Project

General

Profile

1
# Database import/export
2

    
3
import copy
4
import csv
5
import operator
6
import warnings
7
import sys
8

    
9
import csvs
10
import exc
11
import dicts
12
import sql
13
import sql_gen
14
import streams
15
import strings
16
import util
17

    
18
##### Exceptions
19

    
20
# Can't use built-in SyntaxError because it stringifies to only the first line
21
class SyntaxError(Exception): pass
22

    
23
##### Data cleanup
24

    
25
null_strs = ['', '-', r'\N', 'NULL', 'UNKNOWN', 'nulo']
26

    
27
def cleanup_table(db, table):
28
    table = sql_gen.as_Table(table)
29
    cols = filter(lambda c: sql_gen.is_text_col(db, c),
30
        sql.table_cols(db, table))
31
    try: pkey_col = sql.table_pkey_col(db, table)
32
    except sql.DoesNotExistException: pass
33
    else:
34
        try: cols.remove(pkey_col)
35
        except ValueError: pass
36
    if not cols: return
37
    
38
    db.log_debug('Cleaning up table', level=1.5)
39
    
40
    expr = 'trim(both from %s)'
41
    for null in null_strs: expr = 'nullif('+expr+', '+db.esc_value(null)+')'
42
    changes = [(v, sql_gen.CustomCode(expr % v.to_str(db))) for v in cols]
43
    
44
    while True:
45
        try:
46
            sql.update(db, table, changes, in_place=True, recover=True)
47
            break # successful
48
        except sql.NullValueException, e:
49
            db.log_debug('Caught exception: '+exc.str_(e))
50
            col, = e.cols
51
            sql.drop_not_null(db, col)
52
    
53
    db.log_debug('Vacuuming and reanalyzing table', level=1.5)
54
    sql.vacuum(db, table)
55

    
56
##### Error tracking
57

    
58
def track_data_error(db, errors_table, cols, value, error_code, error):
59
    '''
60
    @param errors_table If None, does nothing.
61
    '''
62
    if errors_table == None: return
63
    
64
    col_names = [c.name for c in cols]
65
    if not col_names: col_names = [None] # need at least one entry
66
    for col_name in col_names:
67
        try:
68
            sql.insert(db, errors_table, dict(column=col_name, value=value,
69
                error_code=error_code, error=error), recover=True,
70
                cacheable=True, log_level=4)
71
        except sql.DuplicateKeyException: pass
72

    
73
class ExcToErrorsTable(sql_gen.ExcToWarning):
74
    '''Handles an exception by saving it or converting it to a warning.'''
75
    def __init__(self, return_, srcs, errors_table, value=None):
76
        '''
77
        @param return_ See sql_gen.ExcToWarning
78
        @param srcs The column names for the errors table
79
        @param errors_table None|sql_gen.Table
80
        @param value The value (or an expression for it) that caused the error
81
        @pre The invalid value must be in a local variable "value" of type text.
82
        '''
83
        sql_gen.ExcToWarning.__init__(self, return_)
84
        
85
        value = sql_gen.as_Code(value)
86
        
87
        self.srcs = srcs
88
        self.errors_table = errors_table
89
        self.value = value
90
    
91
    def to_str(self, db):
92
        if not self.srcs or self.errors_table == None:
93
            return sql_gen.ExcToWarning.to_str(self, db)
94
        
95
        errors_table_cols = map(sql_gen.Col,
96
            ['column', 'value', 'error_code', 'error'])
97
        col_names_query = sql.mk_select(db, sql_gen.NamedValues('c', None,
98
            [[c.name] for c in self.srcs]), order_by=None)
99
        insert_query = sql.mk_insert_select(db, self.errors_table,
100
            errors_table_cols,
101
            sql_gen.Values(errors_table_cols).to_str(db))+';\n'
102
        return '''\
103
-- Save error in errors table.
104
DECLARE
105
    error_code text := SQLSTATE;
106
    error text := SQLERRM;
107
    value text := '''+self.value.to_str(db)+''';
108
    "column" text;
109
BEGIN
110
    -- Insert the value and error for *each* source column.
111
'''+strings.indent(sql_gen.RowExcIgnore(None, col_names_query, insert_query,
112
    row_var=errors_table_cols[0]).to_str(db))+'''
113
END;
114

    
115
'''+self.return_.to_str(db)
116

    
117
def data_exception_handler(*args, **kw_args):
118
    '''Handles a data_exception by saving it or converting it to a warning.
119
    For params, see ExcToErrorsTable().
120
    '''
121
    return sql_gen.data_exception_handler(ExcToErrorsTable(*args, **kw_args))
122

