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 2145 aaronmk
        DbException.__init__(self, 'for name: '+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
        DbException.__init__(self,
45 2369 aaronmk
            'for name: '+str(name)+'; value: '+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 2345 aaronmk
        DbException.__init__(self, 'Violated '+name+ ' constraint on columns: '
52 2307 aaronmk
            +(', '.join(cols)), cause)
53 2306 aaronmk
        self.name = name
54 468 aaronmk
        self.cols = cols
55 11 aaronmk
56 2143 aaronmk
class NameException(DbException): pass
57
58 2306 aaronmk
class DuplicateKeyException(ConstraintException): pass
59 13 aaronmk
60 2306 aaronmk
class NullValueException(ConstraintException): pass
61 13 aaronmk
62 2240 aaronmk
class FunctionValueException(ExceptionWithNameValue): pass
63 2239 aaronmk
64 2143 aaronmk
class DuplicateTableException(ExceptionWithName): pass
65
66 2188 aaronmk
class DuplicateFunctionException(ExceptionWithName): pass
67
68 89 aaronmk
class EmptyRowException(DbException): pass
69
70 865 aaronmk
##### Warnings
71
72
class DbWarning(UserWarning): pass
73
74 1930 aaronmk
##### Result retrieval
75
76
def col_names(cur): return (col[0] for col in cur.description)
77
78
def rows(cur): return iter(lambda: cur.fetchone(), None)
79
80
def consume_rows(cur):
81
    '''Used to fetch all rows so result will be cached'''
82
    iters.consume_iter(rows(cur))
83
84
def next_row(cur): return rows(cur).next()
85
86
def row(cur):
87
    row_ = next_row(cur)
88
    consume_rows(cur)
89
    return row_
90
91
def next_value(cur): return next_row(cur)[0]
92
93
def value(cur): return row(cur)[0]
94
95
def values(cur): return iters.func_iter(lambda: next_value(cur))
96
97
def value_or_none(cur):
98
    try: return value(cur)
99
    except StopIteration: return None
100
101 2101 aaronmk
##### Input validation
102
103
def check_name(name):
104
    if re.search(r'\W', name) != None: raise NameException('Name "'+name
105
        +'" may contain only alphanumeric characters and _')
106
107
def esc_name_by_module(module, name, ignore_case=False):
108 2388 aaronmk
    if module == 'psycopg2' or module == None:
109 2101 aaronmk
        if ignore_case:
110
            # Don't enclose in quotes because this disables case-insensitivity
111
            check_name(name)
112
            return name
113
        else: quote = '"'
114
    elif module == 'MySQLdb': quote = '`'
115
    else: raise NotImplementedError("Can't escape name for "+module+' database')
116
    return quote + name.replace(quote, '') + quote
117
118
def esc_name_by_engine(engine, name, **kw_args):
119
    return esc_name_by_module(db_engines[engine][0], name, **kw_args)
120
121
def esc_name(db, name, **kw_args):
122
    return esc_name_by_module(util.root_module(db.db), name, **kw_args)
123
124
def qual_name(db, schema, table):
125
    def esc_name_(name): return esc_name(db, name)
126
    table = esc_name_(table)
127
    if schema != None: return esc_name_(schema)+'.'+table
128
    else: return table
129
130 1869 aaronmk
##### Database connections
131 1849 aaronmk
132 2097 aaronmk
db_config_names = ['engine', 'host', 'user', 'password', 'database', 'schemas']
133 1926 aaronmk
134 1869 aaronmk
db_engines = {
135
    'MySQL': ('MySQLdb', {'password': 'passwd', 'database': 'db'}),
136
    'PostgreSQL': ('psycopg2', {}),
137
}
138
139
DatabaseErrors_set = set([DbException])
140
DatabaseErrors = tuple(DatabaseErrors_set)
141
142
def _add_module(module):
143
    DatabaseErrors_set.add(module.DatabaseError)
144
    global DatabaseErrors
145
    DatabaseErrors = tuple(DatabaseErrors_set)
146
147
def db_config_str(db_config):
148
    return db_config['engine']+' database '+db_config['database']
149
150 1909 aaronmk
def _query_lookup(query, params): return (query, dicts.make_hashable(params))
151 1894 aaronmk
152 1901 aaronmk
log_debug_none = lambda msg: None
153
154 1849 aaronmk
class DbConn:
155 2190 aaronmk
    def __init__(self, db_config, serializable=True, autocommit=False,
156
        caching=True, log_debug=log_debug_none):
157 1869 aaronmk
        self.db_config = db_config
158
        self.serializable = serializable
159 2190 aaronmk
        self.autocommit = autocommit
160
        self.caching = caching
161 1901 aaronmk
        self.log_debug = log_debug
162 2193 aaronmk
        self.debug = log_debug != log_debug_none
163 1869 aaronmk
164
        self.__db = None
165 1889 aaronmk
        self.query_results = {}
166 2139 aaronmk
        self._savepoint = 0
167 1869 aaronmk
168
    def __getattr__(self, name):
169
        if name == '__dict__': raise Exception('getting __dict__')
170
        if name == 'db': return self._db()
171
        else: raise AttributeError()
172
173
    def __getstate__(self):
174
        state = copy.copy(self.__dict__) # shallow copy
175 1915 aaronmk
        state['log_debug'] = None # don't pickle the debug callback
176 1869 aaronmk
        state['_DbConn__db'] = None # don't pickle the connection
177
        return state
178
179 2165 aaronmk
    def connected(self): return self.__db != None
180
181 1869 aaronmk
    def _db(self):
182
        if self.__db == None:
183
            # Process db_config
184
            db_config = self.db_config.copy() # don't modify input!
