Project

General

Profile

1 11 aaronmk
# Database access
2
3 1869 aaronmk
import copy
4 2127 aaronmk
import operator
5 11 aaronmk
import re
6 865 aaronmk
import warnings
7 11 aaronmk
8 300 aaronmk
import exc
9 1909 aaronmk
import dicts
10 1893 aaronmk
import iters
11 1960 aaronmk
import lists
12 1889 aaronmk
from Proxy import Proxy
13 1872 aaronmk
import rand
14 2217 aaronmk
import sql_gen
15 862 aaronmk
import strings
16 131 aaronmk
import util
17 11 aaronmk
18 832 aaronmk
##### Exceptions
19
20 2804 aaronmk
def get_cur_query(cur, input_query=None):
21 2168 aaronmk
    raw_query = None
22
    if hasattr(cur, 'query'): raw_query = cur.query
23
    elif hasattr(cur, '_last_executed'): raw_query = cur._last_executed
24 2170 aaronmk
25
    if raw_query != None: return raw_query
26 2804 aaronmk
    else: return '[input] '+strings.ustr(input_query)
27 14 aaronmk
28 2170 aaronmk
def _add_cursor_info(e, *args, **kw_args):
29
    '''For params, see get_cur_query()'''
30 2771 aaronmk
    exc.add_msg(e, 'query: '+strings.ustr(get_cur_query(*args, **kw_args)))
31 135 aaronmk
32 300 aaronmk
class DbException(exc.ExceptionWithCause):
33 14 aaronmk
    def __init__(self, msg, cause=None, cur=None):
34 2145 aaronmk
        exc.ExceptionWithCause.__init__(self, msg, cause, cause_newline=True)
35 14 aaronmk
        if cur != None: _add_cursor_info(self, cur)
36
37 2143 aaronmk
class ExceptionWithName(DbException):
38
    def __init__(self, name, cause=None):
39 2484 aaronmk
        DbException.__init__(self, 'for name: '+strings.as_tt(str(name)), cause)
40 2143 aaronmk
        self.name = name
41 360 aaronmk
42 2240 aaronmk
class ExceptionWithNameValue(DbException):
43
    def __init__(self, name, value, cause=None):
44 2484 aaronmk
        DbException.__init__(self, 'for name: '+strings.as_tt(str(name))
45
            +'; value: '+strings.as_tt(repr(value)), cause)
46 2240 aaronmk
        self.name = name
47
        self.value = value
48
49 2945 aaronmk
class ExceptionWithNameType(DbException):
50
    def __init__(self, type_, name, cause=None):
51
        DbException.__init__(self, 'for type: '+strings.as_tt(str(type_))
52
            +'; name: '+strings.as_tt(name), cause)
53
        self.type = type_
54
        self.name = name
55
56 2306 aaronmk
class ConstraintException(DbException):
57
    def __init__(self, name, cols, cause=None):
58 2484 aaronmk
        DbException.__init__(self, 'Violated '+strings.as_tt(name)
59
            +' constraint on columns: '+strings.as_tt(', '.join(cols)), cause)
60 2306 aaronmk
        self.name = name
61 468 aaronmk
        self.cols = cols
62 11 aaronmk
63 2523 aaronmk
class MissingCastException(DbException):
64
    def __init__(self, type_, col, cause=None):
65
        DbException.__init__(self, 'Missing cast to type '+strings.as_tt(type_)
66
            +' on column: '+strings.as_tt(col), cause)
67
        self.type = type_
68
        self.col = col
69
70 2143 aaronmk
class NameException(DbException): pass
71
72 2306 aaronmk
class DuplicateKeyException(ConstraintException): pass
73 13 aaronmk
74 2306 aaronmk
class NullValueException(ConstraintException): pass
75 13 aaronmk
76 2240 aaronmk
class FunctionValueException(ExceptionWithNameValue): pass
77 2239 aaronmk
78 2945 aaronmk
class DuplicateException(ExceptionWithNameType): pass
79 2143 aaronmk
80 89 aaronmk
class EmptyRowException(DbException): pass
81
82 865 aaronmk
##### Warnings
83
84
class DbWarning(UserWarning): pass
85
86 1930 aaronmk
##### Result retrieval
87
88
def col_names(cur): return (col[0] for col in cur.description)
89
90
def rows(cur): return iter(lambda: cur.fetchone(), None)
91
92
def consume_rows(cur):
93
    '''Used to fetch all rows so result will be cached'''
94
    iters.consume_iter(rows(cur))
95
96
def next_row(cur): return rows(cur).next()
97
98
def row(cur):
99
    row_ = next_row(cur)
100
    consume_rows(cur)
101
    return row_
102
103
def next_value(cur): return next_row(cur)[0]
104
105
def value(cur): return row(cur)[0]
106
107
def values(cur): return iters.func_iter(lambda: next_value(cur))
108
109
def value_or_none(cur):
110
    try: return value(cur)
111
    except StopIteration: return None
112
113 2762 aaronmk
##### Escaping
114 2101 aaronmk
115 2573 aaronmk
def esc_name_by_module(module, name):
116
    if module == 'psycopg2' or module == None: quote = '"'
117 2101 aaronmk
    elif module == 'MySQLdb': quote = '`'
118
    else: raise NotImplementedError("Can't escape name for "+module+' database')
119 2500 aaronmk
    return sql_gen.esc_name(name, quote)
120 2101 aaronmk
121
def esc_name_by_engine(engine, name, **kw_args):
122
    return esc_name_by_module(db_engines[engine][0], name, **kw_args)
123
124
def esc_name(db, name, **kw_args):
125
    return esc_name_by_module(util.root_module(db.db), name, **kw_args)
126
127
def qual_name(db, schema, table):
128
    def esc_name_(name): return esc_name(db, name)
129
    table = esc_name_(table)
130
    if schema != None: return esc_name_(schema)+'.'+table
131
    else: return table
132
133 1869 aaronmk
##### Database connections
134 1849 aaronmk
135 2097 aaronmk
db_config_names = ['engine', 'host', 'user', 'password', 'database', 'schemas']
136 1926 aaronmk
137 1869 aaronmk
db_engines = {
138
    'MySQL': ('MySQLdb', {'password': 'passwd', 'database': 'db'}),
139
    'PostgreSQL': ('psycopg2', {}),
140
}
141
142
DatabaseErrors_set = set([DbException])
143
DatabaseErrors = tuple(DatabaseErrors_set)
144
145
def _add_module(module):
146
    DatabaseErrors_set.add(module.DatabaseError)
147
    global DatabaseErrors
148
    DatabaseErrors = tuple(DatabaseErrors_set)
149
150
def db_config_str(db_config):
151
    return db_config['engine']+' database '+db_config['database']
152
153 2448 aaronmk
log_debug_none = lambda msg, level=2: None
154 1901 aaronmk
155 1849 aaronmk
class DbConn:
156 2923 aaronmk
    def __init__(self, db_config, autocommit=True, caching=True,
157 2915 aaronmk
        log_debug=log_debug_none, debug_temp=False):
158
        '''
159
        @param debug_temp Whether temporary objects should instead be permanent.
160
            This assists in debugging the internal objects used by the program.
161
        '''
162 1869 aaronmk
        self.db_config = db_config
163 2190 aaronmk
        self.autocommit = autocommit
164
        self.caching = caching
165 1901 aaronmk
        self.log_debug = log_debug
166 2193 aaronmk
        self.debug = log_debug != log_debug_none
167 2915 aaronmk
        self.debug_temp = debug_temp
168 1869 aaronmk
169
        self.__db = None
170 1889 aaronmk
        self.query_results = {}
171 2139 aaronmk
        self._savepoint = 0
172 2671 aaronmk
        self._notices_seen = set()
173 1869 aaronmk
174
    def __getattr__(self, name):
175
        if name == '__dict__': raise Exception('getting __dict__')
176
        if name == 'db': return self._db()
177
        else: raise AttributeError()
178
179
    def __getstate__(self):
180
        state = copy.copy(self.__dict__) # shallow copy
181 1915 aaronmk
        state['log_debug'] = None # don't pickle the debug callback
182 1869 aaronmk
        state['_DbConn__db'] = None # don't pickle the connection
183
        return state
184
185 2165 aaronmk
    def connected(self): return self.__db != None
186
187 1869 aaronmk
    def _db(self):
188
        if self.__db == None:
189
            # Process db_config
190
            db_config = self.db_config.copy() # don't modify input!
