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 2923 aaronmk
    def __init__(self, db_config, autocommit=True, 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 2930 aaronmk
        try: return func()
358 2139 aaronmk
        except:
359 2443 aaronmk
            self.run_query('ROLLBACK TO SAVEPOINT '+savepoint, log_level=4)
360 2139 aaronmk
            raise
361 2930 aaronmk
        finally:
362
            # Always release savepoint, because after ROLLBACK TO SAVEPOINT,
363
            # "The savepoint remains valid and can be rolled back to again"
364
            # (http://www.postgresql.org/docs/8.3/static/sql-rollback-to.html).
365 2443 aaronmk
            self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
366 2930 aaronmk
367
            self._savepoint -= 1
368
            assert self._savepoint >= 0
369
370
            self.do_autocommit() # OK to do this after ROLLBACK TO SAVEPOINT
371 2191 aaronmk
372
    def do_autocommit(self):
373
        '''Autocommits if outside savepoint'''
374
        assert self._savepoint >= 0
375
        if self.autocommit and self._savepoint == 0:
376 2924 aaronmk
            self.log_debug('Autocommitting', level=4)
377 2191 aaronmk
            self.db.commit()
378 2643 aaronmk
379 2819 aaronmk
    def col_info(self, col):
380 2643 aaronmk
        table = sql_gen.Table('columns', 'information_schema')
381 2821 aaronmk
        cols = ['data_type', 'column_default',
382
            cast(self, 'boolean', 'is_nullable')]
383 2643 aaronmk
384
        conds = [('table_name', col.table.name), ('column_name', col.name)]
385
        schema = col.table.schema
386
        if schema != None: conds.append(('table_schema', schema))
387
388 2819 aaronmk
        type_, default, nullable = row(select(self, table, cols, conds,
389 2874 aaronmk
            order_by='table_schema', limit=1, log_level=4))
390 2643 aaronmk
            # TODO: order_by search_path schema order
391 2819 aaronmk
        default = sql_gen.as_Code(default, self)
392
393
        return sql_gen.TypedCol(col.name, type_, default, nullable)
394 2917 aaronmk
395
    def TempFunction(self, name):
396
        if self.debug_temp: schema = None
397
        else: schema = 'pg_temp'
398
        return sql_gen.Function(name, schema)
399 1849 aaronmk
400 1869 aaronmk
connect = DbConn
401
402 832 aaronmk
##### Recoverable querying
403 15 aaronmk
404 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
405 11 aaronmk
406 2791 aaronmk
def run_query(db, query, recover=None, cacheable=False, log_level=2,
407
    log_ignore_excs=None, **kw_args):
408 2794 aaronmk
    '''For params, see DbConn.run_query()'''
409 830 aaronmk
    if recover == None: recover = False
410 2464 aaronmk
    if log_ignore_excs == None: log_ignore_excs = ()
411
    log_ignore_excs = tuple(log_ignore_excs)
412 830 aaronmk
413 2666 aaronmk
    debug_msg_ref = None # usually, db.run_query() logs query before running it
414
    # But if filtering with log_ignore_excs, wait until after exception parsing
415
    if log_ignore_excs != () or not db.can_mogrify(): debug_msg_ref = [None]
416
417 2148 aaronmk
    try:
418 2464 aaronmk
        try:
419 2794 aaronmk
            def run(): return db.run_query(query, cacheable, log_level,
420 2793 aaronmk
                debug_msg_ref, **kw_args)
421 2796 aaronmk
            if recover and not db.is_cached(query):
422 2464 aaronmk
                return with_savepoint(db, run)
423
            else: return run() # don't need savepoint if cached
424
        except Exception, e:
425
            if not recover: raise # need savepoint to run index_cols()
426
            msg = exc.str_(e)
427
428
            match = re.search(r'duplicate key value violates unique constraint '
429 2493 aaronmk
                r'"((_?[^\W_]+)_.+?)"', msg)
430 2464 aaronmk
            if match:
431
                constraint, table = match.groups()
432
                try: cols = index_cols(db, table, constraint)
433
                except NotImplementedError: raise e
434
                else: raise DuplicateKeyException(constraint, cols, e)
435
436 2493 aaronmk
            match = re.search(r'null value in column "(.+?)" violates not-null'
437 2464 aaronmk
                r' constraint', msg)
438
            if match: raise NullValueException('NOT NULL', [match.group(1)], e)
439
440
            match = re.search(r'\b(?:invalid input (?:syntax|value)\b.*?'
441
                r'|date/time field value out of range): "(.+?)"\n'
442 2535 aaronmk
                r'(?:(?s).*?)\bfunction "(.+?)"', msg)
443 2464 aaronmk
            if match:
444
                value, name = match.groups()
445
                raise FunctionValueException(name, strings.to_unicode(value), e)
446
447 2526 aaronmk
            match = re.search(r'column "(.+?)" is of type (.+?) but expression '
448 2523 aaronmk
                r'is of type', msg)
449
            if match:
450
                col, type_ = match.groups()
451
                raise MissingCastException(type_, col, e)
452
453 2493 aaronmk
            match = re.search(r'relation "(.+?)" already exists', msg)
454 2464 aaronmk
            if match: raise DuplicateTableException(match.group(1), e)
455
456 2493 aaronmk
            match = re.search(r'function "(.+?)" already exists', msg)
457 2464 aaronmk
            if match: raise DuplicateFunctionException(match.group(1), e)
458
459
            raise # no specific exception raised
460
    except log_ignore_excs:
461
        log_level += 2
462
        raise
463
    finally:
464 2666 aaronmk
        if debug_msg_ref != None and debug_msg_ref[0] != None:
465
            db.log_debug(debug_msg_ref[0], log_level)
466 830 aaronmk
467 832 aaronmk
##### Basic queries
468
469 2153 aaronmk
def next_version(name):
470 2163 aaronmk
    version = 1 # first existing name was version 0
471 2586 aaronmk
    match = re.match(r'^(.*)#(\d+)$', name)
472 2153 aaronmk
    if match:
473 2586 aaronmk
        name, version = match.groups()
474
        version = int(version)+1
475 2932 aaronmk
    return sql_gen.concat(name, '#'+str(version))
476 2153 aaronmk
477 2899 aaronmk
def lock_table(db, table, mode):
478
    table = sql_gen.as_Table(table)
479
    run_query(db, 'LOCK TABLE '+table.to_str(db)+' IN '+mode+' MODE')
480
481 2789 aaronmk
def run_query_into(db, query, into=None, add_indexes_=False, **kw_args):
482 2085 aaronmk
    '''Outputs a query to a temp table.
