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 2665 aaronmk
    def can_mogrify(self):
273 2663 aaronmk
        module = util.root_module(self.db)
274 2665 aaronmk
        return module == 'psycopg2'
275 2663 aaronmk
276 2665 aaronmk
    def mogrify(self, query, params=None):
277
        if self.can_mogrify(): return self.db.cursor().mogrify(query, params)
278
        else: raise NotImplementedError("Can't mogrify query")
279
280 2445 aaronmk
    def run_query(self, query, params=None, cacheable=False, log_level=2,
281 2464 aaronmk
        debug_msg_ref=None):
282 2445 aaronmk
        '''
283 2464 aaronmk
        @param log_ignore_excs The log_level will be increased by 2 if the query
284
            throws one of these exceptions.
285 2664 aaronmk
        @param debug_msg_ref If specified, the log message will be returned in
286
            this instead of being output. This allows you to filter log messages
287
            depending on the result of the query.
288 2445 aaronmk
        '''
289 2167 aaronmk
        assert query != None
290
291 2047 aaronmk
        if not self.caching: cacheable = False
292 1903 aaronmk
        used_cache = False
293 2664 aaronmk
294
        def log_msg(query):
295
            if used_cache: cache_status = 'cache hit'
296
            elif cacheable: cache_status = 'cache miss'
297
            else: cache_status = 'non-cacheable'
298
            return 'DB query: '+cache_status+':\n'+strings.as_code(query, 'SQL')
299
300 1903 aaronmk
        try:
301 1927 aaronmk
            # Get cursor
302
            if cacheable:
303
                query_lookup = _query_lookup(query, params)
304
                try:
305
                    cur = self.query_results[query_lookup]
306
                    used_cache = True
307
                except KeyError: cur = self.DbCursor(self)
308
            else: cur = self.db.cursor()
309
310 2664 aaronmk
            # Log query
311
            if self.debug and debug_msg_ref == None: # log before running
312
                self.log_debug(log_msg(query), log_level)
313
314 1927 aaronmk
            # Run query
315 2148 aaronmk
            cur.execute(query, params)
316 1903 aaronmk
        finally:
317 2664 aaronmk
            if self.debug and debug_msg_ref != None: # return after running
318
                debug_msg_ref[0] = log_msg(str(get_cur_query(cur, query,
319
                    params)))
320 1903 aaronmk
321
        return cur
322 1914 aaronmk
323
    def is_cached(self, query, params=None):
324
        return _query_lookup(query, params) in self.query_results
325 2139 aaronmk
326
    def with_savepoint(self, func):
327 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
328 2443 aaronmk
        self.run_query('SAVEPOINT '+savepoint, log_level=4)
329 2139 aaronmk
        self._savepoint += 1
330
        try:
331
            try: return_val = func()
332
            finally:
333
                self._savepoint -= 1
334
                assert self._savepoint >= 0
335
        except:
336 2443 aaronmk
            self.run_query('ROLLBACK TO SAVEPOINT '+savepoint, log_level=4)
337 2139 aaronmk
            raise
338
        else:
339 2443 aaronmk
            self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
340 2191 aaronmk
            self.do_autocommit()
341 2139 aaronmk
            return return_val
342 2191 aaronmk
343
    def do_autocommit(self):
344
        '''Autocommits if outside savepoint'''
345
        assert self._savepoint >= 0
346
        if self.autocommit and self._savepoint == 0:
347
            self.log_debug('Autocommiting')
348
            self.db.commit()
349 2643 aaronmk
350
    def col_default(self, col):
351
        table = sql_gen.Table('columns', 'information_schema')
352
353
        conds = [('table_name', col.table.name), ('column_name', col.name)]
354
        schema = col.table.schema
355
        if schema != None: conds.append(('table_schema', schema))
356
357 2660 aaronmk
        return sql_gen.as_Code(value(select(self, table, ['column_default'],
358
            conds, order_by='table_schema', limit=1, log_level=3)))
359 2643 aaronmk
            # TODO: order_by search_path schema order
360 1849 aaronmk
361 1869 aaronmk
connect = DbConn
362
363 832 aaronmk
##### Querying
364
365 1894 aaronmk
def run_raw_query(db, *args, **kw_args):
366 2085 aaronmk
    '''For params, see DbConn.run_query()'''
367 1894 aaronmk
    return db.run_query(*args, **kw_args)
368 11 aaronmk
369 2068 aaronmk
def mogrify(db, query, params):
370
    module = util.root_module(db.db)
371
    if module == 'psycopg2': return db.db.cursor().mogrify(query, params)
372
    else: raise NotImplementedError("Can't mogrify query for "+module+
373
        ' database')
374
375 832 aaronmk
##### Recoverable querying
376 15 aaronmk
377 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
378 11 aaronmk
379 2464 aaronmk
def run_query(db, query, params=None, recover=None, cacheable=False,
380
    log_level=2, log_ignore_excs=None, **kw_args):
381 2441 aaronmk
    '''For params, see run_raw_query()'''
382 830 aaronmk
    if recover == None: recover = False
383 2464 aaronmk
    if log_ignore_excs == None: log_ignore_excs = ()
384
    log_ignore_excs = tuple(log_ignore_excs)
385 830 aaronmk
386 2666 aaronmk
    debug_msg_ref = None # usually, db.run_query() logs query before running it
387
    # But if filtering with log_ignore_excs, wait until after exception parsing
388
    if log_ignore_excs != () or not db.can_mogrify(): debug_msg_ref = [None]
389
390 2148 aaronmk
    try:
391 2464 aaronmk
        try:
392
            def run(): return run_raw_query(db, query, params, cacheable,
393
                log_level, debug_msg_ref, **kw_args)
394
            if recover and not db.is_cached(query, params):
395
                return with_savepoint(db, run)
396
            else: return run() # don't need savepoint if cached
397
        except Exception, e:
398
            if not recover: raise # need savepoint to run index_cols()
399
            msg = exc.str_(e)
400
401
            match = re.search(r'duplicate key value violates unique constraint '
402 2493 aaronmk
                r'"((_?[^\W_]+)_.+?)"', msg)
403 2464 aaronmk
            if match:
404
                constraint, table = match.groups()
405
                try: cols = index_cols(db, table, constraint)
406
                except NotImplementedError: raise e
407
                else: raise DuplicateKeyException(constraint, cols, e)
408
409 2493 aaronmk
            match = re.search(r'null value in column "(.+?)" violates not-null'
410 2464 aaronmk
                r' constraint', msg)
411
            if match: raise NullValueException('NOT NULL', [match.group(1)], e)
412
413
            match = re.search(r'\b(?:invalid input (?:syntax|value)\b.*?'
