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 1869 aaronmk
169
        self.__db = None
170 1889 aaronmk
        self.query_results = {}
171 2139 aaronmk
        self._savepoint = 0
172 2671 aaronmk
        self._notices_seen = set()
173 1869 aaronmk
174
    def __getattr__(self, name):
175
        if name == '__dict__': raise Exception('getting __dict__')
176
        if name == 'db': return self._db()
177
        else: raise AttributeError()
178
179
    def __getstate__(self):
180
        state = copy.copy(self.__dict__) # shallow copy
181 1915 aaronmk
        state['log_debug'] = None # don't pickle the debug callback
182 1869 aaronmk
        state['_DbConn__db'] = None # don't pickle the connection
183
        return state
184
185 2165 aaronmk
    def connected(self): return self.__db != None
186
187 1869 aaronmk
    def _db(self):
188
        if self.__db == None:
189
            # Process db_config
190
            db_config = self.db_config.copy() # don't modify input!
191 2097 aaronmk
            schemas = db_config.pop('schemas', None)
192 1869 aaronmk
            module_name, mappings = db_engines[db_config.pop('engine')]
193
            module = __import__(module_name)
194
            _add_module(module)
195
            for orig, new in mappings.iteritems():
196
                try: util.rename_key(db_config, orig, new)
197
                except KeyError: pass
198
199
            # Connect
200
            self.__db = module.connect(**db_config)
201
202
            # Configure connection
203 2906 aaronmk
            if hasattr(self.db, 'set_isolation_level'):
204
                import psycopg2.extensions
205
                self.db.set_isolation_level(
206
                    psycopg2.extensions.ISOLATION_LEVEL_READ_COMMITTED)
207 2101 aaronmk
            if schemas != None:
208 2893 aaronmk
                search_path = [self.esc_name(s) for s in schemas.split(',')]
209
                search_path.append(value(run_query(self, 'SHOW search_path',
210
                    log_level=4)))
211
                run_query(self, 'SET search_path TO '+(','.join(search_path)),
212
                    log_level=3)
213 1869 aaronmk
214
        return self.__db
215 1889 aaronmk
216 1891 aaronmk
    class DbCursor(Proxy):
217 1927 aaronmk
        def __init__(self, outer):
218 1891 aaronmk
            Proxy.__init__(self, outer.db.cursor())
219 2191 aaronmk
            self.outer = outer
220 1927 aaronmk
            self.query_results = outer.query_results
221 1894 aaronmk
            self.query_lookup = None
222 1891 aaronmk
            self.result = []
223 1889 aaronmk
224 2802 aaronmk
        def execute(self, query):
225 2764 aaronmk
            self._is_insert = query.startswith('INSERT')
226 2797 aaronmk
            self.query_lookup = query
227 2148 aaronmk
            try:
228 2191 aaronmk
                try:
229 2802 aaronmk
                    cur = self.inner.execute(query)
230 2191 aaronmk
                    self.outer.do_autocommit()
231 2802 aaronmk
                finally: self.query = get_cur_query(self.inner, query)
232 1904 aaronmk
            except Exception, e:
233 2802 aaronmk
                _add_cursor_info(e, self, query)
234 1904 aaronmk
                self.result = e # cache the exception as the result
235
                self._cache_result()
236
                raise
237 3004 aaronmk
238
            # Always cache certain queries
239
            if query.startswith('CREATE') or query.startswith('ALTER'):
240 3007 aaronmk
                # structural changes
241 3040 aaronmk
                # Rest of query must be unique in the face of name collisions,
242
                # so don't cache ADD COLUMN unless it has distinguishing comment
243
                if query.find('ADD COLUMN') < 0 or query.endswith('*/'):
244 3007 aaronmk
                    self._cache_result()
245 3004 aaronmk
            elif self.rowcount == 0 and query.startswith('SELECT'): # empty
246 2800 aaronmk
                consume_rows(self) # fetch all rows so result will be cached
247 3004 aaronmk
248 2762 aaronmk
            return cur
249 1894 aaronmk
250 1891 aaronmk
        def fetchone(self):
251
            row = self.inner.fetchone()
252 1899 aaronmk
            if row != None: self.result.append(row)
253
            # otherwise, fetched all rows
254 1904 aaronmk
            else: self._cache_result()
255
            return row
256
257
        def _cache_result(self):
258 2948 aaronmk
            # For inserts that return a result set, don't cache result set since
259
            # inserts are not idempotent. Other non-SELECT queries don't have
260
            # their result set read, so only exceptions will be cached (an
261
            # invalid query will always be invalid).
262 1930 aaronmk
            if self.query_results != None and (not self._is_insert
263 1906 aaronmk
                or isinstance(self.result, Exception)):
264
265 1894 aaronmk
                assert self.query_lookup != None
266 1916 aaronmk
                self.query_results[self.query_lookup] = self.CacheCursor(
267
                    util.dict_subset(dicts.AttrsDictView(self),
268
                    ['query', 'result', 'rowcount', 'description']))
269 1906 aaronmk
270 1916 aaronmk
        class CacheCursor:
271
            def __init__(self, cached_result): self.__dict__ = cached_result
272
273 1927 aaronmk
            def execute(self, *args, **kw_args):
274 1916 aaronmk
                if isinstance(self.result, Exception): raise self.result
275
                # otherwise, result is a rows list
276
                self.iter = iter(self.result)
277
278
            def fetchone(self):
279
                try: return self.iter.next()
280
                except StopIteration: return None
281 1891 aaronmk
282 2212 aaronmk
    def esc_value(self, value):
283 2663 aaronmk
        try: str_ = self.mogrify('%s', [value])
284
        except NotImplementedError, e:
285
            module = util.root_module(self.db)
286
            if module == 'MySQLdb':
287
                import _mysql
288
                str_ = _mysql.escape_string(value)
289
            else: raise e
290 2374 aaronmk
        return strings.to_unicode(str_)
291 2212 aaronmk
292 2347 aaronmk
    def esc_name(self, name): return esc_name(self, name) # calls global func
293
294 2814 aaronmk
    def std_code(self, str_):
295
        '''Standardizes SQL code.
296
        * Ensures that string literals are prefixed by `E`
297
        '''
298
        if str_.startswith("'"): str_ = 'E'+str_
299
        return str_
300
301 2665 aaronmk
    def can_mogrify(self):
302 2663 aaronmk
        module = util.root_module(self.db)
303 2665 aaronmk
        return module == 'psycopg2'
304 2663 aaronmk
305 2665 aaronmk
    def mogrify(self, query, params=None):
306
        if self.can_mogrify(): return self.db.cursor().mogrify(query, params)
307
        else: raise NotImplementedError("Can't mogrify query")
308
309 2671 aaronmk
    def print_notices(self):
310 2725 aaronmk
        if hasattr(self.db, 'notices'):
311
            for msg in self.db.notices:
312
                if msg not in self._notices_seen:
313
                    self._notices_seen.add(msg)
314
                    self.log_debug(msg, level=2)
315 2671 aaronmk
316 2793 aaronmk
    def run_query(self, query, cacheable=False, log_level=2,
317 2464 aaronmk
        debug_msg_ref=None):
318 2445 aaronmk
        '''
319 2464 aaronmk
        @param log_ignore_excs The log_level will be increased by 2 if the query
320
            throws one of these exceptions.
321 2664 aaronmk
        @param debug_msg_ref If specified, the log message will be returned in
322
            this instead of being output. This allows you to filter log messages
323
            depending on the result of the query.
