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
14 862 aaronmk
import strings
15 131 aaronmk
import util
16 11 aaronmk
17 832 aaronmk
##### Exceptions
18
19 2170 aaronmk
def get_cur_query(cur, input_query=None, input_params=None):
20 2168 aaronmk
    raw_query = None
21
    if hasattr(cur, 'query'): raw_query = cur.query
22
    elif hasattr(cur, '_last_executed'): raw_query = cur._last_executed
23 2170 aaronmk
24
    if raw_query != None: return raw_query
25
    else: return repr(input_query)+' % '+repr(input_params)
26 14 aaronmk
27 2170 aaronmk
def _add_cursor_info(e, *args, **kw_args):
28
    '''For params, see get_cur_query()'''
29
    exc.add_msg(e, 'query: '+str(get_cur_query(*args, **kw_args)))
30 135 aaronmk
31 300 aaronmk
class DbException(exc.ExceptionWithCause):
32 14 aaronmk
    def __init__(self, msg, cause=None, cur=None):
33 2145 aaronmk
        exc.ExceptionWithCause.__init__(self, msg, cause, cause_newline=True)
34 14 aaronmk
        if cur != None: _add_cursor_info(self, cur)
35
36 2143 aaronmk
class ExceptionWithName(DbException):
37
    def __init__(self, name, cause=None):
38 2145 aaronmk
        DbException.__init__(self, 'for name: '+str(name), cause)
39 2143 aaronmk
        self.name = name
40 360 aaronmk
41 468 aaronmk
class ExceptionWithColumns(DbException):
42
    def __init__(self, cols, cause=None):
43 2145 aaronmk
        DbException.__init__(self, 'for columns: '+(', '.join(cols)), cause)
44 468 aaronmk
        self.cols = cols
45 11 aaronmk
46 2143 aaronmk
class NameException(DbException): pass
47
48 468 aaronmk
class DuplicateKeyException(ExceptionWithColumns): pass
49 13 aaronmk
50 468 aaronmk
class NullValueException(ExceptionWithColumns): pass
51 13 aaronmk
52 2143 aaronmk
class DuplicateTableException(ExceptionWithName): pass
53
54 89 aaronmk
class EmptyRowException(DbException): pass
55
56 865 aaronmk
##### Warnings
57
58
class DbWarning(UserWarning): pass
59
60 1930 aaronmk
##### Result retrieval
61
62
def col_names(cur): return (col[0] for col in cur.description)
63
64
def rows(cur): return iter(lambda: cur.fetchone(), None)
65
66
def consume_rows(cur):
67
    '''Used to fetch all rows so result will be cached'''
68
    iters.consume_iter(rows(cur))
69
70
def next_row(cur): return rows(cur).next()
71
72
def row(cur):
73
    row_ = next_row(cur)
74
    consume_rows(cur)
75
    return row_
76
77
def next_value(cur): return next_row(cur)[0]
78
79
def value(cur): return row(cur)[0]
80
81
def values(cur): return iters.func_iter(lambda: next_value(cur))
82
83
def value_or_none(cur):
84
    try: return value(cur)
85
    except StopIteration: return None
86
87 2101 aaronmk
##### Input validation
88
89
def clean_name(name): return re.sub(r'\W', r'', name)
90
91
def check_name(name):
92
    if re.search(r'\W', name) != None: raise NameException('Name "'+name
93
        +'" may contain only alphanumeric characters and _')
94
95
def esc_name_by_module(module, name, ignore_case=False):
96
    if module == 'psycopg2':
97
        if ignore_case:
98
            # Don't enclose in quotes because this disables case-insensitivity
99
            check_name(name)
100
            return name
101
        else: quote = '"'
102
    elif module == 'MySQLdb': quote = '`'
103
    else: raise NotImplementedError("Can't escape name for "+module+' database')
104
    return quote + name.replace(quote, '') + quote
105
106
def esc_name_by_engine(engine, name, **kw_args):
107
    return esc_name_by_module(db_engines[engine][0], name, **kw_args)
108
109
def esc_name(db, name, **kw_args):
110
    return esc_name_by_module(util.root_module(db.db), name, **kw_args)
111
112
def qual_name(db, schema, table):
113
    def esc_name_(name): return esc_name(db, name)
114
    table = esc_name_(table)
115
    if schema != None: return esc_name_(schema)+'.'+table
116
    else: return table
117
118 1869 aaronmk
##### Database connections
119 1849 aaronmk
