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 2170 aaronmk
def get_cur_query(cur, input_query=None, input_params=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 2371 aaronmk
    else: return '[input] '+strings.ustr(input_query)+' % '+repr(input_params)
27 14 aaronmk
28 2170 aaronmk
def _add_cursor_info(e, *args, **kw_args):
29
    '''For params, see get_cur_query()'''
30
    exc.add_msg(e, 'query: '+str(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 2101 aaronmk
##### Input validation
109
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 1909 aaronmk
def _query_lookup(query, params): return (query, dicts.make_hashable(params))
149 1894 aaronmk
150 2448 aaronmk
log_debug_none = lambda msg, level=2: None
151 1901 aaronmk
152 1849 aaronmk
class DbConn:
153 2190 aaronmk
    def __init__(self, db_config, serializable=True, autocommit=False,
154
        caching=True, log_debug=log_debug_none):
155 1869 aaronmk
        self.db_config = db_config
156
        self.serializable = serializable
157 2190 aaronmk
        self.autocommit = autocommit
158
        self.caching = caching
159 1901 aaronmk
        self.log_debug = log_debug
160 2193 aaronmk
        self.debug = log_debug != log_debug_none
161 1869 aaronmk
162
        self.__db = None
163 1889 aaronmk
        self.query_results = {}
164 2139 aaronmk
        self._savepoint = 0
165 2671 aaronmk
        self._notices_seen = set()
166 1869 aaronmk
167
    def __getattr__(self, name):
168
        if name == '__dict__': raise Exception('getting __dict__')
169
        if name == 'db': return self._db()
170
        else: raise AttributeError()
171
172
    def __getstate__(self):
173
        state = copy.copy(self.__dict__) # shallow copy
174 1915 aaronmk
        state['log_debug'] = None # don't pickle the debug callback
175 1869 aaronmk
        state['_DbConn__db'] = None # don't pickle the connection
176
        return state
177
178 2165 aaronmk
    def connected(self): return self.__db != None
179
180 1869 aaronmk
    def _db(self):
181
        if self.__db == None:
182
            # Process db_config
183
            db_config = self.db_config.copy() # don't modify input!
184 2097 aaronmk
            schemas = db_config.pop('schemas', None)
185 1869 aaronmk
            module_name, mappings = db_engines[db_config.pop('engine')]
186
            module = __import__(module_name)
187
            _add_module(module)
188
            for orig, new in mappings.iteritems():
189
                try: util.rename_key(db_config, orig, new)
190
                except KeyError: pass
191
192
            # Connect
193
            self.__db = module.connect(**db_config)
194
195
            # Configure connection
196 2676 aaronmk
            if self.serializable and not self.autocommit: self.run_query(
197 2678 aaronmk
                'SET TRANSACTION ISOLATION LEVEL SERIALIZABLE', log_level=4)
198 2101 aaronmk
            if schemas != None:
199
                schemas_ = ''.join((esc_name(self, s)+', '
200
                    for s in schemas.split(',')))
201 2676 aaronmk
                self.run_query("SELECT set_config('search_path', \
202
%s || current_setting('search_path'), false)", [schemas_], 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 1894 aaronmk
        def execute(self, query, params=None):
215 1930 aaronmk
            self._is_insert = query.upper().find('INSERT') >= 0
216 1894 aaronmk
            self.query_lookup = _query_lookup(query, params)
217 2148 aaronmk
            try:
218 2191 aaronmk
                try:
219
                    return_value = self.inner.execute(query, params)
220
                    self.outer.do_autocommit()
221 2648 aaronmk
                finally: self.query = get_cur_query(self.inner, query, params)
222 1904 aaronmk
            except Exception, e:
223 2170 aaronmk
                _add_cursor_info(e, self, query, params)
224 1904 aaronmk
                self.result = e # cache the exception as the result
225
                self._cache_result()
226
                raise
227 1930 aaronmk
            # Fetch all rows so result will be cached
228
            if self.rowcount == 0 and not self._is_insert: consume_rows(self)
229 1894 aaronmk
            return return_value
230
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 2665 aaronmk
    def can_mogrify(self):
274 2663 aaronmk
        module = util.root_module(self.db)
275 2665 aaronmk
        return module == 'psycopg2'
276 2663 aaronmk
277 2665 aaronmk
    def mogrify(self, query, params=None):
278
        if self.can_mogrify(): return self.db.cursor().mogrify(query, params)
279
        else: raise NotImplementedError("Can't mogrify query")
280
281 2671 aaronmk
    def print_notices(self):
282 2725 aaronmk
        if hasattr(self.db, 'notices'):
283
            for msg in self.db.notices:
284
                if msg not in self._notices_seen:
285
                    self._notices_seen.add(msg)
286
                    self.log_debug(msg, level=2)
287 2671 aaronmk
288 2445 aaronmk
    def run_query(self, query, params=None, cacheable=False, log_level=2,
289 2464 aaronmk
        debug_msg_ref=None):
290 2445 aaronmk
        '''
291 2464 aaronmk
        @param log_ignore_excs The log_level will be increased by 2 if the query
292
            throws one of these exceptions.
293 2664 aaronmk
        @param debug_msg_ref If specified, the log message will be returned in
294
            this instead of being output. This allows you to filter log messages
295
            depending on the result of the query.
