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 2945 aaronmk
class ExceptionWithNameType(DbException):
50
    def __init__(self, type_, name, cause=None):
51
        DbException.__init__(self, 'for type: '+strings.as_tt(str(type_))
52
            +'; name: '+strings.as_tt(name), cause)
53
        self.type = type_
54
        self.name = name
55
56 2306 aaronmk
class ConstraintException(DbException):
57
    def __init__(self, name, cols, cause=None):
58 2484 aaronmk
        DbException.__init__(self, 'Violated '+strings.as_tt(name)
59
            +' constraint on columns: '+strings.as_tt(', '.join(cols)), cause)
60 2306 aaronmk
        self.name = name
61 468 aaronmk
        self.cols = cols
62 11 aaronmk
63 2523 aaronmk
class MissingCastException(DbException):
64
    def __init__(self, type_, col, cause=None):
65
        DbException.__init__(self, 'Missing cast to type '+strings.as_tt(type_)
66
            +' on column: '+strings.as_tt(col), cause)
67
        self.type = type_
68
        self.col = col
69
70 2143 aaronmk
class NameException(DbException): pass
71
72 2306 aaronmk
class DuplicateKeyException(ConstraintException): pass
73 13 aaronmk
74 2306 aaronmk
class NullValueException(ConstraintException): pass
75 13 aaronmk
76 2240 aaronmk
class FunctionValueException(ExceptionWithNameValue): pass
77 2239 aaronmk
78 2945 aaronmk
class DuplicateException(ExceptionWithNameType): pass
79 2143 aaronmk
80 89 aaronmk
class EmptyRowException(DbException): pass
81
82 865 aaronmk
##### Warnings
83
84
class DbWarning(UserWarning): pass
85
86 1930 aaronmk
##### Result retrieval
87
88
def col_names(cur): return (col[0] for col in cur.description)
89
90
def rows(cur): return iter(lambda: cur.fetchone(), None)
91
92
def consume_rows(cur):
93
    '''Used to fetch all rows so result will be cached'''
94
    iters.consume_iter(rows(cur))
95
96
def next_row(cur): return rows(cur).next()
97
98
def row(cur):
99
    row_ = next_row(cur)
100
    consume_rows(cur)
101
    return row_
102
103
def next_value(cur): return next_row(cur)[0]
104
105
def value(cur): return row(cur)[0]
106
107
def values(cur): return iters.func_iter(lambda: next_value(cur))
108
109
def value_or_none(cur):
110
    try: return value(cur)
111
    except StopIteration: return None
112
113 2762 aaronmk
##### Escaping
114 2101 aaronmk
115 2573 aaronmk
def esc_name_by_module(module, name):
116
    if module == 'psycopg2' or module == None: quote = '"'
117 2101 aaronmk
    elif module == 'MySQLdb': quote = '`'
118
    else: raise NotImplementedError("Can't escape name for "+module+' database')
119 2500 aaronmk
    return sql_gen.esc_name(name, quote)
120 2101 aaronmk
121
def esc_name_by_engine(engine, name, **kw_args):
122
    return esc_name_by_module(db_engines[engine][0], name, **kw_args)
123
124
def esc_name(db, name, **kw_args):
125
    return esc_name_by_module(util.root_module(db.db), name, **kw_args)
126
127
def qual_name(db, schema, table):
128
    def esc_name_(name): return esc_name(db, name)
129
    table = esc_name_(table)
130
    if schema != None: return esc_name_(schema)+'.'+table
131
    else: return table
132
133 1869 aaronmk
##### Database connections
134 1849 aaronmk
135 2097 aaronmk
db_config_names = ['engine', 'host', 'user', 'password', 'database', 'schemas']
136 1926 aaronmk
137 1869 aaronmk
db_engines = {
138
    'MySQL': ('MySQLdb', {'password': 'passwd', 'database': 'db'}),
139
    'PostgreSQL': ('psycopg2', {}),
140
}
141
142
DatabaseErrors_set = set([DbException])
143
DatabaseErrors = tuple(DatabaseErrors_set)
144
145
def _add_module(module):
146
    DatabaseErrors_set.add(module.DatabaseError)
147
    global DatabaseErrors
148
    DatabaseErrors = tuple(DatabaseErrors_set)
149
150
def db_config_str(db_config):
151
    return db_config['engine']+' database '+db_config['database']
152
153 2448 aaronmk
log_debug_none = lambda msg, level=2: None
154 1901 aaronmk
155 1849 aaronmk
class DbConn:
156 2923 aaronmk
    def __init__(self, db_config, autocommit=True, caching=True,
157 2915 aaronmk
        log_debug=log_debug_none, debug_temp=False):
158
        '''
159
        @param debug_temp Whether temporary objects should instead be permanent.
160
            This assists in debugging the internal objects used by the program.
161
        '''
162 1869 aaronmk
        self.db_config = db_config
163 2190 aaronmk
        self.autocommit = autocommit
164
        self.caching = caching
165 1901 aaronmk
        self.log_debug = log_debug
166 2193 aaronmk
        self.debug = log_debug != log_debug_none
167 2915 aaronmk
        self.debug_temp = debug_temp
168 3074 aaronmk
        self.autoanalyze = False
169 1869 aaronmk
170
        self.__db = None
171 1889 aaronmk
        self.query_results = {}
172 2139 aaronmk
        self._savepoint = 0
173 2671 aaronmk
        self._notices_seen = set()
174 1869 aaronmk
175
    def __getattr__(self, name):
176
        if name == '__dict__': raise Exception('getting __dict__')
177
        if name == 'db': return self._db()
178
        else: raise AttributeError()
179
180
    def __getstate__(self):
181
        state = copy.copy(self.__dict__) # shallow copy
182 1915 aaronmk
        state['log_debug'] = None # don't pickle the debug callback
183 1869 aaronmk
        state['_DbConn__db'] = None # don't pickle the connection
184
        return state
185
186 2165 aaronmk
    def connected(self): return self.__db != None
187
188 1869 aaronmk
    def _db(self):
189
        if self.__db == None:
190
            # Process db_config
191
            db_config = self.db_config.copy() # don't modify input!
192 2097 aaronmk
            schemas = db_config.pop('schemas', None)
193 1869 aaronmk
            module_name, mappings = db_engines[db_config.pop('engine')]
194
            module = __import__(module_name)
195
            _add_module(module)
196
            for orig, new in mappings.iteritems():
197
                try: util.rename_key(db_config, orig, new)
198
                except KeyError: pass
199
200
            # Connect
201
            self.__db = module.connect(**db_config)
202
203
            # Configure connection
204 2906 aaronmk
            if hasattr(self.db, 'set_isolation_level'):
205
                import psycopg2.extensions
206
                self.db.set_isolation_level(
207
                    psycopg2.extensions.ISOLATION_LEVEL_READ_COMMITTED)
208 2101 aaronmk
            if schemas != None:
209 2893 aaronmk
                search_path = [self.esc_name(s) for s in schemas.split(',')]
210
                search_path.append(value(run_query(self, 'SHOW search_path',
211
                    log_level=4)))
212
                run_query(self, 'SET search_path TO '+(','.join(search_path)),
213
                    log_level=3)
214 1869 aaronmk
215
        return self.__db
216 1889 aaronmk
217 1891 aaronmk
    class DbCursor(Proxy):
218 1927 aaronmk
        def __init__(self, outer):
219 1891 aaronmk
            Proxy.__init__(self, outer.db.cursor())
220 2191 aaronmk
            self.outer = outer
221 1927 aaronmk
            self.query_results = outer.query_results
222 1894 aaronmk
            self.query_lookup = None
223 1891 aaronmk
            self.result = []
224 1889 aaronmk
225 2802 aaronmk
        def execute(self, query):
226 2764 aaronmk
            self._is_insert = query.startswith('INSERT')
227 2797 aaronmk
            self.query_lookup = query
228 2148 aaronmk
            try:
229 2191 aaronmk
                try:
230 2802 aaronmk
                    cur = self.inner.execute(query)
231 2191 aaronmk
                    self.outer.do_autocommit()
232 2802 aaronmk
                finally: self.query = get_cur_query(self.inner, query)
233 1904 aaronmk
            except Exception, e:
234 2802 aaronmk
                _add_cursor_info(e, self, query)
235 1904 aaronmk
                self.result = e # cache the exception as the result
236
                self._cache_result()
237
                raise
238 3004 aaronmk
239
            # Always cache certain queries
240
            if query.startswith('CREATE') or query.startswith('ALTER'):
241 3007 aaronmk
                # structural changes
242 3040 aaronmk
                # Rest of query must be unique in the face of name collisions,
243
                # so don't cache ADD COLUMN unless it has distinguishing comment
244
                if query.find('ADD COLUMN') < 0 or query.endswith('*/'):
245 3007 aaronmk
                    self._cache_result()
246 3004 aaronmk
            elif self.rowcount == 0 and query.startswith('SELECT'): # empty
247 2800 aaronmk
                consume_rows(self) # fetch all rows so result will be cached
248 3004 aaronmk
249 2762 aaronmk
            return cur
250 1894 aaronmk
251 1891 aaronmk
        def fetchone(self):
252
            row = self.inner.fetchone()
253 1899 aaronmk
            if row != None: self.result.append(row)
254
            # otherwise, fetched all rows
255 1904 aaronmk
            else: self._cache_result()
256
            return row
257
258
        def _cache_result(self):