483
    For params, see run_query().
484
    '''
485 2789 aaronmk
    if into == None: return run_query(db, query, **kw_args)
486 2790 aaronmk
487
    assert isinstance(into, sql_gen.Table)
488
489
    kw_args['recover'] = True
490
    kw_args.setdefault('log_ignore_excs', (DuplicateTableException,))
491
492 2916 aaronmk
    temp = not db.debug_temp # tables are permanent in debug_temp mode
493 2790 aaronmk
    # "temporary tables cannot specify a schema name", so remove schema
494
    if temp: into.schema = None
495
496
    # Create table
497
    while True:
498
        create_query = 'CREATE'
499
        if temp: create_query += ' TEMP'
500
        create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
501 2385 aaronmk
502 2790 aaronmk
        try:
503
            cur = run_query(db, create_query, **kw_args)
504
                # CREATE TABLE AS sets rowcount to # rows in query
505
            break
506
        except DuplicateTableException, e:
507
            into.name = next_version(into.name)
508
            # try again with next version of name
509
510
    if add_indexes_: add_indexes(db, into)
511
512
    return cur
513 2085 aaronmk
514 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
515
516 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
517
518 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
519 2293 aaronmk
    start=None, order_by=order_by_pkey, default_table=None):
520 1981 aaronmk
    '''
521 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
522 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
523 1981 aaronmk
    @param fields Use None to select all fields in the table
524 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
525 2379 aaronmk
        * container can be any iterable type
526 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
527
        * compare_right_side: sql_gen.ValueCond|literal value
528 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
529
        use all columns
530 2786 aaronmk
    @return query
531 1981 aaronmk
    '''
532 2315 aaronmk
    # Parse tables param
533 2121 aaronmk
    if not lists.is_seq(tables): tables = [tables]
534 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
535 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
536 2121 aaronmk
537 2315 aaronmk
    # Parse other params
538 2376 aaronmk
    if conds == None: conds = []
539 2650 aaronmk
    elif dicts.is_dict(conds): conds = conds.items()
540 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
541 135 aaronmk
    assert limit == None or type(limit) == int
542 865 aaronmk
    assert start == None or type(start) == int
543 2315 aaronmk
    if order_by is order_by_pkey:
544
        if distinct_on != []: order_by = None
545
        else: order_by = pkey(db, table0, recover=True)
546 865 aaronmk
547 2315 aaronmk
    query = 'SELECT'
548 2056 aaronmk
549 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
550 2056 aaronmk
551 2200 aaronmk
    # DISTINCT ON columns
552 2233 aaronmk
    if distinct_on != []:
553 2467 aaronmk
        query += '\nDISTINCT'
554 2254 aaronmk
        if distinct_on is not distinct_on_all:
555 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
556
557
    # Columns
558 2467 aaronmk
    query += '\n'
559 1135 aaronmk
    if fields == None: query += '*'
560 2765 aaronmk
    else:
561
        assert fields != []
562
        query += '\n, '.join(map(parse_col, fields))
563 2200 aaronmk
564
    # Main table
565 2467 aaronmk
    query += '\nFROM '+table0.to_str(db)
566 865 aaronmk
567 2122 aaronmk
    # Add joins
568 2271 aaronmk
    left_table = table0
569 2263 aaronmk
    for join_ in tables:
570
        table = join_.table
571 2238 aaronmk
572 2343 aaronmk
        # Parse special values
573
        if join_.type_ is sql_gen.filter_out: # filter no match
574 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
575 2853 aaronmk
                sql_gen.CompareCond(None, '~=')))
576 2343 aaronmk
577 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
578 2122 aaronmk
579
        left_table = table
580
581 865 aaronmk
    missing = True
582 2376 aaronmk
    if conds != []:
583 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
584
        else: whitespace = '\n'
585 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
586
            .to_str(db) for l, r in conds], 'WHERE')
587 865 aaronmk
        missing = False
588 2227 aaronmk
    if order_by != None:
589 2467 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
590
    if limit != None: query += '\nLIMIT '+str(limit); missing = False
591 865 aaronmk
    if start != None:
592 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
593 865 aaronmk
        missing = False
594
    if missing: warnings.warn(DbWarning(
595
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
596
597 2786 aaronmk
    return query
598 11 aaronmk
599 2054 aaronmk
def select(db, *args, **kw_args):
600
    '''For params, see mk_select() and run_query()'''
601
    recover = kw_args.pop('recover', None)
602
    cacheable = kw_args.pop('cacheable', True)
603 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
604 2054 aaronmk
605 2791 aaronmk
    return run_query(db, mk_select(db, *args, **kw_args), recover, cacheable,
606
        log_level=log_level)
607 2054 aaronmk
608 2788 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, returning=None,
609
    embeddable=False):
610 1960 aaronmk
    '''
611
    @param returning str|None An inserted column (such as pkey) to return
612 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
613 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
614
        query will be fully cached, not just if it raises an exception.