120 2097 aaronmk
db_config_names = ['engine', 'host', 'user', 'password', 'database', 'schemas']
121 1926 aaronmk
122 1869 aaronmk
db_engines = {
123
    'MySQL': ('MySQLdb', {'password': 'passwd', 'database': 'db'}),
124
    'PostgreSQL': ('psycopg2', {}),
125
}
126
127
DatabaseErrors_set = set([DbException])
128
DatabaseErrors = tuple(DatabaseErrors_set)
129
130
def _add_module(module):
131
    DatabaseErrors_set.add(module.DatabaseError)
132
    global DatabaseErrors
133
    DatabaseErrors = tuple(DatabaseErrors_set)
134
135
def db_config_str(db_config):
136
    return db_config['engine']+' database '+db_config['database']
137
138 1909 aaronmk
def _query_lookup(query, params): return (query, dicts.make_hashable(params))
139 1894 aaronmk
140 1901 aaronmk
log_debug_none = lambda msg: None
141
142 1849 aaronmk
class DbConn:
143 2047 aaronmk
    def __init__(self, db_config, serializable=True, log_debug=log_debug_none,
144 2050 aaronmk
        caching=True):
145 1869 aaronmk
        self.db_config = db_config
146
        self.serializable = serializable
147 1901 aaronmk
        self.log_debug = log_debug
148 2047 aaronmk
        self.caching = caching
149 1869 aaronmk
150
        self.__db = None
151 1889 aaronmk
        self.query_results = {}
152 2139 aaronmk
        self._savepoint = 0
153 1869 aaronmk
154
    def __getattr__(self, name):
155
        if name == '__dict__': raise Exception('getting __dict__')
156
        if name == 'db': return self._db()
157
        else: raise AttributeError()
158
159
    def __getstate__(self):
160
        state = copy.copy(self.__dict__) # shallow copy
161 1915 aaronmk
        state['log_debug'] = None # don't pickle the debug callback
162 1869 aaronmk
        state['_DbConn__db'] = None # don't pickle the connection
163
        return state
164
165 2165 aaronmk
    def connected(self): return self.__db != None
166
167 1869 aaronmk
    def _db(self):
168
        if self.__db == None:
169
            # Process db_config
170
            db_config = self.db_config.copy() # don't modify input!
171 2097 aaronmk
            schemas = db_config.pop('schemas', None)
172 1869 aaronmk
            module_name, mappings = db_engines[db_config.pop('engine')]
173
            module = __import__(module_name)
174
            _add_module(module)
175
            for orig, new in mappings.iteritems():
176
                try: util.rename_key(db_config, orig, new)
177
                except KeyError: pass
178
179
            # Connect
180
            self.__db = module.connect(**db_config)
181
182
            # Configure connection
183
            if self.serializable: run_raw_query(self,
184
                'SET TRANSACTION ISOLATION LEVEL SERIALIZABLE')
185 2101 aaronmk
            if schemas != None:
186
                schemas_ = ''.join((esc_name(self, s)+', '
187
                    for s in schemas.split(',')))
188
                run_raw_query(self, "SELECT set_config('search_path', \
189
%s || current_setting('search_path'), false)", [schemas_])
190 1869 aaronmk
191
        return self.__db
192 1889 aaronmk
193 1891 aaronmk
    class DbCursor(Proxy):
194 1927 aaronmk
        def __init__(self, outer):
195 1891 aaronmk
            Proxy.__init__(self, outer.db.cursor())
196 1927 aaronmk
            self.query_results = outer.query_results
197 1894 aaronmk
            self.query_lookup = None
198 1891 aaronmk
            self.result = []
199 1889 aaronmk
200 1894 aaronmk
        def execute(self, query, params=None):
201 1930 aaronmk
            self._is_insert = query.upper().find('INSERT') >= 0
202 1894 aaronmk
            self.query_lookup = _query_lookup(query, params)
203 2148 aaronmk
            try:
204
                try: return_value = self.inner.execute(query, params)
205
                finally: self.query = get_cur_query(self.inner)
206 1904 aaronmk
            except Exception, e:
207 2170 aaronmk
