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 2327 aaronmk
        returning_name = copy.copy(returning)
607
        returning_name.table = None
608
        returning_name = returning_name.to_str(db)
609 2467 aaronmk
        query += '\nRETURNING '+returning_name
610 2063 aaronmk
611 2070 aaronmk
    if embeddable:
612 2327 aaronmk
        assert returning != None
613
614 2070 aaronmk
        # Create function
615 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
616 2327 aaronmk
        return_type = 'SETOF '+returning.to_str(db)+'%TYPE'
617 2189 aaronmk
        while True:
618
            try:
619 2698 aaronmk
                function = sql_gen.TempFunction(function_name, db.autocommit)
620 2194 aaronmk
621 2189 aaronmk
                function_query = '''\
622 2698 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''()
623 2467 aaronmk
RETURNS '''+return_type+'''
624
LANGUAGE sql
625
AS $$
626
'''+mogrify(db, query, params)+''';
627
$$;
628 2070 aaronmk
'''
629 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
630 2464 aaronmk
                    log_ignore_excs=(DuplicateFunctionException,))
631 2189 aaronmk
                break # this version was successful
632
            except DuplicateFunctionException, e:
633
                function_name = next_version(function_name)
634
                # try again with next version of name
635 2070 aaronmk
636 2337 aaronmk
        # Return query that uses function
637 2698 aaronmk
        func_table = sql_gen.NamedTable('f', sql_gen.FunctionCall(function),
638 2742 aaronmk
            [returning]) # AS clause requires function alias
639 2337 aaronmk
        return mk_select(db, func_table, start=0, order_by=None)
640 2070 aaronmk
641 2066 aaronmk
    return (query, params)
642
643
def insert_select(db, *args, **kw_args):
644 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
645 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
646
        values in
647 2072 aaronmk
    '''
648 2386 aaronmk
    into = kw_args.pop('into', None)
649
    if into != None: kw_args['embeddable'] = True
650 2066 aaronmk
    recover = kw_args.pop('recover', None)
651
    cacheable = kw_args.pop('cacheable', True)
652 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
653 2066 aaronmk
654
    query, params = mk_insert_select(db, *args, **kw_args)
655 2386 aaronmk
    return run_query_into(db, query, params, into, recover=recover,
656 2673 aaronmk
        cacheable=cacheable, log_level=log_level)
657 2063 aaronmk
658 2738 aaronmk
default = sql_gen.default # tells insert() to use the default value for a column
659 2066 aaronmk
660 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
661 2085 aaronmk
    '''For params, see insert_select()'''
662 1960 aaronmk
    if lists.is_seq(row): cols = None
663
    else:
664
        cols = row.keys()
665
        row = row.values()
666 2738 aaronmk
    row = list(row) # ensure that "== []" works
667 1960 aaronmk
668 2738 aaronmk
    if row == []: query = None
669
    else: query = sql_gen.Values(row).to_str(db)
670 1961 aaronmk
671 2738 aaronmk
    return insert_select(db, table, cols, query, [], *args, **kw_args)
672 11 aaronmk
673 2402 aaronmk
def mk_update(db, table, changes=None, cond=None):
674
    '''
675
    @param changes [(col, new_value),...]
676
        * container can be any iterable type
677
        * col: sql_gen.Code|str (for col name)
678
        * new_value: sql_gen.Code|literal value
679
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
680
    @return str query
681
    '''
682
    query = 'UPDATE '+sql_gen.as_Table(table).to_str(db)+'\nSET\n'
683 2405 aaronmk
    query += ',\n'.join((sql_gen.to_name_only_col(col, table).to_str(db)+' = '
684 2402 aaronmk
        +sql_gen.as_Value(new_value).to_str(db) for col, new_value in changes))
685 2467 aaronmk
    if cond != None: query += '\nWHERE\n'+cond.to_str(db)
686 2402 aaronmk
687
    return query
688
689
def update(db, *args, **kw_args):
690
    '''For params, see mk_update() and run_query()'''
691
    recover = kw_args.pop('recover', None)
692
693
    return run_query(db, mk_update(db, *args, **kw_args), [], recover)
694
695 135 aaronmk
def last_insert_id(db):
696 1849 aaronmk
    module = util.root_module(db.db)
697 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
698
    elif module == 'MySQLdb': return db.insert_id()
699
    else: return None
700 13 aaronmk
701 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
702 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
703 2415 aaronmk
    to names that will be distinct among the columns' tables.
