Project

General

Profile

1
# SQL code generation
2

    
3
import copy
4
import operator
5
import re
6
import UserDict
7

    
8
import dicts
9
import iters
10
import lists
11
import objects
12
import strings
13
import util
14

    
15
##### Names
16

    
17
identifier_max_len = 63 # works for both PostgreSQL and MySQL
18

    
19
def add_suffix(str_, suffix):
20
    '''Preserves version so that it won't be truncated off the string, leading
21
    to collisions.'''
22
    # Preserve version
23
    before, sep, version = str_.rpartition('#')
24
    if sep != '': # found match
25
        str_ = before
26
        suffix = sep+version+suffix
27
    
28
    return strings.add_suffix(str_, suffix, identifier_max_len)
29

    
30
def truncate(str_): return add_suffix(str_, '')
31

    
32
def is_safe_name(name):
33
    '''A name is safe *and unambiguous* if it:
34
    * contains only *lowercase* word (\w) characters
35
    * doesn't start with a digit
36
    * contains "_", so that it's not a keyword
37
    ''' 
38
    return re.match(r'^(?=.*_)(?!\d)[^\WA-Z]+$', name) 
39

    
40
def esc_name(name, quote='"'):
41
    return quote + name.replace(quote, quote+quote) + quote
42
        # doubling an embedded quote escapes it in both PostgreSQL and MySQL
43

    
44
def clean_name(name): return name.replace('"', '').replace('`', '')
45

    
46
##### General SQL code objects
47

    
48
class MockDb:
49
    def esc_value(self, value): return strings.repr_no_u(value)
50
    
51
    def esc_name(self, name): return esc_name(name)
52
    
53
    def col_info(self, col):
54
        return TypedCol(col.name, '<type>', CustomCode('<default>'), True)
55

    
56
mockDb = MockDb()
57

    
58
class BasicObject(objects.BasicObject):
59
    def __init__(self, value): self.value = value
60
    
61
    def __str__(self): return clean_name(strings.repr_no_u(self))
62

    
63
##### Unparameterized code objects
64

    
65
class Code(BasicObject):
66
    def to_str(self, db): raise NotImplementedError()
67
    
68
    def __repr__(self): return self.to_str(mockDb)
69

    
70
class CustomCode(Code):
71
    def __init__(self, str_): self.str_ = str_
72
    
73
    def to_str(self, db): return self.str_
74

    
75
def as_Code(value, db=None):
76
    '''
77
    @param db If set, runs db.std_code() on the value.
78
    '''
79
    if util.is_str(value):
80
        if db != None: value = db.std_code(value)
81
        return CustomCode(value)
82
    else: return Literal(value)
83

    
84
class Expr(Code):
85
    def __init__(self, expr): self.expr = expr
86
    
87
    def to_str(self, db): return '('+self.expr.to_str(db)+')'
88

    
89
##### Literal values
90

    
91
class Literal(Code):
92
    def __init__(self, value): self.value = value
93
    
94
    def to_str(self, db): return db.esc_value(self.value)
95

    
96
def as_Value(value):
97
    if isinstance(value, Code): return value
98
    else: return Literal(value)
99

    
100
def is_null(value): return isinstance(value, Literal) and value.value == None
101

    
102
##### Derived elements
103

    
104
src_self = object() # tells Col that it is its own source column
105

    
106
class Derived(Code):
107
    def __init__(self, srcs):
108
        '''An element which was derived from some other element(s).
109
        @param srcs See self.set_srcs()
110
        '''
111
        self.set_srcs(srcs)
112
    
113
    def set_srcs(self, srcs, overwrite=True):
114
        '''
115
        @param srcs (self_type...)|src_self The element(s) this is derived from
116
        '''
117
        if not overwrite and self.srcs != (): return # already set
118
        
119
        if srcs == src_self: srcs = (self,)
120
        srcs = tuple(srcs) # make Col hashable
121
        self.srcs = srcs
122
    
123
    def _compare_on(self):
124
        compare_on = self.__dict__.copy()
125
        del compare_on['srcs'] # ignore
126
        return compare_on
127

    
128
def cols_srcs(cols): return lists.uniqify(iters.flatten((v.srcs for v in cols)))
129