    
123
def cast(db, type_, col, errors_table=None):
124
    '''Casts an (unrenamed) column or value.
125
    If errors_table set and col has srcs, saves errors in errors_table (using
126
    col's srcs attr as source columns). Otherwise, converts errors to warnings.
127
    @param col str|sql_gen.Col|sql_gen.Literal
128
    @param errors_table None|sql_gen.Table|str
129
    '''
130
    col = sql_gen.as_Col(col)
131
    
132
    # Don't convert exceptions to warnings for user-supplied constants
133
    if isinstance(col, sql_gen.Literal): return sql_gen.Cast(type_, col)
134
    
135
    assert not isinstance(col, sql_gen.NamedCol)
136
    
137
    function_name = strings.first_word(type_)
138
    srcs = col.srcs
139
    save_errors = errors_table != None and srcs
140
    if save_errors: # function will be unique for the given srcs
141
        function_name = strings.ustr(sql_gen.FunctionCall(function_name,
142
            *map(sql_gen.to_name_only_col, srcs)))
143
    function = db.TempFunction(function_name)
144
    
145
    # Create function definition
146
    modifiers = 'STRICT'
147
    if not save_errors: modifiers = 'IMMUTABLE '+modifiers
148
    value_param = sql_gen.FunctionParam('value', 'anyelement')
149
    handler = data_exception_handler('RETURN NULL;\n', srcs, errors_table,
150
        value_param.name)
151
    body = sql_gen.CustomCode(handler.to_str(db, '''\
152
/* The explicit cast to the return type is needed to make the cast happen
153
inside the try block. (Implicit casts to the return type happen at the end
154
of the function, outside any block.) */
155
RETURN '''+sql_gen.Cast(type_, sql_gen.CustomCode('value')).to_str(db)+''';
156
'''))
157
    body.lang='plpgsql'
158
    sql.define_func(db, sql_gen.FunctionDef(function, type_, body,
159
        [value_param], modifiers))
160
    
161
    return sql_gen.FunctionCall(function, col)
162

    
163
def func_wrapper_exception_handler(db, return_, args, errors_table):
164
    '''Handles a function call's data_exceptions.
165
    Supports PL/Python functions.
166
    @param return_ See data_exception_handler()
167
    @param args [arg...] Function call's args
168
    @param errors_table See data_exception_handler()
169
    '''
170
    args = filter(sql_gen.has_srcs, args)
171
    
172
    srcs = sql_gen.cross_join_srcs(args)
173
    value = sql_gen.merge_not_null(db, ',', args)
174
    return sql_gen.NestedExcHandler(
175
        data_exception_handler(return_, srcs, errors_table, value)
176
        , sql_gen.plpythonu_error_handler
177
        )
178

    
179
def cast_temp_col(db, type_, col, errors_table=None):
180
    '''Like cast(), but creates a new column with the cast values if the input
181
    is a column.
182
    @return The new column or cast value
183
    '''
184
    def cast_(col): return cast(db, type_, col, errors_table)
185
    
186
    try: col = sql_gen.underlying_col(col)
187
    except sql_gen.NoUnderlyingTableException: return sql_gen.wrap(cast_, col)
188
    
189
    table = col.table
190
    new_col = sql_gen.suffixed_col(col, '::'+strings.first_word(type_))
191
    expr = cast_(col)
192
    
193
    # Add column
194
    new_typed_col = sql_gen.TypedCol(new_col.name, type_)
195
    sql.add_col(db, table, new_typed_col, comment=strings.urepr(col)+'::'+type_)
196
    new_col.name = new_typed_col.name # propagate any renaming
197
    
198
    sql.update(db, table, [(new_col, expr)], in_place=True, recover=True)
199
    
200
    return new_col
201

    
202
def errors_table(db, table, if_exists=True):
203
    '''
204
    @param if_exists If set, returns None if the errors table doesn't exist
205
    @return None|sql_gen.Table
206
    '''
207
    table = sql_gen.as_Table(table)
208
    if table.srcs != (): table = table.srcs[0]
209
    
210
    errors_table = sql_gen.suffixed_table(table, '.errors')
211
    if if_exists and not sql.table_exists(db, errors_table): return None
212
    return errors_table
213

    
214
def mk_errors_table(db, table):
215
    errors_table_ = errors_table(db, table, if_exists=False)
216
    if sql.table_exists(db, errors_table_, cacheable=False): return
217
    