704 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
705 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
706
    @param into The table for the new columns.
707 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
708
        columns will be included in the mapping even if they are not in cols.
709
        The tables of the provided Col objects will be changed to into, so make
710
        copies of them if you want to keep the original tables.
711
    @param as_items Whether to return a list of dict items instead of a dict
712 2383 aaronmk
    @return dict(orig_col=new_col, ...)
713
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
714 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
715
        * All mappings use the into table so its name can easily be
716 2383 aaronmk
          changed for all columns at once
717
    '''
718 2415 aaronmk
    cols = lists.uniqify(cols)
719
720 2394 aaronmk
    items = []
721 2389 aaronmk
    for col in preserve:
722 2390 aaronmk
        orig_col = copy.copy(col)
723 2392 aaronmk
        col.table = into
724 2394 aaronmk
        items.append((orig_col, col))
725
    preserve = set(preserve)
726
    for col in cols:
727 2716 aaronmk
        if col not in preserve:
728
            items.append((col, sql_gen.Col(str(col), into, col.srcs)))
729 2394 aaronmk
730
    if not as_items: items = dict(items)
731
    return items
732 2383 aaronmk
733 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
734 2391 aaronmk
    '''For params, see mk_flatten_mapping()
735
    @return See return value of mk_flatten_mapping()
736
    '''
737 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
738
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
739 2391 aaronmk
    run_query_into(db, *mk_select(db, joins, cols, limit=limit, start=start),
740 2392 aaronmk
        into=into)
741 2394 aaronmk
    return dict(items)
742 2391 aaronmk
743 2704 aaronmk
def cast(db, type_, col, errors_table=None):
744
    '''Casts an (unrenamed) column or value.
745
    If a column, converts any errors to warnings.
746
    @param col sql_gen.Col|sql_gen.Literal
747
    @param errors_table None|sql_gen.Table|str
748
        If set and col is a column with srcs, saves any errors in this table,
749
        using column's srcs attr as the source columns.
750
    '''
751
    if isinstance(col, sql_gen.Literal): # literal value, so just cast
752
        return sql_gen.CustomCode(col.to_str(db)+'::'+type_)
753
754
    assert isinstance(col, sql_gen.Col)
755
    assert not isinstance(col, sql_gen.NamedCol)
756
757
    save_errors = errors_table != None and col.srcs != ()
758
    if save_errors:
759
        errors_table = sql_gen.as_Table(errors_table)
760
        srcs = map(sql_gen.to_name_only_col, col.srcs)
761 2717 aaronmk
        function_name = str(sql_gen.FunctionCall(type_, *srcs))
762 2704 aaronmk
    else: function_name = type_
763
    function = sql_gen.TempFunction(function_name, db.autocommit)
764
765 2718 aaronmk
    while True:
766
        # Create function definition
767
        query = '''\
768 2704 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''(value text)
769
RETURNS '''+type_+'''
770
LANGUAGE plpgsql
771
'''
772 2718 aaronmk
        if not save_errors: query += 'IMMUTABLE\n'
773
        query += '''\
774 2704 aaronmk
STRICT
775
AS $$
776
BEGIN
777
    /* The explicit cast to the return type is needed to make the cast happen
778
    inside the try block. (Implicit casts to the return type happen at the end
779
    of the function, outside any block.) */
780
    RETURN value::'''+type_+''';
781
EXCEPTION
782
    WHEN data_exception THEN
783
'''
784 2718 aaronmk
        if save_errors:
785
            cols = [sql_gen.Literal(c.name) for c in srcs]
786
            query += '''\
787 2704 aaronmk
        -- Save error in errors table.
788
        BEGIN
789 2719 aaronmk
            -- Insert the value and error for *each* source column.
