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 2671 aaronmk
        self._notices_seen = set()
166 1869 aaronmk
167
    def __getattr__(self, name):
168
        if name == '__dict__': raise Exception('getting __dict__')
169
        if name == 'db': return self._db()
170
        else: raise AttributeError()
171
172
    def __getstate__(self):
173
        state = copy.copy(self.__dict__) # shallow copy
174 1915 aaronmk
        state['log_debug'] = None # don't pickle the debug callback
175 1869 aaronmk
        state['_DbConn__db'] = None # don't pickle the connection
176
        return state
177
178 2165 aaronmk
    def connected(self): return self.__db != None
179
180 1869 aaronmk
    def _db(self):
181
        if self.__db == None:
182
            # Process db_config
183
            db_config = self.db_config.copy() # don't modify input!
184 2097 aaronmk
            schemas = db_config.pop('schemas', None)
185 1869 aaronmk
            module_name, mappings = db_engines[db_config.pop('engine')]
186
            module = __import__(module_name)
187
            _add_module(module)
188
            for orig, new in mappings.iteritems():
189
                try: util.rename_key(db_config, orig, new)
190
                except KeyError: pass
191
192
            # Connect
193
            self.__db = module.connect(**db_config)
194
195
            # Configure connection
196 2234 aaronmk
            if self.serializable and not self.autocommit: run_raw_query(self,
197
                'SET TRANSACTION ISOLATION LEVEL SERIALIZABLE')
198 2101 aaronmk
            if schemas != None:
199
                schemas_ = ''.join((esc_name(self, s)+', '
200
                    for s in schemas.split(',')))
201
                run_raw_query(self, "SELECT set_config('search_path', \
202
%s || current_setting('search_path'), false)", [schemas_])
203 1869 aaronmk
204
        return self.__db
205 1889 aaronmk
206 1891 aaronmk
    class DbCursor(Proxy):
207 1927 aaronmk
        def __init__(self, outer):
208 1891 aaronmk
            Proxy.__init__(self, outer.db.cursor())
209 2191 aaronmk
            self.outer = outer
210 1927 aaronmk
            self.query_results = outer.query_results
211 1894 aaronmk
            self.query_lookup = None
212 1891 aaronmk
            self.result = []
213 1889 aaronmk
214 1894 aaronmk
        def execute(self, query, params=None):
215 1930 aaronmk
            self._is_insert = query.upper().find('INSERT') >= 0
216 1894 aaronmk
            self.query_lookup = _query_lookup(query, params)
217 2148 aaronmk
            try:
218 2191 aaronmk
                try:
219
                    return_value = self.inner.execute(query, params)
220
                    self.outer.do_autocommit()
221 2648 aaronmk
                finally: self.query = get_cur_query(self.inner, query, params)
222 1904 aaronmk
            except Exception, e:
223 2170 aaronmk
                _add_cursor_info(e, self, query, params)
224 1904 aaronmk
                self.result = e # cache the exception as the result
225
                self._cache_result()
226
                raise
227 1930 aaronmk
            # Fetch all rows so result will be cached
228
            if self.rowcount == 0 and not self._is_insert: consume_rows(self)
229 1894 aaronmk
            return return_value
230
231 1891 aaronmk
        def fetchone(self):
232
            row = self.inner.fetchone()
233 1899 aaronmk
            if row != None: self.result.append(row)
234
            # otherwise, fetched all rows
235 1904 aaronmk
            else: self._cache_result()
236
            return row
237
238
        def _cache_result(self):
239 1906 aaronmk
            # For inserts, only cache exceptions since inserts are not
240
            # idempotent, but an invalid insert will always be invalid
241 1930 aaronmk
            if self.query_results != None and (not self._is_insert
242 1906 aaronmk
                or isinstance(self.result, Exception)):
243
244 1894 aaronmk
                assert self.query_lookup != None
245 1916 aaronmk
                self.query_results[self.query_lookup] = self.CacheCursor(
246
                    util.dict_subset(dicts.AttrsDictView(self),
247
                    ['query', 'result', 'rowcount', 'description']))
248 1906 aaronmk
249 1916 aaronmk
        class CacheCursor:
250
            def __init__(self, cached_result): self.__dict__ = cached_result
251
252 1927 aaronmk
            def execute(self, *args, **kw_args):
253 1916 aaronmk
                if isinstance(self.result, Exception): raise self.result
254
                # otherwise, result is a rows list
255
                self.iter = iter(self.result)
256
257
            def fetchone(self):
258
                try: return self.iter.next()
259
                except StopIteration: return None
260 1891 aaronmk
261 2212 aaronmk
    def esc_value(self, value):
262 2663 aaronmk
        try: str_ = self.mogrify('%s', [value])
263
        except NotImplementedError, e:
264
            module = util.root_module(self.db)
265
            if module == 'MySQLdb':
266
                import _mysql
267
                str_ = _mysql.escape_string(value)
268
            else: raise e
269 2374 aaronmk
        return strings.to_unicode(str_)
270 2212 aaronmk
271 2347 aaronmk
    def esc_name(self, name): return esc_name(self, name) # calls global func
272
273 2665 aaronmk
    def can_mogrify(self):
274 2663 aaronmk
        module = util.root_module(self.db)
275 2665 aaronmk
        return module == 'psycopg2'
276 2663 aaronmk
277 2665 aaronmk
    def mogrify(self, query, params=None):
278
        if self.can_mogrify(): return self.db.cursor().mogrify(query, params)
279
        else: raise NotImplementedError("Can't mogrify query")
280
281 2671 aaronmk
    def print_notices(self):
282
        for msg in self.db.notices:
283
            if msg not in self._notices_seen:
284
                self._notices_seen.add(msg)
285
                self.log_debug(msg, level=2)
286
287 2445 aaronmk
    def run_query(self, query, params=None, cacheable=False, log_level=2,
288 2464 aaronmk
        debug_msg_ref=None):
289 2445 aaronmk
        '''
290 2464 aaronmk
        @param log_ignore_excs The log_level will be increased by 2 if the query
291
            throws one of these exceptions.
