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 2903 aaronmk
            # Prevent concurrent duplicate keys when used with INSERT SELECT
565
            lock_table(db, sql_gen.underlying_table(table), 'EXCLUSIVE')
566 2343 aaronmk
567 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
568 2122 aaronmk
569
        left_table = table
570
571 865 aaronmk
    missing = True
572 2376 aaronmk
    if conds != []:
573 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
574
        else: whitespace = '\n'
575 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
576
            .to_str(db) for l, r in conds], 'WHERE')
577 865 aaronmk
        missing = False
578 2227 aaronmk
    if order_by != None:
579 2467 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
580
    if limit != None: query += '\nLIMIT '+str(limit); missing = False
581 865 aaronmk
    if start != None:
582 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
583 865 aaronmk
        missing = False
584
    if missing: warnings.warn(DbWarning(
585
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
586
587 2786 aaronmk
    return query
588 11 aaronmk
589 2054 aaronmk
def select(db, *args, **kw_args):
590
    '''For params, see mk_select() and run_query()'''
591
    recover = kw_args.pop('recover', None)
592
    cacheable = kw_args.pop('cacheable', True)
593 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
594 2054 aaronmk
595 2791 aaronmk
    return run_query(db, mk_select(db, *args, **kw_args), recover, cacheable,
596
        log_level=log_level)
597 2054 aaronmk
598 2788 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, returning=None,
599
    embeddable=False):
600 1960 aaronmk
    '''
601
    @param returning str|None An inserted column (such as pkey) to return
602 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
603 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
604
        query will be fully cached, not just if it raises an exception.
605 1960 aaronmk
    '''
606 2754 aaronmk
    table = sql_gen.remove_table_rename(sql_gen.as_Table(table))
607 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
608 2646 aaronmk
    if cols != None:
609
        cols = [sql_gen.to_name_only_col(v, table).to_str(db) for v in cols]
610 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
611 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
612 2063 aaronmk
613
    # Build query
614 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
615
    query = first_line
616 2467 aaronmk
    if cols != None: query += '\n('+', '.join(cols)+')'
617
    query += '\n'+select_query
618 2063 aaronmk
619
    if returning != None:
620 2743 aaronmk
        query += '\nRETURNING '+sql_gen.to_name_only_col(returning).to_str(db)
621 2063 aaronmk
622 2070 aaronmk
    if embeddable:
623 2327 aaronmk
        assert returning != None
624
625 2070 aaronmk
        # Create function
626 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
627 2327 aaronmk
        return_type = 'SETOF '+returning.to_str(db)+'%TYPE'
628 2189 aaronmk
        while True:
629
            try:
630 2698 aaronmk
                function = sql_gen.TempFunction(function_name, db.autocommit)
631 2194 aaronmk
632 2189 aaronmk
                function_query = '''\
633 2698 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''()
634 2467 aaronmk
RETURNS '''+return_type+'''
635
LANGUAGE sql
636
AS $$
637 2788 aaronmk
'''+query+''';
638 2467 aaronmk
$$;
639 2070 aaronmk
'''
640 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
641 2464 aaronmk
                    log_ignore_excs=(DuplicateFunctionException,))
642 2189 aaronmk
                break # this version was successful
643
            except DuplicateFunctionException, e:
644
                function_name = next_version(function_name)
645
                # try again with next version of name
646 2070 aaronmk
647 2337 aaronmk
        # Return query that uses function
648 2698 aaronmk
        func_table = sql_gen.NamedTable('f', sql_gen.FunctionCall(function),
649 2742 aaronmk
            [returning]) # AS clause requires function alias
650 2787 aaronmk
        return mk_select(db, func_table, start=0, order_by=None)
651 2070 aaronmk
652 2787 aaronmk
    return query
653 2066 aaronmk
654
def insert_select(db, *args, **kw_args):
655 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
656 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
657
        values in
658 2072 aaronmk
    '''
659 2386 aaronmk
    into = kw_args.pop('into', None)
660
    if into != None: kw_args['embeddable'] = True
661 2066 aaronmk
    recover = kw_args.pop('recover', None)
662
    cacheable = kw_args.pop('cacheable', True)
663 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
664 2066 aaronmk
665 2789 aaronmk
    return run_query_into(db, mk_insert_select(db, *args, **kw_args), into,
666 2787 aaronmk
        recover=recover, cacheable=cacheable, log_level=log_level)
667 2063 aaronmk
668 2738 aaronmk
default = sql_gen.default # tells insert() to use the default value for a column
669 2066 aaronmk
670 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
671 2085 aaronmk
    '''For params, see insert_select()'''
672 1960 aaronmk
    if lists.is_seq(row): cols = None
673
    else:
674
        cols = row.keys()
675
        row = row.values()
676 2738 aaronmk
    row = list(row) # ensure that "== []" works
677 1960 aaronmk
678 2738 aaronmk
    if row == []: query = None
679
    else: query = sql_gen.Values(row).to_str(db)
680 1961 aaronmk
681 2788 aaronmk
    return insert_select(db, table, cols, query, *args, **kw_args)
682 11 aaronmk
683 2402 aaronmk
def mk_update(db, table, changes=None, cond=None):
684
    '''
685
    @param changes [(col, new_value),...]