185 2097 aaronmk
            schemas = db_config.pop('schemas', None)
186 1869 aaronmk
            module_name, mappings = db_engines[db_config.pop('engine')]
187
            module = __import__(module_name)
188
            _add_module(module)
189
            for orig, new in mappings.iteritems():
190
                try: util.rename_key(db_config, orig, new)
191
                except KeyError: pass
192
193
            # Connect
194
            self.__db = module.connect(**db_config)
195
196
            # Configure connection
197 2234 aaronmk
            if self.serializable and not self.autocommit: run_raw_query(self,
198
                'SET TRANSACTION ISOLATION LEVEL SERIALIZABLE')
199 2101 aaronmk
            if schemas != None:
200
                schemas_ = ''.join((esc_name(self, s)+', '
201
                    for s in schemas.split(',')))
202
                run_raw_query(self, "SELECT set_config('search_path', \
203
%s || current_setting('search_path'), false)", [schemas_])
204 1869 aaronmk
205
        return self.__db
206 1889 aaronmk
207 1891 aaronmk
    class DbCursor(Proxy):
208 1927 aaronmk
        def __init__(self, outer):
209 1891 aaronmk
            Proxy.__init__(self, outer.db.cursor())
210 2191 aaronmk
            self.outer = outer
211 1927 aaronmk
            self.query_results = outer.query_results
212 1894 aaronmk
            self.query_lookup = None
213 1891 aaronmk
            self.result = []
214 1889 aaronmk
215 1894 aaronmk
        def execute(self, query, params=None):
216 1930 aaronmk
            self._is_insert = query.upper().find('INSERT') >= 0
217 1894 aaronmk
            self.query_lookup = _query_lookup(query, params)
218 2148 aaronmk
            try:
219 2191 aaronmk
                try:
220
                    return_value = self.inner.execute(query, params)
221
                    self.outer.do_autocommit()
222 2148 aaronmk
                finally: self.query = get_cur_query(self.inner)
223 1904 aaronmk
            except Exception, e:
224 2170 aaronmk
                _add_cursor_info(e, self, query, params)
225 1904 aaronmk
                self.result = e # cache the exception as the result
226
                self._cache_result()
227
                raise
228 1930 aaronmk
            # Fetch all rows so result will be cached
229
            if self.rowcount == 0 and not self._is_insert: consume_rows(self)
230 1894 aaronmk
            return return_value
231
232 1891 aaronmk
        def fetchone(self):
233
            row = self.inner.fetchone()
234 1899 aaronmk
            if row != None: self.result.append(row)
235
            # otherwise, fetched all rows
236 1904 aaronmk
            else: self._cache_result()
237
            return row
238
239
        def _cache_result(self):
240 1906 aaronmk
            # For inserts, only cache exceptions since inserts are not
241
            # idempotent, but an invalid insert will always be invalid
242 1930 aaronmk
            if self.query_results != None and (not self._is_insert
243 1906 aaronmk
                or isinstance(self.result, Exception)):
244
245 1894 aaronmk
                assert self.query_lookup != None
246 1916 aaronmk
                self.query_results[self.query_lookup] = self.CacheCursor(
247
                    util.dict_subset(dicts.AttrsDictView(self),
248
                    ['query', 'result', 'rowcount', 'description']))
249 1906 aaronmk
250 1916 aaronmk
        class CacheCursor:
251
            def __init__(self, cached_result): self.__dict__ = cached_result
252
253 1927 aaronmk
            def execute(self, *args, **kw_args):
254 1916 aaronmk
                if isinstance(self.result, Exception): raise self.result
255
                # otherwise, result is a rows list
256
                self.iter = iter(self.result)
257
258
            def fetchone(self):
259
                try: return self.iter.next()
260
                except StopIteration: return None
261 1891 aaronmk
262 2212 aaronmk
    def esc_value(self, value):
263 2215 aaronmk
        module = util.root_module(self.db)
264 2374 aaronmk
        if module == 'psycopg2': str_ = self.db.cursor().mogrify('%s', [value])
265 2212 aaronmk
        elif module == 'MySQLdb':
266
            import _mysql
267 2374 aaronmk
            str_ = _mysql.escape_string(value)
268 2212 aaronmk
        else: raise NotImplementedError("Can't escape value for "+module
269
            +' database')
270 2374 aaronmk
        return strings.to_unicode(str_)
271 2212 aaronmk
272 2347 aaronmk
    def esc_name(self, name): return esc_name(self, name) # calls global func
273
274 1894 aaronmk
    def run_query(self, query, params=None, cacheable=False):
275 2148 aaronmk
        '''Translates known DB errors to typed exceptions:
276
        See self.DbCursor.execute().'''
