
    if module == 'psycopg2':
903
        return list(values(run_query(db, '''\
904
SELECT attname
905
FROM pg_constraint
906
JOIN pg_class ON pg_class.oid = conrelid
907
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
908
WHERE
909 2783 aaronmk
    relname = '''+db.esc_value(table)+'''
910
    AND conname = '''+db.esc_value(constraint)+'''
911 464 aaronmk
ORDER BY attnum
912 2783 aaronmk
'''
913
            )))
914 464 aaronmk
    else: raise NotImplementedError("Can't list constraint columns for "+module+
915
        ' database')
916
917 2096 aaronmk
row_num_col = '_row_num'
918
919 2847 aaronmk
def add_index(db, exprs, table=None, unique=False, ensure_not_null=True):
920 2688 aaronmk
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
921 2538 aaronmk
    Currently, only function calls are supported as expressions.
922 2847 aaronmk
    @param ensure_not_null If set, translates NULL values to sentinel values.
923
        This allows indexes to be used for comparisons where NULLs are equal.
924 2538 aaronmk
    '''
925 2688 aaronmk
    if not lists.is_seq(exprs): exprs = [exprs]
926 2538 aaronmk
927 2688 aaronmk
    # Parse exprs
928
    old_exprs = exprs[:]
929
    exprs = []
930
    cols = []
931
    for i, expr in enumerate(old_exprs):
932
        expr = copy.deepcopy(expr) # don't modify input!
933 2823 aaronmk
        expr = sql_gen.as_Col(expr, table)
934 2688 aaronmk
935 2823 aaronmk
        # Handle nullable columns
936 2860 aaronmk
        if ensure_not_null:
937
            try: expr = sql_gen.ensure_not_null(db, expr)
938
            except KeyError: pass # unknown type, so just create plain index
939 2823 aaronmk
940 2688 aaronmk
        # Extract col
941
        if isinstance(expr, sql_gen.FunctionCall):
942
            col = expr.args[0]
943
            expr = sql_gen.Expr(expr)
944
        else: col = expr
945 2823 aaronmk
        assert isinstance(col, sql_gen.Col)
946 2688 aaronmk
947
        # Extract table
948
        if table == None:
949
            assert sql_gen.is_table_col(col)
950
            table = col.table
951
952
        col.table = None
953
954
        exprs.append(expr)
955
        cols.append(col)
956 2408 aaronmk
957 2688 aaronmk
    table = sql_gen.as_Table(table)
958
    index = sql_gen.Table(str(sql_gen.Col(','.join(map(str, cols)), table)))
959
960
    str_ = 'CREATE'
961
    if unique: str_ += ' UNIQUE'
962
    str_ += ' INDEX '+index.to_str(db)+' ON '+table.to_str(db)+' ('+(
963
        ', '.join((v.to_str(db) for v in exprs)))+')'
964
965
    try: run_query(db, str_, recover=True, cacheable=True, log_level=3)
966 2408 aaronmk
    except DuplicateTableException: pass # index already existed
967
968 2686 aaronmk
def add_pkey(db, table, cols=None, recover=None):
969
    '''Adds a primary key.
970
    @param cols [sql_gen.Col,...] The columns in the primary key.
971
        Defaults to the first column in the table.
972 2406 aaronmk
    @pre The table must not already have a primary key.
973
    '''
974
    table = sql_gen.as_Table(table)
975 2686 aaronmk
    if cols == None: cols = [pkey(db, table, recover)]
976
    col_strs = [sql_gen.to_name_only_col(v).to_str(db) for v in cols]
977 2406 aaronmk
978 2781 aaronmk
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD PRIMARY KEY ('
979
        +(', '.join(col_strs))+')', recover=True, cacheable=True, log_level=3,
980
        log_ignore_excs=(DuplicateTableException,))
981 2406 aaronmk
982 2772 aaronmk
already_indexed = object() # tells add_indexes() the pkey has already been added
983
984
def add_indexes(db, table, has_pkey=True):
985
    '''Adds an index on all columns in a table.
986
    @param has_pkey bool|already_indexed Whether a pkey instead of a regular
987
        index should be added on the first column.
988
        * If already_indexed, the pkey is assumed to have already been added
989
    '''
990
    cols = table_cols(db, table)
991
    if has_pkey:
992
        if has_pkey is not already_indexed: add_pkey(db, table)
993
        cols = cols[1:]
994
    for col in cols: add_index(db, col, table)
995
996 2086 aaronmk
def add_row_num(db, table):
997 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
998
    be the primary key.'''