686
        * container can be any iterable type
687
        * col: sql_gen.Code|str (for col name)
688
        * new_value: sql_gen.Code|literal value
689
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
690
    @return str query
691
    '''
692
    query = 'UPDATE '+sql_gen.as_Table(table).to_str(db)+'\nSET\n'
693 2405 aaronmk
    query += ',\n'.join((sql_gen.to_name_only_col(col, table).to_str(db)+' = '
694 2402 aaronmk
        +sql_gen.as_Value(new_value).to_str(db) for col, new_value in changes))
695 2467 aaronmk
    if cond != None: query += '\nWHERE\n'+cond.to_str(db)
696 2402 aaronmk
697
    return query
698
699
def update(db, *args, **kw_args):
700
    '''For params, see mk_update() and run_query()'''
701
    recover = kw_args.pop('recover', None)
702
703 2791 aaronmk
    return run_query(db, mk_update(db, *args, **kw_args), recover)
704 2402 aaronmk
705 135 aaronmk
def last_insert_id(db):
706 1849 aaronmk
    module = util.root_module(db.db)
707 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
708
    elif module == 'MySQLdb': return db.insert_id()
709
    else: return None
710 13 aaronmk
711 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
712 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
713 2415 aaronmk
    to names that will be distinct among the columns' tables.
714 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
715 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
716
    @param into The table for the new columns.
717 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
718
        columns will be included in the mapping even if they are not in cols.
719
        The tables of the provided Col objects will be changed to into, so make
720
        copies of them if you want to keep the original tables.
721
    @param as_items Whether to return a list of dict items instead of a dict
722 2383 aaronmk
    @return dict(orig_col=new_col, ...)
723
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
724 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
725
        * All mappings use the into table so its name can easily be
726 2383 aaronmk
          changed for all columns at once
727
    '''
728 2415 aaronmk
    cols = lists.uniqify(cols)
729
730 2394 aaronmk
    items = []
731 2389 aaronmk
    for col in preserve:
732 2390 aaronmk
        orig_col = copy.copy(col)
733 2392 aaronmk
        col.table = into
734 2394 aaronmk
        items.append((orig_col, col))
735
    preserve = set(preserve)
736
    for col in cols:
737 2716 aaronmk
        if col not in preserve:
738
            items.append((col, sql_gen.Col(str(col), into, col.srcs)))
739 2394 aaronmk
740
    if not as_items: items = dict(items)
741
    return items
742 2383 aaronmk
743 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
744 2391 aaronmk
    '''For params, see mk_flatten_mapping()
745
    @return See return value of mk_flatten_mapping()
746
    '''
747 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
748
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
749 2786 aaronmk
    run_query_into(db, mk_select(db, joins, cols, limit=limit, start=start),
750 2846 aaronmk
        into=into, add_indexes_=True)
751 2394 aaronmk
    return dict(items)
752 2391 aaronmk
753 2751 aaronmk
def mk_track_data_error(db, errors_table, cols, value, error_code, error):
754 2810 aaronmk
    assert cols != ()
755
756 2751 aaronmk
    cols = map(sql_gen.to_name_only_col, cols)
757
758
    columns_cols = ['column']
759
    columns = sql_gen.NamedValues('columns', columns_cols,
760
        [[c.name] for c in cols])
761
    values_cols = ['value', 'error_code', 'error']
762
    values = sql_gen.NamedValues('values', values_cols,
763
        [value, error_code, error])
764
765
    select_cols = columns_cols+values_cols
766
    name_only_cols = map(sql_gen.to_name_only_col, select_cols)
767 2755 aaronmk
    errors_table = sql_gen.NamedTable('errors', errors_table)
768 2751 aaronmk
    joins = [columns, sql_gen.Join(values, type_='CROSS'),
769
        sql_gen.Join(errors_table, dict(zip(name_only_cols, select_cols)),
770
        sql_gen.filter_out)]
771
772
    return mk_insert_select(db, errors_table, name_only_cols,
773 2787 aaronmk
        mk_select(db, joins, select_cols, order_by=None))
774 2751 aaronmk
775 2810 aaronmk
def track_data_error(db, errors_table, cols, *args, **kw_args):
776 2758 aaronmk
    '''
777
    @param errors_table If None, does nothing.
778
    '''
779 2810 aaronmk
    if errors_table == None or cols == (): return
780
    run_query(db, mk_track_data_error(db, errors_table, cols, *args, **kw_args),
781 2758 aaronmk
        cacheable=True, log_level=4)
782
783 2704 aaronmk
def cast(db, type_, col, errors_table=None):
784
    '''Casts an (unrenamed) column or value.
785 2769 aaronmk
    If errors_table set and col has srcs, saves errors in errors_table (using
786
    col's srcs attr as the source columns) and converts errors to warnings.