324 2445 aaronmk
        '''
325 2167 aaronmk
        assert query != None
326
327 2047 aaronmk
        if not self.caching: cacheable = False
328 1903 aaronmk
        used_cache = False
329 2664 aaronmk
330
        def log_msg(query):
331
            if used_cache: cache_status = 'cache hit'
332
            elif cacheable: cache_status = 'cache miss'
333
            else: cache_status = 'non-cacheable'
334
            return 'DB query: '+cache_status+':\n'+strings.as_code(query, 'SQL')
335
336 1903 aaronmk
        try:
337 1927 aaronmk
            # Get cursor
338
            if cacheable:
339
                try:
340 2797 aaronmk
                    cur = self.query_results[query]
341 1927 aaronmk
                    used_cache = True
342
                except KeyError: cur = self.DbCursor(self)
343
            else: cur = self.db.cursor()
344
345 2664 aaronmk
            # Log query
346
            if self.debug and debug_msg_ref == None: # log before running
347
                self.log_debug(log_msg(query), log_level)
348
349 1927 aaronmk
            # Run query
350 2793 aaronmk
            cur.execute(query)
351 1903 aaronmk
        finally:
352 2671 aaronmk
            self.print_notices()
353 2664 aaronmk
            if self.debug and debug_msg_ref != None: # return after running
354 2793 aaronmk
                debug_msg_ref[0] = log_msg(str(get_cur_query(cur, query)))
355 1903 aaronmk
356
        return cur
357 1914 aaronmk
358 2797 aaronmk
    def is_cached(self, query): return query in self.query_results
359 2139 aaronmk
360 2907 aaronmk
    def with_autocommit(self, func):
361 2801 aaronmk
        import psycopg2.extensions
362
363
        prev_isolation_level = self.db.isolation_level
364 2907 aaronmk
        self.db.set_isolation_level(
365
            psycopg2.extensions.ISOLATION_LEVEL_AUTOCOMMIT)
366 2683 aaronmk
        try: return func()
367 2801 aaronmk
        finally: self.db.set_isolation_level(prev_isolation_level)
368 2683 aaronmk
369 2139 aaronmk
    def with_savepoint(self, func):
370 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
371 2443 aaronmk
        self.run_query('SAVEPOINT '+savepoint, log_level=4)
372 2139 aaronmk
        self._savepoint += 1
373 2930 aaronmk
        try: return func()
374 2139 aaronmk
        except:
375 2443 aaronmk
            self.run_query('ROLLBACK TO SAVEPOINT '+savepoint, log_level=4)
376 2139 aaronmk
            raise
377 2930 aaronmk
        finally:
378
            # Always release savepoint, because after ROLLBACK TO SAVEPOINT,
379
            # "The savepoint remains valid and can be rolled back to again"
380
            # (http://www.postgresql.org/docs/8.3/static/sql-rollback-to.html).
381 2443 aaronmk
            self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
382 2930 aaronmk
383
            self._savepoint -= 1
384
            assert self._savepoint >= 0
385
386
            self.do_autocommit() # OK to do this after ROLLBACK TO SAVEPOINT
387 2191 aaronmk
388
    def do_autocommit(self):
389
        '''Autocommits if outside savepoint'''
390
        assert self._savepoint >= 0
391
        if self.autocommit and self._savepoint == 0:
392 2924 aaronmk
            self.log_debug('Autocommitting', level=4)
393 2191 aaronmk
            self.db.commit()
394 2643 aaronmk
395 2819 aaronmk
    def col_info(self, col):
396 2643 aaronmk
        table = sql_gen.Table('columns', 'information_schema')
397 2821 aaronmk
        cols = ['data_type', 'column_default',
398
            cast(self, 'boolean', 'is_nullable')]
399 2643 aaronmk
400
        conds = [('table_name', col.table.name), ('column_name', col.name)]
401
        schema = col.table.schema
402
        if schema != None: conds.append(('table_schema', schema))
403
404 2819 aaronmk
        type_, default, nullable = row(select(self, table, cols, conds,
405 2874 aaronmk
            order_by='table_schema', limit=1, log_level=4))
406 2643 aaronmk
            # TODO: order_by search_path schema order
407 2819 aaronmk
        default = sql_gen.as_Code(default, self)
408
409
        return sql_gen.TypedCol(col.name, type_, default, nullable)
410 2917 aaronmk
411
    def TempFunction(self, name):
412
        if self.debug_temp: schema = None
413
        else: schema = 'pg_temp'
414
        return sql_gen.Function(name, schema)
415 1849 aaronmk
416 1869 aaronmk
connect = DbConn
417
418 832 aaronmk
##### Recoverable querying
419 15 aaronmk
420 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
421 11 aaronmk
422 2791 aaronmk
def run_query(db, query, recover=None, cacheable=False, log_level=2,
423
    log_ignore_excs=None, **kw_args):
424 2794 aaronmk
    '''For params, see DbConn.run_query()'''
425 830 aaronmk
    if recover == None: recover = False
426 2464 aaronmk
    if log_ignore_excs == None: log_ignore_excs = ()
427
    log_ignore_excs = tuple(log_ignore_excs)
428 830 aaronmk
429 2666 aaronmk
    debug_msg_ref = None # usually, db.run_query() logs query before running it
430
    # But if filtering with log_ignore_excs, wait until after exception parsing
431 2984 aaronmk
    if log_ignore_excs != () or not db.can_mogrify(): debug_msg_ref = [None]
432 2666 aaronmk
433 2148 aaronmk
    try:
434 2464 aaronmk
        try:
435 2794 aaronmk
            def run(): return db.run_query(query, cacheable, log_level,
436 2793 aaronmk
                debug_msg_ref, **kw_args)
437 2796 aaronmk
            if recover and not db.is_cached(query):
438 2464 aaronmk
                return with_savepoint(db, run)
439
            else: return run() # don't need savepoint if cached
440
        except Exception, e:
441
            msg = exc.str_(e)
442
443
            match = re.search(r'duplicate key value violates unique constraint '
444 2493 aaronmk
                r'"((_?[^\W_]+)_.+?)"', msg)
445 2464 aaronmk
            if match:
446
                constraint, table = match.groups()
447 3025 aaronmk
                cols = []
448
                if recover: # need auto-rollback to run index_cols()
449
                    try: cols = index_cols(db, table, constraint)
450
                    except NotImplementedError: pass
451
                raise DuplicateKeyException(constraint, cols, e)
452 2464 aaronmk
453 2493 aaronmk
            match = re.search(r'null value in column "(.+?)" violates not-null'
454 2464 aaronmk
                r' constraint', msg)
455
            if match: raise NullValueException('NOT NULL', [match.group(1)], e)
456
457
            match = re.search(r'\b(?:invalid input (?:syntax|value)\b.*?'
458
                r'|date/time field value out of range): "(.+?)"\n'
459 2535 aaronmk
                r'(?:(?s).*?)\bfunction "(.+?)"', msg)
460 2464 aaronmk
            if match:
461
                value, name = match.groups()
462
                raise FunctionValueException(name, strings.to_unicode(value), e)
463
464 2526 aaronmk
            match = re.search(r'column "(.+?)" is of type (.+?) but expression '
465 2523 aaronmk
                r'is of type', msg)
466
            if match:
467
                col, type_ = match.groups()
468
                raise MissingCastException(type_, col, e)
469
470 2946 aaronmk
            match = re.search(r'\b(\S+) "(.+?)".*? already exists', msg)
471 2945 aaronmk
            if match:
472
                type_, name = match.groups()
473
                raise DuplicateException(type_, name, e)
474 2464 aaronmk
475
            raise # no specific exception raised
476
    except log_ignore_excs:
477
        log_level += 2
478
        raise
479
    finally:
480 2666 aaronmk
        if debug_msg_ref != None and debug_msg_ref[0] != None:
481
            db.log_debug(debug_msg_ref[0], log_level)
482 830 aaronmk
483 832 aaronmk
##### Basic queries
484
485 2153 aaronmk
def next_version(name):
486 2163 aaronmk
    version = 1 # first existing name was version 0
487 2586 aaronmk
    match = re.match(r'^(.*)#(\d+)$', name)
488 2153 aaronmk
    if match:
489 2586 aaronmk
        name, version = match.groups()
490
        version = int(version)+1
491 2932 aaronmk
    return sql_gen.concat(name, '#'+str(version))
492 2153 aaronmk
493 2899 aaronmk
def lock_table(db, table, mode):
494
    table = sql_gen.as_Table(table)
495
    run_query(db, 'LOCK TABLE '+table.to_str(db)+' IN '+mode+' MODE')
496
497 2789 aaronmk
def run_query_into(db, query, into=None, add_indexes_=False, **kw_args):
498 2085 aaronmk
    '''Outputs a query to a temp table.
499
    For params, see run_query().
