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 2676 aaronmk
            if self.serializable and not self.autocommit: self.run_query(
197 2678 aaronmk
                'SET TRANSACTION ISOLATION LEVEL SERIALIZABLE', log_level=4)
198 2101 aaronmk
            if schemas != None:
199
                schemas_ = ''.join((esc_name(self, s)+', '
200
                    for s in schemas.split(',')))
201 2676 aaronmk
                self.run_query("SELECT set_config('search_path', \
202
%s || current_setting('search_path'), false)", [schemas_], log_level=3)
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 2683 aaronmk
    def with_autocommit(self, func, autocommit=True):
335
        prev_autocommit = self.db.autocommit
336
        self.db.autocommit = autocommit
337
        try: return func()
338
        finally: self.db.autocommit = prev_autocommit
339
340 2139 aaronmk
    def with_savepoint(self, func):
341 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
342 2443 aaronmk
        self.run_query('SAVEPOINT '+savepoint, log_level=4)
343 2139 aaronmk
        self._savepoint += 1
344 2683 aaronmk
        try:
345 2139 aaronmk
            try: return_val = func()
346
            finally:
347
                self._savepoint -= 1
348
                assert self._savepoint >= 0
349
        except:
350 2443 aaronmk
            self.run_query('ROLLBACK TO SAVEPOINT '+savepoint, log_level=4)
351 2139 aaronmk
            raise
352
        else:
353 2443 aaronmk
            self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
354 2191 aaronmk
            self.do_autocommit()
355 2139 aaronmk
            return return_val
356 2191 aaronmk
357
    def do_autocommit(self):
358
        '''Autocommits if outside savepoint'''
359
        assert self._savepoint >= 0
360
        if self.autocommit and self._savepoint == 0:
361 2668 aaronmk
            self.log_debug('Autocommitting')
362 2191 aaronmk
            self.db.commit()
363 2643 aaronmk
364
    def col_default(self, col):
365
        table = sql_gen.Table('columns', 'information_schema')
366
367
        conds = [('table_name', col.table.name), ('column_name', col.name)]
368
        schema = col.table.schema
369
        if schema != None: conds.append(('table_schema', schema))
370
371 2660 aaronmk
        return sql_gen.as_Code(value(select(self, table, ['column_default'],
372
            conds, order_by='table_schema', limit=1, log_level=3)))
373 2643 aaronmk
            # TODO: order_by search_path schema order
374 1849 aaronmk
375 1869 aaronmk
connect = DbConn
376
377 832 aaronmk
##### Querying
378
379 1894 aaronmk
def run_raw_query(db, *args, **kw_args):
380 2085 aaronmk
    '''For params, see DbConn.run_query()'''
381 1894 aaronmk
    return db.run_query(*args, **kw_args)
382 11 aaronmk
383 2068 aaronmk
def mogrify(db, query, params):
384
    module = util.root_module(db.db)
385
    if module == 'psycopg2': return db.db.cursor().mogrify(query, params)
386
    else: raise NotImplementedError("Can't mogrify query for "+module+
387
        ' database')
388
389 832 aaronmk
##### Recoverable querying
390 15 aaronmk
391 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
392 11 aaronmk
393 2464 aaronmk
def run_query(db, query, params=None, recover=None, cacheable=False,
394
    log_level=2, log_ignore_excs=None, **kw_args):
395 2441 aaronmk
    '''For params, see run_raw_query()'''
396 830 aaronmk
    if recover == None: recover = False
397 2464 aaronmk
    if log_ignore_excs == None: log_ignore_excs = ()
398
    log_ignore_excs = tuple(log_ignore_excs)
399 830 aaronmk
400 2666 aaronmk
    debug_msg_ref = None # usually, db.run_query() logs query before running it
401
    # But if filtering with log_ignore_excs, wait until after exception parsing
402
    if log_ignore_excs != () or not db.can_mogrify(): debug_msg_ref = [None]
403
404 2148 aaronmk
    try:
405 2464 aaronmk
        try:
406
            def run(): return run_raw_query(db, query, params, cacheable,
407
                log_level, debug_msg_ref, **kw_args)
408
            if recover and not db.is_cached(query, params):
409
                return with_savepoint(db, run)
410
            else: return run() # don't need savepoint if cached
411
        except Exception, e:
412
            if not recover: raise # need savepoint to run index_cols()
413
            msg = exc.str_(e)
414
415
            match = re.search(r'duplicate key value violates unique constraint '
416 2493 aaronmk
                r'"((_?[^\W_]+)_.+?)"', msg)
417 2464 aaronmk
            if match:
418
                constraint, table = match.groups()
419
                try: cols = index_cols(db, table, constraint)
420
                except NotImplementedError: raise e
421
                else: raise DuplicateKeyException(constraint, cols, e)
422
423 2493 aaronmk
            match = re.search(r'null value in column "(.+?)" violates not-null'
424 2464 aaronmk
                r' constraint', msg)
425
            if match: raise NullValueException('NOT NULL', [match.group(1)], e)
426
427
            match = re.search(r'\b(?:invalid input (?:syntax|value)\b.*?'
