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 2804 aaronmk
def get_cur_query(cur, input_query=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 2804 aaronmk
    else: return '[input] '+strings.ustr(input_query)
27 14 aaronmk
28 2170 aaronmk
def _add_cursor_info(e, *args, **kw_args):
29
    '''For params, see get_cur_query()'''
30 2771 aaronmk
    exc.add_msg(e, 'query: '+strings.ustr(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 2762 aaronmk
##### Escaping
109 2101 aaronmk
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 2448 aaronmk
log_debug_none = lambda msg, level=2: None
149 1901 aaronmk
150 1849 aaronmk
class DbConn:
151 2905 aaronmk
    def __init__(self, db_config, autocommit=False, caching=True,
152 2915 aaronmk
        log_debug=log_debug_none, debug_temp=False):
153
        '''
154
        @param debug_temp Whether temporary objects should instead be permanent.
155
            This assists in debugging the internal objects used by the program.
156
        '''
157 1869 aaronmk
        self.db_config = db_config
158 2190 aaronmk
        self.autocommit = autocommit
159
        self.caching = caching
160 1901 aaronmk
        self.log_debug = log_debug
161 2193 aaronmk
        self.debug = log_debug != log_debug_none
162 2915 aaronmk
        self.debug_temp = debug_temp
163 1869 aaronmk
164
        self.__db = None
165 1889 aaronmk
        self.query_results = {}
166 2139 aaronmk
        self._savepoint = 0
167 2671 aaronmk
        self._notices_seen = set()
168 1869 aaronmk
169
    def __getattr__(self, name):
170
        if name == '__dict__': raise Exception('getting __dict__')
171
        if name == 'db': return self._db()
172
        else: raise AttributeError()
173
174
    def __getstate__(self):
175
        state = copy.copy(self.__dict__) # shallow copy
176 1915 aaronmk
        state['log_debug'] = None # don't pickle the debug callback
177 1869 aaronmk
        state['_DbConn__db'] = None # don't pickle the connection
178
        return state
179
180 2165 aaronmk
    def connected(self): return self.__db != None
181
182 1869 aaronmk
    def _db(self):
183
        if self.__db == None:
184
            # Process db_config
185
            db_config = self.db_config.copy() # don't modify input!
186 2097 aaronmk
            schemas = db_config.pop('schemas', None)
187 1869 aaronmk
            module_name, mappings = db_engines[db_config.pop('engine')]
188
            module = __import__(module_name)
189
            _add_module(module)
190
            for orig, new in mappings.iteritems():
191
                try: util.rename_key(db_config, orig, new)
192
                except KeyError: pass
193
194
            # Connect
195
            self.__db = module.connect(**db_config)
196
197
            # Configure connection
198 2906 aaronmk
            if hasattr(self.db, 'set_isolation_level'):
199
                import psycopg2.extensions
200
                self.db.set_isolation_level(
201
                    psycopg2.extensions.ISOLATION_LEVEL_READ_COMMITTED)
202 2101 aaronmk
            if schemas != None:
203 2893 aaronmk
                search_path = [self.esc_name(s) for s in schemas.split(',')]
204
                search_path.append(value(run_query(self, 'SHOW search_path',
205
                    log_level=4)))
206
                run_query(self, 'SET search_path TO '+(','.join(search_path)),
207
                    log_level=3)
208 1869 aaronmk
209
        return self.__db
210 1889 aaronmk
211 1891 aaronmk
    class DbCursor(Proxy):
212 1927 aaronmk
        def __init__(self, outer):
213 1891 aaronmk
            Proxy.__init__(self, outer.db.cursor())
214 2191 aaronmk
            self.outer = outer
215 1927 aaronmk
            self.query_results = outer.query_results
216 1894 aaronmk
            self.query_lookup = None
217 1891 aaronmk
            self.result = []
218 1889 aaronmk
219 2802 aaronmk
        def execute(self, query):
220 2764 aaronmk
            self._is_insert = query.startswith('INSERT')
221 2797 aaronmk
            self.query_lookup = query
222 2148 aaronmk
            try:
223 2191 aaronmk
                try:
224 2802 aaronmk
                    cur = self.inner.execute(query)
225 2191 aaronmk
                    self.outer.do_autocommit()
226 2802 aaronmk
                finally: self.query = get_cur_query(self.inner, query)
227 1904 aaronmk
            except Exception, e:
228 2802 aaronmk
                _add_cursor_info(e, self, query)
229 1904 aaronmk
                self.result = e # cache the exception as the result
230
                self._cache_result()
231
                raise
232 2800 aaronmk
            if self.rowcount == 0 and query.startswith('SELECT'): # empty SELECT
233
                consume_rows(self) # fetch all rows so result will be cached
234 2762 aaronmk
            return cur
235 1894 aaronmk
236 1891 aaronmk
        def fetchone(self):
237
            row = self.inner.fetchone()
238 1899 aaronmk
            if row != None: self.result.append(row)
239
            # otherwise, fetched all rows
240 1904 aaronmk
            else: self._cache_result()
241
            return row
242
243
        def _cache_result(self):
244 1906 aaronmk
            # For inserts, only cache exceptions since inserts are not
245
            # idempotent, but an invalid insert will always be invalid
246 1930 aaronmk
            if self.query_results != None and (not self._is_insert
247 1906 aaronmk
                or isinstance(self.result, Exception)):
248
249 1894 aaronmk
                assert self.query_lookup != None
250 1916 aaronmk
                self.query_results[self.query_lookup] = self.CacheCursor(
251
                    util.dict_subset(dicts.AttrsDictView(self),
252
                    ['query', 'result', 'rowcount', 'description']))
253 1906 aaronmk
254 1916 aaronmk
        class CacheCursor:
255
            def __init__(self, cached_result): self.__dict__ = cached_result
256
257 1927 aaronmk
            def execute(self, *args, **kw_args):
258 1916 aaronmk
                if isinstance(self.result, Exception): raise self.result
259
                # otherwise, result is a rows list
260
                self.iter = iter(self.result)
261
262
            def fetchone(self):
263
                try: return self.iter.next()
264
                except StopIteration: return None
265 1891 aaronmk
266 2212 aaronmk
    def esc_value(self, value):
267 2663 aaronmk
        try: str_ = self.mogrify('%s', [value])
268
        except NotImplementedError, e:
269
            module = util.root_module(self.db)
270
            if module == 'MySQLdb':
271
                import _mysql
272
                str_ = _mysql.escape_string(value)
273
            else: raise e
274 2374 aaronmk
        return strings.to_unicode(str_)
275 2212 aaronmk
276 2347 aaronmk
    def esc_name(self, name): return esc_name(self, name) # calls global func
277
278 2814 aaronmk
    def std_code(self, str_):
279
        '''Standardizes SQL code.
280
        * Ensures that string literals are prefixed by `E`
281
        '''
282
        if str_.startswith("'"): str_ = 'E'+str_
283
        return str_
284
285 2665 aaronmk
    def can_mogrify(self):
286 2663 aaronmk
        module = util.root_module(self.db)
287 2665 aaronmk
        return module == 'psycopg2'
288 2663 aaronmk
289 2665 aaronmk
    def mogrify(self, query, params=None):
290
        if self.can_mogrify(): return self.db.cursor().mogrify(query, params)
291
        else: raise NotImplementedError("Can't mogrify query")
292
293 2671 aaronmk
    def print_notices(self):
294 2725 aaronmk
        if hasattr(self.db, 'notices'):
295
            for msg in self.db.notices:
296
                if msg not in self._notices_seen:
297
                    self._notices_seen.add(msg)
298
                    self.log_debug(msg, level=2)
299 2671 aaronmk
300 2793 aaronmk
    def run_query(self, query, cacheable=False, log_level=2,
301 2464 aaronmk
        debug_msg_ref=None):
302 2445 aaronmk
        '''
303 2464 aaronmk
        @param log_ignore_excs The log_level will be increased by 2 if the query
304
            throws one of these exceptions.