191 2097 aaronmk
            schemas = db_config.pop('schemas', None)
192 1869 aaronmk
            module_name, mappings = db_engines[db_config.pop('engine')]
193
            module = __import__(module_name)
194
            _add_module(module)
195
            for orig, new in mappings.iteritems():
196
                try: util.rename_key(db_config, orig, new)
197
                except KeyError: pass
198
199
            # Connect
200
            self.__db = module.connect(**db_config)
201
202
            # Configure connection
203 2906 aaronmk
            if hasattr(self.db, 'set_isolation_level'):
204
                import psycopg2.extensions
205
                self.db.set_isolation_level(
206
                    psycopg2.extensions.ISOLATION_LEVEL_READ_COMMITTED)
207 2101 aaronmk
            if schemas != None:
208 2893 aaronmk
                search_path = [self.esc_name(s) for s in schemas.split(',')]
209
                search_path.append(value(run_query(self, 'SHOW search_path',
210
                    log_level=4)))
211
                run_query(self, 'SET search_path TO '+(','.join(search_path)),
212
                    log_level=3)
213 1869 aaronmk
214
        return self.__db
215 1889 aaronmk
216 1891 aaronmk
    class DbCursor(Proxy):
217 1927 aaronmk
        def __init__(self, outer):
218 1891 aaronmk
            Proxy.__init__(self, outer.db.cursor())
219 2191 aaronmk
            self.outer = outer
220 1927 aaronmk
            self.query_results = outer.query_results
221 1894 aaronmk
            self.query_lookup = None
222 1891 aaronmk
            self.result = []
223 1889 aaronmk
224 2802 aaronmk
        def execute(self, query):
225 2764 aaronmk
            self._is_insert = query.startswith('INSERT')
226 2797 aaronmk
            self.query_lookup = query
227 2148 aaronmk
            try:
228 2191 aaronmk
                try:
229 2802 aaronmk
                    cur = self.inner.execute(query)
230 2191 aaronmk
                    self.outer.do_autocommit()
231 2802 aaronmk
                finally: self.query = get_cur_query(self.inner, query)
232 1904 aaronmk
            except Exception, e:
233 2802 aaronmk
                _add_cursor_info(e, self, query)
234 1904 aaronmk
                self.result = e # cache the exception as the result
235
                self._cache_result()
236
                raise
237 3004 aaronmk
238
            # Always cache certain queries
239
            if query.startswith('CREATE') or query.startswith('ALTER'):
240 3007 aaronmk
                # structural changes
241 3040 aaronmk
                # Rest of query must be unique in the face of name collisions,
242
                # so don't cache ADD COLUMN unless it has distinguishing comment
243
                if query.find('ADD COLUMN') < 0 or query.endswith('*/'):
244 3007 aaronmk
                    self._cache_result()
245 3004 aaronmk
            elif self.rowcount == 0 and query.startswith('SELECT'): # empty
246 2800 aaronmk
                consume_rows(self) # fetch all rows so result will be cached
247 3004 aaronmk
248 2762 aaronmk
            return cur
249 1894 aaronmk
250 1891 aaronmk
        def fetchone(self):
251
            row = self.inner.fetchone()
252 1899 aaronmk
            if row != None: self.result.append(row)
253
            # otherwise, fetched all rows
254 1904 aaronmk
            else: self._cache_result()
255
            return row
256
257
        def _cache_result(self):
258 2948 aaronmk
            # For inserts that return a result set, don't cache result set since
259
            # inserts are not idempotent. Other non-SELECT queries don't have
260
            # their result set read, so only exceptions will be cached (an
261
            # invalid query will always be invalid).
262 1930 aaronmk
            if self.query_results != None and (not self._is_insert
263 1906 aaronmk
                or isinstance(self.result, Exception)):
264
265 1894 aaronmk
                assert self.query_lookup != None
266 1916 aaronmk
                self.query_results[self.query_lookup] = self.CacheCursor(
267
                    util.dict_subset(dicts.AttrsDictView(self),
268
                    ['query', 'result', 'rowcount', 'description']))
269 1906 aaronmk
270 1916 aaronmk
        class CacheCursor:
271
            def __init__(self, cached_result): self.__dict__ = cached_result
272
273 1927 aaronmk
            def execute(self, *args, **kw_args):
274 1916 aaronmk
                if isinstance(self.result, Exception): raise self.result
275
                # otherwise, result is a rows list
276
                self.iter = iter(self.result)
277
278
            def fetchone(self):
279
                try: return self.iter.next()
280
                except StopIteration: return None
281 1891 aaronmk
282 2212 aaronmk
    def esc_value(self, value):
283 2663 aaronmk
        try: str_ = self.mogrify('%s', [value])
284
        except NotImplementedError, e:
285
            module = util.root_module(self.db)
286
            if module == 'MySQLdb':
287
                import _mysql
288
                str_ = _mysql.escape_string(value)
289
            else: raise e
290 2374 aaronmk
        return strings.to_unicode(str_)
291 2212 aaronmk
292 2347 aaronmk
    def esc_name(self, name): return esc_name(self, name) # calls global func
293
294 2814 aaronmk
    def std_code(self, str_):
295
        '''Standardizes SQL code.
296
        * Ensures that string literals are prefixed by `E`
297
        '''
298
        if str_.startswith("'"): str_ = 'E'+str_
299
        return str_
300
301 2665 aaronmk
    def can_mogrify(self):
302 2663 aaronmk
        module = util.root_module(self.db)
303 2665 aaronmk
        return module == 'psycopg2'
304 2663 aaronmk
305 2665 aaronmk
    def mogrify(self, query, params=None):
306
        if self.can_mogrify(): return self.db.cursor().mogrify(query, params)
307
        else: raise NotImplementedError("Can't mogrify query")
308
309 2671 aaronmk
    def print_notices(self):
310 2725 aaronmk
        if hasattr(self.db, 'notices'):
311
            for msg in self.db.notices:
312
                if msg not in self._notices_seen:
313
                    self._notices_seen.add(msg)
314
                    self.log_debug(msg, level=2)
315 2671 aaronmk
316 2793 aaronmk
    def run_query(self, query, cacheable=False, log_level=2,
317 2464 aaronmk
        debug_msg_ref=None):
318 2445 aaronmk
        '''
319 2464 aaronmk
        @param log_ignore_excs The log_level will be increased by 2 if the query
320
            throws one of these exceptions.
321 2664 aaronmk
        @param debug_msg_ref If specified, the log message will be returned in
322
            this instead of being output. This allows you to filter log messages
323
            depending on the result of the query.
324 2445 aaronmk
        '''
325 2167 aaronmk
        assert query != None
326
327 2047 aaronmk
        if not self.caching: cacheable = False
328 1903 aaronmk
        used_cache = False
329 2664 aaronmk
330
        def log_msg(query):
331
            if used_cache: cache_status = 'cache hit'
332
            elif cacheable: cache_status = 'cache miss'
333
            else: cache_status = 'non-cacheable'
334
            return 'DB query: '+cache_status+':\n'+strings.as_code(query, 'SQL')
335
336 1903 aaronmk
        try:
337 1927 aaronmk
            # Get cursor
338
            if cacheable:
339
                try:
340 2797 aaronmk
                    cur = self.query_results[query]
341 1927 aaronmk
                    used_cache = True
342
                except KeyError: cur = self.DbCursor(self)
343
            else: cur = self.db.cursor()
344
345 2664 aaronmk
            # Log query
346
            if self.debug and debug_msg_ref == None: # log before running
347
                self.log_debug(log_msg(query), log_level)
348
349 1927 aaronmk
            # Run query
350 2793 aaronmk
            cur.execute(query)
351 1903 aaronmk
        finally:
352 2671 aaronmk
            self.print_notices()
353 2664 aaronmk
            if self.debug and debug_msg_ref != None: # return after running
354 2793 aaronmk
                debug_msg_ref[0] = log_msg(str(get_cur_query(cur, query)))
355 1903 aaronmk
356
        return cur
357 1914 aaronmk
358 2797 aaronmk
    def is_cached(self, query): return query in self.query_results
359 2139 aaronmk
360 2907 aaronmk
    def with_autocommit(self, func):
361 2801 aaronmk
        import psycopg2.extensions
362
363
        prev_isolation_level = self.db.isolation_level
364 2907 aaronmk
        self.db.set_isolation_level(
365
            psycopg2.extensions.ISOLATION_LEVEL_AUTOCOMMIT)
366 2683 aaronmk
        try: return func()
367 2801 aaronmk
        finally: self.db.set_isolation_level(prev_isolation_level)
368 2683 aaronmk
369 2139 aaronmk
    def with_savepoint(self, func):
370 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
371 2443 aaronmk
        self.run_query('SAVEPOINT '+savepoint, log_level=4)
372 2139 aaronmk
        self._savepoint += 1
373 2930 aaronmk
        try: return func()
374 2139 aaronmk
        except:
375 2443 aaronmk
            self.run_query('ROLLBACK TO SAVEPOINT '+savepoint, log_level=4)
376 2139 aaronmk
            raise
377 2930 aaronmk
        finally:
