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 2484 aaronmk
        DbException.__init__(self, 'for name: '+strings.as_tt(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 2484 aaronmk
        DbException.__init__(self, 'for name: '+strings.as_tt(str(name))
45
            +'; value: '+strings.as_tt(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 2484 aaronmk
        DbException.__init__(self, 'Violated '+strings.as_tt(name)
52
            +' constraint on columns: '+strings.as_tt(', '.join(cols)), cause)
53 2306 aaronmk
        self.name = name
54 468 aaronmk
        self.cols = cols
55 11 aaronmk
56 2523 aaronmk
class MissingCastException(DbException):
57
    def __init__(self, type_, col, cause=None):
58
        DbException.__init__(self, 'Missing cast to type '+strings.as_tt(type_)
59
            +' on column: '+strings.as_tt(col), cause)
60
        self.type = type_
61
        self.col = col
62
63 2143 aaronmk
class NameException(DbException): pass
64
65 2306 aaronmk
class DuplicateKeyException(ConstraintException): pass
66 13 aaronmk
67 2306 aaronmk
class NullValueException(ConstraintException): pass
68 13 aaronmk
69 2240 aaronmk
class FunctionValueException(ExceptionWithNameValue): pass
70 2239 aaronmk
71 2143 aaronmk
class DuplicateTableException(ExceptionWithName): pass
72
73 2188 aaronmk
class DuplicateFunctionException(ExceptionWithName): pass
74
75 89 aaronmk
class EmptyRowException(DbException): pass
76
77 865 aaronmk
##### Warnings
78
79
class DbWarning(UserWarning): pass
80
81 1930 aaronmk
##### Result retrieval
82
83
def col_names(cur): return (col[0] for col in cur.description)
84
85
def rows(cur): return iter(lambda: cur.fetchone(), None)
86
87
def consume_rows(cur):
88
    '''Used to fetch all rows so result will be cached'''
89
    iters.consume_iter(rows(cur))
90
91
def next_row(cur): return rows(cur).next()
92
93
def row(cur):
94
    row_ = next_row(cur)
95
    consume_rows(cur)
96
    return row_
97
98
def next_value(cur): return next_row(cur)[0]
99
100
def value(cur): return row(cur)[0]
101
102
def values(cur): return iters.func_iter(lambda: next_value(cur))
103
104
def value_or_none(cur):
105
    try: return value(cur)
106
    except StopIteration: return None
107
108 2101 aaronmk
##### Input validation
109
110 2573 aaronmk
def esc_name_by_module(module, name):
111
    if module == 'psycopg2' or module == None: quote = '"'
112 2101 aaronmk
    elif module == 'MySQLdb': quote = '`'
113
    else: raise NotImplementedError("Can't escape name for "+module+' database')
114 2500 aaronmk
    return sql_gen.esc_name(name, quote)
115 2101 aaronmk
116
def esc_name_by_engine(engine, name, **kw_args):
117
    return esc_name_by_module(db_engines[engine][0], name, **kw_args)
118
119
def esc_name(db, name, **kw_args):
120
    return esc_name_by_module(util.root_module(db.db), name, **kw_args)
121
122
def qual_name(db, schema, table):
123
    def esc_name_(name): return esc_name(db, name)
124
    table = esc_name_(table)
125
    if schema != None: return esc_name_(schema)+'.'+table
126
    else: return table
127
128 1869 aaronmk
##### Database connections
129 1849 aaronmk
130 2097 aaronmk
db_config_names = ['engine', 'host', 'user', 'password', 'database', 'schemas']
131 1926 aaronmk
132 1869 aaronmk
db_engines = {
133
    'MySQL': ('MySQLdb', {'password': 'passwd', 'database': 'db'}),
134
    'PostgreSQL': ('psycopg2', {}),
135
}
136
137
DatabaseErrors_set = set([DbException])
138
DatabaseErrors = tuple(DatabaseErrors_set)
139
140
def _add_module(module):
141
    DatabaseErrors_set.add(module.DatabaseError)
142
    global DatabaseErrors
143
    DatabaseErrors = tuple(DatabaseErrors_set)
144
145
def db_config_str(db_config):
146
    return db_config['engine']+' database '+db_config['database']
147
148 1909 aaronmk
def _query_lookup(query, params): return (query, dicts.make_hashable(params))
149 1894 aaronmk
150 2448 aaronmk
log_debug_none = lambda msg, level=2: None
151 1901 aaronmk
152 1849 aaronmk
class DbConn:
153 2190 aaronmk
    def __init__(self, db_config, serializable=True, autocommit=False,
154
        caching=True, log_debug=log_debug_none):
155 1869 aaronmk
        self.db_config = db_config
156
        self.serializable = serializable
157 2190 aaronmk
        self.autocommit = autocommit
158
        self.caching = caching
159 1901 aaronmk
        self.log_debug = log_debug
160 2193 aaronmk
        self.debug = log_debug != log_debug_none
161 1869 aaronmk
162
        self.__db = None
163 1889 aaronmk
        self.query_results = {}
164 2139 aaronmk
        self._savepoint = 0
165 1869 aaronmk
166
    def __getattr__(self, name):
167
        if name == '__dict__': raise Exception('getting __dict__')
168
        if name == 'db': return self._db()
169
        else: raise AttributeError()
170
171
    def __getstate__(self):
172
        state = copy.copy(self.__dict__) # shallow copy
173 1915 aaronmk
        state['log_debug'] = None # don't pickle the debug callback
174 1869 aaronmk
        state['_DbConn__db'] = None # don't pickle the connection
175
        return state
176
177 2165 aaronmk
    def connected(self): return self.__db != None
178
179 1869 aaronmk
    def _db(self):
180
        if self.__db == None:
181
            # Process db_config
182
            db_config = self.db_config.copy() # don't modify input!