305 2664 aaronmk
        @param debug_msg_ref If specified, the log message will be returned in
306
            this instead of being output. This allows you to filter log messages
307
            depending on the result of the query.
308 2445 aaronmk
        '''
309 2167 aaronmk
        assert query != None
310
311 2047 aaronmk
        if not self.caching: cacheable = False
312 1903 aaronmk
        used_cache = False
313 2664 aaronmk
314
        def log_msg(query):
315
            if used_cache: cache_status = 'cache hit'
316
            elif cacheable: cache_status = 'cache miss'
317
            else: cache_status = 'non-cacheable'
318
            return 'DB query: '+cache_status+':\n'+strings.as_code(query, 'SQL')
319
320 1903 aaronmk
        try:
321 1927 aaronmk
            # Get cursor
322
            if cacheable:
323
                try:
324 2797 aaronmk
                    cur = self.query_results[query]
325 1927 aaronmk
                    used_cache = True
326
                except KeyError: cur = self.DbCursor(self)
327
            else: cur = self.db.cursor()
328
329 2664 aaronmk
            # Log query
330
            if self.debug and debug_msg_ref == None: # log before running
331
                self.log_debug(log_msg(query), log_level)
332
333 1927 aaronmk
            # Run query
334 2793 aaronmk
            cur.execute(query)
335 1903 aaronmk
        finally:
336 2671 aaronmk
            self.print_notices()
337 2664 aaronmk
            if self.debug and debug_msg_ref != None: # return after running
338 2793 aaronmk
                debug_msg_ref[0] = log_msg(str(get_cur_query(cur, query)))
339 1903 aaronmk
340
        return cur
341 1914 aaronmk
342 2797 aaronmk
    def is_cached(self, query): return query in self.query_results
343 2139 aaronmk
344 2907 aaronmk
    def with_autocommit(self, func):
345 2801 aaronmk
        import psycopg2.extensions
346
347
        prev_isolation_level = self.db.isolation_level
348 2907 aaronmk
        self.db.set_isolation_level(
349
            psycopg2.extensions.ISOLATION_LEVEL_AUTOCOMMIT)
350 2683 aaronmk
        try: return func()
351 2801 aaronmk
        finally: self.db.set_isolation_level(prev_isolation_level)
352 2683 aaronmk
353 2139 aaronmk
    def with_savepoint(self, func):
354 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
355 2443 aaronmk
        self.run_query('SAVEPOINT '+savepoint, log_level=4)
356 2139 aaronmk
        self._savepoint += 1
357 2683 aaronmk
        try:
358 2139 aaronmk
            try: return_val = func()
359
            finally:
360
                self._savepoint -= 1
361
                assert self._savepoint >= 0
362
        except:
363 2443 aaronmk
            self.run_query('ROLLBACK TO SAVEPOINT '+savepoint, log_level=4)
364 2139 aaronmk
            raise
365
        else:
366 2443 aaronmk
            self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
367 2191 aaronmk
            self.do_autocommit()
368 2139 aaronmk
            return return_val
369 2191 aaronmk
370
    def do_autocommit(self):
371
        '''Autocommits if outside savepoint'''
372
        assert self._savepoint >= 0
373
        if self.autocommit and self._savepoint == 0:
374 2668 aaronmk
            self.log_debug('Autocommitting')
375 2191 aaronmk
            self.db.commit()
376 2643 aaronmk
377 2819 aaronmk
    def col_info(self, col):
378 2643 aaronmk
        table = sql_gen.Table('columns', 'information_schema')
379 2821 aaronmk
        cols = ['data_type', 'column_default',
380
            cast(self, 'boolean', 'is_nullable')]
381 2643 aaronmk
382
        conds = [('table_name', col.table.name), ('column_name', col.name)]
383
        schema = col.table.schema
384
        if schema != None: conds.append(('table_schema', schema))
385
386 2819 aaronmk
        type_, default, nullable = row(select(self, table, cols, conds,
387 2874 aaronmk
            order_by='table_schema', limit=1, log_level=4))
388 2643 aaronmk
            # TODO: order_by search_path schema order
389 2819 aaronmk
        default = sql_gen.as_Code(default, self)
390
391
        return sql_gen.TypedCol(col.name, type_, default, nullable)
392 2917 aaronmk
393
    def TempFunction(self, name):
394
        if self.debug_temp: schema = None
395
        else: schema = 'pg_temp'
396
        return sql_gen.Function(name, schema)
397 1849 aaronmk
398 1869 aaronmk
connect = DbConn
399
400 832 aaronmk
##### Recoverable querying
401 15 aaronmk
402 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
403 11 aaronmk
404 2791 aaronmk
def run_query(db, query, recover=None, cacheable=False, log_level=2,
405
    log_ignore_excs=None, **kw_args):
406 2794 aaronmk
    '''For params, see DbConn.run_query()'''
407 830 aaronmk
    if recover == None: recover = False
408 2464 aaronmk
    if log_ignore_excs == None: log_ignore_excs = ()
409
    log_ignore_excs = tuple(log_ignore_excs)
410 830 aaronmk
411 2666 aaronmk
    debug_msg_ref = None # usually, db.run_query() logs query before running it
412
    # But if filtering with log_ignore_excs, wait until after exception parsing
413
    if log_ignore_excs != () or not db.can_mogrify(): debug_msg_ref = [None]
414
415 2148 aaronmk
    try:
416 2464 aaronmk
        try:
417 2794 aaronmk
            def run(): return db.run_query(query, cacheable, log_level,
418 2793 aaronmk
                debug_msg_ref, **kw_args)
419 2796 aaronmk
            if recover and not db.is_cached(query):
420 2464 aaronmk
                return with_savepoint(db, run)
421
            else: return run() # don't need savepoint if cached
422
        except Exception, e:
423
            if not recover: raise # need savepoint to run index_cols()
424
            msg = exc.str_(e)
425
426
            match = re.search(r'duplicate key value violates unique constraint '
427 2493 aaronmk
                r'"((_?[^\W_]+)_.+?)"', msg)
428 2464 aaronmk
            if match:
429
                constraint, table = match.groups()
430
                try: cols = index_cols(db, table, constraint)
431
                except NotImplementedError: raise e
432
                else: raise DuplicateKeyException(constraint, cols, e)
433
434 2493 aaronmk
            match = re.search(r'null value in column "(.+?)" violates not-null'
435 2464 aaronmk
                r' constraint', msg)
436
            if match: raise NullValueException('NOT NULL', [match.group(1)], e)
437
438
            match = re.search(r'\b(?:invalid input (?:syntax|value)\b.*?'