218
    typed_cols = [
219
        sql_gen.TypedCol('column', 'text'),
220
        sql_gen.TypedCol('value', 'text'),
221
        sql_gen.TypedCol('error_code', 'character varying(5)', nullable=False),
222
        sql_gen.TypedCol('error', 'text', nullable=False),
223
        ]
224
    sql.create_table(db, errors_table_, typed_cols, has_pkey=False)
225
    index_cols = ['column', 'value', 'error_code', 'error']
226
    sql.add_index(db, index_cols, errors_table_, unique=True)
227

    
228
##### Import
229

    
230
row_num_col_def = copy.copy(sql.row_num_col_def)
231
row_num_col_def.name = 'row_num'
232
row_num_col_def.type = 'integer'
233

    
234
def append_csv(db, table, reader, header):
235
    assert sql.table_col_names(db, table) == header
236
    
237
    def log(msg, level=1): db.log_debug(msg, level)
238
    
239
    # Wrap in standardizing stream
240
    cols_ct = len(header)
241
    stream = csvs.InputRewriter(streams.ProgressInputStream(csvs.StreamFilter(
242
        csvs.ColCtFilter(reader, cols_ct)), sys.stderr, msg='Read %d row(s)',
243
        n=1000))
244
    dialect = stream.dialect # use default dialect
245
    
246
    # Create COPY FROM statement
247
    copy_from = ('COPY '+table.to_str(db)+' FROM STDIN DELIMITER '
248
        +db.esc_value(dialect.delimiter)+' NULL '+db.esc_value(''))
249
    assert not csvs.is_tsv(dialect)
250
    copy_from += ' CSV'
251
    if dialect.quoting != csv.QUOTE_NONE:
252
        quote_str = db.esc_value(dialect.quotechar)
253
        copy_from += ' QUOTE '+quote_str
254
        if dialect.doublequote: copy_from += ' ESCAPE '+quote_str
255
    copy_from += ';\n'
256
    
257
    log(copy_from, level=2)
258
    try: db.db.cursor().copy_expert(copy_from, stream)
259
    except Exception, e: sql.parse_exception(db, e, recover=True)
260

    
261
def import_csv(db, table, reader, header):
262
    def log(msg, level=1): db.log_debug(msg, level)
263
    
264
    # Get format info
265
    col_names = map(strings.to_unicode, header)
266
    for i, col in enumerate(col_names): # replace empty column names
267
        if col == '': col_names[i] = 'column_'+str(i)
268
    
269
    # Select schema and escape names
270
    def esc_name(name): return db.esc_name(name)
271
    
272
    typed_cols = [sql_gen.TypedCol(v, 'text') for v in col_names]
273
    typed_cols.insert(0, row_num_col_def)
274
    header.insert(0, row_num_col_def.name)
275
    reader = csvs.RowNumFilter(reader)
276
    
277
    log('Creating table')
278
    # Note that this is not rolled back if the import fails. Instead, it is
279
    # cached, and will not be re-run if the import is retried.
280
    sql.create_table(db, table, typed_cols, has_pkey=False, col_indexes=False)
281
    
282
    # Free memory used by deleted (rolled back) rows from any failed import.
283
    # This MUST be run so that the rows will be stored in inserted order, and
284
    # the row_num added after import will match up with the CSV's row order.
285
    sql.truncate(db, table)
286
    
287
    # Load the data
288
    def load(): append_csv(db, table, reader, header)
289
    sql.with_savepoint(db, load)
290
    
291
    cleanup_table(db, table)
292

    
293
def put(db, table, row, pkey_=None, row_ct_ref=None, on_error=exc.reraise):
294
    '''Recovers from errors.
295
    Only works under PostgreSQL (uses INSERT RETURNING).
296
    '''
297
    return put_table(db, table, [], row, row_ct_ref, on_error=on_error)
298

    
299
def get(db, table, row, pkey, row_ct_ref=None, create=False):
300
    '''Recovers from errors'''
301
    try:
302
        return sql.value(sql.select(db, table, [pkey], row, limit=1,
303
            recover=True))
304
    except StopIteration:
305
        if not create: raise
306
        return put(db, table, row, pkey, row_ct_ref) # insert new row
307

    
308
def is_func_result(col):
309
    return col.table.name.find('(') >= 0 and col.name == 'result'
310

    
311
def into_table_name(out_table, in_tables0, mapping, is_func):
312
    def in_col_str(in_col):
313
        in_col = sql_gen.remove_col_rename(in_col)
314
        if isinstance(in_col, sql_gen.Col):
315
            table = in_col.table
316
            if table == in_tables0:
317
                in_col = sql_gen.to_name_only_col(in_col)
318
            elif is_func_result(in_col): in_col = table # omit col name
319
        return strings.ustr(in_col)
320
    