296 2445 aaronmk
        '''
297 2167 aaronmk
        assert query != None
298
299 2047 aaronmk
        if not self.caching: cacheable = False
300 1903 aaronmk
        used_cache = False
301 2664 aaronmk
302
        def log_msg(query):
303
            if used_cache: cache_status = 'cache hit'
304
            elif cacheable: cache_status = 'cache miss'
305
            else: cache_status = 'non-cacheable'
306
            return 'DB query: '+cache_status+':\n'+strings.as_code(query, 'SQL')
307
308 1903 aaronmk
        try:
309 1927 aaronmk
            # Get cursor
310
            if cacheable:
311
                query_lookup = _query_lookup(query, params)
312
                try:
313
                    cur = self.query_results[query_lookup]
314
                    used_cache = True
315
                except KeyError: cur = self.DbCursor(self)
316
            else: cur = self.db.cursor()
317
318 2664 aaronmk
            # Log query
319
            if self.debug and debug_msg_ref == None: # log before running
320
                self.log_debug(log_msg(query), log_level)
321
322 1927 aaronmk
            # Run query
323 2148 aaronmk
            cur.execute(query, params)
324 1903 aaronmk
        finally:
325 2671 aaronmk
            self.print_notices()
326 2664 aaronmk
            if self.debug and debug_msg_ref != None: # return after running
327
                debug_msg_ref[0] = log_msg(str(get_cur_query(cur, query,
328
                    params)))
329 1903 aaronmk
330
        return cur
331 1914 aaronmk
332
    def is_cached(self, query, params=None):
333
        return _query_lookup(query, params) in self.query_results
334 2139 aaronmk
335 2683 aaronmk
    def with_autocommit(self, func, autocommit=True):
336
        prev_autocommit = self.db.autocommit
337
        self.db.autocommit = autocommit
338
        try: return func()
339
        finally: self.db.autocommit = prev_autocommit
340
341 2139 aaronmk
    def with_savepoint(self, func):
342 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
343 2443 aaronmk
        self.run_query('SAVEPOINT '+savepoint, log_level=4)
344 2139 aaronmk
        self._savepoint += 1
345 2683 aaronmk
        try:
346 2139 aaronmk
            try: return_val = func()
347
            finally:
348
                self._savepoint -= 1
349
                assert self._savepoint >= 0
350
        except:
351 2443 aaronmk
            self.run_query('ROLLBACK TO SAVEPOINT '+savepoint, log_level=4)
352 2139 aaronmk
            raise
353
        else:
354 2443 aaronmk
            self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
355 2191 aaronmk
            self.do_autocommit()
356 2139 aaronmk
            return return_val
357 2191 aaronmk
358
    def do_autocommit(self):
359
        '''Autocommits if outside savepoint'''
360
        assert self._savepoint >= 0
361
        if self.autocommit and self._savepoint == 0:
362 2668 aaronmk
            self.log_debug('Autocommitting')
363 2191 aaronmk
            self.db.commit()
364 2643 aaronmk
365
    def col_default(self, col):
366
        table = sql_gen.Table('columns', 'information_schema')
367
368
        conds = [('table_name', col.table.name), ('column_name', col.name)]
369
        schema = col.table.schema
370
        if schema != None: conds.append(('table_schema', schema))
371
372 2660 aaronmk
        return sql_gen.as_Code(value(select(self, table, ['column_default'],
373
            conds, order_by='table_schema', limit=1, log_level=3)))
374 2643 aaronmk
            # TODO: order_by search_path schema order
375 1849 aaronmk
376 1869 aaronmk
connect = DbConn
377
378 832 aaronmk
##### Querying
379
380 1894 aaronmk
def run_raw_query(db, *args, **kw_args):
381 2085 aaronmk
    '''For params, see DbConn.run_query()'''
382 1894 aaronmk
    return db.run_query(*args, **kw_args)
383 11 aaronmk
384 2068 aaronmk
def mogrify(db, query, params):
385
    module = util.root_module(db.db)
386
    if module == 'psycopg2': return db.db.cursor().mogrify(query, params)
387
    else: raise NotImplementedError("Can't mogrify query for "+module+
388
        ' database')
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 2464 aaronmk
def run_query(db, query, params=None, recover=None, cacheable=False,
395
    log_level=2, log_ignore_excs=None, **kw_args):
396 2441 aaronmk
    '''For params, see run_raw_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
            def run(): return run_raw_query(db, query, params, cacheable,
408
                log_level, debug_msg_ref, **kw_args)
409
            if recover and not db.is_cached(query, params):
410
                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 2386 aaronmk
def run_query_into(db, query, params, into=None, *args, **kw_args):
466 2085 aaronmk
    '''Outputs a query to a temp table.
467
    For params, see run_query().
468
    '''
469 2386 aaronmk
    if into == None: return run_query(db, query, params, *args, **kw_args)
470 2085 aaronmk
    else: # place rows in temp table
471 2386 aaronmk
        assert isinstance(into, sql_gen.Table)
472 2385 aaronmk
473 2153 aaronmk
        kw_args['recover'] = True
474 2464 aaronmk
        kw_args.setdefault('log_ignore_excs', (DuplicateTableException,))
475 2440 aaronmk
476 2468 aaronmk
        temp = not db.autocommit # tables are permanent in autocommit mode
477 2440 aaronmk
        # "temporary tables cannot specify a schema name", so remove schema
478
        if temp: into.schema = None
479
480 2153 aaronmk
        while True:
481
            try:
482 2194 aaronmk
                create_query = 'CREATE'
483 2440 aaronmk
                if temp: create_query += ' TEMP'
484 2467 aaronmk
                create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
485 2194 aaronmk
486
                return run_query(db, create_query, params, *args, **kw_args)
487 2153 aaronmk
                    # CREATE TABLE AS sets rowcount to # rows in query
488
            except DuplicateTableException, e:
489 2386 aaronmk
                into.name = next_version(into.name)
490 2153 aaronmk
                # try again with next version of name
491 2085 aaronmk
492 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
493
494 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
495
496 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
497 2293 aaronmk
    start=None, order_by=order_by_pkey, default_table=None):
498 1981 aaronmk
    '''
499 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
500 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
501 1981 aaronmk
    @param fields Use None to select all fields in the table
502 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
503 2379 aaronmk
        * container can be any iterable type
504 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
505
        * compare_right_side: sql_gen.ValueCond|literal value
506 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
507
        use all columns
508 2054 aaronmk
    @return tuple(query, params)
509 1981 aaronmk
    '''
510 2315 aaronmk
    # Parse tables param
511 2121 aaronmk
    if not lists.is_seq(tables): tables = [tables]
512 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