277 2167 aaronmk
        assert query != None
278
279 2047 aaronmk
        if not self.caching: cacheable = False
280 1903 aaronmk
        used_cache = False
281
        try:
282 1927 aaronmk
            # Get cursor
283
            if cacheable:
284
                query_lookup = _query_lookup(query, params)
285
                try:
286
                    cur = self.query_results[query_lookup]
287
                    used_cache = True
288
                except KeyError: cur = self.DbCursor(self)
289
            else: cur = self.db.cursor()
290
291
            # Run query
292 2148 aaronmk
            cur.execute(query, params)
293 1903 aaronmk
        finally:
294 2193 aaronmk
            if self.debug: # only compute msg if needed
295 1903 aaronmk
                if used_cache: cache_status = 'Cache hit'
296
                elif cacheable: cache_status = 'Cache miss'
297
                else: cache_status = 'Non-cacheable'
298 1927 aaronmk
                self.log_debug(cache_status+': '
299 2170 aaronmk
                    +strings.one_line(str(get_cur_query(cur, query, params))))
300 1903 aaronmk
301
        return cur
302 1914 aaronmk
303
    def is_cached(self, query, params=None):
304
        return _query_lookup(query, params) in self.query_results
305 2139 aaronmk
306
    def with_savepoint(self, func):
307 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
308 2139 aaronmk
        self.run_query('SAVEPOINT '+savepoint)
309
        self._savepoint += 1
310
        try:
311
            try: return_val = func()
312
            finally:
313
                self._savepoint -= 1
314
                assert self._savepoint >= 0
315
        except:
316
            self.run_query('ROLLBACK TO SAVEPOINT '+savepoint)
317
            raise
318
        else:
319
            self.run_query('RELEASE SAVEPOINT '+savepoint)
320 2191 aaronmk
            self.do_autocommit()
321 2139 aaronmk
            return return_val
322 2191 aaronmk
323
    def do_autocommit(self):
324
        '''Autocommits if outside savepoint'''
325
        assert self._savepoint >= 0
326
        if self.autocommit and self._savepoint == 0:
327
            self.log_debug('Autocommiting')
328
            self.db.commit()
329 1849 aaronmk
330 1869 aaronmk
connect = DbConn
331
332 832 aaronmk
##### Querying
333
334 1894 aaronmk
def run_raw_query(db, *args, **kw_args):
335 2085 aaronmk
    '''For params, see DbConn.run_query()'''
336 1894 aaronmk
    return db.run_query(*args, **kw_args)
337 11 aaronmk
338 2068 aaronmk
def mogrify(db, query, params):
339
    module = util.root_module(db.db)
340
    if module == 'psycopg2': return db.db.cursor().mogrify(query, params)
341
    else: raise NotImplementedError("Can't mogrify query for "+module+
342
        ' database')
343
344 832 aaronmk
##### Recoverable querying
345 15 aaronmk
346 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
347 11 aaronmk
348 1894 aaronmk
def run_query(db, query, params=None, recover=None, cacheable=False):
349 830 aaronmk
    if recover == None: recover = False
350
351 2148 aaronmk
    try:
352
        def run(): return run_raw_query(db, query, params, cacheable)
353
        if recover and not db.is_cached(query, params):
354
            return with_savepoint(db, run)
355
        else: return run() # don't need savepoint if cached
356
    except Exception, e:
357
        if not recover: raise # need savepoint to run index_cols()
358 2247 aaronmk
        msg = exc.str_(e)
359 2240 aaronmk
360 2148 aaronmk
        match = re.search(r'duplicate key value violates unique constraint '
361 2368 aaronmk
            r'"((_?[^\W_]+)_[^"]+?)"', msg)
362 2148 aaronmk
        if match:
363
            constraint, table = match.groups()
364
            try: cols = index_cols(db, table, constraint)
365
            except NotImplementedError: raise e
366 2306 aaronmk
            else: raise DuplicateKeyException(constraint, cols, e)
367 2240 aaronmk
368 2368 aaronmk
        match = re.search(r'null value in column "(\w+?)" violates not-null '
369 2239 aaronmk
            r'constraint', msg)
370 2345 aaronmk
        if match: raise NullValueException('NOT NULL', [match.group(1)], e)
371 2240 aaronmk
372 2381 aaronmk
        match = re.search(r'\b(?:invalid input (?:syntax|value)\b.*?'