321
    str_ = strings.ustr(out_table)
322
    if is_func:
323
        str_ += '('
324
        
325
        try: value_in_col = mapping['value']
326
        except KeyError:
327
            str_ += ', '.join((strings.ustr(k)+'='+in_col_str(v)
328
                for k, v in mapping.iteritems()))
329
        else: str_ += in_col_str(value_in_col)
330
        
331
        str_ += ')'
332
    else:
333
        out_col = 'rank'
334
        try: in_col = mapping[out_col]
335
        except KeyError: str_ += '_pkeys'
336
        else: # has a rank column, so hierarchical
337
            str_ += '['+strings.ustr(out_col)+'='+in_col_str(in_col)+']'
338
    return str_
339

    
340
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None, default=None,
341
    col_defaults={}, on_error=exc.reraise):
342
    '''Recovers from errors.
343
    Only works under PostgreSQL (uses INSERT RETURNING).
344
    
345
    Warning: This function's normalizing algorithm does not support database
346
    triggers that populate fields covered by a unique constraint. Such fields
347
    must be populated by the mappings instead.
348
    (Other fields are not affected by this restriction on triggers.)
349
    
350
    Note that much of the complexity of the normalizing algorithm is due to
351
    PostgreSQL (and other DB systems) not having a native command for
352
    insert/on duplicate select. This operation is a cross between MySQL's
353
    INSERT ON DUPLICATE KEY UPDATE (which does not provide SELECT
354
    functionality), and PostgreSQL's INSERT RETURNING (which throws an error
355
    instead of returning the existing row).
356
    @param in_tables The main input table to select from, followed by a list of
357
        tables to join with it using the main input table's pkey
358
    @param mapping dict(out_table_col=in_table_col, ...)
359
        * out_table_col: str (*not* sql_gen.Col)
360
        * in_table_col: sql_gen.Col|literal-value
361
    @param default The *output* column to use as the pkey for missing rows.
362
        If this output column does not exist in the mapping, uses None.
363
    @param col_defaults Default values for required columns.
364
    @return sql_gen.Col Where the output pkeys are made available
365
    '''
366
    import psycopg2.extensions
367
    
368
    # Special handling for functions with hstore params
369
    if out_table == '_map':
370
        import psycopg2.extras
371
        psycopg2.extras.register_hstore(db.db)
372
        
373
        # Parse args
374
        try: value = mapping.pop('value')
375
        except KeyError: return None # value required
376
        
377
        mapping = dict([(k, sql_gen.get_value(v))
378
            for k, v in mapping.iteritems()]) # unwrap literal value
379
        mapping = dict(map=mapping, value=value) # non-value params -> hstore
380
    
381
    out_table = sql_gen.as_Table(out_table)
382
    
383
    def log_debug(msg): db.log_debug(msg, level=1.5)
384
    def col_ustr(str_):
385
        return strings.repr_no_u(sql_gen.remove_col_rename(str_))
386
    
387
    log_debug('********** New iteration **********')
388
    log_debug('Inserting these input columns into '+strings.as_tt(
389
        out_table.to_str(db))+':\n'+strings.as_table(mapping, ustr=col_ustr))
390
    
391
    is_function = sql.function_exists(db, out_table)
392
    
393
    if is_function: row_ct_ref = None # only track inserted rows
394
    
395
    # Warn if inserting empty table rows
396
    if not mapping and not is_function: # functions with no args OK
397
        warnings.warn(UserWarning('Inserting empty table row(s)'))
398
    
399
    if is_function: out_pkey = 'result'
400
    else: out_pkey = sql.pkey_name(db, out_table, recover=True)
401
    out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
402
    
403
    in_tables_ = copy.copy(in_tables) # don't modify input!
404
    try: in_tables0 = in_tables_.pop(0) # first table is separate
405
    except IndexError: in_tables0 = None
406
    else:
407
        in_pkey = sql.pkey_name(db, in_tables0, recover=True)
408
        in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
409
    
410
    # Determine if can use optimization for only literal values
411
    is_literals = not reduce(operator.or_, map(sql_gen.is_table_col,
412
        mapping.values()), False)
413
    is_literals_or_function = is_literals or is_function
414
    
415
    if in_tables0 == None: errors_table_ = None
416
    else: errors_table_ = errors_table(db, in_tables0)
417
    
418
    # Create input joins from list of input tables
419
    input_joins = [in_tables0]+[sql_gen.Join(v,
420
        {in_pkey: sql_gen.join_same_not_null}) for v in in_tables_]
421
    