428
                r'|date/time field value out of range): "(.+?)"\n'
429 2535 aaronmk
                r'(?:(?s).*?)\bfunction "(.+?)"', msg)
430 2464 aaronmk
            if match:
431
                value, name = match.groups()
432
                raise FunctionValueException(name, strings.to_unicode(value), e)
433
434 2526 aaronmk
            match = re.search(r'column "(.+?)" is of type (.+?) but expression '
435 2523 aaronmk
                r'is of type', msg)
436
            if match:
437
                col, type_ = match.groups()
438
                raise MissingCastException(type_, col, e)
439
440 2493 aaronmk
            match = re.search(r'relation "(.+?)" already exists', msg)
441 2464 aaronmk
            if match: raise DuplicateTableException(match.group(1), e)
442
443 2493 aaronmk
            match = re.search(r'function "(.+?)" already exists', msg)
444 2464 aaronmk
            if match: raise DuplicateFunctionException(match.group(1), e)
445
446
            raise # no specific exception raised
447
    except log_ignore_excs:
448
        log_level += 2
449
        raise
450
    finally:
451 2666 aaronmk
        if debug_msg_ref != None and debug_msg_ref[0] != None:
452
            db.log_debug(debug_msg_ref[0], log_level)
453 830 aaronmk
454 832 aaronmk
##### Basic queries
455
456 2153 aaronmk
def next_version(name):
457 2163 aaronmk
    version = 1 # first existing name was version 0
458 2586 aaronmk
    match = re.match(r'^(.*)#(\d+)$', name)
459 2153 aaronmk
    if match:
460 2586 aaronmk
        name, version = match.groups()
461
        version = int(version)+1
462 2588 aaronmk
    return sql_gen.add_suffix(name, '#'+str(version))
463 2153 aaronmk
464 2386 aaronmk
def run_query_into(db, query, params, into=None, *args, **kw_args):
465 2085 aaronmk
    '''Outputs a query to a temp table.
466
    For params, see run_query().
467
    '''
468 2386 aaronmk
    if into == None: return run_query(db, query, params, *args, **kw_args)
469 2085 aaronmk
    else: # place rows in temp table
470 2386 aaronmk
        assert isinstance(into, sql_gen.Table)
471 2385 aaronmk
472 2153 aaronmk
        kw_args['recover'] = True
473 2464 aaronmk
        kw_args.setdefault('log_ignore_excs', (DuplicateTableException,))
474 2440 aaronmk
475 2468 aaronmk
        temp = not db.autocommit # tables are permanent in autocommit mode
476 2440 aaronmk
        # "temporary tables cannot specify a schema name", so remove schema
477
        if temp: into.schema = None
478
479 2153 aaronmk
        while True:
480
            try:
481 2194 aaronmk
                create_query = 'CREATE'
482 2440 aaronmk
                if temp: create_query += ' TEMP'
483 2467 aaronmk
                create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
484 2194 aaronmk
485
                return run_query(db, create_query, params, *args, **kw_args)
486 2153 aaronmk
                    # CREATE TABLE AS sets rowcount to # rows in query
487
            except DuplicateTableException, e:
488 2386 aaronmk
                into.name = next_version(into.name)
489 2153 aaronmk
                # try again with next version of name
490 2085 aaronmk
491 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
492
493 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
494
495 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
496 2293 aaronmk
    start=None, order_by=order_by_pkey, default_table=None):
497 1981 aaronmk
    '''
498 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
499 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
500 1981 aaronmk
    @param fields Use None to select all fields in the table
501 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
502 2379 aaronmk
        * container can be any iterable type
503 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
504
        * compare_right_side: sql_gen.ValueCond|literal value
505 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
506
        use all columns
507 2054 aaronmk
    @return tuple(query, params)
508 1981 aaronmk
    '''
509 2315 aaronmk
    # Parse tables param
510 2121 aaronmk
    if not lists.is_seq(tables): tables = [tables]
511 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
512 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
513 2121 aaronmk
514 2315 aaronmk
    # Parse other params
515 2376 aaronmk
    if conds == None: conds = []
516 2650 aaronmk
    elif dicts.is_dict(conds): conds = conds.items()
517 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
518 135 aaronmk
    assert limit == None or type(limit) == int
519 865 aaronmk
    assert start == None or type(start) == int
520 2315 aaronmk
    if order_by is order_by_pkey:
521
        if distinct_on != []: order_by = None
522
        else: order_by = pkey(db, table0, recover=True)
523 865 aaronmk
524 2315 aaronmk
    query = 'SELECT'
525 2056 aaronmk
526 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
527 2056 aaronmk
528 2200 aaronmk
    # DISTINCT ON columns
529 2233 aaronmk
    if distinct_on != []:
530 2467 aaronmk
        query += '\nDISTINCT'
531 2254 aaronmk
        if distinct_on is not distinct_on_all:
532 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
533
534
    # Columns
535 2467 aaronmk
    query += '\n'
536 1135 aaronmk
    if fields == None: query += '*'
537 2479 aaronmk
    else: query += '\n, '.join(map(parse_col, fields))
538 2200 aaronmk
539
    # Main table
540 2467 aaronmk
    query += '\nFROM '+table0.to_str(db)
541 865 aaronmk
542 2122 aaronmk
    # Add joins
543 2271 aaronmk
    left_table = table0
544 2263 aaronmk
    for join_ in tables:
545
        table = join_.table
546 2238 aaronmk
547 2343 aaronmk
        # Parse special values
548
        if join_.type_ is sql_gen.filter_out: # filter no match
549 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
550
                None))
551 2343 aaronmk
552 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
553 2122 aaronmk
554
        left_table = table
555
556 865 aaronmk
    missing = True
557 2376 aaronmk
    if conds != []:
558 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
559
        else: whitespace = '\n'
560 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
561
            .to_str(db) for l, r in conds], 'WHERE')
562 865 aaronmk
        missing = False
563 2227 aaronmk
    if order_by != None:
564 2467 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
565
    if limit != None: query += '\nLIMIT '+str(limit); missing = False
566 865 aaronmk
    if start != None:
567 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
568 865 aaronmk
        missing = False
569
    if missing: warnings.warn(DbWarning(
570
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
571
572 2315 aaronmk
    return (query, [])
573 11 aaronmk
574 2054 aaronmk
def select(db, *args, **kw_args):
575
    '''For params, see mk_select() and run_query()'''
576
    recover = kw_args.pop('recover', None)
577
    cacheable = kw_args.pop('cacheable', True)
578 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
579 2054 aaronmk
580
    query, params = mk_select(db, *args, **kw_args)
581 2442 aaronmk
    return run_query(db, query, params, recover, cacheable, log_level=log_level)
582 2054 aaronmk
583 2066 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, params=None,
584 2292 aaronmk
    returning=None, embeddable=False):
585 1960 aaronmk
    '''
586
    @param returning str|None An inserted column (such as pkey) to return
587 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
588 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
589
        query will be fully cached, not just if it raises an exception.