414
                r'|date/time field value out of range): "(.+?)"\n'
415 2535 aaronmk
                r'(?:(?s).*?)\bfunction "(.+?)"', msg)
416 2464 aaronmk
            if match:
417
                value, name = match.groups()
418
                raise FunctionValueException(name, strings.to_unicode(value), e)
419
420 2526 aaronmk
            match = re.search(r'column "(.+?)" is of type (.+?) but expression '
421 2523 aaronmk
                r'is of type', msg)
422
            if match:
423
                col, type_ = match.groups()
424
                raise MissingCastException(type_, col, e)
425
426 2493 aaronmk
            match = re.search(r'relation "(.+?)" already exists', msg)
427 2464 aaronmk
            if match: raise DuplicateTableException(match.group(1), e)
428
429 2493 aaronmk
            match = re.search(r'function "(.+?)" already exists', msg)
430 2464 aaronmk
            if match: raise DuplicateFunctionException(match.group(1), e)
431
432
            raise # no specific exception raised
433
    except log_ignore_excs:
434
        log_level += 2
435
        raise
436
    finally:
437 2666 aaronmk
        if debug_msg_ref != None and debug_msg_ref[0] != None:
438
            db.log_debug(debug_msg_ref[0], log_level)
439 830 aaronmk
440 832 aaronmk
##### Basic queries
441
442 2153 aaronmk
def next_version(name):
443 2163 aaronmk
    version = 1 # first existing name was version 0
444 2586 aaronmk
    match = re.match(r'^(.*)#(\d+)$', name)
445 2153 aaronmk
    if match:
446 2586 aaronmk
        name, version = match.groups()
447
        version = int(version)+1
448 2588 aaronmk
    return sql_gen.add_suffix(name, '#'+str(version))
449 2153 aaronmk
450 2386 aaronmk
def run_query_into(db, query, params, into=None, *args, **kw_args):
451 2085 aaronmk
    '''Outputs a query to a temp table.
452
    For params, see run_query().
453
    '''
454 2386 aaronmk
    if into == None: return run_query(db, query, params, *args, **kw_args)
455 2085 aaronmk
    else: # place rows in temp table
456 2386 aaronmk
        assert isinstance(into, sql_gen.Table)
457 2385 aaronmk
458 2153 aaronmk
        kw_args['recover'] = True
459 2464 aaronmk
        kw_args.setdefault('log_ignore_excs', (DuplicateTableException,))
460 2440 aaronmk
461 2468 aaronmk
        temp = not db.autocommit # tables are permanent in autocommit mode
462 2440 aaronmk
        # "temporary tables cannot specify a schema name", so remove schema
463
        if temp: into.schema = None
464
465 2153 aaronmk
        while True:
466
            try:
467 2194 aaronmk
                create_query = 'CREATE'
468 2440 aaronmk
                if temp: create_query += ' TEMP'
469 2467 aaronmk
                create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
470 2194 aaronmk
471
                return run_query(db, create_query, params, *args, **kw_args)
472 2153 aaronmk
                    # CREATE TABLE AS sets rowcount to # rows in query
473
            except DuplicateTableException, e:
474 2386 aaronmk
                into.name = next_version(into.name)
475 2153 aaronmk
                # try again with next version of name
476 2085 aaronmk
477 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
478
479 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
480
481 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
482 2293 aaronmk
    start=None, order_by=order_by_pkey, default_table=None):
483 1981 aaronmk
    '''
484 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
485 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
486 1981 aaronmk
    @param fields Use None to select all fields in the table
487 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
488 2379 aaronmk
        * container can be any iterable type
489 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
490
        * compare_right_side: sql_gen.ValueCond|literal value
491 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
492
        use all columns
493 2054 aaronmk
    @return tuple(query, params)
494 1981 aaronmk
    '''
495 2315 aaronmk
    # Parse tables param
496 2121 aaronmk
    if not lists.is_seq(tables): tables = [tables]
497 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
498 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
499 2121 aaronmk
500 2315 aaronmk
    # Parse other params
501 2376 aaronmk
    if conds == None: conds = []
502 2650 aaronmk
    elif dicts.is_dict(conds): conds = conds.items()
503 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
504 135 aaronmk
    assert limit == None or type(limit) == int
505 865 aaronmk
    assert start == None or type(start) == int
506 2315 aaronmk
    if order_by is order_by_pkey:
507
        if distinct_on != []: order_by = None
508
        else: order_by = pkey(db, table0, recover=True)
509 865 aaronmk
510 2315 aaronmk
    query = 'SELECT'
511 2056 aaronmk
512 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
513 2056 aaronmk
514 2200 aaronmk
    # DISTINCT ON columns
515 2233 aaronmk
    if distinct_on != []:
516 2467 aaronmk
        query += '\nDISTINCT'
517 2254 aaronmk
        if distinct_on is not distinct_on_all:
518 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
519
520
    # Columns
521 2467 aaronmk
    query += '\n'
522 1135 aaronmk
    if fields == None: query += '*'
523 2479 aaronmk
    else: query += '\n, '.join(map(parse_col, fields))
524 2200 aaronmk
525
    # Main table
526 2467 aaronmk
    query += '\nFROM '+table0.to_str(db)
527 865 aaronmk