615 1960 aaronmk
    '''
616 2754 aaronmk
    table = sql_gen.remove_table_rename(sql_gen.as_Table(table))
617 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
618 2646 aaronmk
    if cols != None:
619
        cols = [sql_gen.to_name_only_col(v, table).to_str(db) for v in cols]
620 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
621 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
622 2063 aaronmk
623
    # Build query
624 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
625
    query = first_line
626 2467 aaronmk
    if cols != None: query += '\n('+', '.join(cols)+')'
627
    query += '\n'+select_query
628 2063 aaronmk
629
    if returning != None:
630 2743 aaronmk
        query += '\nRETURNING '+sql_gen.to_name_only_col(returning).to_str(db)
631 2063 aaronmk
632 2070 aaronmk
    if embeddable:
633 2327 aaronmk
        assert returning != None
634
635 2070 aaronmk
        # Create function
636 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
637 2327 aaronmk
        return_type = 'SETOF '+returning.to_str(db)+'%TYPE'
638 2189 aaronmk
        while True:
639
            try:
640 2918 aaronmk
                function = db.TempFunction(function_name)
641 2194 aaronmk
642 2189 aaronmk
                function_query = '''\
643 2698 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''()
644 2467 aaronmk
RETURNS '''+return_type+'''
645
LANGUAGE sql
646
AS $$
647 2788 aaronmk
'''+query+''';
648 2467 aaronmk
$$;
649 2070 aaronmk
'''
650 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
651 2464 aaronmk
                    log_ignore_excs=(DuplicateFunctionException,))
652 2189 aaronmk
                break # this version was successful
653
            except DuplicateFunctionException, e:
654
                function_name = next_version(function_name)
655
                # try again with next version of name
656 2070 aaronmk
657 2337 aaronmk
        # Return query that uses function
658 2698 aaronmk
        func_table = sql_gen.NamedTable('f', sql_gen.FunctionCall(function),
659 2742 aaronmk
            [returning]) # AS clause requires function alias
660 2787 aaronmk
        return mk_select(db, func_table, start=0, order_by=None)
661 2070 aaronmk
662 2787 aaronmk
    return query
663 2066 aaronmk
664
def insert_select(db, *args, **kw_args):
665 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
666 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
667
        values in
668 2072 aaronmk
    '''
669 2386 aaronmk
    into = kw_args.pop('into', None)
670
    if into != None: kw_args['embeddable'] = True
671 2066 aaronmk
    recover = kw_args.pop('recover', None)
672
    cacheable = kw_args.pop('cacheable', True)
673 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
674 2066 aaronmk
675 2789 aaronmk
    return run_query_into(db, mk_insert_select(db, *args, **kw_args), into,
676 2787 aaronmk
        recover=recover, cacheable=cacheable, log_level=log_level)
677 2063 aaronmk
678 2738 aaronmk
default = sql_gen.default # tells insert() to use the default value for a column
679 2066 aaronmk
680 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
681 2085 aaronmk
    '''For params, see insert_select()'''
682 1960 aaronmk
    if lists.is_seq(row): cols = None
683
    else:
684
        cols = row.keys()
685
        row = row.values()
686 2738 aaronmk
    row = list(row) # ensure that "== []" works
687 1960 aaronmk
688 2738 aaronmk
    if row == []: query = None
689
    else: query = sql_gen.Values(row).to_str(db)
690 1961 aaronmk
691 2788 aaronmk
    return insert_select(db, table, cols, query, *args, **kw_args)
692 11 aaronmk
693 2402 aaronmk
def mk_update(db, table, changes=None, cond=None):
694
    '''
695
    @param changes [(col, new_value),...]
696
        * container can be any iterable type
697
        * col: sql_gen.Code|str (for col name)
698
        * new_value: sql_gen.Code|literal value
699
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
700
    @return str query
701
    '''
702
    query = 'UPDATE '+sql_gen.as_Table(table).to_str(db)+'\nSET\n'
703 2405 aaronmk
    query += ',\n'.join((sql_gen.to_name_only_col(col, table).to_str(db)+' = '
704 2402 aaronmk
        +sql_gen.as_Value(new_value).to_str(db) for col, new_value in changes))
705 2467 aaronmk
    if cond != None: query += '\nWHERE\n'+cond.to_str(db)
706 2402 aaronmk
707
    return query
708
709
def update(db, *args, **kw_args):
710
    '''For params, see mk_update() and run_query()'''
711
    recover = kw_args.pop('recover', None)
712
713 2791 aaronmk
    return run_query(db, mk_update(db, *args, **kw_args), recover)
714 2402 aaronmk
715 135 aaronmk
def last_insert_id(db):
716 1849 aaronmk
    module = util.root_module(db.db)
717 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
718
    elif module == 'MySQLdb': return db.insert_id()
719
    else: return None
720 13 aaronmk
721 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
722 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
723 2415 aaronmk
    to names that will be distinct among the columns' tables.
724 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
725 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
726
    @param into The table for the new columns.
727 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
728
        columns will be included in the mapping even if they are not in cols.
729
        The tables of the provided Col objects will be changed to into, so make
730
        copies of them if you want to keep the original tables.
731
    @param as_items Whether to return a list of dict items instead of a dict
732 2383 aaronmk
    @return dict(orig_col=new_col, ...)
733
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
734 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
735
        * All mappings use the into table so its name can easily be
736 2383 aaronmk
          changed for all columns at once
737
    '''
738 2415 aaronmk
    cols = lists.uniqify(cols)
739
740 2394 aaronmk
    items = []
741 2389 aaronmk
    for col in preserve:
742 2390 aaronmk
        orig_col = copy.copy(col)
743 2392 aaronmk
        col.table = into
744 2394 aaronmk
        items.append((orig_col, col))
745
    preserve = set(preserve)
746
    for col in cols:
747 2716 aaronmk
        if col not in preserve:
748
            items.append((col, sql_gen.Col(str(col), into, col.srcs)))
749 2394 aaronmk
750
    if not as_items: items = dict(items)
751
    return items
752 2383 aaronmk
753 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
754 2391 aaronmk
    '''For params, see mk_flatten_mapping()
755
    @return See return value of mk_flatten_mapping()
756
    '''
757 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
758
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
759 2786 aaronmk
    run_query_into(db, mk_select(db, joins, cols, limit=limit, start=start),
760 2846 aaronmk
        into=into, add_indexes_=True)
761 2394 aaronmk
    return dict(items)
762 2391 aaronmk
763 2751 aaronmk
def mk_track_data_error(db, errors_table, cols, value, error_code, error):
764 2810 aaronmk
    assert cols != ()
765
766 2751 aaronmk
    cols = map(sql_gen.to_name_only_col, cols)
767
768
    columns_cols = ['column']
769
    columns = sql_gen.NamedValues('columns', columns_cols,
770
        [[c.name] for c in cols])
771
    values_cols = ['value', 'error_code', 'error']
772
    values = sql_gen.NamedValues('values', values_cols,
773
        [value, error_code, error])
774
775
    select_cols = columns_cols+values_cols
776
    name_only_cols = map(sql_gen.to_name_only_col, select_cols)
777 2755 aaronmk
    errors_table = sql_gen.NamedTable('errors', errors_table)
778 2751 aaronmk
    joins = [columns, sql_gen.Join(values, type_='CROSS'),
779
        sql_gen.Join(errors_table, dict(zip(name_only_cols, select_cols)),
780
        sql_gen.filter_out)]
781
782
    return mk_insert_select(db, errors_table, name_only_cols,
783 2787 aaronmk
        mk_select(db, joins, select_cols, order_by=None))
784 2751 aaronmk
785 2810 aaronmk
def track_data_error(db, errors_table, cols, *args, **kw_args):
786 2758 aaronmk
    '''
787
    @param errors_table If None, does nothing.
788
    '''
789 2810 aaronmk
    if errors_table == None or cols == (): return
790
    run_query(db, mk_track_data_error(db, errors_table, cols, *args, **kw_args),
791 2758 aaronmk
        cacheable=True, log_level=4)
792
793 2704 aaronmk
def cast(db, type_, col, errors_table=None):
794
    '''Casts an (unrenamed) column or value.
