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 2148 aaronmk
                finally: self.query = get_cur_query(self.inner)
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 2215 aaronmk
        module = util.root_module(self.db)
262 2374 aaronmk
        if module == 'psycopg2': str_ = self.db.cursor().mogrify('%s', [value])
263 2212 aaronmk
        elif module == 'MySQLdb':
264
            import _mysql
265 2374 aaronmk
            str_ = _mysql.escape_string(value)
266 2212 aaronmk
        else: raise NotImplementedError("Can't escape value for "+module
267
            +' database')
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 2445 aaronmk
    def run_query(self, query, params=None, cacheable=False, log_level=2,
273 2464 aaronmk
        debug_msg_ref=None):
274 2445 aaronmk
        '''
275 2464 aaronmk
        @param log_ignore_excs The log_level will be increased by 2 if the query
276
            throws one of these exceptions.
277 2445 aaronmk
        '''
278 2167 aaronmk
        assert query != None
279
280 2047 aaronmk
        if not self.caching: cacheable = False
281 1903 aaronmk
        used_cache = False
282
        try:
283 1927 aaronmk
            # Get cursor
284
            if cacheable:
285
                query_lookup = _query_lookup(query, params)
286
                try:
287
                    cur = self.query_results[query_lookup]
288
                    used_cache = True
289
                except KeyError: cur = self.DbCursor(self)
290
            else: cur = self.db.cursor()
291
292
            # Run query
293 2148 aaronmk
            cur.execute(query, params)
294 1903 aaronmk
        finally:
295 2464 aaronmk
            if self.debug and debug_msg_ref != None:# only compute msg if needed
296 2470 aaronmk
                if used_cache: cache_status = 'cache hit'
297
                elif cacheable: cache_status = 'cache miss'
298
                else: cache_status = 'non-cacheable'
299 2472 aaronmk
                query_code = strings.as_code(str(get_cur_query(cur, query,
300
                    params)), 'SQL')
301
                debug_msg_ref[0] = 'DB query: '+cache_status+':\n'+query_code
302 1903 aaronmk
303
        return cur
304 1914 aaronmk
305
    def is_cached(self, query, params=None):
306
        return _query_lookup(query, params) in self.query_results
307 2139 aaronmk
308
    def with_savepoint(self, func):
309 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
310 2443 aaronmk
        self.run_query('SAVEPOINT '+savepoint, log_level=4)
311 2139 aaronmk
        self._savepoint += 1
312
        try:
313
            try: return_val = func()
314
            finally:
315
                self._savepoint -= 1
316
                assert self._savepoint >= 0
317
        except:
318 2443 aaronmk
            self.run_query('ROLLBACK TO SAVEPOINT '+savepoint, log_level=4)
319 2139 aaronmk
            raise
320
        else:
321 2443 aaronmk
            self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
322 2191 aaronmk
            self.do_autocommit()
323 2139 aaronmk
            return return_val
324 2191 aaronmk
325
    def do_autocommit(self):
326
        '''Autocommits if outside savepoint'''
327
        assert self._savepoint >= 0
328
        if self.autocommit and self._savepoint == 0:
329
            self.log_debug('Autocommiting')
330
            self.db.commit()
331 2643 aaronmk
332
    def col_default(self, col):
333
        table = sql_gen.Table('columns', 'information_schema')
334
335
        conds = [('table_name', col.table.name), ('column_name', col.name)]
336
        schema = col.table.schema
337
        if schema != None: conds.append(('table_schema', schema))
338
339
        return value(select(self, table, ['column_default'], conds,
340
            order_by='table_schema', limit=1, log_level=3))
341
            # TODO: order_by search_path schema order
342 1849 aaronmk
343 1869 aaronmk
connect = DbConn
344
345 832 aaronmk
##### Querying
346
347 1894 aaronmk
def run_raw_query(db, *args, **kw_args):
348 2085 aaronmk
    '''For params, see DbConn.run_query()'''
349 1894 aaronmk
    return db.run_query(*args, **kw_args)
350 11 aaronmk
351 2068 aaronmk
def mogrify(db, query, params):
352
    module = util.root_module(db.db)
353
    if module == 'psycopg2': return db.db.cursor().mogrify(query, params)
354
    else: raise NotImplementedError("Can't mogrify query for "+module+
355
        ' database')
356
357 832 aaronmk
##### Recoverable querying
358 15 aaronmk
359 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
360 11 aaronmk
361 2464 aaronmk
def run_query(db, query, params=None, recover=None, cacheable=False,
362
    log_level=2, log_ignore_excs=None, **kw_args):
363 2441 aaronmk
    '''For params, see run_raw_query()'''
364 830 aaronmk
    if recover == None: recover = False
365 2464 aaronmk
    if log_ignore_excs == None: log_ignore_excs = ()
366
    log_ignore_excs = tuple(log_ignore_excs)
367 830 aaronmk
368 2464 aaronmk
    debug_msg_ref = [None]
369 2148 aaronmk
    try:
370 2464 aaronmk
        try:
371
            def run(): return run_raw_query(db, query, params, cacheable,
372
                log_level, debug_msg_ref, **kw_args)
373
            if recover and not db.is_cached(query, params):
374
                return with_savepoint(db, run)
375
            else: return run() # don't need savepoint if cached
376
        except Exception, e:
377
            if not recover: raise # need savepoint to run index_cols()
378
            msg = exc.str_(e)
379
380
            match = re.search(r'duplicate key value violates unique constraint '
381 2493 aaronmk
                r'"((_?[^\W_]+)_.+?)"', msg)
382 2464 aaronmk
            if match:
383
                constraint, table = match.groups()
384
                try: cols = index_cols(db, table, constraint)
385
                except NotImplementedError: raise e
386
                else: raise DuplicateKeyException(constraint, cols, e)
387
388 2493 aaronmk
            match = re.search(r'null value in column "(.+?)" violates not-null'
389 2464 aaronmk
                r' constraint', msg)
390
            if match: raise NullValueException('NOT NULL', [match.group(1)], e)
391
392
            match = re.search(r'\b(?:invalid input (?:syntax|value)\b.*?'
