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 2396 aaronmk
def clean_name(name): return re.sub(r'\W', r'', name.replace('.', '_'))
104
105 2101 aaronmk
def check_name(name):
106
    if re.search(r'\W', name) != None: raise NameException('Name "'+name
107
        +'" may contain only alphanumeric characters and _')
108
109
def esc_name_by_module(module, name, ignore_case=False):
110 2388 aaronmk
    if module == 'psycopg2' or module == None:
111 2101 aaronmk
        if ignore_case:
112
            # Don't enclose in quotes because this disables case-insensitivity
113
            check_name(name)
114
            return name
115
        else: quote = '"'
116
    elif module == 'MySQLdb': quote = '`'
117
    else: raise NotImplementedError("Can't escape name for "+module+' database')
118
    return quote + name.replace(quote, '') + quote
119
120
def esc_name_by_engine(engine, name, **kw_args):
121
    return esc_name_by_module(db_engines[engine][0], name, **kw_args)
122
123
def esc_name(db, name, **kw_args):
124
    return esc_name_by_module(util.root_module(db.db), name, **kw_args)
125
126
def qual_name(db, schema, table):
127
    def esc_name_(name): return esc_name(db, name)
128
    table = esc_name_(table)
129
    if schema != None: return esc_name_(schema)+'.'+table
130
    else: return table
131
132 1869 aaronmk
##### Database connections
133 1849 aaronmk
134 2097 aaronmk
db_config_names = ['engine', 'host', 'user', 'password', 'database', 'schemas']
135 1926 aaronmk
136 1869 aaronmk
db_engines = {
137
    'MySQL': ('MySQLdb', {'password': 'passwd', 'database': 'db'}),
138
    'PostgreSQL': ('psycopg2', {}),
139
}
140
141
DatabaseErrors_set = set([DbException])
142
DatabaseErrors = tuple(DatabaseErrors_set)
143
144
def _add_module(module):
145
    DatabaseErrors_set.add(module.DatabaseError)
146
    global DatabaseErrors
147
    DatabaseErrors = tuple(DatabaseErrors_set)
148
149
def db_config_str(db_config):
150
    return db_config['engine']+' database '+db_config['database']
151
152 1909 aaronmk
def _query_lookup(query, params): return (query, dicts.make_hashable(params))
153 1894 aaronmk
154 1901 aaronmk
log_debug_none = lambda msg: None
155
156 1849 aaronmk
class DbConn:
157 2190 aaronmk
    def __init__(self, db_config, serializable=True, autocommit=False,
158
        caching=True, log_debug=log_debug_none):
159 1869 aaronmk
        self.db_config = db_config
160
        self.serializable = serializable
161 2190 aaronmk
        self.autocommit = autocommit
162
        self.caching = caching
163 1901 aaronmk
        self.log_debug = log_debug
164 2193 aaronmk
        self.debug = log_debug != log_debug_none
165 1869 aaronmk
166
        self.__db = None
167 1889 aaronmk
        self.query_results = {}
168 2139 aaronmk
        self._savepoint = 0
169 1869 aaronmk
170
    def __getattr__(self, name):
171
        if name == '__dict__': raise Exception('getting __dict__')
172
        if name == 'db': return self._db()
173
        else: raise AttributeError()
174
175
    def __getstate__(self):
176
        state = copy.copy(self.__dict__) # shallow copy
177 1915 aaronmk
        state['log_debug'] = None # don't pickle the debug callback
178 1869 aaronmk
        state['_DbConn__db'] = None # don't pickle the connection
179
        return state
180
181 2165 aaronmk
    def connected(self): return self.__db != None
182
183 1869 aaronmk
    def _db(self):
184
        if self.__db == None:
185
            # Process db_config
186
            db_config = self.db_config.copy() # don't modify input!