                _add_cursor_info(e, self, query, params)
208 1904 aaronmk
                self.result = e # cache the exception as the result
209
                self._cache_result()
210
                raise
211 1930 aaronmk
            # Fetch all rows so result will be cached
212
            if self.rowcount == 0 and not self._is_insert: consume_rows(self)
213 1894 aaronmk
            return return_value
214
215 1891 aaronmk
        def fetchone(self):
216
            row = self.inner.fetchone()
217 1899 aaronmk
            if row != None: self.result.append(row)
218
            # otherwise, fetched all rows
219 1904 aaronmk
            else: self._cache_result()
220
            return row
221
222
        def _cache_result(self):
223 1906 aaronmk
            # For inserts, only cache exceptions since inserts are not
224
            # idempotent, but an invalid insert will always be invalid
225 1930 aaronmk
            if self.query_results != None and (not self._is_insert
226 1906 aaronmk
                or isinstance(self.result, Exception)):
227
228 1894 aaronmk
                assert self.query_lookup != None
229 1916 aaronmk
                self.query_results[self.query_lookup] = self.CacheCursor(
230
                    util.dict_subset(dicts.AttrsDictView(self),
231
                    ['query', 'result', 'rowcount', 'description']))
232 1906 aaronmk
233 1916 aaronmk
        class CacheCursor:
234
            def __init__(self, cached_result): self.__dict__ = cached_result
235
236 1927 aaronmk
            def execute(self, *args, **kw_args):
237 1916 aaronmk
                if isinstance(self.result, Exception): raise self.result
238
                # otherwise, result is a rows list
239
                self.iter = iter(self.result)
240
241
            def fetchone(self):
242
                try: return self.iter.next()
243
                except StopIteration: return None
244 1891 aaronmk
245 1894 aaronmk
    def run_query(self, query, params=None, cacheable=False):
246 2148 aaronmk
        '''Translates known DB errors to typed exceptions:
247
        See self.DbCursor.execute().'''
248 2167 aaronmk
        assert query != None
249
250 2047 aaronmk
        if not self.caching: cacheable = False
251 1903 aaronmk
        used_cache = False
252
        try:
253 1927 aaronmk
            # Get cursor
254
            if cacheable:
255
                query_lookup = _query_lookup(query, params)
256
                try:
257
                    cur = self.query_results[query_lookup]
258
                    used_cache = True
259
                except KeyError: cur = self.DbCursor(self)
260
            else: cur = self.db.cursor()
261
262
            # Run query
263 2148 aaronmk
            cur.execute(query, params)
264 1903 aaronmk
        finally:
265
            if self.log_debug != log_debug_none: # only compute msg if needed
266
                if used_cache: cache_status = 'Cache hit'
267
                elif cacheable: cache_status = 'Cache miss'
268
                else: cache_status = 'Non-cacheable'
269 1927 aaronmk
                self.log_debug(cache_status+': '
270 2170 aaronmk
                    +strings.one_line(str(get_cur_query(cur, query, params))))
271 1903 aaronmk
272
        return cur
273 1914 aaronmk
274
    def is_cached(self, query, params=None):
275
        return _query_lookup(query, params) in self.query_results
276 2139 aaronmk
277
    def with_savepoint(self, func):
278 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
279 2139 aaronmk
        self.run_query('SAVEPOINT '+savepoint)
280
        self._savepoint += 1
281
        try:
282
            try: return_val = func()
283
            finally:
284
                self._savepoint -= 1
285
                assert self._savepoint >= 0
286
        except:
287
            self.run_query('ROLLBACK TO SAVEPOINT '+savepoint)
288
            raise
289
