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 2915 aaronmk
        log_debug=log_debug_none, debug_temp=False):
153
        '''
154
        @param debug_temp Whether temporary objects should instead be permanent.
155
            This assists in debugging the internal objects used by the program.
156
        '''
157 2916 aaronmk
        if debug_temp: autocommit = True
158
159 1869 aaronmk
        self.db_config = db_config
160 2190 aaronmk
        self.autocommit = autocommit
161
        self.caching = caching
162 1901 aaronmk
        self.log_debug = log_debug
163 2193 aaronmk
        self.debug = log_debug != log_debug_none
164 2915 aaronmk
        self.debug_temp = debug_temp
165 1869 aaronmk
166
        self.__db = None
167 1889 aaronmk
        self.query_results = {}
168 2139 aaronmk
        self._savepoint = 0
169 2671 aaronmk
        self._notices_seen = set()
170 1869 aaronmk
171
    def __getattr__(self, name):
172
        if name == '__dict__': raise Exception('getting __dict__')
173
        if name == 'db': return self._db()
174
        else: raise AttributeError()
175
176
    def __getstate__(self):
177
        state = copy.copy(self.__dict__) # shallow copy
178 1915 aaronmk
        state['log_debug'] = None # don't pickle the debug callback
179 1869 aaronmk
        state['_DbConn__db'] = None # don't pickle the connection
180
        return state
181
182 2165 aaronmk
    def connected(self): return self.__db != None
183
184 1869 aaronmk
    def _db(self):
185
        if self.__db == None:
186
            # Process db_config
187
            db_config = self.db_config.copy() # don't modify input!
188 2097 aaronmk
            schemas = db_config.pop('schemas', None)
189 1869 aaronmk
            module_name, mappings = db_engines[db_config.pop('engine')]
190
            module = __import__(module_name)
191
            _add_module(module)
192
            for orig, new in mappings.iteritems():
193
                try: util.rename_key(db_config, orig, new)
194
                except KeyError: pass
195
196
            # Connect
197
            self.__db = module.connect(**db_config)
198
199
            # Configure connection
200 2906 aaronmk
            if hasattr(self.db, 'set_isolation_level'):
201
                import psycopg2.extensions
202
                self.db.set_isolation_level(
203
                    psycopg2.extensions.ISOLATION_LEVEL_READ_COMMITTED)
204 2101 aaronmk
            if schemas != None:
205 2893 aaronmk
                search_path = [self.esc_name(s) for s in schemas.split(',')]
206
                search_path.append(value(run_query(self, 'SHOW search_path',
207
                    log_level=4)))
208
                run_query(self, 'SET search_path TO '+(','.join(search_path)),
209
                    log_level=3)
210 1869 aaronmk
211
        return self.__db
212 1889 aaronmk
213 1891 aaronmk
    class DbCursor(Proxy):
214 1927 aaronmk
        def __init__(self, outer):
215 1891 aaronmk
            Proxy.__init__(self, outer.db.cursor())
216 2191 aaronmk
            self.outer = outer
217 1927 aaronmk
            self.query_results = outer.query_results
218 1894 aaronmk
            self.query_lookup = None
219 1891 aaronmk
            self.result = []
220 1889 aaronmk
221 2802 aaronmk
        def execute(self, query):
222 2764 aaronmk
            self._is_insert = query.startswith('INSERT')
223 2797 aaronmk
            self.query_lookup = query
224 2148 aaronmk
            try:
225 2191 aaronmk
                try:
226 2802 aaronmk
                    cur = self.inner.execute(query)
227 2191 aaronmk
                    self.outer.do_autocommit()
228 2802 aaronmk
                finally: self.query = get_cur_query(self.inner, query)
229 1904 aaronmk
            except Exception, e:
230 2802 aaronmk
                _add_cursor_info(e, self, query)
231 1904 aaronmk
                self.result = e # cache the exception as the result
232
                self._cache_result()
233
                raise
234 2800 aaronmk
            if self.rowcount == 0 and query.startswith('SELECT'): # empty SELECT
235
                consume_rows(self) # fetch all rows so result will be cached
236 2762 aaronmk
            return cur
237 1894 aaronmk
238 1891 aaronmk
        def fetchone(self):
239
            row = self.inner.fetchone()
240 1899 aaronmk
            if row != None: self.result.append(row)
241
            # otherwise, fetched all rows
242 1904 aaronmk
            else: self._cache_result()
243
            return row
244
245
        def _cache_result(self):
246 1906 aaronmk
            # For inserts, only cache exceptions since inserts are not
247
            # idempotent, but an invalid insert will always be invalid
248 1930 aaronmk
            if self.query_results != None and (not self._is_insert
249 1906 aaronmk
                or isinstance(self.result, Exception)):
250
251 1894 aaronmk
                assert self.query_lookup != None
252 1916 aaronmk
                self.query_results[self.query_lookup] = self.CacheCursor(
253
                    util.dict_subset(dicts.AttrsDictView(self),
254
                    ['query', 'result', 'rowcount', 'description']))
255 1906 aaronmk
256 1916 aaronmk
        class CacheCursor:
257
            def __init__(self, cached_result): self.__dict__ = cached_result
258
259 1927 aaronmk
            def execute(self, *args, **kw_args):
260 1916 aaronmk
                if isinstance(self.result, Exception): raise self.result
261
                # otherwise, result is a rows list
262
                self.iter = iter(self.result)
263
264
            def fetchone(self):
265
                try: return self.iter.next()
266
                except StopIteration: return None
267 1891 aaronmk
268 2212 aaronmk
    def esc_value(self, value):
269 2663 aaronmk
        try: str_ = self.mogrify('%s', [value])
270
        except NotImplementedError, e:
271
            module = util.root_module(self.db)
272
            if module == 'MySQLdb':
273
                import _mysql
274
                str_ = _mysql.escape_string(value)
275
            else: raise e
276 2374 aaronmk
        return strings.to_unicode(str_)
277 2212 aaronmk
278 2347 aaronmk
    def esc_name(self, name): return esc_name(self, name) # calls global func
279
280 2814 aaronmk
    def std_code(self, str_):
281
        '''Standardizes SQL code.
282
        * Ensures that string literals are prefixed by `E`
283
        '''
284
        if str_.startswith("'"): str_ = 'E'+str_
285
        return str_
286
287 2665 aaronmk
    def can_mogrify(self):
288 2663 aaronmk
        module = util.root_module(self.db)
289 2665 aaronmk
        return module == 'psycopg2'
290 2663 aaronmk
291 2665 aaronmk
    def mogrify(self, query, params=None):
292
        if self.can_mogrify(): return self.db.cursor().mogrify(query, params)
293
        else: raise NotImplementedError("Can't mogrify query")
294
295 2671 aaronmk
    def print_notices(self):
296 2725 aaronmk
        if hasattr(self.db, 'notices'):
297
            for msg in self.db.notices:
298
                if msg not in self._notices_seen:
299
                    self._notices_seen.add(msg)
300
                    self.log_debug(msg, level=2)
301 2671 aaronmk
302 2793 aaronmk
    def run_query(self, query, cacheable=False, log_level=2,
303 2464 aaronmk
        debug_msg_ref=None):
304 2445 aaronmk
        '''
305 2464 aaronmk
        @param log_ignore_excs The log_level will be increased by 2 if the query
306
            throws one of these exceptions.
307 2664 aaronmk
        @param debug_msg_ref If specified, the log message will be returned in
308
            this instead of being output. This allows you to filter log messages
309
            depending on the result of the query.