187 2097 aaronmk
            schemas = db_config.pop('schemas', None)
188 1869 aaronmk
            module_name, mappings = db_engines[db_config.pop('engine')]
189
            module = __import__(module_name)
190
            _add_module(module)
191
            for orig, new in mappings.iteritems():
192
                try: util.rename_key(db_config, orig, new)
193
                except KeyError: pass
194
195
            # Connect
196
            self.__db = module.connect(**db_config)
197
198
            # Configure connection
199 2234 aaronmk
            if self.serializable and not self.autocommit: run_raw_query(self,
200
                'SET TRANSACTION ISOLATION LEVEL SERIALIZABLE')
201 2101 aaronmk
            if schemas != None:
202
                schemas_ = ''.join((esc_name(self, s)+', '
203
                    for s in schemas.split(',')))
204
                run_raw_query(self, "SELECT set_config('search_path', \
205
%s || current_setting('search_path'), false)", [schemas_])
206 1869 aaronmk
207
        return self.__db
208 1889 aaronmk
209 1891 aaronmk
    class DbCursor(Proxy):
210 1927 aaronmk
        def __init__(self, outer):
211 1891 aaronmk
            Proxy.__init__(self, outer.db.cursor())
212 2191 aaronmk
            self.outer = outer
213 1927 aaronmk
            self.query_results = outer.query_results
214 1894 aaronmk
            self.query_lookup = None
215 1891 aaronmk
            self.result = []
216 1889 aaronmk
217 1894 aaronmk
        def execute(self, query, params=None):
218 1930 aaronmk
            self._is_insert = query.upper().find('INSERT') >= 0
219 1894 aaronmk
            self.query_lookup = _query_lookup(query, params)
220 2148 aaronmk
            try:
221 2191 aaronmk
                try:
222
                    return_value = self.inner.execute(query, params)
223
                    self.outer.do_autocommit()
224 2148 aaronmk
                finally: self.query = get_cur_query(self.inner)
225 1904 aaronmk
            except Exception, e:
226 2170 aaronmk
                _add_cursor_info(e, self, query, params)
227 1904 aaronmk
                self.result = e # cache the exception as the result
228
                self._cache_result()
229
                raise
230 1930 aaronmk
            # Fetch all rows so result will be cached
231
            if self.rowcount == 0 and not self._is_insert: consume_rows(self)
232 1894 aaronmk
            return return_value
233
234 1891 aaronmk
        def fetchone(self):
235
            row = self.inner.fetchone()
236 1899 aaronmk
            if row != None: self.result.append(row)
237
            # otherwise, fetched all rows
238 1904 aaronmk
            else: self._cache_result()
239
            return row
240
241
        def _cache_result(self):
242 1906 aaronmk
            # For inserts, only cache exceptions since inserts are not
243
            # idempotent, but an invalid insert will always be invalid
244 1930 aaronmk
            if self.query_results != None and (not self._is_insert
245 1906 aaronmk
                or isinstance(self.result, Exception)):
246
247 1894 aaronmk
                assert self.query_lookup != None
248 1916 aaronmk
                self.query_results[self.query_lookup] = self.CacheCursor(
249
                    util.dict_subset(dicts.AttrsDictView(self),
250
                    ['query', 'result', 'rowcount', 'description']))
251 1906 aaronmk
252 1916 aaronmk
        class CacheCursor:
253
            def __init__(self, cached_result): self.__dict__ = cached_result
254
255 1927 aaronmk
            def execute(self, *args, **kw_args):
256 1916 aaronmk
                if isinstance(self.result, Exception): raise self.result
257
                # otherwise, result is a rows list
258
                self.iter = iter(self.result)
259
260
            def fetchone(self):
261
                try: return self.iter.next()
262
                except StopIteration: return None
263 1891 aaronmk
264 2212 aaronmk
    def esc_value(self, value):
265 2215 aaronmk
        module = util.root_module(self.db)
266 2374 aaronmk
        if module == 'psycopg2': str_ = self.db.cursor().mogrify('%s', [value])
267 2212 aaronmk
        elif module == 'MySQLdb':
268
            import _mysql
269 2374 aaronmk
            str_ = _mysql.escape_string(value)
270 2212 aaronmk
        else: raise NotImplementedError("Can't escape value for "+module
271
            +' database')
272 2374 aaronmk
        return strings.to_unicode(str_)
273 2212 aaronmk
274 2347 aaronmk
    def esc_name(self, name): return esc_name(self, name) # calls global func
275
276 1894 aaronmk
    def run_query(self, query, params=None, cacheable=False):
277 2148 aaronmk
        '''Translates known DB errors to typed exceptions:
278
        See self.DbCursor.execute().'''