590 1960 aaronmk
    '''
591 2328 aaronmk
    table = sql_gen.as_Table(table)
592 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
593 2646 aaronmk
    if cols != None:
594
        cols = [sql_gen.to_name_only_col(v, table).to_str(db) for v in cols]
595 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
596 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
597 2063 aaronmk
598
    # Build query
599 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
600
    query = first_line
601 2467 aaronmk
    if cols != None: query += '\n('+', '.join(cols)+')'
602
    query += '\n'+select_query
603 2063 aaronmk
604
    if returning != None:
605 2327 aaronmk
        returning_name = copy.copy(returning)
606
        returning_name.table = None
607
        returning_name = returning_name.to_str(db)
608 2467 aaronmk
        query += '\nRETURNING '+returning_name
609 2063 aaronmk
610 2070 aaronmk
    if embeddable:
611 2327 aaronmk
        assert returning != None
612
613 2070 aaronmk
        # Create function
614 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
615 2327 aaronmk
        return_type = 'SETOF '+returning.to_str(db)+'%TYPE'
616 2189 aaronmk
        while True:
617
            try:
618 2327 aaronmk
                func_schema = None
619 2468 aaronmk
                if not db.autocommit: func_schema = 'pg_temp'
620 2327 aaronmk
                function = sql_gen.Table(function_name, func_schema).to_str(db)
621 2194 aaronmk
622 2189 aaronmk
                function_query = '''\
623 2467 aaronmk
CREATE FUNCTION '''+function+'''()
624
RETURNS '''+return_type+'''
625
LANGUAGE sql
626
AS $$
627
'''+mogrify(db, query, params)+''';
628
$$;
629 2070 aaronmk
'''
630 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
631 2464 aaronmk
                    log_ignore_excs=(DuplicateFunctionException,))
632 2189 aaronmk
                break # this version was successful
633
            except DuplicateFunctionException, e:
634
                function_name = next_version(function_name)
635
                # try again with next version of name
636 2070 aaronmk
637 2337 aaronmk
        # Return query that uses function
638
        func_table = sql_gen.NamedTable('f', sql_gen.CustomCode(function+'()'),
639
            [returning_name]) # AS clause requires function alias
640
        return mk_select(db, func_table, start=0, order_by=None)
641 2070 aaronmk
642 2066 aaronmk
    return (query, params)
643
644
def insert_select(db, *args, **kw_args):
645 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
646 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
647
        values in
648 2072 aaronmk
    '''
649 2386 aaronmk
    into = kw_args.pop('into', None)
650
    if into != None: kw_args['embeddable'] = True
651 2066 aaronmk
    recover = kw_args.pop('recover', None)
652
    cacheable = kw_args.pop('cacheable', True)
653 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
654 2066 aaronmk
655
    query, params = mk_insert_select(db, *args, **kw_args)
656 2386 aaronmk
    return run_query_into(db, query, params, into, recover=recover,
657 2673 aaronmk
        cacheable=cacheable, log_level=log_level)
658 2063 aaronmk
659 2066 aaronmk
default = object() # tells insert() to use the default value for a column
660
661 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
662 2085 aaronmk
    '''For params, see insert_select()'''
663 1960 aaronmk
    if lists.is_seq(row): cols = None
664
    else:
665
        cols = row.keys()
666
        row = row.values()
667
    row = list(row) # ensure that "!= []" works
668
669 1961 aaronmk
    # Check for special values
670
    labels = []
671
    values = []
672
    for value in row:
673 2647 aaronmk
        value = sql_gen.remove_col_rename(sql_gen.as_Value(value)).value
674 2254 aaronmk
        if value is default: labels.append('DEFAULT')
675 1961 aaronmk
        else:
676
            labels.append('%s')
677
            values.append(value)
678
679
    # Build query
680 2467 aaronmk
    if values != []: query = 'VALUES ('+(', '.join(labels))+')'
681 2063 aaronmk
    else: query = None
682 1554 aaronmk
683 2064 aaronmk
    return insert_select(db, table, cols, query, values, *args, **kw_args)
684 11 aaronmk
685 2402 aaronmk
def mk_update(db, table, changes=None, cond=None):
686
    '''
687
    @param changes [(col, new_value),...]