373
            r'|date/time field value out of range): "(.+?)"\n'
374 2368 aaronmk
            r'(?:(?s).*?)\bfunction "(\w+?)".*?\bat assignment', msg)
375 2240 aaronmk
        if match:
376 2373 aaronmk
            value, name = match.groups()
377
            raise FunctionValueException(name, strings.to_unicode(value), e)
378 2240 aaronmk
379 2368 aaronmk
        match = re.search(r'relation "(\w+?)" already exists', msg)
380 2148 aaronmk
        if match: raise DuplicateTableException(match.group(1), e)
381 2240 aaronmk
382 2368 aaronmk
        match = re.search(r'function "(\w+?)" already exists', msg)
383 2188 aaronmk
        if match: raise DuplicateFunctionException(match.group(1), e)
384 2240 aaronmk
385 2148 aaronmk
        raise # no specific exception raised
386 830 aaronmk
387 832 aaronmk
##### Basic queries
388
389 2153 aaronmk
def next_version(name):
390
    '''Prepends the version # so it won't be removed if the name is truncated'''
391 2163 aaronmk
    version = 1 # first existing name was version 0
392 2153 aaronmk
    match = re.match(r'^v(\d+)_(.*)$', name)
393
    if match:
394
        version = int(match.group(1))+1
395
        name = match.group(2)
396
    return 'v'+str(version)+'_'+name
397
398 2386 aaronmk
def run_query_into(db, query, params, into=None, *args, **kw_args):
399 2085 aaronmk
    '''Outputs a query to a temp table.
400
    For params, see run_query().
401
    '''
402 2386 aaronmk
    if into == None: return run_query(db, query, params, *args, **kw_args)
403 2085 aaronmk
    else: # place rows in temp table
404 2386 aaronmk
        assert isinstance(into, sql_gen.Table)
405 2385 aaronmk
406 2153 aaronmk
        kw_args['recover'] = True
407
        while True:
408
            try:
409 2194 aaronmk
                create_query = 'CREATE'
410
                if not db.debug: create_query += ' TEMP'
411 2386 aaronmk
                create_query += ' TABLE '+into.to_str(db)+' AS '+query
412 2194 aaronmk
413
                return run_query(db, create_query, params, *args, **kw_args)
414 2153 aaronmk
                    # CREATE TABLE AS sets rowcount to # rows in query
415
            except DuplicateTableException, e:
416 2386 aaronmk
                into.name = next_version(into.name)
417 2153 aaronmk
                # try again with next version of name
418 2085 aaronmk
419 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
420
421 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
422
423 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
424 2293 aaronmk
    start=None, order_by=order_by_pkey, default_table=None):
425 1981 aaronmk
    '''
426 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
427 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
428 1981 aaronmk
    @param fields Use None to select all fields in the table
429 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
430 2379 aaronmk
        * container can be any iterable type
431 2218 aaronmk
        * compare_left_side: Code|str (for col name)
432
        * compare_right_side: ValueCond|literal value
433 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
434
        use all columns
435 2054 aaronmk
    @return tuple(query, params)
436 1981 aaronmk
    '''
437 2315 aaronmk
    # Parse tables param
438 2121 aaronmk
    if not lists.is_seq(tables): tables = [tables]
439 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
440 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
441 2121 aaronmk
442 2315 aaronmk
    # Parse other params
443 2376 aaronmk
    if conds == None: conds = []
444
    elif isinstance(conds, dict): conds = conds.items()
445 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
446 135 aaronmk
    assert limit == None or type(limit) == int
447 865 aaronmk
    assert start == None or type(start) == int
448 2315 aaronmk
    if order_by is order_by_pkey:
449
        if distinct_on != []: order_by = None
450
        else: order_by = pkey(db, table0, recover=True)
451 865 aaronmk
452 2315 aaronmk
    query = 'SELECT'
453 2056 aaronmk
454 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
455 2056 aaronmk
456 2200 aaronmk
    # DISTINCT ON columns
457 2233 aaronmk
    if distinct_on != []:
458 2200 aaronmk
        query += ' DISTINCT'
459 2254 aaronmk
        if distinct_on is not distinct_on_all:
460 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
461
462
    # Columns
463
    query += ' '
464 1135 aaronmk
    if fields == None: query += '*'
465 2056 aaronmk
    else: query += ', '.join(map(parse_col, fields))
466 2200 aaronmk
467
    # Main table
468 2271 aaronmk
    query += ' FROM '+table0.to_str(db)
469 865 aaronmk
470 2122 aaronmk
    # Add joins
471 2271 aaronmk
    left_table = table0
472 2263 aaronmk
    for join_ in tables:
473
        table = join_.table
474 2238 aaronmk
475 2343 aaronmk
        # Parse special values
476
        if join_.type_ is sql_gen.filter_out: # filter no match
477 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
478
                None))
479 2343 aaronmk
480 2277 aaronmk