279 2167 aaronmk
        assert query != None
280
281 2047 aaronmk
        if not self.caching: cacheable = False
282 1903 aaronmk
        used_cache = False
283
        try:
284 1927 aaronmk
            # Get cursor
285
            if cacheable:
286
                query_lookup = _query_lookup(query, params)
287
                try:
288
                    cur = self.query_results[query_lookup]
289
                    used_cache = True
290
                except KeyError: cur = self.DbCursor(self)
291
            else: cur = self.db.cursor()
292
293
            # Run query
294 2148 aaronmk
            cur.execute(query, params)
295 1903 aaronmk
        finally:
296 2193 aaronmk
            if self.debug: # only compute msg if needed
297 1903 aaronmk
                if used_cache: cache_status = 'Cache hit'
298
                elif cacheable: cache_status = 'Cache miss'
299
                else: cache_status = 'Non-cacheable'
300 1927 aaronmk
                self.log_debug(cache_status+': '
301 2170 aaronmk
                    +strings.one_line(str(get_cur_query(cur, query, params))))
302 1903 aaronmk
303
        return cur
304 1914 aaronmk
305
    def is_cached(self, query, params=None):
306
        return _query_lookup(query, params) in self.query_results
307 2139 aaronmk
308
    def with_savepoint(self, func):
309 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
310 2139 aaronmk
        self.run_query('SAVEPOINT '+savepoint)
311
        self._savepoint += 1
312
        try:
313
            try: return_val = func()
314
            finally:
315
                self._savepoint -= 1
316
                assert self._savepoint >= 0
317
        except:
318
            self.run_query('ROLLBACK TO SAVEPOINT '+savepoint)
319
            raise
320
        else:
321
            self.run_query('RELEASE SAVEPOINT '+savepoint)
322 2191 aaronmk
            self.do_autocommit()
323 2139 aaronmk
            return return_val
324 2191 aaronmk
325
    def do_autocommit(self):
326
        '''Autocommits if outside savepoint'''
327
        assert self._savepoint >= 0
328
        if self.autocommit and self._savepoint == 0:
329
            self.log_debug('Autocommiting')
330
            self.db.commit()
331 1849 aaronmk
332 1869 aaronmk
connect = DbConn
333
334 832 aaronmk
##### Querying
335
336 1894 aaronmk
def run_raw_query(db, *args, **kw_args):
337 2085 aaronmk
    '''For params, see DbConn.run_query()'''
338 1894 aaronmk
    return db.run_query(*args, **kw_args)
339 11 aaronmk
340 2068 aaronmk
def mogrify(db, query, params):
341
    module = util.root_module(db.db)
342
    if module == 'psycopg2': return db.db.cursor().mogrify(query, params)
343
    else: raise NotImplementedError("Can't mogrify query for "+module+
344
        ' database')
345
346 832 aaronmk
##### Recoverable querying
347 15 aaronmk
348 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
349 11 aaronmk
350 1894 aaronmk
def run_query(db, query, params=None, recover=None, cacheable=False):
351 830 aaronmk
    if recover == None: recover = False
352
353 2148 aaronmk
    try:
354
        def run(): return run_raw_query(db, query, params, cacheable)
355
        if recover and not db.is_cached(query, params):
356
            return with_savepoint(db, run)
357
        else: return run() # don't need savepoint if cached
358
    except Exception, e:
359
        if not recover: raise # need savepoint to run index_cols()
360 2247 aaronmk
        msg = exc.str_(e)
361 2240 aaronmk
362 2148 aaronmk
        match = re.search(r'duplicate key value violates unique constraint '
363 2368 aaronmk
            r'"((_?[^\W_]+)_[^"]+?)"', msg)
364 2148 aaronmk
        if match:
365
            constraint, table = match.groups()
366
            try: cols = index_cols(db, table, constraint)
367
            except NotImplementedError: raise e
368 2306 aaronmk
            else: raise DuplicateKeyException(constraint, cols, e)
369 2240 aaronmk
370 2368 aaronmk
        match = re.search(r'null value in column "(\w+?)" violates not-null '
371 2239 aaronmk
            r'constraint', msg)
372 2345 aaronmk
        if match: raise NullValueException('NOT NULL', [match.group(1)], e)
373 2240 aaronmk
374 2381 aaronmk
        match = re.search(r'\b(?:invalid input (?:syntax|value)\b.*?'
375
            r'|date/time field value out of range): "(.+?)"\n'
376 2368 aaronmk
            r'(?:(?s).*?)\bfunction "(\w+?)".*?\bat assignment', msg)
377 2240 aaronmk
        if match:
378 2373 aaronmk
            value, name = match.groups()
379
            raise FunctionValueException(name, strings.to_unicode(value), e)
380 2240 aaronmk
381 2368 aaronmk
        match = re.search(r'relation "(\w+?)" already exists', msg)
382 2148 aaronmk
        if match: raise DuplicateTableException(match.group(1), e)
383 2240 aaronmk
384 2368 aaronmk
        match = re.search(r'function "(\w+?)" already exists', msg)
385 2188 aaronmk
        if match: raise DuplicateFunctionException(match.group(1), e)
386 2240 aaronmk
387 2148 aaronmk
        raise # no specific exception raised
388 830 aaronmk
389 832 aaronmk
##### Basic queries
390
391 2153 aaronmk
def next_version(name):
392
    '''Prepends the version # so it won't be removed if the name is truncated'''
393 2163 aaronmk
    version = 1 # first existing name was version 0
394 2153 aaronmk
    match = re.match(r'^v(\d+)_(.*)$', name)
395
    if match:
396
        version = int(match.group(1))+1
397
        name = match.group(2)
398
    return 'v'+str(version)+'_'+name
399
400 2386 aaronmk
def run_query_into(db, query, params, into=None, *args, **kw_args):
401 2085 aaronmk
    '''Outputs a query to a temp table.