183 2097 aaronmk
            schemas = db_config.pop('schemas', None)
184 1869 aaronmk
            module_name, mappings = db_engines[db_config.pop('engine')]
185
            module = __import__(module_name)
186
            _add_module(module)
187
            for orig, new in mappings.iteritems():
188
                try: util.rename_key(db_config, orig, new)
189
                except KeyError: pass
190
191
            # Connect
192
            self.__db = module.connect(**db_config)
193
194
            # Configure connection
195 2234 aaronmk
            if self.serializable and not self.autocommit: run_raw_query(self,
196
                'SET TRANSACTION ISOLATION LEVEL SERIALIZABLE')
197 2101 aaronmk
            if schemas != None:
198
                schemas_ = ''.join((esc_name(self, s)+', '
199
                    for s in schemas.split(',')))
200
                run_raw_query(self, "SELECT set_config('search_path', \
201
%s || current_setting('search_path'), false)", [schemas_])
202 1869 aaronmk
203
        return self.__db
204 1889 aaronmk
205 1891 aaronmk
    class DbCursor(Proxy):
206 1927 aaronmk
        def __init__(self, outer):
207 1891 aaronmk
            Proxy.__init__(self, outer.db.cursor())
208 2191 aaronmk
            self.outer = outer
209 1927 aaronmk
            self.query_results = outer.query_results
210 1894 aaronmk
            self.query_lookup = None
211 1891 aaronmk
            self.result = []
212 1889 aaronmk
213 1894 aaronmk
        def execute(self, query, params=None):
214 1930 aaronmk
            self._is_insert = query.upper().find('INSERT') >= 0
215 1894 aaronmk
            self.query_lookup = _query_lookup(query, params)
216 2148 aaronmk
            try:
217 2191 aaronmk
                try:
218
                    return_value = self.inner.execute(query, params)
219
                    self.outer.do_autocommit()
220 2648 aaronmk
                finally: self.query = get_cur_query(self.inner, query, params)
221 1904 aaronmk
            except Exception, e:
222 2170 aaronmk
                _add_cursor_info(e, self, query, params)
223 1904 aaronmk
                self.result = e # cache the exception as the result
224
                self._cache_result()
225
                raise
226 1930 aaronmk
            # Fetch all rows so result will be cached
227
            if self.rowcount == 0 and not self._is_insert: consume_rows(self)
228 1894 aaronmk
            return return_value
229
230 1891 aaronmk
        def fetchone(self):
231
            row = self.inner.fetchone()
232 1899 aaronmk
            if row != None: self.result.append(row)
233
            # otherwise, fetched all rows
234 1904 aaronmk
            else: self._cache_result()
235
            return row
236
237
        def _cache_result(self):
238 1906 aaronmk
            # For inserts, only cache exceptions since inserts are not
239
            # idempotent, but an invalid insert will always be invalid
240 1930 aaronmk
            if self.query_results != None and (not self._is_insert
241 1906 aaronmk
                or isinstance(self.result, Exception)):
242
243 1894 aaronmk
                assert self.query_lookup != None
244 1916 aaronmk
                self.query_results[self.query_lookup] = self.CacheCursor(
245
                    util.dict_subset(dicts.AttrsDictView(self),
246
                    ['query', 'result', 'rowcount', 'description']))
247 1906 aaronmk
248 1916 aaronmk
        class CacheCursor:
249
            def __init__(self, cached_result): self.__dict__ = cached_result
250
251 1927 aaronmk
            def execute(self, *args, **kw_args):
252 1916 aaronmk
                if isinstance(self.result, Exception): raise self.result
253
                # otherwise, result is a rows list
254
                self.iter = iter(self.result)
255
256
            def fetchone(self):
257
                try: return self.iter.next()
258
                except StopIteration: return None
259 1891 aaronmk
260 2212 aaronmk
    def esc_value(self, value):
261 2663 aaronmk
        try: str_ = self.mogrify('%s', [value])
262
        except NotImplementedError, e:
263
            module = util.root_module(self.db)
264
            if module == 'MySQLdb':
265
                import _mysql
266
                str_ = _mysql.escape_string(value)
267
            else: raise e
268 2374 aaronmk
        return strings.to_unicode(str_)
269 2212 aaronmk
270 2347 aaronmk
    def esc_name(self, name): return esc_name(self, name) # calls global func
271
272 2663 aaronmk
    def mogrify(self, query, params=None):
273
        module = util.root_module(self.db)
274
        if module == 'psycopg2': return self.db.cursor().mogrify(query, params)
275
        else: raise NotImplementedError("Can't mogrify query for "+module
276
            +' database')
277
278 2445 aaronmk
    def run_query(self, query, params=None, cacheable=False, log_level=2,
279 2464 aaronmk
        debug_msg_ref=None):
280 2445 aaronmk
        '''
281 2464 aaronmk
        @param log_ignore_excs The log_level will be increased by 2 if the query
282
            throws one of these exceptions.
283 2664 aaronmk
        @param debug_msg_ref If specified, the log message will be returned in
284
            this instead of being output. This allows you to filter log messages
285
            depending on the result of the query.
286 2445 aaronmk
        '''
287 2167 aaronmk
        assert query != None
288
289 2047 aaronmk
        if not self.caching: cacheable = False
290 1903 aaronmk
        used_cache = False
291 2664 aaronmk
292
        def log_msg(query):
293
            if used_cache: cache_status = 'cache hit'
294
            elif cacheable: cache_status = 'cache miss'
295
            else: cache_status = 'non-cacheable'
296
            return 'DB query: '+cache_status+':\n'+strings.as_code(query, 'SQL')
297
298 1903 aaronmk
        try:
299 1927 aaronmk
            # Get cursor
300
            if cacheable:
301
                query_lookup = _query_lookup(query, params)
302
                try:
303
                    cur = self.query_results[query_lookup]
304
                    used_cache = True
305
                except KeyError: cur = self.DbCursor(self)
306
            else: cur = self.db.cursor()
307
308 2664 aaronmk
            # Log query
309
            if self.debug and debug_msg_ref == None: # log before running