259 2948 aaronmk
            # For inserts that return a result set, don't cache result set since
260
            # inserts are not idempotent. Other non-SELECT queries don't have
261
            # their result set read, so only exceptions will be cached (an
262
            # invalid query will always be invalid).
263 1930 aaronmk
            if self.query_results != None and (not self._is_insert
264 1906 aaronmk
                or isinstance(self.result, Exception)):
265
266 1894 aaronmk
                assert self.query_lookup != None
267 1916 aaronmk
                self.query_results[self.query_lookup] = self.CacheCursor(
268
                    util.dict_subset(dicts.AttrsDictView(self),
269
                    ['query', 'result', 'rowcount', 'description']))
270 1906 aaronmk
271 1916 aaronmk
        class CacheCursor:
272
            def __init__(self, cached_result): self.__dict__ = cached_result
273
274 1927 aaronmk
            def execute(self, *args, **kw_args):
275 1916 aaronmk
                if isinstance(self.result, Exception): raise self.result
276
                # otherwise, result is a rows list
277
                self.iter = iter(self.result)
278
279
            def fetchone(self):
280
                try: return self.iter.next()
281
                except StopIteration: return None
282 1891 aaronmk
283 2212 aaronmk
    def esc_value(self, value):
284 2663 aaronmk
        try: str_ = self.mogrify('%s', [value])
285
        except NotImplementedError, e:
286
            module = util.root_module(self.db)
287
            if module == 'MySQLdb':
288
                import _mysql
289
                str_ = _mysql.escape_string(value)
290
            else: raise e
291 2374 aaronmk
        return strings.to_unicode(str_)
292 2212 aaronmk
293 2347 aaronmk
    def esc_name(self, name): return esc_name(self, name) # calls global func
294
295 2814 aaronmk
    def std_code(self, str_):
296
        '''Standardizes SQL code.
297
        * Ensures that string literals are prefixed by `E`
298
        '''
299
        if str_.startswith("'"): str_ = 'E'+str_
300
        return str_
301
302 2665 aaronmk
    def can_mogrify(self):
303 2663 aaronmk
        module = util.root_module(self.db)
304 2665 aaronmk
        return module == 'psycopg2'
305 2663 aaronmk
306 2665 aaronmk
    def mogrify(self, query, params=None):
307
        if self.can_mogrify(): return self.db.cursor().mogrify(query, params)
308
        else: raise NotImplementedError("Can't mogrify query")
309
310 2671 aaronmk
    def print_notices(self):
311 2725 aaronmk
        if hasattr(self.db, 'notices'):
312
            for msg in self.db.notices:
313
                if msg not in self._notices_seen:
314
                    self._notices_seen.add(msg)
315
                    self.log_debug(msg, level=2)
316 2671 aaronmk
317 2793 aaronmk
    def run_query(self, query, cacheable=False, log_level=2,
318 2464 aaronmk
        debug_msg_ref=None):
319 2445 aaronmk
        '''
320 2464 aaronmk
        @param log_ignore_excs The log_level will be increased by 2 if the query
321
            throws one of these exceptions.
322 2664 aaronmk
        @param debug_msg_ref If specified, the log message will be returned in
323
            this instead of being output. This allows you to filter log messages
324
            depending on the result of the query.
325 2445 aaronmk
        '''
326 2167 aaronmk
        assert query != None
327
328 2047 aaronmk
        if not self.caching: cacheable = False
329 1903 aaronmk
        used_cache = False
330 2664 aaronmk
331
        def log_msg(query):
332
            if used_cache: cache_status = 'cache hit'
333
            elif cacheable: cache_status = 'cache miss'
334
            else: cache_status = 'non-cacheable'
335
            return 'DB query: '+cache_status+':\n'+strings.as_code(query, 'SQL')
336
337 1903 aaronmk
        try:
338 1927 aaronmk
            # Get cursor
339
            if cacheable:
340
                try:
341 2797 aaronmk
                    cur = self.query_results[query]
342 1927 aaronmk
                    used_cache = True
343
                except KeyError: cur = self.DbCursor(self)
344
            else: cur = self.db.cursor()
345
346 2664 aaronmk
            # Log query
347
            if self.debug and debug_msg_ref == None: # log before running
348
                self.log_debug(log_msg(query), log_level)
349
350 1927 aaronmk
            # Run query
351 2793 aaronmk
            cur.execute(query)
352 1903 aaronmk
        finally:
353 2671 aaronmk
            self.print_notices()
354 2664 aaronmk
            if self.debug and debug_msg_ref != None: # return after running
355 2793 aaronmk
                debug_msg_ref[0] = log_msg(str(get_cur_query(cur, query)))
356 1903 aaronmk
357
        return cur
358 1914 aaronmk
359 2797 aaronmk
    def is_cached(self, query): return query in self.query_results
360 2139 aaronmk
361 2907 aaronmk
    def with_autocommit(self, func):
362 2801 aaronmk
        import psycopg2.extensions
363
364
        prev_isolation_level = self.db.isolation_level
365 2907 aaronmk
        self.db.set_isolation_level(
366
            psycopg2.extensions.ISOLATION_LEVEL_AUTOCOMMIT)
367 2683 aaronmk
        try: return func()
368 2801 aaronmk
        finally: self.db.set_isolation_level(prev_isolation_level)
369 2683 aaronmk
370 2139 aaronmk
    def with_savepoint(self, func):
371 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
372 2443 aaronmk
        self.run_query('SAVEPOINT '+savepoint, log_level=4)
373 2139 aaronmk
        self._savepoint += 1
374 2930 aaronmk
        try: return func()
375 2139 aaronmk
        except:
376 2443 aaronmk
            self.run_query('ROLLBACK TO SAVEPOINT '+savepoint, log_level=4)
377 2139 aaronmk
            raise
378 2930 aaronmk
        finally:
379
            # Always release savepoint, because after ROLLBACK TO SAVEPOINT,
380
            # "The savepoint remains valid and can be rolled back to again"
381
            # (http://www.postgresql.org/docs/8.3/static/sql-rollback-to.html).
382 2443 aaronmk
            self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
383 2930 aaronmk
384
            self._savepoint -= 1
385
            assert self._savepoint >= 0
386
387
            self.do_autocommit() # OK to do this after ROLLBACK TO SAVEPOINT
388 2191 aaronmk
389
    def do_autocommit(self):
390
        '''Autocommits if outside savepoint'''
391
        assert self._savepoint >= 0
392
        if self.autocommit and self._savepoint == 0:
393 2924 aaronmk
            self.log_debug('Autocommitting', level=4)
394 2191 aaronmk
            self.db.commit()
395 2643 aaronmk
396 2819 aaronmk
    def col_info(self, col):
397 2643 aaronmk
        table = sql_gen.Table('columns', 'information_schema')
398 3063 aaronmk
        type_ = sql_gen.Coalesce(sql_gen.Nullif(sql_gen.Col('data_type'),
399
            'USER-DEFINED'), sql_gen.Col('udt_name'))
400
        cols = [type_, 'column_default', cast(self, 'boolean', 'is_nullable')]
401 2643 aaronmk
402
        conds = [('table_name', col.table.name), ('column_name', col.name)]
403
        schema = col.table.schema
404
        if schema != None: conds.append(('table_schema', schema))
405
406 2819 aaronmk
        type_, default, nullable = row(select(self, table, cols, conds,
407 3059 aaronmk
            order_by='table_schema', limit=1, cacheable=False, log_level=4))
408 2643 aaronmk
            # TODO: order_by search_path schema order
409 2819 aaronmk
        default = sql_gen.as_Code(default, self)
410
411
        return sql_gen.TypedCol(col.name, type_, default, nullable)
412 2917 aaronmk
413
    def TempFunction(self, name):
414
        if self.debug_temp: schema = None
415
        else: schema = 'pg_temp'
416
        return sql_gen.Function(name, schema)
417 1849 aaronmk
418 1869 aaronmk
connect = DbConn
419
420 832 aaronmk
##### Recoverable querying
421 15 aaronmk
422 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
423 11 aaronmk
424 2791 aaronmk
def run_query(db, query, recover=None, cacheable=False, log_level=2,
425
    log_ignore_excs=None, **kw_args):
426 2794 aaronmk
    '''For params, see DbConn.run_query()'''
427 830 aaronmk
    if recover == None: recover = False
428 2464 aaronmk
    if log_ignore_excs == None: log_ignore_excs = ()
429
    log_ignore_excs = tuple(log_ignore_excs)
430 830 aaronmk
431 2666 aaronmk
    debug_msg_ref = None # usually, db.run_query() logs query before running it
432
    # But if filtering with log_ignore_excs, wait until after exception parsing
433 2984 aaronmk
    if log_ignore_excs != () or not db.can_mogrify(): debug_msg_ref = [None]
434 2666 aaronmk
435 2148 aaronmk
    try:
436 2464 aaronmk
        try:
437 2794 aaronmk
            def run(): return db.run_query(query, cacheable, log_level,
438 2793 aaronmk
                debug_msg_ref, **kw_args)
439 2796 aaronmk
            if recover and not db.is_cached(query):
440 2464 aaronmk
                return with_savepoint(db, run)
441
            else: return run() # don't need savepoint if cached
442
        except Exception, e:
443
            msg = exc.str_(e)
444
445
            match = re.search(r'duplicate key value violates unique constraint '
446 2493 aaronmk
                r'"((_?[^\W_]+)_.+?)"', msg)
447 2464 aaronmk
            if match:
448
                constraint, table = match.groups()
449 3025 aaronmk
                cols = []
450
                if recover: # need auto-rollback to run index_cols()
451
                    try: cols = index_cols(db, table, constraint)
452
                    except NotImplementedError: pass
453
                raise DuplicateKeyException(constraint, cols, e)
454 2464 aaronmk
455 2493 aaronmk
            match = re.search(r'null value in column "(.+?)" violates not-null'
456 2464 aaronmk
                r' constraint', msg)
457
            if match: raise NullValueException('NOT NULL', [match.group(1)], e)
458
459
            match = re.search(r'\b(?:invalid input (?:syntax|value)\b.*?'