795 2769 aaronmk
    If errors_table set and col has srcs, saves errors in errors_table (using
796
    col's srcs attr as the source columns) and converts errors to warnings.
797 2820 aaronmk
    @param col str|sql_gen.Col|sql_gen.Literal
798 2704 aaronmk
    @param errors_table None|sql_gen.Table|str
799
    '''
800 2820 aaronmk
    col = sql_gen.as_Col(col)
801 2769 aaronmk
    save_errors = (errors_table != None and isinstance(col, sql_gen.Col)
802
        and col.srcs != ())
803
    if not save_errors: # can't save errors
804
        return sql_gen.CustomCode(col.to_str(db)+'::'+type_) # just cast
805 2704 aaronmk
806
    assert not isinstance(col, sql_gen.NamedCol)
807
808 2770 aaronmk
    errors_table = sql_gen.as_Table(errors_table)
809
    srcs = map(sql_gen.to_name_only_col, col.srcs)
810
    function_name = str(sql_gen.FunctionCall(type_, *srcs))
811 2918 aaronmk
    function = db.TempFunction(function_name)
812 2704 aaronmk
813 2718 aaronmk
    while True:
814
        # Create function definition
815
        query = '''\
816 2704 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''(value text)
817
RETURNS '''+type_+'''
818
LANGUAGE plpgsql
819
STRICT
820
AS $$
821
BEGIN
822
    /* The explicit cast to the return type is needed to make the cast happen
823
    inside the try block. (Implicit casts to the return type happen at the end
824
    of the function, outside any block.) */
825
    RETURN value::'''+type_+''';
826
EXCEPTION
827
    WHEN data_exception THEN
828
        -- Save error in errors table.
829 2751 aaronmk
        -- Insert the value and error for *each* source column.
830
'''+mk_track_data_error(db, errors_table, srcs,
831
    *map(sql_gen.CustomCode, ['value', 'SQLSTATE', 'SQLERRM']))+''';
832 2704 aaronmk
833
        RAISE WARNING '%', SQLERRM;
834
        RETURN NULL;
835
END;
836
$$;
837
'''
838 2718 aaronmk
839
        # Create function
840
        try:
841
            run_query(db, query, recover=True, cacheable=True,
842
                log_ignore_excs=(DuplicateFunctionException,))
843
            break # successful
844
        except DuplicateFunctionException:
845 2770 aaronmk
            function.name = next_version(function.name)
846
            # try again with next version of name
847 2704 aaronmk
848
    return sql_gen.FunctionCall(function, col)
849
850 2414 aaronmk
##### Database structure queries
851
852 2426 aaronmk
def table_row_count(db, table, recover=None):
853 2786 aaronmk
    return value(run_query(db, mk_select(db, table, [sql_gen.row_count],
854 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
855 2426 aaronmk
856 2414 aaronmk
def table_cols(db, table, recover=None):
857
    return list(col_names(select(db, table, limit=0, order_by=None,
858 2443 aaronmk
        recover=recover, log_level=4)))
859 2414 aaronmk
860 2291 aaronmk
def pkey(db, table, recover=None):
861 832 aaronmk
    '''Assumed to be first column in table'''
862 2339 aaronmk
    return table_cols(db, table, recover)[0]
863 832 aaronmk
864 2559 aaronmk
not_null_col = 'not_null_col'
865 2340 aaronmk
866
def table_not_null_col(db, table, recover=None):
867
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
868
    if not_null_col in table_cols(db, table, recover): return not_null_col
869
    else: return pkey(db, table, recover)
870
871 853 aaronmk
def index_cols(db, table, index):
872
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
873
    automatically created. When you don't know whether something is a UNIQUE
874
    constraint or a UNIQUE index, use this function.'''
875 1909 aaronmk
    module = util.root_module(db.db)
876
    if module == 'psycopg2':
877
        return list(values(run_query(db, '''\
878 853 aaronmk
SELECT attname
879 866 aaronmk
FROM
880
(
881
        SELECT attnum, attname
882
        FROM pg_index
883
        JOIN pg_class index ON index.oid = indexrelid
884
        JOIN pg_class table_ ON table_.oid = indrelid
885
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
886
        WHERE
887 2782 aaronmk
            table_.relname = '''+db.esc_value(table)+'''
888
            AND index.relname = '''+db.esc_value(index)+'''
889 866 aaronmk
    UNION
890
        SELECT attnum, attname
891
        FROM
892
        (
893
            SELECT
894
                indrelid
895
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
896
                    AS indkey
897
            FROM pg_index
898
            JOIN pg_class index ON index.oid = indexrelid
899
            JOIN pg_class table_ ON table_.oid = indrelid
900
            WHERE
901 2782 aaronmk
                table_.relname = '''+db.esc_value(table)+'''
902
                AND index.relname = '''+db.esc_value(index)+'''
903 866 aaronmk
        ) s
904
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
905
) s
906 853 aaronmk
ORDER BY attnum
907 2782 aaronmk
'''
908
            , cacheable=True, log_level=4)))
909 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
910
        ' database')
911 853 aaronmk
912 464 aaronmk
def constraint_cols(db, table, constraint):
913 1849 aaronmk
    module = util.root_module(db.db)
914 464 aaronmk
    if module == 'psycopg2':
915
        return list(values(run_query(db, '''\
916
SELECT attname
917
FROM pg_constraint
918
JOIN pg_class ON pg_class.oid = conrelid
919
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
920
WHERE
921 2783 aaronmk
    relname = '''+db.esc_value(table)+'''
922
    AND conname = '''+db.esc_value(constraint)+'''
923 464 aaronmk
ORDER BY attnum
924 2783 aaronmk
'''
925
            )))
926 464 aaronmk
    else: raise NotImplementedError("Can't list constraint columns for "+module+
927
        ' database')
928
929 2096 aaronmk
row_num_col = '_row_num'
930
931 2847 aaronmk
def add_index(db, exprs, table=None, unique=False, ensure_not_null=True):
932 2688 aaronmk
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
933 2538 aaronmk
    Currently, only function calls are supported as expressions.
934 2847 aaronmk
    @param ensure_not_null If set, translates NULL values to sentinel values.
935
        This allows indexes to be used for comparisons where NULLs are equal.
936 2538 aaronmk
    '''
937 2688 aaronmk
    if not lists.is_seq(exprs): exprs = [exprs]
938 2538 aaronmk
939 2688 aaronmk
    # Parse exprs
940
    old_exprs = exprs[:]
941
    exprs = []
942
    cols = []
943
    for i, expr in enumerate(old_exprs):
944
        expr = copy.deepcopy(expr) # don't modify input!
