Project

General

Profile

1 11 aaronmk
# Database access
2
3 1869 aaronmk
import copy
4 11 aaronmk
import re
5 865 aaronmk
import warnings
6 11 aaronmk
7 300 aaronmk
import exc
8 1909 aaronmk
import dicts
9 1893 aaronmk
import iters
10 1960 aaronmk
import lists
11 1889 aaronmk
from Proxy import Proxy
12 1872 aaronmk
import rand
13 862 aaronmk
import strings
14 131 aaronmk
import util
15 11 aaronmk
16 832 aaronmk
##### Exceptions
17
18 135 aaronmk
def get_cur_query(cur):
19
    if hasattr(cur, 'query'): return cur.query
20
    elif hasattr(cur, '_last_executed'): return cur._last_executed
21
    else: return None
22 14 aaronmk
23 300 aaronmk
def _add_cursor_info(e, cur): exc.add_msg(e, 'query: '+get_cur_query(cur))
24 135 aaronmk
25 300 aaronmk
class DbException(exc.ExceptionWithCause):
26 14 aaronmk
    def __init__(self, msg, cause=None, cur=None):
27 300 aaronmk
        exc.ExceptionWithCause.__init__(self, msg, cause)
28 14 aaronmk
        if cur != None: _add_cursor_info(self, cur)
29
30 360 aaronmk
class NameException(DbException): pass
31
32 468 aaronmk
class ExceptionWithColumns(DbException):
33
    def __init__(self, cols, cause=None):
34
        DbException.__init__(self, 'columns: ' + ', '.join(cols), cause)
35
        self.cols = cols
36 11 aaronmk
37 468 aaronmk
class DuplicateKeyException(ExceptionWithColumns): pass
38 13 aaronmk
39 468 aaronmk
class NullValueException(ExceptionWithColumns): pass
40 13 aaronmk
41 89 aaronmk
class EmptyRowException(DbException): pass
42
43 865 aaronmk
##### Warnings
44
45
class DbWarning(UserWarning): pass
46
47 1930 aaronmk
##### Result retrieval
48
49
def col_names(cur): return (col[0] for col in cur.description)
50
51
def rows(cur): return iter(lambda: cur.fetchone(), None)
52
53
def consume_rows(cur):
54
    '''Used to fetch all rows so result will be cached'''
55
    iters.consume_iter(rows(cur))
56
57
def next_row(cur): return rows(cur).next()
58
59
def row(cur):
60
    row_ = next_row(cur)
61
    consume_rows(cur)
62
    return row_
63
64
def next_value(cur): return next_row(cur)[0]
65
66
def value(cur): return row(cur)[0]
67
68
def values(cur): return iters.func_iter(lambda: next_value(cur))
69
70
def value_or_none(cur):
71
    try: return value(cur)
72
    except StopIteration: return None
73
74 1869 aaronmk
##### Database connections
75 1849 aaronmk
76 1926 aaronmk
db_config_names = ['engine', 'host', 'user', 'password', 'database']
77
78 1869 aaronmk
db_engines = {
79
    'MySQL': ('MySQLdb', {'password': 'passwd', 'database': 'db'}),
80
    'PostgreSQL': ('psycopg2', {}),
81
}
82
83
DatabaseErrors_set = set([DbException])
84
DatabaseErrors = tuple(DatabaseErrors_set)
85
86
def _add_module(module):
87
    DatabaseErrors_set.add(module.DatabaseError)
88
    global DatabaseErrors
89
    DatabaseErrors = tuple(DatabaseErrors_set)
90
91
def db_config_str(db_config):
92
    return db_config['engine']+' database '+db_config['database']
93
94 1909 aaronmk
def _query_lookup(query, params): return (query, dicts.make_hashable(params))
95 1894 aaronmk
96 1901 aaronmk
log_debug_none = lambda msg: None
97
98 1849 aaronmk
class DbConn:
99 1901 aaronmk
    def __init__(self, db_config, serializable=True, log_debug=log_debug_none):
100 1869 aaronmk
        self.db_config = db_config
101
        self.serializable = serializable
102 1901 aaronmk
        self.log_debug = log_debug
103 1869 aaronmk
104
        self.__db = None
105 1889 aaronmk
        self.query_results = {}
106 1869 aaronmk
107
    def __getattr__(self, name):
108
        if name == '__dict__': raise Exception('getting __dict__')
109
        if name == 'db': return self._db()
110
        else: raise AttributeError()
111
112
    def __getstate__(self):
113
        state = copy.copy(self.__dict__) # shallow copy
114 1915 aaronmk
        state['log_debug'] = None # don't pickle the debug callback
115 1869 aaronmk
        state['_DbConn__db'] = None # don't pickle the connection
116
        return state
117
118
    def _db(self):
119
        if self.__db == None:
120
            # Process db_config
121
            db_config = self.db_config.copy() # don't modify input!