439
                r'|date/time field value out of range): "(.+?)"\n'
440 2535 aaronmk
                r'(?:(?s).*?)\bfunction "(.+?)"', msg)
441 2464 aaronmk
            if match:
442
                value, name = match.groups()
443
                raise FunctionValueException(name, strings.to_unicode(value), e)
444
445 2526 aaronmk
            match = re.search(r'column "(.+?)" is of type (.+?) but expression '
446 2523 aaronmk
                r'is of type', msg)
447
            if match:
448
                col, type_ = match.groups()
449
                raise MissingCastException(type_, col, e)
450
451 2493 aaronmk
            match = re.search(r'relation "(.+?)" already exists', msg)
452 2464 aaronmk
            if match: raise DuplicateTableException(match.group(1), e)
453
454 2493 aaronmk
            match = re.search(r'function "(.+?)" already exists', msg)
455 2464 aaronmk
            if match: raise DuplicateFunctionException(match.group(1), e)
456
457
            raise # no specific exception raised
458
    except log_ignore_excs:
459
        log_level += 2
460
        raise
461
    finally:
462 2666 aaronmk
        if debug_msg_ref != None and debug_msg_ref[0] != None:
463
            db.log_debug(debug_msg_ref[0], log_level)
464 830 aaronmk
465 832 aaronmk
##### Basic queries
466
467 2153 aaronmk
def next_version(name):
468 2163 aaronmk
    version = 1 # first existing name was version 0
469 2586 aaronmk
    match = re.match(r'^(.*)#(\d+)$', name)
470 2153 aaronmk
    if match:
471 2586 aaronmk
        name, version = match.groups()
472
        version = int(version)+1
473 2588 aaronmk
    return sql_gen.add_suffix(name, '#'+str(version))
474 2153 aaronmk
475 2899 aaronmk
def lock_table(db, table, mode):
476
    table = sql_gen.as_Table(table)
477
    run_query(db, 'LOCK TABLE '+table.to_str(db)+' IN '+mode+' MODE')
478
479 2789 aaronmk
def run_query_into(db, query, into=None, add_indexes_=False, **kw_args):
480 2085 aaronmk
    '''Outputs a query to a temp table.
481
    For params, see run_query().
482
    '''
483 2789 aaronmk
    if into == None: return run_query(db, query, **kw_args)
484 2790 aaronmk
485
    assert isinstance(into, sql_gen.Table)
486
487
    kw_args['recover'] = True
488
    kw_args.setdefault('log_ignore_excs', (DuplicateTableException,))
489
490 2916 aaronmk
    temp = not db.debug_temp # tables are permanent in debug_temp mode
491 2790 aaronmk
    # "temporary tables cannot specify a schema name", so remove schema
492
    if temp: into.schema = None
493
494
    # Create table
495
    while True:
496
        create_query = 'CREATE'
497
        if temp: create_query += ' TEMP'
498
        create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
499 2385 aaronmk
500 2790 aaronmk
        try:
501
            cur = run_query(db, create_query, **kw_args)
502
                # CREATE TABLE AS sets rowcount to # rows in query
503
            break
504
        except DuplicateTableException, e:
505
            into.name = next_version(into.name)
506
            # try again with next version of name
507
508
    if add_indexes_: add_indexes(db, into)
509
510
    return cur
511 2085 aaronmk
512 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
513
514 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
515
516 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
517 2293 aaronmk
    start=None, order_by=order_by_pkey, default_table=None):
518 1981 aaronmk
    '''
519 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
520 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
521 1981 aaronmk
    @param fields Use None to select all fields in the table
522 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
523 2379 aaronmk
        * container can be any iterable type
524 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
525
        * compare_right_side: sql_gen.ValueCond|literal value
526 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
527
        use all columns
528 2786 aaronmk
    @return query
529 1981 aaronmk
    '''
530 2315 aaronmk
    # Parse tables param
531 2121 aaronmk
    if not lists.is_seq(tables): tables = [tables]
532 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
533 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
534 2121 aaronmk
535 2315 aaronmk
    # Parse other params
536 2376 aaronmk
    if conds == None: conds = []
537 2650 aaronmk
    elif dicts.is_dict(conds): conds = conds.items()
538 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
539 135 aaronmk
    assert limit == None or type(limit) == int
540 865 aaronmk
    assert start == None or type(start) == int
541 2315 aaronmk
    if order_by is order_by_pkey:
542
        if distinct_on != []: order_by = None
543
        else: order_by = pkey(db, table0, recover=True)
544 865 aaronmk
545 2315 aaronmk
    query = 'SELECT'
546 2056 aaronmk
547 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
548 2056 aaronmk
549 2200 aaronmk
    # DISTINCT ON columns
550 2233 aaronmk
    if distinct_on != []:
551 2467 aaronmk
        query += '\nDISTINCT'
552 2254 aaronmk
        if distinct_on is not distinct_on_all:
553 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
554
555
    # Columns
556 2467 aaronmk
    query += '\n'
557 1135 aaronmk
    if fields == None: query += '*'
558 2765 aaronmk
    else:
559
        assert fields != []
560
        query += '\n, '.join(map(parse_col, fields))
561 2200 aaronmk
562
    # Main table
563 2467 aaronmk
    query += '\nFROM '+table0.to_str(db)
564 865 aaronmk
565 2122 aaronmk
    # Add joins
566 2271 aaronmk
    left_table = table0
567 2263 aaronmk
    for join_ in tables:
568
        table = join_.table
569 2238 aaronmk
570 2343 aaronmk
        # Parse special values
571
        if join_.type_ is sql_gen.filter_out: # filter no match
572 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
573 2853 aaronmk
                sql_gen.CompareCond(None, '~=')))
574 2343 aaronmk
575 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
576 2122 aaronmk
577
        left_table = table
578
579 865 aaronmk
    missing = True
580 2376 aaronmk
    if conds != []:
581 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
582
        else: whitespace = '\n'
583 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
584
            .to_str(db) for l, r in conds], 'WHERE')
585 865 aaronmk
        missing = False
586 2227 aaronmk
    if order_by != None:
587 2467 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
588
    if limit != None: query += '\nLIMIT '+str(limit); missing = False
589 865 aaronmk
    if start != None:
590 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
591 865 aaronmk
        missing = False
592
    if missing: warnings.warn(DbWarning(
593
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
594
595 2786 aaronmk
    return query
596 11 aaronmk
597 2054 aaronmk
def select(db, *args, **kw_args):
598
    '''For params, see mk_select() and run_query()'''
599
    recover = kw_args.pop('recover', None)
600
    cacheable = kw_args.pop('cacheable', True)
601 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
602 2054 aaronmk
603 2791 aaronmk
    return run_query(db, mk_select(db, *args, **kw_args), recover, cacheable,
604
        log_level=log_level)
605 2054 aaronmk
606 2788 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, returning=None,
607
    embeddable=False):
608 1960 aaronmk
    '''
609
    @param returning str|None An inserted column (such as pkey) to return
610 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
611 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
612
        query will be fully cached, not just if it raises an exception.