393
                r'|date/time field value out of range): "(.+?)"\n'
394 2535 aaronmk
                r'(?:(?s).*?)\bfunction "(.+?)"', msg)
395 2464 aaronmk
            if match:
396
                value, name = match.groups()
397
                raise FunctionValueException(name, strings.to_unicode(value), e)
398
399 2526 aaronmk
            match = re.search(r'column "(.+?)" is of type (.+?) but expression '
400 2523 aaronmk
                r'is of type', msg)
401
            if match:
402
                col, type_ = match.groups()
403
                raise MissingCastException(type_, col, e)
404
405 2493 aaronmk
            match = re.search(r'relation "(.+?)" already exists', msg)
406 2464 aaronmk
            if match: raise DuplicateTableException(match.group(1), e)
407
408 2493 aaronmk
            match = re.search(r'function "(.+?)" already exists', msg)
409 2464 aaronmk
            if match: raise DuplicateFunctionException(match.group(1), e)
410
411
            raise # no specific exception raised
412
    except log_ignore_excs:
413
        log_level += 2
414
        raise
415
    finally:
416
        if debug_msg_ref[0] != None: db.log_debug(debug_msg_ref[0], log_level)
417 830 aaronmk
418 832 aaronmk
##### Basic queries
419
420 2153 aaronmk
def next_version(name):
421 2163 aaronmk
    version = 1 # first existing name was version 0
422 2586 aaronmk
    match = re.match(r'^(.*)#(\d+)$', name)
423 2153 aaronmk
    if match:
424 2586 aaronmk
        name, version = match.groups()
425
        version = int(version)+1
426 2588 aaronmk
    return sql_gen.add_suffix(name, '#'+str(version))
427 2153 aaronmk
428 2386 aaronmk
def run_query_into(db, query, params, into=None, *args, **kw_args):
429 2085 aaronmk
    '''Outputs a query to a temp table.
430
    For params, see run_query().
431
    '''
432 2386 aaronmk
    if into == None: return run_query(db, query, params, *args, **kw_args)
433 2085 aaronmk
    else: # place rows in temp table
434 2386 aaronmk
        assert isinstance(into, sql_gen.Table)
435 2385 aaronmk
436 2153 aaronmk
        kw_args['recover'] = True
437 2464 aaronmk
        kw_args.setdefault('log_ignore_excs', (DuplicateTableException,))
438 2440 aaronmk
439 2468 aaronmk
        temp = not db.autocommit # tables are permanent in autocommit mode
440 2440 aaronmk
        # "temporary tables cannot specify a schema name", so remove schema
441
        if temp: into.schema = None
442
443 2153 aaronmk
        while True:
444
            try:
445 2194 aaronmk
                create_query = 'CREATE'
446 2440 aaronmk
                if temp: create_query += ' TEMP'
447 2467 aaronmk
                create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
448 2194 aaronmk
449
                return run_query(db, create_query, params, *args, **kw_args)
450 2153 aaronmk
                    # CREATE TABLE AS sets rowcount to # rows in query
451
            except DuplicateTableException, e:
452 2386 aaronmk
                into.name = next_version(into.name)
453 2153 aaronmk
                # try again with next version of name
454 2085 aaronmk
455 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
456
457 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
458
459 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
460 2293 aaronmk
    start=None, order_by=order_by_pkey, default_table=None):
461 1981 aaronmk
    '''
462 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
463 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
464 1981 aaronmk
    @param fields Use None to select all fields in the table
465 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
466 2379 aaronmk
        * container can be any iterable type
467 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
468
        * compare_right_side: sql_gen.ValueCond|literal value
469 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
470
        use all columns
471 2054 aaronmk
    @return tuple(query, params)
472 1981 aaronmk
    '''
473 2315 aaronmk
    # Parse tables param
474 2121 aaronmk
    if not lists.is_seq(tables): tables = [tables]
475 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
476 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
477 2121 aaronmk
478 2315 aaronmk
    # Parse other params
479 2376 aaronmk
    if conds == None: conds = []
480
    elif isinstance(conds, dict): conds = conds.items()
481 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
482 135 aaronmk
    assert limit == None or type(limit) == int
483 865 aaronmk
    assert start == None or type(start) == int
484 2315 aaronmk
    if order_by is order_by_pkey:
485
        if distinct_on != []: order_by = None
486
        else: order_by = pkey(db, table0, recover=True)
487 865 aaronmk
488 2315 aaronmk
    query = 'SELECT'
489 2056 aaronmk
490 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
491 2056 aaronmk
492 2200 aaronmk
    # DISTINCT ON columns
493 2233 aaronmk
    if distinct_on != []:
494 2467 aaronmk
        query += '\nDISTINCT'
495 2254 aaronmk
        if distinct_on is not distinct_on_all:
496 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
497
498
    # Columns
499 2467 aaronmk
    query += '\n'
500 1135 aaronmk
    if fields == None: query += '*'
501 2479 aaronmk
    else: query += '\n, '.join(map(parse_col, fields))
502 2200 aaronmk
503
    # Main table
504 2467 aaronmk
    query += '\nFROM '+table0.to_str(db)
505 865 aaronmk