378
            # Always release savepoint, because after ROLLBACK TO SAVEPOINT,
379
            # "The savepoint remains valid and can be rolled back to again"
380
            # (http://www.postgresql.org/docs/8.3/static/sql-rollback-to.html).
381 2443 aaronmk
            self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
382 2930 aaronmk
383
            self._savepoint -= 1
384
            assert self._savepoint >= 0
385
386
            self.do_autocommit() # OK to do this after ROLLBACK TO SAVEPOINT
387 2191 aaronmk
388
    def do_autocommit(self):
389
        '''Autocommits if outside savepoint'''
390
        assert self._savepoint >= 0
391
        if self.autocommit and self._savepoint == 0:
392 2924 aaronmk
            self.log_debug('Autocommitting', level=4)
393 2191 aaronmk
            self.db.commit()
394 2643 aaronmk
395 2819 aaronmk
    def col_info(self, col):
396 2643 aaronmk
        table = sql_gen.Table('columns', 'information_schema')
397 3063 aaronmk
        type_ = sql_gen.Coalesce(sql_gen.Nullif(sql_gen.Col('data_type'),
398
            'USER-DEFINED'), sql_gen.Col('udt_name'))
399
        cols = [type_, 'column_default', cast(self, 'boolean', 'is_nullable')]
400 2643 aaronmk
401
        conds = [('table_name', col.table.name), ('column_name', col.name)]
402
        schema = col.table.schema
403
        if schema != None: conds.append(('table_schema', schema))
404
405 2819 aaronmk
        type_, default, nullable = row(select(self, table, cols, conds,
406 3059 aaronmk
            order_by='table_schema', limit=1, cacheable=False, log_level=4))
407 2643 aaronmk
            # TODO: order_by search_path schema order
408 2819 aaronmk
        default = sql_gen.as_Code(default, self)
409
410
        return sql_gen.TypedCol(col.name, type_, default, nullable)
411 2917 aaronmk
412
    def TempFunction(self, name):
413
        if self.debug_temp: schema = None
414
        else: schema = 'pg_temp'
415
        return sql_gen.Function(name, schema)
416 1849 aaronmk
417 1869 aaronmk
connect = DbConn
418
419 832 aaronmk
##### Recoverable querying
420 15 aaronmk
421 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
422 11 aaronmk
423 2791 aaronmk
def run_query(db, query, recover=None, cacheable=False, log_level=2,
424
    log_ignore_excs=None, **kw_args):
425 2794 aaronmk
    '''For params, see DbConn.run_query()'''
426 830 aaronmk
    if recover == None: recover = False
427 2464 aaronmk
    if log_ignore_excs == None: log_ignore_excs = ()
428
    log_ignore_excs = tuple(log_ignore_excs)
429 830 aaronmk
430 2666 aaronmk
    debug_msg_ref = None # usually, db.run_query() logs query before running it
431
    # But if filtering with log_ignore_excs, wait until after exception parsing
432 2984 aaronmk
    if log_ignore_excs != () or not db.can_mogrify(): debug_msg_ref = [None]
433 2666 aaronmk
434 2148 aaronmk
    try:
435 2464 aaronmk
        try:
436 2794 aaronmk
            def run(): return db.run_query(query, cacheable, log_level,
437 2793 aaronmk
                debug_msg_ref, **kw_args)
438 2796 aaronmk
            if recover and not db.is_cached(query):
439 2464 aaronmk
                return with_savepoint(db, run)
440
            else: return run() # don't need savepoint if cached
441
        except Exception, e:
442
            msg = exc.str_(e)
443
444
            match = re.search(r'duplicate key value violates unique constraint '
445 2493 aaronmk
                r'"((_?[^\W_]+)_.+?)"', msg)
446 2464 aaronmk
            if match:
447
                constraint, table = match.groups()
448 3025 aaronmk
                cols = []
449
                if recover: # need auto-rollback to run index_cols()
450
                    try: cols = index_cols(db, table, constraint)
451
                    except NotImplementedError: pass
452
                raise DuplicateKeyException(constraint, cols, e)
453 2464 aaronmk
454 2493 aaronmk
            match = re.search(r'null value in column "(.+?)" violates not-null'
455 2464 aaronmk
                r' constraint', msg)
456
            if match: raise NullValueException('NOT NULL', [match.group(1)], e)
457
458
            match = re.search(r'\b(?:invalid input (?:syntax|value)\b.*?'
459
                r'|date/time field value out of range): "(.+?)"\n'
460 2535 aaronmk
                r'(?:(?s).*?)\bfunction "(.+?)"', msg)
461 2464 aaronmk
            if match:
462
                value, name = match.groups()
463
                raise FunctionValueException(name, strings.to_unicode(value), e)
464
465 2526 aaronmk
            match = re.search(r'column "(.+?)" is of type (.+?) but expression '
466 2523 aaronmk
                r'is of type', msg)
467
            if match:
468
                col, type_ = match.groups()
469
                raise MissingCastException(type_, col, e)
470
471 2946 aaronmk
            match = re.search(r'\b(\S+) "(.+?)".*? already exists', msg)
472 2945 aaronmk
            if match:
473
                type_, name = match.groups()
474
                raise DuplicateException(type_, name, e)
475 2464 aaronmk
476
            raise # no specific exception raised
477
    except log_ignore_excs:
478
        log_level += 2
479
        raise
480
    finally:
481 2666 aaronmk
        if debug_msg_ref != None and debug_msg_ref[0] != None:
482
            db.log_debug(debug_msg_ref[0], log_level)
483 830 aaronmk
484 832 aaronmk
##### Basic queries
485
486 2153 aaronmk
def next_version(name):
487 2163 aaronmk
    version = 1 # first existing name was version 0
488 2586 aaronmk
    match = re.match(r'^(.*)#(\d+)$', name)
489 2153 aaronmk
    if match:
490 2586 aaronmk
        name, version = match.groups()
491
        version = int(version)+1
492 2932 aaronmk
    return sql_gen.concat(name, '#'+str(version))
493 2153 aaronmk
494 2899 aaronmk
def lock_table(db, table, mode):
495
    table = sql_gen.as_Table(table)
496
    run_query(db, 'LOCK TABLE '+table.to_str(db)+' IN '+mode+' MODE')
497
498 2789 aaronmk
def run_query_into(db, query, into=None, add_indexes_=False, **kw_args):
499 2085 aaronmk
    '''Outputs a query to a temp table.
500
    For params, see run_query().
501
    '''
502 2789 aaronmk
    if into == None: return run_query(db, query, **kw_args)
503 2790 aaronmk
504
    assert isinstance(into, sql_gen.Table)
505
506 2992 aaronmk
    into.is_temp = True
507 3008 aaronmk
    # "temporary tables cannot specify a schema name", so remove schema
508
    into.schema = None
509 2992 aaronmk
510 2790 aaronmk
    kw_args['recover'] = True
511 2945 aaronmk
    kw_args.setdefault('log_ignore_excs', (DuplicateException,))
512 2790 aaronmk
513 2916 aaronmk
    temp = not db.debug_temp # tables are permanent in debug_temp mode
514 2790 aaronmk
515
    # Create table
516
    while True:
517
        create_query = 'CREATE'
518
        if temp: create_query += ' TEMP'
519
        create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
520 2385 aaronmk
521 2790 aaronmk
        try:
522
            cur = run_query(db, create_query, **kw_args)
523
                # CREATE TABLE AS sets rowcount to # rows in query
524
            break
525 2945 aaronmk
        except DuplicateException, e:
526 2790 aaronmk
            into.name = next_version(into.name)
527
            # try again with next version of name
528
529
    if add_indexes_: add_indexes(db, into)
530
531
    return cur
532 2085 aaronmk
533 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
534
535 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
536
537 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
538 2293 aaronmk
    start=None, order_by=order_by_pkey, default_table=None):
539 1981 aaronmk
    '''
540 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
541 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
542 1981 aaronmk
    @param fields Use None to select all fields in the table
543 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
544 2379 aaronmk
        * container can be any iterable type
545 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
546
        * compare_right_side: sql_gen.ValueCond|literal value
547 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
548
        use all columns
549 2786 aaronmk
    @return query
550 1981 aaronmk
    '''
551 2315 aaronmk
    # Parse tables param
552 2964 aaronmk
    tables = lists.mk_seq(tables)
553 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
554 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
555 2121 aaronmk
556 2315 aaronmk
    # Parse other params
557 2376 aaronmk
    if conds == None: conds = []
558 2650 aaronmk
    elif dicts.is_dict(conds): conds = conds.items()
559 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
560 135 aaronmk
    assert limit == None or type(limit) == int
561 865 aaronmk
    assert start == None or type(start) == int
562 2315 aaronmk
    if order_by is order_by_pkey:
563
        if distinct_on != []: order_by = None
564
        else: order_by = pkey(db, table0, recover=True)
565 865 aaronmk
566 2315 aaronmk
    query = 'SELECT'
567 2056 aaronmk
568 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
569 2056 aaronmk
570 2200 aaronmk
    # DISTINCT ON columns
571 2233 aaronmk
    if distinct_on != []:
572 2467 aaronmk
        query += '\nDISTINCT'
573 2254 aaronmk
        if distinct_on is not distinct_on_all:
574 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
575
576
    # Columns
577 3027 aaronmk
    if fields == None:
578
        if query.find('\n') >= 0: whitespace = '\n'
579
        else: whitespace = ' '
580
        query += whitespace+'*'
581 2765 aaronmk
    else:
582
        assert fields != []
583 3027 aaronmk
        query += '\n'+('\n, '.join(map(parse_col, fields)))
584 2200 aaronmk
585
    # Main table
586 2467 aaronmk
    query += '\nFROM '+table0.to_str(db)
587 865 aaronmk
588 2122 aaronmk
    # Add joins
589 2271 aaronmk
    left_table = table0
590 2263 aaronmk
    for join_ in tables:
591
        table = join_.table
592 2238 aaronmk
593 2343 aaronmk
        # Parse special values
594
        if join_.type_ is sql_gen.filter_out: # filter no match
595 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
596 2853 aaronmk
                sql_gen.CompareCond(None, '~=')))
597 2343 aaronmk
598 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
599 2122 aaronmk
600
        left_table = table
601
602 865 aaronmk
    missing = True
603 2376 aaronmk
    if conds != []:
604 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
605
        else: whitespace = '\n'
606 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
607
            .to_str(db) for l, r in conds], 'WHERE')
608 865 aaronmk
        missing = False
609 2227 aaronmk
    if order_by != None:
610 2467 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by, table0).to_str(db)
611
    if limit != None: query += '\nLIMIT '+str(limit); missing = False
612 865 aaronmk
    if start != None:
613 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
614 865 aaronmk
        missing = False
615
    if missing: warnings.warn(DbWarning(
616
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
617
618 2786 aaronmk
    return query
619 11 aaronmk
620 2054 aaronmk
def select(db, *args, **kw_args):
621
    '''For params, see mk_select() and run_query()'''
622
    recover = kw_args.pop('recover', None)
623
    cacheable = kw_args.pop('cacheable', True)
624 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
625 2054 aaronmk
626 2791 aaronmk
    return run_query(db, mk_select(db, *args, **kw_args), recover, cacheable,
627
        log_level=log_level)
628 2054 aaronmk
629 2788 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, returning=None,
630 3009 aaronmk
    embeddable=False, ignore=False):
631 1960 aaronmk
    '''
632
    @param returning str|None An inserted column (such as pkey) to return
633 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
634 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
635
        query will be fully cached, not just if it raises an exception.
636 3009 aaronmk
    @param ignore Whether to ignore duplicate keys.