999 2320 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
1000 2096 aaronmk
    run_query(db, 'ALTER TABLE '+table+' ADD COLUMN '+row_num_col
1001 2443 aaronmk
        +' serial NOT NULL PRIMARY KEY', log_level=3)
1002 2086 aaronmk
1003 2889 aaronmk
def drop_table(db, table):
1004
    table = sql_gen.as_Table(table)
1005
    return run_query(db, 'DROP TABLE IF EXISTS '+table.to_str(db)+' CASCADE')
1006
1007 2760 aaronmk
def create_table(db, table, cols, has_pkey=True, col_indexes=True):
1008 2675 aaronmk
    '''Creates a table.
1009 2681 aaronmk
    @param cols [sql_gen.TypedCol,...] The column names and types
1010
    @param has_pkey If set, the first column becomes the primary key.
1011 2760 aaronmk
    @param col_indexes bool|[ref]
1012
        * If True, indexes will be added on all non-pkey columns.
1013
        * If a list reference, [0] will be set to a function to do this.
1014
          This can be used to delay index creation until the table is populated.
1015 2675 aaronmk
    '''
1016
    table = sql_gen.as_Table(table)
1017
1018 2681 aaronmk
    if has_pkey:
1019
        cols[0] = pkey = copy.copy(cols[0]) # don't modify input!
1020 2872 aaronmk
        pkey.constraints = 'PRIMARY KEY'
1021 2681 aaronmk
1022 2675 aaronmk
    str_ = 'CREATE TABLE '+table.to_str(db)+' (\n'
1023 2681 aaronmk
    str_ += '\n, '.join(v.to_str(db) for v in cols)
1024
    str_ += '\n);\n'
1025 2675 aaronmk
    run_query(db, str_, cacheable=True, log_level=2)
1026 2760 aaronmk
1027
    # Add indexes
1028 2773 aaronmk
    if has_pkey: has_pkey = already_indexed
1029
    def add_indexes_(): add_indexes(db, table, has_pkey)
1030
    if isinstance(col_indexes, list): col_indexes[0] = add_indexes_ # defer
1031
    elif col_indexes: add_indexes_() # add now
1032 2675 aaronmk
1033 2684 aaronmk
def vacuum(db, table):
1034
    table = sql_gen.as_Table(table)
1035
    db.with_autocommit(lambda: run_query(db, 'VACUUM ANALYZE '+table.to_str(db),
1036
        log_level=3))
1037
1038 2732 aaronmk
def truncate(db, table, schema='public'):
1039 2777 aaronmk
    table = sql_gen.as_Table(table, schema)
1040 2733 aaronmk
    return run_query(db, 'TRUNCATE '+table.to_str(db)+' CASCADE')
1041 2732 aaronmk
1042 2729 aaronmk
def tables(db, schema_like='public', table_like='%', exact=False):
1043
    if exact: compare = '='
1044
    else: compare = 'LIKE'
1045
1046 1849 aaronmk
    module = util.root_module(db.db)
1047 832 aaronmk
    if module == 'psycopg2':
1048 2779 aaronmk
        conds = [('schemaname', sql_gen.CompareCond(schema_like, compare)),
1049
            ('tablename', sql_gen.CompareCond(table_like, compare))]
1050
        return values(select(db, 'pg_tables', ['tablename'], conds,
1051
            order_by='tablename', log_level=4))
1052 1968 aaronmk
    elif module == 'MySQLdb':
1053 2779 aaronmk
        return values(run_query(db, 'SHOW TABLES LIKE '+db.esc_value(table_like)
1054
            , cacheable=True, log_level=4))
1055 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
1056 830 aaronmk
1057 2726 aaronmk
def table_exists(db, table):
1058
    table = sql_gen.as_Table(table)
1059 2730 aaronmk
    return list(tables(db, table.schema, table.name, exact=True)) != []
1060 2726 aaronmk
1061 2913 aaronmk
def function_exists(db, function):
1062
    function = sql_gen.as_Function(function)
1063
1064
    info_table = sql_gen.Table('routines', 'information_schema')
1065
    conds = [('routine_name', function.name)]
1066
    schema = function.schema
1067
    if schema != None: conds.append(('routine_schema', schema))
1068
1069
    return list(values(select(db, info_table, ['routine_name'], conds,
1070
        order_by='routine_schema', limit=1, log_level=4))) != []
1071
        # TODO: order_by search_path schema order
1072
1073 2726 aaronmk
def errors_table(db, table, if_exists=True):
1074
    '''
1075
    @param if_exists If set, returns None if the errors table doesn't exist
1076
    @return None|sql_gen.Table
1077
    '''
1078
    table = sql_gen.as_Table(table)
1079
    if table.srcs != (): table = table.srcs[0]
1080
1081
    errors_table = sql_gen.suffixed_table(table, '.errors')
1082
    if if_exists and not table_exists(db, errors_table): return None
1083
    return errors_table
1084
1085 833 aaronmk
##### Database management
1086
1087 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
1088
    '''For kw_args, see tables()'''
1089
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
1090 833 aaronmk
1091 832 aaronmk
##### Heuristic queries
1092
1093 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
1094 1554 aaronmk
    '''Recovers from errors.