460
                r'|date/time field value out of range): "(.+?)"\n'
461 2535 aaronmk
                r'(?:(?s).*?)\bfunction "(.+?)"', msg)
462 2464 aaronmk
            if match:
463
                value, name = match.groups()
464
                raise FunctionValueException(name, strings.to_unicode(value), e)
465
466 2526 aaronmk
            match = re.search(r'column "(.+?)" is of type (.+?) but expression '
467 2523 aaronmk
                r'is of type', msg)
468
            if match:
469
                col, type_ = match.groups()
470
                raise MissingCastException(type_, col, e)
471
472 2946 aaronmk
            match = re.search(r'\b(\S+) "(.+?)".*? already exists', msg)
473 2945 aaronmk
            if match:
474
                type_, name = match.groups()
475
                raise DuplicateException(type_, name, e)
476 2464 aaronmk
477
            raise # no specific exception raised
478
    except log_ignore_excs:
479
        log_level += 2
480
        raise
481
    finally:
482 2666 aaronmk
        if debug_msg_ref != None and debug_msg_ref[0] != None:
483
            db.log_debug(debug_msg_ref[0], log_level)
484 830 aaronmk
485 832 aaronmk
##### Basic queries
486
487 2153 aaronmk
def next_version(name):
488 2163 aaronmk
    version = 1 # first existing name was version 0
489 2586 aaronmk
    match = re.match(r'^(.*)#(\d+)$', name)
490 2153 aaronmk
    if match:
491 2586 aaronmk
        name, version = match.groups()
492
        version = int(version)+1
493 2932 aaronmk
    return sql_gen.concat(name, '#'+str(version))
494 2153 aaronmk
495 2899 aaronmk
def lock_table(db, table, mode):
496
    table = sql_gen.as_Table(table)
497
    run_query(db, 'LOCK TABLE '+table.to_str(db)+' IN '+mode+' MODE')
498
499 2789 aaronmk
def run_query_into(db, query, into=None, add_indexes_=False, **kw_args):
500 2085 aaronmk
    '''Outputs a query to a temp table.
501
    For params, see run_query().
502
    '''
503 2789 aaronmk
    if into == None: return run_query(db, query, **kw_args)
504 2790 aaronmk
505
    assert isinstance(into, sql_gen.Table)
506
507 2992 aaronmk
    into.is_temp = True
508 3008 aaronmk
    # "temporary tables cannot specify a schema name", so remove schema
509
    into.schema = None
510 2992 aaronmk
511 2790 aaronmk
    kw_args['recover'] = True
512 2945 aaronmk
    kw_args.setdefault('log_ignore_excs', (DuplicateException,))
513 2790 aaronmk
514 2916 aaronmk
    temp = not db.debug_temp # tables are permanent in debug_temp mode
515 2790 aaronmk
516
    # Create table
517
    while True:
518
        create_query = 'CREATE'
519
        if temp: create_query += ' TEMP'
520
        create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
521 2385 aaronmk
522 2790 aaronmk
        try:
523
            cur = run_query(db, create_query, **kw_args)
524
                # CREATE TABLE AS sets rowcount to # rows in query
525
            break
526 2945 aaronmk
        except DuplicateException, e:
527 2790 aaronmk
            into.name = next_version(into.name)
528
            # try again with next version of name
529
530
    if add_indexes_: add_indexes(db, into)
531 3073 aaronmk
    analyze(db, into)
532 2790 aaronmk
533
    return cur
534 2085 aaronmk
535 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
536
537 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
538
539 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
540 2293 aaronmk
    start=None, order_by=order_by_pkey, default_table=None):
541 1981 aaronmk
    '''
542 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
543 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
544 1981 aaronmk
    @param fields Use None to select all fields in the table
545 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
546 2379 aaronmk
        * container can be any iterable type
547 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
548
        * compare_right_side: sql_gen.ValueCond|literal value
549 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
550
        use all columns
551 2786 aaronmk
    @return query
552 1981 aaronmk
    '''
553 2315 aaronmk
    # Parse tables param
554 2964 aaronmk
    tables = lists.mk_seq(tables)
555 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
556 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
557 2121 aaronmk
558 2315 aaronmk
    # Parse other params
559 2376 aaronmk
    if conds == None: conds = []
560 2650 aaronmk
    elif dicts.is_dict(conds): conds = conds.items()
561 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
562 135 aaronmk
    assert limit == None or type(limit) == int
563 865 aaronmk
    assert start == None or type(start) == int
564 2315 aaronmk
    if order_by is order_by_pkey:
565
        if distinct_on != []: order_by = None
566
        else: order_by = pkey(db, table0, recover=True)
567 865 aaronmk
568 2315 aaronmk
    query = 'SELECT'
569 2056 aaronmk
570 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
571 2056 aaronmk
572 2200 aaronmk
    # DISTINCT ON columns
573 2233 aaronmk
    if distinct_on != []:
574 2467 aaronmk
        query += '\nDISTINCT'
575 2254 aaronmk
        if distinct_on is not distinct_on_all:
576 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
577
578
    # Columns
579 3027 aaronmk
    if fields == None:
580
        if query.find('\n') >= 0: whitespace = '\n'
581
        else: whitespace = ' '
582
        query += whitespace+'*'
583 2765 aaronmk
    else:
584
        assert fields != []
585 3027 aaronmk
        query += '\n'+('\n, '.join(map(parse_col, fields)))
586 2200 aaronmk
587
    # Main table
588 2467 aaronmk
    query += '\nFROM '+table0.to_str(db)
589 865 aaronmk
590 2122 aaronmk
    # Add joins
591 2271 aaronmk
    left_table = table0
592 2263 aaronmk
    for join_ in tables:
593
        table = join_.table
594 2238 aaronmk
595 2343 aaronmk
        # Parse special values
596
        if join_.type_ is sql_gen.filter_out: # filter no match
597 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
598 2853 aaronmk
                sql_gen.CompareCond(None, '~=')))
599 2343 aaronmk
600 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
601 2122 aaronmk
602
        left_table = table
603
604 865 aaronmk
    missing = True
605 2376 aaronmk
    if conds != []:
606 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
607
        else: whitespace = '\n'
608 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
609
            .to_str(db) for l, r in conds], 'WHERE')
610 865 aaronmk
        missing = False
611 2227 aaronmk
    if order_by != None:
612 2467 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
613
    if limit != None: query += '\nLIMIT '+str(limit); missing = False
614 865 aaronmk
    if start != None:
615 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
616 865 aaronmk
        missing = False
617
    if missing: warnings.warn(DbWarning(
618
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
619
620 2786 aaronmk
    return query
621 11 aaronmk
622 2054 aaronmk
def select(db, *args, **kw_args):
623
    '''For params, see mk_select() and run_query()'''
624
    recover = kw_args.pop('recover', None)
625
    cacheable = kw_args.pop('cacheable', True)
626 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
627 2054 aaronmk
628 2791 aaronmk
    return run_query(db, mk_select(db, *args, **kw_args), recover, cacheable,
629
        log_level=log_level)
630 2054 aaronmk
631 2788 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, returning=None,
632 3009 aaronmk
    embeddable=False, ignore=False):
633 1960 aaronmk
    '''
634
    @param returning str|None An inserted column (such as pkey) to return
635 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
636 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
637
        query will be fully cached, not just if it raises an exception.
638 3009 aaronmk
    @param ignore Whether to ignore duplicate keys.
639 1960 aaronmk
    '''
640 2754 aaronmk
    table = sql_gen.remove_table_rename(sql_gen.as_Table(table))
641 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
642 3010 aaronmk
    if cols != None: cols = [sql_gen.to_name_only_col(c, table) for c in cols]
643 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
644 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
645 2063 aaronmk
646 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
647 2063 aaronmk
648 3009 aaronmk
    def mk_insert(select_query):
649
        query = first_line
650 3014 aaronmk
        if cols != None:
651
            query += '\n('+(', '.join((c.to_str(db) for c in cols)))+')'
652 3009 aaronmk
        query += '\n'+select_query
653
654
        if returning != None:
655
            returning_name_col = sql_gen.to_name_only_col(returning)
656
            query += '\nRETURNING '+returning_name_col.to_str(db)
657
658
        return query
659 2063 aaronmk
660 3017 aaronmk
    return_type = 'unknown'
661
    if returning != None: return_type = returning.to_str(db)+'%TYPE'
662
663 3009 aaronmk
    lang = 'sql'
664
    if ignore:
665 3017 aaronmk
        assert cols != None
666
        # Always return something to set the correct rowcount
667
        if returning == None: returning = sql_gen.NamedCol('NULL', None)
668
669 3009 aaronmk
        embeddable = True # must use function
670
        lang = 'plpgsql'
671 3014 aaronmk
        row = [sql_gen.Col(c.name, 'row') for c in cols]
672 3010 aaronmk
673 3009 aaronmk
        query = '''\
674 3010 aaronmk
DECLARE
675 3014 aaronmk
    row '''+table.to_str(db)+'''%ROWTYPE;
676 3009 aaronmk
BEGIN
677 3019 aaronmk
    /* Need an EXCEPTION block for each individual row because "When an error is
678
    caught by an EXCEPTION clause, [...] all changes to persistent database
679
    state within the block are rolled back."