613 1960 aaronmk
    '''
614 2754 aaronmk
    table = sql_gen.remove_table_rename(sql_gen.as_Table(table))
615 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
616 2646 aaronmk
    if cols != None:
617
        cols = [sql_gen.to_name_only_col(v, table).to_str(db) for v in cols]
618 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
619 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
620 2063 aaronmk
621
    # Build query
622 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
623
    query = first_line
624 2467 aaronmk
    if cols != None: query += '\n('+', '.join(cols)+')'
625
    query += '\n'+select_query
626 2063 aaronmk
627
    if returning != None:
628 2743 aaronmk
        query += '\nRETURNING '+sql_gen.to_name_only_col(returning).to_str(db)
629 2063 aaronmk
630 2070 aaronmk
    if embeddable:
631 2327 aaronmk
        assert returning != None
632
633 2070 aaronmk
        # Create function
634 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
635 2327 aaronmk
        return_type = 'SETOF '+returning.to_str(db)+'%TYPE'
636 2189 aaronmk
        while True:
637
            try:
638 2918 aaronmk
                function = db.TempFunction(function_name)
639 2194 aaronmk
640 2189 aaronmk
                function_query = '''\
641 2698 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''()
642 2467 aaronmk
RETURNS '''+return_type+'''
643
LANGUAGE sql
644
AS $$
645 2788 aaronmk
'''+query+''';
646 2467 aaronmk
$$;
647 2070 aaronmk
'''
648 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
649 2464 aaronmk
                    log_ignore_excs=(DuplicateFunctionException,))
650 2189 aaronmk
                break # this version was successful
651
            except DuplicateFunctionException, e:
652
                function_name = next_version(function_name)
653
                # try again with next version of name
654 2070 aaronmk
655 2337 aaronmk
        # Return query that uses function
656 2698 aaronmk
        func_table = sql_gen.NamedTable('f', sql_gen.FunctionCall(function),
657 2742 aaronmk
            [returning]) # AS clause requires function alias
658 2787 aaronmk
        return mk_select(db, func_table, start=0, order_by=None)
659 2070 aaronmk
660 2787 aaronmk
    return query
661 2066 aaronmk
662
def insert_select(db, *args, **kw_args):
663 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
664 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
665
        values in
666 2072 aaronmk
    '''
667 2386 aaronmk
    into = kw_args.pop('into', None)
668
    if into != None: kw_args['embeddable'] = True
669 2066 aaronmk
    recover = kw_args.pop('recover', None)
670
    cacheable = kw_args.pop('cacheable', True)
671 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
672 2066 aaronmk
673 2789 aaronmk
    return run_query_into(db, mk_insert_select(db, *args, **kw_args), into,
674 2787 aaronmk
        recover=recover, cacheable=cacheable, log_level=log_level)
675 2063 aaronmk
676 2738 aaronmk
default = sql_gen.default # tells insert() to use the default value for a column
677 2066 aaronmk
678 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
679 2085 aaronmk
    '''For params, see insert_select()'''
680 1960 aaronmk
    if lists.is_seq(row): cols = None
681
    else:
682
        cols = row.keys()
683
        row = row.values()
684 2738 aaronmk
    row = list(row) # ensure that "== []" works
685 1960 aaronmk
686 2738 aaronmk
    if row == []: query = None
687
    else: query = sql_gen.Values(row).to_str(db)
688 1961 aaronmk
689 2788 aaronmk
    return insert_select(db, table, cols, query, *args, **kw_args)
690 11 aaronmk
691 2402 aaronmk
def mk_update(db, table, changes=None, cond=None):
692
    '''
693
    @param changes [(col, new_value),...]
694
        * container can be any iterable type
695
        * col: sql_gen.Code|str (for col name)
696
        * new_value: sql_gen.Code|literal value
697
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
698
    @return str query
699
    '''
700
    query = 'UPDATE '+sql_gen.as_Table(table).to_str(db)+'\nSET\n'
701 2405 aaronmk
    query += ',\n'.join((sql_gen.to_name_only_col(col, table).to_str(db)+' = '
702 2402 aaronmk
        +sql_gen.as_Value(new_value).to_str(db) for col, new_value in changes))
703 2467 aaronmk
    if cond != None: query += '\nWHERE\n'+cond.to_str(db)
704 2402 aaronmk
705
    return query
706
707
def update(db, *args, **kw_args):
708
    '''For params, see mk_update() and run_query()'''
709
    recover = kw_args.pop('recover', None)
710
711 2791 aaronmk
    return run_query(db, mk_update(db, *args, **kw_args), recover)
712 2402 aaronmk
713 135 aaronmk
def last_insert_id(db):
714 1849 aaronmk
    module = util.root_module(db.db)
715 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
716
    elif module == 'MySQLdb': return db.insert_id()
717
    else: return None
718 13 aaronmk
719 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
720 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
721 2415 aaronmk
    to names that will be distinct among the columns' tables.
722 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
723 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
724
    @param into The table for the new columns.
725 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
726
        columns will be included in the mapping even if they are not in cols.
727
        The tables of the provided Col objects will be changed to into, so make
728
        copies of them if you want to keep the original tables.
729
    @param as_items Whether to return a list of dict items instead of a dict
730 2383 aaronmk
    @return dict(orig_col=new_col, ...)
731
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
732 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
733
        * All mappings use the into table so its name can easily be
734 2383 aaronmk
          changed for all columns at once
735
    '''
736 2415 aaronmk
    cols = lists.uniqify(cols)
737
738 2394 aaronmk
    items = []
739 2389 aaronmk
    for col in preserve:
740 2390 aaronmk
        orig_col = copy.copy(col)
741 2392 aaronmk
        col.table = into
742 2394 aaronmk
        items.append((orig_col, col))
743
    preserve = set(preserve)
744
    for col in cols:
745 2716 aaronmk
        if col not in preserve:
746
            items.append((col, sql_gen.Col(str(col), into, col.srcs)))
747 2394 aaronmk
748
    if not as_items: items = dict(items)
749
    return items
750 2383 aaronmk
751 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
752 2391 aaronmk
    '''For params, see mk_flatten_mapping()
753
    @return See return value of mk_flatten_mapping()
754
    '''
755 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
756
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
757 2786 aaronmk
    run_query_into(db, mk_select(db, joins, cols, limit=limit, start=start),
758 2846 aaronmk
        into=into, add_indexes_=True)
759 2394 aaronmk
    return dict(items)
760 2391 aaronmk
761 2751 aaronmk
def mk_track_data_error(db, errors_table, cols, value, error_code, error):
762 2810 aaronmk
    assert cols != ()
763
764 2751 aaronmk
    cols = map(sql_gen.to_name_only_col, cols)
765
766
    columns_cols = ['column']
767
    columns = sql_gen.NamedValues('columns', columns_cols,
768
        [[c.name] for c in cols])
769
    values_cols = ['value', 'error_code', 'error']
770
    values = sql_gen.NamedValues('values', values_cols,
771
        [value, error_code, error])
772
773
    select_cols = columns_cols+values_cols
774
    name_only_cols = map(sql_gen.to_name_only_col, select_cols)
775 2755 aaronmk
    errors_table = sql_gen.NamedTable('errors', errors_table)
776 2751 aaronmk
    joins = [columns, sql_gen.Join(values, type_='CROSS'),
777
        sql_gen.Join(errors_table, dict(zip(name_only_cols, select_cols)),
778
        sql_gen.filter_out)]
779
780
    return mk_insert_select(db, errors_table, name_only_cols,
781 2787 aaronmk
        mk_select(db, joins, select_cols, order_by=None))
782 2751 aaronmk
783 2810 aaronmk
def track_data_error(db, errors_table, cols, *args, **kw_args):
784 2758 aaronmk
    '''
785
    @param errors_table If None, does nothing.
786
    '''
787 2810 aaronmk
    if errors_table == None or cols == (): return
788
    run_query(db, mk_track_data_error(db, errors_table, cols, *args, **kw_args),
789 2758 aaronmk
        cacheable=True, log_level=4)
790
791 2704 aaronmk
def cast(db, type_, col, errors_table=None):
792
    '''Casts an (unrenamed) column or value.
793 2769 aaronmk
    If errors_table set and col has srcs, saves errors in errors_table (using
794
    col's srcs attr as the source columns) and converts errors to warnings.
795 2820 aaronmk
    @param col str|sql_gen.Col|sql_gen.Literal
796 2704 aaronmk
    @param errors_table None|sql_gen.Table|str
797
    '''
798 2820 aaronmk
    col = sql_gen.as_Col(col)
799 2769 aaronmk
    save_errors = (errors_table != None and isinstance(col, sql_gen.Col)
800
        and col.srcs != ())
801
    if not save_errors: # can't save errors
802
        return sql_gen.CustomCode(col.to_str(db)+'::'+type_) # just cast
803 2704 aaronmk
804
    assert not isinstance(col, sql_gen.NamedCol)
805
806 2770 aaronmk
    errors_table = sql_gen.as_Table(errors_table)
807
    srcs = map(sql_gen.to_name_only_col, col.srcs)
808
    function_name = str(sql_gen.FunctionCall(type_, *srcs))
809 2918 aaronmk
    function = db.TempFunction(function_name)
810 2704 aaronmk
811 2718 aaronmk
    while True:
812
        # Create function definition
813
        query = '''\
814 2704 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''(value text)
815
RETURNS '''+type_+'''
816
LANGUAGE plpgsql
817
STRICT
818
AS $$
819
BEGIN
820
    /* The explicit cast to the return type is needed to make the cast happen
821
    inside the try block. (Implicit casts to the return type happen at the end
822
    of the function, outside any block.) */
823
    RETURN value::'''+type_+''';
824
EXCEPTION
825
    WHEN data_exception THEN
826
        -- Save error in errors table.