310 2445 aaronmk
        '''
311 2167 aaronmk
        assert query != None
312
313 2047 aaronmk
        if not self.caching: cacheable = False
314 1903 aaronmk
        used_cache = False
315 2664 aaronmk
316
        def log_msg(query):
317
            if used_cache: cache_status = 'cache hit'
318
            elif cacheable: cache_status = 'cache miss'
319
            else: cache_status = 'non-cacheable'
320
            return 'DB query: '+cache_status+':\n'+strings.as_code(query, 'SQL')
321
322 1903 aaronmk
        try:
323 1927 aaronmk
            # Get cursor
324
            if cacheable:
325
                try:
326 2797 aaronmk
                    cur = self.query_results[query]
327 1927 aaronmk
                    used_cache = True
328
                except KeyError: cur = self.DbCursor(self)
329
            else: cur = self.db.cursor()
330
331 2664 aaronmk
            # Log query
332
            if self.debug and debug_msg_ref == None: # log before running
333
                self.log_debug(log_msg(query), log_level)
334
335 1927 aaronmk
            # Run query
336 2793 aaronmk
            cur.execute(query)
337 1903 aaronmk
        finally:
338 2671 aaronmk
            self.print_notices()
339 2664 aaronmk
            if self.debug and debug_msg_ref != None: # return after running
340 2793 aaronmk
                debug_msg_ref[0] = log_msg(str(get_cur_query(cur, query)))
341 1903 aaronmk
342
        return cur
343 1914 aaronmk
344 2797 aaronmk
    def is_cached(self, query): return query in self.query_results
345 2139 aaronmk
346 2907 aaronmk
    def with_autocommit(self, func):
347 2801 aaronmk
        import psycopg2.extensions
348
349
        prev_isolation_level = self.db.isolation_level
350 2907 aaronmk
        self.db.set_isolation_level(
351
            psycopg2.extensions.ISOLATION_LEVEL_AUTOCOMMIT)
352 2683 aaronmk
        try: return func()
353 2801 aaronmk
        finally: self.db.set_isolation_level(prev_isolation_level)
354 2683 aaronmk
355 2139 aaronmk
    def with_savepoint(self, func):
356 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
357 2443 aaronmk
        self.run_query('SAVEPOINT '+savepoint, log_level=4)
358 2139 aaronmk
        self._savepoint += 1
359 2683 aaronmk
        try:
360 2139 aaronmk
            try: return_val = func()
361
            finally:
362
                self._savepoint -= 1
363
                assert self._savepoint >= 0
364
        except:
365 2443 aaronmk
            self.run_query('ROLLBACK TO SAVEPOINT '+savepoint, log_level=4)
366 2139 aaronmk
            raise
367
        else:
368 2443 aaronmk
            self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
369 2191 aaronmk
            self.do_autocommit()
370 2139 aaronmk
            return return_val
371 2191 aaronmk
372
    def do_autocommit(self):
373
        '''Autocommits if outside savepoint'''
374
        assert self._savepoint >= 0
375
        if self.autocommit and self._savepoint == 0:
376 2668 aaronmk
            self.log_debug('Autocommitting')
377 2191 aaronmk
            self.db.commit()
378 2643 aaronmk
379 2819 aaronmk
    def col_info(self, col):
380 2643 aaronmk
        table = sql_gen.Table('columns', 'information_schema')
381 2821 aaronmk
        cols = ['data_type', 'column_default',
382
            cast(self, 'boolean', 'is_nullable')]
383 2643 aaronmk
384
        conds = [('table_name', col.table.name), ('column_name', col.name)]
385
        schema = col.table.schema
386
        if schema != None: conds.append(('table_schema', schema))
387
388 2819 aaronmk
        type_, default, nullable = row(select(self, table, cols, conds,
389 2874 aaronmk
            order_by='table_schema', limit=1, log_level=4))
390 2643 aaronmk
            # TODO: order_by search_path schema order
391 2819 aaronmk
        default = sql_gen.as_Code(default, self)
392
393
        return sql_gen.TypedCol(col.name, type_, default, nullable)
394 1849 aaronmk
395 1869 aaronmk
connect = DbConn
396
397 832 aaronmk
##### Recoverable querying
398 15 aaronmk
399 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
400 11 aaronmk
401 2791 aaronmk
def run_query(db, query, recover=None, cacheable=False, log_level=2,
402
    log_ignore_excs=None, **kw_args):
403 2794 aaronmk
    '''For params, see DbConn.run_query()'''
404 830 aaronmk
    if recover == None: recover = False
405 2464 aaronmk
    if log_ignore_excs == None: log_ignore_excs = ()
406
    log_ignore_excs = tuple(log_ignore_excs)
407 830 aaronmk
408 2666 aaronmk
    debug_msg_ref = None # usually, db.run_query() logs query before running it
409
    # But if filtering with log_ignore_excs, wait until after exception parsing
410
    if log_ignore_excs != () or not db.can_mogrify(): debug_msg_ref = [None]
411
412 2148 aaronmk
    try:
413 2464 aaronmk
        try:
414 2794 aaronmk
            def run(): return db.run_query(query, cacheable, log_level,
415 2793 aaronmk
                debug_msg_ref, **kw_args)
416 2796 aaronmk
            if recover and not db.is_cached(query):
417 2464 aaronmk
                return with_savepoint(db, run)
418
            else: return run() # don't need savepoint if cached
419
        except Exception, e:
420
            if not recover: raise # need savepoint to run index_cols()
421
            msg = exc.str_(e)
422
423
            match = re.search(r'duplicate key value violates unique constraint '
424 2493 aaronmk
                r'"((_?[^\W_]+)_.+?)"', msg)
425 2464 aaronmk
            if match:
426
                constraint, table = match.groups()
427
                try: cols = index_cols(db, table, constraint)
428
                except NotImplementedError: raise e
429
                else: raise DuplicateKeyException(constraint, cols, e)
430
431 2493 aaronmk
            match = re.search(r'null value in column "(.+?)" violates not-null'
432 2464 aaronmk
                r' constraint', msg)
433
            if match: raise NullValueException('NOT NULL', [match.group(1)], e)
434
435
            match = re.search(r'\b(?:invalid input (?:syntax|value)\b.*?'
436
                r'|date/time field value out of range): "(.+?)"\n'
437 2535 aaronmk
                r'(?:(?s).*?)\bfunction "(.+?)"', msg)
438 2464 aaronmk
            if match:
439
                value, name = match.groups()
440
                raise FunctionValueException(name, strings.to_unicode(value), e)
441
442 2526 aaronmk
            match = re.search(r'column "(.+?)" is of type (.+?) but expression '
443 2523 aaronmk
                r'is of type', msg)
444
            if match:
445
                col, type_ = match.groups()
446
                raise MissingCastException(type_, col, e)
447
448 2493 aaronmk
            match = re.search(r'relation "(.+?)" already exists', msg)
449 2464 aaronmk
            if match: raise DuplicateTableException(match.group(1), e)
450
451 2493 aaronmk
            match = re.search(r'function "(.+?)" already exists', msg)
452 2464 aaronmk
            if match: raise DuplicateFunctionException(match.group(1), e)
453
454
            raise # no specific exception raised
455
    except log_ignore_excs:
456
        log_level += 2
457
        raise
458
    finally:
459 2666 aaronmk
        if debug_msg_ref != None and debug_msg_ref[0] != None:
460
            db.log_debug(debug_msg_ref[0], log_level)
461 830 aaronmk
462 832 aaronmk
##### Basic queries
463
464 2153 aaronmk
def next_version(name):
465 2163 aaronmk
    version = 1 # first existing name was version 0
466 2586 aaronmk
    match = re.match(r'^(.*)#(\d+)$', name)
467 2153 aaronmk
    if match:
468 2586 aaronmk
        name, version = match.groups()
469
        version = int(version)+1
470 2588 aaronmk
    return sql_gen.add_suffix(name, '#'+str(version))
471 2153 aaronmk
472 2899 aaronmk
def lock_table(db, table, mode):
473
    table = sql_gen.as_Table(table)
474
    run_query(db, 'LOCK TABLE '+table.to_str(db)+' IN '+mode+' MODE')
475
476 2789 aaronmk
def run_query_into(db, query, into=None, add_indexes_=False, **kw_args):
477 2085 aaronmk
    '''Outputs a query to a temp table.