513 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
514 2121 aaronmk
515 2315 aaronmk
    # Parse other params
516 2376 aaronmk
    if conds == None: conds = []
517 2650 aaronmk
    elif dicts.is_dict(conds): conds = conds.items()
518 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
519 135 aaronmk
    assert limit == None or type(limit) == int
520 865 aaronmk
    assert start == None or type(start) == int
521 2315 aaronmk
    if order_by is order_by_pkey:
522
        if distinct_on != []: order_by = None
523
        else: order_by = pkey(db, table0, recover=True)
524 865 aaronmk
525 2315 aaronmk
    query = 'SELECT'
526 2056 aaronmk
527 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
528 2056 aaronmk
529 2200 aaronmk
    # DISTINCT ON columns
530 2233 aaronmk
    if distinct_on != []:
531 2467 aaronmk
        query += '\nDISTINCT'
532 2254 aaronmk
        if distinct_on is not distinct_on_all:
533 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
534
535
    # Columns
536 2467 aaronmk
    query += '\n'
537 1135 aaronmk
    if fields == None: query += '*'
538 2479 aaronmk
    else: query += '\n, '.join(map(parse_col, fields))
539 2200 aaronmk
540
    # Main table
541 2467 aaronmk
    query += '\nFROM '+table0.to_str(db)
542 865 aaronmk
543 2122 aaronmk
    # Add joins
544 2271 aaronmk
    left_table = table0
545 2263 aaronmk
    for join_ in tables:
546
        table = join_.table
547 2238 aaronmk
548 2343 aaronmk
        # Parse special values
549
        if join_.type_ is sql_gen.filter_out: # filter no match
550 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
551
                None))
552 2343 aaronmk
553 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
554 2122 aaronmk
555
        left_table = table
556
557 865 aaronmk
    missing = True
558 2376 aaronmk
    if conds != []:
559 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
560
        else: whitespace = '\n'
561 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
562
            .to_str(db) for l, r in conds], 'WHERE')
563 865 aaronmk
        missing = False
564 2227 aaronmk
    if order_by != None:
565 2467 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
566
    if limit != None: query += '\nLIMIT '+str(limit); missing = False
567 865 aaronmk
    if start != None:
568 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
569 865 aaronmk
        missing = False
570
    if missing: warnings.warn(DbWarning(
571
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
572
573 2315 aaronmk
    return (query, [])
574 11 aaronmk
575 2054 aaronmk
def select(db, *args, **kw_args):
576
    '''For params, see mk_select() and run_query()'''
577
    recover = kw_args.pop('recover', None)
578
    cacheable = kw_args.pop('cacheable', True)
579 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
580 2054 aaronmk
581
    query, params = mk_select(db, *args, **kw_args)
582 2442 aaronmk
    return run_query(db, query, params, recover, cacheable, log_level=log_level)
583 2054 aaronmk
584 2066 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, params=None,
585 2292 aaronmk
    returning=None, embeddable=False):
586 1960 aaronmk
    '''
587
    @param returning str|None An inserted column (such as pkey) to return
588 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
589 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
590
        query will be fully cached, not just if it raises an exception.
591 1960 aaronmk
    '''
592 2328 aaronmk
    table = sql_gen.as_Table(table)
593 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
594 2646 aaronmk
    if cols != None:
595
        cols = [sql_gen.to_name_only_col(v, table).to_str(db) for v in cols]
596 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
597 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
598 2063 aaronmk
599
    # Build query
600 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
601
    query = first_line
602 2467 aaronmk
    if cols != None: query += '\n('+', '.join(cols)+')'
603
    query += '\n'+select_query
604 2063 aaronmk
605
    if returning != None:
606 2743 aaronmk
        query += '\nRETURNING '+sql_gen.to_name_only_col(returning).to_str(db)
607 2063 aaronmk
608 2070 aaronmk
    if embeddable:
609 2327 aaronmk
        assert returning != None
610
611 2070 aaronmk
        # Create function
612 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
613 2327 aaronmk
        return_type = 'SETOF '+returning.to_str(db)+'%TYPE'
614 2189 aaronmk
        while True:
615
            try:
616 2698 aaronmk
                function = sql_gen.TempFunction(function_name, db.autocommit)
617 2194 aaronmk
618 2189 aaronmk
                function_query = '''\
619 2698 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''()
620 2467 aaronmk
RETURNS '''+return_type+'''
621
LANGUAGE sql
622
AS $$
623
'''+mogrify(db, query, params)+''';
624
$$;
625 2070 aaronmk
'''
626 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
627 2464 aaronmk
                    log_ignore_excs=(DuplicateFunctionException,))
628 2189 aaronmk
                break # this version was successful
629
            except DuplicateFunctionException, e:
630
                function_name = next_version(function_name)
631
                # try again with next version of name
632 2070 aaronmk
633 2337 aaronmk
        # Return query that uses function
634 2698 aaronmk
        func_table = sql_gen.NamedTable('f', sql_gen.FunctionCall(function),
635 2742 aaronmk
            [returning]) # AS clause requires function alias
636 2337 aaronmk
        return mk_select(db, func_table, start=0, order_by=None)
637 2070 aaronmk
638 2066 aaronmk
    return (query, params)
639
640
def insert_select(db, *args, **kw_args):
641 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
642 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
643
        values in
644 2072 aaronmk
    '''
645 2386 aaronmk
    into = kw_args.pop('into', None)
646
    if into != None: kw_args['embeddable'] = True
647 2066 aaronmk
    recover = kw_args.pop('recover', None)
648
    cacheable = kw_args.pop('cacheable', True)
649 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
650 2066 aaronmk
651
    query, params = mk_insert_select(db, *args, **kw_args)
652 2386 aaronmk
    return run_query_into(db, query, params, into, recover=recover,
653 2673 aaronmk
        cacheable=cacheable, log_level=log_level)
654 2063 aaronmk
655 2738 aaronmk
default = sql_gen.default # tells insert() to use the default value for a column
656 2066 aaronmk
657 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
658 2085 aaronmk
    '''For params, see insert_select()'''
659 1960 aaronmk
    if lists.is_seq(row): cols = None
660
    else:
661
        cols = row.keys()
662
        row = row.values()
663 2738 aaronmk
    row = list(row) # ensure that "== []" works
664 1960 aaronmk
665 2738 aaronmk
    if row == []: query = None
666
    else: query = sql_gen.Values(row).to_str(db)
667 1961 aaronmk
668 2738 aaronmk
    return insert_select(db, table, cols, query, [], *args, **kw_args)
669 11 aaronmk
670 2402 aaronmk
def mk_update(db, table, changes=None, cond=None):
671
    '''
672
    @param changes [(col, new_value),...]