787 2820 aaronmk
    @param col str|sql_gen.Col|sql_gen.Literal
788 2704 aaronmk
    @param errors_table None|sql_gen.Table|str
789
    '''
790 2820 aaronmk
    col = sql_gen.as_Col(col)
791 2769 aaronmk
    save_errors = (errors_table != None and isinstance(col, sql_gen.Col)
792
        and col.srcs != ())
793
    if not save_errors: # can't save errors
794
        return sql_gen.CustomCode(col.to_str(db)+'::'+type_) # just cast
795 2704 aaronmk
796
    assert not isinstance(col, sql_gen.NamedCol)
797
798 2770 aaronmk
    errors_table = sql_gen.as_Table(errors_table)
799
    srcs = map(sql_gen.to_name_only_col, col.srcs)
800
    function_name = str(sql_gen.FunctionCall(type_, *srcs))
801 2704 aaronmk
    function = sql_gen.TempFunction(function_name, db.autocommit)
802
803 2718 aaronmk
    while True:
804
        # Create function definition
805
        query = '''\
806 2704 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''(value text)
807
RETURNS '''+type_+'''
808
LANGUAGE plpgsql
809
STRICT
810
AS $$
811
BEGIN
812
    /* The explicit cast to the return type is needed to make the cast happen
813
    inside the try block. (Implicit casts to the return type happen at the end
814
    of the function, outside any block.) */
815
    RETURN value::'''+type_+''';
816
EXCEPTION
817
    WHEN data_exception THEN
818
        -- Save error in errors table.
819 2751 aaronmk
        -- Insert the value and error for *each* source column.
820
'''+mk_track_data_error(db, errors_table, srcs,
821
    *map(sql_gen.CustomCode, ['value', 'SQLSTATE', 'SQLERRM']))+''';
822 2704 aaronmk
823
        RAISE WARNING '%', SQLERRM;
824
        RETURN NULL;
825
END;
826
$$;
827
'''
828 2718 aaronmk
829
        # Create function
830
        try:
831
            run_query(db, query, recover=True, cacheable=True,
832
                log_ignore_excs=(DuplicateFunctionException,))
833
            break # successful
834
        except DuplicateFunctionException:
835 2770 aaronmk
            function.name = next_version(function.name)
836
            # try again with next version of name
837 2704 aaronmk
838
    return sql_gen.FunctionCall(function, col)
839
840 2414 aaronmk
##### Database structure queries
841
842 2426 aaronmk
def table_row_count(db, table, recover=None):
843 2786 aaronmk
    return value(run_query(db, mk_select(db, table, [sql_gen.row_count],
844 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
845 2426 aaronmk
846 2414 aaronmk
def table_cols(db, table, recover=None):
847
    return list(col_names(select(db, table, limit=0, order_by=None,
848 2443 aaronmk
        recover=recover, log_level=4)))
849 2414 aaronmk
850 2291 aaronmk
def pkey(db, table, recover=None):
851 832 aaronmk
    '''Assumed to be first column in table'''
852 2339 aaronmk
    return table_cols(db, table, recover)[0]
853 832 aaronmk
854 2559 aaronmk
not_null_col = 'not_null_col'
855 2340 aaronmk
856
def table_not_null_col(db, table, recover=None):
857
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
858
    if not_null_col in table_cols(db, table, recover): return not_null_col
859
    else: return pkey(db, table, recover)
860
861 853 aaronmk
def index_cols(db, table, index):
862
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
863
    automatically created. When you don't know whether something is a UNIQUE
864
    constraint or a UNIQUE index, use this function.'''
865 1909 aaronmk
    module = util.root_module(db.db)
866
    if module == 'psycopg2':
867
        return list(values(run_query(db, '''\
868 853 aaronmk
SELECT attname
869 866 aaronmk
FROM
870
(
871
        SELECT attnum, attname
872
        FROM pg_index
873
        JOIN pg_class index ON index.oid = indexrelid
874
        JOIN pg_class table_ ON table_.oid = indrelid
875
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
876
        WHERE
877 2782 aaronmk
            table_.relname = '''+db.esc_value(table)+'''
878
            AND index.relname = '''+db.esc_value(index)+'''
879 866 aaronmk
    UNION
880
        SELECT attnum, attname
881
        FROM
882
        (
883
            SELECT
884
                indrelid
885
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
886
                    AS indkey
887
            FROM pg_index
888
            JOIN pg_class index ON index.oid = indexrelid
889
            JOIN pg_class table_ ON table_.oid = indrelid
890
            WHERE
891 2782 aaronmk
                table_.relname = '''+db.esc_value(table)+'''
892
                AND index.relname = '''+db.esc_value(index)+'''
893 866 aaronmk
        ) s
894
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
895
) s
896 853 aaronmk
ORDER BY attnum
897 2782 aaronmk
'''
898
            , cacheable=True, log_level=4)))
899 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
900
        ' database')
901 853 aaronmk
902 464 aaronmk
def constraint_cols(db, table, constraint):
903 1849 aaronmk
    module = util.root_module(db.db)
904 464 aaronmk
    if module == 'psycopg2':
905
        return list(values(run_query(db, '''\
906
SELECT attname
907
FROM pg_constraint
908
JOIN pg_class ON pg_class.oid = conrelid
909
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
910
WHERE
911 2783 aaronmk
    relname = '''+db.esc_value(table)+'''
912
    AND conname = '''+db.esc_value(constraint)+'''
913 464 aaronmk
ORDER BY attnum
914 2783 aaronmk
'''
915
            )))
916 464 aaronmk
    else: raise NotImplementedError("Can't list constraint columns for "+module+
917
        ' database')
918
919 2096 aaronmk
row_num_col = '_row_num'
920
921 2847 aaronmk
def add_index(db, exprs, table=None, unique=False, ensure_not_null=True):
922 2688 aaronmk
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
923 2538 aaronmk
    Currently, only function calls are supported as expressions.