1095 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
1096
    '''
1097 2642 aaronmk
    row = sql_gen.ColDict(db, table, row)
1098 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
1099
1100 471 aaronmk
    try:
1101 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
1102 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
1103
            row_ct_ref[0] += cur.rowcount
1104
        return value(cur)
1105 471 aaronmk
    except DuplicateKeyException, e:
1106 2653 aaronmk
        row = sql_gen.ColDict(db, table,
1107
            util.dict_subset_right_join(row, e.cols))
1108
        return value(select(db, table, [pkey_], row, recover=True))
1109 471 aaronmk
1110 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
1111 830 aaronmk
    '''Recovers from errors'''
1112 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
1113 14 aaronmk
    except StopIteration:
1114 40 aaronmk
        if not create: raise
1115 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
1116 2078 aaronmk
1117 2593 aaronmk
def is_func_result(col):
1118
    return col.table.name.find('(') >= 0 and col.name == 'result'
1119
1120 2592 aaronmk
def into_table_name(out_table, in_tables0, mapping, is_func):
1121 2615 aaronmk
    def in_col_str(in_col):
1122
        in_col = sql_gen.remove_col_rename(in_col)
1123
        if isinstance(in_col, sql_gen.Col):
1124
            table = in_col.table
1125
            if table == in_tables0:
1126
                in_col = sql_gen.to_name_only_col(in_col)
1127
            elif is_func_result(in_col): in_col = table # omit col name
1128
        return str(in_col)
1129
1130 2580 aaronmk
    str_ = str(out_table)
1131
    if is_func:
1132 2615 aaronmk
        str_ += '('
1133 2580 aaronmk
1134 2615 aaronmk
        try: value_in_col = mapping['value']
1135
        except KeyError:
1136 2654 aaronmk
            str_ += ', '.join((str(k)+'='+in_col_str(v)
1137 2615 aaronmk
                for k, v in mapping.iteritems()))
1138
        else: str_ += in_col_str(value_in_col)
1139
1140
        str_ += ')'
1141 2812 aaronmk
    else:
1142
        out_col = 'rank'
1143
        try: in_col = mapping[out_col]
1144
        except KeyError: str_ += '_pkeys'
1145
        else: # has a rank column, so hierarchical
1146
            str_ += '['+str(out_col)+'='+in_col_str(in_col)+']'
1147 2580 aaronmk
    return str_
1148
1149 2508 aaronmk
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None, into=None,
1150 2805 aaronmk
    default=None, is_func=False, on_error=exc.raise_):
1151 2078 aaronmk
    '''Recovers from errors.
1152
    Only works under PostgreSQL (uses INSERT RETURNING).
1153 2909 aaronmk
    IMPORTANT: Must be run at the *beginning* of a transaction.
1154 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
1155
        tables to join with it using the main input table's pkey
1156 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
1157 2612 aaronmk
        * out_table_col: str (*not* sql_gen.Col)
1158
        * in_table_col: sql_gen.Col|literal-value
1159 2489 aaronmk
    @param into The table to contain the output and input pkeys.
1160 2574 aaronmk
        Defaults to `out_table.name+'_pkeys'`.
1161 2509 aaronmk
    @param default The *output* column to use as the pkey for missing rows.
1162
        If this output column does not exist in the mapping, uses None.