        else:
290
            self.run_query('RELEASE SAVEPOINT '+savepoint)
291
            return return_val
292 1849 aaronmk
293 1869 aaronmk
connect = DbConn
294
295 832 aaronmk
##### Querying
296
297 1894 aaronmk
def run_raw_query(db, *args, **kw_args):
298 2085 aaronmk
    '''For params, see DbConn.run_query()'''
299 1894 aaronmk
    return db.run_query(*args, **kw_args)
300 11 aaronmk
301 2068 aaronmk
def mogrify(db, query, params):
302
    module = util.root_module(db.db)
303
    if module == 'psycopg2': return db.db.cursor().mogrify(query, params)
304
    else: raise NotImplementedError("Can't mogrify query for "+module+
305
        ' database')
306
307 832 aaronmk
##### Recoverable querying
308 15 aaronmk
309 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
310 11 aaronmk
311 1894 aaronmk
def run_query(db, query, params=None, recover=None, cacheable=False):
312 830 aaronmk
    if recover == None: recover = False
313
314 2148 aaronmk
    try:
315
        def run(): return run_raw_query(db, query, params, cacheable)
316
        if recover and not db.is_cached(query, params):
317
            return with_savepoint(db, run)
318
        else: return run() # don't need savepoint if cached
319
    except Exception, e:
320
        if not recover: raise # need savepoint to run index_cols()
321
        msg = str(e)
322
        match = re.search(r'duplicate key value violates unique constraint '
323
            r'"((_?[^\W_]+)_[^"]+)"', msg)
324
        if match:
325
            constraint, table = match.groups()
326
            try: cols = index_cols(db, table, constraint)
327
            except NotImplementedError: raise e
328
            else: raise DuplicateKeyException(cols, e)
329
        match = re.search(r'null value in column "(\w+)" violates not-null '
330
            'constraint', msg)
331
        if match: raise NullValueException([match.group(1)], e)
332
        match = re.search(r'relation "(\w+)" already exists', msg)
333
        if match: raise DuplicateTableException(match.group(1), e)
334
        raise # no specific exception raised
335 830 aaronmk
336 832 aaronmk
##### Basic queries
337
338 2153 aaronmk
def next_version(name):
339
    '''Prepends the version # so it won't be removed if the name is truncated'''
340 2163 aaronmk
    version = 1 # first existing name was version 0
341 2153 aaronmk
    match = re.match(r'^v(\d+)_(.*)$', name)
342
    if match:
343
        version = int(match.group(1))+1
344
        name = match.group(2)
345
    return 'v'+str(version)+'_'+name
346
347 2151 aaronmk
def run_query_into(db, query, params, into_ref=None, *args, **kw_args):
348 2085 aaronmk
    '''Outputs a query to a temp table.
349
    For params, see run_query().
350
    '''
351 2151 aaronmk
    if into_ref == None: return run_query(db, query, params, *args, **kw_args)
352 2085 aaronmk
    else: # place rows in temp table
353 2151 aaronmk
        check_name(into_ref[0])
354 2153 aaronmk
        kw_args['recover'] = True
355
        while True:
356
            try:
357
                return run_query(db, 'CREATE TEMP TABLE '+into_ref[0]+' AS '
358
                    +query, params, *args, **kw_args)
359
                    # CREATE TABLE AS sets rowcount to # rows in query
360
            except DuplicateTableException, e:
361
                into_ref[0] = next_version(into_ref[0])
362
                # try again with next version of name
363 2085 aaronmk
364 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
365
366 2127 aaronmk
join_using = object() # tells mk_select() to join the column with USING
367
368 2121 aaronmk
def mk_select(db, tables, fields=None, conds=None, limit=None, start=None,
369 2120 aaronmk
    order_by=order_by_pkey, table_is_esc=False):
370 1981 aaronmk
    '''
371 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
372 2124 aaronmk
        together: [table0, (table1, dict(right_col=left_col, ...)), ...]