506 2122 aaronmk
    # Add joins
507 2271 aaronmk
    left_table = table0
508 2263 aaronmk
    for join_ in tables:
509
        table = join_.table
510 2238 aaronmk
511 2343 aaronmk
        # Parse special values
512
        if join_.type_ is sql_gen.filter_out: # filter no match
513 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
514
                None))
515 2343 aaronmk
516 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
517 2122 aaronmk
518
        left_table = table
519
520 865 aaronmk
    missing = True
521 2376 aaronmk
    if conds != []:
522 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
523
        else: whitespace = '\n'
524 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
525
            .to_str(db) for l, r in conds], 'WHERE')
526 865 aaronmk
        missing = False
527 2227 aaronmk
    if order_by != None:
528 2467 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
529
    if limit != None: query += '\nLIMIT '+str(limit); missing = False
530 865 aaronmk
    if start != None:
531 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
532 865 aaronmk
        missing = False
533
    if missing: warnings.warn(DbWarning(
534
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
535
536 2315 aaronmk
    return (query, [])
537 11 aaronmk
538 2054 aaronmk
def select(db, *args, **kw_args):
539
    '''For params, see mk_select() and run_query()'''
540
    recover = kw_args.pop('recover', None)
541
    cacheable = kw_args.pop('cacheable', True)
542 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
543 2054 aaronmk
544
    query, params = mk_select(db, *args, **kw_args)
545 2442 aaronmk
    return run_query(db, query, params, recover, cacheable, log_level=log_level)
546 2054 aaronmk
547 2066 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, params=None,
548 2292 aaronmk
    returning=None, embeddable=False):
549 1960 aaronmk
    '''
550
    @param returning str|None An inserted column (such as pkey) to return
551 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
552 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
553
        query will be fully cached, not just if it raises an exception.
554 1960 aaronmk
    '''
555 2328 aaronmk
    table = sql_gen.as_Table(table)
556 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
557
    if cols != None: cols = [sql_gen.as_Col(v).to_str(db) for v in cols]
558 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
559 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
560 2063 aaronmk
561
    # Build query
562 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
563
    query = first_line
564 2467 aaronmk
    if cols != None: query += '\n('+', '.join(cols)+')'
565
    query += '\n'+select_query
566 2063 aaronmk
567
    if returning != None:
568 2327 aaronmk
        returning_name = copy.copy(returning)
569
        returning_name.table = None
570
        returning_name = returning_name.to_str(db)
571 2467 aaronmk
        query += '\nRETURNING '+returning_name
572 2063 aaronmk
573 2070 aaronmk
    if embeddable:
574 2327 aaronmk
        assert returning != None
575
576 2070 aaronmk
        # Create function
577 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
578 2327 aaronmk
        return_type = 'SETOF '+returning.to_str(db)+'%TYPE'
579 2189 aaronmk
        while True:
580
            try:
581 2327 aaronmk
                func_schema = None
582 2468 aaronmk
                if not db.autocommit: func_schema = 'pg_temp'
583 2327 aaronmk
                function = sql_gen.Table(function_name, func_schema).to_str(db)
584 2194 aaronmk
585 2189 aaronmk
                function_query = '''\
586 2467 aaronmk
CREATE FUNCTION '''+function+'''()
587
RETURNS '''+return_type+'''
588
LANGUAGE sql
589
AS $$
590
'''+mogrify(db, query, params)+''';
591
$$;
592 2070 aaronmk
'''
593 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
594 2464 aaronmk
                    log_ignore_excs=(DuplicateFunctionException,))
595 2189 aaronmk
                break # this version was successful
596
            except DuplicateFunctionException, e:
597
                function_name = next_version(function_name)
598
                # try again with next version of name
599 2070 aaronmk
600 2337 aaronmk
        # Return query that uses function
601
        func_table = sql_gen.NamedTable('f', sql_gen.CustomCode(function+'()'),
602
            [returning_name]) # AS clause requires function alias
603
        return mk_select(db, func_table, start=0, order_by=None)
604 2070 aaronmk
605 2066 aaronmk
    return (query, params)
606
607
def insert_select(db, *args, **kw_args):
608 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
609 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
610
        values in
611 2072 aaronmk
    '''
612 2386 aaronmk
    into = kw_args.pop('into', None)
613
    if into != None: kw_args['embeddable'] = True
614 2066 aaronmk
    recover = kw_args.pop('recover', None)
615
    cacheable = kw_args.pop('cacheable', True)
616
617
    query, params = mk_insert_select(db, *args, **kw_args)
618 2386 aaronmk
    return run_query_into(db, query, params, into, recover=recover,
619 2153 aaronmk
        cacheable=cacheable)
620 2063 aaronmk
621 2066 aaronmk
default = object() # tells insert() to use the default value for a column
622
623 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
624 2085 aaronmk
    '''For params, see insert_select()'''