422
    orig_mapping = mapping.copy()
423
    if mapping == {} and not is_function: # need >= one column for INSERT SELECT
424
        mapping = {out_pkey: None} # ColDict will replace with default value
425
    
426
    if not is_literals:
427
        into = sql_gen.as_Table(into_table_name(out_table, in_tables0, mapping,
428
            is_function))
429
        # Ensure into's out_pkey is different from in_pkey by prepending "out."
430
        if is_function: into_out_pkey = out_pkey
431
        else: into_out_pkey = 'out.'+out_pkey
432
        
433
        # Set column sources
434
        in_cols = filter(sql_gen.is_table_col, mapping.values())
435
        for col in in_cols:
436
            if col.table == in_tables0: col.set_srcs(sql_gen.src_self)
437
        
438
        log_debug('Joining together input tables into temp table')
439
        # Place in new table so don't modify input and for speed
440
        in_table = sql_gen.Table('in')
441
        mapping = dicts.join(mapping, sql.flatten(db, in_table, input_joins,
442
            in_cols, preserve=[in_pkey_col]))
443
        input_joins = [in_table]
444
        db.log_debug('Temp table: '+strings.as_tt(in_table.to_str(db)), level=2)
445
    
446
    # Wrap mapping in a sql_gen.ColDict.
447
    # sql_gen.ColDict sanitizes both keys and values passed into it.
448
    # Do after applying dicts.join() because that returns a plain dict.
449
    mapping = sql_gen.ColDict(db, out_table, mapping)
450
    
451
    # Save all rows since in_table may have rows deleted
452
    if is_literals: pass
453
    elif is_function: full_in_table = in_table
454
    else:
455
        full_in_table = sql_gen.suffixed_table(in_table, '_full')
456
        sql.copy_table(db, in_table, full_in_table)
457
    
458
    pkeys_table_exists_ref = [False]
459
    def insert_into_pkeys(query, **kw_args):
460
        if pkeys_table_exists_ref[0]:
461
            sql.insert_select(db, into, [in_pkey, into_out_pkey], query,
462
                **kw_args)
463
        else:
464
            kw_args.setdefault('add_pkey_', True)
465
            
466
            sql.run_query_into(db, query, into=into, **kw_args)
467
            pkeys_table_exists_ref[0] = True
468
    
469
    def mk_main_select(joins, cols): return sql.mk_select(db, joins, cols)
470
    
471
    if is_literals: insert_in_table = None
472
    else:
473
        insert_in_table = in_table
474
        insert_in_tables = [insert_in_table]
475
    join_cols = sql_gen.ColDict(db, out_table)
476
    
477
    exc_strs = set()
478
    def log_exc(e):
479
        e_str = exc.str_(e, first_line_only=True)
480
        log_debug('Caught exception: '+e_str)
481
        if e_str in exc_strs: # avoid infinite loops
482
            log_debug('Exception already seen, handler broken')
483
            on_error(e)
484
            remove_all_rows()
485
            return False
486
        else: exc_strs.add(e_str)
487
        return True
488
    
489
    ignore_all_ref = [False]
490
    def remove_all_rows():
491
        log_debug('Ignoring all rows')
492
        ignore_all_ref[0] = True # just return the default value column
493
    
494
    def handle_unknown_exc(e):
495
        log_debug('No handler for exception')
496
        on_error(e)
497
        remove_all_rows()
498
    
499
    def ensure_cond(cond, e, passed=False, failed=False):
500
        if is_literals: # we know the constraint was applied exactly once
501
            if passed: pass
502
            elif failed: remove_all_rows()
503
            else: raise NotImplementedError()
504
        else:
505
            if not is_function:
506
                out_table_cols = sql_gen.ColDict(db, out_table)
507
                out_table_cols.update(util.dict_subset_right_join({},
508
                    sql.table_col_names(db, out_table)))
509
            
510
            in_cols = []
511
            cond = strings.ustr(cond)
512
            orig_cond = cond
513
            cond = sql_gen.map_expr(db, cond, mapping, in_cols)
514
            if not is_function:
515
                cond = sql_gen.map_expr(db, cond, out_table_cols)
516
            
517
            log_debug('Ignoring rows that do not satisfy '+strings.as_tt(cond))
518
            cur = None
519
            if cond == sql_gen.false_expr:
520
                assert failed
521
                remove_all_rows()
522
            elif cond == sql_gen.true_expr: assert passed
523
            else:
524
                while True:
525
                    not_cond = sql_gen.NotCond(sql_gen.CustomCode(cond))
526
                    try:
527
                        cur = sql.delete(db, insert_in_table, not_cond)
528
                        break
529
                    except sql.DoesNotExistException, e:
530
                        if e.type != 'column': raise
531
                        