402
    For params, see run_query().
403
    '''
404 2386 aaronmk
    if into == None: return run_query(db, query, params, *args, **kw_args)
405 2085 aaronmk
    else: # place rows in temp table
406 2386 aaronmk
        assert isinstance(into, sql_gen.Table)
407 2385 aaronmk
408 2153 aaronmk
        kw_args['recover'] = True
409
        while True:
410
            try:
411 2194 aaronmk
                create_query = 'CREATE'
412
                if not db.debug: create_query += ' TEMP'
413 2386 aaronmk
                create_query += ' TABLE '+into.to_str(db)+' AS '+query
414 2194 aaronmk
415
                return run_query(db, create_query, params, *args, **kw_args)
416 2153 aaronmk
                    # CREATE TABLE AS sets rowcount to # rows in query
417
            except DuplicateTableException, e:
418 2386 aaronmk
                into.name = next_version(into.name)
419 2153 aaronmk
                # try again with next version of name
420 2085 aaronmk
421 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
422
423 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
424
425 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
426 2293 aaronmk
    start=None, order_by=order_by_pkey, default_table=None):
427 1981 aaronmk
    '''
428 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
429 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
430 1981 aaronmk
    @param fields Use None to select all fields in the table
431 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
432 2379 aaronmk
        * container can be any iterable type
433 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
434
        * compare_right_side: sql_gen.ValueCond|literal value
435 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
436
        use all columns
437 2054 aaronmk
    @return tuple(query, params)
438 1981 aaronmk
    '''
439 2315 aaronmk
    # Parse tables param
440 2121 aaronmk
    if not lists.is_seq(tables): tables = [tables]
441 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
442 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
443 2121 aaronmk
444 2315 aaronmk
    # Parse other params
445 2376 aaronmk
    if conds == None: conds = []
446
    elif isinstance(conds, dict): conds = conds.items()
447 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
448 135 aaronmk
    assert limit == None or type(limit) == int
449 865 aaronmk
    assert start == None or type(start) == int
450 2315 aaronmk
    if order_by is order_by_pkey:
451
        if distinct_on != []: order_by = None
452
        else: order_by = pkey(db, table0, recover=True)
453 865 aaronmk
454 2315 aaronmk
    query = 'SELECT'
455 2056 aaronmk
456 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
457 2056 aaronmk
458 2200 aaronmk
    # DISTINCT ON columns
459 2233 aaronmk
    if distinct_on != []:
460 2200 aaronmk
        query += ' DISTINCT'
461 2254 aaronmk
        if distinct_on is not distinct_on_all:
462 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
463
464
    # Columns
465
    query += ' '
466 1135 aaronmk
    if fields == None: query += '*'
467 2056 aaronmk
    else: query += ', '.join(map(parse_col, fields))
468 2200 aaronmk
469
    # Main table
470 2271 aaronmk
    query += ' FROM '+table0.to_str(db)
471 865 aaronmk
472 2122 aaronmk
    # Add joins
473 2271 aaronmk
    left_table = table0
474 2263 aaronmk
    for join_ in tables:
475
        table = join_.table
476 2238 aaronmk
477 2343 aaronmk
        # Parse special values
478
        if join_.type_ is sql_gen.filter_out: # filter no match
479 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
480
                None))
481 2343 aaronmk
482 2277 aaronmk
        query += ' '+join_.to_str(db, left_table)
483 2122 aaronmk
484
        left_table = table
485
486 865 aaronmk
    missing = True
487 2376 aaronmk
    if conds != []:
488 2399 aaronmk
        query += ' WHERE '+(' AND '.join((sql_gen.ColValueCond(l, r).to_str(db)
489 2376 aaronmk
            for l, r in conds)))
490 865 aaronmk
        missing = False
491 2227 aaronmk
    if order_by != None:
492 2285 aaronmk
        query += ' ORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
493 865 aaronmk
    if limit != None: query += ' LIMIT '+str(limit); missing = False
494
    if start != None:
495
        if start != 0: query += ' OFFSET '+str(start)
496
        missing = False
497
    if missing: warnings.warn(DbWarning(
498
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
499
500 2315 aaronmk
    return (query, [])
501 11 aaronmk
502 2054 aaronmk
def select(db, *args, **kw_args):
503
    '''For params, see mk_select() and run_query()'''
504
    recover = kw_args.pop('recover', None)
505
    cacheable = kw_args.pop('cacheable', True)
506
507
    query, params = mk_select(db, *args, **kw_args)
508
    return run_query(db, query, params, recover, cacheable)
509
510 2066 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, params=None,
511 2292 aaronmk
    returning=None, embeddable=False):
512 1960 aaronmk
    '''
513
    @param returning str|None An inserted column (such as pkey) to return
514 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
515 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
516
        query will be fully cached, not just if it raises an exception.