310
                self.log_debug(log_msg(query), log_level)
311
312 1927 aaronmk
            # Run query
313 2148 aaronmk
            cur.execute(query, params)
314 1903 aaronmk
        finally:
315 2664 aaronmk
            if self.debug and debug_msg_ref != None: # return after running
316
                debug_msg_ref[0] = log_msg(str(get_cur_query(cur, query,
317
                    params)))
318 1903 aaronmk
319
        return cur
320 1914 aaronmk
321
    def is_cached(self, query, params=None):
322
        return _query_lookup(query, params) in self.query_results
323 2139 aaronmk
324
    def with_savepoint(self, func):
325 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
326 2443 aaronmk
        self.run_query('SAVEPOINT '+savepoint, log_level=4)
327 2139 aaronmk
        self._savepoint += 1
328
        try:
329
            try: return_val = func()
330
            finally:
331
                self._savepoint -= 1
332
                assert self._savepoint >= 0
333
        except:
334 2443 aaronmk
            self.run_query('ROLLBACK TO SAVEPOINT '+savepoint, log_level=4)
335 2139 aaronmk
            raise
336
        else:
337 2443 aaronmk
            self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
338 2191 aaronmk
            self.do_autocommit()
339 2139 aaronmk
            return return_val
340 2191 aaronmk
341
    def do_autocommit(self):
342
        '''Autocommits if outside savepoint'''
343
        assert self._savepoint >= 0
344
        if self.autocommit and self._savepoint == 0:
345
            self.log_debug('Autocommiting')
346
            self.db.commit()
347 2643 aaronmk
348
    def col_default(self, col):
349
        table = sql_gen.Table('columns', 'information_schema')
350
351
        conds = [('table_name', col.table.name), ('column_name', col.name)]
352
        schema = col.table.schema
353
        if schema != None: conds.append(('table_schema', schema))
354
355 2660 aaronmk
        return sql_gen.as_Code(value(select(self, table, ['column_default'],
356
            conds, order_by='table_schema', limit=1, log_level=3)))
357 2643 aaronmk
            # TODO: order_by search_path schema order
358 1849 aaronmk
359 1869 aaronmk
connect = DbConn
360
361 832 aaronmk
##### Querying
362
363 1894 aaronmk
def run_raw_query(db, *args, **kw_args):
364 2085 aaronmk
    '''For params, see DbConn.run_query()'''
365 1894 aaronmk
    return db.run_query(*args, **kw_args)
366 11 aaronmk
367 2068 aaronmk
def mogrify(db, query, params):
368
    module = util.root_module(db.db)
369
    if module == 'psycopg2': return db.db.cursor().mogrify(query, params)
370
    else: raise NotImplementedError("Can't mogrify query for "+module+
371
        ' database')
372
373 832 aaronmk
##### Recoverable querying
374 15 aaronmk
375 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
376 11 aaronmk
377 2464 aaronmk
def run_query(db, query, params=None, recover=None, cacheable=False,
378
    log_level=2, log_ignore_excs=None, **kw_args):
379 2441 aaronmk
    '''For params, see run_raw_query()'''
380 830 aaronmk
    if recover == None: recover = False
381 2464 aaronmk
    if log_ignore_excs == None: log_ignore_excs = ()
382
    log_ignore_excs = tuple(log_ignore_excs)
383 830 aaronmk
384 2464 aaronmk
    debug_msg_ref = [None]
385 2148 aaronmk
    try:
386 2464 aaronmk
        try:
387
            def run(): return run_raw_query(db, query, params, cacheable,
388
                log_level, debug_msg_ref, **kw_args)
389
            if recover and not db.is_cached(query, params):
390
                return with_savepoint(db, run)
391
            else: return run() # don't need savepoint if cached
392
        except Exception, e:
393
            if not recover: raise # need savepoint to run index_cols()
394
            msg = exc.str_(e)
395
396
            match = re.search(r'duplicate key value violates unique constraint '
397 2493 aaronmk
                r'"((_?[^\W_]+)_.+?)"', msg)
398 2464 aaronmk
            if match:
399
                constraint, table = match.groups()
400
                try: cols = index_cols(db, table, constraint)
401
                except NotImplementedError: raise e
402
                else: raise DuplicateKeyException(constraint, cols, e)
403
404 2493 aaronmk
            match = re.search(r'null value in column "(.+?)" violates not-null'
405 2464 aaronmk
                r' constraint', msg)
406
            if match: raise NullValueException('NOT NULL', [match.group(1)], e)
407
408
            match = re.search(r'\b(?:invalid input (?:syntax|value)\b.*?'
409
                r'|date/time field value out of range): "(.+?)"\n'
410 2535 aaronmk
                r'(?:(?s).*?)\bfunction "(.+?)"', msg)
411 2464 aaronmk
            if match:
412
                value, name = match.groups()
413
                raise FunctionValueException(name, strings.to_unicode(value), e)
414
415 2526 aaronmk
            match = re.search(r'column "(.+?)" is of type (.+?) but expression '
416 2523 aaronmk
                r'is of type', msg)
417
            if match:
418
                col, type_ = match.groups()
419
                raise MissingCastException(type_, col, e)
420
421 2493 aaronmk
            match = re.search(r'relation "(.+?)" already exists', msg)
422 2464 aaronmk
            if match: raise DuplicateTableException(match.group(1), e)
423
424 2493 aaronmk
            match = re.search(r'function "(.+?)" already exists', msg)
425 2464 aaronmk
            if match: raise DuplicateFunctionException(match.group(1), e)
426
427
            raise # no specific exception raised
428
    except log_ignore_excs:
429
        log_level += 2
430
        raise
431
    finally:
432
        if debug_msg_ref[0] != None: db.log_debug(debug_msg_ref[0], log_level)
433 830 aaronmk
434 832 aaronmk
##### Basic queries
435
436 2153 aaronmk
def next_version(name):
437 2163 aaronmk
    version = 1 # first existing name was version 0
438 2586 aaronmk
    match = re.match(r'^(.*)#(\d+)$', name)
439 2153 aaronmk
    if match:
440 2586 aaronmk
        name, version = match.groups()
441
        version = int(version)+1
442 2588 aaronmk
    return sql_gen.add_suffix(name, '#'+str(version))
443 2153 aaronmk
444 2386 aaronmk
def run_query_into(db, query, params, into=None, *args, **kw_args):
445 2085 aaronmk
    '''Outputs a query to a temp table.
446
    For params, see run_query().