790 2704 aaronmk
            INSERT INTO '''+errors_table.to_str(db)+'''
791 2720 aaronmk
            ("column", value, error_code, error)
792 2710 aaronmk
            SELECT *
793
            FROM (VALUES '''+(', '.join(('('+c.to_str(db)+')' for c in cols))
794 2704 aaronmk
                )+''') AS c
795 2720 aaronmk
            CROSS JOIN (VALUES (value, SQLSTATE, SQLERRM)) AS v
796 2704 aaronmk
            ;
797
        EXCEPTION
798
            WHEN unique_violation THEN NULL; -- ignore duplicate key
799
        END;
800
801
'''
802 2718 aaronmk
        query += '''\
803 2704 aaronmk
        RAISE WARNING '%', SQLERRM;
804
        RETURN NULL;
805
END;
806
$$;
807
'''
808 2718 aaronmk
809
        # Create function
810
        try:
811
            run_query(db, query, recover=True, cacheable=True,
812
                log_ignore_excs=(DuplicateFunctionException,))
813
            break # successful
814
        except DuplicateFunctionException:
815
            if save_errors: function.name = next_version(function.name)
816
                # try again with next version of name
817
            else: break # plain cast function, so only need one version
818 2704 aaronmk
819
    return sql_gen.FunctionCall(function, col)
820
821 2414 aaronmk
##### Database structure queries
822
823 2426 aaronmk
def table_row_count(db, table, recover=None):
824
    return value(run_query(db, *mk_select(db, table, [sql_gen.row_count],
825 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
826 2426 aaronmk
827 2414 aaronmk
def table_cols(db, table, recover=None):
828
    return list(col_names(select(db, table, limit=0, order_by=None,
829 2443 aaronmk
        recover=recover, log_level=4)))
830 2414 aaronmk
831 2291 aaronmk
def pkey(db, table, recover=None):
832 832 aaronmk
    '''Assumed to be first column in table'''
833 2339 aaronmk
    return table_cols(db, table, recover)[0]
834 832 aaronmk
835 2559 aaronmk
not_null_col = 'not_null_col'
836 2340 aaronmk
837
def table_not_null_col(db, table, recover=None):
838
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
839
    if not_null_col in table_cols(db, table, recover): return not_null_col
840
    else: return pkey(db, table, recover)
841
842 853 aaronmk
def index_cols(db, table, index):
843
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
844
    automatically created. When you don't know whether something is a UNIQUE
845
    constraint or a UNIQUE index, use this function.'''
846 1909 aaronmk
    module = util.root_module(db.db)
847
    if module == 'psycopg2':
848
        return list(values(run_query(db, '''\
849 853 aaronmk
SELECT attname
850 866 aaronmk
FROM
851
(
852
        SELECT attnum, attname
853
        FROM pg_index
854
        JOIN pg_class index ON index.oid = indexrelid
855
        JOIN pg_class table_ ON table_.oid = indrelid
856
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
857
        WHERE
858
            table_.relname = %(table)s
859
            AND index.relname = %(index)s
860
    UNION
861
        SELECT attnum, attname
862
        FROM
863
        (
864
            SELECT
865
                indrelid
866
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
867
                    AS indkey
868
            FROM pg_index
869
            JOIN pg_class index ON index.oid = indexrelid
870
            JOIN pg_class table_ ON table_.oid = indrelid
871
            WHERE
872
                table_.relname = %(table)s
873
                AND index.relname = %(index)s
874
        ) s
875
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
876
) s
877 853 aaronmk
ORDER BY attnum
878
''',
879 2443 aaronmk
            {'table': table, 'index': index}, cacheable=True, log_level=4)))
880 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
881
        ' database')
882 853 aaronmk
883 464 aaronmk
def constraint_cols(db, table, constraint):
884 1849 aaronmk
    module = util.root_module(db.db)
885 464 aaronmk
    if module == 'psycopg2':
886
        return list(values(run_query(db, '''\
887
SELECT attname
888
FROM pg_constraint
889
JOIN pg_class ON pg_class.oid = conrelid
890
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
891
WHERE
892
    relname = %(table)s
893
    AND conname = %(constraint)s
894
ORDER BY attnum
895
''',
896
            {'table': table, 'constraint': constraint})))
897
    else: raise NotImplementedError("Can't list constraint columns for "+module+
898
        ' database')
899
900 2096 aaronmk
row_num_col = '_row_num'
901
902 2688 aaronmk
def add_index(db, exprs, table=None, unique=False):
903
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
904 2538 aaronmk
    Currently, only function calls are supported as expressions.
905
    '''
906 2688 aaronmk
    if not lists.is_seq(exprs): exprs = [exprs]
907 2538 aaronmk
908 2688 aaronmk
    # Parse exprs
909
    old_exprs = exprs[:]
910
    exprs = []
911
    cols = []
912
    for i, expr in enumerate(old_exprs):
913
        expr = copy.deepcopy(expr) # don't modify input!