292 2664 aaronmk
        @param debug_msg_ref If specified, the log message will be returned in
293
            this instead of being output. This allows you to filter log messages
294
            depending on the result of the query.
295 2445 aaronmk
        '''
296 2167 aaronmk
        assert query != None
297
298 2047 aaronmk
        if not self.caching: cacheable = False
299 1903 aaronmk
        used_cache = False
300 2664 aaronmk
301
        def log_msg(query):
302
            if used_cache: cache_status = 'cache hit'
303
            elif cacheable: cache_status = 'cache miss'
304
            else: cache_status = 'non-cacheable'
305
            return 'DB query: '+cache_status+':\n'+strings.as_code(query, 'SQL')
306
307 1903 aaronmk
        try:
308 1927 aaronmk
            # Get cursor
309
            if cacheable:
310
                query_lookup = _query_lookup(query, params)
311
                try:
312
                    cur = self.query_results[query_lookup]
313
                    used_cache = True
314
                except KeyError: cur = self.DbCursor(self)
315
            else: cur = self.db.cursor()
316
317 2664 aaronmk
            # Log query
318
            if self.debug and debug_msg_ref == None: # log before running
319
                self.log_debug(log_msg(query), log_level)
320
321 1927 aaronmk
            # Run query
322 2148 aaronmk
            cur.execute(query, params)
323 1903 aaronmk
        finally:
324 2671 aaronmk
            self.print_notices()
325 2664 aaronmk
            if self.debug and debug_msg_ref != None: # return after running
326
                debug_msg_ref[0] = log_msg(str(get_cur_query(cur, query,
327
                    params)))
328 1903 aaronmk
329
        return cur
330 1914 aaronmk
331
    def is_cached(self, query, params=None):
332
        return _query_lookup(query, params) in self.query_results
333 2139 aaronmk
334
    def with_savepoint(self, func):
335 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
336 2443 aaronmk
        self.run_query('SAVEPOINT '+savepoint, log_level=4)
337 2139 aaronmk
        self._savepoint += 1
338
        try:
339
            try: return_val = func()
340
            finally:
341
                self._savepoint -= 1
342
                assert self._savepoint >= 0
343
        except:
344 2443 aaronmk
            self.run_query('ROLLBACK TO SAVEPOINT '+savepoint, log_level=4)
345 2139 aaronmk
            raise
346
        else:
347 2443 aaronmk
            self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
348 2191 aaronmk
            self.do_autocommit()
349 2139 aaronmk
            return return_val
350 2191 aaronmk
351
    def do_autocommit(self):
352
        '''Autocommits if outside savepoint'''
353
        assert self._savepoint >= 0
354
        if self.autocommit and self._savepoint == 0:
355 2668 aaronmk
            self.log_debug('Autocommitting')
356 2191 aaronmk
            self.db.commit()
357 2643 aaronmk
358
    def col_default(self, col):
359
        table = sql_gen.Table('columns', 'information_schema')
360
361
        conds = [('table_name', col.table.name), ('column_name', col.name)]
362
        schema = col.table.schema
363
        if schema != None: conds.append(('table_schema', schema))
364
365 2660 aaronmk
        return sql_gen.as_Code(value(select(self, table, ['column_default'],
366
            conds, order_by='table_schema', limit=1, log_level=3)))
367 2643 aaronmk
            # TODO: order_by search_path schema order
368 1849 aaronmk
369 1869 aaronmk
connect = DbConn
370
371 832 aaronmk
##### Querying
372
373 1894 aaronmk
def run_raw_query(db, *args, **kw_args):
374 2085 aaronmk
    '''For params, see DbConn.run_query()'''
375 1894 aaronmk
    return db.run_query(*args, **kw_args)
376 11 aaronmk
377 2068 aaronmk
def mogrify(db, query, params):
378
    module = util.root_module(db.db)
379
    if module == 'psycopg2': return db.db.cursor().mogrify(query, params)
380
    else: raise NotImplementedError("Can't mogrify query for "+module+
381
        ' database')
382
383 832 aaronmk
##### Recoverable querying
384 15 aaronmk
385 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
386 11 aaronmk
387 2464 aaronmk
def run_query(db, query, params=None, recover=None, cacheable=False,
388
    log_level=2, log_ignore_excs=None, **kw_args):
389 2441 aaronmk
    '''For params, see run_raw_query()'''
390 830 aaronmk
    if recover == None: recover = False
391 2464 aaronmk
    if log_ignore_excs == None: log_ignore_excs = ()
392
    log_ignore_excs = tuple(log_ignore_excs)
393 830 aaronmk
394 2666 aaronmk
    debug_msg_ref = None # usually, db.run_query() logs query before running it
395
    # But if filtering with log_ignore_excs, wait until after exception parsing
396
    if log_ignore_excs != () or not db.can_mogrify(): debug_msg_ref = [None]
397
398 2148 aaronmk
    try:
399 2464 aaronmk
        try:
400
            def run(): return run_raw_query(db, query, params, cacheable,
401
                log_level, debug_msg_ref, **kw_args)
402
            if recover and not db.is_cached(query, params):
403
                return with_savepoint(db, run)
404
            else: return run() # don't need savepoint if cached
405
        except Exception, e:
406
            if not recover: raise # need savepoint to run index_cols()
407
            msg = exc.str_(e)
408
409
            match = re.search(r'duplicate key value violates unique constraint '
410 2493 aaronmk
                r'"((_?[^\W_]+)_.+?)"', msg)
411 2464 aaronmk
            if match:
412
                constraint, table = match.groups()
413
                try: cols = index_cols(db, table, constraint)
414
                except NotImplementedError: raise e
415
                else: raise DuplicateKeyException(constraint, cols, e)
416
417 2493 aaronmk
            match = re.search(r'null value in column "(.+?)" violates not-null'
418 2464 aaronmk
                r' constraint', msg)
419
            if match: raise NullValueException('NOT NULL', [match.group(1)], e)
420
421
            match = re.search(r'\b(?:invalid input (?:syntax|value)\b.*?'