945 2823 aaronmk
        expr = sql_gen.as_Col(expr, table)
946 2688 aaronmk
947 2823 aaronmk
        # Handle nullable columns
948 2860 aaronmk
        if ensure_not_null:
949
            try: expr = sql_gen.ensure_not_null(db, expr)
950
            except KeyError: pass # unknown type, so just create plain index
951 2823 aaronmk
952 2688 aaronmk
        # Extract col
953
        if isinstance(expr, sql_gen.FunctionCall):
954
            col = expr.args[0]
955
            expr = sql_gen.Expr(expr)
956
        else: col = expr
957 2823 aaronmk
        assert isinstance(col, sql_gen.Col)
958 2688 aaronmk
959
        # Extract table
960
        if table == None:
961
            assert sql_gen.is_table_col(col)
962
            table = col.table
963
964
        col.table = None
965
966
        exprs.append(expr)
967
        cols.append(col)
968 2408 aaronmk
969 2688 aaronmk
    table = sql_gen.as_Table(table)
970
    index = sql_gen.Table(str(sql_gen.Col(','.join(map(str, cols)), table)))
971
972
    str_ = 'CREATE'
973
    if unique: str_ += ' UNIQUE'
974
    str_ += ' INDEX '+index.to_str(db)+' ON '+table.to_str(db)+' ('+(
975
        ', '.join((v.to_str(db) for v in exprs)))+')'
976
977
    try: run_query(db, str_, recover=True, cacheable=True, log_level=3)
978 2408 aaronmk
    except DuplicateTableException: pass # index already existed
979
980 2686 aaronmk
def add_pkey(db, table, cols=None, recover=None):
981
    '''Adds a primary key.
982
    @param cols [sql_gen.Col,...] The columns in the primary key.
983
        Defaults to the first column in the table.
984 2406 aaronmk
    @pre The table must not already have a primary key.
985
    '''
986
    table = sql_gen.as_Table(table)
987 2686 aaronmk
    if cols == None: cols = [pkey(db, table, recover)]
988
    col_strs = [sql_gen.to_name_only_col(v).to_str(db) for v in cols]
989 2406 aaronmk
990 2781 aaronmk
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD PRIMARY KEY ('
991
        +(', '.join(col_strs))+')', recover=True, cacheable=True, log_level=3,
992
        log_ignore_excs=(DuplicateTableException,))
993 2406 aaronmk
994 2772 aaronmk
already_indexed = object() # tells add_indexes() the pkey has already been added
995
996
def add_indexes(db, table, has_pkey=True):
997
    '''Adds an index on all columns in a table.
998
    @param has_pkey bool|already_indexed Whether a pkey instead of a regular
999
        index should be added on the first column.
1000
        * If already_indexed, the pkey is assumed to have already been added
1001
    '''
1002
    cols = table_cols(db, table)
1003
    if has_pkey:
1004
        if has_pkey is not already_indexed: add_pkey(db, table)
1005
        cols = cols[1:]
1006
    for col in cols: add_index(db, col, table)
1007
1008 2935 aaronmk
def add_col(db, table, col, **kw_args):
1009 2934 aaronmk
    assert isinstance(col, sql_gen.TypedCol)
1010 2935 aaronmk
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD COLUMN '+col.to_str(db),
1011
        **kw_args)
1012 2934 aaronmk
1013 2935 aaronmk
row_num_typed_col = sql_gen.TypedCol(row_num_col, 'serial', nullable=False,
1014
    constraints='PRIMARY KEY')
1015
1016 2086 aaronmk
def add_row_num(db, table):
1017 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
1018
    be the primary key.'''
1019 2935 aaronmk
    add_col(db, table, row_num_typed_col, log_level=3)
1020 2086 aaronmk
1021 2936 aaronmk
def cast_temp_col(db, type_, col, errors_table=None):
1022
    '''Like cast(), but creates a new column with the cast values if the input
1023
    is a column.
1024
    @return The new column or cast value
1025
    '''
1026
    def cast_(col): return cast(db, type_, col, errors_table)
1027
1028
    try: col = sql_gen.underlying_col(col)
1029
    except sql_gen.NoUnderlyingTableException: return sql_gen.wrap(cast_, col)
1030
1031
    table = col.table
1032
    new_col = sql_gen.Col(sql_gen.concat(col.name, '::'+type_), table, col.srcs)
1033
    expr = cast_(col)
1034
    add_col(db, table, sql_gen.TypedCol(new_col.name, type_))
1035
    update(db, table, [(new_col, expr)])
1036
1037
    return new_col
1038
1039 2889 aaronmk
def drop_table(db, table):
1040
    table = sql_gen.as_Table(table)
1041
    return run_query(db, 'DROP TABLE IF EXISTS '+table.to_str(db)+' CASCADE')
1042
1043 2760 aaronmk
def create_table(db, table, cols, has_pkey=True, col_indexes=True):
1044 2675 aaronmk
    '''Creates a table.
1045 2681 aaronmk
    @param cols [sql_gen.TypedCol,...] The column names and types
1046
    @param has_pkey If set, the first column becomes the primary key.
1047 2760 aaronmk
    @param col_indexes bool|[ref]
1048
        * If True, indexes will be added on all non-pkey columns.
1049
        * If a list reference, [0] will be set to a function to do this.
1050
          This can be used to delay index creation until the table is populated.
1051 2675 aaronmk
    '''
1052
    table = sql_gen.as_Table(table)
1053
1054 2681 aaronmk
    if has_pkey:
1055
        cols[0] = pkey = copy.copy(cols[0]) # don't modify input!