122
            module_name, mappings = db_engines[db_config.pop('engine')]
123
            module = __import__(module_name)
124
            _add_module(module)
125
            for orig, new in mappings.iteritems():
126
                try: util.rename_key(db_config, orig, new)
127
                except KeyError: pass
128
129
            # Connect
130
            self.__db = module.connect(**db_config)
131
132
            # Configure connection
133
            if self.serializable: run_raw_query(self,
134
                'SET TRANSACTION ISOLATION LEVEL SERIALIZABLE')
135
136
        return self.__db
137 1889 aaronmk
138 1891 aaronmk
    class DbCursor(Proxy):
139 1927 aaronmk
        def __init__(self, outer):
140 1891 aaronmk
            Proxy.__init__(self, outer.db.cursor())
141 1927 aaronmk
            self.query_results = outer.query_results
142 1894 aaronmk
            self.query_lookup = None
143 1891 aaronmk
            self.result = []
144 1889 aaronmk
145 1894 aaronmk
        def execute(self, query, params=None):
146 1930 aaronmk
            self._is_insert = query.upper().find('INSERT') >= 0
147 1894 aaronmk
            self.query_lookup = _query_lookup(query, params)
148 1904 aaronmk
            try: return_value = self.inner.execute(query, params)
149
            except Exception, e:
150
                self.result = e # cache the exception as the result
151
                self._cache_result()
152
                raise
153
            finally: self.query = get_cur_query(self.inner)
154 1930 aaronmk
            # Fetch all rows so result will be cached
155
            if self.rowcount == 0 and not self._is_insert: consume_rows(self)
156 1894 aaronmk
            return return_value
157
158 1891 aaronmk
        def fetchone(self):
159
            row = self.inner.fetchone()
160 1899 aaronmk
            if row != None: self.result.append(row)
161
            # otherwise, fetched all rows
162 1904 aaronmk
            else: self._cache_result()
163
            return row
164
165
        def _cache_result(self):
166 1906 aaronmk
            # For inserts, only cache exceptions since inserts are not
167
            # idempotent, but an invalid insert will always be invalid
168 1930 aaronmk
            if self.query_results != None and (not self._is_insert
169 1906 aaronmk
                or isinstance(self.result, Exception)):
170
171 1894 aaronmk
                assert self.query_lookup != None
172 1916 aaronmk
                self.query_results[self.query_lookup] = self.CacheCursor(
173
                    util.dict_subset(dicts.AttrsDictView(self),
174
                    ['query', 'result', 'rowcount', 'description']))
175 1906 aaronmk
176 1916 aaronmk
        class CacheCursor:
177
            def __init__(self, cached_result): self.__dict__ = cached_result
178
179 1927 aaronmk
            def execute(self, *args, **kw_args):
180 1916 aaronmk
                if isinstance(self.result, Exception): raise self.result
181
                # otherwise, result is a rows list
182
                self.iter = iter(self.result)
183
184
            def fetchone(self):
185
                try: return self.iter.next()
186
                except StopIteration: return None
187 1891 aaronmk
188 1894 aaronmk
    def run_query(self, query, params=None, cacheable=False):
189 1903 aaronmk
        used_cache = False
190
        try:
191 1927 aaronmk
            # Get cursor
192
            if cacheable:
193
                query_lookup = _query_lookup(query, params)
194
                try:
195
                    cur = self.query_results[query_lookup]
196