924 2847 aaronmk
    @param ensure_not_null If set, translates NULL values to sentinel values.
925
        This allows indexes to be used for comparisons where NULLs are equal.
926 2538 aaronmk
    '''
927 2688 aaronmk
    if not lists.is_seq(exprs): exprs = [exprs]
928 2538 aaronmk
929 2688 aaronmk
    # Parse exprs
930
    old_exprs = exprs[:]
931
    exprs = []
932
    cols = []
933
    for i, expr in enumerate(old_exprs):
934
        expr = copy.deepcopy(expr) # don't modify input!
935 2823 aaronmk
        expr = sql_gen.as_Col(expr, table)
936 2688 aaronmk
937 2823 aaronmk
        # Handle nullable columns
938 2860 aaronmk
        if ensure_not_null:
939
            try: expr = sql_gen.ensure_not_null(db, expr)
940
            except KeyError: pass # unknown type, so just create plain index
941 2823 aaronmk
942 2688 aaronmk
        # Extract col
943
        if isinstance(expr, sql_gen.FunctionCall):
944
            col = expr.args[0]
945
            expr = sql_gen.Expr(expr)
946
        else: col = expr
947 2823 aaronmk
        assert isinstance(col, sql_gen.Col)
948 2688 aaronmk
949
        # Extract table
950
        if table == None:
951
            assert sql_gen.is_table_col(col)
952
            table = col.table
953
954
        col.table = None
955
956
        exprs.append(expr)
957
        cols.append(col)
958 2408 aaronmk
959 2688 aaronmk
    table = sql_gen.as_Table(table)
960
    index = sql_gen.Table(str(sql_gen.Col(','.join(map(str, cols)), table)))
961
962
    str_ = 'CREATE'
963
    if unique: str_ += ' UNIQUE'
964
    str_ += ' INDEX '+index.to_str(db)+' ON '+table.to_str(db)+' ('+(
965
        ', '.join((v.to_str(db) for v in exprs)))+')'
966
967
    try: run_query(db, str_, recover=True, cacheable=True, log_level=3)
968 2408 aaronmk
    except DuplicateTableException: pass # index already existed
969
970 2686 aaronmk
def add_pkey(db, table, cols=None, recover=None):
971
    '''Adds a primary key.
972
    @param cols [sql_gen.Col,...] The columns in the primary key.
973
        Defaults to the first column in the table.
974 2406 aaronmk
    @pre The table must not already have a primary key.
975
    '''
976
    table = sql_gen.as_Table(table)
977 2686 aaronmk
    if cols == None: cols = [pkey(db, table, recover)]
978
    col_strs = [sql_gen.to_name_only_col(v).to_str(db) for v in cols]
979 2406 aaronmk
980 2781 aaronmk
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD PRIMARY KEY ('
981
        +(', '.join(col_strs))+')', recover=True, cacheable=True, log_level=3,
982
        log_ignore_excs=(DuplicateTableException,))
983 2406 aaronmk
984 2772 aaronmk
already_indexed = object() # tells add_indexes() the pkey has already been added
985
986
def add_indexes(db, table, has_pkey=True):
987
    '''Adds an index on all columns in a table.
988
    @param has_pkey bool|already_indexed Whether a pkey instead of a regular
989
        index should be added on the first column.
990
        * If already_indexed, the pkey is assumed to have already been added
991
    '''
992
    cols = table_cols(db, table)
993
    if has_pkey:
994
        if has_pkey is not already_indexed: add_pkey(db, table)
995
        cols = cols[1:]
996
    for col in cols: add_index(db, col, table)
997
998 2086 aaronmk
def add_row_num(db, table):
999 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
1000
    be the primary key.'''
1001 2320 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
1002 2096 aaronmk
    run_query(db, 'ALTER TABLE '+table+' ADD COLUMN '+row_num_col
1003 2443 aaronmk
        +' serial NOT NULL PRIMARY KEY', log_level=3)
1004 2086 aaronmk
1005 2889 aaronmk
def drop_table(db, table):
1006
    table = sql_gen.as_Table(table)
1007
    return run_query(db, 'DROP TABLE IF EXISTS '+table.to_str(db)+' CASCADE')
1008
1009 2760 aaronmk
def create_table(db, table, cols, has_pkey=True, col_indexes=True):
1010 2675 aaronmk
    '''Creates a table.