1163 2552 aaronmk
    @param is_func Whether out_table is the name of a SQL function, not a table
1164 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
1165 2078 aaronmk
    '''
1166 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
1167 2312 aaronmk
1168 2450 aaronmk
    def log_debug(msg): db.log_debug(msg, level=1.5)
1169 2505 aaronmk
    def col_ustr(str_):
1170 2567 aaronmk
        return strings.repr_no_u(sql_gen.remove_col_rename(str_))
1171 2450 aaronmk
1172 2910 aaronmk
    log_debug('********** New iteration **********')
1173
    log_debug('Inserting these input columns into '+strings.as_tt(
1174
        out_table.to_str(db))+':\n'+strings.as_table(mapping, ustr=col_ustr))
1175
1176
    log_debug('Locking output table to prevent concurrent duplicate keys')
1177
    # Must happen before any SELECT query on the table to avoid lock upgrades
1178
    lock_table(db, out_table, 'EXCLUSIVE')
1179
1180 2768 aaronmk
    out_pkey = pkey(db, out_table, recover=True)
1181
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
1182
1183
    if mapping == {}: # need at least one column for INSERT SELECT
1184
        mapping = {out_pkey: None} # ColDict will replace with default value
1185
1186 2382 aaronmk
    # Create input joins from list of input tables
1187
    in_tables_ = in_tables[:] # don't modify input!
1188
    in_tables0 = in_tables_.pop(0) # first table is separate
1189 2735 aaronmk
    errors_table_ = errors_table(db, in_tables0)
1190 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
1191 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
1192 2460 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v,
1193
        {in_pkey: sql_gen.join_same_not_null}) for v in in_tables_]
1194 2131 aaronmk
1195 2592 aaronmk
    if into == None:
1196
        into = into_table_name(out_table, in_tables0, mapping, is_func)
1197
    into = sql_gen.as_Table(into)
1198
1199 2702 aaronmk
    # Set column sources
1200
    in_cols = filter(sql_gen.is_table_col, mapping.values())
1201
    for col in in_cols:
1202
        if col.table == in_tables0: col.set_srcs(sql_gen.src_self)
1203
1204 2486 aaronmk
    log_debug('Joining together input tables into temp table')
1205 2395 aaronmk
    # Place in new table for speed and so don't modify input if values edited
1206 2584 aaronmk
    in_table = sql_gen.Table('in')
1207 2702 aaronmk
    mapping = dicts.join(mapping, flatten(db, in_table, input_joins, in_cols,
1208
        preserve=[in_pkey_col], start=0))
1209 2395 aaronmk
    input_joins = [in_table]
1210 2486 aaronmk
    db.log_debug('Temp table: '+strings.as_tt(in_table.to_str(db)), level=2)
1211 2395 aaronmk
1212 2656 aaronmk
    mapping = sql_gen.ColDict(db, out_table, mapping)
1213
        # after applying dicts.join() because that returns a plain dict
1214
1215 2509 aaronmk
    # Resolve default value column
1216 2841 aaronmk
    if default != None:
1217
        try: default = mapping[default]
1218
        except KeyError:
1219 2509 aaronmk
            db.log_debug('Default value column '
1220
                +strings.as_tt(strings.repr_no_u(default))
1221 2511 aaronmk
                +' does not exist in mapping, falling back to None', level=2.1)
1222 2509 aaronmk
            default = None
1223
1224 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
1225 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
1226
1227 2201 aaronmk
    pkeys_table_exists_ref = [False]
1228 2420 aaronmk
    def insert_into_pkeys(joins, cols):
1229 2786 aaronmk
        query = mk_select(db, joins, cols, order_by=None, start=0)
1230 2201 aaronmk
        if pkeys_table_exists_ref[0]:
1231 2786 aaronmk
            insert_select(db, into, pkeys_names, query)
1232 2201 aaronmk
        else:
1233 2786 aaronmk
            run_query_into(db, query, into=into)
1234 2201 aaronmk
            pkeys_table_exists_ref[0] = True
1235
1236 2429 aaronmk
    limit_ref = [None]
1237 2380 aaronmk
    conds = set()
1238 2811 aaronmk
    distinct_on = sql_gen.ColDict(db, out_table)
1239 2325 aaronmk
    def mk_main_select(joins, cols):