673
        * container can be any iterable type
674
        * col: sql_gen.Code|str (for col name)
675
        * new_value: sql_gen.Code|literal value
676
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
677
    @return str query
678
    '''
679
    query = 'UPDATE '+sql_gen.as_Table(table).to_str(db)+'\nSET\n'
680 2405 aaronmk
    query += ',\n'.join((sql_gen.to_name_only_col(col, table).to_str(db)+' = '
681 2402 aaronmk
        +sql_gen.as_Value(new_value).to_str(db) for col, new_value in changes))
682 2467 aaronmk
    if cond != None: query += '\nWHERE\n'+cond.to_str(db)
683 2402 aaronmk
684
    return query
685
686
def update(db, *args, **kw_args):
687
    '''For params, see mk_update() and run_query()'''
688
    recover = kw_args.pop('recover', None)
689
690
    return run_query(db, mk_update(db, *args, **kw_args), [], recover)
691
692 135 aaronmk
def last_insert_id(db):
693 1849 aaronmk
    module = util.root_module(db.db)
694 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
695
    elif module == 'MySQLdb': return db.insert_id()
696
    else: return None
697 13 aaronmk
698 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
699 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
700 2415 aaronmk
    to names that will be distinct among the columns' tables.
701 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
702 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
703
    @param into The table for the new columns.
704 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
705
        columns will be included in the mapping even if they are not in cols.
706
        The tables of the provided Col objects will be changed to into, so make
707
        copies of them if you want to keep the original tables.
708
    @param as_items Whether to return a list of dict items instead of a dict
709 2383 aaronmk
    @return dict(orig_col=new_col, ...)
710
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
711 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
712
        * All mappings use the into table so its name can easily be
713 2383 aaronmk
          changed for all columns at once
714
    '''
715 2415 aaronmk
    cols = lists.uniqify(cols)
716
717 2394 aaronmk
    items = []
718 2389 aaronmk
    for col in preserve:
719 2390 aaronmk
        orig_col = copy.copy(col)
720 2392 aaronmk
        col.table = into
721 2394 aaronmk
        items.append((orig_col, col))
722
    preserve = set(preserve)
723
    for col in cols:
724 2716 aaronmk
        if col not in preserve:
725
            items.append((col, sql_gen.Col(str(col), into, col.srcs)))
726 2394 aaronmk
727
    if not as_items: items = dict(items)
728
    return items
729 2383 aaronmk
730 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
731 2391 aaronmk
    '''For params, see mk_flatten_mapping()
732
    @return See return value of mk_flatten_mapping()
733
    '''
734 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
735
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
736 2391 aaronmk
    run_query_into(db, *mk_select(db, joins, cols, limit=limit, start=start),
737 2392 aaronmk
        into=into)
738 2394 aaronmk
    return dict(items)
739 2391 aaronmk
740 2751 aaronmk
def mk_track_data_error(db, errors_table, cols, value, error_code, error):
741
    cols = map(sql_gen.to_name_only_col, cols)
742
743
    columns_cols = ['column']
744
    columns = sql_gen.NamedValues('columns', columns_cols,
745
        [[c.name] for c in cols])
746
    values_cols = ['value', 'error_code', 'error']
747
    values = sql_gen.NamedValues('values', values_cols,
748
        [value, error_code, error])
749
750
    select_cols = columns_cols+values_cols
751
    name_only_cols = map(sql_gen.to_name_only_col, select_cols)
752
    joins = [columns, sql_gen.Join(values, type_='CROSS'),
753
        sql_gen.Join(errors_table, dict(zip(name_only_cols, select_cols)),
754
        sql_gen.filter_out)]
755
756
    return mk_insert_select(db, errors_table, name_only_cols,
757
        *mk_select(db, joins, select_cols, order_by=None))[0]
758
759 2704 aaronmk
def cast(db, type_, col, errors_table=None):
760
    '''Casts an (unrenamed) column or value.
761
    If a column, converts any errors to warnings.
762
    @param col sql_gen.Col|sql_gen.Literal
763
    @param errors_table None|sql_gen.Table|str
764
        If set and col is a column with srcs, saves any errors in this table,
765
        using column's srcs attr as the source columns.
766
    '''
767
    if isinstance(col, sql_gen.Literal): # literal value, so just cast
768
        return sql_gen.CustomCode(col.to_str(db)+'::'+type_)
769
770
    assert isinstance(col, sql_gen.Col)
771
    assert not isinstance(col, sql_gen.NamedCol)
772
773
    save_errors = errors_table != None and col.srcs != ()
774
    if save_errors:
775
        errors_table = sql_gen.as_Table(errors_table)
776
        srcs = map(sql_gen.to_name_only_col, col.srcs)
777 2717 aaronmk
        function_name = str(sql_gen.FunctionCall(type_, *srcs))
778 2704 aaronmk
    else: function_name = type_
779
    function = sql_gen.TempFunction(function_name, db.autocommit)
780
781 2718 aaronmk
    while True:
782
        # Create function definition
783
        query = '''\
784 2704 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''(value text)
785
RETURNS '''+type_+'''
786
LANGUAGE plpgsql
787
'''
788 2718 aaronmk
        if not save_errors: query += 'IMMUTABLE\n'
789
        query += '''\
790 2704 aaronmk
STRICT
791
AS $$
792
BEGIN
793
    /* The explicit cast to the return type is needed to make the cast happen
794
    inside the try block. (Implicit casts to the return type happen at the end
795
    of the function, outside any block.) */
796
    RETURN value::'''+type_+''';
797
EXCEPTION
798
    WHEN data_exception THEN
799
'''
800 2718 aaronmk
        if save_errors:
801
            query += '''\
802 2704 aaronmk
        -- Save error in errors table.
803 2751 aaronmk
        -- Insert the value and error for *each* source column.
804
'''+mk_track_data_error(db, errors_table, srcs,
805
    *map(sql_gen.CustomCode, ['value', 'SQLSTATE', 'SQLERRM']))+''';
806 2704 aaronmk
807
'''
808 2718 aaronmk
        query += '''\
809 2704 aaronmk
        RAISE WARNING '%', SQLERRM;
810
        RETURN NULL;
811
END;
812
$$;
813
'''
814 2718 aaronmk
815
        # Create function
816
        try:
817
            run_query(db, query, recover=True, cacheable=True,
818
                log_ignore_excs=(DuplicateFunctionException,))
819
            break # successful
820
        except DuplicateFunctionException:
821
            if save_errors: function.name = next_version(function.name)
822
                # try again with next version of name
823
            else: break # plain cast function, so only need one version
824 2704 aaronmk
825
    return sql_gen.FunctionCall(function, col)
826
827 2414 aaronmk
##### Database structure queries
828
829 2426 aaronmk
def table_row_count(db, table, recover=None):
830
    return value(run_query(db, *mk_select(db, table, [sql_gen.row_count],
831 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
832 2426 aaronmk
833 2414 aaronmk
def table_cols(db, table, recover=None):
834
    return list(col_names(select(db, table, limit=0, order_by=None,
835 2443 aaronmk
        recover=recover, log_level=4)))
836 2414 aaronmk
837 2291 aaronmk
def pkey(db, table, recover=None):
838 832 aaronmk
    '''Assumed to be first column in table'''
839 2339 aaronmk
    return table_cols(db, table, recover)[0]
840 832 aaronmk
841 2559 aaronmk
not_null_col = 'not_null_col'
842 2340 aaronmk
843
def table_not_null_col(db, table, recover=None):
844
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
845
    if not_null_col in table_cols(db, table, recover): return not_null_col
846
    else: return pkey(db, table, recover)
847
848 853 aaronmk
def index_cols(db, table, index):
849
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
850
    automatically created. When you don't know whether something is a UNIQUE
851
    constraint or a UNIQUE index, use this function.'''
