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
                if query.find('ADD COLUMN') < 0: # rest of query must be unique
242
                    self._cache_result()
243 3004 aaronmk
            elif self.rowcount == 0 and query.startswith('SELECT'): # empty
244 2800 aaronmk
                consume_rows(self) # fetch all rows so result will be cached
245 3004 aaronmk
246 2762 aaronmk
            return cur
247 1894 aaronmk
248 1891 aaronmk
        def fetchone(self):
249
            row = self.inner.fetchone()
250 1899 aaronmk
            if row != None: self.result.append(row)
251
            # otherwise, fetched all rows
252 1904 aaronmk
            else: self._cache_result()
253
            return row
254
255
        def _cache_result(self):
256 2948 aaronmk
            # For inserts that return a result set, don't cache result set since
257
            # inserts are not idempotent. Other non-SELECT queries don't have
258
            # their result set read, so only exceptions will be cached (an
259
            # invalid query will always be invalid).
260 1930 aaronmk
            if self.query_results != None and (not self._is_insert
261 1906 aaronmk
                or isinstance(self.result, Exception)):
262
263 1894 aaronmk
                assert self.query_lookup != None
264 1916 aaronmk
                self.query_results[self.query_lookup] = self.CacheCursor(
265
                    util.dict_subset(dicts.AttrsDictView(self),
266
                    ['query', 'result', 'rowcount', 'description']))
267 1906 aaronmk
268 1916 aaronmk
        class CacheCursor:
269
            def __init__(self, cached_result): self.__dict__ = cached_result
270
271 1927 aaronmk
            def execute(self, *args, **kw_args):
272 1916 aaronmk
                if isinstance(self.result, Exception): raise self.result
273
                # otherwise, result is a rows list
274
                self.iter = iter(self.result)
275
276
            def fetchone(self):
277
                try: return self.iter.next()
278
                except StopIteration: return None
279 1891 aaronmk
280 2212 aaronmk
    def esc_value(self, value):
281 2663 aaronmk
        try: str_ = self.mogrify('%s', [value])
282
        except NotImplementedError, e:
283
            module = util.root_module(self.db)
284
            if module == 'MySQLdb':
285
                import _mysql
286
                str_ = _mysql.escape_string(value)
287
            else: raise e
288 2374 aaronmk
        return strings.to_unicode(str_)
289 2212 aaronmk
290 2347 aaronmk
    def esc_name(self, name): return esc_name(self, name) # calls global func
291
292 2814 aaronmk
    def std_code(self, str_):
293
        '''Standardizes SQL code.
294
        * Ensures that string literals are prefixed by `E`
295
        '''
296
        if str_.startswith("'"): str_ = 'E'+str_
297
        return str_
298
299 2665 aaronmk
    def can_mogrify(self):
300 2663 aaronmk
        module = util.root_module(self.db)
301 2665 aaronmk
        return module == 'psycopg2'
302 2663 aaronmk
303 2665 aaronmk
    def mogrify(self, query, params=None):
304
        if self.can_mogrify(): return self.db.cursor().mogrify(query, params)
305
        else: raise NotImplementedError("Can't mogrify query")
306
307 2671 aaronmk
    def print_notices(self):
308 2725 aaronmk
        if hasattr(self.db, 'notices'):
309
            for msg in self.db.notices:
310
                if msg not in self._notices_seen:
311
                    self._notices_seen.add(msg)
312
                    self.log_debug(msg, level=2)
313 2671 aaronmk
314 2793 aaronmk
    def run_query(self, query, cacheable=False, log_level=2,
315 2464 aaronmk
        debug_msg_ref=None):
316 2445 aaronmk
        '''
317 2464 aaronmk
        @param log_ignore_excs The log_level will be increased by 2 if the query
318
            throws one of these exceptions.
319 2664 aaronmk
        @param debug_msg_ref If specified, the log message will be returned in
320
            this instead of being output. This allows you to filter log messages
321
            depending on the result of the query.
322 2445 aaronmk
        '''
323 2167 aaronmk
        assert query != None
324
325 2047 aaronmk
        if not self.caching: cacheable = False
326 1903 aaronmk
        used_cache = False
327 2664 aaronmk
328
        def log_msg(query):
329
            if used_cache: cache_status = 'cache hit'
330
            elif cacheable: cache_status = 'cache miss'
331
            else: cache_status = 'non-cacheable'
332
            return 'DB query: '+cache_status+':\n'+strings.as_code(query, 'SQL')
333
334 1903 aaronmk
        try:
335 1927 aaronmk
            # Get cursor
336
            if cacheable:
337
                try:
338 2797 aaronmk
                    cur = self.query_results[query]
339 1927 aaronmk
                    used_cache = True
340
                except KeyError: cur = self.DbCursor(self)
341
            else: cur = self.db.cursor()
342
343 2664 aaronmk
            # Log query
344
            if self.debug and debug_msg_ref == None: # log before running
345
                self.log_debug(log_msg(query), log_level)
346
347 1927 aaronmk
            # Run query
348 2793 aaronmk
            cur.execute(query)
349 1903 aaronmk
        finally:
350 2671 aaronmk
            self.print_notices()
351 2664 aaronmk
            if self.debug and debug_msg_ref != None: # return after running
352 2793 aaronmk
                debug_msg_ref[0] = log_msg(str(get_cur_query(cur, query)))
353 1903 aaronmk
354
        return cur
355 1914 aaronmk
356 2797 aaronmk
    def is_cached(self, query): return query in self.query_results
357 2139 aaronmk
358 2907 aaronmk
    def with_autocommit(self, func):
359 2801 aaronmk
        import psycopg2.extensions
360
361
        prev_isolation_level = self.db.isolation_level
362 2907 aaronmk
        self.db.set_isolation_level(
363
            psycopg2.extensions.ISOLATION_LEVEL_AUTOCOMMIT)
364 2683 aaronmk
        try: return func()
365 2801 aaronmk
        finally: self.db.set_isolation_level(prev_isolation_level)
366 2683 aaronmk
367 2139 aaronmk
    def with_savepoint(self, func):
368 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
369 2443 aaronmk
        self.run_query('SAVEPOINT '+savepoint, log_level=4)
370 2139 aaronmk
        self._savepoint += 1
371 2930 aaronmk
        try: return func()
372 2139 aaronmk
        except:
373 2443 aaronmk
            self.run_query('ROLLBACK TO SAVEPOINT '+savepoint, log_level=4)
374 2139 aaronmk
            raise
375 2930 aaronmk
        finally:
376
            # Always release savepoint, because after ROLLBACK TO SAVEPOINT,
377
            # "The savepoint remains valid and can be rolled back to again"
378
            # (http://www.postgresql.org/docs/8.3/static/sql-rollback-to.html).
379 2443 aaronmk
            self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
380 2930 aaronmk
381
            self._savepoint -= 1
382
            assert self._savepoint >= 0
383
384
            self.do_autocommit() # OK to do this after ROLLBACK TO SAVEPOINT
385 2191 aaronmk
386
    def do_autocommit(self):
387
        '''Autocommits if outside savepoint'''
388
        assert self._savepoint >= 0
389
        if self.autocommit and self._savepoint == 0:
390 2924 aaronmk
            self.log_debug('Autocommitting', level=4)
391 2191 aaronmk
            self.db.commit()
392 2643 aaronmk
393 2819 aaronmk
    def col_info(self, col):
394 2643 aaronmk
        table = sql_gen.Table('columns', 'information_schema')
395 2821 aaronmk
        cols = ['data_type', 'column_default',
396
            cast(self, 'boolean', 'is_nullable')]
397 2643 aaronmk
398
        conds = [('table_name', col.table.name), ('column_name', col.name)]
399
        schema = col.table.schema
400
        if schema != None: conds.append(('table_schema', schema))
401
402 2819 aaronmk
        type_, default, nullable = row(select(self, table, cols, conds,
403 2874 aaronmk
            order_by='table_schema', limit=1, log_level=4))
404 2643 aaronmk
            # TODO: order_by search_path schema order
405 2819 aaronmk
        default = sql_gen.as_Code(default, self)
406
407
        return sql_gen.TypedCol(col.name, type_, default, nullable)
408 2917 aaronmk
409
    def TempFunction(self, name):
410
        if self.debug_temp: schema = None
411
        else: schema = 'pg_temp'
412
        return sql_gen.Function(name, schema)
413 1849 aaronmk
414 1869 aaronmk
connect = DbConn
415
416 832 aaronmk
##### Recoverable querying
417 15 aaronmk
418 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
419 11 aaronmk
420 2791 aaronmk
def run_query(db, query, recover=None, cacheable=False, log_level=2,
421
    log_ignore_excs=None, **kw_args):
422 2794 aaronmk
    '''For params, see DbConn.run_query()'''
423 830 aaronmk
    if recover == None: recover = False
424 2464 aaronmk
    if log_ignore_excs == None: log_ignore_excs = ()
425
    log_ignore_excs = tuple(log_ignore_excs)
426 830 aaronmk
427 2666 aaronmk
    debug_msg_ref = None # usually, db.run_query() logs query before running it
428
    # But if filtering with log_ignore_excs, wait until after exception parsing
429 2984 aaronmk
    if log_ignore_excs != () or not db.can_mogrify(): debug_msg_ref = [None]
430 2666 aaronmk
431 2148 aaronmk
    try:
432 2464 aaronmk
        try:
433 2794 aaronmk
            def run(): return db.run_query(query, cacheable, log_level,
434 2793 aaronmk
                debug_msg_ref, **kw_args)
435 2796 aaronmk
            if recover and not db.is_cached(query):
436 2464 aaronmk
                return with_savepoint(db, run)
437
            else: return run() # don't need savepoint if cached
438
        except Exception, e:
439
            msg = exc.str_(e)
440
441
            match = re.search(r'duplicate key value violates unique constraint '
442 2493 aaronmk
                r'"((_?[^\W_]+)_.+?)"', msg)
443 2464 aaronmk
            if match:
444
                constraint, table = match.groups()
445 3025 aaronmk
                cols = []
446
                if recover: # need auto-rollback to run index_cols()
447
                    try: cols = index_cols(db, table, constraint)
448
                    except NotImplementedError: pass
449
                raise DuplicateKeyException(constraint, cols, e)
450 2464 aaronmk
451 2493 aaronmk
            match = re.search(r'null value in column "(.+?)" violates not-null'
452 2464 aaronmk
                r' constraint', msg)
453
            if match: raise NullValueException('NOT NULL', [match.group(1)], e)
454
455
            match = re.search(r'\b(?:invalid input (?:syntax|value)\b.*?'
456
                r'|date/time field value out of range): "(.+?)"\n'
457 2535 aaronmk
                r'(?:(?s).*?)\bfunction "(.+?)"', msg)
458 2464 aaronmk
            if match:
459
                value, name = match.groups()
460
                raise FunctionValueException(name, strings.to_unicode(value), e)
461
462 2526 aaronmk
            match = re.search(r'column "(.+?)" is of type (.+?) but expression '
463 2523 aaronmk
                r'is of type', msg)
464
            if match:
465
                col, type_ = match.groups()
466
                raise MissingCastException(type_, col, e)
467
468 2946 aaronmk
            match = re.search(r'\b(\S+) "(.+?)".*? already exists', msg)
469 2945 aaronmk
            if match:
470
                type_, name = match.groups()
471
                raise DuplicateException(type_, name, e)
472 2464 aaronmk
473
            raise # no specific exception raised
474
    except log_ignore_excs:
475
        log_level += 2
476
        raise
477
    finally:
478 2666 aaronmk
        if debug_msg_ref != None and debug_msg_ref[0] != None:
479
            db.log_debug(debug_msg_ref[0], log_level)
480 830 aaronmk
481 832 aaronmk
##### Basic queries
482
483 2153 aaronmk
def next_version(name):
484 2163 aaronmk
    version = 1 # first existing name was version 0
485 2586 aaronmk
    match = re.match(r'^(.*)#(\d+)$', name)
486 2153 aaronmk
    if match:
487 2586 aaronmk
        name, version = match.groups()
488
        version = int(version)+1
489 2932 aaronmk
    return sql_gen.concat(name, '#'+str(version))
490 2153 aaronmk
491 2899 aaronmk
def lock_table(db, table, mode):
492
    table = sql_gen.as_Table(table)
493
    run_query(db, 'LOCK TABLE '+table.to_str(db)+' IN '+mode+' MODE')
494
495 2789 aaronmk
def run_query_into(db, query, into=None, add_indexes_=False, **kw_args):
496 2085 aaronmk
    '''Outputs a query to a temp table.
497
    For params, see run_query().
498
    '''
499 2789 aaronmk
    if into == None: return run_query(db, query, **kw_args)
500 2790 aaronmk
501
    assert isinstance(into, sql_gen.Table)
502
503 2992 aaronmk
    into.is_temp = True
504 3008 aaronmk
    # "temporary tables cannot specify a schema name", so remove schema
505
    into.schema = None
506 2992 aaronmk
507 2790 aaronmk
    kw_args['recover'] = True
508 2945 aaronmk
    kw_args.setdefault('log_ignore_excs', (DuplicateException,))
509 2790 aaronmk
510 2916 aaronmk
    temp = not db.debug_temp # tables are permanent in debug_temp mode
511 2790 aaronmk
512
    # Create table
513
    while True:
514
        create_query = 'CREATE'
515
        if temp: create_query += ' TEMP'
516
        create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
517 2385 aaronmk
518 2790 aaronmk
        try:
519
            cur = run_query(db, create_query, **kw_args)
520
                # CREATE TABLE AS sets rowcount to # rows in query
521
            break
522 2945 aaronmk
        except DuplicateException, e:
523 2790 aaronmk
            into.name = next_version(into.name)
524
            # try again with next version of name
525
526
    if add_indexes_: add_indexes(db, into)
527
528
    return cur
529 2085 aaronmk
530 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
531
532 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
533
534 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
535 2293 aaronmk
    start=None, order_by=order_by_pkey, default_table=None):
536 1981 aaronmk
    '''
537 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
538 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
539 1981 aaronmk
    @param fields Use None to select all fields in the table
540 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
541 2379 aaronmk
        * container can be any iterable type
542 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
543
        * compare_right_side: sql_gen.ValueCond|literal value
544 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
545
        use all columns
546 2786 aaronmk
    @return query
547 1981 aaronmk
    '''
548 2315 aaronmk
    # Parse tables param
549 2964 aaronmk
    tables = lists.mk_seq(tables)
550 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
551 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
552 2121 aaronmk
553 2315 aaronmk
    # Parse other params
554 2376 aaronmk
    if conds == None: conds = []
555 2650 aaronmk
    elif dicts.is_dict(conds): conds = conds.items()
556 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
557 135 aaronmk
    assert limit == None or type(limit) == int
558 865 aaronmk
    assert start == None or type(start) == int
559 2315 aaronmk
    if order_by is order_by_pkey:
560
        if distinct_on != []: order_by = None
561
        else: order_by = pkey(db, table0, recover=True)
562 865 aaronmk
563 2315 aaronmk
    query = 'SELECT'
564 2056 aaronmk
565 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
566 2056 aaronmk
567 2200 aaronmk
    # DISTINCT ON columns
568 2233 aaronmk
    if distinct_on != []:
569 2467 aaronmk
        query += '\nDISTINCT'
570 2254 aaronmk
        if distinct_on is not distinct_on_all:
571 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
572
573
    # Columns
574 2467 aaronmk
    query += '\n'
575 1135 aaronmk
    if fields == None: query += '*'
576 2765 aaronmk
    else:
577
        assert fields != []
578
        query += '\n, '.join(map(parse_col, fields))
579 2200 aaronmk
580
    # Main table
581 2467 aaronmk
    query += '\nFROM '+table0.to_str(db)
582 865 aaronmk
583 2122 aaronmk
    # Add joins
584 2271 aaronmk
    left_table = table0
585 2263 aaronmk
    for join_ in tables:
586
        table = join_.table
587 2238 aaronmk
588 2343 aaronmk
        # Parse special values
589
        if join_.type_ is sql_gen.filter_out: # filter no match
590 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
591 2853 aaronmk
                sql_gen.CompareCond(None, '~=')))
592 2343 aaronmk
593 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
594 2122 aaronmk
595
        left_table = table
596
597 865 aaronmk
    missing = True
598 2376 aaronmk
    if conds != []:
599 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
600
        else: whitespace = '\n'
601 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
602
            .to_str(db) for l, r in conds], 'WHERE')
603 865 aaronmk
        missing = False
604 2227 aaronmk
    if order_by != None:
605 2467 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
606
    if limit != None: query += '\nLIMIT '+str(limit); missing = False
607 865 aaronmk
    if start != None:
608 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
609 865 aaronmk
        missing = False
610
    if missing: warnings.warn(DbWarning(
611
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
612
613 2786 aaronmk
    return query
614 11 aaronmk
615 2054 aaronmk
def select(db, *args, **kw_args):
616
    '''For params, see mk_select() and run_query()'''
617
    recover = kw_args.pop('recover', None)
618
    cacheable = kw_args.pop('cacheable', True)
619 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
620 2054 aaronmk
621 2791 aaronmk
    return run_query(db, mk_select(db, *args, **kw_args), recover, cacheable,
622
        log_level=log_level)
623 2054 aaronmk
624 2788 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, returning=None,
625 3009 aaronmk
    embeddable=False, ignore=False):
626 1960 aaronmk
    '''
627
    @param returning str|None An inserted column (such as pkey) to return
628 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
629 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
630
        query will be fully cached, not just if it raises an exception.