447
    '''
448 2386 aaronmk
    if into == None: return run_query(db, query, params, *args, **kw_args)
449 2085 aaronmk
    else: # place rows in temp table
450 2386 aaronmk
        assert isinstance(into, sql_gen.Table)
451 2385 aaronmk
452 2153 aaronmk
        kw_args['recover'] = True
453 2464 aaronmk
        kw_args.setdefault('log_ignore_excs', (DuplicateTableException,))
454 2440 aaronmk
455 2468 aaronmk
        temp = not db.autocommit # tables are permanent in autocommit mode
456 2440 aaronmk
        # "temporary tables cannot specify a schema name", so remove schema
457
        if temp: into.schema = None
458
459 2153 aaronmk
        while True:
460
            try:
461 2194 aaronmk
                create_query = 'CREATE'
462 2440 aaronmk
                if temp: create_query += ' TEMP'
463 2467 aaronmk
                create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
464 2194 aaronmk
465
                return run_query(db, create_query, params, *args, **kw_args)
466 2153 aaronmk
                    # CREATE TABLE AS sets rowcount to # rows in query
467
            except DuplicateTableException, e:
468 2386 aaronmk
                into.name = next_version(into.name)
469 2153 aaronmk
                # try again with next version of name
470 2085 aaronmk
471 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
472
473 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
474
475 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
476 2293 aaronmk
    start=None, order_by=order_by_pkey, default_table=None):
477 1981 aaronmk
    '''
478 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
479 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
480 1981 aaronmk
    @param fields Use None to select all fields in the table
481 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
482 2379 aaronmk
        * container can be any iterable type
483 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
484
        * compare_right_side: sql_gen.ValueCond|literal value
485 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
486
        use all columns
487 2054 aaronmk
    @return tuple(query, params)
488 1981 aaronmk
    '''
489 2315 aaronmk
    # Parse tables param
490 2121 aaronmk
    if not lists.is_seq(tables): tables = [tables]
491 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
492 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
493 2121 aaronmk
494 2315 aaronmk
    # Parse other params
495 2376 aaronmk
    if conds == None: conds = []
496 2650 aaronmk
    elif dicts.is_dict(conds): conds = conds.items()
497 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
498 135 aaronmk
    assert limit == None or type(limit) == int
499 865 aaronmk
    assert start == None or type(start) == int
500 2315 aaronmk
    if order_by is order_by_pkey:
501
        if distinct_on != []: order_by = None
502
        else: order_by = pkey(db, table0, recover=True)
503 865 aaronmk
504 2315 aaronmk
    query = 'SELECT'
505 2056 aaronmk
506 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
507 2056 aaronmk
508 2200 aaronmk
    # DISTINCT ON columns
509 2233 aaronmk
    if distinct_on != []:
510 2467 aaronmk
        query += '\nDISTINCT'
511 2254 aaronmk
        if distinct_on is not distinct_on_all:
512 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
513
514
    # Columns
515 2467 aaronmk
    query += '\n'
516 1135 aaronmk
    if fields == None: query += '*'
517 2479 aaronmk
    else: query += '\n, '.join(map(parse_col, fields))
518 2200 aaronmk
519
    # Main table
520 2467 aaronmk
    query += '\nFROM '+table0.to_str(db)
521 865 aaronmk
522 2122 aaronmk
    # Add joins
523 2271 aaronmk
    left_table = table0
524 2263 aaronmk
    for join_ in tables:
525
        table = join_.table
526 2238 aaronmk
527 2343 aaronmk
        # Parse special values
528
        if join_.type_ is sql_gen.filter_out: # filter no match
529 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
530
                None))
531 2343 aaronmk
532 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
533 2122 aaronmk
534
        left_table = table
535
536 865 aaronmk
    missing = True
537 2376 aaronmk
    if conds != []:
538 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
539
        else: whitespace = '\n'
540 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
541
            .to_str(db) for l, r in conds], 'WHERE')
542 865 aaronmk
        missing = False
543 2227 aaronmk
    if order_by != None:
544 2467 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
545
    if limit != None: query += '\nLIMIT '+str(limit); missing = False
546 865 aaronmk
    if start != None:
547 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
548 865 aaronmk
        missing = False
549
    if missing: warnings.warn(DbWarning(
550
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
551
552 2315 aaronmk
    return (query, [])
553 11 aaronmk
554 2054 aaronmk
def select(db, *args, **kw_args):
555
    '''For params, see mk_select() and run_query()'''
556
    recover = kw_args.pop('recover', None)
557
    cacheable = kw_args.pop('cacheable', True)
558 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
559 2054 aaronmk
560
    query, params = mk_select(db, *args, **kw_args)
561 2442 aaronmk
    return run_query(db, query, params, recover, cacheable, log_level=log_level)
562 2054 aaronmk
563 2066 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, params=None,
564 2292 aaronmk
    returning=None, embeddable=False):
565 1960 aaronmk
    '''
566
    @param returning str|None An inserted column (such as pkey) to return
567 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
568 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
569
        query will be fully cached, not just if it raises an exception.
570 1960 aaronmk
    '''
571 2328 aaronmk
    table = sql_gen.as_Table(table)
572 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
573 2646 aaronmk
    if cols != None:
574
        cols = [sql_gen.to_name_only_col(v, table).to_str(db) for v in cols]
575 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
576 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
577 2063 aaronmk
578
    # Build query
579 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
580
    query = first_line
581 2467 aaronmk
    if cols != None: query += '\n('+', '.join(cols)+')'
582
    query += '\n'+select_query
583 2063 aaronmk
584
    if returning != None:
585 2327 aaronmk
        returning_name = copy.copy(returning)
586
        returning_name.table = None
587
        returning_name = returning_name.to_str(db)
588 2467 aaronmk
        query += '\nRETURNING '+returning_name
589 2063 aaronmk
590 2070 aaronmk
    if embeddable:
591 2327 aaronmk
        assert returning != None
592
593 2070 aaronmk
        # Create function
594 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
595 2327 aaronmk
        return_type = 'SETOF '+returning.to_str(db)+'%TYPE'
596 2189 aaronmk
        while True:
597
            try:
598 2327 aaronmk
                func_schema = None
599 2468 aaronmk
                if not db.autocommit: func_schema = 'pg_temp'
600 2327 aaronmk
                function = sql_gen.Table(function_name, func_schema).to_str(db)
601 2194 aaronmk
602 2189 aaronmk
                function_query = '''\
603 2467 aaronmk
CREATE FUNCTION '''+function+'''()
604
RETURNS '''+return_type+'''
605
LANGUAGE sql
606
AS $$
607
'''+mogrify(db, query, params)+''';
608
$$;
609 2070 aaronmk
'''
610 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
611 2464 aaronmk
                    log_ignore_excs=(DuplicateFunctionException,))
612 2189 aaronmk
                break # this version was successful
613
            except DuplicateFunctionException, e:
614
                function_name = next_version(function_name)
615
                # try again with next version of name
616 2070 aaronmk
617 2337 aaronmk
        # Return query that uses function
618
        func_table = sql_gen.NamedTable('f', sql_gen.CustomCode(function+'()'),
619
            [returning_name]) # AS clause requires function alias
620
        return mk_select(db, func_table, start=0, order_by=None)
621 2070 aaronmk
622 2066 aaronmk
    return (query, params)
623
624
def insert_select(db, *args, **kw_args):
625 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
626 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
627
        values in
628 2072 aaronmk
    '''
629 2386 aaronmk
    into = kw_args.pop('into', None)
630
    if into != None: kw_args['embeddable'] = True
631 2066 aaronmk
    recover = kw_args.pop('recover', None)
632
    cacheable = kw_args.pop('cacheable', True)
633
634
    query, params = mk_insert_select(db, *args, **kw_args)
635 2386 aaronmk
    return run_query_into(db, query, params, into, recover=recover,
636 2153 aaronmk
        cacheable=cacheable)
637 2063 aaronmk
638 2066 aaronmk
default = object() # tells insert() to use the default value for a column
639
640 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
641 2085 aaronmk
    '''For params, see insert_select()'''
642 1960 aaronmk
    if lists.is_seq(row): cols = None
643
    else:
644
        cols = row.keys()
645
        row = row.values()
646
    row = list(row) # ensure that "!= []" works
647
648 1961 aaronmk
    # Check for special values
649
    labels = []
650
    values = []
651
    for value in row:
652 2647 aaronmk
        value = sql_gen.remove_col_rename(sql_gen.as_Value(value)).value
653 2254 aaronmk
        if value is default: labels.append('DEFAULT')
654 1961 aaronmk
        else:
655
            labels.append('%s')
656
            values.append(value)
657
658
    # Build query
659 2467 aaronmk
    if values != []: query = 'VALUES ('+(', '.join(labels))+')'
660 2063 aaronmk
    else: query = None
661 1554 aaronmk
662 2064 aaronmk
    return insert_select(db, table, cols, query, values, *args, **kw_args)
663 11 aaronmk
664 2402 aaronmk
def mk_update(db, table, changes=None, cond=None):
665
    '''
666
    @param changes [(col, new_value),...]