517 1960 aaronmk
    '''
518 2328 aaronmk
    table = sql_gen.as_Table(table)
519 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
520
    if cols != None: cols = [sql_gen.as_Col(v).to_str(db) for v in cols]
521 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
522 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
523 2063 aaronmk
524
    # Build query
525 2328 aaronmk
    query = 'INSERT INTO '+table.to_str(db)
526 2318 aaronmk
    if cols != None: query += ' ('+', '.join(cols)+')'
527 2063 aaronmk
    query += ' '+select_query
528
529
    if returning != None:
530 2327 aaronmk
        returning_name = copy.copy(returning)
531
        returning_name.table = None
532
        returning_name = returning_name.to_str(db)
533
        query += ' RETURNING '+returning_name
534 2063 aaronmk
535 2070 aaronmk
    if embeddable:
536 2327 aaronmk
        assert returning != None
537
538 2070 aaronmk
        # Create function
539 2330 aaronmk
        function_name = '_'.join(['insert', table.name] + cols)
540 2327 aaronmk
        return_type = 'SETOF '+returning.to_str(db)+'%TYPE'
541 2189 aaronmk
        while True:
542
            try:
543 2327 aaronmk
                func_schema = None
544
                if not db.debug: func_schema = 'pg_temp'
545
                function = sql_gen.Table(function_name, func_schema).to_str(db)
546 2194 aaronmk
547 2189 aaronmk
                function_query = '''\
548
CREATE FUNCTION '''+function+'''() RETURNS '''+return_type+'''
549 2070 aaronmk
    LANGUAGE sql
550
    AS $$'''+mogrify(db, query, params)+''';$$;
551
'''
552 2189 aaronmk
                run_query(db, function_query, recover=True, cacheable=True)
553
                break # this version was successful
554
            except DuplicateFunctionException, e:
555
                function_name = next_version(function_name)
556
                # try again with next version of name
557 2070 aaronmk
558 2337 aaronmk
        # Return query that uses function
559
        func_table = sql_gen.NamedTable('f', sql_gen.CustomCode(function+'()'),
560
            [returning_name]) # AS clause requires function alias
561
        return mk_select(db, func_table, start=0, order_by=None)
562 2070 aaronmk
563 2066 aaronmk
    return (query, params)
564
565
def insert_select(db, *args, **kw_args):
566 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
567 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
568
        values in
569 2072 aaronmk
    '''
570 2386 aaronmk
    into = kw_args.pop('into', None)
571
    if into != None: kw_args['embeddable'] = True
572 2066 aaronmk
    recover = kw_args.pop('recover', None)
573
    cacheable = kw_args.pop('cacheable', True)
574
575
    query, params = mk_insert_select(db, *args, **kw_args)
576 2386 aaronmk
    return run_query_into(db, query, params, into, recover=recover,
577 2153 aaronmk
        cacheable=cacheable)
578 2063 aaronmk
579 2066 aaronmk
default = object() # tells insert() to use the default value for a column
580
581 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
582 2085 aaronmk
    '''For params, see insert_select()'''
583 1960 aaronmk
    if lists.is_seq(row): cols = None
584
    else:
585
        cols = row.keys()
586
        row = row.values()
587
    row = list(row) # ensure that "!= []" works
588
589 1961 aaronmk
    # Check for special values
590
    labels = []
591
    values = []
592
    for value in row:
593 2254 aaronmk
        if value is default: labels.append('DEFAULT')
594 1961 aaronmk
        else:
595
            labels.append('%s')
596
            values.append(value)
597
598
    # Build query
599 2063 aaronmk
    if values != []: query = ' VALUES ('+(', '.join(labels))+')'
600
    else: query = None
601 1554 aaronmk
602 2064 aaronmk
    return insert_select(db, table, cols, query, values, *args, **kw_args)
603 11 aaronmk
604 135 aaronmk
def last_insert_id(db):
605 1849 aaronmk
    module = util.root_module(db.db)
606 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
607
    elif module == 'MySQLdb': return db.insert_id()
608
    else: return None
609 13 aaronmk
610 1968 aaronmk
def truncate(db, table, schema='public'):
611
    return run_query(db, 'TRUNCATE '+qual_name(db, schema, table)+' CASCADE')
612 832 aaronmk
613
##### Database structure queries
614
615 2339 aaronmk
def table_cols(db, table, recover=None):
616
    return list(col_names(select(db, table, limit=0, order_by=None,
617
        recover=recover)))
618
619 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
620 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
621
    to names that will be distinct among the given tables.