528 2122 aaronmk
    # Add joins
529 2271 aaronmk
    left_table = table0
530 2263 aaronmk
    for join_ in tables:
531
        table = join_.table
532 2238 aaronmk
533 2343 aaronmk
        # Parse special values
534
        if join_.type_ is sql_gen.filter_out: # filter no match
535 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
536
                None))
537 2343 aaronmk
538 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
539 2122 aaronmk
540
        left_table = table
541
542 865 aaronmk
    missing = True
543 2376 aaronmk
    if conds != []:
544 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
545
        else: whitespace = '\n'
546 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
547
            .to_str(db) for l, r in conds], 'WHERE')
548 865 aaronmk
        missing = False
549 2227 aaronmk
    if order_by != None:
550 2467 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
551
    if limit != None: query += '\nLIMIT '+str(limit); missing = False
552 865 aaronmk
    if start != None:
553 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
554 865 aaronmk
        missing = False
555
    if missing: warnings.warn(DbWarning(
556
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
557
558 2315 aaronmk
    return (query, [])
559 11 aaronmk
560 2054 aaronmk
def select(db, *args, **kw_args):
561
    '''For params, see mk_select() and run_query()'''
562
    recover = kw_args.pop('recover', None)
563
    cacheable = kw_args.pop('cacheable', True)
564 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
565 2054 aaronmk
566
    query, params = mk_select(db, *args, **kw_args)
567 2442 aaronmk
    return run_query(db, query, params, recover, cacheable, log_level=log_level)
568 2054 aaronmk
569 2066 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, params=None,
570 2292 aaronmk
    returning=None, embeddable=False):
571 1960 aaronmk
    '''
572
    @param returning str|None An inserted column (such as pkey) to return
573 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
574 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
575
        query will be fully cached, not just if it raises an exception.
576 1960 aaronmk
    '''
577 2328 aaronmk
    table = sql_gen.as_Table(table)
578 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
579 2646 aaronmk
    if cols != None:
580
        cols = [sql_gen.to_name_only_col(v, table).to_str(db) for v in cols]
581 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
582 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
583 2063 aaronmk
584
    # Build query
585 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
586
    query = first_line
587 2467 aaronmk
    if cols != None: query += '\n('+', '.join(cols)+')'
588
    query += '\n'+select_query
589 2063 aaronmk
590
    if returning != None:
591 2327 aaronmk
        returning_name = copy.copy(returning)
592
        returning_name.table = None
593
        returning_name = returning_name.to_str(db)
594 2467 aaronmk
        query += '\nRETURNING '+returning_name
595 2063 aaronmk
596 2070 aaronmk
    if embeddable:
597 2327 aaronmk
        assert returning != None
598
599 2070 aaronmk
        # Create function
600 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
601 2327 aaronmk
        return_type = 'SETOF '+returning.to_str(db)+'%TYPE'
602 2189 aaronmk
        while True:
603
            try:
604 2327 aaronmk
                func_schema = None
605 2468 aaronmk
                if not db.autocommit: func_schema = 'pg_temp'
606 2327 aaronmk
                function = sql_gen.Table(function_name, func_schema).to_str(db)
607 2194 aaronmk
608 2189 aaronmk
                function_query = '''\
609 2467 aaronmk
CREATE FUNCTION '''+function+'''()
610
RETURNS '''+return_type+'''
611
LANGUAGE sql
612
AS $$
613
'''+mogrify(db, query, params)+''';
614
$$;
615 2070 aaronmk
'''
616 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
617 2464 aaronmk
                    log_ignore_excs=(DuplicateFunctionException,))
618 2189 aaronmk
                break # this version was successful
619
            except DuplicateFunctionException, e:
620
                function_name = next_version(function_name)
621
                # try again with next version of name
622 2070 aaronmk
623 2337 aaronmk
        # Return query that uses function
624
        func_table = sql_gen.NamedTable('f', sql_gen.CustomCode(function+'()'),
625
            [returning_name]) # AS clause requires function alias
626
        return mk_select(db, func_table, start=0, order_by=None)
627 2070 aaronmk
628 2066 aaronmk
    return (query, params)
629
630
def insert_select(db, *args, **kw_args):
631 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
632 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
633
        values in
634 2072 aaronmk
    '''
635 2386 aaronmk
    into = kw_args.pop('into', None)
636
    if into != None: kw_args['embeddable'] = True
637 2066 aaronmk
    recover = kw_args.pop('recover', None)
638
    cacheable = kw_args.pop('cacheable', True)
639
640
    query, params = mk_insert_select(db, *args, **kw_args)
641 2386 aaronmk
    return run_query_into(db, query, params, into, recover=recover,
642 2153 aaronmk
        cacheable=cacheable)
643 2063 aaronmk
644 2066 aaronmk
default = object() # tells insert() to use the default value for a column
645
646 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
647 2085 aaronmk
    '''For params, see insert_select()'''
648 1960 aaronmk