1056 2872 aaronmk
        pkey.constraints = 'PRIMARY KEY'
1057 2681 aaronmk
1058 2675 aaronmk
    str_ = 'CREATE TABLE '+table.to_str(db)+' (\n'
1059 2681 aaronmk
    str_ += '\n, '.join(v.to_str(db) for v in cols)
1060
    str_ += '\n);\n'
1061 2675 aaronmk
    run_query(db, str_, cacheable=True, log_level=2)
1062 2760 aaronmk
1063
    # Add indexes
1064 2773 aaronmk
    if has_pkey: has_pkey = already_indexed
1065
    def add_indexes_(): add_indexes(db, table, has_pkey)
1066
    if isinstance(col_indexes, list): col_indexes[0] = add_indexes_ # defer
1067
    elif col_indexes: add_indexes_() # add now
1068 2675 aaronmk
1069 2684 aaronmk
def vacuum(db, table):
1070
    table = sql_gen.as_Table(table)
1071
    db.with_autocommit(lambda: run_query(db, 'VACUUM ANALYZE '+table.to_str(db),
1072
        log_level=3))
1073
1074 2732 aaronmk
def truncate(db, table, schema='public'):
1075 2777 aaronmk
    table = sql_gen.as_Table(table, schema)
1076 2733 aaronmk
    return run_query(db, 'TRUNCATE '+table.to_str(db)+' CASCADE')
1077 2732 aaronmk
1078 2729 aaronmk
def tables(db, schema_like='public', table_like='%', exact=False):
1079
    if exact: compare = '='
1080
    else: compare = 'LIKE'
1081
1082 1849 aaronmk
    module = util.root_module(db.db)
1083 832 aaronmk
    if module == 'psycopg2':
1084 2779 aaronmk
        conds = [('schemaname', sql_gen.CompareCond(schema_like, compare)),
1085
            ('tablename', sql_gen.CompareCond(table_like, compare))]
1086
        return values(select(db, 'pg_tables', ['tablename'], conds,
1087
            order_by='tablename', log_level=4))
1088 1968 aaronmk
    elif module == 'MySQLdb':
1089 2779 aaronmk
        return values(run_query(db, 'SHOW TABLES LIKE '+db.esc_value(table_like)
1090
            , cacheable=True, log_level=4))
1091 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
1092 830 aaronmk
1093 2726 aaronmk
def table_exists(db, table):
1094
    table = sql_gen.as_Table(table)
1095 2730 aaronmk
    return list(tables(db, table.schema, table.name, exact=True)) != []
1096 2726 aaronmk
1097 2913 aaronmk
def function_exists(db, function):
1098
    function = sql_gen.as_Function(function)
1099
1100
    info_table = sql_gen.Table('routines', 'information_schema')
1101
    conds = [('routine_name', function.name)]
1102
    schema = function.schema
1103
    if schema != None: conds.append(('routine_schema', schema))
1104 2914 aaronmk
    # Exclude trigger functions, since they cannot be called directly
1105
    conds.append(('data_type', sql_gen.CompareCond('trigger', '!=')))
1106 2913 aaronmk
1107
    return list(values(select(db, info_table, ['routine_name'], conds,
1108
        order_by='routine_schema', limit=1, log_level=4))) != []
1109
        # TODO: order_by search_path schema order
1110
1111 2726 aaronmk
def errors_table(db, table, if_exists=True):
1112
    '''
1113
    @param if_exists If set, returns None if the errors table doesn't exist
1114
    @return None|sql_gen.Table
1115
    '''
1116
    table = sql_gen.as_Table(table)
1117
    if table.srcs != (): table = table.srcs[0]
1118
1119
    errors_table = sql_gen.suffixed_table(table, '.errors')
1120
    if if_exists and not table_exists(db, errors_table): return None
1121
    return errors_table
1122
1123 833 aaronmk
##### Database management
1124
1125 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
1126
    '''For kw_args, see tables()'''
1127
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
1128 833 aaronmk
1129 832 aaronmk
##### Heuristic queries
1130
1131 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
1132 1554 aaronmk
    '''Recovers from errors.
1133 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
1134
    '''
1135 2642 aaronmk
    row = sql_gen.ColDict(db, table, row)
1136 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
1137
1138 471 aaronmk
    try:
1139 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
1140 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
1141
            row_ct_ref[0] += cur.rowcount
1142
        return value(cur)
1143 471 aaronmk
    except DuplicateKeyException, e:
1144 2653 aaronmk
        row = sql_gen.ColDict(db, table,
1145
            util.dict_subset_right_join(row, e.cols))
1146
        return value(select(db, table, [pkey_], row, recover=True))
1147 471 aaronmk
1148 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
1149 830 aaronmk
    '''Recovers from errors'''
1150 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
1151 14 aaronmk
    except StopIteration:
1152 40 aaronmk
        if not create: raise
1153 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
1154 2078 aaronmk
1155 2593 aaronmk
def is_func_result(col):
1156
    return col.table.name.find('(') >= 0 and col.name == 'result'
1157
1158 2592 aaronmk
def into_table_name(out_table, in_tables0, mapping, is_func):
1159 2615 aaronmk
    def in_col_str(in_col):
1160
        in_col = sql_gen.remove_col_rename(in_col)
1161
        if isinstance(in_col, sql_gen.Col):
1162
            table = in_col.table
1163
            if table == in_tables0:
1164
                in_col = sql_gen.to_name_only_col(in_col)
1165
            elif is_func_result(in_col): in_col = table # omit col name
1166
        return str(in_col)
1167
1168 2580 aaronmk
    str_ = str(out_table)
1169
    if is_func:
1170 2615 aaronmk
        str_ += '('
1171 2580 aaronmk
1172 2615 aaronmk
        try: value_in_col = mapping['value']
1173
        except KeyError:
1174 2654 aaronmk
            str_ += ', '.join((str(k)+'='+in_col_str(v)
1175 2615 aaronmk
                for k, v in mapping.iteritems()))
1176
        else: str_ += in_col_str(value_in_col)
1177
1178
        str_ += ')'
1179 2812 aaronmk
    else:
1180
        out_col = 'rank'
1181
        try: in_col = mapping[out_col]
1182
        except KeyError: str_ += '_pkeys'
1183
        else: # has a rank column, so hierarchical
1184
            str_ += '['+str(out_col)+'='+in_col_str(in_col)+']'
1185 2580 aaronmk
    return str_
1186
1187 2508 aaronmk
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None, into=None,
1188 2805 aaronmk
    default=None, is_func=False, on_error=exc.raise_):
1189 2078 aaronmk
    '''Recovers from errors.
1190
    Only works under PostgreSQL (uses INSERT RETURNING).
1191 2909 aaronmk
    IMPORTANT: Must be run at the *beginning* of a transaction.
1192 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
1193
        tables to join with it using the main input table's pkey
1194 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
1195 2612 aaronmk
        * out_table_col: str (*not* sql_gen.Col)
1196
        * in_table_col: sql_gen.Col|literal-value
1197 2489 aaronmk
    @param into The table to contain the output and input pkeys.
1198 2574 aaronmk
        Defaults to `out_table.name+'_pkeys'`.
1199 2509 aaronmk
    @param default The *output* column to use as the pkey for missing rows.
1200
        If this output column does not exist in the mapping, uses None.