422
                r'|date/time field value out of range): "(.+?)"\n'
423 2535 aaronmk
                r'(?:(?s).*?)\bfunction "(.+?)"', msg)
424 2464 aaronmk
            if match:
425
                value, name = match.groups()
426
                raise FunctionValueException(name, strings.to_unicode(value), e)
427
428 2526 aaronmk
            match = re.search(r'column "(.+?)" is of type (.+?) but expression '
429 2523 aaronmk
                r'is of type', msg)
430
            if match:
431
                col, type_ = match.groups()
432
                raise MissingCastException(type_, col, e)
433
434 2493 aaronmk
            match = re.search(r'relation "(.+?)" already exists', msg)
435 2464 aaronmk
            if match: raise DuplicateTableException(match.group(1), e)
436
437 2493 aaronmk
            match = re.search(r'function "(.+?)" already exists', msg)
438 2464 aaronmk
            if match: raise DuplicateFunctionException(match.group(1), e)
439
440
            raise # no specific exception raised
441
    except log_ignore_excs:
442
        log_level += 2
443
        raise
444
    finally:
445 2666 aaronmk
        if debug_msg_ref != None and debug_msg_ref[0] != None:
446
            db.log_debug(debug_msg_ref[0], log_level)
447 830 aaronmk
448 832 aaronmk
##### Basic queries
449
450 2153 aaronmk
def next_version(name):
451 2163 aaronmk
    version = 1 # first existing name was version 0
452 2586 aaronmk
    match = re.match(r'^(.*)#(\d+)$', name)
453 2153 aaronmk
    if match:
454 2586 aaronmk
        name, version = match.groups()
455
        version = int(version)+1
456 2588 aaronmk
    return sql_gen.add_suffix(name, '#'+str(version))
457 2153 aaronmk
458 2386 aaronmk
def run_query_into(db, query, params, into=None, *args, **kw_args):
459 2085 aaronmk
    '''Outputs a query to a temp table.
460
    For params, see run_query().
461
    '''
462 2386 aaronmk
    if into == None: return run_query(db, query, params, *args, **kw_args)
463 2085 aaronmk
    else: # place rows in temp table
464 2386 aaronmk
        assert isinstance(into, sql_gen.Table)
465 2385 aaronmk
466 2153 aaronmk
        kw_args['recover'] = True
467 2464 aaronmk
        kw_args.setdefault('log_ignore_excs', (DuplicateTableException,))
468 2440 aaronmk
469 2468 aaronmk
        temp = not db.autocommit # tables are permanent in autocommit mode
470 2440 aaronmk
        # "temporary tables cannot specify a schema name", so remove schema
471
        if temp: into.schema = None
472
473 2153 aaronmk
        while True:
474
            try:
475 2194 aaronmk
                create_query = 'CREATE'
476 2440 aaronmk
                if temp: create_query += ' TEMP'
477 2467 aaronmk
                create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
478 2194 aaronmk
479
                return run_query(db, create_query, params, *args, **kw_args)
480 2153 aaronmk
                    # CREATE TABLE AS sets rowcount to # rows in query
481
            except DuplicateTableException, e:
482 2386 aaronmk
                into.name = next_version(into.name)
483 2153 aaronmk
                # try again with next version of name
484 2085 aaronmk
485 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
486
487 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
488
489 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
490 2293 aaronmk
    start=None, order_by=order_by_pkey, default_table=None):
491 1981 aaronmk
    '''
492 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
493 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
494 1981 aaronmk
    @param fields Use None to select all fields in the table
495 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
496 2379 aaronmk
        * container can be any iterable type
497 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
498
        * compare_right_side: sql_gen.ValueCond|literal value
499 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
500
        use all columns
501 2054 aaronmk
    @return tuple(query, params)
502 1981 aaronmk
    '''
503 2315 aaronmk
    # Parse tables param
504 2121 aaronmk
    if not lists.is_seq(tables): tables = [tables]
505 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
506 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
507 2121 aaronmk
508 2315 aaronmk
    # Parse other params
509 2376 aaronmk
    if conds == None: conds = []
510 2650 aaronmk
    elif dicts.is_dict(conds): conds = conds.items()
511 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
512 135 aaronmk
    assert limit == None or type(limit) == int
513 865 aaronmk
    assert start == None or type(start) == int
514 2315 aaronmk
    if order_by is order_by_pkey:
515
        if distinct_on != []: order_by = None
516
        else: order_by = pkey(db, table0, recover=True)
517 865 aaronmk
518 2315 aaronmk
    query = 'SELECT'
519 2056 aaronmk
520 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
521 2056 aaronmk
522 2200 aaronmk
    # DISTINCT ON columns
523 2233 aaronmk
    if distinct_on != []:
524 2467 aaronmk
        query += '\nDISTINCT'
525 2254 aaronmk
        if distinct_on is not distinct_on_all:
526 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
527
528
    # Columns
529 2467 aaronmk
    query += '\n'
530 1135 aaronmk
    if fields == None: query += '*'
531 2479 aaronmk
    else: query += '\n, '.join(map(parse_col, fields))
532 2200 aaronmk
533
    # Main table
534 2467 aaronmk
    query += '\nFROM '+table0.to_str(db)
535 865 aaronmk
536 2122 aaronmk
    # Add joins
537 2271 aaronmk
    left_table = table0
538 2263 aaronmk
    for join_ in tables:
539
        table = join_.table
540 2238 aaronmk
541 2343 aaronmk
        # Parse special values
542
        if join_.type_ is sql_gen.filter_out: # filter no match
543 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
544
                None))
545 2343 aaronmk
546 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
547 2122 aaronmk
548
        left_table = table
549
550 865 aaronmk
    missing = True
551 2376 aaronmk
    if conds != []:
552 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
553
        else: whitespace = '\n'
554 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
555
            .to_str(db) for l, r in conds], 'WHERE')
556 865 aaronmk
        missing = False
557 2227 aaronmk
    if order_by != None:
558 2467 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
559
    if limit != None: query += '\nLIMIT '+str(limit); missing = False
560 865 aaronmk
    if start != None:
561 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
562 865 aaronmk
        missing = False
563
    if missing: warnings.warn(DbWarning(
564
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
565
566 2315 aaronmk
    return (query, [])
567 11 aaronmk
568 2054 aaronmk
def select(db, *args, **kw_args):
569
    '''For params, see mk_select() and run_query()'''
570
    recover = kw_args.pop('recover', None)
571
    cacheable = kw_args.pop('cacheable', True)
572 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
573 2054 aaronmk
574
    query, params = mk_select(db, *args, **kw_args)
575 2442 aaronmk
    return run_query(db, query, params, recover, cacheable, log_level=log_level)
576 2054 aaronmk
577 2066 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, params=None,
578 2292 aaronmk
    returning=None, embeddable=False):
579 1960 aaronmk
    '''
580
    @param returning str|None An inserted column (such as pkey) to return
581 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
582 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
583
        query will be fully cached, not just if it raises an exception.