622
    This is meant to be used for several tables that are being joined together.
623 2392 aaronmk
    @param into The table for the new columns
624 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
625
        columns will be included in the mapping even if they are not in cols.
626
        The tables of the provided Col objects will be changed to into, so make
627
        copies of them if you want to keep the original tables.
628
    @param as_items Whether to return a list of dict items instead of a dict
629 2383 aaronmk
    @return dict(orig_col=new_col, ...)
630
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
631 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
632
        * All mappings use the into table so its name can easily be
633 2383 aaronmk
          changed for all columns at once
634
    '''
635 2394 aaronmk
    items = []
636 2389 aaronmk
    for col in preserve:
637 2390 aaronmk
        orig_col = copy.copy(col)
638 2392 aaronmk
        col.table = into
639 2394 aaronmk
        items.append((orig_col, col))
640
    preserve = set(preserve)
641
    for col in cols:
642 2397 aaronmk
        if col not in preserve:
643
            items.append((col, sql_gen.Col(clean_name(str(col)), into)))
644 2394 aaronmk
645
    if not as_items: items = dict(items)
646
    return items
647 2383 aaronmk
648 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
649 2391 aaronmk
    '''For params, see mk_flatten_mapping()
650
    @return See return value of mk_flatten_mapping()
651
    '''
652 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
653
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
654 2391 aaronmk
    run_query_into(db, *mk_select(db, joins, cols, limit=limit, start=start),
655 2392 aaronmk
        into=into)
656 2394 aaronmk
    return dict(items)
657 2391 aaronmk
658 2291 aaronmk
def pkey(db, table, recover=None):
659 832 aaronmk
    '''Assumed to be first column in table'''
660 2339 aaronmk
    return table_cols(db, table, recover)[0]
661 832 aaronmk
662 2340 aaronmk
not_null_col = 'not_null'
663
664
def table_not_null_col(db, table, recover=None):
665
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
666
    if not_null_col in table_cols(db, table, recover): return not_null_col
667
    else: return pkey(db, table, recover)
668
669 853 aaronmk
def index_cols(db, table, index):
670
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
671
    automatically created. When you don't know whether something is a UNIQUE
672
    constraint or a UNIQUE index, use this function.'''
673 1909 aaronmk
    module = util.root_module(db.db)
674
    if module == 'psycopg2':
675
        return list(values(run_query(db, '''\
676 853 aaronmk
SELECT attname
677 866 aaronmk
FROM
678
(
679
        SELECT attnum, attname
680
        FROM pg_index
681
        JOIN pg_class index ON index.oid = indexrelid
682
        JOIN pg_class table_ ON table_.oid = indrelid
683
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
684
        WHERE
685
            table_.relname = %(table)s
686
            AND index.relname = %(index)s
687
    UNION
688
        SELECT attnum, attname
689
        FROM
690
        (
691
            SELECT
692
                indrelid
693
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
694
                    AS indkey
695
            FROM pg_index
696
            JOIN pg_class index ON index.oid = indexrelid
697
            JOIN pg_class table_ ON table_.oid = indrelid
698
            WHERE
699
                table_.relname = %(table)s
700
                AND index.relname = %(index)s
701
        ) s
702
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
703
) s
704 853 aaronmk
ORDER BY attnum
705
''',
706 1909 aaronmk
            {'table': table, 'index': index}, cacheable=True)))
707
    else: raise NotImplementedError("Can't list index columns for "+module+
708
        ' database')
709 853 aaronmk
710 464 aaronmk
def constraint_cols(db, table, constraint):
711 1849 aaronmk
    module = util.root_module(db.db)
712 464 aaronmk
    if module == 'psycopg2':
713
        return list(values(run_query(db, '''\
714
SELECT attname
715
FROM pg_constraint
716
JOIN pg_class ON pg_class.oid = conrelid
717
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
718
WHERE
719
    relname = %(table)s
720
    AND conname = %(constraint)s
721
ORDER BY attnum
722
''',
723
            {'table': table, 'constraint': constraint})))
724
    else: raise NotImplementedError("Can't list constraint columns for "+module+
725
        ' database')
726
727 2096 aaronmk
row_num_col = '_row_num'
728
729 2086 aaronmk
def add_row_num(db, table):
730 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
731
    be the primary key.'''