        query += ' '+join_.to_str(db, left_table)
481 2122 aaronmk
482
        left_table = table
483
484 865 aaronmk
    missing = True
485 2376 aaronmk
    if conds != []:
486 2284 aaronmk
        query += ' WHERE '+(' AND '.join((sql_gen.as_ValueCond(r).to_str(db, l)
487 2376 aaronmk
            for l, r in conds)))
488 865 aaronmk
        missing = False
489 2227 aaronmk
    if order_by != None:
490 2285 aaronmk
        query += ' ORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
491 865 aaronmk
    if limit != None: query += ' LIMIT '+str(limit); missing = False
492
    if start != None:
493
        if start != 0: query += ' OFFSET '+str(start)
494
        missing = False
495
    if missing: warnings.warn(DbWarning(
496
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
497
498 2315 aaronmk
    return (query, [])
499 11 aaronmk
500 2054 aaronmk
def select(db, *args, **kw_args):
501
    '''For params, see mk_select() and run_query()'''
502
    recover = kw_args.pop('recover', None)
503
    cacheable = kw_args.pop('cacheable', True)
504
505
    query, params = mk_select(db, *args, **kw_args)
506
    return run_query(db, query, params, recover, cacheable)
507
508 2066 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, params=None,
509 2292 aaronmk
    returning=None, embeddable=False):
510 1960 aaronmk
    '''
511
    @param returning str|None An inserted column (such as pkey) to return
512 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
513 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
514
        query will be fully cached, not just if it raises an exception.
515 1960 aaronmk
    '''
516 2328 aaronmk
    table = sql_gen.as_Table(table)
517 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
518
    if cols != None: cols = [sql_gen.as_Col(v).to_str(db) for v in cols]
519 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
520 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
521 2063 aaronmk
522
    # Build query
523 2328 aaronmk
    query = 'INSERT INTO '+table.to_str(db)
524 2318 aaronmk
    if cols != None: query += ' ('+', '.join(cols)+')'
525 2063 aaronmk
    query += ' '+select_query
526
527
    if returning != None:
528 2327 aaronmk
        returning_name = copy.copy(returning)
529
        returning_name.table = None
530
        returning_name = returning_name.to_str(db)
531
        query += ' RETURNING '+returning_name
532 2063 aaronmk
533 2070 aaronmk
    if embeddable:
534 2327 aaronmk
        assert returning != None
535
536 2070 aaronmk
        # Create function
537 2330 aaronmk
        function_name = '_'.join(['insert', table.name] + cols)
538 2327 aaronmk
        return_type = 'SETOF '+returning.to_str(db)+'%TYPE'
539 2189 aaronmk
        while True:
540
            try:
541 2327 aaronmk
                func_schema = None
542
                if not db.debug: func_schema = 'pg_temp'
543
                function = sql_gen.Table(function_name, func_schema).to_str(db)
544 2194 aaronmk
545 2189 aaronmk
                function_query = '''\
546
CREATE FUNCTION '''+function+'''() RETURNS '''+return_type+'''
547 2070 aaronmk
    LANGUAGE sql
548
    AS $$'''+mogrify(db, query, params)+''';$$;
549
'''
550 2189 aaronmk
                run_query(db, function_query, recover=True, cacheable=True)
551
                break # this version was successful
552
            except DuplicateFunctionException, e:
553
                function_name = next_version(function_name)
554
                # try again with next version of name
555 2070 aaronmk
556 2337 aaronmk
        # Return query that uses function
557
        func_table = sql_gen.NamedTable('f', sql_gen.CustomCode(function+'()'),
558
            [returning_name]) # AS clause requires function alias
559
        return mk_select(db, func_table, start=0, order_by=None)
560 2070 aaronmk
561 2066 aaronmk
    return (query, params)
562
563
def insert_select(db, *args, **kw_args):
564 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
565 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
566
        values in
567 2072 aaronmk
    '''
568 2386 aaronmk
    into = kw_args.pop('into', None)
569
    if into != None: kw_args['embeddable'] = True
570 2066 aaronmk
    recover = kw_args.pop('recover', None)
571
    cacheable = kw_args.pop('cacheable', True)
572
573
    query, params = mk_insert_select(db, *args, **kw_args)
574 2386 aaronmk
    return run_query_into(db, query, params, into, recover=recover,
575 2153 aaronmk
        cacheable=cacheable)
576 2063 aaronmk
577 2066 aaronmk
default = object() # tells insert() to use the default value for a column
578
579 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
580 2085 aaronmk
    '''For params, see insert_select()'''
581 1960 aaronmk
    if lists.is_seq(row): cols = None
582
    else:
583
        cols = row.keys()
584
        row = row.values()
585
    row = list(row) # ensure that "!= []" works
586
587 1961 aaronmk
    # Check for special values
588
    labels = []
589
    values = []
590
    for value in row:
591 2254 aaronmk
        if value is default: labels.append('DEFAULT')
592 1961 aaronmk
        else:
593
            labels.append('%s')
594
            values.append(value)
595
596
    # Build query
597 2063 aaronmk
    if values != []: query = ' VALUES ('+(', '.join(labels))+')'
598
    else: query = None
599 1554 aaronmk
600 2064 aaronmk
    return insert_select(db, table, cols, query, values, *args, **kw_args)
601 11 aaronmk
602 135 aaronmk
def last_insert_id(db):
603 1849 aaronmk
    module = util.root_module(db.db)
604 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
605
    elif module == 'MySQLdb': return db.insert_id()
606
    else: return None
607 13 aaronmk
608 1968 aaronmk
def truncate(db, table, schema='public'):
609
    return run_query(db, 'TRUNCATE '+qual_name(db, schema, table)+' CASCADE')
610 832 aaronmk
611
##### Database structure queries
612
613 2339 aaronmk
def table_cols(db, table, recover=None):
614
    return list(col_names(select(db, table, limit=0, order_by=None,
615
        recover=recover)))
616
617 2389 aaronmk
def mk_flatten_mapping(db, tables, flat_table, preserve=[]):
618 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
619
    to names that will be distinct among the given tables.
620
    This is meant to be used for several tables that are being joined together.
621
    @param flat_table The table for the new columns
622 2389 aaronmk
    @param preserve List of columns not to rename
623 2383 aaronmk
    @return dict(orig_col=new_col, ...)
624
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
625
        * new_col: sql_gen.Col(orig_col_name, flat_table)
626
        * All mappings use the flat_table table so its name can easily be
627
          changed for all columns at once
628
    '''
629
    flatten_mapping = {}
630
    for table in tables:
631
        for col in table_cols(db, table):
632
            col = sql_gen.as_Col(col, table)
633
            flatten_mapping[col] = sql_gen.Col(str(col), flat_table)
634 2389 aaronmk
    for col in preserve:
635
        flatten_mapping[col] = sql_gen.Col(col.name, flat_table)
636 2383 aaronmk
    return flatten_mapping
637
638 2291 aaronmk
def pkey(db, table, recover=None):
639 832 aaronmk
    '''Assumed to be first column in table'''
640 2339 aaronmk
    return table_cols(db, table, recover)[0]
641 832 aaronmk
642 2340 aaronmk
not_null_col = 'not_null'
643
644
def table_not_null_col(db, table, recover=None):
645
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
646
    if not_null_col in table_cols(db, table, recover): return not_null_col
647
    else: return pkey(db, table, recover)
648
649 853 aaronmk
def index_cols(db, table, index):
650
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
651
    automatically created. When you don't know whether something is a UNIQUE
652
    constraint or a UNIQUE index, use this function.'''
653 1909 aaronmk
    module = util.root_module(db.db)
654
    if module == 'psycopg2':
655
        return list(values(run_query(db, '''\
656 853 aaronmk
SELECT attname
657 866 aaronmk
FROM
658
(
659
        SELECT attnum, attname
660
        FROM pg_index
661
        JOIN pg_class index ON index.oid = indexrelid
662
        JOIN pg_class table_ ON table_.oid = indrelid
663
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
664
        WHERE
665
            table_.relname = %(table)s
666
            AND index.relname = %(index)s
667
    UNION
668
        SELECT attnum, attname
669
        FROM
670
        (
671
            SELECT
672
                indrelid
673
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
674
                    AS indkey
675
            FROM pg_index
676
            JOIN pg_class index ON index.oid = indexrelid
677
            JOIN pg_class table_ ON table_.oid = indrelid
678
            WHERE
679
                table_.relname = %(table)s
680
                AND index.relname = %(index)s
681
        ) s
682
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
683
) s
684 853 aaronmk
ORDER BY attnum
685
''',
686 1909 aaronmk
            {'table': table, 'index': index}, cacheable=True)))
687
    else: raise NotImplementedError("Can't list index columns for "+module+
688
        ' database')
689 853 aaronmk
690 464 aaronmk
def constraint_cols(db, table, constraint):
691 1849 aaronmk
    module = util.root_module(db.db)
692 464 aaronmk
    if module == 'psycopg2':
693
        return list(values(run_query(db, '''\
694
SELECT attname
695
FROM pg_constraint
696
JOIN pg_class ON pg_class.oid = conrelid
697
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
698
WHERE
699
    relname = %(table)s
700
    AND conname = %(constraint)s
701
ORDER BY attnum
702
''',
703
            {'table': table, 'constraint': constraint})))
704
    else: raise NotImplementedError("Can't list constraint columns for "+module+
705
        ' database')
706
707 2096 aaronmk
row_num_col = '_row_num'
708
709 2086 aaronmk
def add_row_num(db, table):
710 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
711
    be the primary key.'''