625 1960 aaronmk
    if lists.is_seq(row): cols = None
626
    else:
627
        cols = row.keys()
628
        row = row.values()
629
    row = list(row) # ensure that "!= []" works
630
631 1961 aaronmk
    # Check for special values
632
    labels = []
633
    values = []
634
    for value in row:
635 2254 aaronmk
        if value is default: labels.append('DEFAULT')
636 1961 aaronmk
        else:
637
            labels.append('%s')
638
            values.append(value)
639
640
    # Build query
641 2467 aaronmk
    if values != []: query = 'VALUES ('+(', '.join(labels))+')'
642 2063 aaronmk
    else: query = None
643 1554 aaronmk
644 2064 aaronmk
    return insert_select(db, table, cols, query, values, *args, **kw_args)
645 11 aaronmk
646 2402 aaronmk
def mk_update(db, table, changes=None, cond=None):
647
    '''
648
    @param changes [(col, new_value),...]
649
        * container can be any iterable type
650
        * col: sql_gen.Code|str (for col name)
651
        * new_value: sql_gen.Code|literal value
652
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
653
    @return str query
654
    '''
655
    query = 'UPDATE '+sql_gen.as_Table(table).to_str(db)+'\nSET\n'
656 2405 aaronmk
    query += ',\n'.join((sql_gen.to_name_only_col(col, table).to_str(db)+' = '
657 2402 aaronmk
        +sql_gen.as_Value(new_value).to_str(db) for col, new_value in changes))
658 2467 aaronmk
    if cond != None: query += '\nWHERE\n'+cond.to_str(db)
659 2402 aaronmk
660
    return query
661
662
def update(db, *args, **kw_args):
663
    '''For params, see mk_update() and run_query()'''
664
    recover = kw_args.pop('recover', None)
665
666
    return run_query(db, mk_update(db, *args, **kw_args), [], recover)
667
668 135 aaronmk
def last_insert_id(db):
669 1849 aaronmk
    module = util.root_module(db.db)
670 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
671
    elif module == 'MySQLdb': return db.insert_id()
672
    else: return None
673 13 aaronmk
674 1968 aaronmk
def truncate(db, table, schema='public'):
675
    return run_query(db, 'TRUNCATE '+qual_name(db, schema, table)+' CASCADE')
676 832 aaronmk
677 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
678 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
679 2415 aaronmk
    to names that will be distinct among the columns' tables.
680 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
681 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
682
    @param into The table for the new columns.
683 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
684
        columns will be included in the mapping even if they are not in cols.
685
        The tables of the provided Col objects will be changed to into, so make
686
        copies of them if you want to keep the original tables.
687
    @param as_items Whether to return a list of dict items instead of a dict
688 2383 aaronmk
    @return dict(orig_col=new_col, ...)
689
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
690 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
691
        * All mappings use the into table so its name can easily be
692 2383 aaronmk
          changed for all columns at once
693
    '''
694 2415 aaronmk
    cols = lists.uniqify(cols)
695
696 2394 aaronmk
    items = []
697 2389 aaronmk
    for col in preserve:
698 2390 aaronmk
        orig_col = copy.copy(col)
699 2392 aaronmk
        col.table = into
700 2394 aaronmk
        items.append((orig_col, col))
701
    preserve = set(preserve)
702
    for col in cols:
703 2515 aaronmk
        if col not in preserve: items.append((col, sql_gen.Col(str(col), into)))
704 2394 aaronmk
705
    if not as_items: items = dict(items)
706
    return items
707 2383 aaronmk
708 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
709 2391 aaronmk
    '''For params, see mk_flatten_mapping()
710
    @return See return value of mk_flatten_mapping()
711
    '''
712 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
713
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
714 2391 aaronmk
    run_query_into(db, *mk_select(db, joins, cols, limit=limit, start=start),
715 2392 aaronmk
        into=into)
716 2394 aaronmk
    return dict(items)
717 2391 aaronmk
718 2414 aaronmk
##### Database structure queries
719
720 2426 aaronmk
def table_row_count(db, table, recover=None):
721
    return value(run_query(db, *mk_select(db, table, [sql_gen.row_count],
722 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
723 2426 aaronmk
724 2414 aaronmk
def table_cols(db, table, recover=None):
725
    return list(col_names(select(db, table, limit=0, order_by=None,
726 2443 aaronmk
        recover=recover, log_level=4)))
727 2414 aaronmk
728 2291 aaronmk
def pkey(db, table, recover=None):
729 832 aaronmk
    '''Assumed to be first column in table'''
730 2339 aaronmk
    return table_cols(db, table, recover)[0]
731 832 aaronmk
732 2559 aaronmk
not_null_col = 'not_null_col'
733 2340 aaronmk
734
def table_not_null_col(db, table, recover=None):
735
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
736
    if not_null_col in table_cols(db, table, recover): return not_null_col
737
    else: return pkey(db, table, recover)
738
739 853 aaronmk
def index_cols(db, table, index):
740
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
741
    automatically created. When you don't know whether something is a UNIQUE
742
    constraint or a UNIQUE index, use this function.'''