584 1960 aaronmk
    '''
585 2328 aaronmk
    table = sql_gen.as_Table(table)
586 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
587 2646 aaronmk
    if cols != None:
588
        cols = [sql_gen.to_name_only_col(v, table).to_str(db) for v in cols]
589 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
590 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
591 2063 aaronmk
592
    # Build query
593 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
594
    query = first_line
595 2467 aaronmk
    if cols != None: query += '\n('+', '.join(cols)+')'
596
    query += '\n'+select_query
597 2063 aaronmk
598
    if returning != None:
599 2327 aaronmk
        returning_name = copy.copy(returning)
600
        returning_name.table = None
601
        returning_name = returning_name.to_str(db)
602 2467 aaronmk
        query += '\nRETURNING '+returning_name
603 2063 aaronmk
604 2070 aaronmk
    if embeddable:
605 2327 aaronmk
        assert returning != None
606
607 2070 aaronmk
        # Create function
608 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
609 2327 aaronmk
        return_type = 'SETOF '+returning.to_str(db)+'%TYPE'
610 2189 aaronmk
        while True:
611
            try:
612 2327 aaronmk
                func_schema = None
613 2468 aaronmk
                if not db.autocommit: func_schema = 'pg_temp'
614 2327 aaronmk
                function = sql_gen.Table(function_name, func_schema).to_str(db)
615 2194 aaronmk
616 2189 aaronmk
                function_query = '''\
617 2467 aaronmk
CREATE FUNCTION '''+function+'''()
618
RETURNS '''+return_type+'''
619
LANGUAGE sql
620
AS $$
621
'''+mogrify(db, query, params)+''';
622
$$;
623 2070 aaronmk
'''
624 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
625 2464 aaronmk
                    log_ignore_excs=(DuplicateFunctionException,))
626 2189 aaronmk
                break # this version was successful
627
            except DuplicateFunctionException, e:
628
                function_name = next_version(function_name)
629
                # try again with next version of name
630 2070 aaronmk
631 2337 aaronmk
        # Return query that uses function
632
        func_table = sql_gen.NamedTable('f', sql_gen.CustomCode(function+'()'),
633
            [returning_name]) # AS clause requires function alias
634
        return mk_select(db, func_table, start=0, order_by=None)
635 2070 aaronmk
636 2066 aaronmk
    return (query, params)
637
638
def insert_select(db, *args, **kw_args):
639 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
640 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
641
        values in
642 2072 aaronmk
    '''
643 2386 aaronmk
    into = kw_args.pop('into', None)
644
    if into != None: kw_args['embeddable'] = True
645 2066 aaronmk
    recover = kw_args.pop('recover', None)
646
    cacheable = kw_args.pop('cacheable', True)
647 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
648 2066 aaronmk
649
    query, params = mk_insert_select(db, *args, **kw_args)
650 2386 aaronmk
    return run_query_into(db, query, params, into, recover=recover,
651 2673 aaronmk
        cacheable=cacheable, log_level=log_level)
652 2063 aaronmk
653 2066 aaronmk
default = object() # tells insert() to use the default value for a column
654
655 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
656 2085 aaronmk
    '''For params, see insert_select()'''
657 1960 aaronmk
    if lists.is_seq(row): cols = None
658
    else:
659
        cols = row.keys()
660
        row = row.values()
661
    row = list(row) # ensure that "!= []" works
662
663 1961 aaronmk
    # Check for special values
664
    labels = []
665
    values = []
666
    for value in row:
667 2647 aaronmk
        value = sql_gen.remove_col_rename(sql_gen.as_Value(value)).value
668 2254 aaronmk
        if value is default: labels.append('DEFAULT')
669 1961 aaronmk
        else:
670
            labels.append('%s')
671
            values.append(value)
672
673
    # Build query
674 2467 aaronmk
    if values != []: query = 'VALUES ('+(', '.join(labels))+')'
675 2063 aaronmk
    else: query = None
676 1554 aaronmk
677 2064 aaronmk
    return insert_select(db, table, cols, query, values, *args, **kw_args)
678 11 aaronmk
679 2402 aaronmk
def mk_update(db, table, changes=None, cond=None):
680
    '''
681
    @param changes [(col, new_value),...]