637 1960 aaronmk
    '''
638 2754 aaronmk
    table = sql_gen.remove_table_rename(sql_gen.as_Table(table))
639 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
640 3010 aaronmk
    if cols != None: cols = [sql_gen.to_name_only_col(c, table) for c in cols]
641 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
642 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
643 2063 aaronmk
644 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
645 2063 aaronmk
646 3009 aaronmk
    def mk_insert(select_query):
647
        query = first_line
648 3014 aaronmk
        if cols != None:
649
            query += '\n('+(', '.join((c.to_str(db) for c in cols)))+')'
650 3009 aaronmk
        query += '\n'+select_query
651
652
        if returning != None:
653
            returning_name_col = sql_gen.to_name_only_col(returning)
654
            query += '\nRETURNING '+returning_name_col.to_str(db)
655
656
        return query
657 2063 aaronmk
658 3017 aaronmk
    return_type = 'unknown'
659
    if returning != None: return_type = returning.to_str(db)+'%TYPE'
660
661 3009 aaronmk
    lang = 'sql'
662
    if ignore:
663 3017 aaronmk
        assert cols != None
664
        # Always return something to set the correct rowcount
665
        if returning == None: returning = sql_gen.NamedCol('NULL', None)
666
667 3009 aaronmk
        embeddable = True # must use function
668
        lang = 'plpgsql'
669 3014 aaronmk
        row = [sql_gen.Col(c.name, 'row') for c in cols]
670 3010 aaronmk
671 3009 aaronmk
        query = '''\
672 3010 aaronmk
DECLARE
673 3014 aaronmk
    row '''+table.to_str(db)+'''%ROWTYPE;
674 3009 aaronmk
BEGIN
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 3034 aaronmk
    FOR '''+(', '.join((c.to_str(db) for c in row)))+''' IN
684
'''+select_query+'''
685
    LOOP
686 3015 aaronmk
        BEGIN
687 3019 aaronmk
            RETURN QUERY
688 3014 aaronmk
'''+mk_insert(sql_gen.Values(row).to_str(db))+'''
689 3010 aaronmk
;
690 3015 aaronmk
        EXCEPTION
691 3019 aaronmk
            WHEN unique_violation THEN NULL; -- continue to next row
692 3015 aaronmk
        END;
693 3010 aaronmk
    END LOOP;
694
END;\
695 3009 aaronmk
'''
696
    else: query = mk_insert(select_query)
697
698 2070 aaronmk
    if embeddable:
699
        # Create function
700 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
701 2189 aaronmk
        while True:
702
            try:
703 2918 aaronmk
                function = db.TempFunction(function_name)
704 2194 aaronmk
705 2189 aaronmk
                function_query = '''\
706 2698 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''()
707 3017 aaronmk
RETURNS SETOF '''+return_type+'''
708 3009 aaronmk
LANGUAGE '''+lang+'''
709 2467 aaronmk
AS $$
710 3009 aaronmk
'''+query+'''
711 2467 aaronmk
$$;
712 2070 aaronmk
'''
713 2446 aaronmk
                run_query(db, function_query, recover=True, cacheable=True,
714 2945 aaronmk
                    log_ignore_excs=(DuplicateException,))
715 2189 aaronmk
                break # this version was successful
716 2945 aaronmk
            except DuplicateException, e:
717 2189 aaronmk
                function_name = next_version(function_name)
718
                # try again with next version of name
719 2070 aaronmk
720 2337 aaronmk
        # Return query that uses function
721 3009 aaronmk
        cols = None
722
        if returning != None: cols = [returning]
723 2698 aaronmk
        func_table = sql_gen.NamedTable('f', sql_gen.FunctionCall(function),
724 3009 aaronmk
            cols) # AS clause requires function alias
725 2787 aaronmk
        return mk_select(db, func_table, start=0, order_by=None)
726 2070 aaronmk
727 2787 aaronmk
    return query
728 2066 aaronmk
729
def insert_select(db, *args, **kw_args):
730 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
731 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
732
        values in
733 2072 aaronmk
    '''
734 2386 aaronmk
    into = kw_args.pop('into', None)
735
    if into != None: kw_args['embeddable'] = True
736 2066 aaronmk
    recover = kw_args.pop('recover', None)
737 3011 aaronmk
    if kw_args.get('ignore', False): recover = True
738 2066 aaronmk
    cacheable = kw_args.pop('cacheable', True)
739 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
740 2066 aaronmk
741 2789 aaronmk
    return run_query_into(db, mk_insert_select(db, *args, **kw_args), into,
742 2787 aaronmk
        recover=recover, cacheable=cacheable, log_level=log_level)
743 2063 aaronmk
744 2738 aaronmk
default = sql_gen.default # tells insert() to use the default value for a column
745 2066 aaronmk
746 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
747 2085 aaronmk
    '''For params, see insert_select()'''
748 1960 aaronmk
    if lists.is_seq(row): cols = None
749
    else:
750
        cols = row.keys()
751
        row = row.values()
752 2738 aaronmk
    row = list(row) # ensure that "== []" works
753 1960 aaronmk
754 2738 aaronmk
    if row == []: query = None
755
    else: query = sql_gen.Values(row).to_str(db)
756 1961 aaronmk
757 2788 aaronmk
    return insert_select(db, table, cols, query, *args, **kw_args)
758 11 aaronmk
759 3056 aaronmk
def mk_update(db, table, changes=None, cond=None, in_place=False):
760 2402 aaronmk
    '''
761
    @param changes [(col, new_value),...]
762
        * container can be any iterable type
763
        * col: sql_gen.Code|str (for col name)
764
        * new_value: sql_gen.Code|literal value
765
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
766 3056 aaronmk
    @param in_place If set, locks the table and updates rows in place.
767
        This avoids creating dead rows in PostgreSQL.
768
        * cond must be None
769 2402 aaronmk
    @return str query
770
    '''
771 3057 aaronmk
    table = sql_gen.as_Table(table)
772
    changes = [(sql_gen.to_name_only_col(c, table), sql_gen.as_Value(v))
773
        for c, v in changes]
774
775 3056 aaronmk
    if in_place:
776
        assert cond == None
777 3058 aaronmk
778 3065 aaronmk
        query = 'ALTER TABLE '+table.to_str(db)+'\n'
779
        query += ',\n'.join(('ALTER COLUMN '+c.to_str(db)+' TYPE '
780
            +db.col_info(sql_gen.with_default_table(c, table)).type
781
            +'\nUSING '+v.to_str(db) for c, v in changes))
782 3058 aaronmk
    else:
783
        query = 'UPDATE '+table.to_str(db)+'\nSET\n'
784
        query += ',\n'.join((c.to_str(db)+' = '+v.to_str(db)
785
            for c, v in changes))
786
        if cond != None: query += '\nWHERE\n'+cond.to_str(db)
787 3056 aaronmk
788 2402 aaronmk
    return query
789
790
def update(db, *args, **kw_args):
791
    '''For params, see mk_update() and run_query()'''
792
    recover = kw_args.pop('recover', None)
793 3043 aaronmk
    cacheable = kw_args.pop('cacheable', False)
794 3030 aaronmk
    log_level = kw_args.pop('log_level', 2)
795 2402 aaronmk
796 3043 aaronmk
    return run_query(db, mk_update(db, *args, **kw_args), recover, cacheable,
797 3030 aaronmk
        log_level=log_level)
798 2402 aaronmk
799 135 aaronmk
def last_insert_id(db):
800 1849 aaronmk
    module = util.root_module(db.db)
801 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
802
    elif module == 'MySQLdb': return db.insert_id()
803
    else: return None
804 13 aaronmk
805 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
806 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
807 2415 aaronmk
    to names that will be distinct among the columns' tables.
808 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
809 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
810
    @param into The table for the new columns.
811 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
812
        columns will be included in the mapping even if they are not in cols.
813
        The tables of the provided Col objects will be changed to into, so make
814
        copies of them if you want to keep the original tables.
815
    @param as_items Whether to return a list of dict items instead of a dict
816 2383 aaronmk
    @return dict(orig_col=new_col, ...)
817
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
818 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
819
        * All mappings use the into table so its name can easily be
820 2383 aaronmk
          changed for all columns at once
821
    '''
822 2415 aaronmk
    cols = lists.uniqify(cols)
823
824 2394 aaronmk
    items = []
825 2389 aaronmk
    for col in preserve:
826 2390 aaronmk
        orig_col = copy.copy(col)
827 2392 aaronmk
        col.table = into
828 2394 aaronmk
        items.append((orig_col, col))
829
    preserve = set(preserve)
830
    for col in cols:
831 2716 aaronmk
        if col not in preserve:
832
            items.append((col, sql_gen.Col(str(col), into, col.srcs)))
833 2394 aaronmk
834
    if not as_items: items = dict(items)
835
    return items
836 2383 aaronmk
837 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
838 2391 aaronmk
    '''For params, see mk_flatten_mapping()
839
    @return See return value of mk_flatten_mapping()
840
    '''
841 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
842
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
843 2786 aaronmk
    run_query_into(db, mk_select(db, joins, cols, limit=limit, start=start),
844 2846 aaronmk
        into=into, add_indexes_=True)
845 2394 aaronmk
    return dict(items)
846 2391 aaronmk
847 3050 aaronmk
def track_data_error(db, errors_table, cols, value, error_code, error):
848 2758 aaronmk
    '''
849
    @param errors_table If None, does nothing.
850
    '''
851 2810 aaronmk
    if errors_table == None or cols == (): return
852 3050 aaronmk
853
    for col in cols:
854
        try:
855
            insert(db, errors_table, dict(column=col.name, value=value,
856
                error_code=error_code, error=error), recover=True,
857
                cacheable=True, log_level=4)
858
        except DuplicateKeyException: pass
859 2758 aaronmk
860 2704 aaronmk
def cast(db, type_, col, errors_table=None):
861
    '''Casts an (unrenamed) column or value.
862 2769 aaronmk
    If errors_table set and col has srcs, saves errors in errors_table (using
863
    col's srcs attr as the source columns) and converts errors to warnings.