732 2320 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
733 2096 aaronmk
    run_query(db, 'ALTER TABLE '+table+' ADD COLUMN '+row_num_col
734 2117 aaronmk
        +' serial NOT NULL PRIMARY KEY')
735 2086 aaronmk
736 1968 aaronmk
def tables(db, schema='public', table_like='%'):
737 1849 aaronmk
    module = util.root_module(db.db)
738 1968 aaronmk
    params = {'schema': schema, 'table_like': table_like}
739 832 aaronmk
    if module == 'psycopg2':
740 1968 aaronmk
        return values(run_query(db, '''\
741
SELECT tablename
742
FROM pg_tables
743
WHERE
744
    schemaname = %(schema)s
745
    AND tablename LIKE %(table_like)s
746
ORDER BY tablename
747
''',
748
            params, cacheable=True))
749
    elif module == 'MySQLdb':
750
        return values(run_query(db, 'SHOW TABLES LIKE %(table_like)s', params,
751
            cacheable=True))
752 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
753 830 aaronmk
754 833 aaronmk
##### Database management
755
756 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
757
    '''For kw_args, see tables()'''
758
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
759 833 aaronmk
760 832 aaronmk
##### Heuristic queries
761
762 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
763 1554 aaronmk
    '''Recovers from errors.
764 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
765
    '''
766 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
767
768 471 aaronmk
    try:
769 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
770 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
771
            row_ct_ref[0] += cur.rowcount
772
        return value(cur)
773 471 aaronmk
    except DuplicateKeyException, e:
774 2104 aaronmk
        return value(select(db, table, [pkey_],
775 1069 aaronmk
            util.dict_subset_right_join(row, e.cols), recover=True))
776 471 aaronmk
777 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
778 830 aaronmk
    '''Recovers from errors'''
779 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
780 14 aaronmk
    except StopIteration:
781 40 aaronmk
        if not create: raise
782 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
783 2078 aaronmk
784 2134 aaronmk
def put_table(db, out_table, in_tables, mapping, limit=None, start=0,
785 2279 aaronmk
    row_ct_ref=None):
786 2078 aaronmk
    '''Recovers from errors.
787
    Only works under PostgreSQL (uses INSERT RETURNING).
788 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
789
        tables to join with it using the main input table's pkey
790 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
791
        * out_table_col: sql_gen.Col|str
792 2323 aaronmk
        * in_table_col: sql_gen.Col Wrap literal values in a sql_gen.NamedCol
793 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
794 2078 aaronmk
    '''
795 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
796 2312 aaronmk
    for in_table_col in mapping.itervalues():
797
        assert isinstance(in_table_col, sql_gen.Col)
798
799 2329 aaronmk
    temp_prefix = out_table.name
800 2387 aaronmk
    pkeys = sql_gen.Table(temp_prefix+'_pkeys')
801 2131 aaronmk
802 2382 aaronmk
    # Create input joins from list of input tables
803
    in_tables_ = in_tables[:] # don't modify input!
804
    in_tables0 = in_tables_.pop(0) # first table is separate
805 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
806 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
807 2352 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v, {in_pkey: sql_gen.join_same})
808 2382 aaronmk
        for v in in_tables_]
809 2131 aaronmk
810 2395 aaronmk
    db.log_debug('Joining together input tables')
811
    # Place in new table for speed and so don't modify input if values edited
812
    in_table = sql_gen.Table(temp_prefix+'_in')
813
    flatten_cols = filter(sql_gen.is_table_col, mapping.values())
814
    mapping = dicts.join(mapping, flatten(db, in_table, input_joins,
815
        flatten_cols, preserve=[in_pkey_col], start=0))
816
    input_joins = [in_table]
817
818 2279 aaronmk
    out_pkey = pkey(db, out_table, recover=True)
819 2285 aaronmk
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
820 2142 aaronmk
821 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
822 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
823
824 2201 aaronmk
    pkeys_table_exists_ref = [False]
825
    def run_query_into_pkeys(query, params):
826
        if pkeys_table_exists_ref[0]:
827 2387 aaronmk
            insert_select(db, pkeys, pkeys_names, query, params)
828 2201 aaronmk
        else:
829 2387 aaronmk
            run_query_into(db, query, params, into=pkeys)
830 2201 aaronmk
            pkeys_table_exists_ref[0] = True
831
832 2380 aaronmk
    conds = set()
833 2233 aaronmk
    distinct_on = []
834 2325 aaronmk
    def mk_main_select(joins, cols):
835
        return mk_select(db, joins, cols, conds, distinct_on, limit=limit,
836 2395 aaronmk
            start=start)
837 2132 aaronmk
838 2309 aaronmk
    def log_exc(e):
839
        db.log_debug('Caught exception: '+exc.str_(e, first_line_only=True))
840 2358 aaronmk
    def ignore(in_col, value):
841 2365 aaronmk
        db.log_debug('Ignoring rows with '+str(in_col)+' = '+repr(value))
842 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
843
        assert cond not in conds # avoid infinite loops
844 2380 aaronmk
        conds.add(cond)
845 2245 aaronmk
846 2206 aaronmk
    # Do inserts and selects
847 2257 aaronmk
    join_cols = {}
848 2387 aaronmk
    out_pkeys = sql_gen.Table(temp_prefix+'_out_pkeys')
849
    in_pkeys = sql_gen.Table(temp_prefix+'_in_pkeys')
850 2206 aaronmk
    while True:
851 2303 aaronmk
        has_joins = join_cols != {}
852
853 2305 aaronmk
        # Prepare to insert new rows
854 2325 aaronmk
        insert_joins = input_joins[:] # don't modify original!