631 3009 aaronmk
    @param ignore Whether to ignore duplicate keys.
632 1960 aaronmk
    '''
633 2754 aaronmk
    table = sql_gen.remove_table_rename(sql_gen.as_Table(table))
634 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
635 3010 aaronmk
    if cols != None: cols = [sql_gen.to_name_only_col(c, table) for c in cols]
636 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
637 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
638 2063 aaronmk
639 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
640 2063 aaronmk
641 3009 aaronmk
    def mk_insert(select_query):
642
        query = first_line
643 3014 aaronmk
        if cols != None:
644
            query += '\n('+(', '.join((c.to_str(db) for c in cols)))+')'
645 3009 aaronmk
        query += '\n'+select_query
646
647
        if returning != None:
648
            returning_name_col = sql_gen.to_name_only_col(returning)
649
            query += '\nRETURNING '+returning_name_col.to_str(db)
650
651
        return query
652 2063 aaronmk
653 3017 aaronmk
    return_type = 'unknown'
654
    if returning != None: return_type = returning.to_str(db)+'%TYPE'
655
656 3009 aaronmk
    lang = 'sql'
657
    if ignore:
658 3017 aaronmk
        assert cols != None
659
        # Always return something to set the correct rowcount
660
        if returning == None: returning = sql_gen.NamedCol('NULL', None)
661
662 3009 aaronmk
        embeddable = True # must use function
663
        lang = 'plpgsql'
664 3014 aaronmk
        row = [sql_gen.Col(c.name, 'row') for c in cols]
665 3010 aaronmk
666 3009 aaronmk
        query = '''\
667 3010 aaronmk
DECLARE
668 3014 aaronmk
    row '''+table.to_str(db)+'''%ROWTYPE;
669 3010 aaronmk
    cur CURSOR FOR
670
'''+select_query+'''
671
;
672 3009 aaronmk
BEGIN
673 3010 aaronmk
    OPEN cur;
674 3015 aaronmk
675 3019 aaronmk
    /* Need an EXCEPTION block for each individual row because "When an error is
676
    caught by an EXCEPTION clause, [...] all changes to persistent database
677
    state within the block are rolled back."
678
    This is unfortunate because "A block containing an EXCEPTION clause is
679
    significantly more expensive to enter and exit than a block without one."
680 3015 aaronmk
    (http://www.postgresql.org/docs/8.3/static/plpgsql-control-structures.html\
681
#PLPGSQL-ERROR-TRAPPING)
682
    */
683 3014 aaronmk
    WHILE true LOOP
684 3015 aaronmk
        BEGIN
685 3019 aaronmk
            FETCH FROM cur INTO \
686 3015 aaronmk
'''+(', '.join((c.to_str(db) for c in row)))+''';
687 3019 aaronmk
            EXIT WHEN NOT FOUND;
688
689
            RETURN QUERY
690 3014 aaronmk
'''+mk_insert(sql_gen.Values(row).to_str(db))+'''
691 3010 aaronmk
;
692 3015 aaronmk
        EXCEPTION
693 3019 aaronmk
            /* Note that cursor stays at current position in spite of automatic
694
            ROLLBACK to before EXCEPTION block: "If a previously opened cursor
695
            is affected by a FETCH command inside a savepoint that is later
696
            rolled back, the cursor position remains at the position that FETCH
697
            left it pointing to (that is, FETCH is not rolled back)."
698
            (http://www.postgresql.org/docs/8.3/static/sql-rollback-to.html)
699
            */
700
            WHEN unique_violation THEN NULL; -- continue to next row
701 3015 aaronmk
        END;
702 3010 aaronmk
    END LOOP;
703 3015 aaronmk
704 3010 aaronmk
    CLOSE cur;
705
END;\
706 3009 aaronmk
'''
707
    else: query = mk_insert(select_query)
708
709 2070 aaronmk
    if embeddable:
710
        # Create function
711 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
712 2189 aaronmk
        while True:
713
            try:
714 2918 aaronmk
                function = db.TempFunction(function_name)
715 2194 aaronmk
716 2189 aaronmk
                function_query = '''\
717 2698 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''()
718 3017 aaronmk
RETURNS SETOF '''+return_type+'''
719 3009 aaronmk
LANGUAGE '''+lang+'''
720 2467 aaronmk
AS $$
721 3009 aaronmk
'''+query+'''
722 2467 aaronmk
$$;
723 2070 aaronmk
'''
724 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
725 2945 aaronmk
                    log_ignore_excs=(DuplicateException,))
726 2189 aaronmk
                break # this version was successful
727 2945 aaronmk
            except DuplicateException, e:
728 2189 aaronmk
                function_name = next_version(function_name)
729
                # try again with next version of name
730 2070 aaronmk
731 2337 aaronmk
        # Return query that uses function
732 3009 aaronmk
        cols = None
733
        if returning != None: cols = [returning]
734 2698 aaronmk
        func_table = sql_gen.NamedTable('f', sql_gen.FunctionCall(function),
735 3009 aaronmk
            cols) # AS clause requires function alias
736 2787 aaronmk
        return mk_select(db, func_table, start=0, order_by=None)
737 2070 aaronmk
738 2787 aaronmk
    return query
739 2066 aaronmk
740
def insert_select(db, *args, **kw_args):
741 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
742 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
743
        values in
744 2072 aaronmk
    '''
745 2386 aaronmk
    into = kw_args.pop('into', None)
746
    if into != None: kw_args['embeddable'] = True
747 2066 aaronmk
    recover = kw_args.pop('recover', None)
748 3011 aaronmk
    if kw_args.get('ignore', False): recover = True
749 2066 aaronmk
    cacheable = kw_args.pop('cacheable', True)
750 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
751 2066 aaronmk
752 2789 aaronmk
    return run_query_into(db, mk_insert_select(db, *args, **kw_args), into,
753 2787 aaronmk
        recover=recover, cacheable=cacheable, log_level=log_level)
754 2063 aaronmk
755 2738 aaronmk
default = sql_gen.default # tells insert() to use the default value for a column
756 2066 aaronmk
757 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
758 2085 aaronmk
    '''For params, see insert_select()'''
759 1960 aaronmk
    if lists.is_seq(row): cols = None
760
    else:
761
        cols = row.keys()
762
        row = row.values()
763 2738 aaronmk
    row = list(row) # ensure that "== []" works
764 1960 aaronmk
765 2738 aaronmk
    if row == []: query = None
766
    else: query = sql_gen.Values(row).to_str(db)
767 1961 aaronmk
768 2788 aaronmk
    return insert_select(db, table, cols, query, *args, **kw_args)
769 11 aaronmk
770 2402 aaronmk
def mk_update(db, table, changes=None, cond=None):
771
    '''
772
    @param changes [(col, new_value),...]