864 2820 aaronmk
    @param col str|sql_gen.Col|sql_gen.Literal
865 2704 aaronmk
    @param errors_table None|sql_gen.Table|str
866
    '''
867 2820 aaronmk
    col = sql_gen.as_Col(col)
868 2769 aaronmk
    save_errors = (errors_table != None and isinstance(col, sql_gen.Col)
869
        and col.srcs != ())
870 2987 aaronmk
    if not save_errors: return sql_gen.Cast(type_, col) # can't save errors
871 2704 aaronmk
872
    assert not isinstance(col, sql_gen.NamedCol)
873
874 2770 aaronmk
    errors_table = sql_gen.as_Table(errors_table)
875
    srcs = map(sql_gen.to_name_only_col, col.srcs)
876
    function_name = str(sql_gen.FunctionCall(type_, *srcs))
877 2918 aaronmk
    function = db.TempFunction(function_name)
878 2704 aaronmk
879 2718 aaronmk
    while True:
880
        # Create function definition
881 3049 aaronmk
        errors_table_cols = map(sql_gen.Col,
882
            ['column', 'value', 'error_code', 'error'])
883 2718 aaronmk
        query = '''\
884 2704 aaronmk
CREATE FUNCTION '''+function.to_str(db)+'''(value text)
885
RETURNS '''+type_+'''
886
LANGUAGE plpgsql
887
STRICT
888
AS $$
889
BEGIN
890
    /* The explicit cast to the return type is needed to make the cast happen
891
    inside the try block. (Implicit casts to the return type happen at the end
892
    of the function, outside any block.) */
893
    RETURN value::'''+type_+''';
894
EXCEPTION
895
    WHEN data_exception THEN
896
        -- Save error in errors table.
897 3049 aaronmk
        DECLARE
898
            error_code text := SQLSTATE;
899
            error text := SQLERRM;
900
            "column" text;
901
        BEGIN
902
            -- Insert the value and error for *each* source column.
903
            FOR "column" IN
904
'''+mk_select(db, sql_gen.NamedValues('c', None, [[c.name] for c in srcs]),
905 3053 aaronmk
    order_by=None, start=0)+'''
906 3049 aaronmk
            LOOP
907
                BEGIN
908
'''+mk_insert_select(db, errors_table, errors_table_cols,
909
    sql_gen.Values(errors_table_cols).to_str(db))+''';
910
                EXCEPTION
911
                    WHEN unique_violation THEN NULL; -- continue to next row
912
                END;
913
            END LOOP;
914
        END;
915 2704 aaronmk
916
        RAISE WARNING '%', SQLERRM;
917
        RETURN NULL;
918
END;
919
$$;
920
'''
921 2718 aaronmk
922
        # Create function
923
        try:
924
            run_query(db, query, recover=True, cacheable=True,
925 2945 aaronmk
                log_ignore_excs=(DuplicateException,))
926 2718 aaronmk
            break # successful
927 2945 aaronmk
        except DuplicateException:
928 2770 aaronmk
            function.name = next_version(function.name)
929
            # try again with next version of name
930 2704 aaronmk
931
    return sql_gen.FunctionCall(function, col)
932
933 2414 aaronmk
##### Database structure queries
934
935 2426 aaronmk
def table_row_count(db, table, recover=None):
936 2786 aaronmk
    return value(run_query(db, mk_select(db, table, [sql_gen.row_count],
937 2443 aaronmk
        order_by=None, start=0), recover=recover, log_level=3))
938 2426 aaronmk
939 2414 aaronmk
def table_cols(db, table, recover=None):
940
    return list(col_names(select(db, table, limit=0, order_by=None,
941 2443 aaronmk
        recover=recover, log_level=4)))
942 2414 aaronmk
943 2291 aaronmk
def pkey(db, table, recover=None):
944 832 aaronmk
    '''Assumed to be first column in table'''
945 2339 aaronmk
    return table_cols(db, table, recover)[0]
946 832 aaronmk
947 2559 aaronmk
not_null_col = 'not_null_col'
948 2340 aaronmk
949
def table_not_null_col(db, table, recover=None):
950
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
951
    if not_null_col in table_cols(db, table, recover): return not_null_col
952
    else: return pkey(db, table, recover)
953
954 853 aaronmk
def index_cols(db, table, index):
955
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
956
    automatically created. When you don't know whether something is a UNIQUE
957
    constraint or a UNIQUE index, use this function.'''
958 1909 aaronmk
    module = util.root_module(db.db)
959
    if module == 'psycopg2':
960
        return list(values(run_query(db, '''\
961 853 aaronmk
SELECT attname
962 866 aaronmk
FROM
963
(
964
        SELECT attnum, attname
965
        FROM pg_index
966
        JOIN pg_class index ON index.oid = indexrelid
967
        JOIN pg_class table_ ON table_.oid = indrelid
968
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
969
        WHERE
970 2782 aaronmk
            table_.relname = '''+db.esc_value(table)+'''
971
            AND index.relname = '''+db.esc_value(index)+'''
972 866 aaronmk
    UNION
973
        SELECT attnum, attname
974
        FROM
975
        (
976
            SELECT
977
                indrelid
978
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
979
                    AS indkey
980
            FROM pg_index
981
            JOIN pg_class index ON index.oid = indexrelid
982
            JOIN pg_class table_ ON table_.oid = indrelid
983
            WHERE
984 2782 aaronmk
                table_.relname = '''+db.esc_value(table)+'''
985
                AND index.relname = '''+db.esc_value(index)+'''
986 866 aaronmk
        ) s
987
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
988
) s
989 853 aaronmk
ORDER BY attnum
990 2782 aaronmk
'''
991
            , cacheable=True, log_level=4)))
992 1909 aaronmk
    else: raise NotImplementedError("Can't list index columns for "+module+
993
        ' database')
994 853 aaronmk
995 464 aaronmk
def constraint_cols(db, table, constraint):
996 1849 aaronmk
    module = util.root_module(db.db)
997 464 aaronmk
    if module == 'psycopg2':
998
        return list(values(run_query(db, '''\
999
SELECT attname
1000
FROM pg_constraint
1001
JOIN pg_class ON pg_class.oid = conrelid
1002
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
1003
WHERE
1004 2783 aaronmk
    relname = '''+db.esc_value(table)+'''
1005
    AND conname = '''+db.esc_value(constraint)+'''
1006 464 aaronmk
ORDER BY attnum
1007 2783 aaronmk
'''
1008
            )))
1009 464 aaronmk
    else: raise NotImplementedError("Can't list constraint columns for "+module+
1010
        ' database')
1011
1012 2096 aaronmk
row_num_col = '_row_num'
1013
1014 2998 aaronmk
def add_index(db, exprs, table=None, unique=False, ensure_not_null_=True):
1015 2688 aaronmk
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
1016 2538 aaronmk
    Currently, only function calls are supported as expressions.
1017 2998 aaronmk
    @param ensure_not_null_ If set, translates NULL values to sentinel values.
1018 2847 aaronmk
        This allows indexes to be used for comparisons where NULLs are equal.
1019 2538 aaronmk
    '''
1020 2964 aaronmk
    exprs = lists.mk_seq(exprs)
1021 2538 aaronmk
1022 2688 aaronmk
    # Parse exprs
1023
    old_exprs = exprs[:]
1024
    exprs = []
1025
    cols = []
1026
    for i, expr in enumerate(old_exprs):
1027 2823 aaronmk
        expr = sql_gen.as_Col(expr, table)
1028 2688 aaronmk
1029 2823 aaronmk
        # Handle nullable columns
1030 2998 aaronmk
        if ensure_not_null_:
1031
            try: expr = ensure_not_null(db, expr)
1032 2860 aaronmk
            except KeyError: pass # unknown type, so just create plain index
1033 2823 aaronmk
1034 2688 aaronmk
        # Extract col
1035 3002 aaronmk
        expr = copy.deepcopy(expr) # don't modify input!
1036 2688 aaronmk
        if isinstance(expr, sql_gen.FunctionCall):
1037
            col = expr.args[0]
1038
            expr = sql_gen.Expr(expr)
1039
        else: col = expr
1040 2823 aaronmk
        assert isinstance(col, sql_gen.Col)
1041 2688 aaronmk
1042
        # Extract table
1043
        if table == None:
1044
            assert sql_gen.is_table_col(col)
1045
            table = col.table
1046
1047
        col.table = None
1048
1049
        exprs.append(expr)
1050
        cols.append(col)
1051 2408 aaronmk
1052 2688 aaronmk
    table = sql_gen.as_Table(table)
1053
    index = sql_gen.Table(str(sql_gen.Col(','.join(map(str, cols)), table)))
1054
1055 3005 aaronmk
    # Add index
1056
    while True:
1057
        str_ = 'CREATE'
1058
        if unique: str_ += ' UNIQUE'
1059
        str_ += ' INDEX '+index.to_str(db)+' ON '+table.to_str(db)+' ('+(
1060
            ', '.join((v.to_str(db) for v in exprs)))+')'
1061
1062
        try:
1063
            run_query(db, str_, recover=True, cacheable=True, log_level=3,
1064
                log_ignore_excs=(DuplicateException,))
1065
            break
1066
        except DuplicateException:
1067
            index.name = next_version(index.name)
1068
            # try again with next version of name
1069 2408 aaronmk
1070 2686 aaronmk
def add_pkey(db, table, cols=None, recover=None):
1071
    '''Adds a primary key.
1072
    @param cols [sql_gen.Col,...] The columns in the primary key.
1073
        Defaults to the first column in the table.
1074 2406 aaronmk
    @pre The table must not already have a primary key.