743 1909 aaronmk
    module = util.root_module(db.db)
744
    if module == 'psycopg2':
745
        return list(values(run_query(db, '''\
746 853 aaronmk
SELECT attname
747 866 aaronmk
FROM
748
(
749
        SELECT attnum, attname
750
        FROM pg_index
751
        JOIN pg_class index ON index.oid = indexrelid
752
        JOIN pg_class table_ ON table_.oid = indrelid
753
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
754
        WHERE
755
            table_.relname = %(table)s
756
            AND index.relname = %(index)s
757
    UNION
758
        SELECT attnum, attname
759
        FROM
760
        (
761
            SELECT
762
                indrelid
763
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
764
                    AS indkey
765
            FROM pg_index
766
            JOIN pg_class index ON index.oid = indexrelid
767
            JOIN pg_class table_ ON table_.oid = indrelid
768
            WHERE
769
                table_.relname = %(table)s
770
                AND index.relname = %(index)s
771
        ) s
772
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
773
) s
774 853 aaronmk
ORDER BY attnum
775
''',
776 2443 aaronmk
            {'table': table, 'index': index}, cacheable=True, log_level=4)))
777 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
778
        ' database')
779 853 aaronmk
780 464 aaronmk
def constraint_cols(db, table, constraint):
781 1849 aaronmk
    module = util.root_module(db.db)
782 464 aaronmk
    if module == 'psycopg2':
783
        return list(values(run_query(db, '''\
784
SELECT attname
785
FROM pg_constraint
786
JOIN pg_class ON pg_class.oid = conrelid
787
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
788
WHERE
789
    relname = %(table)s
790
    AND conname = %(constraint)s
791
ORDER BY attnum
792
''',
793
            {'table': table, 'constraint': constraint})))
794
    else: raise NotImplementedError("Can't list constraint columns for "+module+
795
        ' database')
796
797 2096 aaronmk
row_num_col = '_row_num'
798
799 2538 aaronmk
def add_index(db, expr):
800
    '''Adds an index on a column or expression if it doesn't already exist.
801
    Currently, only function calls are supported as expressions.
802
    '''
803
    expr = copy.copy(expr) # don't modify input!
804
805
    # Extract col
806 2539 aaronmk
    if isinstance(expr, sql_gen.FunctionCall):
807
        col = expr.args[0]
808 2541 aaronmk
        expr = sql_gen.Expr(expr)
809 2538 aaronmk
    else: col = expr
810 2408 aaronmk
    assert sql_gen.is_table_col(col)
811
812 2538 aaronmk
    index = sql_gen.as_Table(str(expr))
813 2408 aaronmk
    table = col.table
814 2538 aaronmk
    col.table = None
815 2408 aaronmk
    try: run_query(db, 'CREATE INDEX '+index.to_str(db)+' ON '+table.to_str(db)
816 2538 aaronmk
        +' ('+expr.to_str(db)+')', recover=True, cacheable=True, log_level=3)
817 2408 aaronmk
    except DuplicateTableException: pass # index already existed
818
819 2594 aaronmk
def add_pkey(db, table, recover=None):
820 2406 aaronmk
    '''Makes the first column in a table the primary key.
821
    @pre The table must not already have a primary key.
822
    '''
823
    table = sql_gen.as_Table(table)
824
825 2590 aaronmk
    index = sql_gen.as_Table(sql_gen.add_suffix(table.name, '_pkey'))
826 2406 aaronmk
    col = sql_gen.to_name_only_col(pkey(db, table, recover))
827 2614 aaronmk
    try:
828
        run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD CONSTRAINT '
829
            +index.to_str(db)+' PRIMARY KEY('+col.to_str(db)+')',
830
            recover=True, cacheable=True, log_level=3,
831
            log_ignore_excs=(DuplicateTableException,))
832
    except DuplicateTableException, e:
833
        index.name = next_version(index.name)
834
        # try again with next version of name
835 2406 aaronmk
836 2086 aaronmk
def add_row_num(db, table):
837 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
838
    be the primary key.'''