500
    '''
501 2789 aaronmk
    if into == None: return run_query(db, query, **kw_args)
502 2790 aaronmk
503
    assert isinstance(into, sql_gen.Table)
504
505 2992 aaronmk
    into.is_temp = True
506 3008 aaronmk
    # "temporary tables cannot specify a schema name", so remove schema
507
    into.schema = None
508 2992 aaronmk
509 2790 aaronmk
    kw_args['recover'] = True
510 2945 aaronmk
    kw_args.setdefault('log_ignore_excs', (DuplicateException,))
511 2790 aaronmk
512 2916 aaronmk
    temp = not db.debug_temp # tables are permanent in debug_temp mode
513 2790 aaronmk
514
    # Create table
515
    while True:
516
        create_query = 'CREATE'
517
        if temp: create_query += ' TEMP'
518
        create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
519 2385 aaronmk
520 2790 aaronmk
        try:
521
            cur = run_query(db, create_query, **kw_args)
522
                # CREATE TABLE AS sets rowcount to # rows in query
523
            break
524 2945 aaronmk
        except DuplicateException, e:
525 2790 aaronmk
            into.name = next_version(into.name)
526
            # try again with next version of name
527
528
    if add_indexes_: add_indexes(db, into)
529
530
    return cur
531 2085 aaronmk
532 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
533
534 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
535
536 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
537 2293 aaronmk
    start=None, order_by=order_by_pkey, default_table=None):
538 1981 aaronmk
    '''
539 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
540 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
541 1981 aaronmk
    @param fields Use None to select all fields in the table
542 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
543 2379 aaronmk
        * container can be any iterable type
544 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
545
        * compare_right_side: sql_gen.ValueCond|literal value
546 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
547
        use all columns
548 2786 aaronmk
    @return query
549 1981 aaronmk
    '''
550 2315 aaronmk
    # Parse tables param
551 2964 aaronmk
    tables = lists.mk_seq(tables)
552 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
553 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
554 2121 aaronmk
555 2315 aaronmk
    # Parse other params
556 2376 aaronmk
    if conds == None: conds = []
557 2650 aaronmk
    elif dicts.is_dict(conds): conds = conds.items()
558 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
559 135 aaronmk
    assert limit == None or type(limit) == int
560 865 aaronmk
    assert start == None or type(start) == int
561 2315 aaronmk
    if order_by is order_by_pkey:
562
        if distinct_on != []: order_by = None
563
        else: order_by = pkey(db, table0, recover=True)
564 865 aaronmk
565 2315 aaronmk
    query = 'SELECT'
566 2056 aaronmk
567 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
568 2056 aaronmk
569 2200 aaronmk
    # DISTINCT ON columns
570 2233 aaronmk
    if distinct_on != []:
571 2467 aaronmk
        query += '\nDISTINCT'
572 2254 aaronmk
        if distinct_on is not distinct_on_all:
573 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
574
575
    # Columns
576 3027 aaronmk
    if fields == None:
577
        if query.find('\n') >= 0: whitespace = '\n'
578
        else: whitespace = ' '
579
        query += whitespace+'*'
580 2765 aaronmk
    else:
581
        assert fields != []
582 3027 aaronmk
        query += '\n'+('\n, '.join(map(parse_col, fields)))
583 2200 aaronmk
584
    # Main table
585 2467 aaronmk
    query += '\nFROM '+table0.to_str(db)
586 865 aaronmk
587 2122 aaronmk
    # Add joins
588 2271 aaronmk
    left_table = table0
589 2263 aaronmk
    for join_ in tables:
590
        table = join_.table
591 2238 aaronmk
592 2343 aaronmk
        # Parse special values
593
        if join_.type_ is sql_gen.filter_out: # filter no match
594 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
595 2853 aaronmk
                sql_gen.CompareCond(None, '~=')))
596 2343 aaronmk
597 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
598 2122 aaronmk
599
        left_table = table
600
601 865 aaronmk
    missing = True
602 2376 aaronmk
    if conds != []:
603 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
604
        else: whitespace = '\n'
605 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
606
            .to_str(db) for l, r in conds], 'WHERE')
607 865 aaronmk
        missing = False
608 2227 aaronmk
    if order_by != None:
609 2467 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
610
    if limit != None: query += '\nLIMIT '+str(limit); missing = False
611 865 aaronmk
    if start != None:
612 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
613 865 aaronmk
        missing = False
614
    if missing: warnings.warn(DbWarning(
615
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
616
617 2786 aaronmk
    return query
618 11 aaronmk
619 2054 aaronmk
def select(db, *args, **kw_args):
620
    '''For params, see mk_select() and run_query()'''
621
    recover = kw_args.pop('recover', None)
622
    cacheable = kw_args.pop('cacheable', True)
623 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
624 2054 aaronmk
625 2791 aaronmk
    return run_query(db, mk_select(db, *args, **kw_args), recover, cacheable,
626
        log_level=log_level)
627 2054 aaronmk
628 2788 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, returning=None,
629 3009 aaronmk
    embeddable=False, ignore=False):
630 1960 aaronmk
    '''
631
    @param returning str|None An inserted column (such as pkey) to return
632 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
633 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
634
        query will be fully cached, not just if it raises an exception.
635 3009 aaronmk
    @param ignore Whether to ignore duplicate keys.
636 1960 aaronmk
    '''
637 2754 aaronmk
    table = sql_gen.remove_table_rename(sql_gen.as_Table(table))
638 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
639 3010 aaronmk
    if cols != None: cols = [sql_gen.to_name_only_col(c, table) for c in cols]
640 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
641 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
642 2063 aaronmk
643 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
644 2063 aaronmk
645 3009 aaronmk
    def mk_insert(select_query):
646
        query = first_line
647 3014 aaronmk
        if cols != None:
648
            query += '\n('+(', '.join((c.to_str(db) for c in cols)))+')'
649 3009 aaronmk
        query += '\n'+select_query
650
651
        if returning != None:
652
            returning_name_col = sql_gen.to_name_only_col(returning)
653
            query += '\nRETURNING '+returning_name_col.to_str(db)
654
655
        return query
656 2063 aaronmk
657 3017 aaronmk
    return_type = 'unknown'
658
    if returning != None: return_type = returning.to_str(db)+'%TYPE'
659
660 3009 aaronmk
    lang = 'sql'
661
    if ignore:
662 3017 aaronmk
        assert cols != None
663
        # Always return something to set the correct rowcount
664
        if returning == None: returning = sql_gen.NamedCol('NULL', None)
665
666 3009 aaronmk
        embeddable = True # must use function
667
        lang = 'plpgsql'
668 3014 aaronmk
        row = [sql_gen.Col(c.name, 'row') for c in cols]
669 3010 aaronmk
670 3009 aaronmk
        query = '''\
671 3010 aaronmk
DECLARE
672 3014 aaronmk
    row '''+table.to_str(db)+'''%ROWTYPE;
673 3009 aaronmk
BEGIN
674 3019 aaronmk
    /* Need an EXCEPTION block for each individual row because "When an error is
675
    caught by an EXCEPTION clause, [...] all changes to persistent database
676
    state within the block are rolled back."
677
    This is unfortunate because "A block containing an EXCEPTION clause is
678
    significantly more expensive to enter and exit than a block without one."
679 3015 aaronmk
    (http://www.postgresql.org/docs/8.3/static/plpgsql-control-structures.html\
680
#PLPGSQL-ERROR-TRAPPING)
681
    */
682 3034 aaronmk
    FOR '''+(', '.join((c.to_str(db) for c in row)))+''' IN
683
'''+select_query+'''
684
    LOOP
685 3015 aaronmk
        BEGIN
686 3019 aaronmk
            RETURN QUERY
687 3014 aaronmk
'''+mk_insert(sql_gen.Values(row).to_str(db))+'''
688 3010 aaronmk
;
689 3015 aaronmk
        EXCEPTION
690 3019 aaronmk
            WHEN unique_violation THEN NULL; -- continue to next row
691 3015 aaronmk
        END;
692 3010 aaronmk
    END LOOP;
693
END;\
694 3009 aaronmk
'''
695
    else: query = mk_insert(select_query)
696
697 2070 aaronmk
    if embeddable:
698
        # Create function
699 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
700 2189 aaronmk
        while True:
701
            try:
702 2918 aaronmk
                function = db.TempFunction(function_name)
703 2194 aaronmk
704 2189 aaronmk
                function_query = '''\
705 2698 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''()
706 3017 aaronmk
RETURNS SETOF '''+return_type+'''
707 3009 aaronmk
LANGUAGE '''+lang+'''
708 2467 aaronmk
AS $$
709 3009 aaronmk
'''+query+'''
710 2467 aaronmk
$$;
711 2070 aaronmk
'''
712 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
713 2945 aaronmk
                    log_ignore_excs=(DuplicateException,))
714 2189 aaronmk
                break # this version was successful
715 2945 aaronmk
            except DuplicateException, e:
716 2189 aaronmk
                function_name = next_version(function_name)
717
                # try again with next version of name
718 2070 aaronmk
719 2337 aaronmk
        # Return query that uses function
720 3009 aaronmk
        cols = None
721
        if returning != None: cols = [returning]
722 2698 aaronmk
        func_table = sql_gen.NamedTable('f', sql_gen.FunctionCall(function),
723 3009 aaronmk
            cols) # AS clause requires function alias
724 2787 aaronmk
        return mk_select(db, func_table, start=0, order_by=None)
725 2070 aaronmk
726 2787 aaronmk
    return query
727 2066 aaronmk
728
def insert_select(db, *args, **kw_args):
729 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
730 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
731
        values in
732 2072 aaronmk
    '''
733 2386 aaronmk
    into = kw_args.pop('into', None)
734
    if into != None: kw_args['embeddable'] = True
735 2066 aaronmk
    recover = kw_args.pop('recover', None)
736 3011 aaronmk
    if kw_args.get('ignore', False): recover = True
737 2066 aaronmk
    cacheable = kw_args.pop('cacheable', True)
738 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
739 2066 aaronmk
740 2789 aaronmk
    return run_query_into(db, mk_insert_select(db, *args, **kw_args), into,
741 2787 aaronmk
        recover=recover, cacheable=cacheable, log_level=log_level)
742 2063 aaronmk
743 2738 aaronmk
default = sql_gen.default # tells insert() to use the default value for a column
744 2066 aaronmk
745 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
746 2085 aaronmk
    '''For params, see insert_select()'''
747 1960 aaronmk
    if lists.is_seq(row): cols = None
748
    else:
749
        cols = row.keys()
750
        row = row.values()
751 2738 aaronmk
    row = list(row) # ensure that "== []" works
752 1960 aaronmk
753 2738 aaronmk
    if row == []: query = None
754
    else: query = sql_gen.Values(row).to_str(db)
755 1961 aaronmk
756 2788 aaronmk
    return insert_select(db, table, cols, query, *args, **kw_args)
757 11 aaronmk
758 2402 aaronmk
def mk_update(db, table, changes=None, cond=None):
759
    '''
760
    @param changes [(col, new_value),...]