773
        * container can be any iterable type
774
        * col: sql_gen.Code|str (for col name)
775
        * new_value: sql_gen.Code|literal value
776
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
777
    @return str query
778
    '''
779
    query = 'UPDATE '+sql_gen.as_Table(table).to_str(db)+'\nSET\n'
780 2405 aaronmk
    query += ',\n'.join((sql_gen.to_name_only_col(col, table).to_str(db)+' = '
781 2402 aaronmk
        +sql_gen.as_Value(new_value).to_str(db) for col, new_value in changes))
782 2467 aaronmk
    if cond != None: query += '\nWHERE\n'+cond.to_str(db)
783 2402 aaronmk
784
    return query
785
786
def update(db, *args, **kw_args):
787
    '''For params, see mk_update() and run_query()'''
788
    recover = kw_args.pop('recover', None)
789
790 2791 aaronmk
    return run_query(db, mk_update(db, *args, **kw_args), recover)
791 2402 aaronmk
792 135 aaronmk
def last_insert_id(db):
793 1849 aaronmk
    module = util.root_module(db.db)
794 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
795
    elif module == 'MySQLdb': return db.insert_id()
796
    else: return None
797 13 aaronmk
798 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
799 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
800 2415 aaronmk
    to names that will be distinct among the columns' tables.
801 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
802 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
803
    @param into The table for the new columns.
804 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
805
        columns will be included in the mapping even if they are not in cols.
806
        The tables of the provided Col objects will be changed to into, so make
807
        copies of them if you want to keep the original tables.
808
    @param as_items Whether to return a list of dict items instead of a dict
809 2383 aaronmk
    @return dict(orig_col=new_col, ...)
810
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
811 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
812
        * All mappings use the into table so its name can easily be
813 2383 aaronmk
          changed for all columns at once
814
    '''
815 2415 aaronmk
    cols = lists.uniqify(cols)
816
817 2394 aaronmk
    items = []
818 2389 aaronmk
    for col in preserve:
819 2390 aaronmk
        orig_col = copy.copy(col)
820 2392 aaronmk
        col.table = into
821 2394 aaronmk
        items.append((orig_col, col))
822
    preserve = set(preserve)
823
    for col in cols:
824 2716 aaronmk
        if col not in preserve:
825
            items.append((col, sql_gen.Col(str(col), into, col.srcs)))
826 2394 aaronmk
827
    if not as_items: items = dict(items)
828
    return items
829 2383 aaronmk
830 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
831 2391 aaronmk
    '''For params, see mk_flatten_mapping()
832
    @return See return value of mk_flatten_mapping()
833
    '''
834 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
835
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
836 2786 aaronmk
    run_query_into(db, mk_select(db, joins, cols, limit=limit, start=start),
837 2846 aaronmk
        into=into, add_indexes_=True)
838 2394 aaronmk
    return dict(items)
839 2391 aaronmk
840 2751 aaronmk
def mk_track_data_error(db, errors_table, cols, value, error_code, error):
841 2810 aaronmk
    assert cols != ()
842
843 2751 aaronmk
    cols = map(sql_gen.to_name_only_col, cols)
844
845
    columns_cols = ['column']
846
    columns = sql_gen.NamedValues('columns', columns_cols,
847
        [[c.name] for c in cols])
848
    values_cols = ['value', 'error_code', 'error']
849
    values = sql_gen.NamedValues('values', values_cols,
850
        [value, error_code, error])
851
852
    select_cols = columns_cols+values_cols
853
    name_only_cols = map(sql_gen.to_name_only_col, select_cols)
854 2755 aaronmk
    errors_table = sql_gen.NamedTable('errors', errors_table)
855 2751 aaronmk
    joins = [columns, sql_gen.Join(values, type_='CROSS'),
856
        sql_gen.Join(errors_table, dict(zip(name_only_cols, select_cols)),
857
        sql_gen.filter_out)]
858
859
    return mk_insert_select(db, errors_table, name_only_cols,
860 2787 aaronmk
        mk_select(db, joins, select_cols, order_by=None))
861 2751 aaronmk
862 2810 aaronmk
def track_data_error(db, errors_table, cols, *args, **kw_args):
863 2758 aaronmk
    '''
864
    @param errors_table If None, does nothing.
865
    '''
866 2810 aaronmk
    if errors_table == None or cols == (): return
867
    run_query(db, mk_track_data_error(db, errors_table, cols, *args, **kw_args),
868 2758 aaronmk
        cacheable=True, log_level=4)
869
870 2704 aaronmk
def cast(db, type_, col, errors_table=None):
871
    '''Casts an (unrenamed) column or value.
872 2769 aaronmk
    If errors_table set and col has srcs, saves errors in errors_table (using
873
    col's srcs attr as the source columns) and converts errors to warnings.
874 2820 aaronmk
    @param col str|sql_gen.Col|sql_gen.Literal
875 2704 aaronmk
    @param errors_table None|sql_gen.Table|str
876
    '''
877 2820 aaronmk
    col = sql_gen.as_Col(col)
878 2769 aaronmk
    save_errors = (errors_table != None and isinstance(col, sql_gen.Col)
879
        and col.srcs != ())
880 2987 aaronmk
    if not save_errors: return sql_gen.Cast(type_, col) # can't save errors
881 2704 aaronmk
882
    assert not isinstance(col, sql_gen.NamedCol)
883
884 2770 aaronmk
    errors_table = sql_gen.as_Table(errors_table)
885
    srcs = map(sql_gen.to_name_only_col, col.srcs)
886
    function_name = str(sql_gen.FunctionCall(type_, *srcs))
887 2918 aaronmk
    function = db.TempFunction(function_name)
888 2704 aaronmk
889 2718 aaronmk
    while True:
890
        # Create function definition
891
        query = '''\
892 2704 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''(value text)
893
RETURNS '''+type_+'''
894
LANGUAGE plpgsql
895
STRICT
896
AS $$
897
BEGIN
898
    /* The explicit cast to the return type is needed to make the cast happen
899
    inside the try block. (Implicit casts to the return type happen at the end
900
    of the function, outside any block.) */
901
    RETURN value::'''+type_+''';
902
EXCEPTION
903
    WHEN data_exception THEN
904
        -- Save error in errors table.
905 2751 aaronmk
        -- Insert the value and error for *each* source column.
906
'''+mk_track_data_error(db, errors_table, srcs,
907
    *map(sql_gen.CustomCode, ['value', 'SQLSTATE', 'SQLERRM']))+''';
908 2704 aaronmk
909
        RAISE WARNING '%', SQLERRM;
910
        RETURN NULL;
911
END;
912
$$;
913
'''
914 2718 aaronmk
915
        # Create function
916
        try:
917
            run_query(db, query, recover=True, cacheable=True,
918 2945 aaronmk
                log_ignore_excs=(DuplicateException,))
919 2718 aaronmk
            break # successful
920 2945 aaronmk
        except DuplicateException:
921 2770 aaronmk
            function.name = next_version(function.name)
922
            # try again with next version of name
923 2704 aaronmk
924
    return sql_gen.FunctionCall(function, col)
925
926 2414 aaronmk
##### Database structure queries
927
928 2426 aaronmk
def table_row_count(db, table, recover=None):
929 2786 aaronmk
    return value(run_query(db, mk_select(db, table, [sql_gen.row_count],
930 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
931 2426 aaronmk
932 2414 aaronmk
def table_cols(db, table, recover=None):
933
    return list(col_names(select(db, table, limit=0, order_by=None,
934 2443 aaronmk
        recover=recover, log_level=4)))
935 2414 aaronmk
936 2291 aaronmk
def pkey(db, table, recover=None):
937 832 aaronmk
    '''Assumed to be first column in table'''
938 2339 aaronmk
    return table_cols(db, table, recover)[0]
939 832 aaronmk
940 2559 aaronmk
not_null_col = 'not_null_col'
941 2340 aaronmk
942
def table_not_null_col(db, table, recover=None):
943
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
944
    if not_null_col in table_cols(db, table, recover): return not_null_col
945
    else: return pkey(db, table, recover)
946
947 853 aaronmk
def index_cols(db, table, index):
948
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
949
    automatically created. When you don't know whether something is a UNIQUE
950
    constraint or a UNIQUE index, use this function.'''