532
                        last_cond = cond
533
                        cond = sql_gen.map_expr(db, cond, {e.name: None})
534
                        if cond == last_cond: raise # not fixable
535
            
536
            # If any rows failed cond
537
            if failed or cur != None and cur.rowcount > 0:
538
                track_data_error(db, errors_table_,
539
                    sql_gen.cross_join_srcs(in_cols), None, e.cause.pgcode,
540
                    strings.ensure_newl(strings.ustr(e.cause.pgerror))
541
                    +'condition: '+orig_cond+'\ntranslated condition: '+cond)
542
    
543
    not_null_cols = set()
544
    def ignore(in_col, value, e):
545
        if sql_gen.is_table_col(in_col):
546
            in_col = sql_gen.with_table(in_col, insert_in_table)
547
            
548
            track_data_error(db, errors_table_, in_col.srcs, value,
549
                e.cause.pgcode, e.cause.pgerror)
550
            
551
            sql.add_index(db, in_col, insert_in_table) # enable fast filtering
552
            if value != None and in_col not in not_null_cols:
553
                log_debug('Replacing invalid value '
554
                    +strings.as_tt(strings.urepr(value))+' with NULL in column '
555
                    +strings.as_tt(in_col.to_str(db)))
556
                sql.update(db, insert_in_table, [(in_col, None)],
557
                    sql_gen.ColValueCond(in_col, value))
558
            else:
559
                log_debug('Ignoring rows with '+strings.as_tt(in_col.to_str(db))
560
                    +' = '+strings.as_tt(strings.urepr(value)))
561
                sql.delete(db, insert_in_table,
562
                    sql_gen.ColValueCond(in_col, value))
563
                if value == None: not_null_cols.add(in_col)
564
        else:
565
            assert isinstance(in_col, sql_gen.NamedCol)
566
            in_value = sql_gen.remove_col_rename(in_col)
567
            assert sql_gen.is_literal(in_value)
568
            if value == in_value.value:
569
                if value != None:
570
                    log_debug('Replacing invalid literal '
571
                        +strings.as_tt(in_col.to_str(db))+' with NULL')
572
                    mapping[in_col.name] = None
573
                else:
574
                    remove_all_rows()
575
            # otherwise, all columns were being ignore()d because the specific
576
            # column couldn't be identified, and this was not the invalid column
577
    
578
    if not is_literals:
579
        def insert_pkeys_table(which):
580
            return sql_gen.Table(sql_gen.concat(in_table.name,
581
                '_insert_'+which+'_pkeys'))
582
        insert_out_pkeys = insert_pkeys_table('out')
583
        insert_in_pkeys = insert_pkeys_table('in')
584
    
585
    def mk_func_call():
586
        args = dict(((k.name, v) for k, v in mapping.iteritems()))
587
        return sql_gen.FunctionCall(out_table, **args), args
588
    
589
    missing_msg = None
590
    
591
    # Do inserts and selects
592
    while True:
593
        has_joins = join_cols != {}
594
        
595
        if ignore_all_ref[0]: break # unrecoverable error, so don't do main case
596
        
597
        # Prepare to insert new rows
598
        if is_function:
599
            if is_literals:
600
                log_debug('Calling function')
601
                func_call, args = mk_func_call()
602
        else:
603
            log_debug('Trying to insert new rows')
604
            insert_args = dict(recover=True, cacheable=False)
605
            if has_joins:
606
                insert_args.update(dict(ignore=True))
607
            else:
608
                insert_args.update(dict(returning=out_pkey))
609
                if not is_literals:
610
                    insert_args.update(dict(into=insert_out_pkeys))
611
            main_select = mk_main_select([insert_in_table], [sql_gen.with_table(
612
                c, insert_in_table) for c in mapping.values()])
613
        
614
        try:
615
            cur = None
616
            if is_function:
617
                if is_literals:
618
                    cur = sql.select(db, fields=[func_call], recover=True,
619
                        cacheable=True)
620
                else:
621
                    log_debug('Defining wrapper function')
622
                    
623
                    func_call, args = mk_func_call()
624
                    func_call = sql_gen.NamedCol(into_out_pkey, func_call)
625
                    