914
        expr = sql_gen.as_Col(expr)
915
916
        # Extract col
917
        if isinstance(expr, sql_gen.FunctionCall):
918
            col = expr.args[0]
919
            expr = sql_gen.Expr(expr)
920
        else: col = expr
921
922
        # Extract table
923
        if table == None:
924
            assert sql_gen.is_table_col(col)
925
            table = col.table
926
927
        col.table = None
928
929
        exprs.append(expr)
930
        cols.append(col)
931 2408 aaronmk
932 2688 aaronmk
    table = sql_gen.as_Table(table)
933
    index = sql_gen.Table(str(sql_gen.Col(','.join(map(str, cols)), table)))
934
935
    str_ = 'CREATE'
936
    if unique: str_ += ' UNIQUE'
937
    str_ += ' INDEX '+index.to_str(db)+' ON '+table.to_str(db)+' ('+(
938
        ', '.join((v.to_str(db) for v in exprs)))+')'
939
940
    try: run_query(db, str_, recover=True, cacheable=True, log_level=3)
941 2408 aaronmk
    except DuplicateTableException: pass # index already existed
942
943 2686 aaronmk
def add_pkey(db, table, cols=None, recover=None):
944
    '''Adds a primary key.
945
    @param cols [sql_gen.Col,...] The columns in the primary key.
946
        Defaults to the first column in the table.
947 2406 aaronmk
    @pre The table must not already have a primary key.
948
    '''
949
    table = sql_gen.as_Table(table)
950 2686 aaronmk
    if cols == None: cols = [pkey(db, table, recover)]
951
    col_strs = [sql_gen.to_name_only_col(v).to_str(db) for v in cols]
952 2406 aaronmk
953 2590 aaronmk
    index = sql_gen.as_Table(sql_gen.add_suffix(table.name, '_pkey'))
954 2614 aaronmk
    try:
955
        run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD CONSTRAINT '
956 2686 aaronmk
            +index.to_str(db)+' PRIMARY KEY ('+(', '.join(col_strs))+')',
957 2614 aaronmk
            recover=True, cacheable=True, log_level=3,
958
            log_ignore_excs=(DuplicateTableException,))
959
    except DuplicateTableException, e:
960
        index.name = next_version(index.name)
961
        # try again with next version of name
962 2406 aaronmk
963 2086 aaronmk
def add_row_num(db, table):
964 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
965
    be the primary key.'''
966 2320 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
967 2096 aaronmk
    run_query(db, 'ALTER TABLE '+table+' ADD COLUMN '+row_num_col
968 2443 aaronmk
        +' serial NOT NULL PRIMARY KEY', log_level=3)
969 2086 aaronmk
970 2681 aaronmk
def create_table(db, table, cols, has_pkey=True):
971 2675 aaronmk
    '''Creates a table.
972 2681 aaronmk
    @param cols [sql_gen.TypedCol,...] The column names and types
973
    @param has_pkey If set, the first column becomes the primary key.
974 2675 aaronmk
    '''
975
    table = sql_gen.as_Table(table)
976
977 2681 aaronmk
    if has_pkey:
978
        cols[0] = pkey = copy.copy(cols[0]) # don't modify input!
979
        pkey.type += ' NOT NULL PRIMARY KEY'
980
981 2675 aaronmk
    str_ = 'CREATE TABLE '+table.to_str(db)+' (\n'
982 2681 aaronmk
    str_ += '\n, '.join(v.to_str(db) for v in cols)
983
    str_ += '\n);\n'
984 2675 aaronmk
    run_query(db, str_, cacheable=True, log_level=2)
985
986 2684 aaronmk
def vacuum(db, table):
987
    table = sql_gen.as_Table(table)
988
    db.with_autocommit(lambda: run_query(db, 'VACUUM ANALYZE '+table.to_str(db),
989
        log_level=3))
990
991 2732 aaronmk
def truncate(db, table, schema='public'):
992 2733 aaronmk
    return run_query(db, 'TRUNCATE '+table.to_str(db)+' CASCADE')
993 2732 aaronmk
994 2729 aaronmk
def tables(db, schema_like='public', table_like='%', exact=False):
995
    if exact: compare = '='
996
    else: compare = 'LIKE'
997
998 1849 aaronmk
    module = util.root_module(db.db)
999 2548 aaronmk
    params = {'schema_like': schema_like, 'table_like': table_like}
1000 832 aaronmk
    if module == 'psycopg2':
1001 1968 aaronmk
        return values(run_query(db, '''\
1002
SELECT tablename
1003
FROM pg_tables
1004
WHERE
1005 2729 aaronmk
    schemaname '''+compare+''' %(schema_like)s
1006
    AND tablename '''+compare+''' %(table_like)s
1007 1968 aaronmk
ORDER BY tablename
1008
''',
1009 2731 aaronmk
            params, cacheable=True, log_level=4))
1010 1968 aaronmk