373 1981 aaronmk
    @param fields Use None to select all fields in the table
374
    @param table_is_esc Whether the table name has already been escaped
375 2054 aaronmk
    @return tuple(query, params)
376 1981 aaronmk
    '''
377 2060 aaronmk
    def esc_name_(name): return esc_name(db, name)
378 2058 aaronmk
379 2121 aaronmk
    if not lists.is_seq(tables): tables = [tables]
380 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
381 2121 aaronmk
    table0 = tables.pop(0) # first table is separate
382
383 1135 aaronmk
    if conds == None: conds = {}
384 135 aaronmk
    assert limit == None or type(limit) == int
385 865 aaronmk
    assert start == None or type(start) == int
386 2120 aaronmk
    if order_by == order_by_pkey:
387 2121 aaronmk
        order_by = pkey(db, table0, recover=True, table_is_esc=table_is_esc)
388
    if not table_is_esc: table0 = esc_name_(table0)
389 865 aaronmk
390 2056 aaronmk
    params = []
391
392 2161 aaronmk
    def parse_col(field, default_table=None):
393 2056 aaronmk
        '''Parses fields'''
394 2157 aaronmk
        is_tuple = isinstance(field, tuple)
395
        if is_tuple and len(field) == 1: # field is literal value
396
            value, = field
397 2056 aaronmk
            sql_ = '%s'
398
            params.append(value)
399 2157 aaronmk
        elif is_tuple and len(field) == 2: # field is col with table
400
            table, col = field
401
            if not table_is_esc: table = esc_name_(table)
402
            sql_ = table+'.'+esc_name_(col)
403 2161 aaronmk
        else:
404
            sql_ = esc_name_(field) # field is col name
405
            if default_table != None: sql_ = default_table+'.'+sql_
406 2056 aaronmk
        return sql_
407 11 aaronmk
    def cond(entry):
408 2056 aaronmk
        '''Parses conditions'''
409 13 aaronmk
        col, value = entry
410 2058 aaronmk
        cond_ = esc_name_(col)+' '
411 11 aaronmk
        if value == None: cond_ += 'IS'
412
        else: cond_ += '='
413
        cond_ += ' %s'
414
        return cond_
415 2056 aaronmk
416 1135 aaronmk
    query = 'SELECT '
417
    if fields == None: query += '*'
418 2056 aaronmk
    else: query += ', '.join(map(parse_col, fields))
419 2121 aaronmk
    query += ' FROM '+table0
420 865 aaronmk
421 2122 aaronmk
    # Add joins
422
    left_table = table0
423
    for table, joins in tables:
424
        if not table_is_esc: table = esc_name_(table)
425 2127 aaronmk
        query += ' JOIN '+table
426 2122 aaronmk
427
        def join(entry):
428 2127 aaronmk
            '''Parses non-USING joins'''
429 2124 aaronmk
            right_col, left_col = entry
430
            right_col = table+'.'+esc_name_(right_col)
431 2161 aaronmk
            left_col = parse_col(left_col, left_table)
432 2123 aaronmk
            return (right_col+' = '+left_col
433
                +' OR ('+right_col+' IS NULL AND '+left_col+' IS NULL)')
434 2122 aaronmk
435 2127 aaronmk
        if reduce(operator.and_, (v == join_using for v in joins.itervalues())):
436
            # all cols w/ USING
437 2130 aaronmk
            query += ' USING ('+(', '.join(joins.iterkeys()))+')'
438 2127 aaronmk
        else: query += ' ON '+(' AND '.join(map(join, joins.iteritems())))
439
440 2122 aaronmk
        left_table = table
441
442 865 aaronmk
    missing = True
443 89 aaronmk
    if conds != {}:
444 2122 aaronmk
        query += ' WHERE '+(' AND '.join(map(cond, conds.iteritems())))
445 2056 aaronmk
        params += conds.values()
446 865 aaronmk
        missing = False
447 2120 aaronmk
    if order_by != None: query += ' ORDER BY '+esc_name_(order_by)
448 865 aaronmk
    if limit != None: query += ' LIMIT '+str(limit); missing = False
449
    if start != None:
450
        if start != 0: query += ' OFFSET '+str(start)
451
        missing = False
452
    if missing: warnings.warn(DbWarning(
453
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
454
455 2056 aaronmk
    return (query, params)
456 11 aaronmk
457 2054 aaronmk
def select(db, *args, **kw_args):
458
    '''For params, see mk_select() and run_query()'''
459
    recover = kw_args.pop('recover', None)
460
    cacheable = kw_args.pop('cacheable', True)
461
462
    query, params = mk_select(db, *args, **kw_args)
463
    return run_query(db, query, params, recover, cacheable)
464
465 2066 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, params=None,
466 2070 aaronmk
    returning=None, embeddable=False, table_is_esc=False):
467 1960 aaronmk
    '''
468
    @param returning str|None An inserted column (such as pkey) to return
469 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
470 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
471
        query will be fully cached, not just if it raises an exception.
