Project

General

Profile

1 11 aaronmk
# Database access
2
3
import random
4
import re
5
import sys
6 865 aaronmk
import warnings
7 11 aaronmk
8 300 aaronmk
import exc
9 862 aaronmk
import strings
10 131 aaronmk
import util
11 11 aaronmk
12 832 aaronmk
##### Exceptions
13
14 135 aaronmk
def get_cur_query(cur):
15
    if hasattr(cur, 'query'): return cur.query
16
    elif hasattr(cur, '_last_executed'): return cur._last_executed
17
    else: return None
18 14 aaronmk
19 300 aaronmk
def _add_cursor_info(e, cur): exc.add_msg(e, 'query: '+get_cur_query(cur))
20 135 aaronmk
21 300 aaronmk
class DbException(exc.ExceptionWithCause):
22 14 aaronmk
    def __init__(self, msg, cause=None, cur=None):
23 300 aaronmk
        exc.ExceptionWithCause.__init__(self, msg, cause)
24 14 aaronmk
        if cur != None: _add_cursor_info(self, cur)
25
26 360 aaronmk
class NameException(DbException): pass
27
28 468 aaronmk
class ExceptionWithColumns(DbException):
29
    def __init__(self, cols, cause=None):
30
        DbException.__init__(self, 'columns: ' + ', '.join(cols), cause)
31
        self.cols = cols
32 11 aaronmk
33 468 aaronmk
class DuplicateKeyException(ExceptionWithColumns): pass
34 13 aaronmk
35 468 aaronmk
class NullValueException(ExceptionWithColumns): pass
36 13 aaronmk
37 89 aaronmk
class EmptyRowException(DbException): pass
38
39 865 aaronmk
##### Warnings
40
41
class DbWarning(UserWarning): pass
42
43 832 aaronmk
##### Input validation
44
45 11 aaronmk
def check_name(name):
46
    if re.search(r'\W', name) != None: raise NameException('Name "'+name
47
        +'" may contain only alphanumeric characters and _')
48
49 643 aaronmk
def esc_name(db, name):
50 1849 aaronmk
    module = util.root_module(db.db)
51 645 aaronmk
    if module == 'psycopg2': return name
52
        # Don't enclose in quotes because this disables case-insensitivity
53 643 aaronmk
    elif module == 'MySQLdb': quote = '`'
54 645 aaronmk
    else: raise NotImplementedError("Can't escape name for "+module+' database')
55 643 aaronmk
    return quote + name.replace(quote, '') + quote
56
57 1849 aaronmk
##### Connection object
58
59
class DbConn:
60
    def __init__(self, db):
61
        self.db = db
62
        self.pkeys = {}
63
        self.index_cols = {}
64
65 832 aaronmk
##### Querying
66
67 830 aaronmk
def run_raw_query(db, query, params=None):
68 1849 aaronmk
    cur = db.db.cursor()
69 11 aaronmk
    try: cur.execute(query, params)
70 46 aaronmk
    except Exception, e:
71
        _add_cursor_info(e, cur)
72 11 aaronmk
        raise
73 867 aaronmk
    if run_raw_query.debug:
74
        sys.stderr.write(strings.one_line(get_cur_query(cur))+'\n')
75 11 aaronmk
    return cur
76
77 832 aaronmk
##### Recoverable querying
78 15 aaronmk
79 11 aaronmk
def with_savepoint(db, func):
80
    savepoint = 'savepoint_'+str(random.randint(0, sys.maxint)) # must be unique
81 830 aaronmk
    run_raw_query(db, 'SAVEPOINT '+savepoint)
82 11 aaronmk
    try: return_val = func()
83
    except:
84 830 aaronmk
        run_raw_query(db, 'ROLLBACK TO SAVEPOINT '+savepoint)
85 11 aaronmk
        raise
86
    else:
87 830 aaronmk
        run_raw_query(db, 'RELEASE SAVEPOINT '+savepoint)
88 11 aaronmk
        return return_val
89
90 830 aaronmk
def run_query(db, query, params=None, recover=None):
91
    if recover == None: recover = False
92
93
    def run(): return run_raw_query(db, query, params)
94
    if recover: return with_savepoint(db, run)
95
    else: return run()
96
97 832 aaronmk
##### Result retrieval
98
99 1135 aaronmk
def col_names(cur): return (col[0] for col in cur.description)
100 832 aaronmk
101
def rows(cur): return iter(lambda: cur.fetchone(), None)