839 2320 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
840 2096 aaronmk
    run_query(db, 'ALTER TABLE '+table+' ADD COLUMN '+row_num_col
841 2443 aaronmk
        +' serial NOT NULL PRIMARY KEY', log_level=3)
842 2086 aaronmk
843 2548 aaronmk
def tables(db, schema_like='public', table_like='%'):
844 1849 aaronmk
    module = util.root_module(db.db)
845 2548 aaronmk
    params = {'schema_like': schema_like, 'table_like': table_like}
846 832 aaronmk
    if module == 'psycopg2':
847 1968 aaronmk
        return values(run_query(db, '''\
848
SELECT tablename
849
FROM pg_tables
850
WHERE
851 2548 aaronmk
    schemaname LIKE %(schema_like)s
852 1968 aaronmk
    AND tablename LIKE %(table_like)s
853
ORDER BY tablename
854
''',
855
            params, cacheable=True))
856
    elif module == 'MySQLdb':
857
        return values(run_query(db, 'SHOW TABLES LIKE %(table_like)s', params,
858
            cacheable=True))
859 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
860 830 aaronmk
861 833 aaronmk
##### Database management
862
863 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
864
    '''For kw_args, see tables()'''
865
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
866 833 aaronmk
867 832 aaronmk
##### Heuristic queries
868
869 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
870 1554 aaronmk
    '''Recovers from errors.
871 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
872
    '''
873 2642 aaronmk
    row = sql_gen.ColDict(db, table, row)
874 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
875
876 471 aaronmk
    try:
877 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
878 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
879
            row_ct_ref[0] += cur.rowcount
880
        return value(cur)
881 471 aaronmk
    except DuplicateKeyException, e:
882 2639 aaronmk
        return value(select(db, table, [pkey_],
883 1069 aaronmk
            util.dict_subset_right_join(row, e.cols), recover=True))
884 471 aaronmk
885 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
886 830 aaronmk
    '''Recovers from errors'''
887 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
888 14 aaronmk
    except StopIteration:
889 40 aaronmk
        if not create: raise
890 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
891 2078 aaronmk
892 2593 aaronmk
def is_func_result(col):
893
    return col.table.name.find('(') >= 0 and col.name == 'result'
894
895 2592 aaronmk
def into_table_name(out_table, in_tables0, mapping, is_func):
896 2615 aaronmk
    def in_col_str(in_col):
897
        in_col = sql_gen.remove_col_rename(in_col)
898
        if isinstance(in_col, sql_gen.Col):
899
            table = in_col.table
900
            if table == in_tables0:
901
                in_col = sql_gen.to_name_only_col(in_col)
902
            elif is_func_result(in_col): in_col = table # omit col name
903
        return str(in_col)
904
905 2580 aaronmk
    str_ = str(out_table)
906
    if is_func:
907 2615 aaronmk
        str_ += '('
908 2580 aaronmk
909 2615 aaronmk
        try: value_in_col = mapping['value']
910
        except KeyError:
911
            str_ += ', '.join((k+'='+in_col_str(v)
912
                for k, v in mapping.iteritems()))
913
        else: str_ += in_col_str(value_in_col)
914
915
        str_ += ')'
916 2580 aaronmk
    else: str_ += '_pkeys'
917
    return str_
918
919 2508 aaronmk
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None, into=None,
920 2552 aaronmk
    default=None, is_func=False):
921 2078 aaronmk
    '''Recovers from errors.
922
    Only works under PostgreSQL (uses INSERT RETURNING).
923 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
924
        tables to join with it using the main input table's pkey
925 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
926 2612 aaronmk
        * out_table_col: str (*not* sql_gen.Col)
927
        * in_table_col: sql_gen.Col|literal-value
928 2489 aaronmk
    @param into The table to contain the output and input pkeys.
929 2574 aaronmk
        Defaults to `out_table.name+'_pkeys'`.
930 2509 aaronmk
    @param default The *output* column to use as the pkey for missing rows.
931
        If this output column does not exist in the mapping, uses None.
932 2552 aaronmk
    @param is_func Whether out_table is the name of a SQL function, not a table
933 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
934 2078 aaronmk
    '''
935 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
936 2642 aaronmk
    mapping = sql_gen.ColDict(db, out_table, mapping)
937 2312 aaronmk
938 2450 aaronmk
    def log_debug(msg): db.log_debug(msg, level=1.5)
939 2505 aaronmk
    def col_ustr(str_):
940 2567 aaronmk
        return strings.repr_no_u(sql_gen.remove_col_rename(str_))
941 2450 aaronmk
942 2486 aaronmk
    log_debug('********** New iteration **********')
943 2505 aaronmk
    log_debug('Inserting these input columns into '+strings.as_tt(
944
        out_table.to_str(db))+':\n'+strings.as_table(mapping, ustr=col_ustr))
945 2463 aaronmk
946 2382 aaronmk
    # Create input joins from list of input tables
947
    in_tables_ = in_tables[:] # don't modify input!