1011 2681 aaronmk
    @param cols [sql_gen.TypedCol,...] The column names and types
1012
    @param has_pkey If set, the first column becomes the primary key.
1013 2760 aaronmk
    @param col_indexes bool|[ref]
1014
        * If True, indexes will be added on all non-pkey columns.
1015
        * If a list reference, [0] will be set to a function to do this.
1016
          This can be used to delay index creation until the table is populated.
1017 2675 aaronmk
    '''
1018
    table = sql_gen.as_Table(table)
1019
1020 2681 aaronmk
    if has_pkey:
1021
        cols[0] = pkey = copy.copy(cols[0]) # don't modify input!
1022 2872 aaronmk
        pkey.constraints = 'PRIMARY KEY'
1023 2681 aaronmk
1024 2675 aaronmk
    str_ = 'CREATE TABLE '+table.to_str(db)+' (\n'
1025 2681 aaronmk
    str_ += '\n, '.join(v.to_str(db) for v in cols)
1026
    str_ += '\n);\n'
1027 2675 aaronmk
    run_query(db, str_, cacheable=True, log_level=2)
1028 2760 aaronmk
1029
    # Add indexes
1030 2773 aaronmk
    if has_pkey: has_pkey = already_indexed
1031
    def add_indexes_(): add_indexes(db, table, has_pkey)
1032
    if isinstance(col_indexes, list): col_indexes[0] = add_indexes_ # defer
1033
    elif col_indexes: add_indexes_() # add now
1034 2675 aaronmk
1035 2684 aaronmk
def vacuum(db, table):
1036
    table = sql_gen.as_Table(table)
1037
    db.with_autocommit(lambda: run_query(db, 'VACUUM ANALYZE '+table.to_str(db),
1038
        log_level=3))
1039
1040 2732 aaronmk
def truncate(db, table, schema='public'):
1041 2777 aaronmk
    table = sql_gen.as_Table(table, schema)
1042 2733 aaronmk
    return run_query(db, 'TRUNCATE '+table.to_str(db)+' CASCADE')
1043 2732 aaronmk
1044 2729 aaronmk
def tables(db, schema_like='public', table_like='%', exact=False):
1045
    if exact: compare = '='
1046
    else: compare = 'LIKE'
1047
1048 1849 aaronmk
    module = util.root_module(db.db)
1049 832 aaronmk
    if module == 'psycopg2':
1050 2779 aaronmk
        conds = [('schemaname', sql_gen.CompareCond(schema_like, compare)),
1051
            ('tablename', sql_gen.CompareCond(table_like, compare))]
1052
        return values(select(db, 'pg_tables', ['tablename'], conds,
1053
            order_by='tablename', log_level=4))
1054 1968 aaronmk
    elif module == 'MySQLdb':
1055 2779 aaronmk
        return values(run_query(db, 'SHOW TABLES LIKE '+db.esc_value(table_like)
1056
            , cacheable=True, log_level=4))
1057 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
1058 830 aaronmk
1059 2726 aaronmk
def table_exists(db, table):
1060
    table = sql_gen.as_Table(table)
1061 2730 aaronmk
    return list(tables(db, table.schema, table.name, exact=True)) != []
1062 2726 aaronmk
1063
def errors_table(db, table, if_exists=True):
1064
    '''
1065
    @param if_exists If set, returns None if the errors table doesn't exist
1066
    @return None|sql_gen.Table
1067
    '''
1068
    table = sql_gen.as_Table(table)
1069
    if table.srcs != (): table = table.srcs[0]
1070
1071
    errors_table = sql_gen.suffixed_table(table, '.errors')
1072
    if if_exists and not table_exists(db, errors_table): return None
1073
    return errors_table
1074
1075 833 aaronmk
##### Database management
1076
1077 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
1078
    '''For kw_args, see tables()'''
1079
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
1080 833 aaronmk
1081 832 aaronmk
##### Heuristic queries
1082
1083 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
1084 1554 aaronmk
    '''Recovers from errors.