472 1960 aaronmk
    @param table_is_esc Whether the table name has already been escaped
473
    '''
474 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
475
    if cols == []: cols = None # no cols (all defaults) = unknown col names
476 1960 aaronmk
    if not table_is_esc: check_name(table)
477 2063 aaronmk
478
    # Build query
479
    query = 'INSERT INTO '+table
480
    if cols != None:
481
        map(check_name, cols)
482
        query += ' ('+', '.join(cols)+')'
483
    query += ' '+select_query
484
485
    if returning != None:
486
        check_name(returning)
487
        query += ' RETURNING '+returning
488
489 2070 aaronmk
    if embeddable:
490
        # Create function
491 2083 aaronmk
        function = 'pg_temp.'+('_'.join(map(clean_name,
492
            ['insert', table] + cols)))
493 2070 aaronmk
        return_type = 'SETOF '+table+'.'+returning+'%TYPE'
494
        function_query = '''\
495 2083 aaronmk
CREATE OR REPLACE FUNCTION '''+function+'''() RETURNS '''+return_type+'''
496 2070 aaronmk
    LANGUAGE sql
497
    AS $$'''+mogrify(db, query, params)+''';$$;
498
'''
499 2172 aaronmk
        run_query(db, function_query)
500
            # don't cache because function def could change and then change back
501 2070 aaronmk
502
        # Return query that uses function
503 2134 aaronmk
        return mk_select(db, function+'() AS f ('+returning+')', start=0,
504
            order_by=None, table_is_esc=True)# AS clause requires function alias
505 2070 aaronmk
506 2066 aaronmk
    return (query, params)
507
508
def insert_select(db, *args, **kw_args):
509 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
510 2152 aaronmk
    @param into_ref List with name of temp table to place RETURNING values in
511 2072 aaronmk
    '''
512 2151 aaronmk
    into_ref = kw_args.pop('into_ref', None)
513
    if into_ref != None: kw_args['embeddable'] = True
514 2066 aaronmk
    recover = kw_args.pop('recover', None)
515
    cacheable = kw_args.pop('cacheable', True)
516
517
    query, params = mk_insert_select(db, *args, **kw_args)
518 2153 aaronmk
    return run_query_into(db, query, params, into_ref, recover=recover,
519
        cacheable=cacheable)
520 2063 aaronmk
521 2066 aaronmk
default = object() # tells insert() to use the default value for a column
522
523 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
524 2085 aaronmk
    '''For params, see insert_select()'''
525 1960 aaronmk
    if lists.is_seq(row): cols = None
526
    else:
527
        cols = row.keys()
528
        row = row.values()
529
    row = list(row) # ensure that "!= []" works
530
531 1961 aaronmk
    # Check for special values
532
    labels = []
533
    values = []
534
    for value in row:
535
        if value == default: labels.append('DEFAULT')
536
        else:
537
            labels.append('%s')
538
            values.append(value)
539
540
    # Build query
541 2063 aaronmk
    if values != []: query = ' VALUES ('+(', '.join(labels))+')'
542
    else: query = None
543 1554 aaronmk
544 2064 aaronmk
    return insert_select(db, table, cols, query, values, *args, **kw_args)
545 11 aaronmk
546 135 aaronmk
def last_insert_id(db):
547 1849 aaronmk
    module = util.root_module(db.db)
548 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
549
    elif module == 'MySQLdb': return db.insert_id()
550
    else: return None
551 13 aaronmk
552 1968 aaronmk
def truncate(db, table, schema='public'):
553
    return run_query(db, 'TRUNCATE '+qual_name(db, schema, table)+' CASCADE')
554 832 aaronmk
555
##### Database structure queries
556
557 2084 aaronmk
def pkey(db, table, recover=None, table_is_esc=False):
558 832 aaronmk
    '''Assumed to be first column in table'''
559 2120 aaronmk
    return col_names(select(db, table, limit=0, order_by=None, recover=recover,
560 2084 aaronmk
        table_is_esc=table_is_esc)).next()
561 832 aaronmk
562 853 aaronmk
def index_cols(db, table, index):
563
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
564
    automatically created. When you don't know whether something is a UNIQUE
565
    constraint or a UNIQUE index, use this function.'''