    elif module == 'MySQLdb':
1011
        return values(run_query(db, 'SHOW TABLES LIKE %(table_like)s', params,
1012 2731 aaronmk
            cacheable=True, log_level=4))
1013 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
1014 830 aaronmk
1015 2726 aaronmk
def table_exists(db, table):
1016
    table = sql_gen.as_Table(table)
1017 2730 aaronmk
    return list(tables(db, table.schema, table.name, exact=True)) != []
1018 2726 aaronmk
1019
def errors_table(db, table, if_exists=True):
1020
    '''
1021
    @param if_exists If set, returns None if the errors table doesn't exist
1022
    @return None|sql_gen.Table
1023
    '''
1024
    table = sql_gen.as_Table(table)
1025
    if table.srcs != (): table = table.srcs[0]
1026
1027
    errors_table = sql_gen.suffixed_table(table, '.errors')
1028
    if if_exists and not table_exists(db, errors_table): return None
1029
    return errors_table
1030
1031 833 aaronmk
##### Database management
1032
1033 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
1034
    '''For kw_args, see tables()'''
1035
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
1036 833 aaronmk
1037 832 aaronmk
##### Heuristic queries
1038
1039 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
1040 1554 aaronmk
    '''Recovers from errors.
1041 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
1042
    '''
1043 2642 aaronmk
    row = sql_gen.ColDict(db, table, row)
1044 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
1045
1046 471 aaronmk
    try:
1047 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
1048 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
1049
            row_ct_ref[0] += cur.rowcount
1050
        return value(cur)
1051 471 aaronmk
    except DuplicateKeyException, e:
1052 2653 aaronmk
        row = sql_gen.ColDict(db, table,
1053
            util.dict_subset_right_join(row, e.cols))
1054
        return value(select(db, table, [pkey_], row, recover=True))
1055 471 aaronmk
1056 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
1057 830 aaronmk
    '''Recovers from errors'''
1058 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
1059 14 aaronmk
    except StopIteration:
1060 40 aaronmk
        if not create: raise
1061 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
1062 2078 aaronmk
1063 2593 aaronmk
def is_func_result(col):
1064
    return col.table.name.find('(') >= 0 and col.name == 'result'
1065
1066 2592 aaronmk
def into_table_name(out_table, in_tables0, mapping, is_func):
1067 2615 aaronmk
    def in_col_str(in_col):
1068
        in_col = sql_gen.remove_col_rename(in_col)
1069
        if isinstance(in_col, sql_gen.Col):
1070
            table = in_col.table
1071
            if table == in_tables0:
1072
                in_col = sql_gen.to_name_only_col(in_col)
1073
            elif is_func_result(in_col): in_col = table # omit col name
1074
        return str(in_col)
1075
1076 2580 aaronmk
    str_ = str(out_table)
1077
    if is_func:
1078 2615 aaronmk
        str_ += '('
1079 2580 aaronmk
1080 2615 aaronmk
        try: value_in_col = mapping['value']
1081
        except KeyError:
1082 2654 aaronmk
            str_ += ', '.join((str(k)+'='+in_col_str(v)
1083 2615 aaronmk
                for k, v in mapping.iteritems()))
1084
        else: str_ += in_col_str(value_in_col)
1085
1086
        str_ += ')'
1087 2580 aaronmk
    else: str_ += '_pkeys'
1088
    return str_
1089
1090 2508 aaronmk
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None, into=None,
1091 2552 aaronmk
    default=None, is_func=False):
1092 2078 aaronmk
    '''Recovers from errors.
1093
    Only works under PostgreSQL (uses INSERT RETURNING).
1094 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
1095
        tables to join with it using the main input table's pkey
1096 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
1097 2612 aaronmk
        * out_table_col: str (*not* sql_gen.Col)
1098
        * in_table_col: sql_gen.Col|literal-value
1099 2489 aaronmk
    @param into The table to contain the output and input pkeys.
1100 2574 aaronmk
        Defaults to `out_table.name+'_pkeys'`.
1101 2509 aaronmk
    @param default The *output* column to use as the pkey for missing rows.