102
103
def row(cur): return rows(cur).next()
104
105
def value(cur): return row(cur)[0]
106
107
def values(cur): return iter(lambda: value(cur), None)
108
109
def value_or_none(cur):
110
    try: return value(cur)
111
    except StopIteration: return None
112
113
##### Basic queries
114
115 1135 aaronmk
def select(db, table, fields=None, conds=None, limit=None, start=None,
116
    recover=None):
117
    '''@param fields Use None to select all fields in the table'''
118
    if conds == None: conds = {}
119 135 aaronmk
    assert limit == None or type(limit) == int
120 865 aaronmk
    assert start == None or type(start) == int
121 15 aaronmk
    check_name(table)
122 1135 aaronmk
    if fields != None: map(check_name, fields)
123 15 aaronmk
    map(check_name, conds.keys())
124 865 aaronmk
125 11 aaronmk
    def cond(entry):
126 13 aaronmk
        col, value = entry
127 644 aaronmk
        cond_ = esc_name(db, col)+' '
128 11 aaronmk
        if value == None: cond_ += 'IS'
129
        else: cond_ += '='
130
        cond_ += ' %s'
131
        return cond_
132 1135 aaronmk
    query = 'SELECT '
133
    if fields == None: query += '*'
134
    else: query += ', '.join([esc_name(db, field) for field in fields])
135
    query += ' FROM '+esc_name(db, table)
136 865 aaronmk
137
    missing = True
138 89 aaronmk
    if conds != {}:
139
        query += ' WHERE '+' AND '.join(map(cond, conds.iteritems()))
140 865 aaronmk
        missing = False
141
    if limit != None: query += ' LIMIT '+str(limit); missing = False
142
    if start != None:
143
        if start != 0: query += ' OFFSET '+str(start)
144
        missing = False
145
    if missing: warnings.warn(DbWarning(
146
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
147
148 830 aaronmk
    return run_query(db, query, conds.values(), recover)
149 11 aaronmk
150 1554 aaronmk
def insert(db, table, row, returning=None, recover=None):
151
    '''@param returning str|None An inserted column (such as pkey) to return'''
152 11 aaronmk
    check_name(table)
153 13 aaronmk
    cols = row.keys()
154 15 aaronmk
    map(check_name, cols)
155 89 aaronmk
    query = 'INSERT INTO '+table
156 1554 aaronmk
157 89 aaronmk
    if row != {}: query += ' ('+', '.join(cols)+') VALUES ('\
158
        +', '.join(['%s']*len(cols))+')'
159
    else: query += ' DEFAULT VALUES'
160 1554 aaronmk
161
    if returning != None:
162
        check_name(returning)
163
        query += ' RETURNING '+returning
164
165 830 aaronmk
    return run_query(db, query, row.values(), recover)
166 11 aaronmk
167 135 aaronmk
def last_insert_id(db):
168 1849 aaronmk
    module = util.root_module(db.db)
169 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
170
    elif module == 'MySQLdb': return db.insert_id()
171
    else: return None
172 13 aaronmk
173 832 aaronmk
def truncate(db, table):
174
    check_name(table)
175 869 aaronmk
    return run_raw_query(db, 'TRUNCATE '+table+' CASCADE')
176 832 aaronmk
177
##### Database structure queries
178
179 1850 aaronmk
def pkey(db, table, recover=None):
180 832 aaronmk
    '''Assumed to be first column in table'''
181
    check_name(table)
182 1850 aaronmk
    if table not in db.pkeys:
183
        db.pkeys[table] = col_names(run_query(db,
184 1135 aaronmk
            'SELECT * FROM '+table+' LIMIT 0', recover=recover)).next()
185 1850 aaronmk
    return db.pkeys[table]
186 832 aaronmk
187 853 aaronmk
def index_cols(db, table, index):
188
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
189
    automatically created. When you don't know whether something is a UNIQUE
190
    constraint or a UNIQUE index, use this function.'''