682
        * container can be any iterable type
683
        * col: sql_gen.Code|str (for col name)
684
        * new_value: sql_gen.Code|literal value
685
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
686
    @return str query
687
    '''
688
    query = 'UPDATE '+sql_gen.as_Table(table).to_str(db)+'\nSET\n'
689 2405 aaronmk
    query += ',\n'.join((sql_gen.to_name_only_col(col, table).to_str(db)+' = '
690 2402 aaronmk
        +sql_gen.as_Value(new_value).to_str(db) for col, new_value in changes))
691 2467 aaronmk
    if cond != None: query += '\nWHERE\n'+cond.to_str(db)
692 2402 aaronmk
693
    return query
694
695
def update(db, *args, **kw_args):
696
    '''For params, see mk_update() and run_query()'''
697
    recover = kw_args.pop('recover', None)
698
699
    return run_query(db, mk_update(db, *args, **kw_args), [], recover)
700
701 135 aaronmk
def last_insert_id(db):
702 1849 aaronmk
    module = util.root_module(db.db)
703 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
704
    elif module == 'MySQLdb': return db.insert_id()
705
    else: return None
706 13 aaronmk
707 1968 aaronmk
def truncate(db, table, schema='public'):
708
    return run_query(db, 'TRUNCATE '+qual_name(db, schema, table)+' CASCADE')
709 832 aaronmk
710 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
711 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
712 2415 aaronmk
    to names that will be distinct among the columns' tables.
713 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
714 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
715
    @param into The table for the new columns.
716 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
717
        columns will be included in the mapping even if they are not in cols.
718
        The tables of the provided Col objects will be changed to into, so make
719
        copies of them if you want to keep the original tables.
720
    @param as_items Whether to return a list of dict items instead of a dict
721 2383 aaronmk
    @return dict(orig_col=new_col, ...)
722
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
723 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
724
        * All mappings use the into table so its name can easily be
725 2383 aaronmk
          changed for all columns at once
726
    '''
727 2415 aaronmk
    cols = lists.uniqify(cols)
728
729 2394 aaronmk
    items = []
730 2389 aaronmk
    for col in preserve:
731 2390 aaronmk
        orig_col = copy.copy(col)
732 2392 aaronmk
        col.table = into
733 2394 aaronmk
        items.append((orig_col, col))
734
    preserve = set(preserve)
735
    for col in cols:
736 2515 aaronmk
        if col not in preserve: items.append((col, sql_gen.Col(str(col), into)))
737 2394 aaronmk
738
    if not as_items: items = dict(items)
739
    return items
740 2383 aaronmk
741 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
742 2391 aaronmk
    '''For params, see mk_flatten_mapping()
743
    @return See return value of mk_flatten_mapping()
744
    '''
745 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
746
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
747 2391 aaronmk
    run_query_into(db, *mk_select(db, joins, cols, limit=limit, start=start),
748 2392 aaronmk
        into=into)
749 2394 aaronmk
    return dict(items)
750 2391 aaronmk
751 2414 aaronmk
##### Database structure queries
752
753 2426 aaronmk
def table_row_count(db, table, recover=None):
754
    return value(run_query(db, *mk_select(db, table, [sql_gen.row_count],
755 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
756 2426 aaronmk
757 2414 aaronmk
def table_cols(db, table, recover=None):
758
    return list(col_names(select(db, table, limit=0, order_by=None,
759 2443 aaronmk
        recover=recover, log_level=4)))
760 2414 aaronmk
761 2291 aaronmk
def pkey(db, table, recover=None):
762 832 aaronmk
    '''Assumed to be first column in table'''
763 2339 aaronmk
    return table_cols(db, table, recover)[0]
764 832 aaronmk
765 2559 aaronmk
not_null_col = 'not_null_col'
766 2340 aaronmk
767
def table_not_null_col(db, table, recover=None):
768
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
769
    if not_null_col in table_cols(db, table, recover): return not_null_col
770
    else: return pkey(db, table, recover)
771
772 853 aaronmk
def index_cols(db, table, index):
773
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
774
    automatically created. When you don't know whether something is a UNIQUE
775
    constraint or a UNIQUE index, use this function.'''
776 1909 aaronmk
    module = util.root_module(db.db)
777
    if module == 'psycopg2':
778
        return list(values(run_query(db, '''\
779 853 aaronmk
SELECT attname
780 866 aaronmk
FROM
781
(
782
        SELECT attnum, attname
783
        FROM pg_index
784
        JOIN pg_class index ON index.oid = indexrelid
785
        JOIN pg_class table_ ON table_.oid = indrelid
786
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
787
        WHERE
788
            table_.relname = %(table)s
789
            AND index.relname = %(index)s
790
    UNION
791
        SELECT attnum, attname
792
        FROM
793
        (
794
            SELECT
795
                indrelid
796
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
797
                    AS indkey
798
            FROM pg_index
799
            JOIN pg_class index ON index.oid = indexrelid
800
            JOIN pg_class table_ ON table_.oid = indrelid
801
            WHERE
802
                table_.relname = %(table)s
803
                AND index.relname = %(index)s
804
        ) s
805
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
806
) s
807 853 aaronmk
ORDER BY attnum
808
''',
809 2443 aaronmk
            {'table': table, 'index': index}, cacheable=True, log_level=4)))
810 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
811
        ' database')
812 853 aaronmk
813 464 aaronmk
def constraint_cols(db, table, constraint):
814 1849 aaronmk
    module = util.root_module(db.db)
815 464 aaronmk
    if module == 'psycopg2':
816
        return list(values(run_query(db, '''\
817
SELECT attname
818
FROM pg_constraint
819
JOIN pg_class ON pg_class.oid = conrelid
820
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
821
WHERE
822
    relname = %(table)s
823
    AND conname = %(constraint)s
824
ORDER BY attnum
825
''',
826
            {'table': table, 'constraint': constraint})))
827
    else: raise NotImplementedError("Can't list constraint columns for "+module+
828
        ' database')
829
830 2096 aaronmk
row_num_col = '_row_num'
831
832 2538 aaronmk
def add_index(db, expr):
833
    '''Adds an index on a column or expression if it doesn't already exist.
834
    Currently, only function calls are supported as expressions.
835
    '''
836
    expr = copy.copy(expr) # don't modify input!