1240 2811 aaronmk
        distinct_on_cols = [c.to_Col() for c in distinct_on.values()]
1241
        return mk_select(db, joins, cols, conds, distinct_on_cols,
1242 2429 aaronmk
            limit=limit_ref[0], start=0)
1243 2132 aaronmk
1244 2519 aaronmk
    exc_strs = set()
1245 2309 aaronmk
    def log_exc(e):
1246 2519 aaronmk
        e_str = exc.str_(e, first_line_only=True)
1247
        log_debug('Caught exception: '+e_str)
1248
        assert e_str not in exc_strs # avoid infinite loops
1249
        exc_strs.add(e_str)
1250 2735 aaronmk
1251 2451 aaronmk
    def remove_all_rows():
1252 2896 aaronmk
        log_debug('Ignoring all rows')
1253 2429 aaronmk
        limit_ref[0] = 0 # just create an empty pkeys table
1254 2735 aaronmk
1255 2758 aaronmk
    def ignore(in_col, value, e):
1256
        track_data_error(db, errors_table_, in_col.srcs, value, e.cause.pgcode,
1257
            e.cause.pgerror)
1258 2895 aaronmk
        log_debug('Ignoring rows with '+strings.as_tt(repr(in_col))+' = '
1259 2545 aaronmk
            +strings.as_tt(repr(value)))
1260 2895 aaronmk
1261 2758 aaronmk
    def remove_rows(in_col, value, e):
1262
        ignore(in_col, value, e)
1263 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
1264
        assert cond not in conds # avoid infinite loops
1265 2380 aaronmk
        conds.add(cond)
1266 2895 aaronmk
1267 2758 aaronmk
    def invalid2null(in_col, value, e):
1268
        ignore(in_col, value, e)
1269 2403 aaronmk
        update(db, in_table, [(in_col, None)],
1270
            sql_gen.ColValueCond(in_col, value))
1271 2245 aaronmk
1272 2589 aaronmk
    def insert_pkeys_table(which):
1273
        return sql_gen.Table(sql_gen.add_suffix(in_table.name,
1274
            '_insert_'+which+'_pkeys'))
1275
    insert_out_pkeys = insert_pkeys_table('out')
1276
    insert_in_pkeys = insert_pkeys_table('in')
1277
1278 2206 aaronmk
    # Do inserts and selects
1279 2642 aaronmk
    join_cols = sql_gen.ColDict(db, out_table)
1280 2206 aaronmk
    while True:
1281 2521 aaronmk
        if limit_ref[0] == 0: # special case
1282
            log_debug('Creating an empty pkeys table')
1283 2786 aaronmk
            cur = run_query_into(db, mk_select(db, out_table, [out_pkey],
1284 2521 aaronmk
                limit=limit_ref[0]), into=insert_out_pkeys)
1285
            break # don't do main case
1286
1287 2303 aaronmk
        has_joins = join_cols != {}
1288
1289 2903 aaronmk
        log_debug('Trying to insert new rows')
1290
1291 2305 aaronmk
        # Prepare to insert new rows
1292 2325 aaronmk
        insert_joins = input_joins[:] # don't modify original!
1293 2403 aaronmk
        insert_args = dict(recover=True, cacheable=False)
1294 2303 aaronmk
        if has_joins:
1295 2325 aaronmk
            insert_joins.append(sql_gen.Join(out_table, join_cols,
1296
                sql_gen.filter_out))
1297
        else:
1298 2404 aaronmk
            insert_args.update(dict(returning=out_pkey, into=insert_out_pkeys))
1299 2786 aaronmk
        main_select = mk_main_select(insert_joins, mapping.values())
1300 2303 aaronmk
1301 2206 aaronmk
        try:
1302 2518 aaronmk
            cur = insert_select(db, out_table, mapping.keys(), main_select,
1303
                **insert_args)
1304 2357 aaronmk
            break # insert successful
1305 2206 aaronmk
        except DuplicateKeyException, e:
1306 2309 aaronmk
            log_exc(e)
1307
1308 2258 aaronmk
            old_join_cols = join_cols.copy()
1309 2811 aaronmk
            distinct_on.update(util.dict_subset(mapping, e.cols))
1310 2565 aaronmk
            join_cols.update(util.dict_subset_right_join(mapping, e.cols))
1311 2486 aaronmk
            log_debug('Ignoring existing rows, comparing on these columns:\n'
1312 2505 aaronmk
                +strings.as_inline_table(join_cols, ustr=col_ustr))
1313 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
1314 2230 aaronmk
        except NullValueException, e:
1315 2309 aaronmk
            log_exc(e)
1316
1317 2230 aaronmk
            out_col, = e.cols
1318
            try: in_col = mapping[out_col]