626
                    # Create empty pkeys table so its row type can be used
627
                    insert_into_pkeys(sql.mk_select(db, input_joins,
628
                        [in_pkey_col, func_call], limit=0), add_pkey_=False,
629
                        recover=True)
630
                    result_type = db.col_info(sql_gen.Col(into_out_pkey,
631
                        into)).type
632
                    
633
                    ## Create error handling wrapper function
634
                    
635
                    wrapper = db.TempFunction(sql_gen.concat(into.name,
636
                        '_wrap'))
637
                    
638
                    select_cols = [in_pkey_col]+args.values()
639
                    row_var = copy.copy(sql_gen.row_var)
640
                    row_var.set_srcs([in_table])
641
                    in_pkey_var = sql_gen.Col(in_pkey, row_var)
642
                    
643
                    args = dict(((k, sql_gen.with_table(v, row_var))
644
                        for k, v in args.iteritems()))
645
                    func_call = sql_gen.FunctionCall(out_table, **args)
646
                    
647
                    def mk_return(result):
648
                        return sql_gen.ReturnQuery(sql.mk_select(db,
649
                            fields=[in_pkey_var, result], explain=False))
650
                    exc_handler = func_wrapper_exception_handler(db,
651
                        mk_return(sql_gen.Cast(result_type, None)),
652
                        args.values(), errors_table_)
653
                    
654
                    sql.define_func(db, sql_gen.FunctionDef(wrapper,
655
                        sql_gen.SetOf(into),
656
                        sql_gen.RowExcIgnore(sql_gen.RowType(in_table),
657
                            sql.mk_select(db, input_joins),
658
                            mk_return(func_call), exc_handler=exc_handler)
659
                        ))
660
                    wrapper_table = sql_gen.FunctionCall(wrapper)
661
                    
662
                    log_debug('Calling function')
663
                    insert_into_pkeys(sql.mk_select(db, wrapper_table,
664
                        order_by=None), recover=True, cacheable=False)
665
                    sql.add_pkey_or_index(db, into, warn=True)
666
            else:
667
                cur = sql.insert_select(db, out_table, mapping.keys(),
668
                    main_select, **insert_args)
669
            break # insert successful
670
        except sql.MissingCastException, e:
671
            if not log_exc(e): break
672
            
673
            type_ = e.type
674
            if e.col == None: out_cols = mapping.keys()
675
            else: out_cols = [e.col]
676
            
677
            for out_col in out_cols:
678
                log_debug('Casting '+strings.as_tt(strings.repr_no_u(out_col))
679
                    +' input to '+strings.as_tt(type_))
680
                in_col = mapping[out_col]
681
                while True:
682
                    try:
683
                        mapping[out_col] = cast_temp_col(db, type_, in_col,
684
                            errors_table_)
685
                        break # cast successful
686
                    except sql.InvalidValueException, e:
687
                        if not log_exc(e): break
688
                        
689
                        ignore(in_col, e.value, e)
690
        except sql.DuplicateKeyException, e:
691
            if not log_exc(e): break
692
            
693
            # Different rows violating different unique constraints not
694
            # supported
695
            assert not join_cols
696
            
697
            if e.cond != None: ensure_cond(e.cond, e, passed=True)
698
            
699
            join_cols.update(util.dict_subset_right_join(mapping, e.cols))
700
            log_debug('Ignoring existing rows, comparing on these columns:\n'
701
                +strings.as_inline_table(join_cols, ustr=col_ustr))
702
            
703
            if is_literals:
704
                return sql.value(sql.select(db, out_table, [out_pkey_col],
705
                    join_cols, order_by=None))
706
            
707
            # Uniquify and filter input table to avoid (most) duplicate keys
708
            # (Additional duplicates may be added concurrently and will be
709
            # filtered out separately upon insert.)
710
            insert_in_table = sql.distinct_table(db, insert_in_table,
711
                join_cols.values(), [insert_in_table,
712
                sql_gen.Join(out_table, join_cols, sql_gen.filter_out)])
713
            insert_in_tables.append(insert_in_table)
714
        except sql.NullValueException, e:
715
            if not log_exc(e): break
716
            
717
            out_col, = e.cols
718
            try: in_col = mapping[out_col]
719
            except KeyError, e:
720
                try: in_col = mapping[out_col] = col_defaults[out_col]
721
                except KeyError:
722
                    missing_msg = 'Missing mapping for NOT NULL column '+out_col
723
                    log_debug(missing_msg)
724
                    remove_all_rows()
725
            else: ignore(in_col, None, e)
726
        except sql.CheckException, e:
727
            if not log_exc(e): break
728
            