680
    This is unfortunate because "A block containing an EXCEPTION clause is
681
    significantly more expensive to enter and exit than a block without one."
682 3015 aaronmk
    (http://www.postgresql.org/docs/8.3/static/plpgsql-control-structures.html\
683
#PLPGSQL-ERROR-TRAPPING)
684
    */
685 3034 aaronmk
    FOR '''+(', '.join((c.to_str(db) for c in row)))+''' IN
686
'''+select_query+'''
687
    LOOP
688 3015 aaronmk
        BEGIN
689 3019 aaronmk
            RETURN QUERY
690 3014 aaronmk
'''+mk_insert(sql_gen.Values(row).to_str(db))+'''
691 3010 aaronmk
;
692 3015 aaronmk
        EXCEPTION
693 3019 aaronmk
            WHEN unique_violation THEN NULL; -- continue to next row
694 3015 aaronmk
        END;
695 3010 aaronmk
    END LOOP;
696
END;\
697 3009 aaronmk
'''
698
    else: query = mk_insert(select_query)
699
700 2070 aaronmk
    if embeddable:
701
        # Create function
702 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
703 2189 aaronmk
        while True:
704
            try:
705 2918 aaronmk
                function = db.TempFunction(function_name)
706 2194 aaronmk
707 2189 aaronmk
                function_query = '''\
708 2698 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''()
709 3017 aaronmk
RETURNS SETOF '''+return_type+'''
710 3009 aaronmk
LANGUAGE '''+lang+'''
711 2467 aaronmk
AS $$
712 3009 aaronmk
'''+query+'''
713 2467 aaronmk
$$;
714 2070 aaronmk
'''
715 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
716 2945 aaronmk
                    log_ignore_excs=(DuplicateException,))
717 2189 aaronmk
                break # this version was successful
718 2945 aaronmk
            except DuplicateException, e:
719 2189 aaronmk
                function_name = next_version(function_name)
720
                # try again with next version of name
721 2070 aaronmk
722 2337 aaronmk
        # Return query that uses function
723 3009 aaronmk
        cols = None
724
        if returning != None: cols = [returning]
725 2698 aaronmk
        func_table = sql_gen.NamedTable('f', sql_gen.FunctionCall(function),
726 3009 aaronmk
            cols) # AS clause requires function alias
727 2787 aaronmk
        return mk_select(db, func_table, start=0, order_by=None)
728 2070 aaronmk
729 2787 aaronmk
    return query
730 2066 aaronmk
731 3074 aaronmk
def insert_select(db, table, *args, **kw_args):
732 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
733 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
734
        values in
735 2072 aaronmk
    '''
736 2386 aaronmk
    into = kw_args.pop('into', None)
737
    if into != None: kw_args['embeddable'] = True
738 2066 aaronmk
    recover = kw_args.pop('recover', None)
739 3011 aaronmk
    if kw_args.get('ignore', False): recover = True
740 2066 aaronmk
    cacheable = kw_args.pop('cacheable', True)
741 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
742 2066 aaronmk
743 3074 aaronmk
    cur = run_query_into(db, mk_insert_select(db, table, *args, **kw_args),
744
        into, recover=recover, cacheable=cacheable, log_level=log_level)
745
    autoanalyze(db, table)
746
    return cur
747 2063 aaronmk
748 2738 aaronmk
default = sql_gen.default # tells insert() to use the default value for a column
749 2066 aaronmk
750 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
751 2085 aaronmk
    '''For params, see insert_select()'''
752 1960 aaronmk
    if lists.is_seq(row): cols = None
753
    else:
754
        cols = row.keys()
755
        row = row.values()
756 2738 aaronmk
    row = list(row) # ensure that "== []" works
757 1960 aaronmk
758 2738 aaronmk
    if row == []: query = None
759
    else: query = sql_gen.Values(row).to_str(db)
760 1961 aaronmk
761 2788 aaronmk
    return insert_select(db, table, cols, query, *args, **kw_args)
762 11 aaronmk
763 3056 aaronmk
def mk_update(db, table, changes=None, cond=None, in_place=False):
764 2402 aaronmk
    '''
765
    @param changes [(col, new_value),...]
766
        * container can be any iterable type
767
        * col: sql_gen.Code|str (for col name)
768
        * new_value: sql_gen.Code|literal value
769
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
770 3056 aaronmk
    @param in_place If set, locks the table and updates rows in place.
771
        This avoids creating dead rows in PostgreSQL.
772
        * cond must be None
773 2402 aaronmk
    @return str query
774
    '''
775 3057 aaronmk
    table = sql_gen.as_Table(table)
776
    changes = [(sql_gen.to_name_only_col(c, table), sql_gen.as_Value(v))
777
        for c, v in changes]
778
779 3056 aaronmk
    if in_place:
780
        assert cond == None
781 3058 aaronmk
782 3065 aaronmk
        query = 'ALTER TABLE '+table.to_str(db)+'\n'
783
        query += ',\n'.join(('ALTER COLUMN '+c.to_str(db)+' TYPE '
784
            +db.col_info(sql_gen.with_default_table(c, table)).type
785
            +'\nUSING '+v.to_str(db) for c, v in changes))
786 3058 aaronmk
    else:
787
        query = 'UPDATE '+table.to_str(db)+'\nSET\n'
788
        query += ',\n'.join((c.to_str(db)+' = '+v.to_str(db)
789
            for c, v in changes))
790
        if cond != None: query += '\nWHERE\n'+cond.to_str(db)
791 3056 aaronmk
792 2402 aaronmk
    return query
793
794 3074 aaronmk
def update(db, table, *args, **kw_args):
795 2402 aaronmk
    '''For params, see mk_update() and run_query()'''
796
    recover = kw_args.pop('recover', None)
797 3043 aaronmk
    cacheable = kw_args.pop('cacheable', False)
798 3030 aaronmk
    log_level = kw_args.pop('log_level', 2)
799 2402 aaronmk
800 3074 aaronmk
    cur = run_query(db, mk_update(db, table, *args, **kw_args), recover,
801
        cacheable, log_level=log_level)
802
    autoanalyze(db, table)
803
    return cur
804 2402 aaronmk
805 135 aaronmk
def last_insert_id(db):
806 1849 aaronmk
    module = util.root_module(db.db)
807 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
808
    elif module == 'MySQLdb': return db.insert_id()
809
    else: return None
810 13 aaronmk
811 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
812 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
813 2415 aaronmk
    to names that will be distinct among the columns' tables.
814 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
815 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
816
    @param into The table for the new columns.
817 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
818
        columns will be included in the mapping even if they are not in cols.
819
        The tables of the provided Col objects will be changed to into, so make
820
        copies of them if you want to keep the original tables.
821
    @param as_items Whether to return a list of dict items instead of a dict
822 2383 aaronmk
    @return dict(orig_col=new_col, ...)
823
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
824 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
825
        * All mappings use the into table so its name can easily be
826 2383 aaronmk
          changed for all columns at once
827
    '''
828 2415 aaronmk
    cols = lists.uniqify(cols)
829
830 2394 aaronmk
    items = []
831 2389 aaronmk
    for col in preserve:
832 2390 aaronmk
        orig_col = copy.copy(col)
833 2392 aaronmk
        col.table = into
834 2394 aaronmk
        items.append((orig_col, col))
835
    preserve = set(preserve)
836
    for col in cols:
837 2716 aaronmk
        if col not in preserve:
838
            items.append((col, sql_gen.Col(str(col), into, col.srcs)))
839 2394 aaronmk
840
    if not as_items: items = dict(items)
841
    return items
842 2383 aaronmk
843 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
844 2391 aaronmk
    '''For params, see mk_flatten_mapping()
845
    @return See return value of mk_flatten_mapping()
846
    '''
847 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
848
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
849 2786 aaronmk
    run_query_into(db, mk_select(db, joins, cols, limit=limit, start=start),
850 2846 aaronmk
        into=into, add_indexes_=True)
851 2394 aaronmk
    return dict(items)
852 2391 aaronmk
853 3050 aaronmk
def track_data_error(db, errors_table, cols, value, error_code, error):
854 2758 aaronmk
    '''
855
    @param errors_table If None, does nothing.
856
    '''
857 2810 aaronmk
    if errors_table == None or cols == (): return
858 3050 aaronmk
859
    for col in cols:
860
        try:
861
            insert(db, errors_table, dict(column=col.name, value=value,
862
                error_code=error_code, error=error), recover=True,
863
                cacheable=True, log_level=4)
864
        except DuplicateKeyException: pass
865 2758 aaronmk
866 2704 aaronmk
def cast(db, type_, col, errors_table=None):
867
    '''Casts an (unrenamed) column or value.
868 2769 aaronmk
    If errors_table set and col has srcs, saves errors in errors_table (using
869
    col's srcs attr as the source columns) and converts errors to warnings.