    if lists.is_seq(row): cols = None
649
    else:
650
        cols = row.keys()
651
        row = row.values()
652
    row = list(row) # ensure that "!= []" works
653
654 1961 aaronmk
    # Check for special values
655
    labels = []
656
    values = []
657
    for value in row:
658 2647 aaronmk
        value = sql_gen.remove_col_rename(sql_gen.as_Value(value)).value
659 2254 aaronmk
        if value is default: labels.append('DEFAULT')
660 1961 aaronmk
        else:
661
            labels.append('%s')
662
            values.append(value)
663
664
    # Build query
665 2467 aaronmk
    if values != []: query = 'VALUES ('+(', '.join(labels))+')'
666 2063 aaronmk
    else: query = None
667 1554 aaronmk
668 2064 aaronmk
    return insert_select(db, table, cols, query, values, *args, **kw_args)
669 11 aaronmk
670 2402 aaronmk
def mk_update(db, table, changes=None, cond=None):
671
    '''
672
    @param changes [(col, new_value),...]
673
        * container can be any iterable type
674
        * col: sql_gen.Code|str (for col name)
675
        * new_value: sql_gen.Code|literal value
676
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
677
    @return str query
678
    '''
679
    query = 'UPDATE '+sql_gen.as_Table(table).to_str(db)+'\nSET\n'
680 2405 aaronmk
    query += ',\n'.join((sql_gen.to_name_only_col(col, table).to_str(db)+' = '
681 2402 aaronmk
        +sql_gen.as_Value(new_value).to_str(db) for col, new_value in changes))
682 2467 aaronmk
    if cond != None: query += '\nWHERE\n'+cond.to_str(db)
683 2402 aaronmk
684
    return query
685
686
def update(db, *args, **kw_args):
687
    '''For params, see mk_update() and run_query()'''
688
    recover = kw_args.pop('recover', None)
689
690
    return run_query(db, mk_update(db, *args, **kw_args), [], recover)
691
692 135 aaronmk
def last_insert_id(db):
693 1849 aaronmk
    module = util.root_module(db.db)
694 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
695
    elif module == 'MySQLdb': return db.insert_id()
696
    else: return None
697 13 aaronmk
698 1968 aaronmk
def truncate(db, table, schema='public'):
699
    return run_query(db, 'TRUNCATE '+qual_name(db, schema, table)+' CASCADE')
700 832 aaronmk
701 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
702 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
703 2415 aaronmk
    to names that will be distinct among the columns' tables.
704 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
705 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
706
    @param into The table for the new columns.
707 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
708
        columns will be included in the mapping even if they are not in cols.
709
        The tables of the provided Col objects will be changed to into, so make
710
        copies of them if you want to keep the original tables.
711
    @param as_items Whether to return a list of dict items instead of a dict
712 2383 aaronmk
    @return dict(orig_col=new_col, ...)
713
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
714 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
715
        * All mappings use the into table so its name can easily be
716 2383 aaronmk
          changed for all columns at once
717
    '''
718 2415 aaronmk
    cols = lists.uniqify(cols)
719
720 2394 aaronmk
    items = []
721 2389 aaronmk
    for col in preserve:
722 2390 aaronmk
        orig_col = copy.copy(col)
723 2392 aaronmk
        col.table = into
724 2394 aaronmk
        items.append((orig_col, col))
725
    preserve = set(preserve)
726
    for col in cols:
727 2515 aaronmk
        if col not in preserve: items.append((col, sql_gen.Col(str(col), into)))
728 2394 aaronmk
729
    if not as_items: items = dict(items)
730
    return items
731 2383 aaronmk
732 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
733 2391 aaronmk
    '''For params, see mk_flatten_mapping()
734
    @return See return value of mk_flatten_mapping()
735
    '''
736 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
737
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
738 2391 aaronmk
    run_query_into(db, *mk_select(db, joins, cols, limit=limit, start=start),
739 2392 aaronmk
        into=into)
740 2394 aaronmk
    return dict(items)
741 2391 aaronmk
742 2414 aaronmk
##### Database structure queries
743
744 2426 aaronmk
def table_row_count(db, table, recover=None):
745
    return value(run_query(db, *mk_select(db, table, [sql_gen.row_count],
746 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
747 2426 aaronmk
748 2414 aaronmk
def table_cols(db, table, recover=None):
749
    return list(col_names(select(db, table, limit=0, order_by=None,
750 2443 aaronmk
        recover=recover, log_level=4)))
751 2414 aaronmk
752 2291 aaronmk
def pkey(db, table, recover=None):
753 832 aaronmk
    '''Assumed to be first column in table'''
754 2339 aaronmk
    return table_cols(db, table, recover)[0]
755 832 aaronmk
756 2559 aaronmk
not_null_col = 'not_null_col'
757 2340 aaronmk
758
def table_not_null_col(db, table, recover=None):
759
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
760
    if not_null_col in table_cols(db, table, recover): return not_null_col
761
    else: return pkey(db, table, recover)
762
763 853 aaronmk
def index_cols(db, table, index):
764
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
765
    automatically created. When you don't know whether something is a UNIQUE
766
    constraint or a UNIQUE index, use this function.'''