1085 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
1086
    '''
1087 2642 aaronmk
    row = sql_gen.ColDict(db, table, row)
1088 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
1089
1090 471 aaronmk
    try:
1091 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
1092 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
1093
            row_ct_ref[0] += cur.rowcount
1094
        return value(cur)
1095 471 aaronmk
    except DuplicateKeyException, e:
1096 2653 aaronmk
        row = sql_gen.ColDict(db, table,
1097
            util.dict_subset_right_join(row, e.cols))
1098
        return value(select(db, table, [pkey_], row, recover=True))
1099 471 aaronmk
1100 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
1101 830 aaronmk
    '''Recovers from errors'''
1102 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
1103 14 aaronmk
    except StopIteration:
1104 40 aaronmk
        if not create: raise
1105 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
1106 2078 aaronmk
1107 2593 aaronmk
def is_func_result(col):
1108
    return col.table.name.find('(') >= 0 and col.name == 'result'
1109
1110 2592 aaronmk
def into_table_name(out_table, in_tables0, mapping, is_func):
1111 2615 aaronmk
    def in_col_str(in_col):
1112
        in_col = sql_gen.remove_col_rename(in_col)
1113
        if isinstance(in_col, sql_gen.Col):
1114
            table = in_col.table
1115
            if table == in_tables0:
1116
                in_col = sql_gen.to_name_only_col(in_col)
1117
            elif is_func_result(in_col): in_col = table # omit col name
1118
        return str(in_col)
1119
1120 2580 aaronmk
    str_ = str(out_table)
1121
    if is_func:
1122 2615 aaronmk
        str_ += '('
1123 2580 aaronmk
1124 2615 aaronmk
        try: value_in_col = mapping['value']
1125
        except KeyError:
1126 2654 aaronmk
            str_ += ', '.join((str(k)+'='+in_col_str(v)
1127 2615 aaronmk
                for k, v in mapping.iteritems()))
1128
        else: str_ += in_col_str(value_in_col)
1129
1130
        str_ += ')'
1131 2812 aaronmk
    else:
1132
        out_col = 'rank'
1133
        try: in_col = mapping[out_col]
1134
        except KeyError: str_ += '_pkeys'
1135
        else: # has a rank column, so hierarchical
1136
            str_ += '['+str(out_col)+'='+in_col_str(in_col)+']'
1137 2580 aaronmk
    return str_
1138
1139 2508 aaronmk
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None, into=None,
1140 2805 aaronmk
    default=None, is_func=False, on_error=exc.raise_):
1141 2078 aaronmk
    '''Recovers from errors.
1142
    Only works under PostgreSQL (uses INSERT RETURNING).
1143 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
1144
        tables to join with it using the main input table's pkey
1145 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
1146 2612 aaronmk
        * out_table_col: str (*not* sql_gen.Col)
1147
        * in_table_col: sql_gen.Col|literal-value
1148 2489 aaronmk
    @param into The table to contain the output and input pkeys.
1149 2574 aaronmk
        Defaults to `out_table.name+'_pkeys'`.
1150 2509 aaronmk
    @param default The *output* column to use as the pkey for missing rows.
1151
        If this output column does not exist in the mapping, uses None.
1152 2552 aaronmk
    @param is_func Whether out_table is the name of a SQL function, not a table
1153 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
1154 2078 aaronmk
    '''
1155 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
1156 2312 aaronmk
1157 2450 aaronmk
    def log_debug(msg): db.log_debug(msg, level=1.5)
1158 2505 aaronmk
    def col_ustr(str_):
1159 2567 aaronmk
        return strings.repr_no_u(sql_gen.remove_col_rename(str_))
1160 2450 aaronmk
1161 2768 aaronmk
    out_pkey = pkey(db, out_table, recover=True)
1162
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
1163
1164
    if mapping == {}: # need at least one column for INSERT SELECT
1165
        mapping = {out_pkey: None} # ColDict will replace with default value
1166
1167 2486 aaronmk
    log_debug('********** New iteration **********')
1168 2505 aaronmk
    log_debug('Inserting these input columns into '+strings.as_tt(
1169
        out_table.to_str(db))+':\n'+strings.as_table(mapping, ustr=col_ustr))
1170 2463 aaronmk
1171 2382 aaronmk
    # Create input joins from list of input tables
1172
    in_tables_ = in_tables[:] # don't modify input!