1201 2552 aaronmk
    @param is_func Whether out_table is the name of a SQL function, not a table
1202 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
1203 2078 aaronmk
    '''
1204 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
1205 2312 aaronmk
1206 2450 aaronmk
    def log_debug(msg): db.log_debug(msg, level=1.5)
1207 2505 aaronmk
    def col_ustr(str_):
1208 2567 aaronmk
        return strings.repr_no_u(sql_gen.remove_col_rename(str_))
1209 2450 aaronmk
1210 2910 aaronmk
    log_debug('********** New iteration **********')
1211
    log_debug('Inserting these input columns into '+strings.as_tt(
1212
        out_table.to_str(db))+':\n'+strings.as_table(mapping, ustr=col_ustr))
1213
1214 2768 aaronmk
    out_pkey = pkey(db, out_table, recover=True)
1215
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
1216
1217
    if mapping == {}: # need at least one column for INSERT SELECT
1218
        mapping = {out_pkey: None} # ColDict will replace with default value
1219
1220 2382 aaronmk
    # Create input joins from list of input tables
1221
    in_tables_ = in_tables[:] # don't modify input!
1222
    in_tables0 = in_tables_.pop(0) # first table is separate
1223 2735 aaronmk
    errors_table_ = errors_table(db, in_tables0)
1224 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
1225 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
1226 2460 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v,
1227
        {in_pkey: sql_gen.join_same_not_null}) for v in in_tables_]
1228 2131 aaronmk
1229 2592 aaronmk
    if into == None:
1230
        into = into_table_name(out_table, in_tables0, mapping, is_func)
1231
    into = sql_gen.as_Table(into)
1232
1233 2702 aaronmk
    # Set column sources
1234
    in_cols = filter(sql_gen.is_table_col, mapping.values())
1235
    for col in in_cols:
1236
        if col.table == in_tables0: col.set_srcs(sql_gen.src_self)
1237
1238 2486 aaronmk
    log_debug('Joining together input tables into temp table')
1239 2395 aaronmk
    # Place in new table for speed and so don't modify input if values edited
1240 2584 aaronmk
    in_table = sql_gen.Table('in')
1241 2702 aaronmk
    mapping = dicts.join(mapping, flatten(db, in_table, input_joins, in_cols,
1242
        preserve=[in_pkey_col], start=0))
1243 2395 aaronmk
    input_joins = [in_table]
1244 2486 aaronmk
    db.log_debug('Temp table: '+strings.as_tt(in_table.to_str(db)), level=2)
1245 2395 aaronmk
1246 2656 aaronmk
    mapping = sql_gen.ColDict(db, out_table, mapping)
1247
        # after applying dicts.join() because that returns a plain dict
1248
1249 2509 aaronmk
    # Resolve default value column
1250 2841 aaronmk
    if default != None:
1251
        try: default = mapping[default]
1252
        except KeyError:
1253 2509 aaronmk
            db.log_debug('Default value column '
1254
                +strings.as_tt(strings.repr_no_u(default))
1255 2511 aaronmk
                +' does not exist in mapping, falling back to None', level=2.1)
1256 2509 aaronmk
            default = None
1257
1258 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
1259 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
1260
1261 2201 aaronmk
    pkeys_table_exists_ref = [False]
1262 2420 aaronmk
    def insert_into_pkeys(joins, cols):
1263 2786 aaronmk
        query = mk_select(db, joins, cols, order_by=None, start=0)
1264 2201 aaronmk
        if pkeys_table_exists_ref[0]:
1265 2786 aaronmk
            insert_select(db, into, pkeys_names, query)
1266 2201 aaronmk
        else:
1267 2786 aaronmk
            run_query_into(db, query, into=into)
1268 2201 aaronmk
            pkeys_table_exists_ref[0] = True
1269
1270 2429 aaronmk
    limit_ref = [None]
1271 2380 aaronmk
    conds = set()
1272 2811 aaronmk
    distinct_on = sql_gen.ColDict(db, out_table)
1273 2325 aaronmk
    def mk_main_select(joins, cols):
1274 2811 aaronmk
        distinct_on_cols = [c.to_Col() for c in distinct_on.values()]
1275
        return mk_select(db, joins, cols, conds, distinct_on_cols,
1276 2429 aaronmk
            limit=limit_ref[0], start=0)
1277 2132 aaronmk
1278 2519 aaronmk
    exc_strs = set()
1279 2309 aaronmk
    def log_exc(e):
1280 2519 aaronmk
        e_str = exc.str_(e, first_line_only=True)
1281
        log_debug('Caught exception: '+e_str)
1282
        assert e_str not in exc_strs # avoid infinite loops
1283
        exc_strs.add(e_str)
1284 2735 aaronmk
1285 2451 aaronmk
    def remove_all_rows():
1286 2896 aaronmk
        log_debug('Ignoring all rows')
1287 2429 aaronmk
        limit_ref[0] = 0 # just create an empty pkeys table
1288 2735 aaronmk
1289 2758 aaronmk
    def ignore(in_col, value, e):
1290
        track_data_error(db, errors_table_, in_col.srcs, value, e.cause.pgcode,
1291
            e.cause.pgerror)
1292 2895 aaronmk
        log_debug('Ignoring rows with '+strings.as_tt(repr(in_col))+' = '
1293 2545 aaronmk
            +strings.as_tt(repr(value)))
1294 2895 aaronmk
1295 2758 aaronmk
    def remove_rows(in_col, value, e):
1296
        ignore(in_col, value, e)
1297 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
1298
        assert cond not in conds # avoid infinite loops
1299 2380 aaronmk
        conds.add(cond)
1300 2895 aaronmk
1301 2758 aaronmk
    def invalid2null(in_col, value, e):
1302
        ignore(in_col, value, e)
1303 2403 aaronmk
        update(db, in_table, [(in_col, None)],
1304
            sql_gen.ColValueCond(in_col, value))
1305 2245 aaronmk
1306 2589 aaronmk
    def insert_pkeys_table(which):
1307 2932 aaronmk
        return sql_gen.Table(sql_gen.concat(in_table.name,
1308 2589 aaronmk
            '_insert_'+which+'_pkeys'))
1309
    insert_out_pkeys = insert_pkeys_table('out')
1310
    insert_in_pkeys = insert_pkeys_table('in')
1311
1312 2206 aaronmk
    # Do inserts and selects
1313 2642 aaronmk
    join_cols = sql_gen.ColDict(db, out_table)
1314 2206 aaronmk
    while True:
1315 2521 aaronmk
        if limit_ref[0] == 0: # special case
1316
            log_debug('Creating an empty pkeys table')
1317 2786 aaronmk
            cur = run_query_into(db, mk_select(db, out_table, [out_pkey],
1318 2521 aaronmk
                limit=limit_ref[0]), into=insert_out_pkeys)
1319
            break # don't do main case
1320
1321 2303 aaronmk
        has_joins = join_cols != {}
1322
1323 2903 aaronmk
        log_debug('Trying to insert new rows')
1324
1325 2305 aaronmk
        # Prepare to insert new rows
1326 2325 aaronmk
        insert_joins = input_joins[:] # don't modify original!