767 1909 aaronmk
    module = util.root_module(db.db)
768
    if module == 'psycopg2':
769
        return list(values(run_query(db, '''\
770 853 aaronmk
SELECT attname
771 866 aaronmk
FROM
772
(
773
        SELECT attnum, attname
774
        FROM pg_index
775
        JOIN pg_class index ON index.oid = indexrelid
776
        JOIN pg_class table_ ON table_.oid = indrelid
777
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
778
        WHERE
779
            table_.relname = %(table)s
780
            AND index.relname = %(index)s
781
    UNION
782
        SELECT attnum, attname
783
        FROM
784
        (
785
            SELECT
786
                indrelid
787
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
788
                    AS indkey
789
            FROM pg_index
790
            JOIN pg_class index ON index.oid = indexrelid
791
            JOIN pg_class table_ ON table_.oid = indrelid
792
            WHERE
793
                table_.relname = %(table)s
794
                AND index.relname = %(index)s
795
        ) s
796
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
797
) s
798 853 aaronmk
ORDER BY attnum
799
''',
800 2443 aaronmk
            {'table': table, 'index': index}, cacheable=True, log_level=4)))
801 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
802
        ' database')
803 853 aaronmk
804 464 aaronmk
def constraint_cols(db, table, constraint):
805 1849 aaronmk
    module = util.root_module(db.db)
806 464 aaronmk
    if module == 'psycopg2':
807
        return list(values(run_query(db, '''\
808
SELECT attname
809
FROM pg_constraint
810
JOIN pg_class ON pg_class.oid = conrelid
811
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
812
WHERE
813
    relname = %(table)s
814
    AND conname = %(constraint)s
815
ORDER BY attnum
816
''',
817
            {'table': table, 'constraint': constraint})))
818
    else: raise NotImplementedError("Can't list constraint columns for "+module+
819
        ' database')
820
821 2096 aaronmk
row_num_col = '_row_num'
822
823 2538 aaronmk
def add_index(db, expr):
824
    '''Adds an index on a column or expression if it doesn't already exist.
825
    Currently, only function calls are supported as expressions.
826
    '''
827
    expr = copy.copy(expr) # don't modify input!
828
829
    # Extract col
830 2539 aaronmk
    if isinstance(expr, sql_gen.FunctionCall):
831
        col = expr.args[0]
832 2541 aaronmk
        expr = sql_gen.Expr(expr)
833 2538 aaronmk
    else: col = expr
834 2408 aaronmk
    assert sql_gen.is_table_col(col)
835
836 2538 aaronmk
    index = sql_gen.as_Table(str(expr))
837 2408 aaronmk
    table = col.table
838 2538 aaronmk
    col.table = None
839 2408 aaronmk
    try: run_query(db, 'CREATE INDEX '+index.to_str(db)+' ON '+table.to_str(db)
840 2538 aaronmk
        +' ('+expr.to_str(db)+')', recover=True, cacheable=True, log_level=3)
841 2408 aaronmk
    except DuplicateTableException: pass # index already existed
842
843 2594 aaronmk
def add_pkey(db, table, recover=None):
844 2406 aaronmk
    '''Makes the first column in a table the primary key.
845
    @pre The table must not already have a primary key.
846
    '''
847
    table = sql_gen.as_Table(table)
848
849 2590 aaronmk
    index = sql_gen.as_Table(sql_gen.add_suffix(table.name, '_pkey'))
850 2406 aaronmk
    col = sql_gen.to_name_only_col(pkey(db, table, recover))
851 2614 aaronmk
    try:
852
        run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD CONSTRAINT '
853
            +index.to_str(db)+' PRIMARY KEY('+col.to_str(db)+')',
854
            recover=True, cacheable=True, log_level=3,
855
            log_ignore_excs=(DuplicateTableException,))
856
    except DuplicateTableException, e:
857
        index.name = next_version(index.name)
858
        # try again with next version of name
859 2406 aaronmk
860 2086 aaronmk
def add_row_num(db, table):
861 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
862
    be the primary key.'''