1173
    in_tables0 = in_tables_.pop(0) # first table is separate
1174 2735 aaronmk
    errors_table_ = errors_table(db, in_tables0)
1175 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
1176 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
1177 2460 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v,
1178
        {in_pkey: sql_gen.join_same_not_null}) for v in in_tables_]
1179 2131 aaronmk
1180 2592 aaronmk
    if into == None:
1181
        into = into_table_name(out_table, in_tables0, mapping, is_func)
1182
    into = sql_gen.as_Table(into)
1183
1184 2702 aaronmk
    # Set column sources
1185
    in_cols = filter(sql_gen.is_table_col, mapping.values())
1186
    for col in in_cols:
1187
        if col.table == in_tables0: col.set_srcs(sql_gen.src_self)
1188
1189 2486 aaronmk
    log_debug('Joining together input tables into temp table')
1190 2395 aaronmk
    # Place in new table for speed and so don't modify input if values edited
1191 2584 aaronmk
    in_table = sql_gen.Table('in')
1192 2702 aaronmk
    mapping = dicts.join(mapping, flatten(db, in_table, input_joins, in_cols,
1193
        preserve=[in_pkey_col], start=0))
1194 2395 aaronmk
    input_joins = [in_table]
1195 2486 aaronmk
    db.log_debug('Temp table: '+strings.as_tt(in_table.to_str(db)), level=2)
1196 2395 aaronmk
1197 2656 aaronmk
    mapping = sql_gen.ColDict(db, out_table, mapping)
1198
        # after applying dicts.join() because that returns a plain dict
1199
1200 2509 aaronmk
    # Resolve default value column
1201 2841 aaronmk
    if default != None:
1202
        try: default = mapping[default]
1203
        except KeyError:
1204 2509 aaronmk
            db.log_debug('Default value column '
1205
                +strings.as_tt(strings.repr_no_u(default))
1206 2511 aaronmk
                +' does not exist in mapping, falling back to None', level=2.1)
1207 2509 aaronmk
            default = None
1208
1209 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
1210 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
1211
1212 2201 aaronmk
    pkeys_table_exists_ref = [False]
1213 2420 aaronmk
    def insert_into_pkeys(joins, cols):
1214 2786 aaronmk
        query = mk_select(db, joins, cols, order_by=None, start=0)
1215 2201 aaronmk
        if pkeys_table_exists_ref[0]:
1216 2786 aaronmk
            insert_select(db, into, pkeys_names, query)
1217 2201 aaronmk
        else:
1218 2786 aaronmk
            run_query_into(db, query, into=into)
1219 2201 aaronmk
            pkeys_table_exists_ref[0] = True
1220
1221 2429 aaronmk
    limit_ref = [None]
1222 2380 aaronmk
    conds = set()
1223 2811 aaronmk
    distinct_on = sql_gen.ColDict(db, out_table)
1224 2325 aaronmk
    def mk_main_select(joins, cols):
1225 2811 aaronmk
        distinct_on_cols = [c.to_Col() for c in distinct_on.values()]
1226
        return mk_select(db, joins, cols, conds, distinct_on_cols,
1227 2429 aaronmk
            limit=limit_ref[0], start=0)
1228 2132 aaronmk
1229 2519 aaronmk
    exc_strs = set()
1230 2309 aaronmk
    def log_exc(e):
1231 2519 aaronmk
        e_str = exc.str_(e, first_line_only=True)
1232
        log_debug('Caught exception: '+e_str)
1233
        assert e_str not in exc_strs # avoid infinite loops
1234
        exc_strs.add(e_str)
1235 2735 aaronmk
1236 2451 aaronmk
    def remove_all_rows():
1237 2896 aaronmk
        log_debug('Ignoring all rows')
1238 2429 aaronmk
        limit_ref[0] = 0 # just create an empty pkeys table
1239 2735 aaronmk
1240 2758 aaronmk
    def ignore(in_col, value, e):
1241
        track_data_error(db, errors_table_, in_col.srcs, value, e.cause.pgcode,
1242
            e.cause.pgerror)
1243 2895 aaronmk
        log_debug('Ignoring rows with '+strings.as_tt(repr(in_col))+' = '
1244 2545 aaronmk
            +strings.as_tt(repr(value)))
1245 2895 aaronmk
1246 2758 aaronmk
    def remove_rows(in_col, value, e):
1247
        ignore(in_col, value, e)
1248 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
1249
        assert cond not in conds # avoid infinite loops
1250 2380 aaronmk
        conds.add(cond)
1251 2895 aaronmk
1252 2758 aaronmk
    def invalid2null(in_col, value, e):
1253
        ignore(in_col, value, e)
1254 2403 aaronmk
        update(db, in_table, [(in_col, None)],
1255
            sql_gen.ColValueCond(in_col, value))
1256 2245 aaronmk
1257 2589 aaronmk
    def insert_pkeys_table(which):
1258
        return sql_gen.Table(sql_gen.add_suffix(in_table.name,
1259
            '_insert_'+which+'_pkeys'))
1260
    insert_out_pkeys = insert_pkeys_table('out')
1261
    insert_in_pkeys = insert_pkeys_table('in')
1262
1263 2206 aaronmk
    # Do inserts and selects
1264 2642 aaronmk
    join_cols = sql_gen.ColDict(db, out_table)
1265 2206 aaronmk
    while True:
1266 2521 aaronmk
        if limit_ref[0] == 0: # special case
1267
            log_debug('Creating an empty pkeys table')
1268 2786 aaronmk
            cur = run_query_into(db, mk_select(db, out_table, [out_pkey],
1269 2521 aaronmk
                limit=limit_ref[0]), into=insert_out_pkeys)
1270
            break # don't do main case
1271
1272 2303 aaronmk
        has_joins = join_cols != {}
1273
1274 2903 aaronmk