667
        * container can be any iterable type
668
        * col: sql_gen.Code|str (for col name)
669
        * new_value: sql_gen.Code|literal value
670
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
671
    @return str query
672
    '''
673
    query = 'UPDATE '+sql_gen.as_Table(table).to_str(db)+'\nSET\n'
674 2405 aaronmk
    query += ',\n'.join((sql_gen.to_name_only_col(col, table).to_str(db)+' = '
675 2402 aaronmk
        +sql_gen.as_Value(new_value).to_str(db) for col, new_value in changes))
676 2467 aaronmk
    if cond != None: query += '\nWHERE\n'+cond.to_str(db)
677 2402 aaronmk
678
    return query
679
680
def update(db, *args, **kw_args):
681
    '''For params, see mk_update() and run_query()'''
682
    recover = kw_args.pop('recover', None)
683
684
    return run_query(db, mk_update(db, *args, **kw_args), [], recover)
685
686 135 aaronmk
def last_insert_id(db):
687 1849 aaronmk
    module = util.root_module(db.db)
688 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
689
    elif module == 'MySQLdb': return db.insert_id()
690
    else: return None
691 13 aaronmk
692 1968 aaronmk
def truncate(db, table, schema='public'):
693
    return run_query(db, 'TRUNCATE '+qual_name(db, schema, table)+' CASCADE')
694 832 aaronmk
695 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
696 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
697 2415 aaronmk
    to names that will be distinct among the columns' tables.
698 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
699 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
700
    @param into The table for the new columns.
701 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
702
        columns will be included in the mapping even if they are not in cols.
703
        The tables of the provided Col objects will be changed to into, so make
704
        copies of them if you want to keep the original tables.
705
    @param as_items Whether to return a list of dict items instead of a dict
706 2383 aaronmk
    @return dict(orig_col=new_col, ...)
707
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
708 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
709
        * All mappings use the into table so its name can easily be
710 2383 aaronmk
          changed for all columns at once
711
    '''
712 2415 aaronmk
    cols = lists.uniqify(cols)
713
714 2394 aaronmk
    items = []
715 2389 aaronmk
    for col in preserve:
716 2390 aaronmk
        orig_col = copy.copy(col)
717 2392 aaronmk
        col.table = into
718 2394 aaronmk
        items.append((orig_col, col))
719
    preserve = set(preserve)
720
    for col in cols:
721 2515 aaronmk
        if col not in preserve: items.append((col, sql_gen.Col(str(col), into)))
722 2394 aaronmk
723
    if not as_items: items = dict(items)
724
    return items
725 2383 aaronmk
726 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
727 2391 aaronmk
    '''For params, see mk_flatten_mapping()
728
    @return See return value of mk_flatten_mapping()
729
    '''
730 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
731
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
732 2391 aaronmk
    run_query_into(db, *mk_select(db, joins, cols, limit=limit, start=start),
733 2392 aaronmk
        into=into)
734 2394 aaronmk
    return dict(items)
735 2391 aaronmk
736 2414 aaronmk
##### Database structure queries
737
738 2426 aaronmk
def table_row_count(db, table, recover=None):
739
    return value(run_query(db, *mk_select(db, table, [sql_gen.row_count],
740 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
741 2426 aaronmk
742 2414 aaronmk
def table_cols(db, table, recover=None):
743
    return list(col_names(select(db, table, limit=0, order_by=None,
744 2443 aaronmk
        recover=recover, log_level=4)))
745 2414 aaronmk
746 2291 aaronmk
def pkey(db, table, recover=None):
747 832 aaronmk
    '''Assumed to be first column in table'''
748 2339 aaronmk
    return table_cols(db, table, recover)[0]
749 832 aaronmk
750 2559 aaronmk
not_null_col = 'not_null_col'
751 2340 aaronmk
752
def table_not_null_col(db, table, recover=None):
753
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
754
    if not_null_col in table_cols(db, table, recover): return not_null_col
755
    else: return pkey(db, table, recover)
756
757 853 aaronmk
def index_cols(db, table, index):
758
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
759
    automatically created. When you don't know whether something is a UNIQUE
760
    constraint or a UNIQUE index, use this function.'''