1075
    '''
1076
    table = sql_gen.as_Table(table)
1077 2686 aaronmk
    if cols == None: cols = [pkey(db, table, recover)]
1078
    col_strs = [sql_gen.to_name_only_col(v).to_str(db) for v in cols]
1079 2406 aaronmk
1080 2781 aaronmk
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD PRIMARY KEY ('
1081
        +(', '.join(col_strs))+')', recover=True, cacheable=True, log_level=3,
1082 2945 aaronmk
        log_ignore_excs=(DuplicateException,))
1083 2406 aaronmk
1084 2997 aaronmk
def add_not_null(db, col):
1085
    table = col.table
1086
    col = sql_gen.to_name_only_col(col)
1087
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ALTER COLUMN '
1088 3031 aaronmk
        +col.to_str(db)+' SET NOT NULL', cacheable=True, log_level=3)
1089 2997 aaronmk
1090
def add_index_col(db, col, suffix, expr, nullable=True):
1091 3000 aaronmk
    if sql_gen.index_col(col) != None: return # already has index col
1092 2997 aaronmk
1093
    new_col = sql_gen.suffixed_col(col, suffix)
1094
1095 3006 aaronmk
    # Add column
1096 3038 aaronmk
    new_typed_col = sql_gen.TypedCol(new_col.name, db.col_info(col).type)
1097 3045 aaronmk
    add_col(db, col.table, new_typed_col, comment='src: '+repr(col),
1098
        log_level=3)
1099 3037 aaronmk
    new_col.name = new_typed_col.name # propagate any renaming
1100 3006 aaronmk
1101 3064 aaronmk
    update(db, col.table, [(new_col, expr)], in_place=True, cacheable=True,
1102
        log_level=3)
1103 2997 aaronmk
    if not nullable: add_not_null(db, new_col)
1104
    add_index(db, new_col)
1105
1106 3000 aaronmk
    col.table.index_cols[col.name] = new_col
1107 2997 aaronmk
1108 3047 aaronmk
# Controls when ensure_not_null() will use index columns
1109
not_null_index_cols_min_rows = 0 # rows; initially always use index columns
1110
1111 2997 aaronmk
def ensure_not_null(db, col):
1112
    '''For params, see sql_gen.ensure_not_null()'''
1113
    expr = sql_gen.ensure_not_null(db, col)
1114
1115 3047 aaronmk
    # If a nullable column in a temp table, add separate index column instead.
1116
    # Note that for small datasources, this adds 6-25% to the total import time.
1117
    if (sql_gen.is_temp_col(col) and isinstance(expr, sql_gen.EnsureNotNull)
1118
        and table_row_count(db, col.table) >= not_null_index_cols_min_rows):
1119 2997 aaronmk
        add_index_col(db, col, '::NOT NULL', expr, nullable=False)
1120 3000 aaronmk
        expr = sql_gen.index_col(col)
1121 2997 aaronmk
1122
    return expr
1123
1124 2772 aaronmk
already_indexed = object() # tells add_indexes() the pkey has already been added
1125
1126
def add_indexes(db, table, has_pkey=True):
1127
    '''Adds an index on all columns in a table.
1128
    @param has_pkey bool|already_indexed Whether a pkey instead of a regular
1129
        index should be added on the first column.
1130
        * If already_indexed, the pkey is assumed to have already been added
1131
    '''
1132
    cols = table_cols(db, table)
1133
    if has_pkey:
1134
        if has_pkey is not already_indexed: add_pkey(db, table)
1135
        cols = cols[1:]
1136
    for col in cols: add_index(db, col, table)
1137
1138 3042 aaronmk
def add_col(db, table, col, comment=None, **kw_args):
1139 3039 aaronmk
    '''
1140 3042 aaronmk
    @param col TypedCol Name may be versioned, so be sure to propagate any
1141
        renaming back to any source column for the TypedCol.
1142
    @param comment None|str SQL comment used to distinguish columns of the same
1143
        name from each other when they contain different data, to allow the
1144
        ADD COLUMN query to be cached. If not set, query will not be cached.
1145 3039 aaronmk
    '''
1146 2934 aaronmk
    assert isinstance(col, sql_gen.TypedCol)
1147 3036 aaronmk
1148
    while True:
1149 3042 aaronmk
        str_ = 'ALTER TABLE '+table.to_str(db)+' ADD COLUMN '+col.to_str(db)
1150
        if comment != None: str_ += ' '+sql_gen.esc_comment(comment)
1151
1152 3036 aaronmk
        try:
1153 3042 aaronmk
            run_query(db, str_, recover=True, cacheable=True, **kw_args)
1154 3036 aaronmk
            break
1155
        except DuplicateException:
1156
            col.name = next_version(col.name)
1157
            # try again with next version of name
1158 2934 aaronmk
1159 2935 aaronmk
row_num_typed_col = sql_gen.TypedCol(row_num_col, 'serial', nullable=False,
1160
    constraints='PRIMARY KEY')
1161
1162 2086 aaronmk
def add_row_num(db, table):
1163 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
1164
    be the primary key.'''
1165 2935 aaronmk
    add_col(db, table, row_num_typed_col, log_level=3)
1166 2086 aaronmk
1167 2936 aaronmk
def cast_temp_col(db, type_, col, errors_table=None):
1168
    '''Like cast(), but creates a new column with the cast values if the input
1169
    is a column.
1170
    @return The new column or cast value
1171
    '''
1172
    def cast_(col): return cast(db, type_, col, errors_table)
1173
1174
    try: col = sql_gen.underlying_col(col)
1175
    except sql_gen.NoUnderlyingTableException: return sql_gen.wrap(cast_, col)
1176
1177
    table = col.table
1178
    new_col = sql_gen.Col(sql_gen.concat(col.name, '::'+type_), table, col.srcs)
1179
    expr = cast_(col)
1180 3035 aaronmk
1181
    # Add column
1182 3038 aaronmk
    new_typed_col = sql_gen.TypedCol(new_col.name, type_)
1183 3045 aaronmk
    add_col(db, table, new_typed_col, comment='src: '+repr(col))
1184 3037 aaronmk
    new_col.name = new_typed_col.name # propagate any renaming
1185 3035 aaronmk
1186 3064 aaronmk
    update(db, table, [(new_col, expr)], in_place=True, cacheable=True)
1187 2955 aaronmk
    add_index(db, new_col)
1188 2936 aaronmk
1189
    return new_col
1190
1191 2889 aaronmk
def drop_table(db, table):
1192
    table = sql_gen.as_Table(table)
1193
    return run_query(db, 'DROP TABLE IF EXISTS '+table.to_str(db)+' CASCADE')
1194
1195 2760 aaronmk
def create_table(db, table, cols, has_pkey=True, col_indexes=True):
1196 2675 aaronmk
    '''Creates a table.
1197 2681 aaronmk
    @param cols [sql_gen.TypedCol,...] The column names and types
1198
    @param has_pkey If set, the first column becomes the primary key.
1199 2760 aaronmk
    @param col_indexes bool|[ref]
1200
        * If True, indexes will be added on all non-pkey columns.
1201
        * If a list reference, [0] will be set to a function to do this.
1202
          This can be used to delay index creation until the table is populated.
1203 2675 aaronmk
    '''
1204
    table = sql_gen.as_Table(table)
1205
1206 2681 aaronmk
    if has_pkey:
1207
        cols[0] = pkey = copy.copy(cols[0]) # don't modify input!
1208 2872 aaronmk
        pkey.constraints = 'PRIMARY KEY'
1209 2681 aaronmk
1210 2675 aaronmk
    str_ = 'CREATE TABLE '+table.to_str(db)+' (\n'
1211 2681 aaronmk
    str_ += '\n, '.join(v.to_str(db) for v in cols)
1212
    str_ += '\n);\n'
1213 2675 aaronmk
    run_query(db, str_, cacheable=True, log_level=2)
1214 2760 aaronmk
1215
    # Add indexes
1216 2773 aaronmk
    if has_pkey: has_pkey = already_indexed
1217
    def add_indexes_(): add_indexes(db, table, has_pkey)
1218
    if isinstance(col_indexes, list): col_indexes[0] = add_indexes_ # defer
1219
    elif col_indexes: add_indexes_() # add now
1220 2675 aaronmk
1221 2684 aaronmk
def vacuum(db, table):
1222
    table = sql_gen.as_Table(table)
1223
    db.with_autocommit(lambda: run_query(db, 'VACUUM ANALYZE '+table.to_str(db),
1224
        log_level=3))
1225
1226 2970 aaronmk
def truncate(db, table, schema='public', **kw_args):
1227
    '''For params, see run_query()'''
1228 2777 aaronmk
    table = sql_gen.as_Table(table, schema)
1229 2970 aaronmk
    return run_query(db, 'TRUNCATE '+table.to_str(db)+' CASCADE', **kw_args)
1230 2732 aaronmk
1231 2965 aaronmk
def empty_temp(db, tables):
1232 2972 aaronmk
    if db.debug_temp: return # leave temp tables there for debugging
1233 2965 aaronmk
    tables = lists.mk_seq(tables)
1234 2971 aaronmk
    for table in tables: truncate(db, table, log_level=3)
1235 2965 aaronmk
1236 2729 aaronmk
def tables(db, schema_like='public', table_like='%', exact=False):
1237
    if exact: compare = '='
1238
    else: compare = 'LIKE'
1239
1240 1849 aaronmk
    module = util.root_module(db.db)
1241 832 aaronmk
    if module == 'psycopg2':
1242 2779 aaronmk
        conds = [('schemaname', sql_gen.CompareCond(schema_like, compare)),
1243
            ('tablename', sql_gen.CompareCond(table_like, compare))]
1244
        return values(select(db, 'pg_tables', ['tablename'], conds,
1245
            order_by='tablename', log_level=4))
1246 1968 aaronmk
    elif module == 'MySQLdb':
1247 2779 aaronmk
        return values(run_query(db, 'SHOW TABLES LIKE '+db.esc_value(table_like)
1248
            , cacheable=True, log_level=4))
1249 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
1250 830 aaronmk
1251 2726 aaronmk
def table_exists(db, table):
1252
    table = sql_gen.as_Table(table)
1253 2730 aaronmk
    return list(tables(db, table.schema, table.name, exact=True)) != []
1254 2726 aaronmk
1255 2913 aaronmk
def function_exists(db, function):
1256
    function = sql_gen.as_Function(function)
1257
1258
    info_table = sql_gen.Table('routines', 'information_schema')
1259
    conds = [('routine_name', function.name)]
1260
    schema = function.schema
1261
    if schema != None: conds.append(('routine_schema', schema))
1262 2914 aaronmk
    # Exclude trigger functions, since they cannot be called directly
1263
    conds.append(('data_type', sql_gen.CompareCond('trigger', '!=')))
1264 2913 aaronmk
1265
    return list(values(select(db, info_table, ['routine_name'], conds,
1266
        order_by='routine_schema', limit=1, log_level=4))) != []
1267
        # TODO: order_by search_path schema order
1268
1269 2726 aaronmk
def errors_table(db, table, if_exists=True):
1270
    '''
1271
    @param if_exists If set, returns None if the errors table doesn't exist
1272
    @return None|sql_gen.Table
1273
    '''
1274
    table = sql_gen.as_Table(table)
1275
    if table.srcs != (): table = table.srcs[0]
1276
1277
    errors_table = sql_gen.suffixed_table(table, '.errors')
1278
    if if_exists and not table_exists(db, errors_table): return None
1279
    return errors_table
1280
1281 833 aaronmk
##### Database management
1282
1283 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
1284
    '''For kw_args, see tables()'''
1285
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
1286 833 aaronmk
1287 832 aaronmk
##### Heuristic queries
1288
1289 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
1290 1554 aaronmk
    '''Recovers from errors.