191
    check_name(table)
192
    check_name(index)
193 1849 aaronmk
    module = util.root_module(db.db)
194 853 aaronmk
    if module == 'psycopg2':
195
        return list(values(run_query(db, '''\
196
SELECT attname
197 866 aaronmk
FROM
198
(
199
        SELECT attnum, attname
200
        FROM pg_index
201
        JOIN pg_class index ON index.oid = indexrelid
202
        JOIN pg_class table_ ON table_.oid = indrelid
203
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
204
        WHERE
205
            table_.relname = %(table)s
206
            AND index.relname = %(index)s
207
    UNION
208
        SELECT attnum, attname
209
        FROM
210
        (
211
            SELECT
212
                indrelid
213
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
214
                    AS indkey
215
            FROM pg_index
216
            JOIN pg_class index ON index.oid = indexrelid
217
            JOIN pg_class table_ ON table_.oid = indrelid
218
            WHERE
219
                table_.relname = %(table)s
220
                AND index.relname = %(index)s
221
        ) s
222
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
223
) s
224 853 aaronmk
ORDER BY attnum
225
''',
226
            {'table': table, 'index': index})))
227
    else: raise NotImplementedError("Can't list index columns for "+module+
228
        ' database')
229
230 464 aaronmk
def constraint_cols(db, table, constraint):
231
    check_name(table)
232
    check_name(constraint)
233 1849 aaronmk
    module = util.root_module(db.db)
234 464 aaronmk
    if module == 'psycopg2':
235
        return list(values(run_query(db, '''\
236
SELECT attname
237
FROM pg_constraint
238
JOIN pg_class ON pg_class.oid = conrelid
239
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
240
WHERE
241
    relname = %(table)s
242
    AND conname = %(constraint)s
243
ORDER BY attnum
244
''',
245
            {'table': table, 'constraint': constraint})))
246
    else: raise NotImplementedError("Can't list constraint columns for "+module+
247
        ' database')
248
249 832 aaronmk
def tables(db):
250 1849 aaronmk
    module = util.root_module(db.db)
251 832 aaronmk
    if module == 'psycopg2':
252
        return values(run_query(db, "SELECT tablename from pg_tables "
253
            "WHERE schemaname = 'public' ORDER BY tablename"))
254
    elif module == 'MySQLdb': return values(run_query(db, 'SHOW TABLES'))
255
    else: raise NotImplementedError("Can't list tables for "+module+' database')
256 830 aaronmk
257 833 aaronmk
##### Database management
258
259
def empty_db(db):
260
    for table in tables(db): truncate(db, table)
261
262 832 aaronmk
##### Heuristic queries
263
264 1554 aaronmk
def try_insert(db, table, row, returning=None):
265 830 aaronmk
    '''Recovers from errors'''
266 1554 aaronmk
    try: return insert(db, table, row, returning, recover=True)
267 46 aaronmk
    except Exception, e:
268
        msg = str(e)
269 465 aaronmk
        match = re.search(r'duplicate key value violates unique constraint '
270
            r'"(([^\W_]+)_[^"]+)"', msg)
271
        if match:
272
            constraint, table = match.groups()
273 854 aaronmk
            try: cols = index_cols(db, table, constraint)
274 465 aaronmk
            except NotImplementedError: raise e
275 851 aaronmk
            else: raise DuplicateKeyException(cols, e)
276 13 aaronmk
        match = re.search(r'null value in column "(\w+)" violates not-null '
277
            'constraint', msg)
278 470 aaronmk
        if match: raise NullValueException([match.group(1)], e)
279 13 aaronmk
        raise # no specific exception raised
280 11 aaronmk
281 471 aaronmk
def put(db, table, row, pkey, row_ct_ref=None):
282 1554 aaronmk
    '''Recovers from errors.
283
    Only works under PostgreSQL (uses `INSERT ... RETURNING`)'''
284 471 aaronmk
    try:
285 1554 aaronmk
        cur = try_insert(db, table, row, pkey)
286
        if row_ct_ref != None and cur.rowcount >= 0:
287
            row_ct_ref[0] += cur.rowcount
288
        return value(cur)
289 471 aaronmk
    except DuplicateKeyException, e:
290 1069 aaronmk
        return value(select(db, table, [pkey],
291
            util.dict_subset_right_join(row, e.cols), recover=True))
292 471 aaronmk
293 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
294 830 aaronmk
    '''Recovers from errors'''
295
    try: return value(select(db, table, [pkey], row, 1, recover=True))
296 14 aaronmk
    except StopIteration:
297 40 aaronmk
        if not create: raise
298 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
299 131 aaronmk
300 832 aaronmk
##### Database connections
301
302 131 aaronmk
db_engines = {
303
    'MySQL': ('MySQLdb', {'password': 'passwd', 'database': 'db'}),
304
    'PostgreSQL': ('psycopg2', {}),
305
}
306
307 360 aaronmk
DatabaseErrors_set = set([DbException])
308
DatabaseErrors = tuple(DatabaseErrors_set)
309 342 aaronmk
310
def _add_module(module):
311
    DatabaseErrors_set.add(module.DatabaseError)
312
    global DatabaseErrors
313
    DatabaseErrors = tuple(DatabaseErrors_set)
314
315 646 aaronmk
def connect(db_config, serializable=True):
316 131 aaronmk
    db_config = db_config.copy() # don't modify input!
317 342 aaronmk
    module_name, mappings = db_engines[db_config.pop('engine')]
318
    module = __import__(module_name)
319
    _add_module(module)
320 330 aaronmk
    for orig, new in mappings.iteritems():
321
        try: util.rename_key(db_config, orig, new)
322
        except KeyError: pass
323 1849 aaronmk
    db = DbConn(module.connect(**db_config))
324 646 aaronmk
    if serializable:
325 830 aaronmk
        run_raw_query(db, 'SET TRANSACTION ISOLATION LEVEL SERIALIZABLE')
326 646 aaronmk
    return db
327
328
def db_config_str(db_config):
329
    return db_config['engine']+' database '+db_config['database']