                    used_cache = True
197
                except KeyError: cur = self.DbCursor(self)
198
            else: cur = self.db.cursor()
199
200
            # Run query
201
            try: cur.execute(query, params)
202
            except Exception, e:
203
                _add_cursor_info(e, cur)
204
                raise
205 1903 aaronmk
        finally:
206
            if self.log_debug != log_debug_none: # only compute msg if needed
207
                if used_cache: cache_status = 'Cache hit'
208
                elif cacheable: cache_status = 'Cache miss'
209
                else: cache_status = 'Non-cacheable'
210 1927 aaronmk
                self.log_debug(cache_status+': '
211
                    +strings.one_line(get_cur_query(cur)))
212 1903 aaronmk
213
        return cur
214 1914 aaronmk
215
    def is_cached(self, query, params=None):
216
        return _query_lookup(query, params) in self.query_results
217 1849 aaronmk
218 1869 aaronmk
connect = DbConn
219
220 1919 aaronmk
##### Input validation
221
222
def check_name(name):
223
    if re.search(r'\W', name) != None: raise NameException('Name "'+name
224
        +'" may contain only alphanumeric characters and _')
225
226
def esc_name_by_module(module, name, preserve_case=False):
227
    if module == 'psycopg2':
228
        if preserve_case: quote = '"'
229
        # Don't enclose in quotes because this disables case-insensitivity
230
        else: return name
231
    elif module == 'MySQLdb': quote = '`'
232
    else: raise NotImplementedError("Can't escape name for "+module+' database')
233
    return quote + name.replace(quote, '') + quote
234
235
def esc_name_by_engine(engine, name, **kw_args):
236
    return esc_name_by_module(db_engines[engine][0], name, **kw_args)
237
238
def esc_name(db, name, **kw_args):
239
    return esc_name_by_module(util.root_module(db.db), name, **kw_args)
240
241 1968 aaronmk
def qual_name(db, schema, table):
242
    def esc_name(name): return sql.esc_name(db, name, preserve_case=True)
243
    return esc_name(schema)+'.'+esc_name(table)
244
245 832 aaronmk
##### Querying
246
247 1894 aaronmk
def run_raw_query(db, *args, **kw_args):
248
    '''For args, see DbConn.run_query()'''
249
    return db.run_query(*args, **kw_args)
250 11 aaronmk
251 832 aaronmk
##### Recoverable querying
252 15 aaronmk
253 11 aaronmk
def with_savepoint(db, func):
254 1872 aaronmk
    savepoint = 'savepoint_'+str(rand.rand_int()) # must be unique
255 830 aaronmk
    run_raw_query(db, 'SAVEPOINT '+savepoint)
256 11 aaronmk
    try: return_val = func()
257
    except:
258 830 aaronmk
        run_raw_query(db, 'ROLLBACK TO SAVEPOINT '+savepoint)
259 11 aaronmk
        raise
260
    else:
261 830 aaronmk
        run_raw_query(db, 'RELEASE SAVEPOINT '+savepoint)
262 11 aaronmk
        return return_val
263
264 1894 aaronmk
def run_query(db, query, params=None, recover=None, cacheable=False):
265 830 aaronmk
    if recover == None: recover = False
266
267 1894 aaronmk
    def run(): return run_raw_query(db, query, params, cacheable)
268 1914 aaronmk
    if recover and not db.is_cached(query, params):
269
        return with_savepoint(db, run)
270
    else: return run() # don't need savepoint if cached
271 830 aaronmk
272 832 aaronmk
##### Basic queries
273
274 1135 aaronmk
def select(db, table, fields=None, conds=None, limit=None, start=None,
275 1894 aaronmk
    recover=None, cacheable=True):
276 1135 aaronmk
    '''@param fields Use None to select all fields in the table'''
277
    if conds == None: conds = {}
278 135 aaronmk
    assert limit == None or type(limit) == int
279 865 aaronmk