951 1909 aaronmk
    module = util.root_module(db.db)
952
    if module == 'psycopg2':
953
        return list(values(run_query(db, '''\
954 853 aaronmk
SELECT attname
955 866 aaronmk
FROM
956
(
957
        SELECT attnum, attname
958
        FROM pg_index
959
        JOIN pg_class index ON index.oid = indexrelid
960
        JOIN pg_class table_ ON table_.oid = indrelid
961
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
962
        WHERE
963 2782 aaronmk
            table_.relname = '''+db.esc_value(table)+'''
964
            AND index.relname = '''+db.esc_value(index)+'''
965 866 aaronmk
    UNION
966
        SELECT attnum, attname
967
        FROM
968
        (
969
            SELECT
970
                indrelid
971
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
972
                    AS indkey
973
            FROM pg_index
974
            JOIN pg_class index ON index.oid = indexrelid
975
            JOIN pg_class table_ ON table_.oid = indrelid
976
            WHERE
977 2782 aaronmk
                table_.relname = '''+db.esc_value(table)+'''
978
                AND index.relname = '''+db.esc_value(index)+'''
979 866 aaronmk
        ) s
980
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
981
) s
982 853 aaronmk
ORDER BY attnum
983 2782 aaronmk
'''
984
            , cacheable=True, log_level=4)))
985 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
986
        ' database')
987 853 aaronmk
988 464 aaronmk
def constraint_cols(db, table, constraint):
989 1849 aaronmk
    module = util.root_module(db.db)
990 464 aaronmk
    if module == 'psycopg2':
991
        return list(values(run_query(db, '''\
992
SELECT attname
993
FROM pg_constraint
994
JOIN pg_class ON pg_class.oid = conrelid
995
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
996
WHERE
997 2783 aaronmk
    relname = '''+db.esc_value(table)+'''
998
    AND conname = '''+db.esc_value(constraint)+'''
999 464 aaronmk
ORDER BY attnum
1000 2783 aaronmk
'''
1001
            )))
1002 464 aaronmk
    else: raise NotImplementedError("Can't list constraint columns for "+module+
1003
        ' database')
1004
1005 2096 aaronmk
row_num_col = '_row_num'
1006
1007 2998 aaronmk
def add_index(db, exprs, table=None, unique=False, ensure_not_null_=True):
1008 2688 aaronmk
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
1009 2538 aaronmk
    Currently, only function calls are supported as expressions.
1010 2998 aaronmk
    @param ensure_not_null_ If set, translates NULL values to sentinel values.
1011 2847 aaronmk
        This allows indexes to be used for comparisons where NULLs are equal.
1012 2538 aaronmk
    '''
1013 2964 aaronmk
    exprs = lists.mk_seq(exprs)
1014 2538 aaronmk
1015 2688 aaronmk
    # Parse exprs
1016
    old_exprs = exprs[:]
1017
    exprs = []
1018
    cols = []
1019
    for i, expr in enumerate(old_exprs):
1020 2823 aaronmk
        expr = sql_gen.as_Col(expr, table)
1021 2688 aaronmk
1022 2823 aaronmk
        # Handle nullable columns
1023 2998 aaronmk
        if ensure_not_null_:
1024
            try: expr = ensure_not_null(db, expr)
1025 2860 aaronmk
            except KeyError: pass # unknown type, so just create plain index
1026 2823 aaronmk
1027 2688 aaronmk
        # Extract col
1028 3002 aaronmk
        expr = copy.deepcopy(expr) # don't modify input!
1029 2688 aaronmk
        if isinstance(expr, sql_gen.FunctionCall):
1030
            col = expr.args[0]
1031
            expr = sql_gen.Expr(expr)
1032
        else: col = expr
1033 2823 aaronmk
        assert isinstance(col, sql_gen.Col)
1034 2688 aaronmk
1035
        # Extract table
1036
        if table == None:
1037
            assert sql_gen.is_table_col(col)
1038
            table = col.table
1039
1040
        col.table = None
1041
1042
        exprs.append(expr)
1043
        cols.append(col)
1044 2408 aaronmk
1045 2688 aaronmk
    table = sql_gen.as_Table(table)
1046
    index = sql_gen.Table(str(sql_gen.Col(','.join(map(str, cols)), table)))
1047
1048 3005 aaronmk
    # Add index
1049
    while True:
1050
        str_ = 'CREATE'
1051
        if unique: str_ += ' UNIQUE'
1052
        str_ += ' INDEX '+index.to_str(db)+' ON '+table.to_str(db)+' ('+(
1053
            ', '.join((v.to_str(db) for v in exprs)))+')'
1054
1055
        try:
1056
            run_query(db, str_, recover=True, cacheable=True, log_level=3,
1057
                log_ignore_excs=(DuplicateException,))
1058
            break
1059
        except DuplicateException:
1060
            index.name = next_version(index.name)
1061
            # try again with next version of name
1062 2408 aaronmk
1063 2686 aaronmk
def add_pkey(db, table, cols=None, recover=None):
1064
    '''Adds a primary key.
1065
    @param cols [sql_gen.Col,...] The columns in the primary key.
1066
        Defaults to the first column in the table.
1067 2406 aaronmk
    @pre The table must not already have a primary key.
1068
    '''
1069
    table = sql_gen.as_Table(table)
1070 2686 aaronmk
    if cols == None: cols = [pkey(db, table, recover)]
1071
    col_strs = [sql_gen.to_name_only_col(v).to_str(db) for v in cols]
1072 2406 aaronmk
1073 2781 aaronmk
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD PRIMARY KEY ('
1074
        +(', '.join(col_strs))+')', recover=True, cacheable=True, log_level=3,
1075 2945 aaronmk
        log_ignore_excs=(DuplicateException,))
1076 2406 aaronmk
1077 2997 aaronmk
def add_not_null(db, col):
1078
    table = col.table
1079
    col = sql_gen.to_name_only_col(col)
1080
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ALTER COLUMN '
1081
        +col.to_str(db)+' SET NOT NULL', cacheable=True)
1082
1083
def add_index_col(db, col, suffix, expr, nullable=True):
1084 3000 aaronmk
    if sql_gen.index_col(col) != None: return # already has index col
1085 2997 aaronmk
1086
    new_col = sql_gen.suffixed_col(col, suffix)
1087
1088 3006 aaronmk
    # Add column
1089
    while True:
1090
        new_typed_col = sql_gen.TypedCol(new_col.name, db.col_info(col).type)
1091
        try:
1092
            add_col(db, col.table, new_typed_col,
1093
                log_ignore_excs=(DuplicateException,))
1094
            break
1095
        except DuplicateException:
1096
            new_col.name = next_version(new_col.name)
1097
            # try again with next version of name
1098
1099 2997 aaronmk
    update(db, col.table, [(new_col, expr)])
1100
    if not nullable: add_not_null(db, new_col)
1101
    add_index(db, new_col)
1102
1103 3000 aaronmk
    col.table.index_cols[col.name] = new_col
1104 2997 aaronmk
1105
def ensure_not_null(db, col):
1106
    '''For params, see sql_gen.ensure_not_null()'''
1107
    expr = sql_gen.ensure_not_null(db, col)
1108
1109
    # If nullable column in a temp table, add separate column instead
1110
    if sql_gen.is_temp_col(col) and isinstance(expr, sql_gen.EnsureNotNull):
1111
        add_index_col(db, col, '::NOT NULL', expr, nullable=False)
1112 3000 aaronmk
        expr = sql_gen.index_col(col)
1113 2997 aaronmk
1114
    return expr
1115
1116 2772 aaronmk
already_indexed = object() # tells add_indexes() the pkey has already been added
1117
1118
def add_indexes(db, table, has_pkey=True):
1119
    '''Adds an index on all columns in a table.
1120
    @param has_pkey bool|already_indexed Whether a pkey instead of a regular
1121
        index should be added on the first column.