870 2820 aaronmk
    @param col str|sql_gen.Col|sql_gen.Literal
871 2704 aaronmk
    @param errors_table None|sql_gen.Table|str
872
    '''
873 2820 aaronmk
    col = sql_gen.as_Col(col)
874 2769 aaronmk
    save_errors = (errors_table != None and isinstance(col, sql_gen.Col)
875
        and col.srcs != ())
876 2987 aaronmk
    if not save_errors: return sql_gen.Cast(type_, col) # can't save errors
877 2704 aaronmk
878
    assert not isinstance(col, sql_gen.NamedCol)
879
880 2770 aaronmk
    errors_table = sql_gen.as_Table(errors_table)
881
    srcs = map(sql_gen.to_name_only_col, col.srcs)
882
    function_name = str(sql_gen.FunctionCall(type_, *srcs))
883 2918 aaronmk
    function = db.TempFunction(function_name)
884 2704 aaronmk
885 2718 aaronmk
    while True:
886
        # Create function definition
887 3049 aaronmk
        errors_table_cols = map(sql_gen.Col,
888
            ['column', 'value', 'error_code', 'error'])
889 2718 aaronmk
        query = '''\
890 2704 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''(value text)
891
RETURNS '''+type_+'''
892
LANGUAGE plpgsql
893
STRICT
894
AS $$
895
BEGIN
896
    /* The explicit cast to the return type is needed to make the cast happen
897
    inside the try block. (Implicit casts to the return type happen at the end
898
    of the function, outside any block.) */
899
    RETURN value::'''+type_+''';
900
EXCEPTION
901
    WHEN data_exception THEN
902
        -- Save error in errors table.
903 3049 aaronmk
        DECLARE
904
            error_code text := SQLSTATE;
905
            error text := SQLERRM;
906
            "column" text;
907
        BEGIN
908
            -- Insert the value and error for *each* source column.
909
            FOR "column" IN
910
'''+mk_select(db, sql_gen.NamedValues('c', None, [[c.name] for c in srcs]),
911 3053 aaronmk
    order_by=None, start=0)+'''
912 3049 aaronmk
            LOOP
913
                BEGIN
914
'''+mk_insert_select(db, errors_table, errors_table_cols,
915
    sql_gen.Values(errors_table_cols).to_str(db))+''';
916
                EXCEPTION
917
                    WHEN unique_violation THEN NULL; -- continue to next row
918
                END;
919
            END LOOP;
920
        END;
921 2704 aaronmk
922
        RAISE WARNING '%', SQLERRM;
923
        RETURN NULL;
924
END;
925
$$;
926
'''
927 2718 aaronmk
928
        # Create function
929
        try:
930
            run_query(db, query, recover=True, cacheable=True,
931 2945 aaronmk
                log_ignore_excs=(DuplicateException,))
932 2718 aaronmk
            break # successful
933 2945 aaronmk
        except DuplicateException:
934 2770 aaronmk
            function.name = next_version(function.name)
935
            # try again with next version of name
936 2704 aaronmk
937
    return sql_gen.FunctionCall(function, col)
938
939 2414 aaronmk
##### Database structure queries
940
941 2426 aaronmk
def table_row_count(db, table, recover=None):
942 2786 aaronmk
    return value(run_query(db, mk_select(db, table, [sql_gen.row_count],
943 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
944 2426 aaronmk
945 2414 aaronmk
def table_cols(db, table, recover=None):
946
    return list(col_names(select(db, table, limit=0, order_by=None,
947 2443 aaronmk
        recover=recover, log_level=4)))
948 2414 aaronmk
949 2291 aaronmk
def pkey(db, table, recover=None):
950 832 aaronmk
    '''Assumed to be first column in table'''
951 2339 aaronmk
    return table_cols(db, table, recover)[0]
952 832 aaronmk
953 2559 aaronmk
not_null_col = 'not_null_col'
954 2340 aaronmk
955
def table_not_null_col(db, table, recover=None):
956
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
957
    if not_null_col in table_cols(db, table, recover): return not_null_col
958
    else: return pkey(db, table, recover)
959
960 853 aaronmk
def index_cols(db, table, index):
961
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
962
    automatically created. When you don't know whether something is a UNIQUE
963
    constraint or a UNIQUE index, use this function.'''
964 1909 aaronmk
    module = util.root_module(db.db)
965
    if module == 'psycopg2':
966
        return list(values(run_query(db, '''\
967 853 aaronmk
SELECT attname
968 866 aaronmk
FROM
969
(
970
        SELECT attnum, attname
971
        FROM pg_index
972
        JOIN pg_class index ON index.oid = indexrelid
973
        JOIN pg_class table_ ON table_.oid = indrelid
974
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
975
        WHERE
976 2782 aaronmk
            table_.relname = '''+db.esc_value(table)+'''
977
            AND index.relname = '''+db.esc_value(index)+'''
978 866 aaronmk
    UNION
979
        SELECT attnum, attname
980
        FROM
981
        (
982
            SELECT
983
                indrelid
984
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
985
                    AS indkey
986
            FROM pg_index
987
            JOIN pg_class index ON index.oid = indexrelid
988
            JOIN pg_class table_ ON table_.oid = indrelid
989
            WHERE
990 2782 aaronmk
                table_.relname = '''+db.esc_value(table)+'''
991
                AND index.relname = '''+db.esc_value(index)+'''
992 866 aaronmk
        ) s
993
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
994
) s
995 853 aaronmk
ORDER BY attnum
996 2782 aaronmk
'''
997
            , cacheable=True, log_level=4)))
998 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
999
        ' database')
1000 853 aaronmk
1001 464 aaronmk
def constraint_cols(db, table, constraint):
1002 1849 aaronmk
    module = util.root_module(db.db)
1003 464 aaronmk
    if module == 'psycopg2':
1004
        return list(values(run_query(db, '''\
1005
SELECT attname
1006
FROM pg_constraint
1007
JOIN pg_class ON pg_class.oid = conrelid
1008
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
1009
WHERE
1010 2783 aaronmk
    relname = '''+db.esc_value(table)+'''
1011
    AND conname = '''+db.esc_value(constraint)+'''
1012 464 aaronmk
ORDER BY attnum
1013 2783 aaronmk
'''
1014
            )))
1015 464 aaronmk
    else: raise NotImplementedError("Can't list constraint columns for "+module+
1016
        ' database')
1017
1018 2096 aaronmk
row_num_col = '_row_num'
1019
1020 2998 aaronmk
def add_index(db, exprs, table=None, unique=False, ensure_not_null_=True):
1021 2688 aaronmk
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
1022 2538 aaronmk
    Currently, only function calls are supported as expressions.
1023 2998 aaronmk
    @param ensure_not_null_ If set, translates NULL values to sentinel values.
1024 2847 aaronmk
        This allows indexes to be used for comparisons where NULLs are equal.
1025 2538 aaronmk
    '''
1026 2964 aaronmk
    exprs = lists.mk_seq(exprs)
1027 2538 aaronmk
1028 2688 aaronmk
    # Parse exprs
1029
    old_exprs = exprs[:]
1030
    exprs = []
1031
    cols = []
1032
    for i, expr in enumerate(old_exprs):
1033 2823 aaronmk
        expr = sql_gen.as_Col(expr, table)
1034 2688 aaronmk
1035 2823 aaronmk
        # Handle nullable columns
1036 2998 aaronmk
        if ensure_not_null_:
1037
            try: expr = ensure_not_null(db, expr)
1038 2860 aaronmk
            except KeyError: pass # unknown type, so just create plain index
1039 2823 aaronmk
1040 2688 aaronmk
        # Extract col
1041 3002 aaronmk
        expr = copy.deepcopy(expr) # don't modify input!
1042 2688 aaronmk
        if isinstance(expr, sql_gen.FunctionCall):
1043
            col = expr.args[0]
1044
            expr = sql_gen.Expr(expr)
1045
        else: col = expr
1046 2823 aaronmk
        assert isinstance(col, sql_gen.Col)
1047 2688 aaronmk
1048
        # Extract table
1049
        if table == None:
1050
            assert sql_gen.is_table_col(col)
1051
            table = col.table
1052
1053
        col.table = None
1054
1055
        exprs.append(expr)
1056
        cols.append(col)
1057 2408 aaronmk
1058 2688 aaronmk
    table = sql_gen.as_Table(table)
1059
    index = sql_gen.Table(str(sql_gen.Col(','.join(map(str, cols)), table)))
1060
1061 3005 aaronmk
    # Add index
1062
    while True:
1063
        str_ = 'CREATE'
1064
        if unique: str_ += ' UNIQUE'
1065
        str_ += ' INDEX '+index.to_str(db)+' ON '+table.to_str(db)+' ('+(
1066
            ', '.join((v.to_str(db) for v in exprs)))+')'
1067
1068
        try:
1069
            run_query(db, str_, recover=True, cacheable=True, log_level=3,
1070
                log_ignore_excs=(DuplicateException,))
1071
            break
1072
        except DuplicateException:
1073
            index.name = next_version(index.name)
1074
            # try again with next version of name
1075 2408 aaronmk
1076 2686 aaronmk
def add_pkey(db, table, cols=None, recover=None):
1077
    '''Adds a primary key.
1078
    @param cols [sql_gen.Col,...] The columns in the primary key.
1079
        Defaults to the first column in the table.
1080 2406 aaronmk
    @pre The table must not already have a primary key.