    assert start == None or type(start) == int
280 15 aaronmk
    check_name(table)
281 1135 aaronmk
    if fields != None: map(check_name, fields)
282 15 aaronmk
    map(check_name, conds.keys())
283 865 aaronmk
284 11 aaronmk
    def cond(entry):
285 13 aaronmk
        col, value = entry
286 644 aaronmk
        cond_ = esc_name(db, col)+' '
287 11 aaronmk
        if value == None: cond_ += 'IS'
288
        else: cond_ += '='
289
        cond_ += ' %s'
290
        return cond_
291 1135 aaronmk
    query = 'SELECT '
292
    if fields == None: query += '*'
293
    else: query += ', '.join([esc_name(db, field) for field in fields])
294
    query += ' FROM '+esc_name(db, table)
295 865 aaronmk
296
    missing = True
297 89 aaronmk
    if conds != {}:
298
        query += ' WHERE '+' AND '.join(map(cond, conds.iteritems()))
299 865 aaronmk
        missing = False
300
    if limit != None: query += ' LIMIT '+str(limit); missing = False
301
    if start != None:
302
        if start != 0: query += ' OFFSET '+str(start)
303
        missing = False
304
    if missing: warnings.warn(DbWarning(
305
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
306
307 1905 aaronmk
    return run_query(db, query, conds.values(), recover, cacheable)
308 11 aaronmk
309 1961 aaronmk
default = object() # tells insert() to use the default value for a column
310
311 1960 aaronmk
def insert(db, table, row, returning=None, recover=None, cacheable=True,
312
    table_is_esc=False):
313
    '''
314
    @param returning str|None An inserted column (such as pkey) to return
315
    @param table_is_esc Whether the table name has already been escaped
316
    '''
317
    if not table_is_esc: check_name(table)
318
    if lists.is_seq(row): cols = None
319
    else:
320
        cols = row.keys()
321
        row = row.values()
322
        map(check_name, cols)
323
    row = list(row) # ensure that "!= []" works
324
325 1961 aaronmk
    # Check for special values
326
    labels = []
327
    values = []
328
    for value in row:
329
        if value == default: labels.append('DEFAULT')
330
        else:
331
            labels.append('%s')
332
            values.append(value)
333
334
    # Build query
335 89 aaronmk
    query = 'INSERT INTO '+table
336 1961 aaronmk
    if values != []:
337 1960 aaronmk
        if cols != None: query += ' ('+', '.join(cols)+')'
338 1961 aaronmk
        query += ' VALUES ('+(', '.join(labels))+')'
339 89 aaronmk
    else: query += ' DEFAULT VALUES'
340 1554 aaronmk
341
    if returning != None:
342
        check_name(returning)
343
        query += ' RETURNING '+returning
344
345 1961 aaronmk
    return run_query(db, query, values, recover, cacheable)
346 11 aaronmk
347 135 aaronmk
def last_insert_id(db):
348 1849 aaronmk
    module = util.root_module(db.db)
349 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
350
    elif module == 'MySQLdb': return db.insert_id()
351
    else: return None
352 13 aaronmk
353 1968 aaronmk
def truncate(db, table, schema='public'):
354
    return run_query(db, 'TRUNCATE '+qual_name(db, schema, table)+' CASCADE')
355 832 aaronmk
356
##### Database structure queries
357
358 1850 aaronmk
def pkey(db, table, recover=None):
359 832 aaronmk
    '''Assumed to be first column in table'''
360
    check_name(table)
361 1929 aaronmk
    return col_names(select(db, table, limit=0, recover=recover)).next()
362 832 aaronmk
363 853 aaronmk
def index_cols(db, table, index):
364
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
365
    automatically created. When you don't know whether something is a UNIQUE
366
    constraint or a UNIQUE index, use this function.'''