827 2751 aaronmk
        -- Insert the value and error for *each* source column.
828
'''+mk_track_data_error(db, errors_table, srcs,
829
    *map(sql_gen.CustomCode, ['value', 'SQLSTATE', 'SQLERRM']))+''';
830 2704 aaronmk
831
        RAISE WARNING '%', SQLERRM;
832
        RETURN NULL;
833
END;
834
$$;
835
'''
836 2718 aaronmk
837
        # Create function
838
        try:
839
            run_query(db, query, recover=True, cacheable=True,
840
                log_ignore_excs=(DuplicateFunctionException,))
841
            break # successful
842
        except DuplicateFunctionException:
843 2770 aaronmk
            function.name = next_version(function.name)
844
            # try again with next version of name
845 2704 aaronmk
846
    return sql_gen.FunctionCall(function, col)
847
848 2414 aaronmk
##### Database structure queries
849
850 2426 aaronmk
def table_row_count(db, table, recover=None):
851 2786 aaronmk
    return value(run_query(db, mk_select(db, table, [sql_gen.row_count],
852 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
853 2426 aaronmk
854 2414 aaronmk
def table_cols(db, table, recover=None):
855
    return list(col_names(select(db, table, limit=0, order_by=None,
856 2443 aaronmk
        recover=recover, log_level=4)))
857 2414 aaronmk
858 2291 aaronmk
def pkey(db, table, recover=None):
859 832 aaronmk
    '''Assumed to be first column in table'''
860 2339 aaronmk
    return table_cols(db, table, recover)[0]
861 832 aaronmk
862 2559 aaronmk
not_null_col = 'not_null_col'
863 2340 aaronmk
864
def table_not_null_col(db, table, recover=None):
865
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
866
    if not_null_col in table_cols(db, table, recover): return not_null_col
867
    else: return pkey(db, table, recover)
868
869 853 aaronmk
def index_cols(db, table, index):
870
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
871
    automatically created. When you don't know whether something is a UNIQUE
872
    constraint or a UNIQUE index, use this function.'''
873 1909 aaronmk
    module = util.root_module(db.db)
874
    if module == 'psycopg2':
875
        return list(values(run_query(db, '''\
876 853 aaronmk
SELECT attname
877 866 aaronmk
FROM
878
(
879
        SELECT attnum, attname
880
        FROM pg_index
881
        JOIN pg_class index ON index.oid = indexrelid
882
        JOIN pg_class table_ ON table_.oid = indrelid
883
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
884
        WHERE
885 2782 aaronmk
            table_.relname = '''+db.esc_value(table)+'''
886
            AND index.relname = '''+db.esc_value(index)+'''
887 866 aaronmk
    UNION
888
        SELECT attnum, attname
889
        FROM
890
        (
891
            SELECT
892
                indrelid
893
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
894
                    AS indkey
895
            FROM pg_index
896
            JOIN pg_class index ON index.oid = indexrelid
897
            JOIN pg_class table_ ON table_.oid = indrelid
898
            WHERE
899 2782 aaronmk
                table_.relname = '''+db.esc_value(table)+'''
900
                AND index.relname = '''+db.esc_value(index)+'''
901 866 aaronmk
        ) s
902
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
903
) s
904 853 aaronmk
ORDER BY attnum
905 2782 aaronmk
'''
906
            , cacheable=True, log_level=4)))
907 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
908
        ' database')
909 853 aaronmk
910 464 aaronmk
def constraint_cols(db, table, constraint):
911 1849 aaronmk
    module = util.root_module(db.db)
912 464 aaronmk
    if module == 'psycopg2':
913
        return list(values(run_query(db, '''\
914
SELECT attname
915
FROM pg_constraint
916
JOIN pg_class ON pg_class.oid = conrelid
917
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
918
WHERE
919 2783 aaronmk
    relname = '''+db.esc_value(table)+'''
920
    AND conname = '''+db.esc_value(constraint)+'''
921 464 aaronmk
ORDER BY attnum
922 2783 aaronmk
'''
923
            )))
924 464 aaronmk
    else: raise NotImplementedError("Can't list constraint columns for "+module+
925
        ' database')
926
927 2096 aaronmk
row_num_col = '_row_num'
928
929 2847 aaronmk
def add_index(db, exprs, table=None, unique=False, ensure_not_null=True):
930 2688 aaronmk
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
931 2538 aaronmk
    Currently, only function calls are supported as expressions.
932 2847 aaronmk
    @param ensure_not_null If set, translates NULL values to sentinel values.
933
        This allows indexes to be used for comparisons where NULLs are equal.
934 2538 aaronmk
    '''
935 2688 aaronmk
    if not lists.is_seq(exprs): exprs = [exprs]
936 2538 aaronmk
937 2688 aaronmk
    # Parse exprs
938
    old_exprs = exprs[:]
939
    exprs = []
940
    cols = []
941
    for i, expr in enumerate(old_exprs):
942
        expr = copy.deepcopy(expr) # don't modify input!
943 2823 aaronmk
        expr = sql_gen.as_Col(expr, table)
944 2688 aaronmk
945 2823 aaronmk
        # Handle nullable columns
946 2860 aaronmk
        if ensure_not_null:
947
            try: expr = sql_gen.ensure_not_null(db, expr)
948
            except KeyError: pass # unknown type, so just create plain index
949 2823 aaronmk
950 2688 aaronmk
        # Extract col
951
        if isinstance(expr, sql_gen.FunctionCall):
952
            col = expr.args[0]
953
            expr = sql_gen.Expr(expr)
954
        else: col = expr
955 2823 aaronmk
        assert isinstance(col, sql_gen.Col)
956 2688 aaronmk
957
        # Extract table
958
        if table == None:
959
            assert sql_gen.is_table_col(col)
960
            table = col.table
961
962
        col.table = None
963
964
        exprs.append(expr)
965
        cols.append(col)
966 2408 aaronmk
967 2688 aaronmk
    table = sql_gen.as_Table(table)
968
    index = sql_gen.Table(str(sql_gen.Col(','.join(map(str, cols)), table)))
969
970
    str_ = 'CREATE'
971
    if unique: str_ += ' UNIQUE'
972
    str_ += ' INDEX '+index.to_str(db)+' ON '+table.to_str(db)+' ('+(
973
        ', '.join((v.to_str(db) for v in exprs)))+')'
974
975
    try: run_query(db, str_, recover=True, cacheable=True, log_level=3)
976 2408 aaronmk
    except DuplicateTableException: pass # index already existed
977
978 2686 aaronmk
def add_pkey(db, table, cols=None, recover=None):
979
    '''Adds a primary key.
980
    @param cols [sql_gen.Col,...] The columns in the primary key.
981
        Defaults to the first column in the table.
982 2406 aaronmk
    @pre The table must not already have a primary key.