948
    in_tables0 = in_tables_.pop(0) # first table is separate
949 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
950 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
951 2460 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v,
952
        {in_pkey: sql_gen.join_same_not_null}) for v in in_tables_]
953 2131 aaronmk
954 2592 aaronmk
    if into == None:
955
        into = into_table_name(out_table, in_tables0, mapping, is_func)
956
    into = sql_gen.as_Table(into)
957
958 2486 aaronmk
    log_debug('Joining together input tables into temp table')
959 2395 aaronmk
    # Place in new table for speed and so don't modify input if values edited
960 2584 aaronmk
    in_table = sql_gen.Table('in')
961 2395 aaronmk
    flatten_cols = filter(sql_gen.is_table_col, mapping.values())
962
    mapping = dicts.join(mapping, flatten(db, in_table, input_joins,
963
        flatten_cols, preserve=[in_pkey_col], start=0))
964
    input_joins = [in_table]
965 2486 aaronmk
    db.log_debug('Temp table: '+strings.as_tt(in_table.to_str(db)), level=2)
966 2395 aaronmk
967 2509 aaronmk
    # Resolve default value column
968
    try: default = mapping[default]
969
    except KeyError:
970
        if default != None:
971
            db.log_debug('Default value column '
972
                +strings.as_tt(strings.repr_no_u(default))
973 2511 aaronmk
                +' does not exist in mapping, falling back to None', level=2.1)
974 2509 aaronmk
            default = None
975
976 2279 aaronmk
    out_pkey = pkey(db, out_table, recover=True)
977 2285 aaronmk
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
978 2142 aaronmk
979 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
980 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
981
982 2201 aaronmk
    pkeys_table_exists_ref = [False]
983 2420 aaronmk
    def insert_into_pkeys(joins, cols):
984
        query, params = mk_select(db, joins, cols, order_by=None, start=0)
985 2201 aaronmk
        if pkeys_table_exists_ref[0]:
986 2489 aaronmk
            insert_select(db, into, pkeys_names, query, params)
987 2201 aaronmk
        else:
988 2489 aaronmk
            run_query_into(db, query, params, into=into)
989 2201 aaronmk
            pkeys_table_exists_ref[0] = True
990
991 2429 aaronmk
    limit_ref = [None]
992 2380 aaronmk
    conds = set()
993 2233 aaronmk
    distinct_on = []
994 2325 aaronmk
    def mk_main_select(joins, cols):
995 2429 aaronmk
        return mk_select(db, joins, cols, conds, distinct_on,
996
            limit=limit_ref[0], start=0)
997 2132 aaronmk
998 2519 aaronmk
    exc_strs = set()
999 2309 aaronmk
    def log_exc(e):
1000 2519 aaronmk
        e_str = exc.str_(e, first_line_only=True)
1001
        log_debug('Caught exception: '+e_str)
1002
        assert e_str not in exc_strs # avoid infinite loops
1003
        exc_strs.add(e_str)
1004 2451 aaronmk
    def remove_all_rows():
1005 2450 aaronmk
        log_debug('Returning NULL for all rows')
1006 2429 aaronmk
        limit_ref[0] = 0 # just create an empty pkeys table
1007 2409 aaronmk
    def ignore(in_col, value):
1008 2545 aaronmk
        in_col_str = strings.as_tt(repr(in_col))
1009 2544 aaronmk
        db.log_debug('Adding index on '+in_col_str+' to enable fast filtering',
1010
            level=2.5)
1011 2537 aaronmk
        add_index(db, in_col)
1012 2545 aaronmk
        log_debug('Ignoring rows with '+in_col_str+' = '
1013
            +strings.as_tt(repr(value)))
1014 2403 aaronmk
    def remove_rows(in_col, value):
1015 2409 aaronmk
        ignore(in_col, value)
1016 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
1017
        assert cond not in conds # avoid infinite loops
1018 2380 aaronmk
        conds.add(cond)
1019 2403 aaronmk
    def invalid2null(in_col, value):
1020 2409 aaronmk
        ignore(in_col, value)
1021 2403 aaronmk
        update(db, in_table, [(in_col, None)],
1022
            sql_gen.ColValueCond(in_col, value))
1023 2245 aaronmk
1024 2589 aaronmk
    def insert_pkeys_table(which):
1025
        return sql_gen.Table(sql_gen.add_suffix(in_table.name,
1026
            '_insert_'+which+'_pkeys'))
1027
    insert_out_pkeys = insert_pkeys_table('out')
1028
    insert_in_pkeys = insert_pkeys_table('in')
1029
1030 2206 aaronmk
    # Do inserts and selects
1031 2642 aaronmk
    join_cols = sql_gen.ColDict(db, out_table)
1032 2206 aaronmk
    while True:
1033 2521 aaronmk
        if limit_ref[0] == 0: # special case
1034
            log_debug('Creating an empty pkeys table')
1035
            cur = run_query_into(db, *mk_select(db, out_table, [out_pkey],
1036
                limit=limit_ref[0]), into=insert_out_pkeys)
1037
            break # don't do main case
1038
1039 2303 aaronmk
        has_joins = join_cols != {}
1040
1041 2305 aaronmk
        # Prepare to insert new rows
1042 2325 aaronmk
        insert_joins = input_joins[:] # don't modify original!