1319 2356 aaronmk
            except KeyError:
1320 2486 aaronmk
                log_debug('Missing mapping for NOT NULL column '+out_col)
1321 2451 aaronmk
                remove_all_rows()
1322 2758 aaronmk
            else: remove_rows(in_col, None, e)
1323 2542 aaronmk
        except FunctionValueException, e:
1324
            log_exc(e)
1325
1326
            func_name = e.name
1327
            value = e.value
1328
            for out_col, in_col in mapping.iteritems():
1329 2758 aaronmk
                in_col = sql_gen.unwrap_func_call(in_col, func_name)
1330
                invalid2null(in_col, value, e)
1331 2525 aaronmk
        except MissingCastException, e:
1332
            log_exc(e)
1333
1334
            out_col = e.col
1335 2706 aaronmk
            type_ = e.type
1336
1337 2724 aaronmk
            log_debug('Casting '+strings.as_tt(out_col)+' input to '
1338
                +strings.as_tt(type_))
1339 2735 aaronmk
            def wrap_func(col): return cast(db, type_, col, errors_table_)
1340 2705 aaronmk
            mapping[out_col] = sql_gen.wrap(wrap_func, mapping[out_col])
1341 2429 aaronmk
        except DatabaseErrors, e:
1342
            log_exc(e)
1343
1344 2808 aaronmk
            log_debug('No handler for exception')
1345
            on_error(e)
1346 2451 aaronmk
            remove_all_rows()
1347 2358 aaronmk
        # after exception handled, rerun loop with additional constraints
1348 2132 aaronmk
1349 2357 aaronmk
    if row_ct_ref != None and cur.rowcount >= 0:
1350
        row_ct_ref[0] += cur.rowcount
1351
1352
    if has_joins:
1353
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
1354 2486 aaronmk
        log_debug('Getting output table pkeys of existing/inserted rows')
1355 2420 aaronmk
        insert_into_pkeys(select_joins, pkeys_cols)
1356 2357 aaronmk
    else:
1357 2404 aaronmk
        add_row_num(db, insert_out_pkeys) # for joining with input pkeys
1358 2357 aaronmk
1359 2486 aaronmk
        log_debug('Getting input table pkeys of inserted rows')
1360 2786 aaronmk
        run_query_into(db, mk_main_select(input_joins, [in_pkey]),
1361 2404 aaronmk
            into=insert_in_pkeys)
1362
        add_row_num(db, insert_in_pkeys) # for joining with output pkeys
1363 2357 aaronmk
1364 2428 aaronmk
        assert table_row_count(db, insert_out_pkeys) == table_row_count(db,
1365
            insert_in_pkeys)
1366
1367 2486 aaronmk
        log_debug('Combining output and input pkeys in inserted order')
1368 2404 aaronmk
        pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
1369 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
1370 2420 aaronmk
        insert_into_pkeys(pkey_joins, pkeys_names)
1371 2357 aaronmk
1372 2486 aaronmk
    db.log_debug('Adding pkey on pkeys table to enable fast joins', level=2.5)
1373 2594 aaronmk
    add_pkey(db, into)
1374 2407 aaronmk
1375 2508 aaronmk
    log_debug('Setting pkeys of missing rows to '+strings.as_tt(repr(default)))
1376 2489 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(into,
1377 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
1378
        # must use join_same_not_null or query will take forever
1379 2420 aaronmk
    insert_into_pkeys(missing_rows_joins,
1380 2508 aaronmk
        [in_pkey_col, sql_gen.NamedCol(out_pkey, default)])
1381 2357 aaronmk
1382 2489 aaronmk
    assert table_row_count(db, into) == table_row_count(db, in_table)
1383 2428 aaronmk
1384 2702 aaronmk
    srcs = []
1385
    if is_func: srcs = sql_gen.cols_srcs(in_cols)
1386
    return sql_gen.Col(out_pkey, into, srcs)
1387 2115 aaronmk
1388
##### Data cleanup
1389
1390 2290 aaronmk
def cleanup_table(db, table, cols):
1391 2677 aaronmk
    table = sql_gen.as_Table(table)
1392
    cols = map(sql_gen.as_Col, cols)
1393 2115 aaronmk
1394 2677 aaronmk
    expr = ('nullif(nullif(trim(both from %s), '+db.esc_value('')+'), '
1395
        +db.esc_value(r'\N')+')')
1396
    changes = [(v, sql_gen.CustomCode(expr % v.to_str(db)))
1397
        for v in cols]
1398 2115 aaronmk
1399 2677 aaronmk
    update(db, table, changes)