983
    '''
984
    table = sql_gen.as_Table(table)
985 2686 aaronmk
    if cols == None: cols = [pkey(db, table, recover)]
986
    col_strs = [sql_gen.to_name_only_col(v).to_str(db) for v in cols]
987 2406 aaronmk
988 2781 aaronmk
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD PRIMARY KEY ('
989
        +(', '.join(col_strs))+')', recover=True, cacheable=True, log_level=3,
990
        log_ignore_excs=(DuplicateTableException,))
991 2406 aaronmk
992 2772 aaronmk
already_indexed = object() # tells add_indexes() the pkey has already been added
993
994
def add_indexes(db, table, has_pkey=True):
995
    '''Adds an index on all columns in a table.
996
    @param has_pkey bool|already_indexed Whether a pkey instead of a regular
997
        index should be added on the first column.
998
        * If already_indexed, the pkey is assumed to have already been added
999
    '''
1000
    cols = table_cols(db, table)
1001
    if has_pkey:
1002
        if has_pkey is not already_indexed: add_pkey(db, table)
1003
        cols = cols[1:]
1004
    for col in cols: add_index(db, col, table)
1005
1006 2086 aaronmk
def add_row_num(db, table):
1007 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
1008
    be the primary key.'''
1009 2320 aaronmk
    table = sql_gen.as_Table(table).to_str(db)
1010 2096 aaronmk
    run_query(db, 'ALTER TABLE '+table+' ADD COLUMN '+row_num_col
1011 2443 aaronmk
        +' serial NOT NULL PRIMARY KEY', log_level=3)
1012 2086 aaronmk
1013 2889 aaronmk
def drop_table(db, table):
1014
    table = sql_gen.as_Table(table)
1015
    return run_query(db, 'DROP TABLE IF EXISTS '+table.to_str(db)+' CASCADE')
1016
1017 2760 aaronmk
def create_table(db, table, cols, has_pkey=True, col_indexes=True):
1018 2675 aaronmk
    '''Creates a table.
1019 2681 aaronmk
    @param cols [sql_gen.TypedCol,...] The column names and types
1020
    @param has_pkey If set, the first column becomes the primary key.
1021 2760 aaronmk
    @param col_indexes bool|[ref]
1022
        * If True, indexes will be added on all non-pkey columns.
1023
        * If a list reference, [0] will be set to a function to do this.
1024
          This can be used to delay index creation until the table is populated.
1025 2675 aaronmk
    '''
1026
    table = sql_gen.as_Table(table)
1027
1028 2681 aaronmk
    if has_pkey:
1029
        cols[0] = pkey = copy.copy(cols[0]) # don't modify input!
1030 2872 aaronmk
        pkey.constraints = 'PRIMARY KEY'
1031 2681 aaronmk
1032 2675 aaronmk
    str_ = 'CREATE TABLE '+table.to_str(db)+' (\n'
1033 2681 aaronmk
    str_ += '\n, '.join(v.to_str(db) for v in cols)
1034
    str_ += '\n);\n'
1035 2675 aaronmk
    run_query(db, str_, cacheable=True, log_level=2)
1036 2760 aaronmk
1037
    # Add indexes
1038 2773 aaronmk
    if has_pkey: has_pkey = already_indexed
1039
    def add_indexes_(): add_indexes(db, table, has_pkey)
1040
    if isinstance(col_indexes, list): col_indexes[0] = add_indexes_ # defer
1041
    elif col_indexes: add_indexes_() # add now
1042 2675 aaronmk
1043 2684 aaronmk
def vacuum(db, table):
1044
    table = sql_gen.as_Table(table)
1045
    db.with_autocommit(lambda: run_query(db, 'VACUUM ANALYZE '+table.to_str(db),
1046
        log_level=3))
1047
1048 2732 aaronmk
def truncate(db, table, schema='public'):
1049 2777 aaronmk
    table = sql_gen.as_Table(table, schema)
1050 2733 aaronmk
    return run_query(db, 'TRUNCATE '+table.to_str(db)+' CASCADE')
1051 2732 aaronmk
1052 2729 aaronmk
def tables(db, schema_like='public', table_like='%', exact=False):
1053
    if exact: compare = '='
1054
    else: compare = 'LIKE'
1055
1056 1849 aaronmk
    module = util.root_module(db.db)
1057 832 aaronmk
    if module == 'psycopg2':
1058 2779 aaronmk
        conds = [('schemaname', sql_gen.CompareCond(schema_like, compare)),
1059
            ('tablename', sql_gen.CompareCond(table_like, compare))]
1060
        return values(select(db, 'pg_tables', ['tablename'], conds,
1061
            order_by='tablename', log_level=4))
1062 1968 aaronmk
    elif module == 'MySQLdb':
1063 2779 aaronmk
        return values(run_query(db, 'SHOW TABLES LIKE '+db.esc_value(table_like)
1064
            , cacheable=True, log_level=4))
1065 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
1066 830 aaronmk
1067 2726 aaronmk
def table_exists(db, table):
1068
    table = sql_gen.as_Table(table)
1069 2730 aaronmk
    return list(tables(db, table.schema, table.name, exact=True)) != []
1070 2726 aaronmk
1071 2913 aaronmk
def function_exists(db, function):
1072
    function = sql_gen.as_Function(function)
1073
1074
    info_table = sql_gen.Table('routines', 'information_schema')
1075
    conds = [('routine_name', function.name)]
1076
    schema = function.schema
1077
    if schema != None: conds.append(('routine_schema', schema))
1078 2914 aaronmk
    # Exclude trigger functions, since they cannot be called directly
1079
    conds.append(('data_type', sql_gen.CompareCond('trigger', '!=')))
1080 2913 aaronmk
1081
    return list(values(select(db, info_table, ['routine_name'], conds,
1082
        order_by='routine_schema', limit=1, log_level=4))) != []
1083
        # TODO: order_by search_path schema order
1084
1085 2726 aaronmk
def errors_table(db, table, if_exists=True):
1086
    '''
1087
    @param if_exists If set, returns None if the errors table doesn't exist
1088
    @return None|sql_gen.Table
1089
    '''
1090
    table = sql_gen.as_Table(table)
1091
    if table.srcs != (): table = table.srcs[0]
1092
1093
    errors_table = sql_gen.suffixed_table(table, '.errors')
1094
    if if_exists and not table_exists(db, errors_table): return None
1095
    return errors_table
1096
1097 833 aaronmk
##### Database management
1098
1099 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
1100
    '''For kw_args, see tables()'''
1101
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
1102 833 aaronmk
1103 832 aaronmk
##### Heuristic queries
1104
1105 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
1106 1554 aaronmk
    '''Recovers from errors.
1107 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
1108
    '''
1109 2642 aaronmk
    row = sql_gen.ColDict(db, table, row)
1110 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
1111
1112 471 aaronmk
    try:
1113 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
1114 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
1115
            row_ct_ref[0] += cur.rowcount
1116
        return value(cur)
1117 471 aaronmk
    except DuplicateKeyException, e:
1118 2653 aaronmk
        row = sql_gen.ColDict(db, table,
1119
            util.dict_subset_right_join(row, e.cols))
1120
        return value(select(db, table, [pkey_], row, recover=True))
1121 471 aaronmk
1122 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
1123 830 aaronmk
    '''Recovers from errors'''
1124 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
1125 14 aaronmk
    except StopIteration:
1126 40 aaronmk
        if not create: raise
1127 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
1128 2078 aaronmk
1129 2593 aaronmk
def is_func_result(col):
1130
    return col.table.name.find('(') >= 0 and col.name == 'result'
1131
1132 2592 aaronmk
def into_table_name(out_table, in_tables0, mapping, is_func):
1133 2615 aaronmk
    def in_col_str(in_col):
1134
        in_col = sql_gen.remove_col_rename(in_col)
1135
        if isinstance(in_col, sql_gen.Col):
1136
            table = in_col.table
1137
            if table == in_tables0:
1138
                in_col = sql_gen.to_name_only_col(in_col)
1139
            elif is_func_result(in_col): in_col = table # omit col name
1140
        return str(in_col)
1141
1142 2580 aaronmk
    str_ = str(out_table)
1143
    if is_func:
1144 2615 aaronmk
        str_ += '('
1145 2580 aaronmk
1146 2615 aaronmk
        try: value_in_col = mapping['value']
1147
        except KeyError:
1148 2654 aaronmk
            str_ += ', '.join((str(k)+'='+in_col_str(v)
1149 2615 aaronmk
                for k, v in mapping.iteritems()))
1150
        else: str_ += in_col_str(value_in_col)
1151
1152
        str_ += ')'
1153 2812 aaronmk
    else:
1154
        out_col = 'rank'
1155
        try: in_col = mapping[out_col]
1156
        except KeyError: str_ += '_pkeys'
1157
        else: # has a rank column, so hierarchical
1158
            str_ += '['+str(out_col)+'='+in_col_str(in_col)+']'
1159 2580 aaronmk
    return str_
1160
1161 2508 aaronmk
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None, into=None,
1162 2805 aaronmk
    default=None, is_func=False, on_error=exc.raise_):
1163 2078 aaronmk
    '''Recovers from errors.