761 1909 aaronmk
    module = util.root_module(db.db)
762
    if module == 'psycopg2':
763
        return list(values(run_query(db, '''\
764 853 aaronmk
SELECT attname
765 866 aaronmk
FROM
766
(
767
        SELECT attnum, attname
768
        FROM pg_index
769
        JOIN pg_class index ON index.oid = indexrelid
770
        JOIN pg_class table_ ON table_.oid = indrelid
771
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
772
        WHERE
773
            table_.relname = %(table)s
774
            AND index.relname = %(index)s
775
    UNION
776
        SELECT attnum, attname
777
        FROM
778
        (
779
            SELECT
780
                indrelid
781
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
782
                    AS indkey
783
            FROM pg_index
784
            JOIN pg_class index ON index.oid = indexrelid
785
            JOIN pg_class table_ ON table_.oid = indrelid
786
            WHERE
787
                table_.relname = %(table)s
788
                AND index.relname = %(index)s
789
        ) s
790
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
791
) s
792 853 aaronmk
ORDER BY attnum
793
''',
794 2443 aaronmk
            {'table': table, 'index': index}, cacheable=True, log_level=4)))
795 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
796
        ' database')
797 853 aaronmk
798 464 aaronmk
def constraint_cols(db, table, constraint):
799 1849 aaronmk
    module = util.root_module(db.db)
800 464 aaronmk
    if module == 'psycopg2':
801
        return list(values(run_query(db, '''\
802
SELECT attname
803
FROM pg_constraint
804
JOIN pg_class ON pg_class.oid = conrelid
805
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
806
WHERE
807
    relname = %(table)s
808
    AND conname = %(constraint)s
809
ORDER BY attnum
810
''',
811
            {'table': table, 'constraint': constraint})))
812
    else: raise NotImplementedError("Can't list constraint columns for "+module+
813
        ' database')
814
815 2096 aaronmk
row_num_col = '_row_num'
816
817 2538 aaronmk
def add_index(db, expr):
818
    '''Adds an index on a column or expression if it doesn't already exist.
819
    Currently, only function calls are supported as expressions.
820
    '''
821
    expr = copy.copy(expr) # don't modify input!
822
823
    # Extract col
824 2539 aaronmk
    if isinstance(expr, sql_gen.FunctionCall):
825
        col = expr.args[0]
826 2541 aaronmk
        expr = sql_gen.Expr(expr)
827 2538 aaronmk
    else: col = expr
828 2408 aaronmk
    assert sql_gen.is_table_col(col)
829
830 2538 aaronmk
    index = sql_gen.as_Table(str(expr))
831 2408 aaronmk
    table = col.table
832 2538 aaronmk
    col.table = None
833 2408 aaronmk
    try: run_query(db, 'CREATE INDEX '+index.to_str(db)+' ON '+table.to_str(db)
834 2538 aaronmk
        +' ('+expr.to_str(db)+')', recover=True, cacheable=True, log_level=3)
835 2408 aaronmk
    except DuplicateTableException: pass # index already existed
836
837 2594 aaronmk
def add_pkey(db, table, recover=None):
838 2406 aaronmk
    '''Makes the first column in a table the primary key.
839
    @pre The table must not already have a primary key.
840
    '''
841
    table = sql_gen.as_Table(table)
842
843 2590 aaronmk
    index = sql_gen.as_Table(sql_gen.add_suffix(table.name, '_pkey'))
844 2406 aaronmk
    col = sql_gen.to_name_only_col(pkey(db, table, recover))
845 2614 aaronmk
    try:
846
        run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD CONSTRAINT '
847
            +index.to_str(db)+' PRIMARY KEY('+col.to_str(db)+')',
848
            recover=True, cacheable=True, log_level=3,
849
            log_ignore_excs=(DuplicateTableException,))
850
    except DuplicateTableException, e:
851
        index.name = next_version(index.name)
852
        # try again with next version of name
853 2406 aaronmk
854 2086 aaronmk
def add_row_num(db, table):
855 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
856
    be the primary key.'''
857 2320 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
858 2096 aaronmk
    run_query(db, 'ALTER TABLE '+table+' ADD COLUMN '+row_num_col
859 2443 aaronmk
        +' serial NOT NULL PRIMARY KEY', log_level=3)
860 2086 aaronmk
861 2548 aaronmk
def tables(db, schema_like='public', table_like='%'):
862 1849 aaronmk
    module = util.root_module(db.db)
863 2548 aaronmk
    params = {'schema_like': schema_like, 'table_like': table_like}
864 832 aaronmk
    if module == 'psycopg2':
865 1968 aaronmk
        return values(run_query(db, '''\
866
SELECT tablename
867
FROM pg_tables
868
WHERE
869 2548 aaronmk
    schemaname LIKE %(schema_like)s
870 1968 aaronmk
    AND tablename LIKE %(table_like)s
871
ORDER BY tablename
872
''',
873
            params, cacheable=True))
874
    elif module == 'MySQLdb':
875
        return values(run_query(db, 'SHOW TABLES LIKE %(table_like)s', params,
876
            cacheable=True))
877 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
878 830 aaronmk
879 833 aaronmk
##### Database management
880
881 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
882
    '''For kw_args, see tables()'''
883
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
884 833 aaronmk
885 832 aaronmk
##### Heuristic queries
886
887 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
888 1554 aaronmk
    '''Recovers from errors.
889 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
890
    '''
891 2642 aaronmk
    row = sql_gen.ColDict(db, table, row)
892 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
893
894 471 aaronmk
    try:
895 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
896 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
897
            row_ct_ref[0] += cur.rowcount
898
        return value(cur)
899 471 aaronmk
    except DuplicateKeyException, e:
900 2653 aaronmk
        row = sql_gen.ColDict(db, table,
901
            util.dict_subset_right_join(row, e.cols))
902
        return value(select(db, table, [pkey_], row, recover=True))
903 471 aaronmk
904 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
905 830 aaronmk
    '''Recovers from errors'''
906 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
907 14 aaronmk
    except StopIteration:
908 40 aaronmk
        if not create: raise
909 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
910 2078 aaronmk
911 2593 aaronmk
def is_func_result(col):
912
    return col.table.name.find('(') >= 0 and col.name == 'result'
913
914 2592 aaronmk
def into_table_name(out_table, in_tables0, mapping, is_func):
915 2615 aaronmk
    def in_col_str(in_col):
916
        in_col = sql_gen.remove_col_rename(in_col)
917
        if isinstance(in_col, sql_gen.Col):
918
            table = in_col.table
919
            if table == in_tables0:
920
                in_col = sql_gen.to_name_only_col(in_col)
921
            elif is_func_result(in_col): in_col = table # omit col name
922
        return str(in_col)
923
924 2580 aaronmk
    str_ = str(out_table)
925
    if is_func:
926 2615 aaronmk
        str_ += '('
927 2580 aaronmk
928 2615 aaronmk
        try: value_in_col = mapping['value']
929
        except KeyError:
930 2654 aaronmk
            str_ += ', '.join((str(k)+'='+in_col_str(v)
931 2615 aaronmk
                for k, v in mapping.iteritems()))
932
        else: str_ += in_col_str(value_in_col)
933
934
        str_ += ')'
935 2580 aaronmk
    else: str_ += '_pkeys'
936
    return str_
937
938 2508 aaronmk
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None, into=None,
939 2552 aaronmk
    default=None, is_func=False):
940 2078 aaronmk
    '''Recovers from errors.