    
130
##### Tables
131

    
132
class Table(Derived):
133
    def __init__(self, name, schema=None, srcs=()):
134
        '''
135
        @param schema str|None (for no schema)
136
        @param srcs (Table...)|src_self See Derived.set_srcs()
137
        '''
138
        Derived.__init__(self, srcs)
139
        
140
        name = truncate(name)
141
        
142
        self.name = name
143
        self.schema = schema
144
    
145
    def to_str(self, db):
146
        str_ = ''
147
        if self.schema != None: str_ += db.esc_name(self.schema)+'.'
148
        str_ += db.esc_name(self.name)
149
        return str_
150
    
151
    def to_Table(self): return self
152

    
153
def is_underlying_table(table):
154
    return isinstance(table, Table) and table.to_Table() is table
155

    
156
class NoUnderlyingTableException(Exception): pass
157

    
158
def underlying_table(table):
159
    table = remove_table_rename(table)
160
    if not is_underlying_table(table): raise NoUnderlyingTableException
161
    return table
162

    
163
def as_Table(table, schema=None):
164
    if table == None or isinstance(table, Code): return table
165
    else: return Table(table, schema)
166

    
167
def suffixed_table(table, suffix): return Table(table.name+suffix, table.schema)
168

    
169
class NamedTable(Table):
170
    def __init__(self, name, code, cols=None):
171
        Table.__init__(self, name)
172
        
173
        if not isinstance(code, Code): code = Table(code)
174
        if not isinstance(code, (Table, FunctionCall, Expr)): code = Expr(code)
175
        if cols != None: cols = map(to_name_only_col, cols)
176
        
177
        self.code = code
178
        self.cols = cols
179
    
180
    def to_str(self, db):
181
        str_ = self.code.to_str(db)+'\nAS '+Table.to_str(self, db)
182
        if self.cols != None:
183
            str_ += ' ('+(', '.join((c.to_str(db) for c in self.cols)))+')'
184
        return str_
185
    
186
    def to_Table(self): return Table(self.name)
187

    
188
def remove_table_rename(table):
189
    if isinstance(table, NamedTable): table = table.code
190
    return table
191

    
192
##### Columns
193

    
194
class Col(Derived):
195
    def __init__(self, name, table=None, srcs=()):
196
        '''
197
        @param table Table|None (for no table)
198
        @param srcs (Col...)|src_self See Derived.set_srcs()
199
        '''
200
        Derived.__init__(self, srcs)
201
        
202
        name = truncate(name)
203
        if util.is_str(table): table = Table(table)
204
        assert table == None or isinstance(table, Table)
205
        
206
        self.name = name
207
        self.table = table
208
    
209
    def to_str(self, db):
210
        str_ = ''
211
        if self.table != None: str_ += self.table.to_Table().to_str(db)+'.'
212
        str_ += db.esc_name(self.name)
213
        return str_
214
    
215
    def to_Col(self): return self
216

    
217
def is_table_col(col): return isinstance(col, Col) and col.table != None
218

    
219
def as_Col(col, table=None, name=None):
220
    '''
221
    @param name If not None, any non-Col input will be renamed using NamedCol.
222
    '''
223
    if name != None:
224
        col = as_Value(col)
225
        if not isinstance(col, Col): col = NamedCol(name, col)
226
    
227
    if isinstance(col, Code): return col
228
    else: return Col(col, table)
229

    
230
def with_default_table(col, table, overwrite=False):
231
    col = as_Col(col)
232
    if not isinstance(col, NamedCol) and (overwrite or col.table == None):
233
        col = copy.copy(col) # don't modify input!
234
        col.table = table
235
    return col
236

    
237
def set_cols_table(table, cols):
238
    table = as_Table(table)
239
    
240
    for i, col in enumerate(cols):
241
        col = cols[i] = as_Col(col)
242
        col.table = table
243

    
244
def to_name_only_col(col, check_table=None):
245
    col = as_Col(col)
246
    if not isinstance(col, Col): return col
247
    
248
    if check_table != None:
249
        table = col.table
250
        assert table == None or table == check_table
251
    return Col(col.name)
252

    
253
class NamedCol(Col):
254
    def __init__(self, name, code):
255
        Col.__init__(self, name)
256
        