1291 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
1292
    '''
1293 2642 aaronmk
    row = sql_gen.ColDict(db, table, row)
1294 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
1295
1296 471 aaronmk
    try:
1297 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
1298 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
1299
            row_ct_ref[0] += cur.rowcount
1300
        return value(cur)
1301 471 aaronmk
    except DuplicateKeyException, e:
1302 2653 aaronmk
        row = sql_gen.ColDict(db, table,
1303
            util.dict_subset_right_join(row, e.cols))
1304
        return value(select(db, table, [pkey_], row, recover=True))
1305 471 aaronmk
1306 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
1307 830 aaronmk
    '''Recovers from errors'''
1308 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
1309 14 aaronmk
    except StopIteration:
1310 40 aaronmk
        if not create: raise
1311 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
1312 2078 aaronmk
1313 2593 aaronmk
def is_func_result(col):
1314
    return col.table.name.find('(') >= 0 and col.name == 'result'
1315
1316 2592 aaronmk
def into_table_name(out_table, in_tables0, mapping, is_func):
1317 2615 aaronmk
    def in_col_str(in_col):
1318
        in_col = sql_gen.remove_col_rename(in_col)
1319
        if isinstance(in_col, sql_gen.Col):
1320
            table = in_col.table
1321
            if table == in_tables0:
1322
                in_col = sql_gen.to_name_only_col(in_col)
1323
            elif is_func_result(in_col): in_col = table # omit col name
1324
        return str(in_col)
1325
1326 2580 aaronmk
    str_ = str(out_table)
1327
    if is_func:
1328 2615 aaronmk
        str_ += '('
1329 2580 aaronmk
1330 2615 aaronmk
        try: value_in_col = mapping['value']
1331
        except KeyError:
1332 2654 aaronmk
            str_ += ', '.join((str(k)+'='+in_col_str(v)
1333 2615 aaronmk
                for k, v in mapping.iteritems()))
1334
        else: str_ += in_col_str(value_in_col)
1335
1336
        str_ += ')'
1337 2812 aaronmk
    else:
1338
        out_col = 'rank'
1339
        try: in_col = mapping[out_col]
1340
        except KeyError: str_ += '_pkeys'
1341
        else: # has a rank column, so hierarchical
1342
            str_ += '['+str(out_col)+'='+in_col_str(in_col)+']'
1343 2580 aaronmk
    return str_
1344
1345 2508 aaronmk
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None, into=None,
1346 2805 aaronmk
    default=None, is_func=False, on_error=exc.raise_):
1347 2078 aaronmk
    '''Recovers from errors.
1348
    Only works under PostgreSQL (uses INSERT RETURNING).
1349 2909 aaronmk
    IMPORTANT: Must be run at the *beginning* of a transaction.
1350 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
1351
        tables to join with it using the main input table's pkey
1352 2312 aaronmk
    @param mapping dict(out_table_col=in_table_col, ...)
1353 2612 aaronmk
        * out_table_col: str (*not* sql_gen.Col)
1354
        * in_table_col: sql_gen.Col|literal-value
1355 2489 aaronmk
    @param into The table to contain the output and input pkeys.
1356 2574 aaronmk
        Defaults to `out_table.name+'_pkeys'`.
1357 2509 aaronmk
    @param default The *output* column to use as the pkey for missing rows.
1358
        If this output column does not exist in the mapping, uses None.
1359 2552 aaronmk
    @param is_func Whether out_table is the name of a SQL function, not a table
1360 2312 aaronmk
    @return sql_gen.Col Where the output pkeys are made available
1361 2078 aaronmk
    '''
1362 2329 aaronmk
    out_table = sql_gen.as_Table(out_table)
1363 2312 aaronmk
1364 2450 aaronmk
    def log_debug(msg): db.log_debug(msg, level=1.5)
1365 2505 aaronmk
    def col_ustr(str_):
1366 2567 aaronmk
        return strings.repr_no_u(sql_gen.remove_col_rename(str_))
1367 2450 aaronmk
1368 2910 aaronmk
    log_debug('********** New iteration **********')
1369
    log_debug('Inserting these input columns into '+strings.as_tt(
1370
        out_table.to_str(db))+':\n'+strings.as_table(mapping, ustr=col_ustr))
1371
1372 2942 aaronmk
    is_function = function_exists(db, out_table)
1373
1374
    if is_function: out_pkey = 'result'
1375
    else: out_pkey = pkey(db, out_table, recover=True)
1376 2768 aaronmk
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
1377
1378
    if mapping == {}: # need at least one column for INSERT SELECT
1379
        mapping = {out_pkey: None} # ColDict will replace with default value
1380
1381 2382 aaronmk
    # Create input joins from list of input tables
1382
    in_tables_ = in_tables[:] # don't modify input!
1383
    in_tables0 = in_tables_.pop(0) # first table is separate
1384 2735 aaronmk
    errors_table_ = errors_table(db, in_tables0)
1385 2279 aaronmk
    in_pkey = pkey(db, in_tables0, recover=True)
1386 2285 aaronmk
    in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
1387 2460 aaronmk
    input_joins = [in_tables0]+[sql_gen.Join(v,
1388
        {in_pkey: sql_gen.join_same_not_null}) for v in in_tables_]
1389 2131 aaronmk
1390 2592 aaronmk
    if into == None:
1391
        into = into_table_name(out_table, in_tables0, mapping, is_func)
1392
    into = sql_gen.as_Table(into)
1393
1394 2702 aaronmk
    # Set column sources
1395
    in_cols = filter(sql_gen.is_table_col, mapping.values())
1396
    for col in in_cols:
1397
        if col.table == in_tables0: col.set_srcs(sql_gen.src_self)
1398
1399 2486 aaronmk
    log_debug('Joining together input tables into temp table')
1400 2395 aaronmk
    # Place in new table for speed and so don't modify input if values edited
1401 2584 aaronmk
    in_table = sql_gen.Table('in')
1402 2702 aaronmk
    mapping = dicts.join(mapping, flatten(db, in_table, input_joins, in_cols,
1403
        preserve=[in_pkey_col], start=0))
1404 2395 aaronmk
    input_joins = [in_table]
1405 2486 aaronmk
    db.log_debug('Temp table: '+strings.as_tt(in_table.to_str(db)), level=2)
1406 2395 aaronmk
1407 2656 aaronmk
    mapping = sql_gen.ColDict(db, out_table, mapping)
1408
        # after applying dicts.join() because that returns a plain dict
1409
1410 2509 aaronmk
    # Resolve default value column
1411 2841 aaronmk
    if default != None:
1412
        try: default = mapping[default]
1413
        except KeyError:
1414 2509 aaronmk
            db.log_debug('Default value column '
1415
                +strings.as_tt(strings.repr_no_u(default))
1416 2511 aaronmk
                +' does not exist in mapping, falling back to None', level=2.1)
1417 2509 aaronmk
            default = None
1418
1419 2387 aaronmk
    pkeys_names = [in_pkey, out_pkey]
1420 2236 aaronmk
    pkeys_cols = [in_pkey_col, out_pkey_col]
1421
1422 2201 aaronmk
    pkeys_table_exists_ref = [False]
1423 2960 aaronmk
    def insert_into_pkeys(joins, cols, distinct=False):
1424
        kw_args = {}
1425
        if distinct: kw_args.update(dict(distinct_on=[in_pkey_col]))
1426
        query = mk_select(db, joins, cols, order_by=None, start=0, **kw_args)
1427
1428 2201 aaronmk
        if pkeys_table_exists_ref[0]:
1429 2786 aaronmk
            insert_select(db, into, pkeys_names, query)
1430 2201 aaronmk
        else:
1431 2786 aaronmk
            run_query_into(db, query, into=into)
1432 2201 aaronmk
            pkeys_table_exists_ref[0] = True
1433
1434 2429 aaronmk
    limit_ref = [None]
1435 2380 aaronmk
    conds = set()
1436 2811 aaronmk
    distinct_on = sql_gen.ColDict(db, out_table)
1437 2325 aaronmk
    def mk_main_select(joins, cols):
1438 2811 aaronmk
        distinct_on_cols = [c.to_Col() for c in distinct_on.values()]
1439
        return mk_select(db, joins, cols, conds, distinct_on_cols,
1440 2429 aaronmk
            limit=limit_ref[0], start=0)
1441 2132 aaronmk
1442 2519 aaronmk
    exc_strs = set()
1443 2309 aaronmk
    def log_exc(e):
1444 2519 aaronmk
        e_str = exc.str_(e, first_line_only=True)
1445
        log_debug('Caught exception: '+e_str)
1446
        assert e_str not in exc_strs # avoid infinite loops
1447
        exc_strs.add(e_str)
1448 2735 aaronmk
1449 2451 aaronmk
    def remove_all_rows():
1450 2896 aaronmk
        log_debug('Ignoring all rows')
1451 2429 aaronmk
        limit_ref[0] = 0 # just create an empty pkeys table
1452 2735 aaronmk
1453 2758 aaronmk
    def ignore(in_col, value, e):
1454
        track_data_error(db, errors_table_, in_col.srcs, value, e.cause.pgcode,
1455
            e.cause.pgerror)
1456 2895 aaronmk
        log_debug('Ignoring rows with '+strings.as_tt(repr(in_col))+' = '
1457 2545 aaronmk
            +strings.as_tt(repr(value)))
1458 2895 aaronmk
1459 2758 aaronmk
    def remove_rows(in_col, value, e):
1460
        ignore(in_col, value, e)
1461 2378 aaronmk
        cond = (in_col, sql_gen.CompareCond(value, '!='))
1462
        assert cond not in conds # avoid infinite loops
1463 2380 aaronmk
        conds.add(cond)
1464 2895 aaronmk
1465 2758 aaronmk
    def invalid2null(in_col, value, e):
1466
        ignore(in_col, value, e)
1467 2403 aaronmk
        update(db, in_table, [(in_col, None)],
1468
            sql_gen.ColValueCond(in_col, value))
1469 2245 aaronmk
1470 2589 aaronmk
    def insert_pkeys_table(which):
1471 2932 aaronmk
        return sql_gen.Table(sql_gen.concat(in_table.name,
1472 2589 aaronmk
            '_insert_'+which+'_pkeys'))
1473
    insert_out_pkeys = insert_pkeys_table('out')
1474
    insert_in_pkeys = insert_pkeys_table('in')
1475
1476 2206 aaronmk
    # Do inserts and selects
1477 2642 aaronmk
    join_cols = sql_gen.ColDict(db, out_table)
1478 2206 aaronmk
    while True:
1479 2521 aaronmk
        if limit_ref[0] == 0: # special case
1480
            log_debug('Creating an empty pkeys table')
1481 2786 aaronmk
            cur = run_query_into(db, mk_select(db, out_table, [out_pkey],
1482 2521 aaronmk
                limit=limit_ref[0]), into=insert_out_pkeys)
1483
            break # don't do main case
1484
1485 2303 aaronmk
        has_joins = join_cols != {}
1486
1487 2903 aaronmk
        log_debug('Trying to insert new rows')
1488
1489 2305 aaronmk
        # Prepare to insert new rows
1490 2325 aaronmk
        insert_joins = input_joins[:] # don't modify original!