688
        * container can be any iterable type
689
        * col: sql_gen.Code|str (for col name)
690
        * new_value: sql_gen.Code|literal value
691
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
692
    @return str query
693
    '''
694
    query = 'UPDATE '+sql_gen.as_Table(table).to_str(db)+'\nSET\n'
695 2405 aaronmk
    query += ',\n'.join((sql_gen.to_name_only_col(col, table).to_str(db)+' = '
696 2402 aaronmk
        +sql_gen.as_Value(new_value).to_str(db) for col, new_value in changes))
697 2467 aaronmk
    if cond != None: query += '\nWHERE\n'+cond.to_str(db)
698 2402 aaronmk
699
    return query
700
701
def update(db, *args, **kw_args):
702
    '''For params, see mk_update() and run_query()'''
703
    recover = kw_args.pop('recover', None)
704
705
    return run_query(db, mk_update(db, *args, **kw_args), [], recover)
706
707 135 aaronmk
def last_insert_id(db):
708 1849 aaronmk
    module = util.root_module(db.db)
709 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
710
    elif module == 'MySQLdb': return db.insert_id()
711
    else: return None
712 13 aaronmk
713 1968 aaronmk
def truncate(db, table, schema='public'):
714
    return run_query(db, 'TRUNCATE '+qual_name(db, schema, table)+' CASCADE')
715 832 aaronmk
716 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
717 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
718 2415 aaronmk
    to names that will be distinct among the columns' tables.
719 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
720 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
721
    @param into The table for the new columns.
722 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
723
        columns will be included in the mapping even if they are not in cols.
724
        The tables of the provided Col objects will be changed to into, so make
725
        copies of them if you want to keep the original tables.
726
    @param as_items Whether to return a list of dict items instead of a dict
727 2383 aaronmk
    @return dict(orig_col=new_col, ...)
728
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
729 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
730
        * All mappings use the into table so its name can easily be
731 2383 aaronmk
          changed for all columns at once
732
    '''
733 2415 aaronmk
    cols = lists.uniqify(cols)
734
735 2394 aaronmk
    items = []
736 2389 aaronmk
    for col in preserve:
737 2390 aaronmk
        orig_col = copy.copy(col)
738 2392 aaronmk
        col.table = into
739 2394 aaronmk
        items.append((orig_col, col))
740
    preserve = set(preserve)
741
    for col in cols:
742 2515 aaronmk
        if col not in preserve: items.append((col, sql_gen.Col(str(col), into)))
743 2394 aaronmk
744
    if not as_items: items = dict(items)
745
    return items
746 2383 aaronmk
747 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
748 2391 aaronmk
    '''For params, see mk_flatten_mapping()
749
    @return See return value of mk_flatten_mapping()
750
    '''
751 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
752
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
753 2391 aaronmk
    run_query_into(db, *mk_select(db, joins, cols, limit=limit, start=start),
754 2392 aaronmk
        into=into)
755 2394 aaronmk
    return dict(items)
756 2391 aaronmk
757 2414 aaronmk
##### Database structure queries
758
759 2426 aaronmk
def table_row_count(db, table, recover=None):
760
    return value(run_query(db, *mk_select(db, table, [sql_gen.row_count],
761 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
762 2426 aaronmk
763 2414 aaronmk
def table_cols(db, table, recover=None):
764
    return list(col_names(select(db, table, limit=0, order_by=None,
765 2443 aaronmk
        recover=recover, log_level=4)))
766 2414 aaronmk
767 2291 aaronmk
def pkey(db, table, recover=None):
768 832 aaronmk
    '''Assumed to be first column in table'''
769 2339 aaronmk
    return table_cols(db, table, recover)[0]
770 832 aaronmk
771 2559 aaronmk
not_null_col = 'not_null_col'
772 2340 aaronmk
773
def table_not_null_col(db, table, recover=None):
774
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
775
    if not_null_col in table_cols(db, table, recover): return not_null_col
776
    else: return pkey(db, table, recover)
777
778 853 aaronmk
def index_cols(db, table, index):
779
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
780
    automatically created. When you don't know whether something is a UNIQUE
781
    constraint or a UNIQUE index, use this function.'''