1164
    Only works under PostgreSQL (uses INSERT RETURNING).
1165 2909 aaronmk
    IMPORTANT: Must be run at the *beginning* of a transaction.
1166 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
1167
        tables to join with it using the main input table's pkey
1168 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
1169 2612 aaronmk
        * out_table_col: str (*not* sql_gen.Col)
1170
        * in_table_col: sql_gen.Col|literal-value
1171 2489 aaronmk
    @param into The table to contain the output and input pkeys.
1172 2574 aaronmk
        Defaults to `out_table.name+'_pkeys'`.
1173 2509 aaronmk
    @param default The *output* column to use as the pkey for missing rows.
1174
        If this output column does not exist in the mapping, uses None.
1175 2552 aaronmk
    @param is_func Whether out_table is the name of a SQL function, not a table
1176 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
1177 2078 aaronmk
    '''
1178 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
1179 2312 aaronmk
1180 2450 aaronmk
    def log_debug(msg): db.log_debug(msg, level=1.5)
1181 2505 aaronmk
    def col_ustr(str_):
1182 2567 aaronmk
        return strings.repr_no_u(sql_gen.remove_col_rename(str_))
1183 2450 aaronmk
1184 2910 aaronmk
    log_debug('********** New iteration **********')
1185
    log_debug('Inserting these input columns into '+strings.as_tt(
1186
        out_table.to_str(db))+':\n'+strings.as_table(mapping, ustr=col_ustr))
1187
1188
    log_debug('Locking output table to prevent concurrent duplicate keys')
1189
    # Must happen before any SELECT query on the table to avoid lock upgrades
1190
    lock_table(db, out_table, 'EXCLUSIVE')
1191
1192 2768 aaronmk
    out_pkey = pkey(db, out_table, recover=True)
1193
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
1194
1195
    if mapping == {}: # need at least one column for INSERT SELECT
1196
        mapping = {out_pkey: None} # ColDict will replace with default value
1197
1198 2382 aaronmk
    # Create input joins from list of input tables
1199
    in_tables_ = in_tables[:] # don't modify input!
1200
    in_tables0 = in_tables_.pop(0) # first table is separate
1201 2735 aaronmk
    errors_table_ = errors_table(db, in_tables0)
1202 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
1203 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
1204 2460 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v,
1205
        {in_pkey: sql_gen.join_same_not_null}) for v in in_tables_]
1206 2131 aaronmk
1207 2592 aaronmk
    if into == None:
1208
        into = into_table_name(out_table, in_tables0, mapping, is_func)
1209
    into = sql_gen.as_Table(into)
1210
1211 2702 aaronmk
    # Set column sources
1212
    in_cols = filter(sql_gen.is_table_col, mapping.values())
1213
    for col in in_cols:
1214
        if col.table == in_tables0: col.set_srcs(sql_gen.src_self)
1215
1216 2486 aaronmk
    log_debug('Joining together input tables into temp table')
1217 2395 aaronmk
    # Place in new table for speed and so don't modify input if values edited
1218 2584 aaronmk
    in_table = sql_gen.Table('in')
1219 2702 aaronmk
    mapping = dicts.join(mapping, flatten(db, in_table, input_joins, in_cols,
1220
        preserve=[in_pkey_col], start=0))
1221 2395 aaronmk
    input_joins = [in_table]
1222 2486 aaronmk
    db.log_debug('Temp table: '+strings.as_tt(in_table.to_str(db)), level=2)
1223 2395 aaronmk
1224 2656 aaronmk
    mapping = sql_gen.ColDict(db, out_table, mapping)
1225
        # after applying dicts.join() because that returns a plain dict
1226
1227 2509 aaronmk
    # Resolve default value column
1228 2841 aaronmk
    if default != None:
1229
        try: default = mapping[default]
1230
        except KeyError:
1231 2509 aaronmk
            db.log_debug('Default value column '
1232
                +strings.as_tt(strings.repr_no_u(default))
1233 2511 aaronmk
                +' does not exist in mapping, falling back to None', level=2.1)
1234 2509 aaronmk
            default = None
1235
1236 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
1237 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
1238
1239 2201 aaronmk
    pkeys_table_exists_ref = [False]
1240 2420 aaronmk
    def insert_into_pkeys(joins, cols):
1241 2786 aaronmk
        query = mk_select(db, joins, cols, order_by=None, start=0)
1242 2201 aaronmk
        if pkeys_table_exists_ref[0]:
1243 2786 aaronmk
            insert_select(db, into, pkeys_names, query)
1244 2201 aaronmk
        else:
1245 2786 aaronmk
            run_query_into(db, query, into=into)
1246 2201 aaronmk
            pkeys_table_exists_ref[0] = True
1247
1248 2429 aaronmk
    limit_ref = [None]
1249 2380 aaronmk
    conds = set()
1250 2811 aaronmk
    distinct_on = sql_gen.ColDict(db, out_table)
1251 2325 aaronmk
    def mk_main_select(joins, cols):
1252 2811 aaronmk
        distinct_on_cols = [c.to_Col() for c in distinct_on.values()]
1253
        return mk_select(db, joins, cols, conds, distinct_on_cols,
1254 2429 aaronmk
            limit=limit_ref[0], start=0)
1255 2132 aaronmk
1256 2519 aaronmk
    exc_strs = set()
1257 2309 aaronmk
    def log_exc(e):
1258 2519 aaronmk
        e_str = exc.str_(e, first_line_only=True)
1259
        log_debug('Caught exception: '+e_str)
1260
        assert e_str not in exc_strs # avoid infinite loops
1261
        exc_strs.add(e_str)
1262 2735 aaronmk
1263 2451 aaronmk
    def remove_all_rows():
1264 2896 aaronmk
        log_debug('Ignoring all rows')
1265 2429 aaronmk
        limit_ref[0] = 0 # just create an empty pkeys table
1266 2735 aaronmk
1267 2758 aaronmk
    def ignore(in_col, value, e):
1268
        track_data_error(db, errors_table_, in_col.srcs, value, e.cause.pgcode,
1269
            e.cause.pgerror)
1270 2895 aaronmk
        log_debug('Ignoring rows with '+strings.as_tt(repr(in_col))+' = '
1271 2545 aaronmk
            +strings.as_tt(repr(value)))
1272 2895 aaronmk
1273 2758 aaronmk
    def remove_rows(in_col, value, e):
1274
        ignore(in_col, value, e)
1275 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
1276
        assert cond not in conds # avoid infinite loops
1277 2380 aaronmk
        conds.add(cond)
1278 2895 aaronmk
1279 2758 aaronmk
    def invalid2null(in_col, value, e):
1280
        ignore(in_col, value, e)
1281 2403 aaronmk
        update(db, in_table, [(in_col, None)],
1282
            sql_gen.ColValueCond(in_col, value))
1283 2245 aaronmk
1284 2589 aaronmk
    def insert_pkeys_table(which):
1285
        return sql_gen.Table(sql_gen.add_suffix(in_table.name,
1286
            '_insert_'+which+'_pkeys'))
1287
    insert_out_pkeys = insert_pkeys_table('out')
1288
    insert_in_pkeys = insert_pkeys_table('in')
1289
1290 2206 aaronmk
    # Do inserts and selects
1291 2642 aaronmk
    join_cols = sql_gen.ColDict(db, out_table)
1292 2206 aaronmk
    while True:
1293 2521 aaronmk
        if limit_ref[0] == 0: # special case
1294
            log_debug('Creating an empty pkeys table')
1295 2786 aaronmk
            cur = run_query_into(db, mk_select(db, out_table, [out_pkey],
1296 2521 aaronmk
                limit=limit_ref[0]), into=insert_out_pkeys)
1297
            break # don't do main case
1298
1299 2303 aaronmk
        has_joins = join_cols != {}
1300
1301 2903 aaronmk
        log_debug('Trying to insert new rows')
1302
1303 2305 aaronmk
        # Prepare to insert new rows
1304 2325 aaronmk
        insert_joins = input_joins[:] # don't modify original!