837
838
    # Extract col
839 2539 aaronmk
    if isinstance(expr, sql_gen.FunctionCall):
840
        col = expr.args[0]
841 2541 aaronmk
        expr = sql_gen.Expr(expr)
842 2538 aaronmk
    else: col = expr
843 2408 aaronmk
    assert sql_gen.is_table_col(col)
844
845 2538 aaronmk
    index = sql_gen.as_Table(str(expr))
846 2408 aaronmk
    table = col.table
847 2538 aaronmk
    col.table = None
848 2408 aaronmk
    try: run_query(db, 'CREATE INDEX '+index.to_str(db)+' ON '+table.to_str(db)
849 2538 aaronmk
        +' ('+expr.to_str(db)+')', recover=True, cacheable=True, log_level=3)
850 2408 aaronmk
    except DuplicateTableException: pass # index already existed
851
852 2594 aaronmk
def add_pkey(db, table, recover=None):
853 2406 aaronmk
    '''Makes the first column in a table the primary key.
854
    @pre The table must not already have a primary key.
855
    '''
856
    table = sql_gen.as_Table(table)
857
858 2590 aaronmk
    index = sql_gen.as_Table(sql_gen.add_suffix(table.name, '_pkey'))
859 2406 aaronmk
    col = sql_gen.to_name_only_col(pkey(db, table, recover))
860 2614 aaronmk
    try:
861
        run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD CONSTRAINT '
862
            +index.to_str(db)+' PRIMARY KEY('+col.to_str(db)+')',
863
            recover=True, cacheable=True, log_level=3,
864
            log_ignore_excs=(DuplicateTableException,))
865
    except DuplicateTableException, e:
866
        index.name = next_version(index.name)
867
        # try again with next version of name
868 2406 aaronmk
869 2086 aaronmk
def add_row_num(db, table):
870 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
871
    be the primary key.'''
872 2320 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
873 2096 aaronmk
    run_query(db, 'ALTER TABLE '+table+' ADD COLUMN '+row_num_col
874 2443 aaronmk
        +' serial NOT NULL PRIMARY KEY', log_level=3)
875 2086 aaronmk
876 2548 aaronmk
def tables(db, schema_like='public', table_like='%'):
877 1849 aaronmk
    module = util.root_module(db.db)
878 2548 aaronmk
    params = {'schema_like': schema_like, 'table_like': table_like}
879 832 aaronmk
    if module == 'psycopg2':
880 1968 aaronmk
        return values(run_query(db, '''\
881
SELECT tablename
882
FROM pg_tables
883
WHERE
884 2548 aaronmk
    schemaname LIKE %(schema_like)s
885 1968 aaronmk
    AND tablename LIKE %(table_like)s
886
ORDER BY tablename
887
''',
888
            params, cacheable=True))
889
    elif module == 'MySQLdb':
890
        return values(run_query(db, 'SHOW TABLES LIKE %(table_like)s', params,
891
            cacheable=True))
892 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
893 830 aaronmk
894 833 aaronmk
##### Database management
895
896 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
897
    '''For kw_args, see tables()'''
898
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
899 833 aaronmk
900 832 aaronmk
##### Heuristic queries
901
902 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
903 1554 aaronmk
    '''Recovers from errors.
904 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
905
    '''
906 2642 aaronmk
    row = sql_gen.ColDict(db, table, row)
907 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
908
909 471 aaronmk
    try:
910 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
911 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
912
            row_ct_ref[0] += cur.rowcount
913
        return value(cur)
914 471 aaronmk
    except DuplicateKeyException, e:
915 2653 aaronmk
        row = sql_gen.ColDict(db, table,
916
            util.dict_subset_right_join(row, e.cols))
917
        return value(select(db, table, [pkey_], row, recover=True))
918 471 aaronmk
919 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
920 830 aaronmk
    '''Recovers from errors'''
921 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
922 14 aaronmk
    except StopIteration:
923 40 aaronmk
        if not create: raise
924 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
925 2078 aaronmk
926 2593 aaronmk
def is_func_result(col):
927
    return col.table.name.find('(') >= 0 and col.name == 'result'
928
929 2592 aaronmk
def into_table_name(out_table, in_tables0, mapping, is_func):
930 2615 aaronmk
    def in_col_str(in_col):
931
        in_col = sql_gen.remove_col_rename(in_col)
932
        if isinstance(in_col, sql_gen.Col):
933
            table = in_col.table
934
            if table == in_tables0:
935
                in_col = sql_gen.to_name_only_col(in_col)
936
            elif is_func_result(in_col): in_col = table # omit col name
937
        return str(in_col)
938
939 2580 aaronmk
    str_ = str(out_table)
940
    if is_func:
941 2615 aaronmk
        str_ += '('
942 2580 aaronmk
943 2615 aaronmk
        try: value_in_col = mapping['value']
944
        except KeyError:
945 2654 aaronmk
            str_ += ', '.join((str(k)+'='+in_col_str(v)
946 2615 aaronmk
                for k, v in mapping.iteritems()))
947
        else: str_ += in_col_str(value_in_col)
948
949
        str_ += ')'
950 2580 aaronmk
    else: str_ += '_pkeys'
951
    return str_
952
953 2508 aaronmk
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None, into=None,
954 2552 aaronmk
    default=None, is_func=False):
955 2078 aaronmk
    '''Recovers from errors.
956
    Only works under PostgreSQL (uses INSERT RETURNING).
957 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
958
        tables to join with it using the main input table's pkey
959 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
960 2612 aaronmk
        * out_table_col: str (*not* sql_gen.Col)
961
        * in_table_col: sql_gen.Col|literal-value
962 2489 aaronmk
    @param into The table to contain the output and input pkeys.
963 2574 aaronmk
        Defaults to `out_table.name+'_pkeys'`.
964 2509 aaronmk
    @param default The *output* column to use as the pkey for missing rows.
965
        If this output column does not exist in the mapping, uses None.