782 1909 aaronmk
    module = util.root_module(db.db)
783
    if module == 'psycopg2':
784
        return list(values(run_query(db, '''\
785 853 aaronmk
SELECT attname
786 866 aaronmk
FROM
787
(
788
        SELECT attnum, attname
789
        FROM pg_index
790
        JOIN pg_class index ON index.oid = indexrelid
791
        JOIN pg_class table_ ON table_.oid = indrelid
792
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
793
        WHERE
794
            table_.relname = %(table)s
795
            AND index.relname = %(index)s
796
    UNION
797
        SELECT attnum, attname
798
        FROM
799
        (
800
            SELECT
801
                indrelid
802
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
803
                    AS indkey
804
            FROM pg_index
805
            JOIN pg_class index ON index.oid = indexrelid
806
            JOIN pg_class table_ ON table_.oid = indrelid
807
            WHERE
808
                table_.relname = %(table)s
809
                AND index.relname = %(index)s
810
        ) s
811
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
812
) s
813 853 aaronmk
ORDER BY attnum
814
''',
815 2443 aaronmk
            {'table': table, 'index': index}, cacheable=True, log_level=4)))
816 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
817
        ' database')
818 853 aaronmk
819 464 aaronmk
def constraint_cols(db, table, constraint):
820 1849 aaronmk
    module = util.root_module(db.db)
821 464 aaronmk
    if module == 'psycopg2':
822
        return list(values(run_query(db, '''\
823
SELECT attname
824
FROM pg_constraint
825
JOIN pg_class ON pg_class.oid = conrelid
826
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
827
WHERE
828
    relname = %(table)s
829
    AND conname = %(constraint)s
830
ORDER BY attnum
831
''',
832
            {'table': table, 'constraint': constraint})))
833
    else: raise NotImplementedError("Can't list constraint columns for "+module+
834
        ' database')
835
836 2096 aaronmk
row_num_col = '_row_num'
837
838 2688 aaronmk
def add_index(db, exprs, table=None, unique=False):
839
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
840 2538 aaronmk
    Currently, only function calls are supported as expressions.
841
    '''
842 2688 aaronmk
    if not lists.is_seq(exprs): exprs = [exprs]
843 2538 aaronmk
844 2688 aaronmk
    # Parse exprs
845
    old_exprs = exprs[:]
846
    exprs = []
847
    cols = []
848
    for i, expr in enumerate(old_exprs):
849
        expr = copy.deepcopy(expr) # don't modify input!
850
        expr = sql_gen.as_Col(expr)
851
852
        # Extract col
853
        if isinstance(expr, sql_gen.FunctionCall):
854
            col = expr.args[0]
855
            expr = sql_gen.Expr(expr)
856
        else: col = expr
857
858
        # Extract table
859
        if table == None:
860
            assert sql_gen.is_table_col(col)
861
            table = col.table
862
863
        col.table = None
864
865
        exprs.append(expr)
866
        cols.append(col)
867 2408 aaronmk
868 2688 aaronmk
    table = sql_gen.as_Table(table)
869
    index = sql_gen.Table(str(sql_gen.Col(','.join(map(str, cols)), table)))
870
871
    str_ = 'CREATE'
872
    if unique: str_ += ' UNIQUE'
873
    str_ += ' INDEX '+index.to_str(db)+' ON '+table.to_str(db)+' ('+(
874
        ', '.join((v.to_str(db) for v in exprs)))+')'
875
876
    try: run_query(db, str_, recover=True, cacheable=True, log_level=3)
877 2408 aaronmk
    except DuplicateTableException: pass # index already existed
878
879 2686 aaronmk
def add_pkey(db, table, cols=None, recover=None):
880
    '''Adds a primary key.
881
    @param cols [sql_gen.Col,...] The columns in the primary key.
882
        Defaults to the first column in the table.
883 2406 aaronmk
    @pre The table must not already have a primary key.
884
    '''
885
    table = sql_gen.as_Table(table)
886 2686 aaronmk
    if cols == None: cols = [pkey(db, table, recover)]
887
    col_strs = [sql_gen.to_name_only_col(v).to_str(db) for v in cols]
888 2406 aaronmk
889 2590 aaronmk
    index = sql_gen.as_Table(sql_gen.add_suffix(table.name, '_pkey'))
890 2614 aaronmk
    try:
891
        run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD CONSTRAINT '
892 2686 aaronmk
            +index.to_str(db)+' PRIMARY KEY ('+(', '.join(col_strs))+')',
893 2614 aaronmk
            recover=True, cacheable=True, log_level=3,
894
            log_ignore_excs=(DuplicateTableException,))
895
    except DuplicateTableException, e:
896
        index.name = next_version(index.name)
897
        # try again with next version of name
898 2406 aaronmk
899 2086 aaronmk
def add_row_num(db, table):
900 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
901
    be the primary key.'''
902 2320 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
903 2096 aaronmk
    run_query(db, 'ALTER TABLE '+table+' ADD COLUMN '+row_num_col
904 2443 aaronmk
        +' serial NOT NULL PRIMARY KEY', log_level=3)
905 2086 aaronmk
906 2681 aaronmk
def create_table(db, table, cols, has_pkey=True):
907 2675 aaronmk
    '''Creates a table.
908 2681 aaronmk
    @param cols [sql_gen.TypedCol,...] The column names and types
909
    @param has_pkey If set, the first column becomes the primary key.
910 2675 aaronmk
    '''
911
    table = sql_gen.as_Table(table)
912
913 2681 aaronmk
    if has_pkey:
914
        cols[0] = pkey = copy.copy(cols[0]) # don't modify input!