1122
        * If already_indexed, the pkey is assumed to have already been added
1123
    '''
1124
    cols = table_cols(db, table)
1125
    if has_pkey:
1126
        if has_pkey is not already_indexed: add_pkey(db, table)
1127
        cols = cols[1:]
1128
    for col in cols: add_index(db, col, table)
1129
1130 2935 aaronmk
def add_col(db, table, col, **kw_args):
1131 2934 aaronmk
    assert isinstance(col, sql_gen.TypedCol)
1132 3006 aaronmk
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD COLUMN '
1133 2947 aaronmk
        +col.to_str(db), recover=True, cacheable=True, **kw_args)
1134 2934 aaronmk
1135 2935 aaronmk
row_num_typed_col = sql_gen.TypedCol(row_num_col, 'serial', nullable=False,
1136
    constraints='PRIMARY KEY')
1137
1138 2086 aaronmk
def add_row_num(db, table):
1139 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
1140
    be the primary key.'''
1141 2935 aaronmk
    add_col(db, table, row_num_typed_col, log_level=3)
1142 2086 aaronmk
1143 2936 aaronmk
def cast_temp_col(db, type_, col, errors_table=None):
1144
    '''Like cast(), but creates a new column with the cast values if the input
1145
    is a column.
1146
    @return The new column or cast value
1147
    '''
1148
    def cast_(col): return cast(db, type_, col, errors_table)
1149
1150
    try: col = sql_gen.underlying_col(col)
1151
    except sql_gen.NoUnderlyingTableException: return sql_gen.wrap(cast_, col)
1152
1153
    table = col.table
1154
    new_col = sql_gen.Col(sql_gen.concat(col.name, '::'+type_), table, col.srcs)
1155
    expr = cast_(col)
1156
    add_col(db, table, sql_gen.TypedCol(new_col.name, type_))
1157
    update(db, table, [(new_col, expr)])
1158 2955 aaronmk
    add_index(db, new_col)
1159 2936 aaronmk
1160
    return new_col
1161
1162 2889 aaronmk
def drop_table(db, table):
1163
    table = sql_gen.as_Table(table)
1164
    return run_query(db, 'DROP TABLE IF EXISTS '+table.to_str(db)+' CASCADE')
1165
1166 2760 aaronmk
def create_table(db, table, cols, has_pkey=True, col_indexes=True):
1167 2675 aaronmk
    '''Creates a table.
1168 2681 aaronmk
    @param cols [sql_gen.TypedCol,...] The column names and types
1169
    @param has_pkey If set, the first column becomes the primary key.
1170 2760 aaronmk
    @param col_indexes bool|[ref]
1171
        * If True, indexes will be added on all non-pkey columns.
1172
        * If a list reference, [0] will be set to a function to do this.
1173
          This can be used to delay index creation until the table is populated.
1174 2675 aaronmk
    '''
1175
    table = sql_gen.as_Table(table)
1176
1177 2681 aaronmk
    if has_pkey:
1178
        cols[0] = pkey = copy.copy(cols[0]) # don't modify input!
1179 2872 aaronmk
        pkey.constraints = 'PRIMARY KEY'
1180 2681 aaronmk
1181 2675 aaronmk
    str_ = 'CREATE TABLE '+table.to_str(db)+' (\n'
1182 2681 aaronmk
    str_ += '\n, '.join(v.to_str(db) for v in cols)
1183
    str_ += '\n);\n'
1184 2675 aaronmk
    run_query(db, str_, cacheable=True, log_level=2)
1185 2760 aaronmk
1186
    # Add indexes
1187 2773 aaronmk
    if has_pkey: has_pkey = already_indexed
1188
    def add_indexes_(): add_indexes(db, table, has_pkey)
1189
    if isinstance(col_indexes, list): col_indexes[0] = add_indexes_ # defer
1190
    elif col_indexes: add_indexes_() # add now
1191 2675 aaronmk
1192 2684 aaronmk
def vacuum(db, table):
1193
    table = sql_gen.as_Table(table)
1194
    db.with_autocommit(lambda: run_query(db, 'VACUUM ANALYZE '+table.to_str(db),
1195
        log_level=3))
1196
1197 2970 aaronmk
def truncate(db, table, schema='public', **kw_args):
1198
    '''For params, see run_query()'''
1199 2777 aaronmk
    table = sql_gen.as_Table(table, schema)
1200 2970 aaronmk
    return run_query(db, 'TRUNCATE '+table.to_str(db)+' CASCADE', **kw_args)
1201 2732 aaronmk
1202 2965 aaronmk
def empty_temp(db, tables):
1203 2972 aaronmk
    if db.debug_temp: return # leave temp tables there for debugging
1204 2965 aaronmk
    tables = lists.mk_seq(tables)
1205 2971 aaronmk
    for table in tables: truncate(db, table, log_level=3)
1206 2965 aaronmk
1207 2729 aaronmk
def tables(db, schema_like='public', table_like='%', exact=False):
1208
    if exact: compare = '='
1209
    else: compare = 'LIKE'
1210
1211 1849 aaronmk
    module = util.root_module(db.db)
1212 832 aaronmk
    if module == 'psycopg2':
1213 2779 aaronmk
        conds = [('schemaname', sql_gen.CompareCond(schema_like, compare)),
1214
            ('tablename', sql_gen.CompareCond(table_like, compare))]
1215
        return values(select(db, 'pg_tables', ['tablename'], conds,
1216
            order_by='tablename', log_level=4))
1217 1968 aaronmk
    elif module == 'MySQLdb':
1218 2779 aaronmk
        return values(run_query(db, 'SHOW TABLES LIKE '+db.esc_value(table_like)
1219
            , cacheable=True, log_level=4))
1220 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
1221 830 aaronmk
1222 2726 aaronmk
def table_exists(db, table):
1223
    table = sql_gen.as_Table(table)
1224 2730 aaronmk
    return list(tables(db, table.schema, table.name, exact=True)) != []
1225 2726 aaronmk
1226 2913 aaronmk
def function_exists(db, function):
1227
    function = sql_gen.as_Function(function)
1228
1229
    info_table = sql_gen.Table('routines', 'information_schema')
1230
    conds = [('routine_name', function.name)]
1231
    schema = function.schema
1232
    if schema != None: conds.append(('routine_schema', schema))
1233 2914 aaronmk
    # Exclude trigger functions, since they cannot be called directly
1234
    conds.append(('data_type', sql_gen.CompareCond('trigger', '!=')))
1235 2913 aaronmk
1236
    return list(values(select(db, info_table, ['routine_name'], conds,
1237
        order_by='routine_schema', limit=1, log_level=4))) != []
1238
        # TODO: order_by search_path schema order
1239
1240 2726 aaronmk
def errors_table(db, table, if_exists=True):
1241
    '''
1242
    @param if_exists If set, returns None if the errors table doesn't exist
1243
    @return None|sql_gen.Table
1244
    '''
1245
    table = sql_gen.as_Table(table)
1246
    if table.srcs != (): table = table.srcs[0]
1247
1248
    errors_table = sql_gen.suffixed_table(table, '.errors')
1249
    if if_exists and not table_exists(db, errors_table): return None
1250
    return errors_table
1251
1252 833 aaronmk
##### Database management
1253
1254 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
1255
    '''For kw_args, see tables()'''
1256
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
1257 833 aaronmk
1258 832 aaronmk
##### Heuristic queries
1259
1260 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
1261 1554 aaronmk
    '''Recovers from errors.
