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
241 832 aaronmk
##### Querying
242
243 1894 aaronmk
def run_raw_query(db, *args, **kw_args):
244
    '''For args, see DbConn.run_query()'''
245
    return db.run_query(*args, **kw_args)
246 11 aaronmk
247 832 aaronmk
##### Recoverable querying
248 15 aaronmk
249 11 aaronmk
def with_savepoint(db, func):
250 1872 aaronmk
    savepoint = 'savepoint_'+str(rand.rand_int()) # must be unique
251 830 aaronmk
    run_raw_query(db, 'SAVEPOINT '+savepoint)
252 11 aaronmk
    try: return_val = func()
253
    except:
254 830 aaronmk
        run_raw_query(db, 'ROLLBACK TO SAVEPOINT '+savepoint)
255 11 aaronmk
        raise
256
    else:
257 830 aaronmk
        run_raw_query(db, 'RELEASE SAVEPOINT '+savepoint)
258 11 aaronmk
        return return_val
259
260 1894 aaronmk
def run_query(db, query, params=None, recover=None, cacheable=False):
261 830 aaronmk
    if recover == None: recover = False
262
263 1894 aaronmk
    def run(): return run_raw_query(db, query, params, cacheable)
264 1914 aaronmk
    if recover and not db.is_cached(query, params):
265
        return with_savepoint(db, run)
266
    else: return run() # don't need savepoint if cached
267 830 aaronmk
268 832 aaronmk
##### Basic queries
269
270 1135 aaronmk
def select(db, table, fields=None, conds=None, limit=None, start=None,
271 1894 aaronmk
    recover=None, cacheable=True):
272 1135 aaronmk
    '''@param fields Use None to select all fields in the table'''
273
    if conds == None: conds = {}
274 135 aaronmk
    assert limit == None or type(limit) == int
275 865 aaronmk
    assert start == None or type(start) == int
276 15 aaronmk
    check_name(table)
277 1135 aaronmk
    if fields != None: map(check_name, fields)
278 15 aaronmk
    map(check_name, conds.keys())
279 865 aaronmk
280 11 aaronmk
    def cond(entry):
281 13 aaronmk
        col, value = entry
282 644 aaronmk
        cond_ = esc_name(db, col)+' '
283 11 aaronmk
        if value == None: cond_ += 'IS'
284
        else: cond_ += '='
285
        cond_ += ' %s'
286
        return cond_
287 1135 aaronmk
    query = 'SELECT '
288
    if fields == None: query += '*'
289
    else: query += ', '.join([esc_name(db, field) for field in fields])
290
    query += ' FROM '+esc_name(db, table)
291 865 aaronmk
292
    missing = True
293 89 aaronmk
    if conds != {}:
294
        query += ' WHERE '+' AND '.join(map(cond, conds.iteritems()))
295 865 aaronmk
        missing = False
296
    if limit != None: query += ' LIMIT '+str(limit); missing = False
297
    if start != None:
298
        if start != 0: query += ' OFFSET '+str(start)
299
        missing = False
300
    if missing: warnings.warn(DbWarning(
301
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
302
303 1905 aaronmk
    return run_query(db, query, conds.values(), recover, cacheable)
304 11 aaronmk
305 1960 aaronmk
def insert(db, table, row, returning=None, recover=None, cacheable=True,
306
    table_is_esc=False):
307
    '''
308
    @param returning str|None An inserted column (such as pkey) to return
309
    @param table_is_esc Whether the table name has already been escaped
310
    '''
311
    if not table_is_esc: check_name(table)
312
    if lists.is_seq(row): cols = None
313
    else:
314
        cols = row.keys()
315
        row = row.values()
316
        map(check_name, cols)
317
    row = list(row) # ensure that "!= []" works
318
319 89 aaronmk
    query = 'INSERT INTO '+table
320 1960 aaronmk
    if row != []:
321
        if cols != None: query += ' ('+', '.join(cols)+')'
322
        query += ' VALUES ('+(', '.join(['%s']*len(row)))+')'
323 89 aaronmk
    else: query += ' DEFAULT VALUES'
324 1554 aaronmk
325
    if returning != None:
326
        check_name(returning)
327
        query += ' RETURNING '+returning
328
329 1960 aaronmk
    return run_query(db, query, row, recover, cacheable)
330 11 aaronmk
331 135 aaronmk
def last_insert_id(db):
332 1849 aaronmk
    module = util.root_module(db.db)
333 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
334
    elif module == 'MySQLdb': return db.insert_id()
335
    else: return None
336 13 aaronmk
337 832 aaronmk
def truncate(db, table):
338
    check_name(table)
339 869 aaronmk
    return run_raw_query(db, 'TRUNCATE '+table+' CASCADE')
340 832 aaronmk
341
##### Database structure queries
342
343 1850 aaronmk
def pkey(db, table, recover=None):
344 832 aaronmk
    '''Assumed to be first column in table'''
345
    check_name(table)
346 1929 aaronmk
    return col_names(select(db, table, limit=0, recover=recover)).next()
347 832 aaronmk
348 853 aaronmk
def index_cols(db, table, index):
349
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
350
    automatically created. When you don't know whether something is a UNIQUE
351
    constraint or a UNIQUE index, use this function.'''