915
        pkey.type += ' NOT NULL PRIMARY KEY'
916
917 2675 aaronmk
    str_ = 'CREATE TABLE '+table.to_str(db)+' (\n'
918 2681 aaronmk
    str_ += '\n, '.join(v.to_str(db) for v in cols)
919
    str_ += '\n);\n'
920 2675 aaronmk
    run_query(db, str_, cacheable=True, log_level=2)
921
922 2684 aaronmk
def vacuum(db, table):
923
    table = sql_gen.as_Table(table)
924
    db.with_autocommit(lambda: run_query(db, 'VACUUM ANALYZE '+table.to_str(db),
925
        log_level=3))
926
927 2548 aaronmk
def tables(db, schema_like='public', table_like='%'):
928 1849 aaronmk
    module = util.root_module(db.db)
929 2548 aaronmk
    params = {'schema_like': schema_like, 'table_like': table_like}
930 832 aaronmk
    if module == 'psycopg2':
931 1968 aaronmk
        return values(run_query(db, '''\
932
SELECT tablename
933
FROM pg_tables
934
WHERE
935 2548 aaronmk
    schemaname LIKE %(schema_like)s
936 1968 aaronmk
    AND tablename LIKE %(table_like)s
937
ORDER BY tablename
938
''',
939
            params, cacheable=True))
940
    elif module == 'MySQLdb':
941
        return values(run_query(db, 'SHOW TABLES LIKE %(table_like)s', params,
942
            cacheable=True))
943 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
944 830 aaronmk
945 833 aaronmk
##### Database management
946
947 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
948
    '''For kw_args, see tables()'''
949
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
950 833 aaronmk
951 832 aaronmk
##### Heuristic queries
952
953 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
954 1554 aaronmk
    '''Recovers from errors.
955 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
956
    '''
957 2642 aaronmk
    row = sql_gen.ColDict(db, table, row)
958 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
959
960 471 aaronmk
    try:
961 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
962 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
963
            row_ct_ref[0] += cur.rowcount
964
        return value(cur)
965 471 aaronmk
    except DuplicateKeyException, e:
966 2653 aaronmk
        row = sql_gen.ColDict(db, table,
967
            util.dict_subset_right_join(row, e.cols))
968
        return value(select(db, table, [pkey_], row, recover=True))
969 471 aaronmk
970 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
971 830 aaronmk
    '''Recovers from errors'''
972 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
973 14 aaronmk
    except StopIteration:
974 40 aaronmk
        if not create: raise
975 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
976 2078 aaronmk
977 2593 aaronmk
def is_func_result(col):
978
    return col.table.name.find('(') >= 0 and col.name == 'result'
979
980 2592 aaronmk
def into_table_name(out_table, in_tables0, mapping, is_func):
981 2615 aaronmk
    def in_col_str(in_col):
982
        in_col = sql_gen.remove_col_rename(in_col)
983
        if isinstance(in_col, sql_gen.Col):
984
            table = in_col.table
985
            if table == in_tables0:
986
                in_col = sql_gen.to_name_only_col(in_col)
987
            elif is_func_result(in_col): in_col = table # omit col name
988
        return str(in_col)
989
990 2580 aaronmk
    str_ = str(out_table)
991
    if is_func:
992 2615 aaronmk
        str_ += '('
993 2580 aaronmk
994 2615 aaronmk
        try: value_in_col = mapping['value']
995
        except KeyError:
996 2654 aaronmk
            str_ += ', '.join((str(k)+'='+in_col_str(v)
997 2615 aaronmk
                for k, v in mapping.iteritems()))
998
        else: str_ += in_col_str(value_in_col)
999
1000
        str_ += ')'
1001 2580 aaronmk
    else: str_ += '_pkeys'
1002
    return str_
1003
1004 2508 aaronmk
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None, into=None,
1005 2552 aaronmk
    default=None, is_func=False):
1006 2078 aaronmk
    '''Recovers from errors.
1007
    Only works under PostgreSQL (uses INSERT RETURNING).
1008 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
1009
        tables to join with it using the main input table's pkey
1010 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
1011 2612 aaronmk
        * out_table_col: str (*not* sql_gen.Col)
1012
        * in_table_col: sql_gen.Col|literal-value
1013 2489 aaronmk
    @param into The table to contain the output and input pkeys.
1014 2574 aaronmk
        Defaults to `out_table.name+'_pkeys'`.
1015 2509 aaronmk
    @param default The *output* column to use as the pkey for missing rows.
1016
        If this output column does not exist in the mapping, uses None.
1017 2552 aaronmk
    @param is_func Whether out_table is the name of a SQL function, not a table
1018 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
1019 2078 aaronmk
    '''
1020 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
1021 2312 aaronmk
1022 2450 aaronmk
    def log_debug(msg): db.log_debug(msg, level=1.5)
1023 2505 aaronmk
    def col_ustr(str_):
1024 2567 aaronmk
        return strings.repr_no_u(sql_gen.remove_col_rename(str_))
1025 2450 aaronmk
1026 2486 aaronmk
    log_debug('********** New iteration **********')
1027 2505 aaronmk
    log_debug('Inserting these input columns into '+strings.as_tt(
1028
        out_table.to_str(db))+':\n'+strings.as_table(mapping, ustr=col_ustr))
1029 2463 aaronmk
1030 2382 aaronmk
    # Create input joins from list of input tables
1031
    in_tables_ = in_tables[:] # don't modify input!