761
        * container can be any iterable type
762
        * col: sql_gen.Code|str (for col name)
763
        * new_value: sql_gen.Code|literal value
764
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
765
    @return str query
766
    '''
767
    query = 'UPDATE '+sql_gen.as_Table(table).to_str(db)+'\nSET\n'
768 2405 aaronmk
    query += ',\n'.join((sql_gen.to_name_only_col(col, table).to_str(db)+' = '
769 2402 aaronmk
        +sql_gen.as_Value(new_value).to_str(db) for col, new_value in changes))
770 2467 aaronmk
    if cond != None: query += '\nWHERE\n'+cond.to_str(db)
771 2402 aaronmk
772
    return query
773
774
def update(db, *args, **kw_args):
775
    '''For params, see mk_update() and run_query()'''
776
    recover = kw_args.pop('recover', None)
777 3043 aaronmk
    cacheable = kw_args.pop('cacheable', False)
778 3030 aaronmk
    log_level = kw_args.pop('log_level', 2)
779 2402 aaronmk
780 3043 aaronmk
    return run_query(db, mk_update(db, *args, **kw_args), recover, cacheable,
781 3030 aaronmk
        log_level=log_level)
782 2402 aaronmk
783 135 aaronmk
def last_insert_id(db):
784 1849 aaronmk
    module = util.root_module(db.db)
785 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
786
    elif module == 'MySQLdb': return db.insert_id()
787
    else: return None
788 13 aaronmk
789 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
790 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
791 2415 aaronmk
    to names that will be distinct among the columns' tables.
792 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
793 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
794
    @param into The table for the new columns.
795 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
796
        columns will be included in the mapping even if they are not in cols.
797
        The tables of the provided Col objects will be changed to into, so make
798
        copies of them if you want to keep the original tables.
799
    @param as_items Whether to return a list of dict items instead of a dict
800 2383 aaronmk
    @return dict(orig_col=new_col, ...)
801
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
802 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
803
        * All mappings use the into table so its name can easily be
804 2383 aaronmk
          changed for all columns at once
805
    '''
806 2415 aaronmk
    cols = lists.uniqify(cols)
807
808 2394 aaronmk
    items = []
809 2389 aaronmk
    for col in preserve:
810 2390 aaronmk
        orig_col = copy.copy(col)
811 2392 aaronmk
        col.table = into
812 2394 aaronmk
        items.append((orig_col, col))
813
    preserve = set(preserve)
814
    for col in cols:
815 2716 aaronmk
        if col not in preserve:
816
            items.append((col, sql_gen.Col(str(col), into, col.srcs)))
817 2394 aaronmk
818
    if not as_items: items = dict(items)
819
    return items
820 2383 aaronmk
821 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
822 2391 aaronmk
    '''For params, see mk_flatten_mapping()
823
    @return See return value of mk_flatten_mapping()
824
    '''
825 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
826
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
827 2786 aaronmk
    run_query_into(db, mk_select(db, joins, cols, limit=limit, start=start),
828 2846 aaronmk
        into=into, add_indexes_=True)
829 2394 aaronmk
    return dict(items)
830 2391 aaronmk
831 3050 aaronmk
def track_data_error(db, errors_table, cols, value, error_code, error):
832 2758 aaronmk
    '''
833
    @param errors_table If None, does nothing.
834
    '''
835 2810 aaronmk
    if errors_table == None or cols == (): return
836 3050 aaronmk
837
    for col in cols:
838
        try:
839
            insert(db, errors_table, dict(column=col.name, value=value,
840
                error_code=error_code, error=error), recover=True,
841
                cacheable=True, log_level=4)
842
        except DuplicateKeyException: pass
843 2758 aaronmk
844 2704 aaronmk
def cast(db, type_, col, errors_table=None):
845
    '''Casts an (unrenamed) column or value.
846 2769 aaronmk
    If errors_table set and col has srcs, saves errors in errors_table (using
847
    col's srcs attr as the source columns) and converts errors to warnings.
848 2820 aaronmk
    @param col str|sql_gen.Col|sql_gen.Literal
849 2704 aaronmk
    @param errors_table None|sql_gen.Table|str
850
    '''
851 2820 aaronmk
    col = sql_gen.as_Col(col)
852 2769 aaronmk
    save_errors = (errors_table != None and isinstance(col, sql_gen.Col)
853
        and col.srcs != ())
854 2987 aaronmk
    if not save_errors: return sql_gen.Cast(type_, col) # can't save errors
855 2704 aaronmk
856
    assert not isinstance(col, sql_gen.NamedCol)
857
858 2770 aaronmk
    errors_table = sql_gen.as_Table(errors_table)
859
    srcs = map(sql_gen.to_name_only_col, col.srcs)
860
    function_name = str(sql_gen.FunctionCall(type_, *srcs))
861 2918 aaronmk
    function = db.TempFunction(function_name)
862 2704 aaronmk
863 2718 aaronmk
    while True:
864
        # Create function definition
865 3049 aaronmk
        errors_table_cols = map(sql_gen.Col,
866
            ['column', 'value', 'error_code', 'error'])
867 2718 aaronmk
        query = '''\
868 2704 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''(value text)
869
RETURNS '''+type_+'''
870
LANGUAGE plpgsql
871
STRICT
872
AS $$
873
BEGIN
874
    /* The explicit cast to the return type is needed to make the cast happen
875
    inside the try block. (Implicit casts to the return type happen at the end
876
    of the function, outside any block.) */
877
    RETURN value::'''+type_+''';
878
EXCEPTION
879
    WHEN data_exception THEN
880
        -- Save error in errors table.
881 3049 aaronmk
        DECLARE
882
            error_code text := SQLSTATE;
883
            error text := SQLERRM;
884
            "column" text;
885
        BEGIN
886
            -- Insert the value and error for *each* source column.
887
            FOR "column" IN
888
'''+mk_select(db, sql_gen.NamedValues('c', None, [[c.name] for c in srcs]),
889 3053 aaronmk
    order_by=None, start=0)+'''
890 3049 aaronmk
            LOOP
891
                BEGIN
892
'''+mk_insert_select(db, errors_table, errors_table_cols,
893
    sql_gen.Values(errors_table_cols).to_str(db))+''';
894
                EXCEPTION
895
                    WHEN unique_violation THEN NULL; -- continue to next row
896
                END;
897
            END LOOP;
898
        END;
899 2704 aaronmk
900
        RAISE WARNING '%', SQLERRM;
901
        RETURN NULL;
902
END;
903
$$;
904
'''
905 2718 aaronmk
906
        # Create function
907
        try:
908
            run_query(db, query, recover=True, cacheable=True,
909 2945 aaronmk
                log_ignore_excs=(DuplicateException,))
910 2718 aaronmk
            break # successful
911 2945 aaronmk
        except DuplicateException:
912 2770 aaronmk
            function.name = next_version(function.name)
913
            # try again with next version of name
914 2704 aaronmk
915
    return sql_gen.FunctionCall(function, col)
916
917 2414 aaronmk
##### Database structure queries
918
919 2426 aaronmk
def table_row_count(db, table, recover=None):
920 2786 aaronmk
    return value(run_query(db, mk_select(db, table, [sql_gen.row_count],
921 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
922 2426 aaronmk
923 2414 aaronmk
def table_cols(db, table, recover=None):
924
    return list(col_names(select(db, table, limit=0, order_by=None,
925 2443 aaronmk
        recover=recover, log_level=4)))
926 2414 aaronmk
927 2291 aaronmk
def pkey(db, table, recover=None):
928 832 aaronmk
    '''Assumed to be first column in table'''
929 2339 aaronmk
    return table_cols(db, table, recover)[0]
930 832 aaronmk
931 2559 aaronmk
not_null_col = 'not_null_col'
932 2340 aaronmk
933
def table_not_null_col(db, table, recover=None):
934
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
935
    if not_null_col in table_cols(db, table, recover): return not_null_col
936
    else: return pkey(db, table, recover)
937
938 853 aaronmk
def index_cols(db, table, index):
939
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
940
    automatically created. When you don't know whether something is a UNIQUE
941
    constraint or a UNIQUE index, use this function.'''