729
            ensure_cond(e.cond, e, failed=True)
730
        except sql.InvalidValueException, e:
731
            if not log_exc(e): break
732
            
733
            for in_col in mapping.values(): ignore(in_col, e.value, e)
734
        except psycopg2.extensions.TransactionRollbackError, e:
735
            if not log_exc(e): break
736
            # retry
737
        except sql.DatabaseErrors, e:
738
            if not log_exc(e): break
739
            
740
            handle_unknown_exc(e)
741
        # after exception handled, rerun loop with additional constraints
742
    
743
    # Resolve default value column
744
    if default != None:
745
        if ignore_all_ref[0]: mapping.update(orig_mapping) # use input cols
746
        try: default = mapping[default]
747
        except KeyError:
748
            db.log_debug('Default value column '
749
                +strings.as_tt(strings.repr_no_u(default))
750
                +' does not exist in mapping, falling back to None', level=2.1)
751
            default = None
752
        else: default = sql_gen.remove_col_rename(default)
753
    
754
    if missing_msg != None and default == None:
755
        warnings.warn(UserWarning(missing_msg))
756
        # not an error because sometimes the mappings include
757
        # extra tables which aren't used by the dataset
758
    
759
    # Handle unrecoverable errors
760
    if ignore_all_ref[0]:
761
        log_debug('Returning default: '+strings.as_tt(strings.urepr(default)))
762
        return default
763
    
764
    if cur != None and row_ct_ref != None and cur.rowcount >= 0:
765
        row_ct_ref[0] += cur.rowcount
766
    
767
    if is_literals: return sql.value(cur)
768
    
769
    if is_function: pass # pkeys table already created
770
    elif has_joins:
771
        select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
772
        log_debug('Getting output table pkeys of existing/inserted rows')
773
        insert_into_pkeys(sql.mk_select(db, select_joins, [in_pkey_col,
774
            sql_gen.NamedCol(into_out_pkey, out_pkey_col)], order_by=None))
775
    else:
776
        sql.add_row_num(db, insert_out_pkeys) # for joining with input pkeys
777
        
778
        log_debug('Getting input table pkeys of inserted rows')
779
        # Note that mk_main_select() does not use ORDER BY. Instead, assume that
780
        # since the SELECT query is identical to the one used in INSERT SELECT,
781
        # its rows will be retrieved in the same order.
782
        sql.run_query_into(db, mk_main_select(input_joins, [in_pkey]),
783
            into=insert_in_pkeys)
784
        sql.add_row_num(db, insert_in_pkeys) # for joining with output pkeys
785
        
786
        assert sql.table_row_count(db, insert_out_pkeys) == sql.table_row_count(
787
            db, insert_in_pkeys)
788
        
789
        log_debug('Combining output and input pkeys in inserted order')
790
        pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
791
            {sql.row_num_col: sql_gen.join_same_not_null})]
792
        in_col = sql_gen.Col(in_pkey, insert_in_pkeys)
793
        out_col = sql_gen.NamedCol(into_out_pkey,
794
            sql_gen.Col(out_pkey, insert_out_pkeys))
795
        insert_into_pkeys(sql.mk_select(db, pkey_joins, [in_col, out_col],
796
            order_by=None))
797
        
798
        sql.empty_temp(db, [insert_out_pkeys, insert_in_pkeys])
799
    
800
    if not is_function: # is_function doesn't leave holes
801
        log_debug('Setting pkeys of missing rows to '
802
            +strings.as_tt(strings.urepr(default)))
803
        
804
        full_in_pkey_col = sql_gen.Col(in_pkey, full_in_table)
805
        if sql_gen.is_table_col(default):
806
            default = sql_gen.with_table(default, full_in_table)
807
        missing_rows_joins = [full_in_table, sql_gen.Join(into,
808
            {in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
809
            # must use join_same_not_null or query will take forever
810
        
811
        insert_args = dict(order_by=None)
812
        if not sql.table_has_pkey(db, full_in_table): # in_table has duplicates
813
            insert_args.update(dict(distinct_on=[full_in_pkey_col]))
814
        
815
        insert_into_pkeys(sql.mk_select(db, missing_rows_joins,
816
            [full_in_pkey_col, sql_gen.NamedCol(into_out_pkey, default)],
817
            **insert_args))
818
    # otherwise, there is already an entry for every row
819
    
820
    sql.empty_temp(db, insert_in_tables+[full_in_table])
821
    
822
    srcs = []
823
    if is_function: srcs = sql_gen.cols_srcs(in_cols)
824
    return sql_gen.Col(into_out_pkey, into, srcs)
(35-35/47)