367
    check_name(table)
368
    check_name(index)
369 1909 aaronmk
    module = util.root_module(db.db)
370
    if module == 'psycopg2':
371
        return list(values(run_query(db, '''\
372 853 aaronmk
SELECT attname
373 866 aaronmk
FROM
374
(
375
        SELECT attnum, attname
376
        FROM pg_index
377
        JOIN pg_class index ON index.oid = indexrelid
378
        JOIN pg_class table_ ON table_.oid = indrelid
379
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
380
        WHERE
381
            table_.relname = %(table)s
382
            AND index.relname = %(index)s
383
    UNION
384
        SELECT attnum, attname
385
        FROM
386
        (
387
            SELECT
388
                indrelid
389
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
390
                    AS indkey
391
            FROM pg_index
392
            JOIN pg_class index ON index.oid = indexrelid
393
            JOIN pg_class table_ ON table_.oid = indrelid
394
            WHERE
395
                table_.relname = %(table)s
396
                AND index.relname = %(index)s
397
        ) s
398
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
399
) s
400 853 aaronmk
ORDER BY attnum
401
''',
402 1909 aaronmk
            {'table': table, 'index': index}, cacheable=True)))
403
    else: raise NotImplementedError("Can't list index columns for "+module+
404
        ' database')
405 853 aaronmk
406 464 aaronmk
def constraint_cols(db, table, constraint):
407
    check_name(table)
408
    check_name(constraint)
409 1849 aaronmk
    module = util.root_module(db.db)
410 464 aaronmk
    if module == 'psycopg2':
411
        return list(values(run_query(db, '''\
412
SELECT attname
413
FROM pg_constraint
414
JOIN pg_class ON pg_class.oid = conrelid
415
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
416
WHERE
417
    relname = %(table)s
418
    AND conname = %(constraint)s
419
ORDER BY attnum
420
''',
421
            {'table': table, 'constraint': constraint})))
422
    else: raise NotImplementedError("Can't list constraint columns for "+module+
423
        ' database')
424
425 1968 aaronmk
def tables(db, schema='public', table_like='%'):
426 1849 aaronmk
    module = util.root_module(db.db)
427 1968 aaronmk
    params = {'schema': schema, 'table_like': table_like}
428 832 aaronmk
    if module == 'psycopg2':
429 1968 aaronmk
        return values(run_query(db, '''\
430
SELECT tablename
431
FROM pg_tables
432
WHERE
433
    schemaname = %(schema)s
434
    AND tablename LIKE %(table_like)s
435
ORDER BY tablename
436
''',
437
            params, cacheable=True))
438
    elif module == 'MySQLdb':
439
        return values(run_query(db, 'SHOW TABLES LIKE %(table_like)s', params,
440
            cacheable=True))
441 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
442 830 aaronmk
443 833 aaronmk
##### Database management
444
445 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
446
    '''For kw_args, see tables()'''
447
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
448 833 aaronmk
449 832 aaronmk
##### Heuristic queries
450
451 1554 aaronmk
def try_insert(db, table, row, returning=None):
452 830 aaronmk
    '''Recovers from errors'''
453 1554 aaronmk
    try: return insert(db, table, row, returning, recover=True)
454 46 aaronmk
    except Exception, e:
455
        msg = str(e)
456 465 aaronmk
        match = re.search(r'duplicate key value violates unique constraint '
457
            r'"(([^\W_]+)_[^"]+)"', msg)
458
        if match:
459
            constraint, table = match.groups()
460 854 aaronmk
            try: cols = index_cols(db, table, constraint)
461 465 aaronmk
            except NotImplementedError: raise e
462 851 aaronmk
            else: raise DuplicateKeyException(cols, e)
463 13 aaronmk
        match = re.search(r'null value in column "(\w+)" violates not-null '
464
            'constraint', msg)
465 470 aaronmk
        if match: raise NullValueException([match.group(1)], e)
466 13 aaronmk
        raise # no specific exception raised
467 11 aaronmk
468 471 aaronmk
def put(db, table, row, pkey, row_ct_ref=None):
469 1554 aaronmk
    '''Recovers from errors.
470
    Only works under PostgreSQL (uses `INSERT ... RETURNING`)'''
471 471 aaronmk
    try:
472 1554 aaronmk
        cur = try_insert(db, table, row, pkey)
473
        if row_ct_ref != None and cur.rowcount >= 0:
474
            row_ct_ref[0] += cur.rowcount
475
        return value(cur)
476 471 aaronmk
    except DuplicateKeyException, e:
477 1069 aaronmk
        return value(select(db, table, [pkey],
478
            util.dict_subset_right_join(row, e.cols), recover=True))
479 471 aaronmk
480 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
481 830 aaronmk
    '''Recovers from errors'''
482
    try: return value(select(db, table, [pkey], row, 1, recover=True))
483 14 aaronmk
    except StopIteration:
484 40 aaronmk
        if not create: raise
485 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row