478
    For params, see run_query().
479
    '''
480 2789 aaronmk
    if into == None: return run_query(db, query, **kw_args)
481 2790 aaronmk
482
    assert isinstance(into, sql_gen.Table)
483
484
    kw_args['recover'] = True
485
    kw_args.setdefault('log_ignore_excs', (DuplicateTableException,))
486
487 2916 aaronmk
    temp = not db.debug_temp # tables are permanent in debug_temp mode
488 2790 aaronmk
    # "temporary tables cannot specify a schema name", so remove schema
489
    if temp: into.schema = None
490
491
    # Create table
492
    while True:
493
        create_query = 'CREATE'
494
        if temp: create_query += ' TEMP'
495
        create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
496 2385 aaronmk
497 2790 aaronmk
        try:
498
            cur = run_query(db, create_query, **kw_args)
499
                # CREATE TABLE AS sets rowcount to # rows in query
500
            break
501
        except DuplicateTableException, e:
502
            into.name = next_version(into.name)
503
            # try again with next version of name
504
505
    if add_indexes_: add_indexes(db, into)
506
507
    return cur
508 2085 aaronmk
509 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
510
511 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
512
513 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
514 2293 aaronmk
    start=None, order_by=order_by_pkey, default_table=None):
515 1981 aaronmk
    '''
516 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
517 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
518 1981 aaronmk
    @param fields Use None to select all fields in the table
519 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
520 2379 aaronmk
        * container can be any iterable type
521 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
522
        * compare_right_side: sql_gen.ValueCond|literal value
523 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
524
        use all columns
525 2786 aaronmk
    @return query
526 1981 aaronmk
    '''
527 2315 aaronmk
    # Parse tables param
528 2121 aaronmk
    if not lists.is_seq(tables): tables = [tables]
529 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
530 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
531 2121 aaronmk
532 2315 aaronmk
    # Parse other params
533 2376 aaronmk
    if conds == None: conds = []
534 2650 aaronmk
    elif dicts.is_dict(conds): conds = conds.items()
535 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
536 135 aaronmk
    assert limit == None or type(limit) == int
537 865 aaronmk
    assert start == None or type(start) == int
538 2315 aaronmk
    if order_by is order_by_pkey:
539
        if distinct_on != []: order_by = None
540
        else: order_by = pkey(db, table0, recover=True)
541 865 aaronmk
542 2315 aaronmk
    query = 'SELECT'
543 2056 aaronmk
544 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
545 2056 aaronmk
546 2200 aaronmk
    # DISTINCT ON columns
547 2233 aaronmk
    if distinct_on != []:
548 2467 aaronmk
        query += '\nDISTINCT'
549 2254 aaronmk
        if distinct_on is not distinct_on_all:
550 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
551
552
    # Columns
553 2467 aaronmk
    query += '\n'
554 1135 aaronmk
    if fields == None: query += '*'
555 2765 aaronmk
    else:
556
        assert fields != []
557
        query += '\n, '.join(map(parse_col, fields))
558 2200 aaronmk
559
    # Main table
560 2467 aaronmk
    query += '\nFROM '+table0.to_str(db)
561 865 aaronmk
562 2122 aaronmk
    # Add joins
563 2271 aaronmk
    left_table = table0
564 2263 aaronmk
    for join_ in tables:
565
        table = join_.table
566 2238 aaronmk
567 2343 aaronmk
        # Parse special values
568
        if join_.type_ is sql_gen.filter_out: # filter no match
569 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
570 2853 aaronmk
                sql_gen.CompareCond(None, '~=')))
571 2343 aaronmk
572 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
573 2122 aaronmk
574
        left_table = table
575
576 865 aaronmk
    missing = True
577 2376 aaronmk
    if conds != []:
578 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
579
        else: whitespace = '\n'
580 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
581
            .to_str(db) for l, r in conds], 'WHERE')
582 865 aaronmk
        missing = False
583 2227 aaronmk
    if order_by != None:
584 2467 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
585
    if limit != None: query += '\nLIMIT '+str(limit); missing = False
586 865 aaronmk
    if start != None:
587 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
588 865 aaronmk
        missing = False
589
    if missing: warnings.warn(DbWarning(
590
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
591
592 2786 aaronmk
    return query
593 11 aaronmk
594 2054 aaronmk
def select(db, *args, **kw_args):
595
    '''For params, see mk_select() and run_query()'''
596
    recover = kw_args.pop('recover', None)
597
    cacheable = kw_args.pop('cacheable', True)
598 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
599 2054 aaronmk
600 2791 aaronmk
    return run_query(db, mk_select(db, *args, **kw_args), recover, cacheable,
601
        log_level=log_level)
602 2054 aaronmk
603 2788 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, returning=None,
604
    embeddable=False):
605 1960 aaronmk
    '''
606
    @param returning str|None An inserted column (such as pkey) to return
607 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
608 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
609
        query will be fully cached, not just if it raises an exception.
610 1960 aaronmk
    '''
611 2754 aaronmk
    table = sql_gen.remove_table_rename(sql_gen.as_Table(table))
612 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
613 2646 aaronmk
    if cols != None:
614
        cols = [sql_gen.to_name_only_col(v, table).to_str(db) for v in cols]
615 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
616 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
617 2063 aaronmk
618
    # Build query
619 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
620
    query = first_line
621 2467 aaronmk
    if cols != None: query += '\n('+', '.join(cols)+')'
622
    query += '\n'+select_query
623 2063 aaronmk
624
    if returning != None:
625 2743 aaronmk
        query += '\nRETURNING '+sql_gen.to_name_only_col(returning).to_str(db)
626 2063 aaronmk
627 2070 aaronmk
    if embeddable:
628 2327 aaronmk
        assert returning != None
629
630 2070 aaronmk
        # Create function
631 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
632 2327 aaronmk
        return_type = 'SETOF '+returning.to_str(db)+'%TYPE'
633 2189 aaronmk
        while True:
634
            try:
635 2916 aaronmk
                function = sql_gen.TempFunction(function_name, db.debug_temp)
636 2194 aaronmk
637 2189 aaronmk
                function_query = '''\
638 2698 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''()
639 2467 aaronmk
RETURNS '''+return_type+'''
640
LANGUAGE sql
641
AS $$
642 2788 aaronmk
'''+query+''';
643 2467 aaronmk
$$;
644 2070 aaronmk
'''
645 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
646 2464 aaronmk
                    log_ignore_excs=(DuplicateFunctionException,))
647 2189 aaronmk
                break # this version was successful
648
            except DuplicateFunctionException, e:
649
                function_name = next_version(function_name)
650
                # try again with next version of name
651 2070 aaronmk
652 2337 aaronmk
        # Return query that uses function
653 2698 aaronmk
        func_table = sql_gen.NamedTable('f', sql_gen.FunctionCall(function),
654 2742 aaronmk
            [returning]) # AS clause requires function alias
655 2787 aaronmk
        return mk_select(db, func_table, start=0, order_by=None)
656 2070 aaronmk
657 2787 aaronmk
    return query
658 2066 aaronmk
659
def insert_select(db, *args, **kw_args):
660 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
661 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
662
        values in
663 2072 aaronmk
    '''
664 2386 aaronmk
    into = kw_args.pop('into', None)
665
    if into != None: kw_args['embeddable'] = True
666 2066 aaronmk
    recover = kw_args.pop('recover', None)
667
    cacheable = kw_args.pop('cacheable', True)
668 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
669 2066 aaronmk
670 2789 aaronmk
    return run_query_into(db, mk_insert_select(db, *args, **kw_args), into,
671 2787 aaronmk
        recover=recover, cacheable=cacheable, log_level=log_level)
672 2063 aaronmk
673 2738 aaronmk
default = sql_gen.default # tells insert() to use the default value for a column
674 2066 aaronmk
675 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
676 2085 aaronmk
    '''For params, see insert_select()'''
677 1960 aaronmk
    if lists.is_seq(row): cols = None
678
    else:
679
        cols = row.keys()
680
        row = row.values()
681 2738 aaronmk
    row = list(row) # ensure that "== []" works
682 1960 aaronmk
683 2738 aaronmk
    if row == []: query = None
684
    else: query = sql_gen.Values(row).to_str(db)
685 1961 aaronmk
686 2788 aaronmk
    return insert_select(db, table, cols, query, *args, **kw_args)
687 11 aaronmk
688 2402 aaronmk
def mk_update(db, table, changes=None, cond=None):
689
    '''
690
    @param changes [(col, new_value),...]