        log_debug('Trying to insert new rows')
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 2206 aaronmk
        try:
1287 2518 aaronmk
            cur = insert_select(db, out_table, mapping.keys(), main_select,
1288
                **insert_args)
1289 2357 aaronmk
            break # insert successful
1290 2206 aaronmk
        except DuplicateKeyException, e:
1291 2309 aaronmk
            log_exc(e)
1292
1293 2258 aaronmk
            old_join_cols = join_cols.copy()
1294 2811 aaronmk
            distinct_on.update(util.dict_subset(mapping, e.cols))
1295 2565 aaronmk
            join_cols.update(util.dict_subset_right_join(mapping, e.cols))
1296 2486 aaronmk
            log_debug('Ignoring existing rows, comparing on these columns:\n'
1297 2505 aaronmk
                +strings.as_inline_table(join_cols, ustr=col_ustr))
1298 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
1299 2230 aaronmk
        except NullValueException, e:
1300 2309 aaronmk
            log_exc(e)
1301
1302 2230 aaronmk
            out_col, = e.cols
1303
            try: in_col = mapping[out_col]
1304 2356 aaronmk
            except KeyError:
1305 2486 aaronmk
                log_debug('Missing mapping for NOT NULL column '+out_col)
1306 2451 aaronmk
                remove_all_rows()
1307 2758 aaronmk
            else: remove_rows(in_col, None, e)
1308 2542 aaronmk
        except FunctionValueException, e:
1309
            log_exc(e)
1310
1311
            func_name = e.name
1312
            value = e.value
1313
            for out_col, in_col in mapping.iteritems():
1314 2758 aaronmk
                in_col = sql_gen.unwrap_func_call(in_col, func_name)
1315
                invalid2null(in_col, value, e)
1316 2525 aaronmk
        except MissingCastException, e:
1317
            log_exc(e)
1318
1319
            out_col = e.col
1320 2706 aaronmk
            type_ = e.type
1321
1322 2724 aaronmk
            log_debug('Casting '+strings.as_tt(out_col)+' input to '
1323
                +strings.as_tt(type_))
1324 2735 aaronmk
            def wrap_func(col): return cast(db, type_, col, errors_table_)
1325 2705 aaronmk
            mapping[out_col] = sql_gen.wrap(wrap_func, mapping[out_col])
1326 2429 aaronmk
        except DatabaseErrors, e:
1327
            log_exc(e)
1328
1329 2808 aaronmk
            log_debug('No handler for exception')
1330
            on_error(e)
1331 2451 aaronmk
            remove_all_rows()
1332 2358 aaronmk
        # after exception handled, rerun loop with additional constraints
1333 2132 aaronmk
1334 2357 aaronmk
    if row_ct_ref != None and cur.rowcount >= 0:
1335
        row_ct_ref[0] += cur.rowcount
1336
1337
    if has_joins:
1338
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
1339 2486 aaronmk
        log_debug('Getting output table pkeys of existing/inserted rows')
1340 2420 aaronmk
        insert_into_pkeys(select_joins, pkeys_cols)
1341 2357 aaronmk
    else:
1342 2404 aaronmk
        add_row_num(db, insert_out_pkeys) # for joining with input pkeys
1343 2357 aaronmk
1344 2486 aaronmk
        log_debug('Getting input table pkeys of inserted rows')
1345 2786 aaronmk
        run_query_into(db, mk_main_select(input_joins, [in_pkey]),
1346 2404 aaronmk
            into=insert_in_pkeys)
1347
        add_row_num(db, insert_in_pkeys) # for joining with output pkeys
1348 2357 aaronmk
1349 2428 aaronmk
        assert table_row_count(db, insert_out_pkeys) == table_row_count(db,
1350
            insert_in_pkeys)
1351
1352 2486 aaronmk
        log_debug('Combining output and input pkeys in inserted order')
1353 2404 aaronmk
        pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
1354 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
1355 2420 aaronmk
        insert_into_pkeys(pkey_joins, pkeys_names)
1356 2357 aaronmk
1357 2486 aaronmk
    db.log_debug('Adding pkey on pkeys table to enable fast joins', level=2.5)
1358 2594 aaronmk
    add_pkey(db, into)
1359 2407 aaronmk
1360 2508 aaronmk
    log_debug('Setting pkeys of missing rows to '+strings.as_tt(repr(default)))
1361 2489 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(into,
1362 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
1363
        # must use join_same_not_null or query will take forever
1364 2420 aaronmk
    insert_into_pkeys(missing_rows_joins,
1365 2508 aaronmk
        [in_pkey_col, sql_gen.NamedCol(out_pkey, default)])
1366 2357 aaronmk
1367 2489 aaronmk
    assert table_row_count(db, into) == table_row_count(db, in_table)
1368 2428 aaronmk
1369 2702 aaronmk
    srcs = []
1370
    if is_func: srcs = sql_gen.cols_srcs(in_cols)
1371
    return sql_gen.Col(out_pkey, into, srcs)
1372 2115 aaronmk
1373
##### Data cleanup
1374
1375 2290 aaronmk
def cleanup_table(db, table, cols):
1376 2677 aaronmk
    table = sql_gen.as_Table(table)
1377
    cols = map(sql_gen.as_Col, cols)
1378 2115 aaronmk
1379 2677 aaronmk
    expr = ('nullif(nullif(trim(both from %s), '+db.esc_value('')+'), '
1380
        +db.esc_value(r'\N')+')')
1381
    changes = [(v, sql_gen.CustomCode(expr % v.to_str(db)))
1382
        for v in cols]
1383 2115 aaronmk
1384 2677 aaronmk
    update(db, table, changes)