1102
        If this output column does not exist in the mapping, uses None.
1103 2552 aaronmk
    @param is_func Whether out_table is the name of a SQL function, not a table
1104 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
1105 2078 aaronmk
    '''
1106 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
1107 2312 aaronmk
1108 2450 aaronmk
    def log_debug(msg): db.log_debug(msg, level=1.5)
1109 2505 aaronmk
    def col_ustr(str_):
1110 2567 aaronmk
        return strings.repr_no_u(sql_gen.remove_col_rename(str_))
1111 2450 aaronmk
1112 2486 aaronmk
    log_debug('********** New iteration **********')
1113 2505 aaronmk
    log_debug('Inserting these input columns into '+strings.as_tt(
1114
        out_table.to_str(db))+':\n'+strings.as_table(mapping, ustr=col_ustr))
1115 2463 aaronmk
1116 2382 aaronmk
    # Create input joins from list of input tables
1117
    in_tables_ = in_tables[:] # don't modify input!
1118
    in_tables0 = in_tables_.pop(0) # first table is separate
1119 2735 aaronmk
    errors_table_ = errors_table(db, in_tables0)
1120 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
1121 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
1122 2460 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v,
1123
        {in_pkey: sql_gen.join_same_not_null}) for v in in_tables_]
1124 2131 aaronmk
1125 2592 aaronmk
    if into == None:
1126
        into = into_table_name(out_table, in_tables0, mapping, is_func)
1127
    into = sql_gen.as_Table(into)
1128
1129 2702 aaronmk
    # Set column sources
1130
    in_cols = filter(sql_gen.is_table_col, mapping.values())
1131
    for col in in_cols:
1132
        if col.table == in_tables0: col.set_srcs(sql_gen.src_self)
1133
1134 2486 aaronmk
    log_debug('Joining together input tables into temp table')
1135 2395 aaronmk
    # Place in new table for speed and so don't modify input if values edited
1136 2584 aaronmk
    in_table = sql_gen.Table('in')
1137 2702 aaronmk
    mapping = dicts.join(mapping, flatten(db, in_table, input_joins, in_cols,
1138
        preserve=[in_pkey_col], start=0))
1139 2395 aaronmk
    input_joins = [in_table]
1140 2486 aaronmk
    db.log_debug('Temp table: '+strings.as_tt(in_table.to_str(db)), level=2)
1141 2395 aaronmk
1142 2656 aaronmk
    mapping = sql_gen.ColDict(db, out_table, mapping)
1143
        # after applying dicts.join() because that returns a plain dict
1144
1145 2509 aaronmk
    # Resolve default value column
1146
    try: default = mapping[default]
1147
    except KeyError:
1148
        if default != None:
1149
            db.log_debug('Default value column '
1150
                +strings.as_tt(strings.repr_no_u(default))
1151 2511 aaronmk
                +' does not exist in mapping, falling back to None', level=2.1)
1152 2509 aaronmk
            default = None
1153
1154 2279 aaronmk
    out_pkey = pkey(db, out_table, recover=True)
1155 2285 aaronmk
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
1156 2142 aaronmk
1157 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
1158 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
1159
1160 2201 aaronmk
    pkeys_table_exists_ref = [False]
1161 2420 aaronmk
    def insert_into_pkeys(joins, cols):
1162
        query, params = mk_select(db, joins, cols, order_by=None, start=0)
1163 2201 aaronmk
        if pkeys_table_exists_ref[0]:
1164 2489 aaronmk
            insert_select(db, into, pkeys_names, query, params)
1165 2201 aaronmk
        else:
1166 2489 aaronmk
            run_query_into(db, query, params, into=into)
1167 2201 aaronmk
            pkeys_table_exists_ref[0] = True
1168
1169 2429 aaronmk
    limit_ref = [None]
1170 2380 aaronmk
    conds = set()
1171 2233 aaronmk
    distinct_on = []
1172 2325 aaronmk
    def mk_main_select(joins, cols):
1173 2429 aaronmk
        return mk_select(db, joins, cols, conds, distinct_on,
1174
            limit=limit_ref[0], start=0)
1175 2132 aaronmk
1176 2519 aaronmk
    exc_strs = set()
1177 2309 aaronmk