852 1909 aaronmk
    module = util.root_module(db.db)
853
    if module == 'psycopg2':
854
        return list(values(run_query(db, '''\
855 853 aaronmk
SELECT attname
856 866 aaronmk
FROM
857
(
858
        SELECT attnum, attname
859
        FROM pg_index
860
        JOIN pg_class index ON index.oid = indexrelid
861
        JOIN pg_class table_ ON table_.oid = indrelid
862
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
863
        WHERE
864
            table_.relname = %(table)s
865
            AND index.relname = %(index)s
866
    UNION
867
        SELECT attnum, attname
868
        FROM
869
        (
870
            SELECT
871
                indrelid
872
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
873
                    AS indkey
874
            FROM pg_index
875
            JOIN pg_class index ON index.oid = indexrelid
876
            JOIN pg_class table_ ON table_.oid = indrelid
877
            WHERE
878
                table_.relname = %(table)s
879
                AND index.relname = %(index)s
880
        ) s
881
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
882
) s
883 853 aaronmk
ORDER BY attnum
884
''',
885 2443 aaronmk
            {'table': table, 'index': index}, cacheable=True, log_level=4)))
886 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
887
        ' database')
888 853 aaronmk
889 464 aaronmk
def constraint_cols(db, table, constraint):
890 1849 aaronmk
    module = util.root_module(db.db)
891 464 aaronmk
    if module == 'psycopg2':
892
        return list(values(run_query(db, '''\
893
SELECT attname
894
FROM pg_constraint
895
JOIN pg_class ON pg_class.oid = conrelid
896
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
897
WHERE
898
    relname = %(table)s
899
    AND conname = %(constraint)s
900
ORDER BY attnum
901
''',
902
            {'table': table, 'constraint': constraint})))
903
    else: raise NotImplementedError("Can't list constraint columns for "+module+
904
        ' database')
905
906 2096 aaronmk
row_num_col = '_row_num'
907
908 2688 aaronmk
def add_index(db, exprs, table=None, unique=False):
909
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
910 2538 aaronmk
    Currently, only function calls are supported as expressions.
911
    '''
912 2688 aaronmk
    if not lists.is_seq(exprs): exprs = [exprs]
913 2538 aaronmk
914 2688 aaronmk
    # Parse exprs
915
    old_exprs = exprs[:]
916
    exprs = []
917
    cols = []
918
    for i, expr in enumerate(old_exprs):
919
        expr = copy.deepcopy(expr) # don't modify input!
920
        expr = sql_gen.as_Col(expr)
921
922
        # Extract col
923
        if isinstance(expr, sql_gen.FunctionCall):
924
            col = expr.args[0]
925
            expr = sql_gen.Expr(expr)
926
        else: col = expr
927
928
        # Extract table
929
        if table == None:
930
            assert sql_gen.is_table_col(col)
931
            table = col.table
932
933
        col.table = None
934
935
        exprs.append(expr)
936
        cols.append(col)
937 2408 aaronmk
938 2688 aaronmk
    table = sql_gen.as_Table(table)
939
    index = sql_gen.Table(str(sql_gen.Col(','.join(map(str, cols)), table)))
940
941
    str_ = 'CREATE'
942
    if unique: str_ += ' UNIQUE'
943
    str_ += ' INDEX '+index.to_str(db)+' ON '+table.to_str(db)+' ('+(
944
        ', '.join((v.to_str(db) for v in exprs)))+')'
945
946
    try: run_query(db, str_, recover=True, cacheable=True, log_level=3)
947 2408 aaronmk
    except DuplicateTableException: pass # index already existed
948
949 2686 aaronmk
def add_pkey(db, table, cols=None, recover=None):
950
    '''Adds a primary key.
951
    @param cols [sql_gen.Col,...] The columns in the primary key.
952
        Defaults to the first column in the table.
953 2406 aaronmk
    @pre The table must not already have a primary key.
954
    '''
955
    table = sql_gen.as_Table(table)
956 2686 aaronmk
    if cols == None: cols = [pkey(db, table, recover)]
957
    col_strs = [sql_gen.to_name_only_col(v).to_str(db) for v in cols]
958 2406 aaronmk
959 2590 aaronmk
    index = sql_gen.as_Table(sql_gen.add_suffix(table.name, '_pkey'))
960 2614 aaronmk
    try:
961
        run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD CONSTRAINT '
962 2686 aaronmk
            +index.to_str(db)+' PRIMARY KEY ('+(', '.join(col_strs))+')',
963 2614 aaronmk
            recover=True, cacheable=True, log_level=3,
964
            log_ignore_excs=(DuplicateTableException,))
965
    except DuplicateTableException, e:
966
        index.name = next_version(index.name)
967
        # try again with next version of name
968 2406 aaronmk
969 2086 aaronmk
def add_row_num(db, table):
970 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
971
    be the primary key.'''
972 2320 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
973 2096 aaronmk
    run_query(db, 'ALTER TABLE '+table+' ADD COLUMN '+row_num_col
974 2443 aaronmk
        +' serial NOT NULL PRIMARY KEY', log_level=3)
975 2086 aaronmk
976 2681 aaronmk
def create_table(db, table, cols, has_pkey=True):
977 2675 aaronmk
    '''Creates a table.
978 2681 aaronmk
    @param cols [sql_gen.TypedCol,...] The column names and types
979
    @param has_pkey If set, the first column becomes the primary key.