691
        * container can be any iterable type
692
        * col: sql_gen.Code|str (for col name)
693
        * new_value: sql_gen.Code|literal value
694
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
695
    @return str query
696
    '''
697
    query = 'UPDATE '+sql_gen.as_Table(table).to_str(db)+'\nSET\n'
698 2405 aaronmk
    query += ',\n'.join((sql_gen.to_name_only_col(col, table).to_str(db)+' = '
699 2402 aaronmk
        +sql_gen.as_Value(new_value).to_str(db) for col, new_value in changes))
700 2467 aaronmk
    if cond != None: query += '\nWHERE\n'+cond.to_str(db)
701 2402 aaronmk
702
    return query
703
704
def update(db, *args, **kw_args):
705
    '''For params, see mk_update() and run_query()'''
706
    recover = kw_args.pop('recover', None)
707
708 2791 aaronmk
    return run_query(db, mk_update(db, *args, **kw_args), recover)
709 2402 aaronmk
710 135 aaronmk
def last_insert_id(db):
711 1849 aaronmk
    module = util.root_module(db.db)
712 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
713
    elif module == 'MySQLdb': return db.insert_id()
714
    else: return None
715 13 aaronmk
716 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
717 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
718 2415 aaronmk
    to names that will be distinct among the columns' tables.
719 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
720 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
721
    @param into The table for the new columns.
722 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
723
        columns will be included in the mapping even if they are not in cols.
724
        The tables of the provided Col objects will be changed to into, so make
725
        copies of them if you want to keep the original tables.
726
    @param as_items Whether to return a list of dict items instead of a dict
727 2383 aaronmk
    @return dict(orig_col=new_col, ...)
728
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
729 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
730
        * All mappings use the into table so its name can easily be
731 2383 aaronmk
          changed for all columns at once
732
    '''
733 2415 aaronmk
    cols = lists.uniqify(cols)
734
735 2394 aaronmk
    items = []
736 2389 aaronmk
    for col in preserve:
737 2390 aaronmk
        orig_col = copy.copy(col)
738 2392 aaronmk
        col.table = into
739 2394 aaronmk
        items.append((orig_col, col))
740
    preserve = set(preserve)
741
    for col in cols:
742 2716 aaronmk
        if col not in preserve:
743
            items.append((col, sql_gen.Col(str(col), into, col.srcs)))
744 2394 aaronmk
745
    if not as_items: items = dict(items)
746
    return items
747 2383 aaronmk
748 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
749 2391 aaronmk
    '''For params, see mk_flatten_mapping()
750
    @return See return value of mk_flatten_mapping()
751
    '''
752 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
753
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
754 2786 aaronmk
    run_query_into(db, mk_select(db, joins, cols, limit=limit, start=start),
755 2846 aaronmk
        into=into, add_indexes_=True)
756 2394 aaronmk
    return dict(items)
757 2391 aaronmk
758 2751 aaronmk
def mk_track_data_error(db, errors_table, cols, value, error_code, error):
759 2810 aaronmk
    assert cols != ()
760
761 2751 aaronmk
    cols = map(sql_gen.to_name_only_col, cols)
762
763
    columns_cols = ['column']
764
    columns = sql_gen.NamedValues('columns', columns_cols,
765
        [[c.name] for c in cols])
766
    values_cols = ['value', 'error_code', 'error']
767
    values = sql_gen.NamedValues('values', values_cols,
768
        [value, error_code, error])
769
770
    select_cols = columns_cols+values_cols
771
    name_only_cols = map(sql_gen.to_name_only_col, select_cols)
772 2755 aaronmk
    errors_table = sql_gen.NamedTable('errors', errors_table)
773 2751 aaronmk
    joins = [columns, sql_gen.Join(values, type_='CROSS'),
774
        sql_gen.Join(errors_table, dict(zip(name_only_cols, select_cols)),
775
        sql_gen.filter_out)]
776
777
    return mk_insert_select(db, errors_table, name_only_cols,
778 2787 aaronmk
        mk_select(db, joins, select_cols, order_by=None))
779 2751 aaronmk
780 2810 aaronmk
def track_data_error(db, errors_table, cols, *args, **kw_args):
781 2758 aaronmk
    '''
782
    @param errors_table If None, does nothing.
783
    '''
784 2810 aaronmk
    if errors_table == None or cols == (): return
785
    run_query(db, mk_track_data_error(db, errors_table, cols, *args, **kw_args),
786 2758 aaronmk
        cacheable=True, log_level=4)
787
788 2704 aaronmk
def cast(db, type_, col, errors_table=None):
789
    '''Casts an (unrenamed) column or value.
790 2769 aaronmk
    If errors_table set and col has srcs, saves errors in errors_table (using
791
    col's srcs attr as the source columns) and converts errors to warnings.
792 2820 aaronmk
    @param col str|sql_gen.Col|sql_gen.Literal
793 2704 aaronmk
    @param errors_table None|sql_gen.Table|str
794
    '''
795 2820 aaronmk
    col = sql_gen.as_Col(col)
796 2769 aaronmk
    save_errors = (errors_table != None and isinstance(col, sql_gen.Col)
797
        and col.srcs != ())
798
    if not save_errors: # can't save errors
799
        return sql_gen.CustomCode(col.to_str(db)+'::'+type_) # just cast
800 2704 aaronmk
801
    assert not isinstance(col, sql_gen.NamedCol)
802
803 2770 aaronmk
    errors_table = sql_gen.as_Table(errors_table)
804
    srcs = map(sql_gen.to_name_only_col, col.srcs)
805
    function_name = str(sql_gen.FunctionCall(type_, *srcs))
806 2916 aaronmk
    function = sql_gen.TempFunction(function_name, db.debug_temp)
807 2704 aaronmk
808 2718 aaronmk
    while True:
809
        # Create function definition
810
        query = '''\
811 2704 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''(value text)
812
RETURNS '''+type_+'''
813
LANGUAGE plpgsql
814
STRICT
815
AS $$
816
BEGIN
817
    /* The explicit cast to the return type is needed to make the cast happen
818
    inside the try block. (Implicit casts to the return type happen at the end
819
    of the function, outside any block.) */
820
    RETURN value::'''+type_+''';
821
EXCEPTION
822
    WHEN data_exception THEN
823
        -- Save error in errors table.