1032
    in_tables0 = in_tables_.pop(0) # first table is separate
1033 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
1034 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
1035 2460 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v,
1036
        {in_pkey: sql_gen.join_same_not_null}) for v in in_tables_]
1037 2131 aaronmk
1038 2592 aaronmk
    if into == None:
1039
        into = into_table_name(out_table, in_tables0, mapping, is_func)
1040
    into = sql_gen.as_Table(into)
1041
1042 2486 aaronmk
    log_debug('Joining together input tables into temp table')
1043 2395 aaronmk
    # Place in new table for speed and so don't modify input if values edited
1044 2584 aaronmk
    in_table = sql_gen.Table('in')
1045 2395 aaronmk
    flatten_cols = filter(sql_gen.is_table_col, mapping.values())
1046
    mapping = dicts.join(mapping, flatten(db, in_table, input_joins,
1047
        flatten_cols, preserve=[in_pkey_col], start=0))
1048
    input_joins = [in_table]
1049 2486 aaronmk
    db.log_debug('Temp table: '+strings.as_tt(in_table.to_str(db)), level=2)
1050 2395 aaronmk
1051 2656 aaronmk
    mapping = sql_gen.ColDict(db, out_table, mapping)
1052
        # after applying dicts.join() because that returns a plain dict
1053
1054 2509 aaronmk
    # Resolve default value column
1055
    try: default = mapping[default]
1056
    except KeyError:
1057
        if default != None:
1058
            db.log_debug('Default value column '
1059
                +strings.as_tt(strings.repr_no_u(default))
1060 2511 aaronmk
                +' does not exist in mapping, falling back to None', level=2.1)
1061 2509 aaronmk
            default = None
1062
1063 2279 aaronmk
    out_pkey = pkey(db, out_table, recover=True)
1064 2285 aaronmk
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
1065 2142 aaronmk
1066 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
1067 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
1068
1069 2201 aaronmk
    pkeys_table_exists_ref = [False]
1070 2420 aaronmk
    def insert_into_pkeys(joins, cols):
1071
        query, params = mk_select(db, joins, cols, order_by=None, start=0)
1072 2201 aaronmk
        if pkeys_table_exists_ref[0]:
1073 2489 aaronmk
            insert_select(db, into, pkeys_names, query, params)
1074 2201 aaronmk
        else:
1075 2489 aaronmk
            run_query_into(db, query, params, into=into)
1076 2201 aaronmk
            pkeys_table_exists_ref[0] = True
1077
1078 2429 aaronmk
    limit_ref = [None]
1079 2380 aaronmk
    conds = set()
1080 2233 aaronmk
    distinct_on = []
1081 2325 aaronmk
    def mk_main_select(joins, cols):
1082 2429 aaronmk
        return mk_select(db, joins, cols, conds, distinct_on,
1083
            limit=limit_ref[0], start=0)
1084 2132 aaronmk
1085 2519 aaronmk
    exc_strs = set()
1086 2309 aaronmk
    def log_exc(e):
1087 2519 aaronmk
        e_str = exc.str_(e, first_line_only=True)
1088
        log_debug('Caught exception: '+e_str)
1089
        assert e_str not in exc_strs # avoid infinite loops
1090
        exc_strs.add(e_str)
1091 2451 aaronmk
    def remove_all_rows():
1092 2450 aaronmk
        log_debug('Returning NULL for all rows')
1093 2429 aaronmk
        limit_ref[0] = 0 # just create an empty pkeys table
1094 2409 aaronmk
    def ignore(in_col, value):
1095 2545 aaronmk
        in_col_str = strings.as_tt(repr(in_col))
1096 2544 aaronmk
        db.log_debug('Adding index on '+in_col_str+' to enable fast filtering',
1097
            level=2.5)
1098 2537 aaronmk
        add_index(db, in_col)
1099 2545 aaronmk
        log_debug('Ignoring rows with '+in_col_str+' = '
1100
            +strings.as_tt(repr(value)))
1101 2403 aaronmk
    def remove_rows(in_col, value):
1102 2409 aaronmk
        ignore(in_col, value)
1103 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
1104
        assert cond not in conds # avoid infinite loops
1105 2380 aaronmk
        conds.add(cond)
1106 2403 aaronmk
    def invalid2null(in_col, value):
1107 2409 aaronmk
        ignore(in_col, value)
1108 2403 aaronmk
        update(db, in_table, [(in_col, None)],
1109
            sql_gen.ColValueCond(in_col, value))
1110 2245 aaronmk
1111 2589 aaronmk
    def insert_pkeys_table(which):
1112
        return sql_gen.Table(sql_gen.add_suffix(in_table.name,
1113
            '_insert_'+which+'_pkeys'))
1114
    insert_out_pkeys = insert_pkeys_table('out')
1115
    insert_in_pkeys = insert_pkeys_table('in')
1116
1117 2206 aaronmk
    # Do inserts and selects
1118 2642 aaronmk
    join_cols = sql_gen.ColDict(db, out_table)
1119 2206 aaronmk
    while True:
1120 2521 aaronmk
        if limit_ref[0] == 0: # special case
1121
            log_debug('Creating an empty pkeys table')
1122
            cur = run_query_into(db, *mk_select(db, out_table, [out_pkey],
1123
                limit=limit_ref[0]), into=insert_out_pkeys)
1124
            break # don't do main case
1125
1126 2303 aaronmk
        has_joins = join_cols != {}
1127
1128 2305 aaronmk
        # Prepare to insert new rows
1129 2325 aaronmk
        insert_joins = input_joins[:] # don't modify original!