1327 2403 aaronmk
        insert_args = dict(recover=True, cacheable=False)
1328 2303 aaronmk
        if has_joins:
1329 2325 aaronmk
            insert_joins.append(sql_gen.Join(out_table, join_cols,
1330
                sql_gen.filter_out))
1331
        else:
1332 2404 aaronmk
            insert_args.update(dict(returning=out_pkey, into=insert_out_pkeys))
1333 2786 aaronmk
        main_select = mk_main_select(insert_joins, mapping.values())
1334 2303 aaronmk
1335 2929 aaronmk
        def main_insert():
1336
            lock_table(db, out_table, 'EXCLUSIVE')
1337
            return insert_select(db, out_table, mapping.keys(), main_select,
1338
                **insert_args)
1339
1340 2206 aaronmk
        try:
1341 2929 aaronmk
            cur = with_savepoint(db, main_insert)
1342 2357 aaronmk
            break # insert successful
1343 2931 aaronmk
        except MissingCastException, e:
1344
            log_exc(e)
1345
1346
            out_col = e.col
1347
            type_ = e.type
1348
1349
            log_debug('Casting '+strings.as_tt(out_col)+' input to '
1350
                +strings.as_tt(type_))
1351
            def wrap_func(col): return cast(db, type_, col, errors_table_)
1352
            mapping[out_col] = sql_gen.wrap(wrap_func, mapping[out_col])
1353 2206 aaronmk
        except DuplicateKeyException, e:
1354 2309 aaronmk
            log_exc(e)
1355
1356 2258 aaronmk
            old_join_cols = join_cols.copy()
1357 2811 aaronmk
            distinct_on.update(util.dict_subset(mapping, e.cols))
1358 2565 aaronmk
            join_cols.update(util.dict_subset_right_join(mapping, e.cols))
1359 2486 aaronmk
            log_debug('Ignoring existing rows, comparing on these columns:\n'
1360 2505 aaronmk
                +strings.as_inline_table(join_cols, ustr=col_ustr))
1361 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
1362 2230 aaronmk
        except NullValueException, e:
1363 2309 aaronmk
            log_exc(e)
1364
1365 2230 aaronmk
            out_col, = e.cols
1366
            try: in_col = mapping[out_col]
1367 2356 aaronmk
            except KeyError:
1368 2486 aaronmk
                log_debug('Missing mapping for NOT NULL column '+out_col)
1369 2451 aaronmk
                remove_all_rows()
1370 2758 aaronmk
            else: remove_rows(in_col, None, e)
1371 2542 aaronmk
        except FunctionValueException, e:
1372
            log_exc(e)
1373
1374
            func_name = e.name
1375
            value = e.value
1376
            for out_col, in_col in mapping.iteritems():
1377 2758 aaronmk
                in_col = sql_gen.unwrap_func_call(in_col, func_name)
1378
                invalid2null(in_col, value, e)
1379 2429 aaronmk
        except DatabaseErrors, e:
1380
            log_exc(e)
1381
1382 2808 aaronmk
            log_debug('No handler for exception')
1383
            on_error(e)
1384 2451 aaronmk
            remove_all_rows()
1385 2358 aaronmk
        # after exception handled, rerun loop with additional constraints
1386 2132 aaronmk
1387 2357 aaronmk
    if row_ct_ref != None and cur.rowcount >= 0:
1388
        row_ct_ref[0] += cur.rowcount
1389
1390
    if has_joins:
1391
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
1392 2486 aaronmk
        log_debug('Getting output table pkeys of existing/inserted rows')
1393 2420 aaronmk
        insert_into_pkeys(select_joins, pkeys_cols)
1394 2357 aaronmk
    else:
1395 2404 aaronmk
        add_row_num(db, insert_out_pkeys) # for joining with input pkeys
1396 2357 aaronmk
1397 2486 aaronmk
        log_debug('Getting input table pkeys of inserted rows')
1398 2786 aaronmk
        run_query_into(db, mk_main_select(input_joins, [in_pkey]),
1399 2404 aaronmk
            into=insert_in_pkeys)
1400
        add_row_num(db, insert_in_pkeys) # for joining with output pkeys
1401 2357 aaronmk
1402 2428 aaronmk
        assert table_row_count(db, insert_out_pkeys) == table_row_count(db,
1403
            insert_in_pkeys)
1404
1405 2486 aaronmk
        log_debug('Combining output and input pkeys in inserted order')
1406 2404 aaronmk
        pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
1407 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
1408 2420 aaronmk
        insert_into_pkeys(pkey_joins, pkeys_names)
1409 2357 aaronmk
1410 2486 aaronmk
    db.log_debug('Adding pkey on pkeys table to enable fast joins', level=2.5)
1411 2594 aaronmk
    add_pkey(db, into)
1412 2407 aaronmk
1413 2508 aaronmk
    log_debug('Setting pkeys of missing rows to '+strings.as_tt(repr(default)))
1414 2489 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(into,
1415 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
1416
        # must use join_same_not_null or query will take forever
1417 2420 aaronmk
    insert_into_pkeys(missing_rows_joins,
1418 2508 aaronmk
        [in_pkey_col, sql_gen.NamedCol(out_pkey, default)])
1419 2357 aaronmk
1420 2489 aaronmk
    assert table_row_count(db, into) == table_row_count(db, in_table)
1421 2428 aaronmk
1422 2702 aaronmk
    srcs = []
1423
    if is_func: srcs = sql_gen.cols_srcs(in_cols)
1424
    return sql_gen.Col(out_pkey, into, srcs)
1425 2115 aaronmk
1426
##### Data cleanup
1427
1428 2290 aaronmk
def cleanup_table(db, table, cols):
1429 2677 aaronmk
    table = sql_gen.as_Table(table)
1430
    cols = map(sql_gen.as_Col, cols)
1431 2115 aaronmk
1432 2677 aaronmk
    expr = ('nullif(nullif(trim(both from %s), '+db.esc_value('')+'), '
1433
        +db.esc_value(r'\N')+')')
1434
    changes = [(v, sql_gen.CustomCode(expr % v.to_str(db)))
1435
        for v in cols]
1436 2115 aaronmk
1437 2677 aaronmk
    update(db, table, changes)