824 2751 aaronmk
        -- Insert the value and error for *each* source column.
825
'''+mk_track_data_error(db, errors_table, srcs,
826
    *map(sql_gen.CustomCode, ['value', 'SQLSTATE', 'SQLERRM']))+''';
827 2704 aaronmk
828
        RAISE WARNING '%', SQLERRM;
829
        RETURN NULL;
830
END;
831
$$;
832
'''
833 2718 aaronmk
834
        # Create function
835
        try:
836
            run_query(db, query, recover=True, cacheable=True,
837
                log_ignore_excs=(DuplicateFunctionException,))
838
            break # successful
839
        except DuplicateFunctionException:
840 2770 aaronmk
            function.name = next_version(function.name)
841
            # try again with next version of name
842 2704 aaronmk
843
    return sql_gen.FunctionCall(function, col)
844
845 2414 aaronmk
##### Database structure queries
846
847 2426 aaronmk
def table_row_count(db, table, recover=None):
848 2786 aaronmk
    return value(run_query(db, mk_select(db, table, [sql_gen.row_count],
849 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
850 2426 aaronmk
851 2414 aaronmk
def table_cols(db, table, recover=None):
852
    return list(col_names(select(db, table, limit=0, order_by=None,
853 2443 aaronmk
        recover=recover, log_level=4)))
854 2414 aaronmk
855 2291 aaronmk
def pkey(db, table, recover=None):
856 832 aaronmk
    '''Assumed to be first column in table'''
857 2339 aaronmk
    return table_cols(db, table, recover)[0]
858 832 aaronmk
859 2559 aaronmk
not_null_col = 'not_null_col'
860 2340 aaronmk
861
def table_not_null_col(db, table, recover=None):
862
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
863
    if not_null_col in table_cols(db, table, recover): return not_null_col
864
    else: return pkey(db, table, recover)
865
866 853 aaronmk
def index_cols(db, table, index):
867
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
868
    automatically created. When you don't know whether something is a UNIQUE
869
    constraint or a UNIQUE index, use this function.'''
870 1909 aaronmk
    module = util.root_module(db.db)
871
    if module == 'psycopg2':
872
        return list(values(run_query(db, '''\
873 853 aaronmk
SELECT attname
874 866 aaronmk
FROM
875
(
876
        SELECT attnum, attname
877
        FROM pg_index
878
        JOIN pg_class index ON index.oid = indexrelid
879
        JOIN pg_class table_ ON table_.oid = indrelid
880
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
881
        WHERE
882 2782 aaronmk
            table_.relname = '''+db.esc_value(table)+'''
883
            AND index.relname = '''+db.esc_value(index)+'''
884 866 aaronmk
    UNION
885
        SELECT attnum, attname
886
        FROM
887
        (
888
            SELECT
889
                indrelid
890
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
891
                    AS indkey
892
            FROM pg_index
893
            JOIN pg_class index ON index.oid = indexrelid
894
            JOIN pg_class table_ ON table_.oid = indrelid
895
            WHERE
896 2782 aaronmk
                table_.relname = '''+db.esc_value(table)+'''
897
                AND index.relname = '''+db.esc_value(index)+'''
898 866 aaronmk
        ) s
899
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
900
) s
901 853 aaronmk
ORDER BY attnum
902 2782 aaronmk
'''
903
            , cacheable=True, log_level=4)))
904 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
905
        ' database')
906 853 aaronmk
907 464 aaronmk
def constraint_cols(db, table, constraint):
908 1849 aaronmk
    module = util.root_module(db.db)
909 464 aaronmk
    if module == 'psycopg2':
910
        return list(values(run_query(db, '''\
911
SELECT attname
912
FROM pg_constraint
913
JOIN pg_class ON pg_class.oid = conrelid
914
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
915
WHERE
916 2783 aaronmk
    relname = '''+db.esc_value(table)+'''
917
    AND conname = '''+db.esc_value(constraint)+'''
918 464 aaronmk
ORDER BY attnum
919 2783 aaronmk
'''
920
            )))
921 464 aaronmk
    else: raise NotImplementedError("Can't list constraint columns for "+module+
922
        ' database')
923
924 2096 aaronmk
row_num_col = '_row_num'
925
926 2847 aaronmk
def add_index(db, exprs, table=None, unique=False, ensure_not_null=True):
927 2688 aaronmk
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
928 2538 aaronmk
    Currently, only function calls are supported as expressions.
929 2847 aaronmk
    @param ensure_not_null If set, translates NULL values to sentinel values.
930
        This allows indexes to be used for comparisons where NULLs are equal.
931 2538 aaronmk
    '''
932 2688 aaronmk
    if not lists.is_seq(exprs): exprs = [exprs]
933 2538 aaronmk
934 2688 aaronmk
    # Parse exprs
935
    old_exprs = exprs[:]
936
    exprs = []
937
    cols = []
938
    for i, expr in enumerate(old_exprs):
939
        expr = copy.deepcopy(expr) # don't modify input!
940 2823 aaronmk
        expr = sql_gen.as_Col(expr, table)
941 2688 aaronmk
942 2823 aaronmk
        # Handle nullable columns
943 2860 aaronmk
        if ensure_not_null:
944
            try: expr = sql_gen.ensure_not_null(db, expr)
945
            except KeyError: pass # unknown type, so just create plain index
946 2823 aaronmk
947 2688 aaronmk
        # Extract col
948
        if isinstance(expr, sql_gen.FunctionCall):
949
            col = expr.args[0]
950
            expr = sql_gen.Expr(expr)
951
        else: col = expr
952 2823 aaronmk
        assert isinstance(col, sql_gen.Col)
953 2688 aaronmk
954
        # Extract table
955
        if table == None:
956
            assert sql_gen.is_table_col(col)
957
            table = col.table
958
959
        col.table = None
960
961
        exprs.append(expr)
962
        cols.append(col)
963 2408 aaronmk
964 2688 aaronmk
    table = sql_gen.as_Table(table)
965
    index = sql_gen.Table(str(sql_gen.Col(','.join(map(str, cols)), table)))
966
967
    str_ = 'CREATE'
968
    if unique: str_ += ' UNIQUE'
969
    str_ += ' INDEX '+index.to_str(db)+' ON '+table.to_str(db)+' ('+(
970
        ', '.join((v.to_str(db) for v in exprs)))+')'
971
972
    try: run_query(db, str_, recover=True, cacheable=True, log_level=3)
973 2408 aaronmk
    except DuplicateTableException: pass # index already existed
974
975 2686 aaronmk
def add_pkey(db, table, cols=None, recover=None):
976
    '''Adds a primary key.
977
    @param cols [sql_gen.Col,...] The columns in the primary key.
978
        Defaults to the first column in the table.
979 2406 aaronmk
    @pre The table must not already have a primary key.
980
    '''
981
    table = sql_gen.as_Table(table)
982 2686 aaronmk
    if cols == None: cols = [pkey(db, table, recover)]
983
    col_strs = [sql_gen.to_name_only_col(v).to_str(db) for v in cols]
984 2406 aaronmk
985 2781 aaronmk
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD PRIMARY KEY ('
986
        +(', '.join(col_strs))+')', recover=True, cacheable=True, log_level=3,
987
        log_ignore_excs=(DuplicateTableException,))
988 2406 aaronmk
989 2772 aaronmk
already_indexed = object() # tells add_indexes() the pkey has already been added
990
991
def add_indexes(db, table, has_pkey=True):
992
    '''Adds an index on all columns in a table.