1262 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
1263
    '''
1264 2642 aaronmk
    row = sql_gen.ColDict(db, table, row)
1265 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
1266
1267 471 aaronmk
    try:
1268 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
1269 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
1270
            row_ct_ref[0] += cur.rowcount
1271
        return value(cur)
1272 471 aaronmk
    except DuplicateKeyException, e:
1273 2653 aaronmk
        row = sql_gen.ColDict(db, table,
1274
            util.dict_subset_right_join(row, e.cols))
1275
        return value(select(db, table, [pkey_], row, recover=True))
1276 471 aaronmk
1277 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
1278 830 aaronmk
    '''Recovers from errors'''
1279 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
1280 14 aaronmk
    except StopIteration:
1281 40 aaronmk
        if not create: raise
1282 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
1283 2078 aaronmk
1284 2593 aaronmk
def is_func_result(col):
1285
    return col.table.name.find('(') >= 0 and col.name == 'result'
1286
1287 2592 aaronmk
def into_table_name(out_table, in_tables0, mapping, is_func):
1288 2615 aaronmk
    def in_col_str(in_col):
1289
        in_col = sql_gen.remove_col_rename(in_col)
1290
        if isinstance(in_col, sql_gen.Col):
1291
            table = in_col.table
1292
            if table == in_tables0:
1293
                in_col = sql_gen.to_name_only_col(in_col)
1294
            elif is_func_result(in_col): in_col = table # omit col name
1295
        return str(in_col)
1296
1297 2580 aaronmk
    str_ = str(out_table)
1298
    if is_func:
1299 2615 aaronmk
        str_ += '('
1300 2580 aaronmk
1301 2615 aaronmk
        try: value_in_col = mapping['value']
1302
        except KeyError:
1303 2654 aaronmk
            str_ += ', '.join((str(k)+'='+in_col_str(v)
1304 2615 aaronmk
                for k, v in mapping.iteritems()))
1305
        else: str_ += in_col_str(value_in_col)
1306
1307
        str_ += ')'
1308 2812 aaronmk
    else:
1309
        out_col = 'rank'
1310
        try: in_col = mapping[out_col]
1311
        except KeyError: str_ += '_pkeys'
1312
        else: # has a rank column, so hierarchical
1313
            str_ += '['+str(out_col)+'='+in_col_str(in_col)+']'
1314 2580 aaronmk
    return str_
1315
1316 2508 aaronmk
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None, into=None,
1317 2805 aaronmk
    default=None, is_func=False, on_error=exc.raise_):
1318 2078 aaronmk
    '''Recovers from errors.
1319
    Only works under PostgreSQL (uses INSERT RETURNING).
1320 2909 aaronmk
    IMPORTANT: Must be run at the *beginning* of a transaction.
1321 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
1322
        tables to join with it using the main input table's pkey
1323 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
1324 2612 aaronmk
        * out_table_col: str (*not* sql_gen.Col)
1325
        * in_table_col: sql_gen.Col|literal-value
1326 2489 aaronmk
    @param into The table to contain the output and input pkeys.
1327 2574 aaronmk
        Defaults to `out_table.name+'_pkeys'`.
1328 2509 aaronmk
    @param default The *output* column to use as the pkey for missing rows.
1329
        If this output column does not exist in the mapping, uses None.
1330 2552 aaronmk
    @param is_func Whether out_table is the name of a SQL function, not a table
1331 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
1332 2078 aaronmk
    '''
1333 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
1334 2312 aaronmk
1335 2450 aaronmk
    def log_debug(msg): db.log_debug(msg, level=1.5)
1336 2505 aaronmk
    def col_ustr(str_):
1337 2567 aaronmk
        return strings.repr_no_u(sql_gen.remove_col_rename(str_))
1338 2450 aaronmk
1339 2910 aaronmk
    log_debug('********** New iteration **********')
1340
    log_debug('Inserting these input columns into '+strings.as_tt(
1341
        out_table.to_str(db))+':\n'+strings.as_table(mapping, ustr=col_ustr))
1342
1343 2942 aaronmk
    is_function = function_exists(db, out_table)
1344
1345
    if is_function: out_pkey = 'result'
1346
    else: out_pkey = pkey(db, out_table, recover=True)
1347 2768 aaronmk
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
1348
1349
    if mapping == {}: # need at least one column for INSERT SELECT
1350
        mapping = {out_pkey: None} # ColDict will replace with default value
1351
1352 2382 aaronmk
    # Create input joins from list of input tables
1353
    in_tables_ = in_tables[:] # don't modify input!
1354
    in_tables0 = in_tables_.pop(0) # first table is separate
1355 2735 aaronmk
    errors_table_ = errors_table(db, in_tables0)
1356 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
1357 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
1358 2460 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v,
1359
        {in_pkey: sql_gen.join_same_not_null}) for v in in_tables_]
1360 2131 aaronmk
1361 2592 aaronmk
    if into == None:
1362
        into = into_table_name(out_table, in_tables0, mapping, is_func)
1363
    into = sql_gen.as_Table(into)
1364
1365 2702 aaronmk
    # Set column sources
1366
    in_cols = filter(sql_gen.is_table_col, mapping.values())
1367
    for col in in_cols:
1368
        if col.table == in_tables0: col.set_srcs(sql_gen.src_self)
1369
1370 2486 aaronmk
    log_debug('Joining together input tables into temp table')
1371 2395 aaronmk
    # Place in new table for speed and so don't modify input if values edited
1372 2584 aaronmk
    in_table = sql_gen.Table('in')
1373 2702 aaronmk
    mapping = dicts.join(mapping, flatten(db, in_table, input_joins, in_cols,
1374
        preserve=[in_pkey_col], start=0))
1375 2395 aaronmk
    input_joins = [in_table]
1376 2486 aaronmk
    db.log_debug('Temp table: '+strings.as_tt(in_table.to_str(db)), level=2)
1377 2395 aaronmk
1378 2656 aaronmk
    mapping = sql_gen.ColDict(db, out_table, mapping)
1379
        # after applying dicts.join() because that returns a plain dict
1380
1381 2509 aaronmk
    # Resolve default value column
1382 2841 aaronmk
    if default != None:
1383
        try: default = mapping[default]
1384
        except KeyError:
1385 2509 aaronmk
            db.log_debug('Default value column '
1386
                +strings.as_tt(strings.repr_no_u(default))
1387 2511 aaronmk
                +' does not exist in mapping, falling back to None', level=2.1)
1388 2509 aaronmk
            default = None
1389
1390 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
1391 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
1392
1393 2201 aaronmk
    pkeys_table_exists_ref = [False]
1394 2960 aaronmk
    def insert_into_pkeys(joins, cols, distinct=False):
1395
        kw_args = {}
1396
        if distinct: kw_args.update(dict(distinct_on=[in_pkey_col]))
1397
        query = mk_select(db, joins, cols, order_by=None, start=0, **kw_args)
1398
1399 2201 aaronmk
        if pkeys_table_exists_ref[0]:
1400 2786 aaronmk
            insert_select(db, into, pkeys_names, query)
1401 2201 aaronmk
        else:
1402 2786 aaronmk
            run_query_into(db, query, into=into)
1403 2201 aaronmk
            pkeys_table_exists_ref[0] = True
1404
1405 2429 aaronmk
    limit_ref = [None]
1406 2380 aaronmk
    conds = set()
1407 2811 aaronmk
    distinct_on = sql_gen.ColDict(db, out_table)
1408 2325 aaronmk
    def mk_main_select(joins, cols):
1409 2811 aaronmk
        distinct_on_cols = [c.to_Col() for c in distinct_on.values()]
1410
        return mk_select(db, joins, cols, conds, distinct_on_cols,
1411 2429 aaronmk
            limit=limit_ref[0], start=0)
1412 2132 aaronmk
1413 2519 aaronmk
    exc_strs = set()
1414 2309 aaronmk
    def log_exc(e):
1415 2519 aaronmk
        e_str = exc.str_(e, first_line_only=True)
1416
        log_debug('Caught exception: '+e_str)
1417
        assert e_str not in exc_strs # avoid infinite loops
1418
        exc_strs.add(e_str)
1419 2735 aaronmk
1420 2451 aaronmk
    def remove_all_rows():
1421 2896 aaronmk
        log_debug('Ignoring all rows')
1422 2429 aaronmk
        limit_ref[0] = 0 # just create an empty pkeys table
1423 2735 aaronmk
1424 2758 aaronmk
    def ignore(in_col, value, e):
1425
        track_data_error(db, errors_table_, in_col.srcs, value, e.cause.pgcode,
1426
            e.cause.pgerror)
1427 2895 aaronmk
        log_debug('Ignoring rows with '+strings.as_tt(repr(in_col))+' = '
1428 2545 aaronmk
            +strings.as_tt(repr(value)))
1429 2895 aaronmk
1430 2758 aaronmk
    def remove_rows(in_col, value, e):
1431
        ignore(in_col, value, e)
1432 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
1433
        assert cond not in conds # avoid infinite loops
1434 2380 aaronmk
        conds.add(cond)
1435 2895 aaronmk
1436 2758 aaronmk
    def invalid2null(in_col, value, e):
1437
        ignore(in_col, value, e)
1438 2403 aaronmk
        update(db, in_table, [(in_col, None)],
1439
            sql_gen.ColValueCond(in_col, value))
1440 2245 aaronmk
1441 2589 aaronmk
    def insert_pkeys_table(which):
1442 2932 aaronmk
        return sql_gen.Table(sql_gen.concat(in_table.name,
1443 2589 aaronmk
            '_insert_'+which+'_pkeys'))
1444
    insert_out_pkeys = insert_pkeys_table('out')
1445
    insert_in_pkeys = insert_pkeys_table('in')
1446
1447 2206 aaronmk
    # Do inserts and selects
1448 2642 aaronmk
    join_cols = sql_gen.ColDict(db, out_table)
1449 2206 aaronmk
    while True:
1450 2521 aaronmk
        if limit_ref[0] == 0: # special case
1451
            log_debug('Creating an empty pkeys table')
1452 2786 aaronmk
            cur = run_query_into(db, mk_select(db, out_table, [out_pkey],
1453 2521 aaronmk
                limit=limit_ref[0]), into=insert_out_pkeys)
1454
            break # don't do main case
1455
1456 2303 aaronmk
        has_joins = join_cols != {}
1457
1458 2903 aaronmk
        log_debug('Trying to insert new rows')
1459
1460 2305 aaronmk
        # Prepare to insert new rows
1461 2325 aaronmk
        insert_joins = input_joins[:] # don't modify original!