863 2320 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
864 2096 aaronmk
    run_query(db, 'ALTER TABLE '+table+' ADD COLUMN '+row_num_col
865 2443 aaronmk
        +' serial NOT NULL PRIMARY KEY', log_level=3)
866 2086 aaronmk
867 2548 aaronmk
def tables(db, schema_like='public', table_like='%'):
868 1849 aaronmk
    module = util.root_module(db.db)
869 2548 aaronmk
    params = {'schema_like': schema_like, 'table_like': table_like}
870 832 aaronmk
    if module == 'psycopg2':
871 1968 aaronmk
        return values(run_query(db, '''\
872
SELECT tablename
873
FROM pg_tables
874
WHERE
875 2548 aaronmk
    schemaname LIKE %(schema_like)s
876 1968 aaronmk
    AND tablename LIKE %(table_like)s
877
ORDER BY tablename
878
''',
879
            params, cacheable=True))
880
    elif module == 'MySQLdb':
881
        return values(run_query(db, 'SHOW TABLES LIKE %(table_like)s', params,
882
            cacheable=True))
883 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
884 830 aaronmk
885 833 aaronmk
##### Database management
886
887 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
888
    '''For kw_args, see tables()'''
889
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
890 833 aaronmk
891 832 aaronmk
##### Heuristic queries
892
893 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
894 1554 aaronmk
    '''Recovers from errors.
895 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
896
    '''
897 2642 aaronmk
    row = sql_gen.ColDict(db, table, row)
898 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
899
900 471 aaronmk
    try:
901 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
902 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
903
            row_ct_ref[0] += cur.rowcount
904
        return value(cur)
905 471 aaronmk
    except DuplicateKeyException, e:
906 2653 aaronmk
        row = sql_gen.ColDict(db, table,
907
            util.dict_subset_right_join(row, e.cols))
908
        return value(select(db, table, [pkey_], row, recover=True))
909 471 aaronmk
910 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
911 830 aaronmk
    '''Recovers from errors'''
912 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
913 14 aaronmk
    except StopIteration:
914 40 aaronmk
        if not create: raise
915 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
916 2078 aaronmk
917 2593 aaronmk
def is_func_result(col):
918
    return col.table.name.find('(') >= 0 and col.name == 'result'
919
920 2592 aaronmk
def into_table_name(out_table, in_tables0, mapping, is_func):
921 2615 aaronmk
    def in_col_str(in_col):
922
        in_col = sql_gen.remove_col_rename(in_col)
923
        if isinstance(in_col, sql_gen.Col):
924
            table = in_col.table
925
            if table == in_tables0:
926
                in_col = sql_gen.to_name_only_col(in_col)
927
            elif is_func_result(in_col): in_col = table # omit col name
928
        return str(in_col)
929
930 2580 aaronmk
    str_ = str(out_table)
931
    if is_func:
932 2615 aaronmk
        str_ += '('
933 2580 aaronmk
934 2615 aaronmk
        try: value_in_col = mapping['value']
935
        except KeyError:
936 2654 aaronmk
            str_ += ', '.join((str(k)+'='+in_col_str(v)
937 2615 aaronmk
                for k, v in mapping.iteritems()))
938
        else: str_ += in_col_str(value_in_col)
939
940
        str_ += ')'
941 2580 aaronmk
    else: str_ += '_pkeys'
942
    return str_
943
944 2508 aaronmk
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None, into=None,
945 2552 aaronmk
    default=None, is_func=False):
946 2078 aaronmk
    '''Recovers from errors.
947
    Only works under PostgreSQL (uses INSERT RETURNING).
948 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
949
        tables to join with it using the main input table's pkey
950 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
951 2612 aaronmk
        * out_table_col: str (*not* sql_gen.Col)
952
        * in_table_col: sql_gen.Col|literal-value
953 2489 aaronmk
    @param into The table to contain the output and input pkeys.
954 2574 aaronmk
        Defaults to `out_table.name+'_pkeys'`.
955 2509 aaronmk
    @param default The *output* column to use as the pkey for missing rows.
956
        If this output column does not exist in the mapping, uses None.
957 2552 aaronmk
    @param is_func Whether out_table is the name of a SQL function, not a table
958 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
959 2078 aaronmk
    '''
960 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
961 2312 aaronmk
962 2450 aaronmk
    def log_debug(msg): db.log_debug(msg, level=1.5)
963 2505 aaronmk
    def col_ustr(str_):
964 2567 aaronmk
        return strings.repr_no_u(sql_gen.remove_col_rename(str_))
965 2450 aaronmk
966 2486 aaronmk
    log_debug('********** New iteration **********')
967 2505 aaronmk
    log_debug('Inserting these input columns into '+strings.as_tt(
968
        out_table.to_str(db))+':\n'+strings.as_table(mapping, ustr=col_ustr))
969 2463 aaronmk
970 2382 aaronmk
    # Create input joins from list of input tables
971
    in_tables_ = in_tables[:] # don't modify input!