993
    @param has_pkey bool|already_indexed Whether a pkey instead of a regular
994
        index should be added on the first column.
995
        * If already_indexed, the pkey is assumed to have already been added
996
    '''
997
    cols = table_cols(db, table)
998
    if has_pkey:
999
        if has_pkey is not already_indexed: add_pkey(db, table)
1000
        cols = cols[1:]
1001
    for col in cols: add_index(db, col, table)
1002
1003 2086 aaronmk
def add_row_num(db, table):
1004 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
1005
    be the primary key.'''
1006 2320 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
1007 2096 aaronmk
    run_query(db, 'ALTER TABLE '+table+' ADD COLUMN '+row_num_col
1008 2443 aaronmk
        +' serial NOT NULL PRIMARY KEY', log_level=3)
1009 2086 aaronmk
1010 2889 aaronmk
def drop_table(db, table):
1011
    table = sql_gen.as_Table(table)
1012
    return run_query(db, 'DROP TABLE IF EXISTS '+table.to_str(db)+' CASCADE')
1013
1014 2760 aaronmk
def create_table(db, table, cols, has_pkey=True, col_indexes=True):
1015 2675 aaronmk
    '''Creates a table.
1016 2681 aaronmk
    @param cols [sql_gen.TypedCol,...] The column names and types
1017
    @param has_pkey If set, the first column becomes the primary key.
1018 2760 aaronmk
    @param col_indexes bool|[ref]
1019
        * If True, indexes will be added on all non-pkey columns.
1020
        * If a list reference, [0] will be set to a function to do this.
1021
          This can be used to delay index creation until the table is populated.
1022 2675 aaronmk
    '''
1023
    table = sql_gen.as_Table(table)
1024
1025 2681 aaronmk
    if has_pkey:
1026
        cols[0] = pkey = copy.copy(cols[0]) # don't modify input!
1027 2872 aaronmk
        pkey.constraints = 'PRIMARY KEY'
1028 2681 aaronmk
1029 2675 aaronmk
    str_ = 'CREATE TABLE '+table.to_str(db)+' (\n'
1030 2681 aaronmk
    str_ += '\n, '.join(v.to_str(db) for v in cols)
1031
    str_ += '\n);\n'
1032 2675 aaronmk
    run_query(db, str_, cacheable=True, log_level=2)
1033 2760 aaronmk
1034
    # Add indexes
1035 2773 aaronmk
    if has_pkey: has_pkey = already_indexed
1036
    def add_indexes_(): add_indexes(db, table, has_pkey)
1037
    if isinstance(col_indexes, list): col_indexes[0] = add_indexes_ # defer
1038
    elif col_indexes: add_indexes_() # add now
1039 2675 aaronmk
1040 2684 aaronmk
def vacuum(db, table):
1041
    table = sql_gen.as_Table(table)
1042
    db.with_autocommit(lambda: run_query(db, 'VACUUM ANALYZE '+table.to_str(db),
1043
        log_level=3))
1044
1045 2732 aaronmk
def truncate(db, table, schema='public'):
1046 2777 aaronmk
    table = sql_gen.as_Table(table, schema)
1047 2733 aaronmk
    return run_query(db, 'TRUNCATE '+table.to_str(db)+' CASCADE')
1048 2732 aaronmk
1049 2729 aaronmk
def tables(db, schema_like='public', table_like='%', exact=False):
1050
    if exact: compare = '='
1051
    else: compare = 'LIKE'
1052
1053 1849 aaronmk
    module = util.root_module(db.db)
1054 832 aaronmk
    if module == 'psycopg2':
1055 2779 aaronmk
        conds = [('schemaname', sql_gen.CompareCond(schema_like, compare)),
1056
            ('tablename', sql_gen.CompareCond(table_like, compare))]
1057
        return values(select(db, 'pg_tables', ['tablename'], conds,
1058
            order_by='tablename', log_level=4))
1059 1968 aaronmk
    elif module == 'MySQLdb':
1060 2779 aaronmk
        return values(run_query(db, 'SHOW TABLES LIKE '+db.esc_value(table_like)
1061
            , cacheable=True, log_level=4))
1062 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
1063 830 aaronmk
1064 2726 aaronmk
def table_exists(db, table):
1065
    table = sql_gen.as_Table(table)
1066 2730 aaronmk
    return list(tables(db, table.schema, table.name, exact=True)) != []
1067 2726 aaronmk
1068 2913 aaronmk
def function_exists(db, function):
1069
    function = sql_gen.as_Function(function)
1070
1071
    info_table = sql_gen.Table('routines', 'information_schema')
1072
    conds = [('routine_name', function.name)]
1073
    schema = function.schema
1074
    if schema != None: conds.append(('routine_schema', schema))
1075 2914 aaronmk
    # Exclude trigger functions, since they cannot be called directly
1076
    conds.append(('data_type', sql_gen.CompareCond('trigger', '!=')))
1077 2913 aaronmk
1078
    return list(values(select(db, info_table, ['routine_name'], conds,
1079
        order_by='routine_schema', limit=1, log_level=4))) != []
1080
        # TODO: order_by search_path schema order
1081
1082 2726 aaronmk
def errors_table(db, table, if_exists=True):
1083
    '''
1084
    @param if_exists If set, returns None if the errors table doesn't exist
1085
    @return None|sql_gen.Table
1086
    '''
1087
    table = sql_gen.as_Table(table)
1088
    if table.srcs != (): table = table.srcs[0]
1089
1090
    errors_table = sql_gen.suffixed_table(table, '.errors')
1091
    if if_exists and not table_exists(db, errors_table): return None
1092
    return errors_table
1093
1094 833 aaronmk
##### Database management
1095
1096 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
1097
    '''For kw_args, see tables()'''
1098
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
1099 833 aaronmk
1100 832 aaronmk
##### Heuristic queries
1101
1102 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
1103 1554 aaronmk
    '''Recovers from errors.
1104 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
1105
    '''
1106 2642 aaronmk
    row = sql_gen.ColDict(db, table, row)
1107 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
1108
1109 471 aaronmk
    try:
1110 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
1111 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
1112
            row_ct_ref[0] += cur.rowcount
1113
        return value(cur)
1114 471 aaronmk
    except DuplicateKeyException, e:
1115 2653 aaronmk
        row = sql_gen.ColDict(db, table,
1116
            util.dict_subset_right_join(row, e.cols))
1117
        return value(select(db, table, [pkey_], row, recover=True))
1118 471 aaronmk
1119 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
1120 830 aaronmk
    '''Recovers from errors'''
1121 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
1122 14 aaronmk
    except StopIteration:
1123 40 aaronmk
        if not create: raise
1124 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
1125 2078 aaronmk
1126 2593 aaronmk
def is_func_result(col):
1127
    return col.table.name.find('(') >= 0 and col.name == 'result'
1128
1129 2592 aaronmk
def into_table_name(out_table, in_tables0, mapping, is_func):
1130 2615 aaronmk
    def in_col_str(in_col):
1131
        in_col = sql_gen.remove_col_rename(in_col)
1132
        if isinstance(in_col, sql_gen.Col):
1133
            table = in_col.table
1134
            if table == in_tables0:
1135
                in_col = sql_gen.to_name_only_col(in_col)
1136
            elif is_func_result(in_col): in_col = table # omit col name
1137
        return str(in_col)
1138
1139 2580 aaronmk
    str_ = str(out_table)
1140
    if is_func:
1141 2615 aaronmk
        str_ += '('
1142 2580 aaronmk
1143 2615 aaronmk
        try: value_in_col = mapping['value']
1144
        except KeyError:
1145 2654 aaronmk
            str_ += ', '.join((str(k)+'='+in_col_str(v)
1146 2615 aaronmk
                for k, v in mapping.iteritems()))
1147
        else: str_ += in_col_str(value_in_col)
1148
1149
        str_ += ')'
1150 2812 aaronmk
    else:
1151
        out_col = 'rank'
1152
        try: in_col = mapping[out_col]
1153
        except KeyError: str_ += '_pkeys'
1154
        else: # has a rank column, so hierarchical
1155
            str_ += '['+str(out_col)+'='+in_col_str(in_col)+']'
1156 2580 aaronmk
    return str_
1157
1158 2508 aaronmk
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None, into=None,
1159 2805 aaronmk
    default=None, is_func=False, on_error=exc.raise_):
1160 2078 aaronmk
    '''Recovers from errors.