1081
    '''
1082
    table = sql_gen.as_Table(table)
1083 2686 aaronmk
    if cols == None: cols = [pkey(db, table, recover)]
1084
    col_strs = [sql_gen.to_name_only_col(v).to_str(db) for v in cols]
1085 2406 aaronmk
1086 2781 aaronmk
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD PRIMARY KEY ('
1087
        +(', '.join(col_strs))+')', recover=True, cacheable=True, log_level=3,
1088 2945 aaronmk
        log_ignore_excs=(DuplicateException,))
1089 2406 aaronmk
1090 2997 aaronmk
def add_not_null(db, col):
1091
    table = col.table
1092
    col = sql_gen.to_name_only_col(col)
1093
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ALTER COLUMN '
1094 3031 aaronmk
        +col.to_str(db)+' SET NOT NULL', cacheable=True, log_level=3)
1095 2997 aaronmk
1096
def add_index_col(db, col, suffix, expr, nullable=True):
1097 3000 aaronmk
    if sql_gen.index_col(col) != None: return # already has index col
1098 2997 aaronmk
1099
    new_col = sql_gen.suffixed_col(col, suffix)
1100
1101 3006 aaronmk
    # Add column
1102 3038 aaronmk
    new_typed_col = sql_gen.TypedCol(new_col.name, db.col_info(col).type)
1103 3045 aaronmk
    add_col(db, col.table, new_typed_col, comment='src: '+repr(col),
1104
        log_level=3)
1105 3037 aaronmk
    new_col.name = new_typed_col.name # propagate any renaming
1106 3006 aaronmk
1107 3064 aaronmk
    update(db, col.table, [(new_col, expr)], in_place=True, cacheable=True,
1108
        log_level=3)
1109 2997 aaronmk
    if not nullable: add_not_null(db, new_col)
1110
    add_index(db, new_col)
1111
1112 3000 aaronmk
    col.table.index_cols[col.name] = new_col
1113 2997 aaronmk
1114 3047 aaronmk
# Controls when ensure_not_null() will use index columns
1115
not_null_index_cols_min_rows = 0 # rows; initially always use index columns
1116
1117 2997 aaronmk
def ensure_not_null(db, col):
1118
    '''For params, see sql_gen.ensure_not_null()'''
1119
    expr = sql_gen.ensure_not_null(db, col)
1120
1121 3047 aaronmk
    # If a nullable column in a temp table, add separate index column instead.
1122
    # Note that for small datasources, this adds 6-25% to the total import time.
1123
    if (sql_gen.is_temp_col(col) and isinstance(expr, sql_gen.EnsureNotNull)
1124
        and table_row_count(db, col.table) >= not_null_index_cols_min_rows):
1125 2997 aaronmk
        add_index_col(db, col, '::NOT NULL', expr, nullable=False)
1126 3000 aaronmk
        expr = sql_gen.index_col(col)
1127 2997 aaronmk
1128
    return expr
1129
1130 2772 aaronmk
already_indexed = object() # tells add_indexes() the pkey has already been added
1131
1132
def add_indexes(db, table, has_pkey=True):
1133
    '''Adds an index on all columns in a table.
1134
    @param has_pkey bool|already_indexed Whether a pkey instead of a regular
1135
        index should be added on the first column.
1136
        * If already_indexed, the pkey is assumed to have already been added
1137
    '''
1138
    cols = table_cols(db, table)
1139
    if has_pkey:
1140
        if has_pkey is not already_indexed: add_pkey(db, table)
1141
        cols = cols[1:]
1142
    for col in cols: add_index(db, col, table)
1143
1144 3042 aaronmk
def add_col(db, table, col, comment=None, **kw_args):
1145 3039 aaronmk
    '''
1146 3042 aaronmk
    @param col TypedCol Name may be versioned, so be sure to propagate any
1147
        renaming back to any source column for the TypedCol.
1148
    @param comment None|str SQL comment used to distinguish columns of the same
1149
        name from each other when they contain different data, to allow the
1150
        ADD COLUMN query to be cached. If not set, query will not be cached.
1151 3039 aaronmk
    '''
1152 2934 aaronmk
    assert isinstance(col, sql_gen.TypedCol)
1153 3036 aaronmk
1154
    while True:
1155 3042 aaronmk
        str_ = 'ALTER TABLE '+table.to_str(db)+' ADD COLUMN '+col.to_str(db)
1156
        if comment != None: str_ += ' '+sql_gen.esc_comment(comment)
1157
1158 3036 aaronmk
        try:
1159 3042 aaronmk
            run_query(db, str_, recover=True, cacheable=True, **kw_args)
1160 3036 aaronmk
            break
1161
        except DuplicateException:
1162
            col.name = next_version(col.name)
1163
            # try again with next version of name
1164 2934 aaronmk
1165 2935 aaronmk
row_num_typed_col = sql_gen.TypedCol(row_num_col, 'serial', nullable=False,
1166
    constraints='PRIMARY KEY')
1167
1168 2086 aaronmk
def add_row_num(db, table):
1169 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
1170
    be the primary key.'''
1171 2935 aaronmk
    add_col(db, table, row_num_typed_col, log_level=3)
1172 2086 aaronmk
1173 2936 aaronmk
def cast_temp_col(db, type_, col, errors_table=None):
1174
    '''Like cast(), but creates a new column with the cast values if the input
1175
    is a column.
1176
    @return The new column or cast value
1177
    '''
1178
    def cast_(col): return cast(db, type_, col, errors_table)
1179
1180
    try: col = sql_gen.underlying_col(col)
1181
    except sql_gen.NoUnderlyingTableException: return sql_gen.wrap(cast_, col)
1182
1183
    table = col.table
1184
    new_col = sql_gen.Col(sql_gen.concat(col.name, '::'+type_), table, col.srcs)
1185
    expr = cast_(col)
1186 3035 aaronmk
1187
    # Add column
1188 3038 aaronmk
    new_typed_col = sql_gen.TypedCol(new_col.name, type_)
1189 3045 aaronmk
    add_col(db, table, new_typed_col, comment='src: '+repr(col))
1190 3037 aaronmk
    new_col.name = new_typed_col.name # propagate any renaming
1191 3035 aaronmk
1192 3064 aaronmk
    update(db, table, [(new_col, expr)], in_place=True, cacheable=True)
1193 2955 aaronmk
    add_index(db, new_col)
1194 2936 aaronmk
1195
    return new_col
1196
1197 2889 aaronmk
def drop_table(db, table):
1198
    table = sql_gen.as_Table(table)
1199
    return run_query(db, 'DROP TABLE IF EXISTS '+table.to_str(db)+' CASCADE')
1200
1201 2760 aaronmk
def create_table(db, table, cols, has_pkey=True, col_indexes=True):
1202 2675 aaronmk
    '''Creates a table.
1203 2681 aaronmk
    @param cols [sql_gen.TypedCol,...] The column names and types
1204
    @param has_pkey If set, the first column becomes the primary key.
1205 2760 aaronmk
    @param col_indexes bool|[ref]
1206
        * If True, indexes will be added on all non-pkey columns.
1207
        * If a list reference, [0] will be set to a function to do this.
1208
          This can be used to delay index creation until the table is populated.
1209 2675 aaronmk
    '''
1210
    table = sql_gen.as_Table(table)
1211
1212 2681 aaronmk
    if has_pkey:
1213
        cols[0] = pkey = copy.copy(cols[0]) # don't modify input!
1214 2872 aaronmk
        pkey.constraints = 'PRIMARY KEY'
1215 2681 aaronmk
1216 2675 aaronmk
    str_ = 'CREATE TABLE '+table.to_str(db)+' (\n'
1217 2681 aaronmk
    str_ += '\n, '.join(v.to_str(db) for v in cols)
1218
    str_ += '\n);\n'
1219 2675 aaronmk
    run_query(db, str_, cacheable=True, log_level=2)
1220 2760 aaronmk
1221
    # Add indexes
1222 2773 aaronmk
    if has_pkey: has_pkey = already_indexed
1223
    def add_indexes_(): add_indexes(db, table, has_pkey)
1224
    if isinstance(col_indexes, list): col_indexes[0] = add_indexes_ # defer
1225
    elif col_indexes: add_indexes_() # add now
1226 2675 aaronmk
1227 3068 aaronmk
def analyze(db, table):
1228
    table = sql_gen.as_Table(table)
1229
    run_query(db, 'ANALYZE '+table.to_str(db), log_level=3)
1230
1231 3074 aaronmk
def autoanalyze(db, table):
1232
    if db.autoanalyze: analyze(db, table)
1233
1234 2684 aaronmk
def vacuum(db, table):
1235
    table = sql_gen.as_Table(table)
1236
    db.with_autocommit(lambda: run_query(db, 'VACUUM ANALYZE '+table.to_str(db),
1237
        log_level=3))
1238
1239 2970 aaronmk
def truncate(db, table, schema='public', **kw_args):
1240
    '''For params, see run_query()'''
1241 2777 aaronmk
    table = sql_gen.as_Table(table, schema)
1242 2970 aaronmk
    return run_query(db, 'TRUNCATE '+table.to_str(db)+' CASCADE', **kw_args)
1243 2732 aaronmk
1244 2965 aaronmk
def empty_temp(db, tables):
1245 2972 aaronmk
    if db.debug_temp: return # leave temp tables there for debugging
1246 2965 aaronmk
    tables = lists.mk_seq(tables)
1247 2971 aaronmk
    for table in tables: truncate(db, table, log_level=3)
1248 2965 aaronmk
1249 2729 aaronmk
def tables(db, schema_like='public', table_like='%', exact=False):
1250
    if exact: compare = '='
1251
    else: compare = 'LIKE'
1252
1253 1849 aaronmk
    module = util.root_module(db.db)
1254 832 aaronmk
    if module == 'psycopg2':
1255 2779 aaronmk
        conds = [('schemaname', sql_gen.CompareCond(schema_like, compare)),
1256
            ('tablename', sql_gen.CompareCond(table_like, compare))]
1257
        return values(select(db, 'pg_tables', ['tablename'], conds,
1258
            order_by='tablename', log_level=4))
1259 1968 aaronmk
    elif module == 'MySQLdb':
1260 2779 aaronmk
        return values(run_query(db, 'SHOW TABLES LIKE '+db.esc_value(table_like)
1261
            , cacheable=True, log_level=4))
1262 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
1263 830 aaronmk
1264 2726 aaronmk
def table_exists(db, table):
1265
    table = sql_gen.as_Table(table)
1266 2730 aaronmk
    return list(tables(db, table.schema, table.name, exact=True)) != []
1267 2726 aaronmk
1268 2913 aaronmk
def function_exists(db, function):
1269
    function = sql_gen.as_Function(function)
1270
1271
    info_table = sql_gen.Table('routines', 'information_schema')
1272
    conds = [('routine_name', function.name)]
1273
    schema = function.schema
1274
    if schema != None: conds.append(('routine_schema', schema))
1275 2914 aaronmk
    # Exclude trigger functions, since they cannot be called directly
1276
    conds.append(('data_type', sql_gen.CompareCond('trigger', '!=')))
1277 2913 aaronmk
1278
    return list(values(select(db, info_table, ['routine_name'], conds,
1279
        order_by='routine_schema', limit=1, log_level=4))) != []
1280
        # TODO: order_by search_path schema order
1281
1282 2726 aaronmk
def errors_table(db, table, if_exists=True):
1283
    '''
1284
    @param if_exists If set, returns None if the errors table doesn't exist
1285
    @return None|sql_gen.Table
1286
    '''
1287
    table = sql_gen.as_Table(table)
1288
    if table.srcs != (): table = table.srcs[0]
1289
1290
    errors_table = sql_gen.suffixed_table(table, '.errors')
1291
    if if_exists and not table_exists(db, errors_table): return None
1292
    return errors_table
1293
1294 833 aaronmk
##### Database management
1295
1296 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
1297
    '''For kw_args, see tables()'''
1298
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
1299 833 aaronmk
1300 832 aaronmk
##### Heuristic queries
1301
1302 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
1303 1554 aaronmk
    '''Recovers from errors.