712 2320 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
713 2096 aaronmk
    run_query(db, 'ALTER TABLE '+table+' ADD COLUMN '+row_num_col
714 2117 aaronmk
        +' serial NOT NULL PRIMARY KEY')
715 2086 aaronmk
716 1968 aaronmk
def tables(db, schema='public', table_like='%'):
717 1849 aaronmk
    module = util.root_module(db.db)
718 1968 aaronmk
    params = {'schema': schema, 'table_like': table_like}
719 832 aaronmk
    if module == 'psycopg2':
720 1968 aaronmk
        return values(run_query(db, '''\
721
SELECT tablename
722
FROM pg_tables
723
WHERE
724
    schemaname = %(schema)s
725
    AND tablename LIKE %(table_like)s
726
ORDER BY tablename
727
''',
728
            params, cacheable=True))
729
    elif module == 'MySQLdb':
730
        return values(run_query(db, 'SHOW TABLES LIKE %(table_like)s', params,
731
            cacheable=True))
732 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
733 830 aaronmk
734 833 aaronmk
##### Database management
735
736 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
737
    '''For kw_args, see tables()'''
738
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
739 833 aaronmk
740 832 aaronmk
##### Heuristic queries
741
742 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
743 1554 aaronmk
    '''Recovers from errors.
744 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
745
    '''
746 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
747
748 471 aaronmk
    try:
749 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
750 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
751
            row_ct_ref[0] += cur.rowcount
752
        return value(cur)
753 471 aaronmk
    except DuplicateKeyException, e:
754 2104 aaronmk
        return value(select(db, table, [pkey_],
755 1069 aaronmk
            util.dict_subset_right_join(row, e.cols), recover=True))
756 471 aaronmk
757 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
758 830 aaronmk
    '''Recovers from errors'''
759 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
760 14 aaronmk
    except StopIteration:
761 40 aaronmk
        if not create: raise
762 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
763 2078 aaronmk
764 2134 aaronmk
def put_table(db, out_table, in_tables, mapping, limit=None, start=0,
765 2279 aaronmk
    row_ct_ref=None):
766 2078 aaronmk
    '''Recovers from errors.
767
    Only works under PostgreSQL (uses INSERT RETURNING).
768 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
769
        tables to join with it using the main input table's pkey
770 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
771
        * out_table_col: sql_gen.Col|str
772 2323 aaronmk
        * in_table_col: sql_gen.Col Wrap literal values in a sql_gen.NamedCol
773 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
774 2078 aaronmk
    '''
775 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
776 2312 aaronmk
    for in_table_col in mapping.itervalues():
777
        assert isinstance(in_table_col, sql_gen.Col)
778
779 2329 aaronmk
    temp_prefix = out_table.name
780 2387 aaronmk
    pkeys = sql_gen.Table(temp_prefix+'_pkeys')
781 2131 aaronmk
782 2382 aaronmk
    # Create input joins from list of input tables
783
    in_tables_ = in_tables[:] # don't modify input!
784
    in_tables0 = in_tables_.pop(0) # first table is separate
785 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
786 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
787 2352 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v, {in_pkey: sql_gen.join_same})
788 2382 aaronmk
        for v in in_tables_]
789 2131 aaronmk
790 2279 aaronmk
    out_pkey = pkey(db, out_table, recover=True)
791 2285 aaronmk
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
792 2142 aaronmk
793 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
794 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
795
796 2201 aaronmk
    pkeys_table_exists_ref = [False]
797
    def run_query_into_pkeys(query, params):
798
        if pkeys_table_exists_ref[0]:
799 2387 aaronmk
            insert_select(db, pkeys, pkeys_names, query, params)
800 2201 aaronmk
        else:
801 2387 aaronmk
            run_query_into(db, query, params, into=pkeys)
802 2201 aaronmk
            pkeys_table_exists_ref[0] = True
803
804 2380 aaronmk
    conds = set()
805 2233 aaronmk
    distinct_on = []
806 2325 aaronmk
    def mk_main_select(joins, cols):
807
        return mk_select(db, joins, cols, conds, distinct_on, limit=limit,
808
            start=start, default_table=in_tables0)
809 2132 aaronmk
810 2309 aaronmk
    def log_exc(e):
811
        db.log_debug('Caught exception: '+exc.str_(e, first_line_only=True))
812 2358 aaronmk
    def ignore(in_col, value):
813 2365 aaronmk
        db.log_debug('Ignoring rows with '+str(in_col)+' = '+repr(value))
814 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
815
        assert cond not in conds # avoid infinite loops
816 2380 aaronmk
        conds.add(cond)
817 2245 aaronmk
818 2206 aaronmk
    # Do inserts and selects
819 2257 aaronmk
    join_cols = {}
820 2387 aaronmk
    out_pkeys = sql_gen.Table(temp_prefix+'_out_pkeys')
821
    in_pkeys = sql_gen.Table(temp_prefix+'_in_pkeys')
822 2206 aaronmk
    while True:
823 2303 aaronmk
        has_joins = join_cols != {}
824
825 2305 aaronmk
        # Prepare to insert new rows
826 2325 aaronmk
        insert_joins = input_joins[:] # don't modify original!