942 1909 aaronmk
    module = util.root_module(db.db)
943
    if module == 'psycopg2':
944
        return list(values(run_query(db, '''\
945 853 aaronmk
SELECT attname
946 866 aaronmk
FROM
947
(
948
        SELECT attnum, attname
949
        FROM pg_index
950
        JOIN pg_class index ON index.oid = indexrelid
951
        JOIN pg_class table_ ON table_.oid = indrelid
952
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
953
        WHERE
954 2782 aaronmk
            table_.relname = '''+db.esc_value(table)+'''
955
            AND index.relname = '''+db.esc_value(index)+'''
956 866 aaronmk
    UNION
957
        SELECT attnum, attname
958
        FROM
959
        (
960
            SELECT
961
                indrelid
962
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
963
                    AS indkey
964
            FROM pg_index
965
            JOIN pg_class index ON index.oid = indexrelid
966
            JOIN pg_class table_ ON table_.oid = indrelid
967
            WHERE
968 2782 aaronmk
                table_.relname = '''+db.esc_value(table)+'''
969
                AND index.relname = '''+db.esc_value(index)+'''
970 866 aaronmk
        ) s
971
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
972
) s
973 853 aaronmk
ORDER BY attnum
974 2782 aaronmk
'''
975
            , cacheable=True, log_level=4)))
976 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
977
        ' database')
978 853 aaronmk
979 464 aaronmk
def constraint_cols(db, table, constraint):
980 1849 aaronmk
    module = util.root_module(db.db)
981 464 aaronmk
    if module == 'psycopg2':
982
        return list(values(run_query(db, '''\
983
SELECT attname
984
FROM pg_constraint
985
JOIN pg_class ON pg_class.oid = conrelid
986
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
987
WHERE
988 2783 aaronmk
    relname = '''+db.esc_value(table)+'''
989
    AND conname = '''+db.esc_value(constraint)+'''
990 464 aaronmk
ORDER BY attnum
991 2783 aaronmk
'''
992
            )))
993 464 aaronmk
    else: raise NotImplementedError("Can't list constraint columns for "+module+
994
        ' database')
995
996 2096 aaronmk
row_num_col = '_row_num'
997
998 2998 aaronmk
def add_index(db, exprs, table=None, unique=False, ensure_not_null_=True):
999 2688 aaronmk
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
1000 2538 aaronmk
    Currently, only function calls are supported as expressions.
1001 2998 aaronmk
    @param ensure_not_null_ If set, translates NULL values to sentinel values.
1002 2847 aaronmk
        This allows indexes to be used for comparisons where NULLs are equal.
1003 2538 aaronmk
    '''
1004 2964 aaronmk
    exprs = lists.mk_seq(exprs)
1005 2538 aaronmk
1006 2688 aaronmk
    # Parse exprs
1007
    old_exprs = exprs[:]
1008
    exprs = []
1009
    cols = []
1010
    for i, expr in enumerate(old_exprs):
1011 2823 aaronmk
        expr = sql_gen.as_Col(expr, table)
1012 2688 aaronmk
1013 2823 aaronmk
        # Handle nullable columns
1014 2998 aaronmk
        if ensure_not_null_:
1015
            try: expr = ensure_not_null(db, expr)
1016 2860 aaronmk
            except KeyError: pass # unknown type, so just create plain index
1017 2823 aaronmk
1018 2688 aaronmk
        # Extract col
1019 3002 aaronmk
        expr = copy.deepcopy(expr) # don't modify input!
1020 2688 aaronmk
        if isinstance(expr, sql_gen.FunctionCall):
1021
            col = expr.args[0]
1022
            expr = sql_gen.Expr(expr)
1023
        else: col = expr
1024 2823 aaronmk
        assert isinstance(col, sql_gen.Col)
1025 2688 aaronmk
1026
        # Extract table
1027
        if table == None:
1028
            assert sql_gen.is_table_col(col)
1029
            table = col.table
1030
1031
        col.table = None
1032
1033
        exprs.append(expr)
1034
        cols.append(col)
1035 2408 aaronmk
1036 2688 aaronmk
    table = sql_gen.as_Table(table)
1037
    index = sql_gen.Table(str(sql_gen.Col(','.join(map(str, cols)), table)))
1038
1039 3005 aaronmk
    # Add index
1040
    while True:
1041
        str_ = 'CREATE'
1042
        if unique: str_ += ' UNIQUE'
1043
        str_ += ' INDEX '+index.to_str(db)+' ON '+table.to_str(db)+' ('+(
1044
            ', '.join((v.to_str(db) for v in exprs)))+')'
1045
1046
        try:
1047
            run_query(db, str_, recover=True, cacheable=True, log_level=3,
1048
                log_ignore_excs=(DuplicateException,))
1049
            break
1050
        except DuplicateException:
1051
            index.name = next_version(index.name)
1052
            # try again with next version of name
1053 2408 aaronmk
1054 2686 aaronmk
def add_pkey(db, table, cols=None, recover=None):
1055
    '''Adds a primary key.
1056
    @param cols [sql_gen.Col,...] The columns in the primary key.
1057
        Defaults to the first column in the table.
1058 2406 aaronmk
    @pre The table must not already have a primary key.
1059
    '''
1060
    table = sql_gen.as_Table(table)
1061 2686 aaronmk
    if cols == None: cols = [pkey(db, table, recover)]
1062
    col_strs = [sql_gen.to_name_only_col(v).to_str(db) for v in cols]
1063 2406 aaronmk
1064 2781 aaronmk
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD PRIMARY KEY ('
1065
        +(', '.join(col_strs))+')', recover=True, cacheable=True, log_level=3,
1066 2945 aaronmk
        log_ignore_excs=(DuplicateException,))
1067 2406 aaronmk
1068 2997 aaronmk
def add_not_null(db, col):
1069
    table = col.table
1070
    col = sql_gen.to_name_only_col(col)
1071
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ALTER COLUMN '
1072 3031 aaronmk
        +col.to_str(db)+' SET NOT NULL', cacheable=True, log_level=3)
1073 2997 aaronmk
1074
def add_index_col(db, col, suffix, expr, nullable=True):
1075 3000 aaronmk
    if sql_gen.index_col(col) != None: return # already has index col
1076 2997 aaronmk
1077
    new_col = sql_gen.suffixed_col(col, suffix)
1078
1079 3006 aaronmk
    # Add column
1080 3038 aaronmk
    new_typed_col = sql_gen.TypedCol(new_col.name, db.col_info(col).type)
1081 3045 aaronmk
    add_col(db, col.table, new_typed_col, comment='src: '+repr(col),
1082
        log_level=3)
1083 3037 aaronmk
    new_col.name = new_typed_col.name # propagate any renaming
1084 3006 aaronmk
1085 3044 aaronmk
    update(db, col.table, [(new_col, expr)], cacheable=True, log_level=3)
1086 2997 aaronmk
    if not nullable: add_not_null(db, new_col)
1087
    add_index(db, new_col)
1088
1089 3000 aaronmk
    col.table.index_cols[col.name] = new_col
1090 2997 aaronmk
1091 3047 aaronmk
# Controls when ensure_not_null() will use index columns
1092
not_null_index_cols_min_rows = 0 # rows; initially always use index columns
1093
1094 2997 aaronmk
def ensure_not_null(db, col):
1095
    '''For params, see sql_gen.ensure_not_null()'''
1096
    expr = sql_gen.ensure_not_null(db, col)
1097
1098 3047 aaronmk
    # If a nullable column in a temp table, add separate index column instead.
1099
    # Note that for small datasources, this adds 6-25% to the total import time.
1100
    if (sql_gen.is_temp_col(col) and isinstance(expr, sql_gen.EnsureNotNull)
1101
        and table_row_count(db, col.table) >= not_null_index_cols_min_rows):
1102 2997 aaronmk
        add_index_col(db, col, '::NOT NULL', expr, nullable=False)
1103 3000 aaronmk
        expr = sql_gen.index_col(col)
1104 2997 aaronmk
1105
    return expr
1106
1107 2772 aaronmk
already_indexed = object() # tells add_indexes() the pkey has already been added
1108
1109
def add_indexes(db, table, has_pkey=True):
1110
    '''Adds an index on all columns in a table.