1305 2403 aaronmk
        insert_args = dict(recover=True, cacheable=False)
1306 2303 aaronmk
        if has_joins:
1307 2325 aaronmk
            insert_joins.append(sql_gen.Join(out_table, join_cols,
1308
                sql_gen.filter_out))
1309
        else:
1310 2404 aaronmk
            insert_args.update(dict(returning=out_pkey, into=insert_out_pkeys))
1311 2786 aaronmk
        main_select = mk_main_select(insert_joins, mapping.values())
1312 2303 aaronmk
1313 2206 aaronmk
        try:
1314 2518 aaronmk
            cur = insert_select(db, out_table, mapping.keys(), main_select,
1315
                **insert_args)
1316 2357 aaronmk
            break # insert successful
1317 2206 aaronmk
        except DuplicateKeyException, e:
1318 2309 aaronmk
            log_exc(e)
1319
1320 2258 aaronmk
            old_join_cols = join_cols.copy()
1321 2811 aaronmk
            distinct_on.update(util.dict_subset(mapping, e.cols))
1322 2565 aaronmk
            join_cols.update(util.dict_subset_right_join(mapping, e.cols))
1323 2486 aaronmk
            log_debug('Ignoring existing rows, comparing on these columns:\n'
1324 2505 aaronmk
                +strings.as_inline_table(join_cols, ustr=col_ustr))
1325 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
1326 2230 aaronmk
        except NullValueException, e:
1327 2309 aaronmk
            log_exc(e)
1328
1329 2230 aaronmk
            out_col, = e.cols
1330
            try: in_col = mapping[out_col]
1331 2356 aaronmk
            except KeyError:
1332 2486 aaronmk
                log_debug('Missing mapping for NOT NULL column '+out_col)
1333 2451 aaronmk
                remove_all_rows()
1334 2758 aaronmk
            else: remove_rows(in_col, None, e)
1335 2542 aaronmk
        except FunctionValueException, e:
1336
            log_exc(e)
1337
1338
            func_name = e.name
1339
            value = e.value
1340
            for out_col, in_col in mapping.iteritems():
1341 2758 aaronmk
                in_col = sql_gen.unwrap_func_call(in_col, func_name)
1342
                invalid2null(in_col, value, e)
1343 2525 aaronmk
        except MissingCastException, e:
1344
            log_exc(e)
1345
1346
            out_col = e.col
1347 2706 aaronmk
            type_ = e.type
1348
1349 2724 aaronmk
            log_debug('Casting '+strings.as_tt(out_col)+' input to '
1350
                +strings.as_tt(type_))
1351 2735 aaronmk
            def wrap_func(col): return cast(db, type_, col, errors_table_)
1352 2705 aaronmk
            mapping[out_col] = sql_gen.wrap(wrap_func, mapping[out_col])
1353 2429 aaronmk
        except DatabaseErrors, e:
1354
            log_exc(e)
1355
1356 2808 aaronmk
            log_debug('No handler for exception')
1357
            on_error(e)
1358 2451 aaronmk
            remove_all_rows()
1359 2358 aaronmk
        # after exception handled, rerun loop with additional constraints
1360 2132 aaronmk
1361 2357 aaronmk
    if row_ct_ref != None and cur.rowcount >= 0:
1362
        row_ct_ref[0] += cur.rowcount
1363
1364
    if has_joins:
1365
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
1366 2486 aaronmk
        log_debug('Getting output table pkeys of existing/inserted rows')
1367 2420 aaronmk
        insert_into_pkeys(select_joins, pkeys_cols)
1368 2357 aaronmk
    else:
1369 2404 aaronmk
        add_row_num(db, insert_out_pkeys) # for joining with input pkeys
1370 2357 aaronmk
1371 2486 aaronmk
        log_debug('Getting input table pkeys of inserted rows')
1372 2786 aaronmk
        run_query_into(db, mk_main_select(input_joins, [in_pkey]),
1373 2404 aaronmk
            into=insert_in_pkeys)
1374
        add_row_num(db, insert_in_pkeys) # for joining with output pkeys
1375 2357 aaronmk
1376 2428 aaronmk
        assert table_row_count(db, insert_out_pkeys) == table_row_count(db,
1377
            insert_in_pkeys)
1378
1379 2486 aaronmk
        log_debug('Combining output and input pkeys in inserted order')
1380 2404 aaronmk
        pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
1381 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
1382 2420 aaronmk
        insert_into_pkeys(pkey_joins, pkeys_names)
1383 2357 aaronmk
1384 2486 aaronmk
    db.log_debug('Adding pkey on pkeys table to enable fast joins', level=2.5)
1385 2594 aaronmk
    add_pkey(db, into)
1386 2407 aaronmk
1387 2508 aaronmk
    log_debug('Setting pkeys of missing rows to '+strings.as_tt(repr(default)))
1388 2489 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(into,
1389 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
1390
        # must use join_same_not_null or query will take forever
1391 2420 aaronmk
    insert_into_pkeys(missing_rows_joins,
1392 2508 aaronmk
        [in_pkey_col, sql_gen.NamedCol(out_pkey, default)])
1393 2357 aaronmk
1394 2489 aaronmk
    assert table_row_count(db, into) == table_row_count(db, in_table)
1395 2428 aaronmk
1396 2702 aaronmk
    srcs = []
1397
    if is_func: srcs = sql_gen.cols_srcs(in_cols)
1398
    return sql_gen.Col(out_pkey, into, srcs)
1399 2115 aaronmk
1400
##### Data cleanup
1401
1402 2290 aaronmk
def cleanup_table(db, table, cols):
1403 2677 aaronmk
    table = sql_gen.as_Table(table)
1404
    cols = map(sql_gen.as_Col, cols)
1405 2115 aaronmk
1406 2677 aaronmk
    expr = ('nullif(nullif(trim(both from %s), '+db.esc_value('')+'), '
1407
        +db.esc_value(r'\N')+')')
1408
    changes = [(v, sql_gen.CustomCode(expr % v.to_str(db)))
1409
        for v in cols]
1410 2115 aaronmk
1411 2677 aaronmk
    update(db, table, changes)