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