980 2675 aaronmk
    '''
981
    table = sql_gen.as_Table(table)
982
983 2681 aaronmk
    if has_pkey:
984
        cols[0] = pkey = copy.copy(cols[0]) # don't modify input!
985
        pkey.type += ' NOT NULL PRIMARY KEY'
986
987 2675 aaronmk
    str_ = 'CREATE TABLE '+table.to_str(db)+' (\n'
988 2681 aaronmk
    str_ += '\n, '.join(v.to_str(db) for v in cols)
989
    str_ += '\n);\n'
990 2675 aaronmk
    run_query(db, str_, cacheable=True, log_level=2)
991
992 2684 aaronmk
def vacuum(db, table):
993
    table = sql_gen.as_Table(table)
994
    db.with_autocommit(lambda: run_query(db, 'VACUUM ANALYZE '+table.to_str(db),
995
        log_level=3))
996
997 2732 aaronmk
def truncate(db, table, schema='public'):
998 2733 aaronmk
    return run_query(db, 'TRUNCATE '+table.to_str(db)+' CASCADE')
999 2732 aaronmk
1000 2729 aaronmk
def tables(db, schema_like='public', table_like='%', exact=False):
1001
    if exact: compare = '='
1002
    else: compare = 'LIKE'
1003
1004 1849 aaronmk
    module = util.root_module(db.db)
1005 2548 aaronmk
    params = {'schema_like': schema_like, 'table_like': table_like}
1006 832 aaronmk
    if module == 'psycopg2':
1007 1968 aaronmk
        return values(run_query(db, '''\
1008
SELECT tablename
1009
FROM pg_tables
1010
WHERE
1011 2729 aaronmk
    schemaname '''+compare+''' %(schema_like)s
1012
    AND tablename '''+compare+''' %(table_like)s
1013 1968 aaronmk
ORDER BY tablename
1014
''',
1015 2731 aaronmk
            params, cacheable=True, log_level=4))
1016 1968 aaronmk
    elif module == 'MySQLdb':
1017
        return values(run_query(db, 'SHOW TABLES LIKE %(table_like)s', params,
1018 2731 aaronmk
            cacheable=True, log_level=4))
1019 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
1020 830 aaronmk
1021 2726 aaronmk
def table_exists(db, table):
1022
    table = sql_gen.as_Table(table)
1023 2730 aaronmk
    return list(tables(db, table.schema, table.name, exact=True)) != []
1024 2726 aaronmk
1025
def errors_table(db, table, if_exists=True):
1026
    '''
1027
    @param if_exists If set, returns None if the errors table doesn't exist
1028
    @return None|sql_gen.Table
1029
    '''
1030
    table = sql_gen.as_Table(table)
1031
    if table.srcs != (): table = table.srcs[0]
1032
1033
    errors_table = sql_gen.suffixed_table(table, '.errors')
1034
    if if_exists and not table_exists(db, errors_table): return None
1035
    return errors_table
1036
1037 833 aaronmk
##### Database management
1038
1039 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
1040
    '''For kw_args, see tables()'''
1041
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
1042 833 aaronmk
1043 832 aaronmk
##### Heuristic queries
1044
1045 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
1046 1554 aaronmk
    '''Recovers from errors.
1047 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
1048
    '''
1049 2642 aaronmk
    row = sql_gen.ColDict(db, table, row)
1050 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
1051
1052 471 aaronmk
    try:
1053 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
1054 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
1055
            row_ct_ref[0] += cur.rowcount
1056
        return value(cur)
1057 471 aaronmk
    except DuplicateKeyException, e:
1058 2653 aaronmk
        row = sql_gen.ColDict(db, table,
1059
            util.dict_subset_right_join(row, e.cols))
1060
        return value(select(db, table, [pkey_], row, recover=True))
1061 471 aaronmk
1062 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
1063 830 aaronmk
    '''Recovers from errors'''
1064 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
1065 14 aaronmk
    except StopIteration:
1066 40 aaronmk
        if not create: raise
1067 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
1068 2078 aaronmk
1069 2593 aaronmk
def is_func_result(col):
1070
    return col.table.name.find('(') >= 0 and col.name == 'result'
1071
1072 2592 aaronmk
def into_table_name(out_table, in_tables0, mapping, is_func):
1073 2615 aaronmk
    def in_col_str(in_col):
1074
        in_col = sql_gen.remove_col_rename(in_col)
1075
        if isinstance(in_col, sql_gen.Col):
1076
            table = in_col.table
1077
            if table == in_tables0:
1078
                in_col = sql_gen.to_name_only_col(in_col)
1079
            elif is_func_result(in_col): in_col = table # omit col name
1080
        return str(in_col)
1081
1082 2580 aaronmk
    str_ = str(out_table)
1083
    if is_func:
1084 2615 aaronmk
        str_ += '('
1085 2580 aaronmk
1086 2615 aaronmk
        try: value_in_col = mapping['value']
1087
        except KeyError:
1088 2654 aaronmk
            str_ += ', '.join((str(k)+'='+in_col_str(v)
1089 2615 aaronmk
                for k, v in mapping.iteritems()))
1090
        else: str_ += in_col_str(value_in_col)
1091
1092
        str_ += ')'
1093 2580 aaronmk
    else: str_ += '_pkeys'
1094
    return str_
1095
1096 2508 aaronmk
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None, into=None,
1097 2552 aaronmk
    default=None, is_func=False):
1098 2078 aaronmk
    '''Recovers from errors.
1099
    Only works under PostgreSQL (uses INSERT RETURNING).
1100 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
1101
        tables to join with it using the main input table's pkey
1102 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
1103 2612 aaronmk
        * out_table_col: str (*not* sql_gen.Col)
1104
        * in_table_col: sql_gen.Col|literal-value
1105 2489 aaronmk
    @param into The table to contain the output and input pkeys.
1106 2574 aaronmk
        Defaults to `out_table.name+'_pkeys'`.
1107 2509 aaronmk
    @param default The *output* column to use as the pkey for missing rows.
1108
        If this output column does not exist in the mapping, uses None.