966 2552 aaronmk
    @param is_func Whether out_table is the name of a SQL function, not a table
967 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
968 2078 aaronmk
    '''
969 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
970 2312 aaronmk
971 2450 aaronmk
    def log_debug(msg): db.log_debug(msg, level=1.5)
972 2505 aaronmk
    def col_ustr(str_):
973 2567 aaronmk
        return strings.repr_no_u(sql_gen.remove_col_rename(str_))
974 2450 aaronmk
975 2486 aaronmk
    log_debug('********** New iteration **********')
976 2505 aaronmk
    log_debug('Inserting these input columns into '+strings.as_tt(
977
        out_table.to_str(db))+':\n'+strings.as_table(mapping, ustr=col_ustr))
978 2463 aaronmk
979 2382 aaronmk
    # Create input joins from list of input tables
980
    in_tables_ = in_tables[:] # don't modify input!
981
    in_tables0 = in_tables_.pop(0) # first table is separate
982 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
983 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
984 2460 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v,
985
        {in_pkey: sql_gen.join_same_not_null}) for v in in_tables_]
986 2131 aaronmk
987 2592 aaronmk
    if into == None:
988
        into = into_table_name(out_table, in_tables0, mapping, is_func)
989
    into = sql_gen.as_Table(into)
990
991 2486 aaronmk
    log_debug('Joining together input tables into temp table')
992 2395 aaronmk
    # Place in new table for speed and so don't modify input if values edited
993 2584 aaronmk
    in_table = sql_gen.Table('in')
994 2395 aaronmk
    flatten_cols = filter(sql_gen.is_table_col, mapping.values())
995
    mapping = dicts.join(mapping, flatten(db, in_table, input_joins,
996
        flatten_cols, preserve=[in_pkey_col], start=0))
997
    input_joins = [in_table]
998 2486 aaronmk
    db.log_debug('Temp table: '+strings.as_tt(in_table.to_str(db)), level=2)
999 2395 aaronmk
1000 2656 aaronmk
    mapping = sql_gen.ColDict(db, out_table, mapping)
1001
        # after applying dicts.join() because that returns a plain dict
1002
1003 2509 aaronmk
    # Resolve default value column
1004
    try: default = mapping[default]
1005
    except KeyError:
1006
        if default != None:
1007
            db.log_debug('Default value column '
1008
                +strings.as_tt(strings.repr_no_u(default))
1009 2511 aaronmk
                +' does not exist in mapping, falling back to None', level=2.1)
1010 2509 aaronmk
            default = None
1011
1012 2279 aaronmk
    out_pkey = pkey(db, out_table, recover=True)
1013 2285 aaronmk
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
1014 2142 aaronmk
1015 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
1016 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
1017
1018 2201 aaronmk
    pkeys_table_exists_ref = [False]
1019 2420 aaronmk
    def insert_into_pkeys(joins, cols):
1020
        query, params = mk_select(db, joins, cols, order_by=None, start=0)
1021 2201 aaronmk
        if pkeys_table_exists_ref[0]:
1022 2489 aaronmk
            insert_select(db, into, pkeys_names, query, params)
1023 2201 aaronmk
        else:
1024 2489 aaronmk
            run_query_into(db, query, params, into=into)
1025 2201 aaronmk
            pkeys_table_exists_ref[0] = True
1026
1027 2429 aaronmk
    limit_ref = [None]
1028 2380 aaronmk
    conds = set()
1029 2233 aaronmk
    distinct_on = []
1030 2325 aaronmk
    def mk_main_select(joins, cols):
1031 2429 aaronmk
        return mk_select(db, joins, cols, conds, distinct_on,
1032
            limit=limit_ref[0], start=0)
1033 2132 aaronmk
1034 2519 aaronmk
    exc_strs = set()
1035 2309 aaronmk
    def log_exc(e):
1036 2519 aaronmk
        e_str = exc.str_(e, first_line_only=True)
1037
        log_debug('Caught exception: '+e_str)
1038
        assert e_str not in exc_strs # avoid infinite loops
1039
        exc_strs.add(e_str)
1040 2451 aaronmk
    def remove_all_rows():
1041 2450 aaronmk
        log_debug('Returning NULL for all rows')
1042 2429 aaronmk
        limit_ref[0] = 0 # just create an empty pkeys table
1043 2409 aaronmk
    def ignore(in_col, value):
1044 2545 aaronmk
        in_col_str = strings.as_tt(repr(in_col))
1045 2544 aaronmk
        db.log_debug('Adding index on '+in_col_str+' to enable fast filtering',
1046
            level=2.5)
1047 2537 aaronmk
        add_index(db, in_col)
1048 2545 aaronmk
        log_debug('Ignoring rows with '+in_col_str+' = '
1049
            +strings.as_tt(repr(value)))
1050 2403 aaronmk
    def remove_rows(in_col, value):
1051 2409 aaronmk
        ignore(in_col, value)
1052 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
1053
        assert cond not in conds # avoid infinite loops
1054 2380 aaronmk
        conds.add(cond)
1055 2403 aaronmk
    def invalid2null(in_col, value):
1056 2409 aaronmk
        ignore(in_col, value)
1057 2403 aaronmk
        update(db, in_table, [(in_col, None)],
1058
            sql_gen.ColValueCond(in_col, value))
1059 2245 aaronmk
1060 2589 aaronmk
    def insert_pkeys_table(which):
1061
        return sql_gen.Table(sql_gen.add_suffix(in_table.name,
1062
            '_insert_'+which+'_pkeys'))
1063
    insert_out_pkeys = insert_pkeys_table('out')
1064
    insert_in_pkeys = insert_pkeys_table('in')
1065
1066 2206 aaronmk
    # Do inserts and selects
1067 2642 aaronmk
    join_cols = sql_gen.ColDict(db, out_table)
1068 2206 aaronmk
    while True:
1069 2521 aaronmk
        if limit_ref[0] == 0: # special case
1070
            log_debug('Creating an empty pkeys table')
1071
            cur = run_query_into(db, *mk_select(db, out_table, [out_pkey],
1072
                limit=limit_ref[0]), into=insert_out_pkeys)
1073
            break # don't do main case
1074
1075 2303 aaronmk
        has_joins = join_cols != {}
1076
1077 2305 aaronmk
        # Prepare to insert new rows
1078 2325 aaronmk
        insert_joins = input_joins[:] # don't modify original!