1130 2403 aaronmk
        insert_args = dict(recover=True, cacheable=False)
1131 2303 aaronmk
        if has_joins:
1132 2317 aaronmk
            distinct_on = [v.to_Col() for v in join_cols.values()]
1133 2325 aaronmk
            insert_joins.append(sql_gen.Join(out_table, join_cols,
1134
                sql_gen.filter_out))
1135
        else:
1136 2404 aaronmk
            insert_args.update(dict(returning=out_pkey, into=insert_out_pkeys))
1137 2520 aaronmk
        main_select = mk_main_select(insert_joins, mapping.values())[0]
1138 2303 aaronmk
1139 2486 aaronmk
        log_debug('Trying to insert new rows')
1140 2206 aaronmk
        try:
1141 2518 aaronmk
            cur = insert_select(db, out_table, mapping.keys(), main_select,
1142
                **insert_args)
1143 2357 aaronmk
            break # insert successful
1144 2206 aaronmk
        except DuplicateKeyException, e:
1145 2309 aaronmk
            log_exc(e)
1146
1147 2258 aaronmk
            old_join_cols = join_cols.copy()
1148 2565 aaronmk
            join_cols.update(util.dict_subset_right_join(mapping, e.cols))
1149 2486 aaronmk
            log_debug('Ignoring existing rows, comparing on these columns:\n'
1150 2505 aaronmk
                +strings.as_inline_table(join_cols, ustr=col_ustr))
1151 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
1152 2230 aaronmk
        except NullValueException, e:
1153 2309 aaronmk
            log_exc(e)
1154
1155 2230 aaronmk
            out_col, = e.cols
1156
            try: in_col = mapping[out_col]
1157 2356 aaronmk
            except KeyError:
1158 2486 aaronmk
                log_debug('Missing mapping for NOT NULL column '+out_col)
1159 2451 aaronmk
                remove_all_rows()
1160 2403 aaronmk
            else: remove_rows(in_col, None)
1161 2542 aaronmk
        except FunctionValueException, e:
1162
            log_exc(e)
1163
1164
            func_name = e.name
1165
            value = e.value
1166
            for out_col, in_col in mapping.iteritems():
1167 2562 aaronmk
                invalid2null(sql_gen.unwrap_func_call(in_col, func_name), value)
1168 2525 aaronmk
        except MissingCastException, e:
1169
            log_exc(e)
1170
1171
            out_col = e.col
1172 2534 aaronmk
            mapping[out_col] = sql_gen.wrap_in_func(e.type, mapping[out_col])
1173 2429 aaronmk
        except DatabaseErrors, e:
1174
            log_exc(e)
1175
1176 2531 aaronmk
            msg = 'No handler for exception: '+exc.str_(e)
1177 2451 aaronmk
            warnings.warn(DbWarning(msg))
1178
            log_debug(msg)
1179
            remove_all_rows()
1180 2358 aaronmk
        # after exception handled, rerun loop with additional constraints
1181 2132 aaronmk
1182 2357 aaronmk
    if row_ct_ref != None and cur.rowcount >= 0:
1183
        row_ct_ref[0] += cur.rowcount
1184
1185
    if has_joins:
1186
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
1187 2486 aaronmk
        log_debug('Getting output table pkeys of existing/inserted rows')
1188 2420 aaronmk
        insert_into_pkeys(select_joins, pkeys_cols)
1189 2357 aaronmk
    else:
1190 2404 aaronmk
        add_row_num(db, insert_out_pkeys) # for joining with input pkeys
1191 2357 aaronmk
1192 2486 aaronmk
        log_debug('Getting input table pkeys of inserted rows')
1193 2357 aaronmk
        run_query_into(db, *mk_main_select(input_joins, [in_pkey]),
1194 2404 aaronmk
            into=insert_in_pkeys)
1195
        add_row_num(db, insert_in_pkeys) # for joining with output pkeys
1196 2357 aaronmk
1197 2428 aaronmk
        assert table_row_count(db, insert_out_pkeys) == table_row_count(db,
1198
            insert_in_pkeys)
1199
1200 2486 aaronmk
        log_debug('Combining output and input pkeys in inserted order')
1201 2404 aaronmk
        pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
1202 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
1203 2420 aaronmk
        insert_into_pkeys(pkey_joins, pkeys_names)
1204 2357 aaronmk
1205 2486 aaronmk
    db.log_debug('Adding pkey on pkeys table to enable fast joins', level=2.5)
1206 2594 aaronmk
    add_pkey(db, into)
1207 2407 aaronmk
1208 2508 aaronmk
    log_debug('Setting pkeys of missing rows to '+strings.as_tt(repr(default)))
1209 2489 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(into,
1210 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
1211
        # must use join_same_not_null or query will take forever
1212 2420 aaronmk
    insert_into_pkeys(missing_rows_joins,
1213 2508 aaronmk
        [in_pkey_col, sql_gen.NamedCol(out_pkey, default)])
1214 2357 aaronmk
1215 2489 aaronmk
    assert table_row_count(db, into) == table_row_count(db, in_table)
1216 2428 aaronmk
1217 2489 aaronmk
    return sql_gen.Col(out_pkey, into)
1218 2115 aaronmk
1219
##### Data cleanup
1220
1221 2290 aaronmk
def cleanup_table(db, table, cols):
1222 2677 aaronmk
    table = sql_gen.as_Table(table)
1223
    cols = map(sql_gen.as_Col, cols)
1224 2115 aaronmk
1225 2677 aaronmk
    expr = ('nullif(nullif(trim(both from %s), '+db.esc_value('')+'), '
1226
        +db.esc_value(r'\N')+')')
1227
    changes = [(v, sql_gen.CustomCode(expr % v.to_str(db)))
1228
        for v in cols]
1229 2115 aaronmk
1230 2677 aaronmk
    update(db, table, changes)