1109 2552 aaronmk
    @param is_func Whether out_table is the name of a SQL function, not a table
1110 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
1111 2078 aaronmk
    '''
1112 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
1113 2312 aaronmk
1114 2450 aaronmk
    def log_debug(msg): db.log_debug(msg, level=1.5)
1115 2505 aaronmk
    def col_ustr(str_):
1116 2567 aaronmk
        return strings.repr_no_u(sql_gen.remove_col_rename(str_))
1117 2450 aaronmk
1118 2486 aaronmk
    log_debug('********** New iteration **********')
1119 2505 aaronmk
    log_debug('Inserting these input columns into '+strings.as_tt(
1120
        out_table.to_str(db))+':\n'+strings.as_table(mapping, ustr=col_ustr))
1121 2463 aaronmk
1122 2382 aaronmk
    # Create input joins from list of input tables
1123
    in_tables_ = in_tables[:] # don't modify input!
1124
    in_tables0 = in_tables_.pop(0) # first table is separate
1125 2735 aaronmk
    errors_table_ = errors_table(db, in_tables0)
1126 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
1127 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
1128 2460 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v,
1129
        {in_pkey: sql_gen.join_same_not_null}) for v in in_tables_]
1130 2131 aaronmk
1131 2592 aaronmk
    if into == None:
1132
        into = into_table_name(out_table, in_tables0, mapping, is_func)
1133
    into = sql_gen.as_Table(into)
1134
1135 2702 aaronmk
    # Set column sources
1136
    in_cols = filter(sql_gen.is_table_col, mapping.values())
1137
    for col in in_cols:
1138
        if col.table == in_tables0: col.set_srcs(sql_gen.src_self)
1139
1140 2486 aaronmk
    log_debug('Joining together input tables into temp table')
1141 2395 aaronmk
    # Place in new table for speed and so don't modify input if values edited
1142 2584 aaronmk
    in_table = sql_gen.Table('in')
1143 2702 aaronmk
    mapping = dicts.join(mapping, flatten(db, in_table, input_joins, in_cols,
1144
        preserve=[in_pkey_col], start=0))
1145 2395 aaronmk
    input_joins = [in_table]
1146 2486 aaronmk
    db.log_debug('Temp table: '+strings.as_tt(in_table.to_str(db)), level=2)
1147 2395 aaronmk
1148 2656 aaronmk
    mapping = sql_gen.ColDict(db, out_table, mapping)
1149
        # after applying dicts.join() because that returns a plain dict
1150
1151 2509 aaronmk
    # Resolve default value column
1152
    try: default = mapping[default]
1153
    except KeyError:
1154
        if default != None:
1155
            db.log_debug('Default value column '
1156
                +strings.as_tt(strings.repr_no_u(default))
1157 2511 aaronmk
                +' does not exist in mapping, falling back to None', level=2.1)
1158 2509 aaronmk
            default = None
1159
1160 2279 aaronmk
    out_pkey = pkey(db, out_table, recover=True)
1161 2285 aaronmk
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
1162 2142 aaronmk
1163 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
1164 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
1165
1166 2201 aaronmk
    pkeys_table_exists_ref = [False]
1167 2420 aaronmk
    def insert_into_pkeys(joins, cols):
1168
        query, params = mk_select(db, joins, cols, order_by=None, start=0)
1169 2201 aaronmk
        if pkeys_table_exists_ref[0]:
1170 2489 aaronmk
            insert_select(db, into, pkeys_names, query, params)
1171 2201 aaronmk
        else:
1172 2489 aaronmk
            run_query_into(db, query, params, into=into)
1173 2201 aaronmk
            pkeys_table_exists_ref[0] = True
1174
1175 2429 aaronmk
    limit_ref = [None]
1176 2380 aaronmk
    conds = set()
1177 2233 aaronmk
    distinct_on = []
1178 2325 aaronmk
    def mk_main_select(joins, cols):
1179 2429 aaronmk
        return mk_select(db, joins, cols, conds, distinct_on,
1180
            limit=limit_ref[0], start=0)
1181 2132 aaronmk
1182 2519 aaronmk
    exc_strs = set()
1183 2309 aaronmk
    def log_exc(e):
1184 2519 aaronmk
        e_str = exc.str_(e, first_line_only=True)
1185
        log_debug('Caught exception: '+e_str)
1186
        assert e_str not in exc_strs # avoid infinite loops
1187
        exc_strs.add(e_str)
1188 2735 aaronmk
1189 2451 aaronmk
    def remove_all_rows():
1190 2450 aaronmk
        log_debug('Returning NULL for all rows')
1191 2429 aaronmk
        limit_ref[0] = 0 # just create an empty pkeys table
1192 2735 aaronmk
1193 2409 aaronmk
    def ignore(in_col, value):
1194 2545 aaronmk
        in_col_str = strings.as_tt(repr(in_col))
1195 2544 aaronmk
        db.log_debug('Adding index on '+in_col_str+' to enable fast filtering',
1196
            level=2.5)
1197 2537 aaronmk
        add_index(db, in_col)
1198 2735 aaronmk
1199 2545 aaronmk
        log_debug('Ignoring rows with '+in_col_str+' = '
1200
            +strings.as_tt(repr(value)))
1201 2403 aaronmk
    def remove_rows(in_col, value):
1202 2409 aaronmk
        ignore(in_col, value)
1203 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
1204
        assert cond not in conds # avoid infinite loops
1205 2380 aaronmk
        conds.add(cond)
1206 2403 aaronmk
    def invalid2null(in_col, value):
1207 2409 aaronmk
        ignore(in_col, value)
1208 2403 aaronmk
        update(db, in_table, [(in_col, None)],
1209
            sql_gen.ColValueCond(in_col, value))
1210 2245 aaronmk
1211 2589 aaronmk
    def insert_pkeys_table(which):
1212
        return sql_gen.Table(sql_gen.add_suffix(in_table.name,
1213
            '_insert_'+which+'_pkeys'))
1214
    insert_out_pkeys = insert_pkeys_table('out')
1215
    insert_in_pkeys = insert_pkeys_table('in')
1216
1217 2206 aaronmk
    # Do inserts and selects
1218 2642 aaronmk
    join_cols = sql_gen.ColDict(db, out_table)
1219 2206 aaronmk
    while True:
1220 2521 aaronmk
        if limit_ref[0] == 0: # special case
1221
            log_debug('Creating an empty pkeys table')
1222
            cur = run_query_into(db, *mk_select(db, out_table, [out_pkey],
1223
                limit=limit_ref[0]), into=insert_out_pkeys)
1224
            break # don't do main case
1225
1226 2303 aaronmk
        has_joins = join_cols != {}
1227
1228 2305 aaronmk
        # Prepare to insert new rows
1229 2325 aaronmk
        insert_joins = input_joins[:] # don't modify original!