1462 2403 aaronmk
        insert_args = dict(recover=True, cacheable=False)
1463 2303 aaronmk
        if has_joins:
1464 3021 aaronmk
            insert_args.update(dict(ignore=True))
1465 2325 aaronmk
        else:
1466 2404 aaronmk
            insert_args.update(dict(returning=out_pkey, into=insert_out_pkeys))
1467 2786 aaronmk
        main_select = mk_main_select(insert_joins, mapping.values())
1468 2303 aaronmk
1469 2929 aaronmk
        def main_insert():
1470 2942 aaronmk
            if is_function:
1471
                log_debug('Calling function on input rows')
1472
                args = dict(((k.name, v) for k, v in mapping.iteritems()))
1473
                func_call = sql_gen.NamedCol(out_pkey,
1474
                    sql_gen.FunctionCall(out_table, **args))
1475
                insert_into_pkeys(input_joins, [in_pkey_col, func_call])
1476
                return None
1477
            else:
1478
                lock_table(db, out_table, 'EXCLUSIVE')
1479
                return insert_select(db, out_table, mapping.keys(), main_select,
1480
                    **insert_args)
1481 2929 aaronmk
1482 2206 aaronmk
        try:
1483 2929 aaronmk
            cur = with_savepoint(db, main_insert)
1484 2357 aaronmk
            break # insert successful
1485 2931 aaronmk
        except MissingCastException, e:
1486
            log_exc(e)
1487
1488
            out_col = e.col
1489
            type_ = e.type
1490
1491
            log_debug('Casting '+strings.as_tt(out_col)+' input to '
1492
                +strings.as_tt(type_))
1493 2937 aaronmk
            mapping[out_col] = cast_temp_col(db, type_, mapping[out_col],
1494
                errors_table_)
1495 2206 aaronmk
        except DuplicateKeyException, e:
1496 2309 aaronmk
            log_exc(e)
1497
1498 2258 aaronmk
            old_join_cols = join_cols.copy()
1499 2811 aaronmk
            distinct_on.update(util.dict_subset(mapping, e.cols))
1500 2565 aaronmk
            join_cols.update(util.dict_subset_right_join(mapping, e.cols))
1501 2486 aaronmk
            log_debug('Ignoring existing rows, comparing on these columns:\n'
1502 2505 aaronmk
                +strings.as_inline_table(join_cols, ustr=col_ustr))
1503 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
1504 2230 aaronmk
        except NullValueException, e:
1505 2309 aaronmk
            log_exc(e)
1506
1507 2230 aaronmk
            out_col, = e.cols
1508
            try: in_col = mapping[out_col]
1509 2356 aaronmk
            except KeyError:
1510 2486 aaronmk
                log_debug('Missing mapping for NOT NULL column '+out_col)
1511 2451 aaronmk
                remove_all_rows()
1512 2758 aaronmk
            else: remove_rows(in_col, None, e)
1513 2542 aaronmk
        except FunctionValueException, e:
1514
            log_exc(e)
1515
1516
            func_name = e.name
1517
            value = e.value
1518
            for out_col, in_col in mapping.iteritems():
1519 2758 aaronmk
                in_col = sql_gen.unwrap_func_call(in_col, func_name)
1520
                invalid2null(in_col, value, e)
1521 2429 aaronmk
        except DatabaseErrors, e:
1522
            log_exc(e)
1523
1524 2808 aaronmk
            log_debug('No handler for exception')
1525
            on_error(e)
1526 2451 aaronmk
            remove_all_rows()
1527 2358 aaronmk
        # after exception handled, rerun loop with additional constraints
1528 2132 aaronmk
1529 2942 aaronmk
    if cur != None and row_ct_ref != None and cur.rowcount >= 0:
1530 2357 aaronmk
        row_ct_ref[0] += cur.rowcount
1531
1532 2942 aaronmk
    if is_function: pass # pkeys table already created
1533
    elif has_joins:
1534 2357 aaronmk
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
1535 2486 aaronmk
        log_debug('Getting output table pkeys of existing/inserted rows')
1536 2960 aaronmk
        insert_into_pkeys(select_joins, pkeys_cols, distinct=True)
1537 2357 aaronmk
    else:
1538 2404 aaronmk
        add_row_num(db, insert_out_pkeys) # for joining with input pkeys
1539 2357 aaronmk
1540 2486 aaronmk
        log_debug('Getting input table pkeys of inserted rows')
1541 2786 aaronmk
        run_query_into(db, mk_main_select(input_joins, [in_pkey]),
1542 2404 aaronmk
            into=insert_in_pkeys)
1543
        add_row_num(db, insert_in_pkeys) # for joining with output pkeys
1544 2357 aaronmk
1545 2428 aaronmk
        assert table_row_count(db, insert_out_pkeys) == table_row_count(db,
1546
            insert_in_pkeys)
1547
1548 2486 aaronmk
        log_debug('Combining output and input pkeys in inserted order')
1549 2404 aaronmk
        pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
1550 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
1551 2420 aaronmk
        insert_into_pkeys(pkey_joins, pkeys_names)
1552 2959 aaronmk
1553 2974 aaronmk
        empty_temp(db, [insert_out_pkeys, insert_in_pkeys])
1554 2357 aaronmk
1555 2486 aaronmk
    db.log_debug('Adding pkey on pkeys table to enable fast joins', level=2.5)
1556 2594 aaronmk
    add_pkey(db, into)
1557 2407 aaronmk
1558 2508 aaronmk
    log_debug('Setting pkeys of missing rows to '+strings.as_tt(repr(default)))
1559 2489 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(into,
1560 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
1561
        # must use join_same_not_null or query will take forever
1562 2420 aaronmk
    insert_into_pkeys(missing_rows_joins,
1563 2508 aaronmk
        [in_pkey_col, sql_gen.NamedCol(out_pkey, default)])
1564 2357 aaronmk
1565 2489 aaronmk
    assert table_row_count(db, into) == table_row_count(db, in_table)
1566 2428 aaronmk
1567 2974 aaronmk
    empty_temp(db, in_table)
1568 2959 aaronmk
1569 2702 aaronmk
    srcs = []
1570
    if is_func: srcs = sql_gen.cols_srcs(in_cols)
1571
    return sql_gen.Col(out_pkey, into, srcs)
1572 2115 aaronmk
1573
##### Data cleanup
1574
1575 2290 aaronmk
def cleanup_table(db, table, cols):
1576 2677 aaronmk
    table = sql_gen.as_Table(table)
1577
    cols = map(sql_gen.as_Col, cols)
1578 2115 aaronmk
1579 2677 aaronmk
    expr = ('nullif(nullif(trim(both from %s), '+db.esc_value('')+'), '
1580
        +db.esc_value(r'\N')+')')
1581
    changes = [(v, sql_gen.CustomCode(expr % v.to_str(db)))
1582
        for v in cols]
1583 2115 aaronmk
1584 2677 aaronmk
    update(db, table, changes)