    def log_exc(e):
1178 2519 aaronmk
        e_str = exc.str_(e, first_line_only=True)
1179
        log_debug('Caught exception: '+e_str)
1180
        assert e_str not in exc_strs # avoid infinite loops
1181
        exc_strs.add(e_str)
1182 2735 aaronmk
1183 2451 aaronmk
    def remove_all_rows():
1184 2450 aaronmk
        log_debug('Returning NULL for all rows')
1185 2429 aaronmk
        limit_ref[0] = 0 # just create an empty pkeys table
1186 2735 aaronmk
1187 2409 aaronmk
    def ignore(in_col, value):
1188 2545 aaronmk
        in_col_str = strings.as_tt(repr(in_col))
1189 2544 aaronmk
        db.log_debug('Adding index on '+in_col_str+' to enable fast filtering',
1190
            level=2.5)
1191 2537 aaronmk
        add_index(db, in_col)
1192 2735 aaronmk
1193 2545 aaronmk
        log_debug('Ignoring rows with '+in_col_str+' = '
1194
            +strings.as_tt(repr(value)))
1195 2403 aaronmk
    def remove_rows(in_col, value):
1196 2409 aaronmk
        ignore(in_col, value)
1197 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
1198
        assert cond not in conds # avoid infinite loops
1199 2380 aaronmk
        conds.add(cond)
1200 2403 aaronmk
    def invalid2null(in_col, value):
1201 2409 aaronmk
        ignore(in_col, value)
1202 2403 aaronmk
        update(db, in_table, [(in_col, None)],
1203
            sql_gen.ColValueCond(in_col, value))
1204 2245 aaronmk
1205 2589 aaronmk
    def insert_pkeys_table(which):
1206
        return sql_gen.Table(sql_gen.add_suffix(in_table.name,
1207
            '_insert_'+which+'_pkeys'))
1208
    insert_out_pkeys = insert_pkeys_table('out')
1209
    insert_in_pkeys = insert_pkeys_table('in')
1210
1211 2206 aaronmk
    # Do inserts and selects
1212 2642 aaronmk
    join_cols = sql_gen.ColDict(db, out_table)
1213 2206 aaronmk
    while True:
1214 2521 aaronmk
        if limit_ref[0] == 0: # special case
1215
            log_debug('Creating an empty pkeys table')
1216
            cur = run_query_into(db, *mk_select(db, out_table, [out_pkey],
1217
                limit=limit_ref[0]), into=insert_out_pkeys)
1218
            break # don't do main case
1219
1220 2303 aaronmk
        has_joins = join_cols != {}
1221
1222 2305 aaronmk
        # Prepare to insert new rows
1223 2325 aaronmk
        insert_joins = input_joins[:] # don't modify original!
1224 2403 aaronmk
        insert_args = dict(recover=True, cacheable=False)
1225 2303 aaronmk
        if has_joins:
1226 2317 aaronmk
            distinct_on = [v.to_Col() for v in join_cols.values()]
1227 2325 aaronmk
            insert_joins.append(sql_gen.Join(out_table, join_cols,
1228
                sql_gen.filter_out))
1229
        else:
1230 2404 aaronmk
            insert_args.update(dict(returning=out_pkey, into=insert_out_pkeys))
1231 2520 aaronmk
        main_select = mk_main_select(insert_joins, mapping.values())[0]
1232 2303 aaronmk
1233 2486 aaronmk
        log_debug('Trying to insert new rows')
1234 2206 aaronmk
        try:
1235 2518 aaronmk
            cur = insert_select(db, out_table, mapping.keys(), main_select,
1236
                **insert_args)
1237 2357 aaronmk
            break # insert successful
1238 2206 aaronmk
        except DuplicateKeyException, e:
1239 2309 aaronmk
            log_exc(e)
1240
1241 2258 aaronmk
            old_join_cols = join_cols.copy()
1242 2565 aaronmk
            join_cols.update(util.dict_subset_right_join(mapping, e.cols))
1243 2486 aaronmk
            log_debug('Ignoring existing rows, comparing on these columns:\n'
1244 2505 aaronmk
                +strings.as_inline_table(join_cols, ustr=col_ustr))
1245 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
1246 2230 aaronmk
        except NullValueException, e:
1247 2309 aaronmk
            log_exc(e)
1248
1249 2230 aaronmk
            out_col, = e.cols
1250
            try: in_col = mapping[out_col]
1251 2356 aaronmk
            except KeyError:
1252 2486 aaronmk
                log_debug('Missing mapping for NOT NULL column '+out_col)