1043 2403 aaronmk
        insert_args = dict(recover=True, cacheable=False)
1044 2303 aaronmk
        if has_joins:
1045 2317 aaronmk
            distinct_on = [v.to_Col() for v in join_cols.values()]
1046 2325 aaronmk
            insert_joins.append(sql_gen.Join(out_table, join_cols,
1047
                sql_gen.filter_out))
1048
        else:
1049 2404 aaronmk
            insert_args.update(dict(returning=out_pkey, into=insert_out_pkeys))
1050 2520 aaronmk
        main_select = mk_main_select(insert_joins, mapping.values())[0]
1051 2303 aaronmk
1052 2486 aaronmk
        log_debug('Trying to insert new rows')
1053 2206 aaronmk
        try:
1054 2518 aaronmk
            cur = insert_select(db, out_table, mapping.keys(), main_select,
1055
                **insert_args)
1056 2357 aaronmk
            break # insert successful
1057 2206 aaronmk
        except DuplicateKeyException, e:
1058 2309 aaronmk
            log_exc(e)
1059
1060 2258 aaronmk
            old_join_cols = join_cols.copy()
1061 2565 aaronmk
            join_cols.update(util.dict_subset_right_join(mapping, e.cols))
1062 2486 aaronmk
            log_debug('Ignoring existing rows, comparing on these columns:\n'
1063 2505 aaronmk
                +strings.as_inline_table(join_cols, ustr=col_ustr))
1064 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
1065 2230 aaronmk
        except NullValueException, e:
1066 2309 aaronmk
            log_exc(e)
1067
1068 2230 aaronmk
            out_col, = e.cols
1069
            try: in_col = mapping[out_col]
1070 2356 aaronmk
            except KeyError:
1071 2486 aaronmk
                log_debug('Missing mapping for NOT NULL column '+out_col)
1072 2451 aaronmk
                remove_all_rows()
1073 2403 aaronmk
            else: remove_rows(in_col, None)
1074 2542 aaronmk
        except FunctionValueException, e:
1075
            log_exc(e)
1076
1077
            func_name = e.name
1078
            value = e.value
1079
            for out_col, in_col in mapping.iteritems():
1080 2562 aaronmk
                invalid2null(sql_gen.unwrap_func_call(in_col, func_name), value)
1081 2525 aaronmk
        except MissingCastException, e:
1082
            log_exc(e)
1083
1084
            out_col = e.col
1085 2534 aaronmk
            mapping[out_col] = sql_gen.wrap_in_func(e.type, mapping[out_col])
1086 2429 aaronmk
        except DatabaseErrors, e:
1087
            log_exc(e)
1088
1089 2531 aaronmk
            msg = 'No handler for exception: '+exc.str_(e)
1090 2451 aaronmk
            warnings.warn(DbWarning(msg))
1091
            log_debug(msg)
1092
            remove_all_rows()
1093 2358 aaronmk
        # after exception handled, rerun loop with additional constraints
1094 2132 aaronmk
1095 2357 aaronmk
    if row_ct_ref != None and cur.rowcount >= 0:
1096
        row_ct_ref[0] += cur.rowcount
1097
1098
    if has_joins:
1099
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
1100 2486 aaronmk
        log_debug('Getting output table pkeys of existing/inserted rows')
1101 2420 aaronmk
        insert_into_pkeys(select_joins, pkeys_cols)
1102 2357 aaronmk
    else:
1103 2404 aaronmk
        add_row_num(db, insert_out_pkeys) # for joining with input pkeys
1104 2357 aaronmk
1105 2486 aaronmk
        log_debug('Getting input table pkeys of inserted rows')
1106 2357 aaronmk
        run_query_into(db, *mk_main_select(input_joins, [in_pkey]),
1107 2404 aaronmk
            into=insert_in_pkeys)
1108
        add_row_num(db, insert_in_pkeys) # for joining with output pkeys
1109 2357 aaronmk
1110 2428 aaronmk
        assert table_row_count(db, insert_out_pkeys) == table_row_count(db,
1111
            insert_in_pkeys)
1112
1113 2486 aaronmk
        log_debug('Combining output and input pkeys in inserted order')
1114 2404 aaronmk
        pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
1115 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
1116 2420 aaronmk
        insert_into_pkeys(pkey_joins, pkeys_names)
1117 2357 aaronmk
1118 2486 aaronmk
    db.log_debug('Adding pkey on pkeys table to enable fast joins', level=2.5)
1119 2594 aaronmk
    add_pkey(db, into)
1120 2407 aaronmk
1121 2508 aaronmk
    log_debug('Setting pkeys of missing rows to '+strings.as_tt(repr(default)))
1122 2489 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(into,
1123 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
1124
        # must use join_same_not_null or query will take forever
1125 2420 aaronmk
    insert_into_pkeys(missing_rows_joins,
1126 2508 aaronmk
        [in_pkey_col, sql_gen.NamedCol(out_pkey, default)])
1127 2357 aaronmk
1128 2489 aaronmk
    assert table_row_count(db, into) == table_row_count(db, in_table)
1129 2428 aaronmk
1130 2489 aaronmk
    return sql_gen.Col(out_pkey, into)
1131 2115 aaronmk
1132
##### Data cleanup
1133
1134 2290 aaronmk
def cleanup_table(db, table, cols):
1135 2115 aaronmk
    def esc_name_(name): return esc_name(db, name)
1136
1137 2290 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
1138 2115 aaronmk
    cols = map(esc_name_, cols)
1139
1140
    run_query(db, 'UPDATE '+table+' SET\n'+(',\n'.join(('\n'+col
1141
        +' = nullif(nullif(trim(both from '+col+"), %(null0)s), %(null1)s)"
1142
            for col in cols))),
1143
        dict(null0='', null1=r'\N'))