566
    check_name(table)
567
    check_name(index)
568 1909 aaronmk
    module = util.root_module(db.db)
569
    if module == 'psycopg2':
570
        return list(values(run_query(db, '''\
571 853 aaronmk
SELECT attname
572 866 aaronmk
FROM
573
(
574
        SELECT attnum, attname
575
        FROM pg_index
576
        JOIN pg_class index ON index.oid = indexrelid
577
        JOIN pg_class table_ ON table_.oid = indrelid
578
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
579
        WHERE
580
            table_.relname = %(table)s
581
            AND index.relname = %(index)s
582
    UNION
583
        SELECT attnum, attname
584
        FROM
585
        (
586
            SELECT
587
                indrelid
588
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
589
                    AS indkey
590
            FROM pg_index
591
            JOIN pg_class index ON index.oid = indexrelid
592
            JOIN pg_class table_ ON table_.oid = indrelid
593
            WHERE
594
                table_.relname = %(table)s
595
                AND index.relname = %(index)s
596
        ) s
597
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
598
) s
599 853 aaronmk
ORDER BY attnum
600
''',
601 1909 aaronmk
            {'table': table, 'index': index}, cacheable=True)))
602
    else: raise NotImplementedError("Can't list index columns for "+module+
603
        ' database')
604 853 aaronmk
605 464 aaronmk
def constraint_cols(db, table, constraint):
606
    check_name(table)
607
    check_name(constraint)
608 1849 aaronmk
    module = util.root_module(db.db)
609 464 aaronmk
    if module == 'psycopg2':
610
        return list(values(run_query(db, '''\
611
SELECT attname
612
FROM pg_constraint
613
JOIN pg_class ON pg_class.oid = conrelid
614
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
615
WHERE
616
    relname = %(table)s
617
    AND conname = %(constraint)s
618
ORDER BY attnum
619
''',
620
            {'table': table, 'constraint': constraint})))
621
    else: raise NotImplementedError("Can't list constraint columns for "+module+
622
        ' database')
623
624 2096 aaronmk
row_num_col = '_row_num'
625
626 2086 aaronmk
def add_row_num(db, table):
627 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
628
    be the primary key.'''
629 2086 aaronmk
    check_name(table)
630 2096 aaronmk
    run_query(db, 'ALTER TABLE '+table+' ADD COLUMN '+row_num_col
631 2117 aaronmk
        +' serial NOT NULL PRIMARY KEY')
632 2086 aaronmk
633 1968 aaronmk
def tables(db, schema='public', table_like='%'):
634 1849 aaronmk
    module = util.root_module(db.db)
635 1968 aaronmk
    params = {'schema': schema, 'table_like': table_like}
636 832 aaronmk
    if module == 'psycopg2':
637 1968 aaronmk
        return values(run_query(db, '''\
638
SELECT tablename
639
FROM pg_tables
640
WHERE
641
    schemaname = %(schema)s
642
    AND tablename LIKE %(table_like)s
643
ORDER BY tablename
644
''',
645
            params, cacheable=True))
646
    elif module == 'MySQLdb':
647
        return values(run_query(db, 'SHOW TABLES LIKE %(table_like)s', params,
648
            cacheable=True))
649 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
650 830 aaronmk
651 833 aaronmk
##### Database management
652
653 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
654
    '''For kw_args, see tables()'''
655
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
656 833 aaronmk
657 832 aaronmk
##### Heuristic queries
658
659 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
660 1554 aaronmk
    '''Recovers from errors.
661 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
662
    '''
663 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
664
665 471 aaronmk
    try:
666 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
667 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
668
            row_ct_ref[0] += cur.rowcount
669
        return value(cur)
670 471 aaronmk
    except DuplicateKeyException, e:
671 2104 aaronmk
        return value(select(db, table, [pkey_],
672 1069 aaronmk
            util.dict_subset_right_join(row, e.cols), recover=True))
673 471 aaronmk
674 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
675 830 aaronmk
    '''Recovers from errors'''
676
    try: return value(select(db, table, [pkey], row, 1, recover=True))
677 14 aaronmk
    except StopIteration:
678 40 aaronmk
        if not create: raise
679 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
680 2078 aaronmk
681 2134 aaronmk
def put_table(db, out_table, in_tables, mapping, limit=None, start=0,
682
    row_ct_ref=None, table_is_esc=False):
683 2078 aaronmk
    '''Recovers from errors.