972
    in_tables0 = in_tables_.pop(0) # first table is separate
973 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
974 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
975 2460 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v,
976
        {in_pkey: sql_gen.join_same_not_null}) for v in in_tables_]
977 2131 aaronmk
978 2592 aaronmk
    if into == None:
979
        into = into_table_name(out_table, in_tables0, mapping, is_func)
980
    into = sql_gen.as_Table(into)
981
982 2486 aaronmk
    log_debug('Joining together input tables into temp table')
983 2395 aaronmk
    # Place in new table for speed and so don't modify input if values edited
984 2584 aaronmk
    in_table = sql_gen.Table('in')
985 2395 aaronmk
    flatten_cols = filter(sql_gen.is_table_col, mapping.values())
986
    mapping = dicts.join(mapping, flatten(db, in_table, input_joins,
987
        flatten_cols, preserve=[in_pkey_col], start=0))
988
    input_joins = [in_table]
989 2486 aaronmk
    db.log_debug('Temp table: '+strings.as_tt(in_table.to_str(db)), level=2)
990 2395 aaronmk
991 2656 aaronmk
    mapping = sql_gen.ColDict(db, out_table, mapping)
992
        # after applying dicts.join() because that returns a plain dict
993
994 2509 aaronmk
    # Resolve default value column
995
    try: default = mapping[default]
996
    except KeyError:
997
        if default != None:
998
            db.log_debug('Default value column '
999
                +strings.as_tt(strings.repr_no_u(default))
1000 2511 aaronmk
                +' does not exist in mapping, falling back to None', level=2.1)
1001 2509 aaronmk
            default = None
1002
1003 2279 aaronmk
    out_pkey = pkey(db, out_table, recover=True)
1004 2285 aaronmk
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
1005 2142 aaronmk
1006 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
1007 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
1008
1009 2201 aaronmk
    pkeys_table_exists_ref = [False]
1010 2420 aaronmk
    def insert_into_pkeys(joins, cols):
1011
        query, params = mk_select(db, joins, cols, order_by=None, start=0)
1012 2201 aaronmk
        if pkeys_table_exists_ref[0]:
1013 2489 aaronmk
            insert_select(db, into, pkeys_names, query, params)
1014 2201 aaronmk
        else:
1015 2489 aaronmk
            run_query_into(db, query, params, into=into)
1016 2201 aaronmk
            pkeys_table_exists_ref[0] = True
1017
1018 2429 aaronmk
    limit_ref = [None]
1019 2380 aaronmk
    conds = set()
1020 2233 aaronmk
    distinct_on = []
1021 2325 aaronmk
    def mk_main_select(joins, cols):
1022 2429 aaronmk
        return mk_select(db, joins, cols, conds, distinct_on,
1023
            limit=limit_ref[0], start=0)
1024 2132 aaronmk
1025 2519 aaronmk
    exc_strs = set()
1026 2309 aaronmk
    def log_exc(e):
1027 2519 aaronmk
        e_str = exc.str_(e, first_line_only=True)
1028
        log_debug('Caught exception: '+e_str)
1029
        assert e_str not in exc_strs # avoid infinite loops
1030
        exc_strs.add(e_str)
1031 2451 aaronmk
    def remove_all_rows():
1032 2450 aaronmk
        log_debug('Returning NULL for all rows')
1033 2429 aaronmk
        limit_ref[0] = 0 # just create an empty pkeys table
1034 2409 aaronmk
    def ignore(in_col, value):
1035 2545 aaronmk
        in_col_str = strings.as_tt(repr(in_col))
1036 2544 aaronmk
        db.log_debug('Adding index on '+in_col_str+' to enable fast filtering',
1037
            level=2.5)
1038 2537 aaronmk
        add_index(db, in_col)
1039 2545 aaronmk
        log_debug('Ignoring rows with '+in_col_str+' = '
1040
            +strings.as_tt(repr(value)))
1041 2403 aaronmk
    def remove_rows(in_col, value):
1042 2409 aaronmk
        ignore(in_col, value)
1043 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
1044
        assert cond not in conds # avoid infinite loops
1045 2380 aaronmk
        conds.add(cond)
1046 2403 aaronmk
    def invalid2null(in_col, value):
1047 2409 aaronmk
        ignore(in_col, value)
1048 2403 aaronmk
        update(db, in_table, [(in_col, None)],
1049
            sql_gen.ColValueCond(in_col, value))
1050 2245 aaronmk
1051 2589 aaronmk
    def insert_pkeys_table(which):
1052
        return sql_gen.Table(sql_gen.add_suffix(in_table.name,
1053
            '_insert_'+which+'_pkeys'))
1054
    insert_out_pkeys = insert_pkeys_table('out')
1055
    insert_in_pkeys = insert_pkeys_table('in')
1056
1057 2206 aaronmk
    # Do inserts and selects
1058 2642 aaronmk
    join_cols = sql_gen.ColDict(db, out_table)
1059 2206 aaronmk
    while True:
1060 2521 aaronmk
        if limit_ref[0] == 0: # special case
1061
            log_debug('Creating an empty pkeys table')
1062
            cur = run_query_into(db, *mk_select(db, out_table, [out_pkey],
1063
                limit=limit_ref[0]), into=insert_out_pkeys)
1064
            break # don't do main case
1065
1066 2303 aaronmk
        has_joins = join_cols != {}
1067
1068 2305 aaronmk
        # Prepare to insert new rows
1069 2325 aaronmk
        insert_joins = input_joins[:] # don't modify original!