1253 2451 aaronmk
                remove_all_rows()
1254 2403 aaronmk
            else: remove_rows(in_col, None)
1255 2542 aaronmk
        except FunctionValueException, e:
1256
            log_exc(e)
1257
1258
            func_name = e.name
1259
            value = e.value
1260
            for out_col, in_col in mapping.iteritems():
1261 2562 aaronmk
                invalid2null(sql_gen.unwrap_func_call(in_col, func_name), value)
1262 2525 aaronmk
        except MissingCastException, e:
1263
            log_exc(e)
1264
1265
            out_col = e.col
1266 2706 aaronmk
            type_ = e.type
1267
1268 2724 aaronmk
            log_debug('Casting '+strings.as_tt(out_col)+' input to '
1269
                +strings.as_tt(type_))
1270 2735 aaronmk
            def wrap_func(col): return cast(db, type_, col, errors_table_)
1271 2705 aaronmk
            mapping[out_col] = sql_gen.wrap(wrap_func, mapping[out_col])
1272 2429 aaronmk
        except DatabaseErrors, e:
1273
            log_exc(e)
1274
1275 2531 aaronmk
            msg = 'No handler for exception: '+exc.str_(e)
1276 2451 aaronmk
            warnings.warn(DbWarning(msg))
1277
            log_debug(msg)
1278
            remove_all_rows()
1279 2358 aaronmk
        # after exception handled, rerun loop with additional constraints
1280 2132 aaronmk
1281 2357 aaronmk
    if row_ct_ref != None and cur.rowcount >= 0:
1282
        row_ct_ref[0] += cur.rowcount
1283
1284
    if has_joins:
1285
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
1286 2486 aaronmk
        log_debug('Getting output table pkeys of existing/inserted rows')
1287 2420 aaronmk
        insert_into_pkeys(select_joins, pkeys_cols)
1288 2357 aaronmk
    else:
1289 2404 aaronmk
        add_row_num(db, insert_out_pkeys) # for joining with input pkeys
1290 2357 aaronmk
1291 2486 aaronmk
        log_debug('Getting input table pkeys of inserted rows')
1292 2357 aaronmk
        run_query_into(db, *mk_main_select(input_joins, [in_pkey]),
1293 2404 aaronmk
            into=insert_in_pkeys)
1294
        add_row_num(db, insert_in_pkeys) # for joining with output pkeys
1295 2357 aaronmk
1296 2428 aaronmk
        assert table_row_count(db, insert_out_pkeys) == table_row_count(db,
1297
            insert_in_pkeys)
1298
1299 2486 aaronmk
        log_debug('Combining output and input pkeys in inserted order')
1300 2404 aaronmk
        pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
1301 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
1302 2420 aaronmk
        insert_into_pkeys(pkey_joins, pkeys_names)
1303 2357 aaronmk
1304 2486 aaronmk
    db.log_debug('Adding pkey on pkeys table to enable fast joins', level=2.5)
1305 2594 aaronmk
    add_pkey(db, into)
1306 2407 aaronmk
1307 2508 aaronmk
    log_debug('Setting pkeys of missing rows to '+strings.as_tt(repr(default)))
1308 2489 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(into,
1309 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
1310
        # must use join_same_not_null or query will take forever
1311 2420 aaronmk
    insert_into_pkeys(missing_rows_joins,
1312 2508 aaronmk
        [in_pkey_col, sql_gen.NamedCol(out_pkey, default)])
1313 2357 aaronmk
1314 2489 aaronmk
    assert table_row_count(db, into) == table_row_count(db, in_table)
1315 2428 aaronmk
1316 2702 aaronmk
    srcs = []
1317
    if is_func: srcs = sql_gen.cols_srcs(in_cols)
1318
    return sql_gen.Col(out_pkey, into, srcs)
1319 2115 aaronmk
1320
##### Data cleanup
1321
1322 2290 aaronmk
def cleanup_table(db, table, cols):
1323 2677 aaronmk
    table = sql_gen.as_Table(table)
1324
    cols = map(sql_gen.as_Col, cols)
1325 2115 aaronmk
1326 2677 aaronmk
    expr = ('nullif(nullif(trim(both from %s), '+db.esc_value('')+'), '
1327
        +db.esc_value(r'\N')+')')
1328
    changes = [(v, sql_gen.CustomCode(expr % v.to_str(db)))
1329
        for v in cols]
1330 2115 aaronmk
1331 2677 aaronmk
    update(db, table, changes)