941
    Only works under PostgreSQL (uses INSERT RETURNING).
942 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
943
        tables to join with it using the main input table's pkey
944 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
945 2612 aaronmk
        * out_table_col: str (*not* sql_gen.Col)
946
        * in_table_col: sql_gen.Col|literal-value
947 2489 aaronmk
    @param into The table to contain the output and input pkeys.
948 2574 aaronmk
        Defaults to `out_table.name+'_pkeys'`.
949 2509 aaronmk
    @param default The *output* column to use as the pkey for missing rows.
950
        If this output column does not exist in the mapping, uses None.
951 2552 aaronmk
    @param is_func Whether out_table is the name of a SQL function, not a table
952 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
953 2078 aaronmk
    '''
954 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
955 2312 aaronmk
956 2450 aaronmk
    def log_debug(msg): db.log_debug(msg, level=1.5)
957 2505 aaronmk
    def col_ustr(str_):
958 2567 aaronmk
        return strings.repr_no_u(sql_gen.remove_col_rename(str_))
959 2450 aaronmk
960 2486 aaronmk
    log_debug('********** New iteration **********')
961 2505 aaronmk
    log_debug('Inserting these input columns into '+strings.as_tt(
962
        out_table.to_str(db))+':\n'+strings.as_table(mapping, ustr=col_ustr))
963 2463 aaronmk
964 2382 aaronmk
    # Create input joins from list of input tables
965
    in_tables_ = in_tables[:] # don't modify input!
966
    in_tables0 = in_tables_.pop(0) # first table is separate
967 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
968 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
969 2460 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v,
970
        {in_pkey: sql_gen.join_same_not_null}) for v in in_tables_]
971 2131 aaronmk
972 2592 aaronmk
    if into == None:
973
        into = into_table_name(out_table, in_tables0, mapping, is_func)
974
    into = sql_gen.as_Table(into)
975
976 2486 aaronmk
    log_debug('Joining together input tables into temp table')
977 2395 aaronmk
    # Place in new table for speed and so don't modify input if values edited
978 2584 aaronmk
    in_table = sql_gen.Table('in')
979 2395 aaronmk
    flatten_cols = filter(sql_gen.is_table_col, mapping.values())
980
    mapping = dicts.join(mapping, flatten(db, in_table, input_joins,
981
        flatten_cols, preserve=[in_pkey_col], start=0))
982
    input_joins = [in_table]
983 2486 aaronmk
    db.log_debug('Temp table: '+strings.as_tt(in_table.to_str(db)), level=2)
984 2395 aaronmk
985 2656 aaronmk
    mapping = sql_gen.ColDict(db, out_table, mapping)
986
        # after applying dicts.join() because that returns a plain dict
987
988 2509 aaronmk
    # Resolve default value column
989
    try: default = mapping[default]
990
    except KeyError:
991
        if default != None:
992
            db.log_debug('Default value column '
993
                +strings.as_tt(strings.repr_no_u(default))
994 2511 aaronmk
                +' does not exist in mapping, falling back to None', level=2.1)
995 2509 aaronmk
            default = None
996
997 2279 aaronmk
    out_pkey = pkey(db, out_table, recover=True)
998 2285 aaronmk
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
999 2142 aaronmk
1000 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
1001 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
1002
1003 2201 aaronmk
    pkeys_table_exists_ref = [False]
1004 2420 aaronmk
    def insert_into_pkeys(joins, cols):
1005
        query, params = mk_select(db, joins, cols, order_by=None, start=0)
1006 2201 aaronmk
        if pkeys_table_exists_ref[0]:
1007 2489 aaronmk
            insert_select(db, into, pkeys_names, query, params)
1008 2201 aaronmk
        else:
1009 2489 aaronmk
            run_query_into(db, query, params, into=into)
1010 2201 aaronmk
            pkeys_table_exists_ref[0] = True
1011
1012 2429 aaronmk
    limit_ref = [None]
1013 2380 aaronmk
    conds = set()
1014 2233 aaronmk
    distinct_on = []
1015 2325 aaronmk
    def mk_main_select(joins, cols):
1016 2429 aaronmk
        return mk_select(db, joins, cols, conds, distinct_on,
1017
            limit=limit_ref[0], start=0)
1018 2132 aaronmk
1019 2519 aaronmk
    exc_strs = set()
1020 2309 aaronmk
    def log_exc(e):
1021 2519 aaronmk
        e_str = exc.str_(e, first_line_only=True)
1022
        log_debug('Caught exception: '+e_str)
1023
        assert e_str not in exc_strs # avoid infinite loops
1024
        exc_strs.add(e_str)
1025 2451 aaronmk
    def remove_all_rows():
1026 2450 aaronmk
        log_debug('Returning NULL for all rows')
1027 2429 aaronmk
        limit_ref[0] = 0 # just create an empty pkeys table
1028 2409 aaronmk
    def ignore(in_col, value):
1029 2545 aaronmk
        in_col_str = strings.as_tt(repr(in_col))
1030 2544 aaronmk
        db.log_debug('Adding index on '+in_col_str+' to enable fast filtering',
1031
            level=2.5)
1032 2537 aaronmk
        add_index(db, in_col)
1033 2545 aaronmk
        log_debug('Ignoring rows with '+in_col_str+' = '
1034
            +strings.as_tt(repr(value)))
1035 2403 aaronmk
    def remove_rows(in_col, value):
1036 2409 aaronmk
        ignore(in_col, value)
1037 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
1038
        assert cond not in conds # avoid infinite loops
1039 2380 aaronmk
        conds.add(cond)
1040 2403 aaronmk
    def invalid2null(in_col, value):
1041 2409 aaronmk
        ignore(in_col, value)
1042 2403 aaronmk
        update(db, in_table, [(in_col, None)],
1043
            sql_gen.ColValueCond(in_col, value))
1044 2245 aaronmk
1045 2589 aaronmk
    def insert_pkeys_table(which):
1046
        return sql_gen.Table(sql_gen.add_suffix(in_table.name,
1047
            '_insert_'+which+'_pkeys'))
1048
    insert_out_pkeys = insert_pkeys_table('out')
1049
    insert_in_pkeys = insert_pkeys_table('in')
1050
1051 2206 aaronmk
    # Do inserts and selects
1052 2642 aaronmk
    join_cols = sql_gen.ColDict(db, out_table)
1053 2206 aaronmk
    while True:
1054 2521 aaronmk
        if limit_ref[0] == 0: # special case
1055
            log_debug('Creating an empty pkeys table')
1056
            cur = run_query_into(db, *mk_select(db, out_table, [out_pkey],
1057
                limit=limit_ref[0]), into=insert_out_pkeys)
1058
            break # don't do main case
1059
1060 2303 aaronmk
        has_joins = join_cols != {}
1061
1062 2305 aaronmk
        # Prepare to insert new rows
1063 2325 aaronmk
        insert_joins = input_joins[:] # don't modify original!