352
    check_name(table)
353
    check_name(index)
354 1909 aaronmk
    module = util.root_module(db.db)
355
    if module == 'psycopg2':
356
        return list(values(run_query(db, '''\
357 853 aaronmk
SELECT attname
358 866 aaronmk
FROM
359
(
360
        SELECT attnum, attname
361
        FROM pg_index
362
        JOIN pg_class index ON index.oid = indexrelid
363
        JOIN pg_class table_ ON table_.oid = indrelid
364
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
365
        WHERE
366
            table_.relname = %(table)s
367
            AND index.relname = %(index)s
368
    UNION
369
        SELECT attnum, attname
370
        FROM
371
        (
372
            SELECT
373
                indrelid
374
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
375
                    AS indkey
376
            FROM pg_index
377
            JOIN pg_class index ON index.oid = indexrelid
378
            JOIN pg_class table_ ON table_.oid = indrelid
379
            WHERE
380
                table_.relname = %(table)s
381
                AND index.relname = %(index)s
382
        ) s
383
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
384
) s
385 853 aaronmk
ORDER BY attnum
386
''',
387 1909 aaronmk
            {'table': table, 'index': index}, cacheable=True)))
388
    else: raise NotImplementedError("Can't list index columns for "+module+
389
        ' database')
390 853 aaronmk
391 464 aaronmk
def constraint_cols(db, table, constraint):
392
    check_name(table)
393
    check_name(constraint)
394 1849 aaronmk
    module = util.root_module(db.db)
395 464 aaronmk
    if module == 'psycopg2':
396
        return list(values(run_query(db, '''\
397
SELECT attname
398
FROM pg_constraint
399
JOIN pg_class ON pg_class.oid = conrelid
400
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
401
WHERE
402
    relname = %(table)s
403
    AND conname = %(constraint)s
404
ORDER BY attnum
405
''',
406
            {'table': table, 'constraint': constraint})))
407
    else: raise NotImplementedError("Can't list constraint columns for "+module+
408
        ' database')
409
410 832 aaronmk
def tables(db):
411 1849 aaronmk
    module = util.root_module(db.db)
412 832 aaronmk
    if module == 'psycopg2':
413
        return values(run_query(db, "SELECT tablename from pg_tables "
414
            "WHERE schemaname = 'public' ORDER BY tablename"))
415
    elif module == 'MySQLdb': return values(run_query(db, 'SHOW TABLES'))
416
    else: raise NotImplementedError("Can't list tables for "+module+' database')
417 830 aaronmk
418 833 aaronmk
##### Database management
419
420
def empty_db(db):
421
    for table in tables(db): truncate(db, table)
422
423 832 aaronmk
##### Heuristic queries
424
425 1554 aaronmk
def try_insert(db, table, row, returning=None):
426 830 aaronmk
    '''Recovers from errors'''
427 1554 aaronmk
    try: return insert(db, table, row, returning, recover=True)
428 46 aaronmk
    except Exception, e:
429
        msg = str(e)
430 465 aaronmk
        match = re.search(r'duplicate key value violates unique constraint '
431
            r'"(([^\W_]+)_[^"]+)"', msg)
432
        if match:
433
            constraint, table = match.groups()
434 854 aaronmk
            try: cols = index_cols(db, table, constraint)
435 465 aaronmk
            except NotImplementedError: raise e
436 851 aaronmk
            else: raise DuplicateKeyException(cols, e)
437 13 aaronmk
        match = re.search(r'null value in column "(\w+)" violates not-null '
438
            'constraint', msg)
439 470 aaronmk
        if match: raise NullValueException([match.group(1)], e)
440 13 aaronmk
        raise # no specific exception raised
441 11 aaronmk
442 471 aaronmk
def put(db, table, row, pkey, row_ct_ref=None):
443 1554 aaronmk
    '''Recovers from errors.
444
    Only works under PostgreSQL (uses `INSERT ... RETURNING`)'''
445 471 aaronmk
    try:
446 1554 aaronmk
        cur = try_insert(db, table, row, pkey)
447
        if row_ct_ref != None and cur.rowcount >= 0:
448
            row_ct_ref[0] += cur.rowcount
449
        return value(cur)
450 471 aaronmk
    except DuplicateKeyException, e:
451 1069 aaronmk
        return value(select(db, table, [pkey],
452
            util.dict_subset_right_join(row, e.cols), recover=True))
453 471 aaronmk
454 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
455 830 aaronmk
    '''Recovers from errors'''
456
    try: return value(select(db, table, [pkey], row, 1, recover=True))
457 14 aaronmk
    except StopIteration:
458 40 aaronmk
        if not create: raise
459 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row