1111
    @param has_pkey bool|already_indexed Whether a pkey instead of a regular
1112
        index should be added on the first column.
1113
        * If already_indexed, the pkey is assumed to have already been added
1114
    '''
1115
    cols = table_cols(db, table)
1116
    if has_pkey:
1117
        if has_pkey is not already_indexed: add_pkey(db, table)
1118
        cols = cols[1:]
1119
    for col in cols: add_index(db, col, table)
1120
1121 3042 aaronmk
def add_col(db, table, col, comment=None, **kw_args):
1122 3039 aaronmk
    '''
1123 3042 aaronmk
    @param col TypedCol Name may be versioned, so be sure to propagate any
1124
        renaming back to any source column for the TypedCol.
1125
    @param comment None|str SQL comment used to distinguish columns of the same
1126
        name from each other when they contain different data, to allow the
1127
        ADD COLUMN query to be cached. If not set, query will not be cached.
1128 3039 aaronmk
    '''
1129 2934 aaronmk
    assert isinstance(col, sql_gen.TypedCol)
1130 3036 aaronmk
1131
    while True:
1132 3042 aaronmk
        str_ = 'ALTER TABLE '+table.to_str(db)+' ADD COLUMN '+col.to_str(db)
1133
        if comment != None: str_ += ' '+sql_gen.esc_comment(comment)
1134
1135 3036 aaronmk
        try:
1136 3042 aaronmk
            run_query(db, str_, recover=True, cacheable=True, **kw_args)
1137 3036 aaronmk
            break
1138
        except DuplicateException:
1139
            col.name = next_version(col.name)
1140
            # try again with next version of name
1141 2934 aaronmk
1142 2935 aaronmk
row_num_typed_col = sql_gen.TypedCol(row_num_col, 'serial', nullable=False,
1143
    constraints='PRIMARY KEY')
1144
1145 2086 aaronmk
def add_row_num(db, table):
1146 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
1147
    be the primary key.'''
1148 2935 aaronmk
    add_col(db, table, row_num_typed_col, log_level=3)
1149 2086 aaronmk
1150 2936 aaronmk
def cast_temp_col(db, type_, col, errors_table=None):
1151
    '''Like cast(), but creates a new column with the cast values if the input
1152
    is a column.
1153
    @return The new column or cast value
1154
    '''
1155
    def cast_(col): return cast(db, type_, col, errors_table)
1156
1157
    try: col = sql_gen.underlying_col(col)
1158
    except sql_gen.NoUnderlyingTableException: return sql_gen.wrap(cast_, col)
1159
1160
    table = col.table
1161
    new_col = sql_gen.Col(sql_gen.concat(col.name, '::'+type_), table, col.srcs)
1162
    expr = cast_(col)
1163 3035 aaronmk
1164
    # Add column
1165 3038 aaronmk
    new_typed_col = sql_gen.TypedCol(new_col.name, type_)
1166 3045 aaronmk
    add_col(db, table, new_typed_col, comment='src: '+repr(col))
1167 3037 aaronmk
    new_col.name = new_typed_col.name # propagate any renaming
1168 3035 aaronmk
1169 3044 aaronmk
    update(db, table, [(new_col, expr)], cacheable=True)
1170 2955 aaronmk
    add_index(db, new_col)
1171 2936 aaronmk
1172
    return new_col
1173
1174 2889 aaronmk
def drop_table(db, table):
1175
    table = sql_gen.as_Table(table)
1176
    return run_query(db, 'DROP TABLE IF EXISTS '+table.to_str(db)+' CASCADE')
1177
1178 2760 aaronmk
def create_table(db, table, cols, has_pkey=True, col_indexes=True):
1179 2675 aaronmk
    '''Creates a table.
1180 2681 aaronmk
    @param cols [sql_gen.TypedCol,...] The column names and types
1181
    @param has_pkey If set, the first column becomes the primary key.
1182 2760 aaronmk
    @param col_indexes bool|[ref]
1183
        * If True, indexes will be added on all non-pkey columns.
1184
        * If a list reference, [0] will be set to a function to do this.
1185
          This can be used to delay index creation until the table is populated.
1186 2675 aaronmk
    '''
1187
    table = sql_gen.as_Table(table)
1188
1189 2681 aaronmk
    if has_pkey:
1190
        cols[0] = pkey = copy.copy(cols[0]) # don't modify input!
1191 2872 aaronmk
        pkey.constraints = 'PRIMARY KEY'
1192 2681 aaronmk
1193 2675 aaronmk
    str_ = 'CREATE TABLE '+table.to_str(db)+' (\n'
1194 2681 aaronmk
    str_ += '\n, '.join(v.to_str(db) for v in cols)
1195
    str_ += '\n);\n'
1196 2675 aaronmk
    run_query(db, str_, cacheable=True, log_level=2)
1197 2760 aaronmk
1198
    # Add indexes
1199 2773 aaronmk
    if has_pkey: has_pkey = already_indexed
1200
    def add_indexes_(): add_indexes(db, table, has_pkey)
1201
    if isinstance(col_indexes, list): col_indexes[0] = add_indexes_ # defer
1202
    elif col_indexes: add_indexes_() # add now
1203 2675 aaronmk
1204 2684 aaronmk
def vacuum(db, table):
1205
    table = sql_gen.as_Table(table)
1206
    db.with_autocommit(lambda: run_query(db, 'VACUUM ANALYZE '+table.to_str(db),
1207
        log_level=3))
1208
1209 2970 aaronmk
def truncate(db, table, schema='public', **kw_args):
1210
    '''For params, see run_query()'''
1211 2777 aaronmk
    table = sql_gen.as_Table(table, schema)
1212 2970 aaronmk
    return run_query(db, 'TRUNCATE '+table.to_str(db)+' CASCADE', **kw_args)
1213 2732 aaronmk
1214 2965 aaronmk
def empty_temp(db, tables):
1215 2972 aaronmk
    if db.debug_temp: return # leave temp tables there for debugging
1216 2965 aaronmk
    tables = lists.mk_seq(tables)
1217 2971 aaronmk
    for table in tables: truncate(db, table, log_level=3)
1218 2965 aaronmk
1219 2729 aaronmk
def tables(db, schema_like='public', table_like='%', exact=False):
1220
    if exact: compare = '='
1221
    else: compare = 'LIKE'
1222
1223 1849 aaronmk
    module = util.root_module(db.db)
1224 832 aaronmk
    if module == 'psycopg2':
1225 2779 aaronmk
        conds = [('schemaname', sql_gen.CompareCond(schema_like, compare)),
1226
            ('tablename', sql_gen.CompareCond(table_like, compare))]
1227
        return values(select(db, 'pg_tables', ['tablename'], conds,
1228
            order_by='tablename', log_level=4))
1229 1968 aaronmk
    elif module == 'MySQLdb':
1230 2779 aaronmk
        return values(run_query(db, 'SHOW TABLES LIKE '+db.esc_value(table_like)
1231
            , cacheable=True, log_level=4))
1232 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
1233 830 aaronmk
1234 2726 aaronmk
def table_exists(db, table):
1235
    table = sql_gen.as_Table(table)
1236 2730 aaronmk
    return list(tables(db, table.schema, table.name, exact=True)) != []
1237 2726 aaronmk
1238 2913 aaronmk
def function_exists(db, function):
1239
    function = sql_gen.as_Function(function)
1240
1241
    info_table = sql_gen.Table('routines', 'information_schema')
1242
    conds = [('routine_name', function.name)]
1243
    schema = function.schema
1244
    if schema != None: conds.append(('routine_schema', schema))
1245 2914 aaronmk
    # Exclude trigger functions, since they cannot be called directly
1246
    conds.append(('data_type', sql_gen.CompareCond('trigger', '!=')))
1247 2913 aaronmk
1248
    return list(values(select(db, info_table, ['routine_name'], conds,
1249
        order_by='routine_schema', limit=1, log_level=4))) != []
1250
        # TODO: order_by search_path schema order
1251
1252 2726 aaronmk
def errors_table(db, table, if_exists=True):
1253
    '''
1254
    @param if_exists If set, returns None if the errors table doesn't exist
1255
    @return None|sql_gen.Table
1256
    '''
1257
    table = sql_gen.as_Table(table)
1258
    if table.srcs != (): table = table.srcs[0]
1259
1260
    errors_table = sql_gen.suffixed_table(table, '.errors')
1261
    if if_exists and not table_exists(db, errors_table): return None
1262
    return errors_table
1263
1264 833 aaronmk
##### Database management
1265
1266 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
1267
    '''For kw_args, see tables()'''
1268
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
1269 833 aaronmk
1270 832 aaronmk
##### Heuristic queries
1271
1272 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
1273 1554 aaronmk
    '''Recovers from errors.