827
        insert_args = dict(recover=True)
828 2303 aaronmk
        if has_joins:
829 2317 aaronmk
            distinct_on = [v.to_Col() for v in join_cols.values()]
830 2325 aaronmk
            insert_joins.append(sql_gen.Join(out_table, join_cols,
831
                sql_gen.filter_out))
832
        else:
833 2387 aaronmk
            insert_args.update(dict(returning=out_pkey, into=out_pkeys))
834 2303 aaronmk
835 2305 aaronmk
        db.log_debug('Inserting new rows')
836 2206 aaronmk
        try:
837
            cur = insert_select(db, out_table, mapping.keys(),
838 2325 aaronmk
                *mk_main_select(insert_joins, mapping.values()), **insert_args)
839 2357 aaronmk
            break # insert successful
840 2206 aaronmk
        except DuplicateKeyException, e:
841 2309 aaronmk
            log_exc(e)
842
843 2258 aaronmk
            old_join_cols = join_cols.copy()
844 2334 aaronmk
            join_cols.update(util.dict_subset(mapping, e.cols))
845 2245 aaronmk
            db.log_debug('Ignoring existing rows, comparing on '+str(join_cols))
846 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
847 2230 aaronmk
        except NullValueException, e:
848 2309 aaronmk
            log_exc(e)
849
850 2230 aaronmk
            out_col, = e.cols
851
            try: in_col = mapping[out_col]
852 2356 aaronmk
            except KeyError:
853
                db.log_debug('Missing mapping for NOT NULL '+out_col)
854
                limit = 0 # just create an empty pkeys table
855 2365 aaronmk
            else: ignore(in_col, None)
856 2243 aaronmk
        except FunctionValueException, e:
857 2309 aaronmk
            log_exc(e)
858
859 2344 aaronmk
            assert e.name == out_table.name
860 2243 aaronmk
            out_col = 'value' # assume function param was named "value"
861 2358 aaronmk
            ignore(mapping[out_col], e.value)
862
        # after exception handled, rerun loop with additional constraints
863 2132 aaronmk
864 2357 aaronmk
    if row_ct_ref != None and cur.rowcount >= 0:
865
        row_ct_ref[0] += cur.rowcount
866
867
    if has_joins:
868
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
869
        db.log_debug('Getting output pkeys of existing/inserted rows')
870 2387 aaronmk
        run_query_into_pkeys(*mk_select(db, select_joins, pkeys_cols, start=0))
871 2357 aaronmk
    else:
872 2387 aaronmk
        add_row_num(db, out_pkeys) # for joining with input pkeys
873 2357 aaronmk
874
        db.log_debug('Getting input pkeys for rows in insert')
875
        run_query_into(db, *mk_main_select(input_joins, [in_pkey]),
876 2387 aaronmk
            into=in_pkeys)
877
        add_row_num(db, in_pkeys) # for joining with output pkeys
878 2357 aaronmk
879
        db.log_debug('Joining together output and input pkeys')
880 2387 aaronmk
        pkey_joins = [in_pkeys, sql_gen.Join(out_pkeys,
881 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
882 2387 aaronmk
        run_query_into_pkeys(*mk_select(db, pkey_joins, pkeys_names, start=0))
883 2357 aaronmk
884
    db.log_debug("Setting missing rows' pkeys to NULL")
885 2387 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(pkeys,
886 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
887
        # must use join_same_not_null or query will take forever
888
    run_query_into_pkeys(*mk_select(db, missing_rows_joins,
889
        [in_pkey_col, sql_gen.NamedCol(out_pkey, None)], start=0))
890
891 2387 aaronmk
    return sql_gen.Col(out_pkey, pkeys)
892 2115 aaronmk
893
##### Data cleanup
894
895 2290 aaronmk
def cleanup_table(db, table, cols):
896 2115 aaronmk
    def esc_name_(name): return esc_name(db, name)
897
898 2290 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
899 2115 aaronmk
    cols = map(esc_name_, cols)
900
901
    run_query(db, 'UPDATE '+table+' SET\n'+(',\n'.join(('\n'+col
902
        +' = nullif(nullif(trim(both from '+col+"), %(null0)s), %(null1)s)"
903
            for col in cols))),
904
        dict(null0='', null1=r'\N'))