855
        insert_args = dict(recover=True)
856 2303 aaronmk
        if has_joins:
857 2317 aaronmk
            distinct_on = [v.to_Col() for v in join_cols.values()]
858 2325 aaronmk
            insert_joins.append(sql_gen.Join(out_table, join_cols,
859
                sql_gen.filter_out))
860
        else:
861 2387 aaronmk
            insert_args.update(dict(returning=out_pkey, into=out_pkeys))
862 2303 aaronmk
863 2305 aaronmk
        db.log_debug('Inserting new rows')
864 2206 aaronmk
        try:
865
            cur = insert_select(db, out_table, mapping.keys(),
866 2325 aaronmk
                *mk_main_select(insert_joins, mapping.values()), **insert_args)
867 2357 aaronmk
            break # insert successful
868 2206 aaronmk
        except DuplicateKeyException, e:
869 2309 aaronmk
            log_exc(e)
870
871 2258 aaronmk
            old_join_cols = join_cols.copy()
872 2334 aaronmk
            join_cols.update(util.dict_subset(mapping, e.cols))
873 2245 aaronmk
            db.log_debug('Ignoring existing rows, comparing on '+str(join_cols))
874 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
875 2230 aaronmk
        except NullValueException, e:
876 2309 aaronmk
            log_exc(e)
877
878 2230 aaronmk
            out_col, = e.cols
879
            try: in_col = mapping[out_col]
880 2356 aaronmk
            except KeyError:
881
                db.log_debug('Missing mapping for NOT NULL '+out_col)
882
                limit = 0 # just create an empty pkeys table
883 2365 aaronmk
            else: ignore(in_col, None)
884 2243 aaronmk
        except FunctionValueException, e:
885 2309 aaronmk
            log_exc(e)
886
887 2344 aaronmk
            assert e.name == out_table.name
888 2243 aaronmk
            out_col = 'value' # assume function param was named "value"
889 2358 aaronmk
            ignore(mapping[out_col], e.value)
890
        # after exception handled, rerun loop with additional constraints
891 2132 aaronmk
892 2357 aaronmk
    if row_ct_ref != None and cur.rowcount >= 0:
893
        row_ct_ref[0] += cur.rowcount
894
895
    if has_joins:
896
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
897
        db.log_debug('Getting output pkeys of existing/inserted rows')
898 2387 aaronmk
        run_query_into_pkeys(*mk_select(db, select_joins, pkeys_cols, start=0))
899 2357 aaronmk
    else:
900 2387 aaronmk
        add_row_num(db, out_pkeys) # for joining with input pkeys
901 2357 aaronmk
902
        db.log_debug('Getting input pkeys for rows in insert')
903
        run_query_into(db, *mk_main_select(input_joins, [in_pkey]),
904 2387 aaronmk
            into=in_pkeys)
905
        add_row_num(db, in_pkeys) # for joining with output pkeys
906 2357 aaronmk
907
        db.log_debug('Joining together output and input pkeys')
908 2387 aaronmk
        pkey_joins = [in_pkeys, sql_gen.Join(out_pkeys,
909 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
910 2387 aaronmk
        run_query_into_pkeys(*mk_select(db, pkey_joins, pkeys_names, start=0))
911 2357 aaronmk
912
    db.log_debug("Setting missing rows' pkeys to NULL")
913 2387 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(pkeys,
914 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
915
        # must use join_same_not_null or query will take forever
916
    run_query_into_pkeys(*mk_select(db, missing_rows_joins,
917
        [in_pkey_col, sql_gen.NamedCol(out_pkey, None)], start=0))
918
919 2387 aaronmk
    return sql_gen.Col(out_pkey, pkeys)
920 2115 aaronmk
921
##### Data cleanup
922
923 2290 aaronmk
def cleanup_table(db, table, cols):
924 2115 aaronmk
    def esc_name_(name): return esc_name(db, name)
925
926 2290 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
927 2115 aaronmk
    cols = map(esc_name_, cols)
928
929
    run_query(db, 'UPDATE '+table+' SET\n'+(',\n'.join(('\n'+col
930
        +' = nullif(nullif(trim(both from '+col+"), %(null0)s), %(null1)s)"
931
            for col in cols))),
932
        dict(null0='', null1=r'\N'))