684
    Only works under PostgreSQL (uses INSERT RETURNING).
685 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
686
        tables to join with it using the main input table's pkey
687 2133 aaronmk
    @return (table, col) Where the pkeys (from INSERT RETURNING) are made
688 2078 aaronmk
        available
689
    '''
690 2162 aaronmk
    temp_suffix = clean_name(out_table)
691 2158 aaronmk
        # suffix, not prefix, so main name won't be removed if name is truncated
692
    pkeys_ref = ['pkeys_'+temp_suffix]
693 2131 aaronmk
694 2132 aaronmk
    # Join together input tables
695 2131 aaronmk
    in_tables = in_tables[:] # don't modify input!
696
    in_tables0 = in_tables.pop(0) # first table is separate
697
    in_pkey = pkey(db, in_tables0, recover=True, table_is_esc=table_is_esc)
698 2142 aaronmk
    in_joins = [in_tables0] + [(t, {in_pkey: join_using}) for t in in_tables]
699 2131 aaronmk
700 2142 aaronmk
    out_pkey = pkey(db, out_table, recover=True, table_is_esc=table_is_esc)
701
    pkeys_cols = [in_pkey, out_pkey]
702
703 2132 aaronmk
    def mk_select_(cols):
704 2142 aaronmk
        return mk_select(db, in_joins, cols, limit=limit, start=start,
705 2134 aaronmk
            table_is_esc=table_is_esc)
706 2132 aaronmk
707 2158 aaronmk
    out_pkeys_ref = ['out_pkeys_'+temp_suffix]
708 2078 aaronmk
    def insert_():
709 2148 aaronmk
        '''Inserts and capture output pkeys.'''
710 2132 aaronmk
        cur = insert_select(db, out_table, mapping.keys(),
711 2133 aaronmk
            *mk_select_(mapping.values()), returning=out_pkey,
712 2155 aaronmk
            into_ref=out_pkeys_ref, recover=True, table_is_esc=table_is_esc)
713 2078 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
714
            row_ct_ref[0] += cur.rowcount
715 2155 aaronmk
            add_row_num(db, out_pkeys_ref[0]) # for joining it with input pkeys
716 2086 aaronmk
717 2132 aaronmk
        # Get input pkeys corresponding to rows in insert
718 2158 aaronmk
        in_pkeys_ref = ['in_pkeys_'+temp_suffix]
719 2155 aaronmk
        run_query_into(db, *mk_select_([in_pkey]), into_ref=in_pkeys_ref)
720
        add_row_num(db, in_pkeys_ref[0]) # for joining it with output pkeys
721 2086 aaronmk
722 2155 aaronmk
        # Join together output and input pkeys
723 2132 aaronmk
        run_query_into(db, *mk_select(db,
724 2155 aaronmk
            [in_pkeys_ref[0], (out_pkeys_ref[0], {row_num_col: join_using})],
725 2154 aaronmk
            pkeys_cols, start=0), into_ref=pkeys_ref)
726 2132 aaronmk
727 2142 aaronmk
    # Do inserts and selects
728 2148 aaronmk
    try: insert_()
729 2142 aaronmk
    except DuplicateKeyException, e:
730
        join_cols = util.dict_subset_right_join(mapping, e.cols)
731
        joins = in_joins + [(out_table, join_cols)]
732
        run_query_into(db, *mk_select(db, joins, pkeys_cols,
733 2154 aaronmk
            table_is_esc=table_is_esc), into_ref=pkeys_ref, recover=True)
734 2142 aaronmk
735 2154 aaronmk
    return (pkeys_ref[0], out_pkey)
736 2115 aaronmk
737
##### Data cleanup
738
739
def cleanup_table(db, table, cols, table_is_esc=False):
740
    def esc_name_(name): return esc_name(db, name)
741
742
    if not table_is_esc: check_name(table)
743
    cols = map(esc_name_, cols)
744
745
    run_query(db, 'UPDATE '+table+' SET\n'+(',\n'.join(('\n'+col
746
        +' = nullif(nullif(trim(both from '+col+"), %(null0)s), %(null1)s)"
747
            for col in cols))),
748
        dict(null0='', null1=r'\N'))