1161
    Only works under PostgreSQL (uses INSERT RETURNING).
1162 2909 aaronmk
    IMPORTANT: Must be run at the *beginning* of a transaction.
1163 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
1164
        tables to join with it using the main input table's pkey
1165 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
1166 2612 aaronmk
        * out_table_col: str (*not* sql_gen.Col)
1167
        * in_table_col: sql_gen.Col|literal-value
1168 2489 aaronmk
    @param into The table to contain the output and input pkeys.
1169 2574 aaronmk
        Defaults to `out_table.name+'_pkeys'`.
1170 2509 aaronmk
    @param default The *output* column to use as the pkey for missing rows.
1171
        If this output column does not exist in the mapping, uses None.
1172 2552 aaronmk
    @param is_func Whether out_table is the name of a SQL function, not a table
1173 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
1174 2078 aaronmk
    '''
1175 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
1176 2312 aaronmk
1177 2450 aaronmk
    def log_debug(msg): db.log_debug(msg, level=1.5)
1178 2505 aaronmk
    def col_ustr(str_):
1179 2567 aaronmk
        return strings.repr_no_u(sql_gen.remove_col_rename(str_))
1180 2450 aaronmk
1181 2910 aaronmk
    log_debug('********** New iteration **********')
1182
    log_debug('Inserting these input columns into '+strings.as_tt(
1183
        out_table.to_str(db))+':\n'+strings.as_table(mapping, ustr=col_ustr))
1184
1185
    log_debug('Locking output table to prevent concurrent duplicate keys')
1186
    # Must happen before any SELECT query on the table to avoid lock upgrades
1187
    lock_table(db, out_table, 'EXCLUSIVE')
1188
1189 2768 aaronmk
    out_pkey = pkey(db, out_table, recover=True)
1190
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
1191
1192
    if mapping == {}: # need at least one column for INSERT SELECT
1193
        mapping = {out_pkey: None} # ColDict will replace with default value
1194
1195 2382 aaronmk
    # Create input joins from list of input tables
1196
    in_tables_ = in_tables[:] # don't modify input!
1197
    in_tables0 = in_tables_.pop(0) # first table is separate
1198 2735 aaronmk
    errors_table_ = errors_table(db, in_tables0)
1199 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
1200 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
1201 2460 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v,
1202
        {in_pkey: sql_gen.join_same_not_null}) for v in in_tables_]
1203 2131 aaronmk
1204 2592 aaronmk
    if into == None:
1205
        into = into_table_name(out_table, in_tables0, mapping, is_func)
1206
    into = sql_gen.as_Table(into)
1207
1208 2702 aaronmk
    # Set column sources
1209
    in_cols = filter(sql_gen.is_table_col, mapping.values())
1210
    for col in in_cols:
1211
        if col.table == in_tables0: col.set_srcs(sql_gen.src_self)
1212
1213 2486 aaronmk
    log_debug('Joining together input tables into temp table')
1214 2395 aaronmk
    # Place in new table for speed and so don't modify input if values edited
1215 2584 aaronmk
    in_table = sql_gen.Table('in')
1216 2702 aaronmk
    mapping = dicts.join(mapping, flatten(db, in_table, input_joins, in_cols,
1217
        preserve=[in_pkey_col], start=0))
1218 2395 aaronmk
    input_joins = [in_table]
1219 2486 aaronmk
    db.log_debug('Temp table: '+strings.as_tt(in_table.to_str(db)), level=2)
1220 2395 aaronmk
1221 2656 aaronmk
    mapping = sql_gen.ColDict(db, out_table, mapping)
1222
        # after applying dicts.join() because that returns a plain dict
1223
1224 2509 aaronmk
    # Resolve default value column
1225 2841 aaronmk
    if default != None:
1226
        try: default = mapping[default]
1227
        except KeyError:
1228 2509 aaronmk
            db.log_debug('Default value column '
1229
                +strings.as_tt(strings.repr_no_u(default))
1230 2511 aaronmk
                +' does not exist in mapping, falling back to None', level=2.1)
1231 2509 aaronmk
            default = None
1232
1233 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
1234 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
1235
1236 2201 aaronmk
    pkeys_table_exists_ref = [False]
1237 2420 aaronmk
    def insert_into_pkeys(joins, cols):
1238 2786 aaronmk
        query = mk_select(db, joins, cols, order_by=None, start=0)
1239 2201 aaronmk
        if pkeys_table_exists_ref[0]:
1240 2786 aaronmk
            insert_select(db, into, pkeys_names, query)
1241 2201 aaronmk
        else:
1242 2786 aaronmk
            run_query_into(db, query, into=into)
1243 2201 aaronmk
            pkeys_table_exists_ref[0] = True
1244
1245 2429 aaronmk
    limit_ref = [None]
1246 2380 aaronmk
    conds = set()
1247 2811 aaronmk
    distinct_on = sql_gen.ColDict(db, out_table)
1248 2325 aaronmk
    def mk_main_select(joins, cols):
1249 2811 aaronmk
        distinct_on_cols = [c.to_Col() for c in distinct_on.values()]
1250
        return mk_select(db, joins, cols, conds, distinct_on_cols,
1251 2429 aaronmk
            limit=limit_ref[0], start=0)
1252 2132 aaronmk
1253 2519 aaronmk
    exc_strs = set()
1254 2309 aaronmk
    def log_exc(e):
1255 2519 aaronmk
        e_str = exc.str_(e, first_line_only=True)
1256
        log_debug('Caught exception: '+e_str)
1257
        assert e_str not in exc_strs # avoid infinite loops
1258
        exc_strs.add(e_str)
1259 2735 aaronmk
1260 2451 aaronmk
    def remove_all_rows():
1261 2896 aaronmk
        log_debug('Ignoring all rows')
1262 2429 aaronmk
        limit_ref[0] = 0 # just create an empty pkeys table
1263 2735 aaronmk
1264 2758 aaronmk
    def ignore(in_col, value, e):
1265
        track_data_error(db, errors_table_, in_col.srcs, value, e.cause.pgcode,
1266
            e.cause.pgerror)
1267 2895 aaronmk
        log_debug('Ignoring rows with '+strings.as_tt(repr(in_col))+' = '
1268 2545 aaronmk
            +strings.as_tt(repr(value)))
1269 2895 aaronmk
1270 2758 aaronmk
    def remove_rows(in_col, value, e):
1271
        ignore(in_col, value, e)
1272 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
1273
        assert cond not in conds # avoid infinite loops
1274 2380 aaronmk
        conds.add(cond)
1275 2895 aaronmk
1276 2758 aaronmk
    def invalid2null(in_col, value, e):
1277
        ignore(in_col, value, e)
1278 2403 aaronmk
        update(db, in_table, [(in_col, None)],
1279
            sql_gen.ColValueCond(in_col, value))
1280 2245 aaronmk
1281 2589 aaronmk
    def insert_pkeys_table(which):
1282
        return sql_gen.Table(sql_gen.add_suffix(in_table.name,
1283
            '_insert_'+which+'_pkeys'))
1284
    insert_out_pkeys = insert_pkeys_table('out')
1285
    insert_in_pkeys = insert_pkeys_table('in')
1286
1287 2206 aaronmk
    # Do inserts and selects
1288 2642 aaronmk
    join_cols = sql_gen.ColDict(db, out_table)
1289 2206 aaronmk
    while True:
1290 2521 aaronmk
        if limit_ref[0] == 0: # special case
1291
            log_debug('Creating an empty pkeys table')
1292 2786 aaronmk
            cur = run_query_into(db, mk_select(db, out_table, [out_pkey],
1293 2521 aaronmk
                limit=limit_ref[0]), into=insert_out_pkeys)
1294
            break # don't do main case
1295
1296 2303 aaronmk
        has_joins = join_cols != {}
1297
1298 2903 aaronmk
        log_debug('Trying to insert new rows')
1299
1300 2305 aaronmk
        # Prepare to insert new rows
1301 2325 aaronmk
        insert_joins = input_joins[:] # don't modify original!