1304 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
1305
    '''
1306 2642 aaronmk
    row = sql_gen.ColDict(db, table, row)
1307 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
1308
1309 471 aaronmk
    try:
1310 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
1311 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
1312
            row_ct_ref[0] += cur.rowcount
1313
        return value(cur)
1314 471 aaronmk
    except DuplicateKeyException, e:
1315 2653 aaronmk
        row = sql_gen.ColDict(db, table,
1316
            util.dict_subset_right_join(row, e.cols))
1317
        return value(select(db, table, [pkey_], row, recover=True))
1318 471 aaronmk
1319 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
1320 830 aaronmk
    '''Recovers from errors'''
1321 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
1322 14 aaronmk
    except StopIteration:
1323 40 aaronmk
        if not create: raise
1324 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
1325 2078 aaronmk
1326 2593 aaronmk
def is_func_result(col):
1327
    return col.table.name.find('(') >= 0 and col.name == 'result'
1328
1329 2592 aaronmk
def into_table_name(out_table, in_tables0, mapping, is_func):
1330 2615 aaronmk
    def in_col_str(in_col):
1331
        in_col = sql_gen.remove_col_rename(in_col)
1332
        if isinstance(in_col, sql_gen.Col):
1333
            table = in_col.table
1334
            if table == in_tables0:
1335
                in_col = sql_gen.to_name_only_col(in_col)
1336
            elif is_func_result(in_col): in_col = table # omit col name
1337
        return str(in_col)
1338
1339 2580 aaronmk
    str_ = str(out_table)
1340
    if is_func:
1341 2615 aaronmk
        str_ += '('
1342 2580 aaronmk
1343 2615 aaronmk
        try: value_in_col = mapping['value']
1344
        except KeyError:
1345 2654 aaronmk
            str_ += ', '.join((str(k)+'='+in_col_str(v)
1346 2615 aaronmk
                for k, v in mapping.iteritems()))
1347
        else: str_ += in_col_str(value_in_col)
1348
1349
        str_ += ')'
1350 2812 aaronmk
    else:
1351
        out_col = 'rank'
1352
        try: in_col = mapping[out_col]
1353
        except KeyError: str_ += '_pkeys'
1354
        else: # has a rank column, so hierarchical
1355
            str_ += '['+str(out_col)+'='+in_col_str(in_col)+']'
1356 2580 aaronmk
    return str_
1357
1358 2508 aaronmk
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None, into=None,
1359 2805 aaronmk
    default=None, is_func=False, on_error=exc.raise_):
1360 2078 aaronmk
    '''Recovers from errors.
1361
    Only works under PostgreSQL (uses INSERT RETURNING).
1362 2909 aaronmk
    IMPORTANT: Must be run at the *beginning* of a transaction.
1363 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
1364
        tables to join with it using the main input table's pkey
1365 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
1366 2612 aaronmk
        * out_table_col: str (*not* sql_gen.Col)
1367
        * in_table_col: sql_gen.Col|literal-value
1368 2489 aaronmk
    @param into The table to contain the output and input pkeys.
1369 2574 aaronmk
        Defaults to `out_table.name+'_pkeys'`.
1370 2509 aaronmk
    @param default The *output* column to use as the pkey for missing rows.
1371
        If this output column does not exist in the mapping, uses None.
1372 2552 aaronmk
    @param is_func Whether out_table is the name of a SQL function, not a table
1373 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
1374 2078 aaronmk
    '''
1375 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
1376 2312 aaronmk
1377 2450 aaronmk
    def log_debug(msg): db.log_debug(msg, level=1.5)
1378 2505 aaronmk
    def col_ustr(str_):
1379 2567 aaronmk
        return strings.repr_no_u(sql_gen.remove_col_rename(str_))
1380 2450 aaronmk
1381 2910 aaronmk
    log_debug('********** New iteration **********')
1382
    log_debug('Inserting these input columns into '+strings.as_tt(
1383
        out_table.to_str(db))+':\n'+strings.as_table(mapping, ustr=col_ustr))
1384
1385 2942 aaronmk
    is_function = function_exists(db, out_table)
1386
1387
    if is_function: out_pkey = 'result'
1388
    else: out_pkey = pkey(db, out_table, recover=True)
1389 2768 aaronmk
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
1390
1391
    if mapping == {}: # need at least one column for INSERT SELECT
1392
        mapping = {out_pkey: None} # ColDict will replace with default value
1393
1394 2382 aaronmk
    # Create input joins from list of input tables
1395
    in_tables_ = in_tables[:] # don't modify input!
1396
    in_tables0 = in_tables_.pop(0) # first table is separate
1397 2735 aaronmk
    errors_table_ = errors_table(db, in_tables0)
1398 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
1399 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
1400 2460 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v,
1401
        {in_pkey: sql_gen.join_same_not_null}) for v in in_tables_]
1402 2131 aaronmk
1403 2592 aaronmk
    if into == None:
1404
        into = into_table_name(out_table, in_tables0, mapping, is_func)
1405
    into = sql_gen.as_Table(into)
1406
1407 2702 aaronmk
    # Set column sources
1408
    in_cols = filter(sql_gen.is_table_col, mapping.values())
1409
    for col in in_cols:
1410
        if col.table == in_tables0: col.set_srcs(sql_gen.src_self)
1411
1412 2486 aaronmk
    log_debug('Joining together input tables into temp table')
1413 2395 aaronmk
    # Place in new table for speed and so don't modify input if values edited
1414 2584 aaronmk
    in_table = sql_gen.Table('in')
1415 2702 aaronmk
    mapping = dicts.join(mapping, flatten(db, in_table, input_joins, in_cols,
1416
        preserve=[in_pkey_col], start=0))
1417 2395 aaronmk
    input_joins = [in_table]
1418 2486 aaronmk
    db.log_debug('Temp table: '+strings.as_tt(in_table.to_str(db)), level=2)
1419 2395 aaronmk
1420 2656 aaronmk
    mapping = sql_gen.ColDict(db, out_table, mapping)
1421
        # after applying dicts.join() because that returns a plain dict
1422
1423 2509 aaronmk
    # Resolve default value column
1424 2841 aaronmk
    if default != None:
1425
        try: default = mapping[default]
1426
        except KeyError:
1427 2509 aaronmk
            db.log_debug('Default value column '
1428
                +strings.as_tt(strings.repr_no_u(default))
1429 2511 aaronmk
                +' does not exist in mapping, falling back to None', level=2.1)
1430 2509 aaronmk
            default = None
1431
1432 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
1433 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
1434
1435 2201 aaronmk
    pkeys_table_exists_ref = [False]
1436 2960 aaronmk
    def insert_into_pkeys(joins, cols, distinct=False):
1437
        kw_args = {}
1438
        if distinct: kw_args.update(dict(distinct_on=[in_pkey_col]))
1439
        query = mk_select(db, joins, cols, order_by=None, start=0, **kw_args)
1440
1441 2201 aaronmk
        if pkeys_table_exists_ref[0]:
1442 2786 aaronmk
            insert_select(db, into, pkeys_names, query)
1443 2201 aaronmk
        else:
1444 2786 aaronmk
            run_query_into(db, query, into=into)
1445 2201 aaronmk
            pkeys_table_exists_ref[0] = True
1446
1447 2429 aaronmk
    limit_ref = [None]
1448 2380 aaronmk
    conds = set()
1449 2811 aaronmk
    distinct_on = sql_gen.ColDict(db, out_table)
1450 2325 aaronmk
    def mk_main_select(joins, cols):
1451 2811 aaronmk
        distinct_on_cols = [c.to_Col() for c in distinct_on.values()]
1452
        return mk_select(db, joins, cols, conds, distinct_on_cols,
1453 2429 aaronmk
            limit=limit_ref[0], start=0)
1454 2132 aaronmk
1455 2519 aaronmk
    exc_strs = set()
1456 2309 aaronmk
    def log_exc(e):
1457 2519 aaronmk
        e_str = exc.str_(e, first_line_only=True)
1458
        log_debug('Caught exception: '+e_str)
1459
        assert e_str not in exc_strs # avoid infinite loops
1460
        exc_strs.add(e_str)
1461 2735 aaronmk
1462 2451 aaronmk
    def remove_all_rows():
1463 2896 aaronmk
        log_debug('Ignoring all rows')
1464 2429 aaronmk
        limit_ref[0] = 0 # just create an empty pkeys table
1465 2735 aaronmk
1466 2758 aaronmk
    def ignore(in_col, value, e):
1467
        track_data_error(db, errors_table_, in_col.srcs, value, e.cause.pgcode,
1468
            e.cause.pgerror)
1469 2895 aaronmk
        log_debug('Ignoring rows with '+strings.as_tt(repr(in_col))+' = '
1470 2545 aaronmk
            +strings.as_tt(repr(value)))
1471 2895 aaronmk
1472 2758 aaronmk
    def remove_rows(in_col, value, e):
1473
        ignore(in_col, value, e)
1474 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
1475
        assert cond not in conds # avoid infinite loops
1476 2380 aaronmk
        conds.add(cond)
1477 2895 aaronmk
1478 2758 aaronmk
    def invalid2null(in_col, value, e):
1479
        ignore(in_col, value, e)
1480 2403 aaronmk
        update(db, in_table, [(in_col, None)],
1481
            sql_gen.ColValueCond(in_col, value))
1482 2245 aaronmk
1483 2589 aaronmk
    def insert_pkeys_table(which):
1484 2932 aaronmk
        return sql_gen.Table(sql_gen.concat(in_table.name,
1485 2589 aaronmk
            '_insert_'+which+'_pkeys'))
1486
    insert_out_pkeys = insert_pkeys_table('out')
1487
    insert_in_pkeys = insert_pkeys_table('in')
1488
1489 2206 aaronmk
    # Do inserts and selects
1490 2642 aaronmk
    join_cols = sql_gen.ColDict(db, out_table)
1491 2206 aaronmk
    while True:
1492 2521 aaronmk
        if limit_ref[0] == 0: # special case
1493
            log_debug('Creating an empty pkeys table')
1494 2786 aaronmk
            cur = run_query_into(db, mk_select(db, out_table, [out_pkey],
1495 2521 aaronmk
                limit=limit_ref[0]), into=insert_out_pkeys)
1496
            break # don't do main case
1497
1498 2303 aaronmk
        has_joins = join_cols != {}
1499
1500 2903 aaronmk
        log_debug('Trying to insert new rows')
1501
1502 2305 aaronmk
        # Prepare to insert new rows
1503 2325 aaronmk
        insert_joins = input_joins[:] # don't modify original!