1070 2403 aaronmk
        insert_args = dict(recover=True, cacheable=False)
1071 2303 aaronmk
        if has_joins:
1072 2317 aaronmk
            distinct_on = [v.to_Col() for v in join_cols.values()]
1073 2325 aaronmk
            insert_joins.append(sql_gen.Join(out_table, join_cols,
1074
                sql_gen.filter_out))
1075
        else:
1076 2404 aaronmk
            insert_args.update(dict(returning=out_pkey, into=insert_out_pkeys))
1077 2520 aaronmk
        main_select = mk_main_select(insert_joins, mapping.values())[0]
1078 2303 aaronmk
1079 2486 aaronmk
        log_debug('Trying to insert new rows')
1080 2206 aaronmk
        try:
1081 2518 aaronmk
            cur = insert_select(db, out_table, mapping.keys(), main_select,
1082
                **insert_args)
1083 2357 aaronmk
            break # insert successful
1084 2206 aaronmk
        except DuplicateKeyException, e:
1085 2309 aaronmk
            log_exc(e)
1086
1087 2258 aaronmk
            old_join_cols = join_cols.copy()
1088 2565 aaronmk
            join_cols.update(util.dict_subset_right_join(mapping, e.cols))
1089 2486 aaronmk
            log_debug('Ignoring existing rows, comparing on these columns:\n'
1090 2505 aaronmk
                +strings.as_inline_table(join_cols, ustr=col_ustr))
1091 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
1092 2230 aaronmk
        except NullValueException, e:
1093 2309 aaronmk
            log_exc(e)
1094
1095 2230 aaronmk
            out_col, = e.cols
1096
            try: in_col = mapping[out_col]
1097 2356 aaronmk
            except KeyError:
1098 2486 aaronmk
                log_debug('Missing mapping for NOT NULL column '+out_col)
1099 2451 aaronmk
                remove_all_rows()
1100 2403 aaronmk
            else: remove_rows(in_col, None)
1101 2542 aaronmk
        except FunctionValueException, e:
1102
            log_exc(e)
1103
1104
            func_name = e.name
1105
            value = e.value
1106
            for out_col, in_col in mapping.iteritems():
1107 2562 aaronmk
                invalid2null(sql_gen.unwrap_func_call(in_col, func_name), value)
1108 2525 aaronmk
        except MissingCastException, e:
1109
            log_exc(e)
1110
1111
            out_col = e.col
1112 2534 aaronmk
            mapping[out_col] = sql_gen.wrap_in_func(e.type, mapping[out_col])
1113 2429 aaronmk
        except DatabaseErrors, e:
1114
            log_exc(e)
1115
1116 2531 aaronmk
            msg = 'No handler for exception: '+exc.str_(e)
1117 2451 aaronmk
            warnings.warn(DbWarning(msg))
1118
            log_debug(msg)
1119
            remove_all_rows()
1120 2358 aaronmk
        # after exception handled, rerun loop with additional constraints
1121 2132 aaronmk
1122 2357 aaronmk
    if row_ct_ref != None and cur.rowcount >= 0:
1123
        row_ct_ref[0] += cur.rowcount
1124
1125
    if has_joins:
1126
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
1127 2486 aaronmk
        log_debug('Getting output table pkeys of existing/inserted rows')
1128 2420 aaronmk
        insert_into_pkeys(select_joins, pkeys_cols)
1129 2357 aaronmk
    else:
1130 2404 aaronmk
        add_row_num(db, insert_out_pkeys) # for joining with input pkeys
1131 2357 aaronmk
1132 2486 aaronmk
        log_debug('Getting input table pkeys of inserted rows')
1133 2357 aaronmk
        run_query_into(db, *mk_main_select(input_joins, [in_pkey]),
1134 2404 aaronmk
            into=insert_in_pkeys)
1135
        add_row_num(db, insert_in_pkeys) # for joining with output pkeys
1136 2357 aaronmk
1137 2428 aaronmk
        assert table_row_count(db, insert_out_pkeys) == table_row_count(db,
1138
            insert_in_pkeys)
1139
1140 2486 aaronmk
        log_debug('Combining output and input pkeys in inserted order')
1141 2404 aaronmk
        pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
1142 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
1143 2420 aaronmk
        insert_into_pkeys(pkey_joins, pkeys_names)
1144 2357 aaronmk
1145 2486 aaronmk
    db.log_debug('Adding pkey on pkeys table to enable fast joins', level=2.5)
1146 2594 aaronmk
    add_pkey(db, into)
1147 2407 aaronmk
1148 2508 aaronmk
    log_debug('Setting pkeys of missing rows to '+strings.as_tt(repr(default)))
1149 2489 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(into,
1150 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
1151
        # must use join_same_not_null or query will take forever
1152 2420 aaronmk
    insert_into_pkeys(missing_rows_joins,
1153 2508 aaronmk
        [in_pkey_col, sql_gen.NamedCol(out_pkey, default)])
1154 2357 aaronmk
1155 2489 aaronmk
    assert table_row_count(db, into) == table_row_count(db, in_table)
1156 2428 aaronmk
1157 2489 aaronmk
    return sql_gen.Col(out_pkey, into)
1158 2115 aaronmk
1159
##### Data cleanup
1160
1161 2290 aaronmk
def cleanup_table(db, table, cols):
1162 2115 aaronmk
    def esc_name_(name): return esc_name(db, name)
1163
1164 2290 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
1165 2115 aaronmk
    cols = map(esc_name_, cols)
1166
1167
    run_query(db, 'UPDATE '+table+' SET\n'+(',\n'.join(('\n'+col
1168
        +' = nullif(nullif(trim(both from '+col+"), %(null0)s), %(null1)s)"
1169
            for col in cols))),
1170
        dict(null0='', null1=r'\N'))