1302 2403 aaronmk
        insert_args = dict(recover=True, cacheable=False)
1303 2303 aaronmk
        if has_joins:
1304 2325 aaronmk
            insert_joins.append(sql_gen.Join(out_table, join_cols,
1305
                sql_gen.filter_out))
1306
        else:
1307 2404 aaronmk
            insert_args.update(dict(returning=out_pkey, into=insert_out_pkeys))
1308 2786 aaronmk
        main_select = mk_main_select(insert_joins, mapping.values())
1309 2303 aaronmk
1310 2206 aaronmk
        try:
1311 2518 aaronmk
            cur = insert_select(db, out_table, mapping.keys(), main_select,
1312
                **insert_args)
1313 2357 aaronmk
            break # insert successful
1314 2206 aaronmk
        except DuplicateKeyException, e:
1315 2309 aaronmk
            log_exc(e)
1316
1317 2258 aaronmk
            old_join_cols = join_cols.copy()
1318 2811 aaronmk
            distinct_on.update(util.dict_subset(mapping, e.cols))
1319 2565 aaronmk
            join_cols.update(util.dict_subset_right_join(mapping, e.cols))
1320 2486 aaronmk
            log_debug('Ignoring existing rows, comparing on these columns:\n'
1321 2505 aaronmk
                +strings.as_inline_table(join_cols, ustr=col_ustr))
1322 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
1323 2230 aaronmk
        except NullValueException, e:
1324 2309 aaronmk
            log_exc(e)
1325
1326 2230 aaronmk
            out_col, = e.cols
1327
            try: in_col = mapping[out_col]
1328 2356 aaronmk
            except KeyError:
1329 2486 aaronmk
                log_debug('Missing mapping for NOT NULL column '+out_col)
1330 2451 aaronmk
                remove_all_rows()
1331 2758 aaronmk
            else: remove_rows(in_col, None, e)
1332 2542 aaronmk
        except FunctionValueException, e:
1333
            log_exc(e)
1334
1335
            func_name = e.name
1336
            value = e.value
1337
            for out_col, in_col in mapping.iteritems():
1338 2758 aaronmk
                in_col = sql_gen.unwrap_func_call(in_col, func_name)
1339
                invalid2null(in_col, value, e)
1340 2525 aaronmk
        except MissingCastException, e:
1341
            log_exc(e)
1342
1343
            out_col = e.col
1344 2706 aaronmk
            type_ = e.type
1345
1346 2724 aaronmk
            log_debug('Casting '+strings.as_tt(out_col)+' input to '
1347
                +strings.as_tt(type_))
1348 2735 aaronmk
            def wrap_func(col): return cast(db, type_, col, errors_table_)
1349 2705 aaronmk
            mapping[out_col] = sql_gen.wrap(wrap_func, mapping[out_col])
1350 2429 aaronmk
        except DatabaseErrors, e:
1351
            log_exc(e)
1352
1353 2808 aaronmk
            log_debug('No handler for exception')
1354
            on_error(e)
1355 2451 aaronmk
            remove_all_rows()
1356 2358 aaronmk
        # after exception handled, rerun loop with additional constraints
1357 2132 aaronmk
1358 2357 aaronmk
    if row_ct_ref != None and cur.rowcount >= 0:
1359
        row_ct_ref[0] += cur.rowcount
1360
1361
    if has_joins:
1362
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
1363 2486 aaronmk
        log_debug('Getting output table pkeys of existing/inserted rows')
1364 2420 aaronmk
        insert_into_pkeys(select_joins, pkeys_cols)
1365 2357 aaronmk
    else:
1366 2404 aaronmk
        add_row_num(db, insert_out_pkeys) # for joining with input pkeys
1367 2357 aaronmk
1368 2486 aaronmk
        log_debug('Getting input table pkeys of inserted rows')
1369 2786 aaronmk
        run_query_into(db, mk_main_select(input_joins, [in_pkey]),
1370 2404 aaronmk
            into=insert_in_pkeys)
1371
        add_row_num(db, insert_in_pkeys) # for joining with output pkeys
1372 2357 aaronmk
1373 2428 aaronmk
        assert table_row_count(db, insert_out_pkeys) == table_row_count(db,
1374
            insert_in_pkeys)
1375
1376 2486 aaronmk
        log_debug('Combining output and input pkeys in inserted order')
1377 2404 aaronmk
        pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
1378 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
1379 2420 aaronmk
        insert_into_pkeys(pkey_joins, pkeys_names)
1380 2357 aaronmk
1381 2486 aaronmk
    db.log_debug('Adding pkey on pkeys table to enable fast joins', level=2.5)
1382 2594 aaronmk
    add_pkey(db, into)
1383 2407 aaronmk
1384 2508 aaronmk
    log_debug('Setting pkeys of missing rows to '+strings.as_tt(repr(default)))
1385 2489 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(into,
1386 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
1387
        # must use join_same_not_null or query will take forever
1388 2420 aaronmk
    insert_into_pkeys(missing_rows_joins,
1389 2508 aaronmk
        [in_pkey_col, sql_gen.NamedCol(out_pkey, default)])
1390 2357 aaronmk
1391 2489 aaronmk
    assert table_row_count(db, into) == table_row_count(db, in_table)
1392 2428 aaronmk
1393 2702 aaronmk
    srcs = []
1394
    if is_func: srcs = sql_gen.cols_srcs(in_cols)
1395
    return sql_gen.Col(out_pkey, into, srcs)
1396 2115 aaronmk
1397
##### Data cleanup
1398
1399 2290 aaronmk
def cleanup_table(db, table, cols):
1400 2677 aaronmk
    table = sql_gen.as_Table(table)
1401
    cols = map(sql_gen.as_Col, cols)
1402 2115 aaronmk
1403 2677 aaronmk
    expr = ('nullif(nullif(trim(both from %s), '+db.esc_value('')+'), '
1404
        +db.esc_value(r'\N')+')')
1405
    changes = [(v, sql_gen.CustomCode(expr % v.to_str(db)))
1406
        for v in cols]
1407 2115 aaronmk
1408 2677 aaronmk
    update(db, table, changes)