1064 2403 aaronmk
        insert_args = dict(recover=True, cacheable=False)
1065 2303 aaronmk
        if has_joins:
1066 2317 aaronmk
            distinct_on = [v.to_Col() for v in join_cols.values()]
1067 2325 aaronmk
            insert_joins.append(sql_gen.Join(out_table, join_cols,
1068
                sql_gen.filter_out))
1069
        else:
1070 2404 aaronmk
            insert_args.update(dict(returning=out_pkey, into=insert_out_pkeys))
1071 2520 aaronmk
        main_select = mk_main_select(insert_joins, mapping.values())[0]
1072 2303 aaronmk
1073 2486 aaronmk
        log_debug('Trying to insert new rows')
1074 2206 aaronmk
        try:
1075 2518 aaronmk
            cur = insert_select(db, out_table, mapping.keys(), main_select,
1076
                **insert_args)
1077 2357 aaronmk
            break # insert successful
1078 2206 aaronmk
        except DuplicateKeyException, e:
1079 2309 aaronmk
            log_exc(e)
1080
1081 2258 aaronmk
            old_join_cols = join_cols.copy()
1082 2565 aaronmk
            join_cols.update(util.dict_subset_right_join(mapping, e.cols))
1083 2486 aaronmk
            log_debug('Ignoring existing rows, comparing on these columns:\n'
1084 2505 aaronmk
                +strings.as_inline_table(join_cols, ustr=col_ustr))
1085 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
1086 2230 aaronmk
        except NullValueException, e:
1087 2309 aaronmk
            log_exc(e)
1088
1089 2230 aaronmk
            out_col, = e.cols
1090
            try: in_col = mapping[out_col]
1091 2356 aaronmk
            except KeyError:
1092 2486 aaronmk
                log_debug('Missing mapping for NOT NULL column '+out_col)
1093 2451 aaronmk
                remove_all_rows()
1094 2403 aaronmk
            else: remove_rows(in_col, None)
1095 2542 aaronmk
        except FunctionValueException, e:
1096
            log_exc(e)
1097
1098
            func_name = e.name
1099
            value = e.value
1100
            for out_col, in_col in mapping.iteritems():
1101 2562 aaronmk
                invalid2null(sql_gen.unwrap_func_call(in_col, func_name), value)
1102 2525 aaronmk
        except MissingCastException, e:
1103
            log_exc(e)
1104
1105
            out_col = e.col
1106 2534 aaronmk
            mapping[out_col] = sql_gen.wrap_in_func(e.type, mapping[out_col])
1107 2429 aaronmk
        except DatabaseErrors, e:
1108
            log_exc(e)
1109
1110 2531 aaronmk
            msg = 'No handler for exception: '+exc.str_(e)
1111 2451 aaronmk
            warnings.warn(DbWarning(msg))
1112
            log_debug(msg)
1113
            remove_all_rows()
1114 2358 aaronmk
        # after exception handled, rerun loop with additional constraints
1115 2132 aaronmk
1116 2357 aaronmk
    if row_ct_ref != None and cur.rowcount >= 0:
1117
        row_ct_ref[0] += cur.rowcount
1118
1119
    if has_joins:
1120
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
1121 2486 aaronmk
        log_debug('Getting output table pkeys of existing/inserted rows')
1122 2420 aaronmk
        insert_into_pkeys(select_joins, pkeys_cols)
1123 2357 aaronmk
    else:
1124 2404 aaronmk
        add_row_num(db, insert_out_pkeys) # for joining with input pkeys
1125 2357 aaronmk
1126 2486 aaronmk
        log_debug('Getting input table pkeys of inserted rows')
1127 2357 aaronmk
        run_query_into(db, *mk_main_select(input_joins, [in_pkey]),
1128 2404 aaronmk
            into=insert_in_pkeys)
1129
        add_row_num(db, insert_in_pkeys) # for joining with output pkeys
1130 2357 aaronmk
1131 2428 aaronmk
        assert table_row_count(db, insert_out_pkeys) == table_row_count(db,
1132
            insert_in_pkeys)
1133
1134 2486 aaronmk
        log_debug('Combining output and input pkeys in inserted order')
1135 2404 aaronmk
        pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
1136 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
1137 2420 aaronmk
        insert_into_pkeys(pkey_joins, pkeys_names)
1138 2357 aaronmk
1139 2486 aaronmk
    db.log_debug('Adding pkey on pkeys table to enable fast joins', level=2.5)
1140 2594 aaronmk
    add_pkey(db, into)
1141 2407 aaronmk
1142 2508 aaronmk
    log_debug('Setting pkeys of missing rows to '+strings.as_tt(repr(default)))
1143 2489 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(into,
1144 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
1145
        # must use join_same_not_null or query will take forever
1146 2420 aaronmk
    insert_into_pkeys(missing_rows_joins,
1147 2508 aaronmk
        [in_pkey_col, sql_gen.NamedCol(out_pkey, default)])
1148 2357 aaronmk
1149 2489 aaronmk
    assert table_row_count(db, into) == table_row_count(db, in_table)
1150 2428 aaronmk
1151 2489 aaronmk
    return sql_gen.Col(out_pkey, into)
1152 2115 aaronmk
1153
##### Data cleanup
1154
1155 2290 aaronmk
def cleanup_table(db, table, cols):
1156 2115 aaronmk
    def esc_name_(name): return esc_name(db, name)
1157
1158 2290 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
1159 2115 aaronmk
    cols = map(esc_name_, cols)
1160
1161
    run_query(db, 'UPDATE '+table+' SET\n'+(',\n'.join(('\n'+col
1162
        +' = nullif(nullif(trim(both from '+col+"), %(null0)s), %(null1)s)"
1163
            for col in cols))),
1164
        dict(null0='', null1=r'\N'))