1079 2403 aaronmk
        insert_args = dict(recover=True, cacheable=False)
1080 2303 aaronmk
        if has_joins:
1081 2317 aaronmk
            distinct_on = [v.to_Col() for v in join_cols.values()]
1082 2325 aaronmk
            insert_joins.append(sql_gen.Join(out_table, join_cols,
1083
                sql_gen.filter_out))
1084
        else:
1085 2404 aaronmk
            insert_args.update(dict(returning=out_pkey, into=insert_out_pkeys))
1086 2520 aaronmk
        main_select = mk_main_select(insert_joins, mapping.values())[0]
1087 2303 aaronmk
1088 2486 aaronmk
        log_debug('Trying to insert new rows')
1089 2206 aaronmk
        try:
1090 2518 aaronmk
            cur = insert_select(db, out_table, mapping.keys(), main_select,
1091
                **insert_args)
1092 2357 aaronmk
            break # insert successful
1093 2206 aaronmk
        except DuplicateKeyException, e:
1094 2309 aaronmk
            log_exc(e)
1095
1096 2258 aaronmk
            old_join_cols = join_cols.copy()
1097 2565 aaronmk
            join_cols.update(util.dict_subset_right_join(mapping, e.cols))
1098 2486 aaronmk
            log_debug('Ignoring existing rows, comparing on these columns:\n'
1099 2505 aaronmk
                +strings.as_inline_table(join_cols, ustr=col_ustr))
1100 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
1101 2230 aaronmk
        except NullValueException, e:
1102 2309 aaronmk
            log_exc(e)
1103
1104 2230 aaronmk
            out_col, = e.cols
1105
            try: in_col = mapping[out_col]
1106 2356 aaronmk
            except KeyError:
1107 2486 aaronmk
                log_debug('Missing mapping for NOT NULL column '+out_col)
1108 2451 aaronmk
                remove_all_rows()
1109 2403 aaronmk
            else: remove_rows(in_col, None)
1110 2542 aaronmk
        except FunctionValueException, e:
1111
            log_exc(e)
1112
1113
            func_name = e.name
1114
            value = e.value
1115
            for out_col, in_col in mapping.iteritems():
1116 2562 aaronmk
                invalid2null(sql_gen.unwrap_func_call(in_col, func_name), value)
1117 2525 aaronmk
        except MissingCastException, e:
1118
            log_exc(e)
1119
1120
            out_col = e.col
1121 2534 aaronmk
            mapping[out_col] = sql_gen.wrap_in_func(e.type, mapping[out_col])
1122 2429 aaronmk
        except DatabaseErrors, e:
1123
            log_exc(e)
1124
1125 2531 aaronmk
            msg = 'No handler for exception: '+exc.str_(e)
1126 2451 aaronmk
            warnings.warn(DbWarning(msg))
1127
            log_debug(msg)
1128
            remove_all_rows()
1129 2358 aaronmk
        # after exception handled, rerun loop with additional constraints
1130 2132 aaronmk
1131 2357 aaronmk
    if row_ct_ref != None and cur.rowcount >= 0:
1132
        row_ct_ref[0] += cur.rowcount
1133
1134
    if has_joins:
1135
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
1136 2486 aaronmk
        log_debug('Getting output table pkeys of existing/inserted rows')
1137 2420 aaronmk
        insert_into_pkeys(select_joins, pkeys_cols)
1138 2357 aaronmk
    else:
1139 2404 aaronmk
        add_row_num(db, insert_out_pkeys) # for joining with input pkeys
1140 2357 aaronmk
1141 2486 aaronmk
        log_debug('Getting input table pkeys of inserted rows')
1142 2357 aaronmk
        run_query_into(db, *mk_main_select(input_joins, [in_pkey]),
1143 2404 aaronmk
            into=insert_in_pkeys)
1144
        add_row_num(db, insert_in_pkeys) # for joining with output pkeys
1145 2357 aaronmk
1146 2428 aaronmk
        assert table_row_count(db, insert_out_pkeys) == table_row_count(db,
1147
            insert_in_pkeys)
1148
1149 2486 aaronmk
        log_debug('Combining output and input pkeys in inserted order')
1150 2404 aaronmk
        pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
1151 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
1152 2420 aaronmk
        insert_into_pkeys(pkey_joins, pkeys_names)
1153 2357 aaronmk
1154 2486 aaronmk
    db.log_debug('Adding pkey on pkeys table to enable fast joins', level=2.5)
1155 2594 aaronmk
    add_pkey(db, into)
1156 2407 aaronmk
1157 2508 aaronmk
    log_debug('Setting pkeys of missing rows to '+strings.as_tt(repr(default)))
1158 2489 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(into,
1159 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
1160
        # must use join_same_not_null or query will take forever
1161 2420 aaronmk
    insert_into_pkeys(missing_rows_joins,
1162 2508 aaronmk
        [in_pkey_col, sql_gen.NamedCol(out_pkey, default)])
1163 2357 aaronmk
1164 2489 aaronmk
    assert table_row_count(db, into) == table_row_count(db, in_table)
1165 2428 aaronmk
1166 2489 aaronmk
    return sql_gen.Col(out_pkey, into)
1167 2115 aaronmk
1168
##### Data cleanup
1169
1170 2290 aaronmk
def cleanup_table(db, table, cols):
1171 2115 aaronmk
    def esc_name_(name): return esc_name(db, name)
1172
1173 2290 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
1174 2115 aaronmk
    cols = map(esc_name_, cols)
1175
1176
    run_query(db, 'UPDATE '+table+' SET\n'+(',\n'.join(('\n'+col
1177
        +' = nullif(nullif(trim(both from '+col+"), %(null0)s), %(null1)s)"
1178
            for col in cols))),
1179
        dict(null0='', null1=r'\N'))