1230 2403 aaronmk
        insert_args = dict(recover=True, cacheable=False)
1231 2303 aaronmk
        if has_joins:
1232 2317 aaronmk
            distinct_on = [v.to_Col() for v in join_cols.values()]
1233 2325 aaronmk
            insert_joins.append(sql_gen.Join(out_table, join_cols,
1234
                sql_gen.filter_out))
1235
        else:
1236 2404 aaronmk
            insert_args.update(dict(returning=out_pkey, into=insert_out_pkeys))
1237 2520 aaronmk
        main_select = mk_main_select(insert_joins, mapping.values())[0]
1238 2303 aaronmk
1239 2486 aaronmk
        log_debug('Trying to insert new rows')
1240 2206 aaronmk
        try:
1241 2518 aaronmk
            cur = insert_select(db, out_table, mapping.keys(), main_select,
1242
                **insert_args)
1243 2357 aaronmk
            break # insert successful
1244 2206 aaronmk
        except DuplicateKeyException, e:
1245 2309 aaronmk
            log_exc(e)
1246
1247 2258 aaronmk
            old_join_cols = join_cols.copy()
1248 2565 aaronmk
            join_cols.update(util.dict_subset_right_join(mapping, e.cols))
1249 2486 aaronmk
            log_debug('Ignoring existing rows, comparing on these columns:\n'
1250 2505 aaronmk
                +strings.as_inline_table(join_cols, ustr=col_ustr))
1251 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
1252 2230 aaronmk
        except NullValueException, e:
1253 2309 aaronmk
            log_exc(e)
1254
1255 2230 aaronmk
            out_col, = e.cols
1256
            try: in_col = mapping[out_col]
1257 2356 aaronmk
            except KeyError:
1258 2486 aaronmk
                log_debug('Missing mapping for NOT NULL column '+out_col)
1259 2451 aaronmk
                remove_all_rows()
1260 2403 aaronmk
            else: remove_rows(in_col, None)
1261 2542 aaronmk
        except FunctionValueException, e:
1262
            log_exc(e)
1263
1264
            func_name = e.name
1265
            value = e.value
1266
            for out_col, in_col in mapping.iteritems():
1267 2562 aaronmk
                invalid2null(sql_gen.unwrap_func_call(in_col, func_name), value)
1268 2525 aaronmk
        except MissingCastException, e:
1269
            log_exc(e)
1270
1271
            out_col = e.col
1272 2706 aaronmk
            type_ = e.type
1273
1274 2724 aaronmk
            log_debug('Casting '+strings.as_tt(out_col)+' input to '
1275
                +strings.as_tt(type_))
1276 2735 aaronmk
            def wrap_func(col): return cast(db, type_, col, errors_table_)
1277 2705 aaronmk
            mapping[out_col] = sql_gen.wrap(wrap_func, mapping[out_col])
1278 2429 aaronmk
        except DatabaseErrors, e:
1279
            log_exc(e)
1280
1281 2531 aaronmk
            msg = 'No handler for exception: '+exc.str_(e)
1282 2451 aaronmk
            warnings.warn(DbWarning(msg))
1283
            log_debug(msg)
1284
            remove_all_rows()
1285 2358 aaronmk
        # after exception handled, rerun loop with additional constraints
1286 2132 aaronmk
1287 2357 aaronmk
    if row_ct_ref != None and cur.rowcount >= 0:
1288
        row_ct_ref[0] += cur.rowcount
1289
1290
    if has_joins:
1291
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
1292 2486 aaronmk
        log_debug('Getting output table pkeys of existing/inserted rows')
1293 2420 aaronmk
        insert_into_pkeys(select_joins, pkeys_cols)
1294 2357 aaronmk
    else:
1295 2404 aaronmk
        add_row_num(db, insert_out_pkeys) # for joining with input pkeys
1296 2357 aaronmk
1297 2486 aaronmk
        log_debug('Getting input table pkeys of inserted rows')
1298 2357 aaronmk
        run_query_into(db, *mk_main_select(input_joins, [in_pkey]),
1299 2404 aaronmk
            into=insert_in_pkeys)
1300
        add_row_num(db, insert_in_pkeys) # for joining with output pkeys
1301 2357 aaronmk
1302 2428 aaronmk
        assert table_row_count(db, insert_out_pkeys) == table_row_count(db,
1303
            insert_in_pkeys)
1304
1305 2486 aaronmk
        log_debug('Combining output and input pkeys in inserted order')
1306 2404 aaronmk
        pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
1307 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
1308 2420 aaronmk
        insert_into_pkeys(pkey_joins, pkeys_names)
1309 2357 aaronmk
1310 2486 aaronmk
    db.log_debug('Adding pkey on pkeys table to enable fast joins', level=2.5)
1311 2594 aaronmk
    add_pkey(db, into)
1312 2407 aaronmk
1313 2508 aaronmk
    log_debug('Setting pkeys of missing rows to '+strings.as_tt(repr(default)))
1314 2489 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(into,
1315 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
1316
        # must use join_same_not_null or query will take forever
1317 2420 aaronmk
    insert_into_pkeys(missing_rows_joins,
1318 2508 aaronmk
        [in_pkey_col, sql_gen.NamedCol(out_pkey, default)])
1319 2357 aaronmk
1320 2489 aaronmk
    assert table_row_count(db, into) == table_row_count(db, in_table)
1321 2428 aaronmk
1322 2702 aaronmk
    srcs = []
1323
    if is_func: srcs = sql_gen.cols_srcs(in_cols)
1324
    return sql_gen.Col(out_pkey, into, srcs)
1325 2115 aaronmk
1326
##### Data cleanup
1327
1328 2290 aaronmk
def cleanup_table(db, table, cols):
1329 2677 aaronmk
    table = sql_gen.as_Table(table)
1330
    cols = map(sql_gen.as_Col, cols)
1331 2115 aaronmk
1332 2677 aaronmk
    expr = ('nullif(nullif(trim(both from %s), '+db.esc_value('')+'), '
1333
        +db.esc_value(r'\N')+')')
1334
    changes = [(v, sql_gen.CustomCode(expr % v.to_str(db)))
1335
        for v in cols]
1336 2115 aaronmk
1337 2677 aaronmk
    update(db, table, changes)