1274 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
1275
    '''
1276 2642 aaronmk
    row = sql_gen.ColDict(db, table, row)
1277 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
1278
1279 471 aaronmk
    try:
1280 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
1281 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
1282
            row_ct_ref[0] += cur.rowcount
1283
        return value(cur)
1284 471 aaronmk
    except DuplicateKeyException, e:
1285 2653 aaronmk
        row = sql_gen.ColDict(db, table,
1286
            util.dict_subset_right_join(row, e.cols))
1287
        return value(select(db, table, [pkey_], row, recover=True))
1288 471 aaronmk
1289 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
1290 830 aaronmk
    '''Recovers from errors'''
1291 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
1292 14 aaronmk
    except StopIteration:
1293 40 aaronmk
        if not create: raise
1294 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
1295 2078 aaronmk
1296 2593 aaronmk
def is_func_result(col):
1297
    return col.table.name.find('(') >= 0 and col.name == 'result'
1298
1299 2592 aaronmk
def into_table_name(out_table, in_tables0, mapping, is_func):
1300 2615 aaronmk
    def in_col_str(in_col):
1301
        in_col = sql_gen.remove_col_rename(in_col)
1302
        if isinstance(in_col, sql_gen.Col):
1303
            table = in_col.table
1304
            if table == in_tables0:
1305
                in_col = sql_gen.to_name_only_col(in_col)
1306
            elif is_func_result(in_col): in_col = table # omit col name
1307
        return str(in_col)
1308
1309 2580 aaronmk
    str_ = str(out_table)
1310
    if is_func:
1311 2615 aaronmk
        str_ += '('
1312 2580 aaronmk
1313 2615 aaronmk
        try: value_in_col = mapping['value']
1314
        except KeyError:
1315 2654 aaronmk
            str_ += ', '.join((str(k)+'='+in_col_str(v)
1316 2615 aaronmk
                for k, v in mapping.iteritems()))
1317
        else: str_ += in_col_str(value_in_col)
1318
1319
        str_ += ')'
1320 2812 aaronmk
    else:
1321
        out_col = 'rank'
1322
        try: in_col = mapping[out_col]
1323
        except KeyError: str_ += '_pkeys'
1324
        else: # has a rank column, so hierarchical
1325
            str_ += '['+str(out_col)+'='+in_col_str(in_col)+']'
1326 2580 aaronmk
    return str_
1327
1328 2508 aaronmk
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None, into=None,
1329 2805 aaronmk
    default=None, is_func=False, on_error=exc.raise_):
1330 2078 aaronmk
    '''Recovers from errors.
1331
    Only works under PostgreSQL (uses INSERT RETURNING).
1332 2909 aaronmk
    IMPORTANT: Must be run at the *beginning* of a transaction.
1333 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
1334
        tables to join with it using the main input table's pkey
1335 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
1336 2612 aaronmk
        * out_table_col: str (*not* sql_gen.Col)
1337
        * in_table_col: sql_gen.Col|literal-value
1338 2489 aaronmk
    @param into The table to contain the output and input pkeys.
1339 2574 aaronmk
        Defaults to `out_table.name+'_pkeys'`.
1340 2509 aaronmk
    @param default The *output* column to use as the pkey for missing rows.
1341
        If this output column does not exist in the mapping, uses None.
1342 2552 aaronmk
    @param is_func Whether out_table is the name of a SQL function, not a table
1343 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
1344 2078 aaronmk
    '''
1345 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
1346 2312 aaronmk
1347 2450 aaronmk
    def log_debug(msg): db.log_debug(msg, level=1.5)
1348 2505 aaronmk
    def col_ustr(str_):
1349 2567 aaronmk
        return strings.repr_no_u(sql_gen.remove_col_rename(str_))
1350 2450 aaronmk
1351 2910 aaronmk
    log_debug('********** New iteration **********')
1352
    log_debug('Inserting these input columns into '+strings.as_tt(
1353
        out_table.to_str(db))+':\n'+strings.as_table(mapping, ustr=col_ustr))
1354
1355 2942 aaronmk
    is_function = function_exists(db, out_table)
1356
1357
    if is_function: out_pkey = 'result'
1358
    else: out_pkey = pkey(db, out_table, recover=True)
1359 2768 aaronmk
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
1360
1361
    if mapping == {}: # need at least one column for INSERT SELECT
1362
        mapping = {out_pkey: None} # ColDict will replace with default value
1363
1364 2382 aaronmk
    # Create input joins from list of input tables
1365
    in_tables_ = in_tables[:] # don't modify input!
1366
    in_tables0 = in_tables_.pop(0) # first table is separate
1367 2735 aaronmk
    errors_table_ = errors_table(db, in_tables0)
1368 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
1369 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
1370 2460 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v,
1371
        {in_pkey: sql_gen.join_same_not_null}) for v in in_tables_]
1372 2131 aaronmk
1373 2592 aaronmk
    if into == None:
1374
        into = into_table_name(out_table, in_tables0, mapping, is_func)
1375
    into = sql_gen.as_Table(into)
1376
1377 2702 aaronmk
    # Set column sources
1378
    in_cols = filter(sql_gen.is_table_col, mapping.values())
1379
    for col in in_cols:
1380
        if col.table == in_tables0: col.set_srcs(sql_gen.src_self)
1381
1382 2486 aaronmk
    log_debug('Joining together input tables into temp table')
1383 2395 aaronmk
    # Place in new table for speed and so don't modify input if values edited
1384 2584 aaronmk
    in_table = sql_gen.Table('in')
1385 2702 aaronmk
    mapping = dicts.join(mapping, flatten(db, in_table, input_joins, in_cols,
1386
        preserve=[in_pkey_col], start=0))
1387 2395 aaronmk
    input_joins = [in_table]
1388 2486 aaronmk
    db.log_debug('Temp table: '+strings.as_tt(in_table.to_str(db)), level=2)
1389 2395 aaronmk
1390 2656 aaronmk
    mapping = sql_gen.ColDict(db, out_table, mapping)
1391
        # after applying dicts.join() because that returns a plain dict
1392
1393 2509 aaronmk
    # Resolve default value column
1394 2841 aaronmk
    if default != None:
1395
        try: default = mapping[default]
1396
        except KeyError:
1397 2509 aaronmk
            db.log_debug('Default value column '
1398
                +strings.as_tt(strings.repr_no_u(default))
1399 2511 aaronmk
                +' does not exist in mapping, falling back to None', level=2.1)
1400 2509 aaronmk
            default = None
1401
1402 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
1403 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
1404
1405 2201 aaronmk
    pkeys_table_exists_ref = [False]
1406 2960 aaronmk
    def insert_into_pkeys(joins, cols, distinct=False):
1407
        kw_args = {}
1408
        if distinct: kw_args.update(dict(distinct_on=[in_pkey_col]))
1409
        query = mk_select(db, joins, cols, order_by=None, start=0, **kw_args)
1410
1411 2201 aaronmk
        if pkeys_table_exists_ref[0]:
1412 2786 aaronmk
            insert_select(db, into, pkeys_names, query)
1413 2201 aaronmk
        else:
1414 2786 aaronmk
            run_query_into(db, query, into=into)
1415 2201 aaronmk
            pkeys_table_exists_ref[0] = True
1416
1417 2429 aaronmk
    limit_ref = [None]
1418 2380 aaronmk
    conds = set()
1419 2811 aaronmk
    distinct_on = sql_gen.ColDict(db, out_table)
1420 2325 aaronmk
    def mk_main_select(joins, cols):
1421 2811 aaronmk
        distinct_on_cols = [c.to_Col() for c in distinct_on.values()]
1422
        return mk_select(db, joins, cols, conds, distinct_on_cols,
1423 2429 aaronmk
            limit=limit_ref[0], start=0)
1424 2132 aaronmk
1425 2519 aaronmk
    exc_strs = set()
1426 2309 aaronmk
    def log_exc(e):
1427 2519 aaronmk
        e_str = exc.str_(e, first_line_only=True)
1428
        log_debug('Caught exception: '+e_str)
1429
        assert e_str not in exc_strs # avoid infinite loops
1430
        exc_strs.add(e_str)
1431 2735 aaronmk
1432 2451 aaronmk
    def remove_all_rows():
1433 2896 aaronmk
        log_debug('Ignoring all rows')
1434 2429 aaronmk
        limit_ref[0] = 0 # just create an empty pkeys table
1435 2735 aaronmk
1436 2758 aaronmk
    def ignore(in_col, value, e):
1437
        track_data_error(db, errors_table_, in_col.srcs, value, e.cause.pgcode,
1438
            e.cause.pgerror)
1439 2895 aaronmk
        log_debug('Ignoring rows with '+strings.as_tt(repr(in_col))+' = '
1440 2545 aaronmk
            +strings.as_tt(repr(value)))
1441 2895 aaronmk
1442 2758 aaronmk
    def remove_rows(in_col, value, e):
1443
        ignore(in_col, value, e)
1444 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
1445
        assert cond not in conds # avoid infinite loops
1446 2380 aaronmk
        conds.add(cond)
1447 2895 aaronmk
1448 2758 aaronmk
    def invalid2null(in_col, value, e):
1449
        ignore(in_col, value, e)
1450 2403 aaronmk
        update(db, in_table, [(in_col, None)],
1451
            sql_gen.ColValueCond(in_col, value))
1452 2245 aaronmk
1453 2589 aaronmk
    def insert_pkeys_table(which):
1454 2932 aaronmk
        return sql_gen.Table(sql_gen.concat(in_table.name,
1455 2589 aaronmk
            '_insert_'+which+'_pkeys'))
1456
    insert_out_pkeys = insert_pkeys_table('out')
1457
    insert_in_pkeys = insert_pkeys_table('in')
1458
1459 2206 aaronmk
    # Do inserts and selects
1460 2642 aaronmk
    join_cols = sql_gen.ColDict(db, out_table)
1461 2206 aaronmk
    while True:
1462 2521 aaronmk
        if limit_ref[0] == 0: # special case
1463
            log_debug('Creating an empty pkeys table')
1464 2786 aaronmk
            cur = run_query_into(db, mk_select(db, out_table, [out_pkey],
1465 2521 aaronmk
                limit=limit_ref[0]), into=insert_out_pkeys)
1466
            break # don't do main case
1467
1468 2303 aaronmk
        has_joins = join_cols != {}
1469
1470 2903 aaronmk
        log_debug('Trying to insert new rows')
1471
1472 2305 aaronmk
        # Prepare to insert new rows
1473 2325 aaronmk
        insert_joins = input_joins[:] # don't modify original!