1504 2403 aaronmk
        insert_args = dict(recover=True, cacheable=False)
1505 2303 aaronmk
        if has_joins:
1506 3021 aaronmk
            insert_args.update(dict(ignore=True))
1507 2325 aaronmk
        else:
1508 2404 aaronmk
            insert_args.update(dict(returning=out_pkey, into=insert_out_pkeys))
1509 2786 aaronmk
        main_select = mk_main_select(insert_joins, mapping.values())
1510 2303 aaronmk
1511 2929 aaronmk
        def main_insert():
1512 2942 aaronmk
            if is_function:
1513
                log_debug('Calling function on input rows')
1514
                args = dict(((k.name, v) for k, v in mapping.iteritems()))
1515
                func_call = sql_gen.NamedCol(out_pkey,
1516
                    sql_gen.FunctionCall(out_table, **args))
1517
                insert_into_pkeys(input_joins, [in_pkey_col, func_call])
1518
                return None
1519
            else:
1520
                return insert_select(db, out_table, mapping.keys(), main_select,
1521
                    **insert_args)
1522 2929 aaronmk
1523 2206 aaronmk
        try:
1524 2929 aaronmk
            cur = with_savepoint(db, main_insert)
1525 2357 aaronmk
            break # insert successful
1526 2931 aaronmk
        except MissingCastException, e:
1527
            log_exc(e)
1528
1529
            out_col = e.col
1530
            type_ = e.type
1531
1532
            log_debug('Casting '+strings.as_tt(out_col)+' input to '
1533
                +strings.as_tt(type_))
1534 2937 aaronmk
            mapping[out_col] = cast_temp_col(db, type_, mapping[out_col],
1535
                errors_table_)
1536 2206 aaronmk
        except DuplicateKeyException, e:
1537 2309 aaronmk
            log_exc(e)
1538
1539 2258 aaronmk
            old_join_cols = join_cols.copy()
1540 2811 aaronmk
            distinct_on.update(util.dict_subset(mapping, e.cols))
1541 2565 aaronmk
            join_cols.update(util.dict_subset_right_join(mapping, e.cols))
1542 2486 aaronmk
            log_debug('Ignoring existing rows, comparing on these columns:\n'
1543 2505 aaronmk
                +strings.as_inline_table(join_cols, ustr=col_ustr))
1544 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
1545 2230 aaronmk
        except NullValueException, e:
1546 2309 aaronmk
            log_exc(e)
1547
1548 2230 aaronmk
            out_col, = e.cols
1549
            try: in_col = mapping[out_col]
1550 2356 aaronmk
            except KeyError:
1551 2486 aaronmk
                log_debug('Missing mapping for NOT NULL column '+out_col)
1552 2451 aaronmk
                remove_all_rows()
1553 2758 aaronmk
            else: remove_rows(in_col, None, e)
1554 2542 aaronmk
        except FunctionValueException, e:
1555
            log_exc(e)
1556
1557
            func_name = e.name
1558
            value = e.value
1559
            for out_col, in_col in mapping.iteritems():
1560 2758 aaronmk
                in_col = sql_gen.unwrap_func_call(in_col, func_name)
1561
                invalid2null(in_col, value, e)
1562 2429 aaronmk
        except DatabaseErrors, e:
1563
            log_exc(e)
1564
1565 2808 aaronmk
            log_debug('No handler for exception')
1566
            on_error(e)
1567 2451 aaronmk
            remove_all_rows()
1568 2358 aaronmk
        # after exception handled, rerun loop with additional constraints
1569 2132 aaronmk
1570 2942 aaronmk
    if cur != None and row_ct_ref != None and cur.rowcount >= 0:
1571 2357 aaronmk
        row_ct_ref[0] += cur.rowcount
1572
1573 2942 aaronmk
    if is_function: pass # pkeys table already created
1574
    elif has_joins:
1575 2357 aaronmk
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
1576 2486 aaronmk
        log_debug('Getting output table pkeys of existing/inserted rows')
1577 2960 aaronmk
        insert_into_pkeys(select_joins, pkeys_cols, distinct=True)
1578 2357 aaronmk
    else:
1579 2404 aaronmk
        add_row_num(db, insert_out_pkeys) # for joining with input pkeys
1580 2357 aaronmk
1581 2486 aaronmk
        log_debug('Getting input table pkeys of inserted rows')
1582 2786 aaronmk
        run_query_into(db, mk_main_select(input_joins, [in_pkey]),
1583 2404 aaronmk
            into=insert_in_pkeys)
1584
        add_row_num(db, insert_in_pkeys) # for joining with output pkeys
1585 2357 aaronmk
1586 2428 aaronmk
        assert table_row_count(db, insert_out_pkeys) == table_row_count(db,
1587
            insert_in_pkeys)
1588
1589 2486 aaronmk
        log_debug('Combining output and input pkeys in inserted order')
1590 2404 aaronmk
        pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
1591 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
1592 2420 aaronmk
        insert_into_pkeys(pkey_joins, pkeys_names)
1593 2959 aaronmk
1594 2974 aaronmk
        empty_temp(db, [insert_out_pkeys, insert_in_pkeys])
1595 2357 aaronmk
1596 2486 aaronmk
    db.log_debug('Adding pkey on pkeys table to enable fast joins', level=2.5)
1597 2594 aaronmk
    add_pkey(db, into)
1598 2407 aaronmk
1599 2508 aaronmk
    log_debug('Setting pkeys of missing rows to '+strings.as_tt(repr(default)))
1600 2489 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(into,
1601 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
1602
        # must use join_same_not_null or query will take forever
1603 2420 aaronmk
    insert_into_pkeys(missing_rows_joins,
1604 2508 aaronmk
        [in_pkey_col, sql_gen.NamedCol(out_pkey, default)])
1605 2357 aaronmk
1606 2489 aaronmk
    assert table_row_count(db, into) == table_row_count(db, in_table)
1607 2428 aaronmk
1608 2974 aaronmk
    empty_temp(db, in_table)
1609 2959 aaronmk
1610 2702 aaronmk
    srcs = []
1611
    if is_func: srcs = sql_gen.cols_srcs(in_cols)
1612
    return sql_gen.Col(out_pkey, into, srcs)
1613 2115 aaronmk
1614
##### Data cleanup
1615
1616 2290 aaronmk
def cleanup_table(db, table, cols):
1617 2677 aaronmk
    table = sql_gen.as_Table(table)
1618
    cols = map(sql_gen.as_Col, cols)
1619 2115 aaronmk
1620 2677 aaronmk
    expr = ('nullif(nullif(trim(both from %s), '+db.esc_value('')+'), '
1621
        +db.esc_value(r'\N')+')')
1622
    changes = [(v, sql_gen.CustomCode(expr % v.to_str(db)))
1623
        for v in cols]
1624 2115 aaronmk
1625 3066 aaronmk
    update(db, table, changes, in_place=True)