1491 2403 aaronmk
        insert_args = dict(recover=True, cacheable=False)
1492 2303 aaronmk
        if has_joins:
1493 3021 aaronmk
            insert_args.update(dict(ignore=True))
1494 2325 aaronmk
        else:
1495 2404 aaronmk
            insert_args.update(dict(returning=out_pkey, into=insert_out_pkeys))
1496 2786 aaronmk
        main_select = mk_main_select(insert_joins, mapping.values())
1497 2303 aaronmk
1498 2929 aaronmk
        def main_insert():
1499 2942 aaronmk
            if is_function:
1500
                log_debug('Calling function on input rows')
1501
                args = dict(((k.name, v) for k, v in mapping.iteritems()))
1502
                func_call = sql_gen.NamedCol(out_pkey,
1503
                    sql_gen.FunctionCall(out_table, **args))
1504
                insert_into_pkeys(input_joins, [in_pkey_col, func_call])
1505
                return None
1506
            else:
1507
                return insert_select(db, out_table, mapping.keys(), main_select,
1508
                    **insert_args)
1509 2929 aaronmk
1510 2206 aaronmk
        try:
1511 2929 aaronmk
            cur = with_savepoint(db, main_insert)
1512 2357 aaronmk
            break # insert successful
1513 2931 aaronmk
        except MissingCastException, e:
1514
            log_exc(e)
1515
1516
            out_col = e.col
1517
            type_ = e.type
1518
1519
            log_debug('Casting '+strings.as_tt(out_col)+' input to '
1520
                +strings.as_tt(type_))
1521 2937 aaronmk
            mapping[out_col] = cast_temp_col(db, type_, mapping[out_col],
1522
                errors_table_)
1523 2206 aaronmk
        except DuplicateKeyException, e:
1524 2309 aaronmk
            log_exc(e)
1525
1526 2258 aaronmk
            old_join_cols = join_cols.copy()
1527 2811 aaronmk
            distinct_on.update(util.dict_subset(mapping, e.cols))
1528 2565 aaronmk
            join_cols.update(util.dict_subset_right_join(mapping, e.cols))
1529 2486 aaronmk
            log_debug('Ignoring existing rows, comparing on these columns:\n'
1530 2505 aaronmk
                +strings.as_inline_table(join_cols, ustr=col_ustr))
1531 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
1532 2230 aaronmk
        except NullValueException, e:
1533 2309 aaronmk
            log_exc(e)
1534
1535 2230 aaronmk
            out_col, = e.cols
1536
            try: in_col = mapping[out_col]
1537 2356 aaronmk
            except KeyError:
1538 2486 aaronmk
                log_debug('Missing mapping for NOT NULL column '+out_col)
1539 2451 aaronmk
                remove_all_rows()
1540 2758 aaronmk
            else: remove_rows(in_col, None, e)
1541 2542 aaronmk
        except FunctionValueException, e:
1542
            log_exc(e)
1543
1544
            func_name = e.name
1545
            value = e.value
1546
            for out_col, in_col in mapping.iteritems():
1547 2758 aaronmk
                in_col = sql_gen.unwrap_func_call(in_col, func_name)
1548
                invalid2null(in_col, value, e)
1549 2429 aaronmk
        except DatabaseErrors, e:
1550
            log_exc(e)
1551
1552 2808 aaronmk
            log_debug('No handler for exception')
1553
            on_error(e)
1554 2451 aaronmk
            remove_all_rows()
1555 2358 aaronmk
        # after exception handled, rerun loop with additional constraints
1556 2132 aaronmk
1557 2942 aaronmk
    if cur != None and row_ct_ref != None and cur.rowcount >= 0:
1558 2357 aaronmk
        row_ct_ref[0] += cur.rowcount
1559
1560 2942 aaronmk
    if is_function: pass # pkeys table already created
1561
    elif has_joins:
1562 2357 aaronmk
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
1563 2486 aaronmk
        log_debug('Getting output table pkeys of existing/inserted rows')
1564 2960 aaronmk
        insert_into_pkeys(select_joins, pkeys_cols, distinct=True)
1565 2357 aaronmk
    else:
1566 2404 aaronmk
        add_row_num(db, insert_out_pkeys) # for joining with input pkeys
1567 2357 aaronmk
1568 2486 aaronmk
        log_debug('Getting input table pkeys of inserted rows')
1569 2786 aaronmk
        run_query_into(db, mk_main_select(input_joins, [in_pkey]),
1570 2404 aaronmk
            into=insert_in_pkeys)
1571
        add_row_num(db, insert_in_pkeys) # for joining with output pkeys
1572 2357 aaronmk
1573 2428 aaronmk
        assert table_row_count(db, insert_out_pkeys) == table_row_count(db,
1574
            insert_in_pkeys)
1575
1576 2486 aaronmk
        log_debug('Combining output and input pkeys in inserted order')
1577 2404 aaronmk
        pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
1578 2357 aaronmk
            {row_num_col: sql_gen.join_same_not_null})]
1579 2420 aaronmk
        insert_into_pkeys(pkey_joins, pkeys_names)
1580 2959 aaronmk
1581 2974 aaronmk
        empty_temp(db, [insert_out_pkeys, insert_in_pkeys])
1582 2357 aaronmk
1583 2486 aaronmk
    db.log_debug('Adding pkey on pkeys table to enable fast joins', level=2.5)
1584 2594 aaronmk
    add_pkey(db, into)
1585 2407 aaronmk
1586 2508 aaronmk
    log_debug('Setting pkeys of missing rows to '+strings.as_tt(repr(default)))
1587 2489 aaronmk
    missing_rows_joins = input_joins+[sql_gen.Join(into,
1588 2357 aaronmk
        {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
1589
        # must use join_same_not_null or query will take forever
1590 2420 aaronmk
    insert_into_pkeys(missing_rows_joins,
1591 2508 aaronmk
        [in_pkey_col, sql_gen.NamedCol(out_pkey, default)])
1592 2357 aaronmk
1593 2489 aaronmk
    assert table_row_count(db, into) == table_row_count(db, in_table)
1594 2428 aaronmk
1595 2974 aaronmk
    empty_temp(db, in_table)
1596 2959 aaronmk
1597 2702 aaronmk
    srcs = []
1598
    if is_func: srcs = sql_gen.cols_srcs(in_cols)
1599
    return sql_gen.Col(out_pkey, into, srcs)
1600 2115 aaronmk
1601
##### Data cleanup
1602
1603 2290 aaronmk
def cleanup_table(db, table, cols):
1604 2677 aaronmk
    table = sql_gen.as_Table(table)
1605
    cols = map(sql_gen.as_Col, cols)
1606 2115 aaronmk
1607 2677 aaronmk
    expr = ('nullif(nullif(trim(both from %s), '+db.esc_value('')+'), '
1608
        +db.esc_value(r'\N')+')')
1609
    changes = [(v, sql_gen.CustomCode(expr % v.to_str(db)))
1610
        for v in cols]
1611 2115 aaronmk
1612 3066 aaronmk
    update(db, table, changes, in_place=True)