1474 2403 aaronmk
        insert_args = dict(recover=True, cacheable=False)
1475 2303 aaronmk
        if has_joins:
1476 3021 aaronmk
            insert_args.update(dict(ignore=True))
1477 2325 aaronmk
        else:
1478 2404 aaronmk
            insert_args.update(dict(returning=out_pkey, into=insert_out_pkeys))
1479 2786 aaronmk
        main_select = mk_main_select(insert_joins, mapping.values())
1480 2303 aaronmk
1481 2929 aaronmk
        def main_insert():
1482 2942 aaronmk
            if is_function:
1483
                log_debug('Calling function on input rows')
1484
                args = dict(((k.name, v) for k, v in mapping.iteritems()))
1485
                func_call = sql_gen.NamedCol(out_pkey,
1486
                    sql_gen.FunctionCall(out_table, **args))
1487
                insert_into_pkeys(input_joins, [in_pkey_col, func_call])
1488
                return None
1489
            else:
1490
                return insert_select(db, out_table, mapping.keys(), main_select,
1491
                    **insert_args)
1492 2929 aaronmk
1493 2206 aaronmk
        try:
1494 2929 aaronmk
            cur = with_savepoint(db, main_insert)
1495 2357 aaronmk
            break # insert successful
1496 2931 aaronmk
        except MissingCastException, e:
1497
            log_exc(e)
1498
1499
            out_col = e.col
1500
            type_ = e.type
1501
1502
            log_debug('Casting '+strings.as_tt(out_col)+' input to '
1503
                +strings.as_tt(type_))
1504 2937 aaronmk
            mapping[out_col] = cast_temp_col(db, type_, mapping[out_col],
1505
                errors_table_)
1506 2206 aaronmk
        except DuplicateKeyException, e:
1507 2309 aaronmk
            log_exc(e)
1508
1509 2258 aaronmk
            old_join_cols = join_cols.copy()
1510 2811 aaronmk
            distinct_on.update(util.dict_subset(mapping, e.cols))
1511 2565 aaronmk
            join_cols.update(util.dict_subset_right_join(mapping, e.cols))
1512 2486 aaronmk
            log_debug('Ignoring existing rows, comparing on these columns:\n'
1513 2505 aaronmk
                +strings.as_inline_table(join_cols, ustr=col_ustr))
1514 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
1515 2230 aaronmk
        except NullValueException, e:
1516 2309 aaronmk
            log_exc(e)
1517
1518 2230 aaronmk
            out_col, = e.cols
1519
            try: in_col = mapping[out_col]
1520 2356 aaronmk
            except KeyError:
1521 2486 aaronmk
                log_debug('Missing mapping for NOT NULL column '+out_col)
1522 2451 aaronmk
                remove_all_rows()
1523 2758 aaronmk
            else: remove_rows(in_col, None, e)
1524 2542 aaronmk
        except FunctionValueException, e:
1525
            log_exc(e)
1526
1527
            func_name = e.name
1528
            value = e.value
1529
            for out_col, in_col in mapping.iteritems():
1530 2758 aaronmk
                in_col = sql_gen.unwrap_func_call(in_col, func_name)
1531
                invalid2null(in_col, value, e)
1532 2429 aaronmk
        except DatabaseErrors, e:
1533
            log_exc(e)
1534
1535 2808 aaronmk
            log_debug('No handler for exception')
1536
            on_error(e)
1537 2451 aaronmk
            remove_all_rows()
1538 2358 aaronmk
        # after exception handled, rerun loop with additional constraints
1539 2132 aaronmk
1540 2942 aaronmk
    if cur != None and row_ct_ref != None and cur.rowcount >= 0:
1541 2357 aaronmk
        row_ct_ref[0] += cur.rowcount
1542
1543 2942 aaronmk
    if is_function: pass # pkeys table already created
1544
    elif has_joins:
1545 2357 aaronmk
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
1546 2486 aaronmk
        log_debug('Getting output table pkeys of existing/inserted rows')
1547 2960 aaronmk
        insert_into_pkeys(select_joins, pkeys_cols, distinct=True)
1548 2357 aaronmk
    else:
1549 2404 aaronmk
        add_row_num(db, insert_out_pkeys) # for joining with input pkeys
1550 2357 aaronmk
1551 2486 aaronmk
        log_debug('Getting input table pkeys of inserted rows')
1552 2786 aaronmk
        run_query_into(db, mk_main_select(input_joins, [in_pkey]),
1553 2404 aaronmk
            into=insert_in_pkeys)
1554
        add_row_num(db, insert_in_pkeys) # for joining with output pkeys
1555 2357 aaronmk
1556 2428 aaronmk
        assert table_row_count(db, insert_out_pkeys) == table_row_count(db,
1557
            insert_in_pkeys)
1558
1559 2486 aaronmk
        log_debug('Combining output and input pkeys in inserted order')
1560 2404 aaronmk
        pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
1561 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
1562 2420 aaronmk
        insert_into_pkeys(pkey_joins, pkeys_names)
1563 2959 aaronmk
1564 2974 aaronmk
        empty_temp(db, [insert_out_pkeys, insert_in_pkeys])
1565 2357 aaronmk
1566 2486 aaronmk
    db.log_debug('Adding pkey on pkeys table to enable fast joins', level=2.5)
1567 2594 aaronmk
    add_pkey(db, into)
1568 2407 aaronmk
1569 2508 aaronmk
    log_debug('Setting pkeys of missing rows to '+strings.as_tt(repr(default)))
1570 2489 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(into,
1571 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
1572
        # must use join_same_not_null or query will take forever
1573 2420 aaronmk
    insert_into_pkeys(missing_rows_joins,
1574 2508 aaronmk
        [in_pkey_col, sql_gen.NamedCol(out_pkey, default)])
1575 2357 aaronmk
1576 2489 aaronmk
    assert table_row_count(db, into) == table_row_count(db, in_table)
1577 2428 aaronmk
1578 2974 aaronmk
    empty_temp(db, in_table)
1579 2959 aaronmk
1580 2702 aaronmk
    srcs = []
1581
    if is_func: srcs = sql_gen.cols_srcs(in_cols)
1582
    return sql_gen.Col(out_pkey, into, srcs)
1583 2115 aaronmk
1584
##### Data cleanup
1585
1586 2290 aaronmk
def cleanup_table(db, table, cols):
1587 2677 aaronmk
    table = sql_gen.as_Table(table)
1588
    cols = map(sql_gen.as_Col, cols)
1589 2115 aaronmk
1590 2677 aaronmk
    expr = ('nullif(nullif(trim(both from %s), '+db.esc_value('')+'), '
1591
        +db.esc_value(r'\N')+')')
1592
    changes = [(v, sql_gen.CustomCode(expr % v.to_str(db)))
1593
        for v in cols]
1594 2115 aaronmk
1595 2677 aaronmk
    update(db, table, changes)