257
        if not isinstance(code, Code): code = Literal(code)
258
        
259
        self.code = code
260
    
261
    def to_str(self, db):
262
        return self.code.to_str(db)+' AS '+Col.to_str(self, db)
263
    
264
    def to_Col(self): return Col(self.name)
265

    
266
def remove_col_rename(col):
267
    if isinstance(col, NamedCol): col = col.code
268
    return col
269

    
270
def underlying_col(col):
271
    col = remove_col_rename(col)
272
    if not isinstance(col, Col): raise NoUnderlyingTableException
273
    
274
    return Col(col.name, underlying_table(col.table), col.srcs)
275

    
276
def wrap(wrap_func, value):
277
    '''Wraps a value, propagating any column renaming to the returned value.'''
278
    if isinstance(value, NamedCol):
279
        return NamedCol(value.name, wrap_func(value.code))
280
    else: return wrap_func(value)
281

    
282
class ColDict(dicts.DictProxy):
283
    '''A dict that automatically makes inserted entries Col objects'''
284
    
285
    def __init__(self, db, keys_table, dict_={}):
286
        dicts.DictProxy.__init__(self, {})
287
        
288
        keys_table = as_Table(keys_table)
289
        
290
        self.db = db
291
        self.table = keys_table
292
        self.update(dict_) # after setting vars because __setitem__() needs them
293
    
294
    def copy(self): return ColDict(self.db, self.table, self.inner.copy())
295
    
296
    def __getitem__(self, key):
297
        return dicts.DictProxy.__getitem__(self, self._key(key))
298
    
299
    def __setitem__(self, key, value):
300
        key = self._key(key)
301
        if value == None: value = self.db.col_info(key).default
302
        dicts.DictProxy.__setitem__(self, key, as_Col(value, name=key.name))
303
    
304
    def _key(self, key): return as_Col(key, self.table)
305

    
306
##### Functions
307

    
308
class Function(Table): pass
309

    
310
def TempFunction(name, autocommit):
311
    schema = None
312
    if not autocommit: schema = 'pg_temp'
313
    return Function(name, schema)
314

    
315
class InternalFunction(CustomCode): pass
316

    
317
class FunctionCall(Code):
318
    def __init__(self, function, *args):
319
        '''
320
        @param args [Code|literal-value...] The function's arguments
321
        '''
322
        if not isinstance(function, Code): function = Function(function)
323
        args = map(remove_col_rename, map(as_Value, args))
324
        
325
        self.function = function
326
        self.args = args
327
    
328
    def to_str(self, db):
329
        args_str = ', '.join((v.to_str(db) for v in self.args))
330
        return self.function.to_str(db)+'('+args_str+')'
331

    
332
def wrap_in_func(function, value):
333
    '''Wraps a value inside a function call.
334
    Propagates any column renaming to the returned value.
335
    '''
336
    return wrap(lambda v: FunctionCall(function, v), value)
337

    
338
def unwrap_func_call(func_call, check_name=None):
339
    '''Unwraps any function call to its first argument.
340
    Also removes any column renaming.
341
    '''
342
    func_call = remove_col_rename(func_call)
343
    if not isinstance(func_call, FunctionCall): return func_call
344
    
345
    if check_name != None:
346
        name = func_call.function.name
347
        assert name == None or name == check_name
348
    return func_call.args[0]
349

    
350
##### Conditions
351

    
352
class ColValueCond(Code):
353
    def __init__(self, col, value):
354
        value = as_ValueCond(value)
355
        
356
        self.col = col
357
        self.value = value
358
    
359
    def to_str(self, db): return self.value.to_str(db, self.col)
360

    
361
def combine_conds(conds, keyword=None):
362
    '''
363
    @param keyword The keyword to add before the conditions, if any
364
    '''
365
    str_ = ''
366
    if keyword != None:
367
        if conds == []: whitespace = ''
368
        elif len(conds) == 1: whitespace = ' '
369
        else: whitespace = '\n'
370
        str_ += keyword+whitespace
371
    
372
    str_ += '\nAND '.join(conds)
373
    return str_
374

    
375
##### Condition column comparisons
376

    
377
class ValueCond(BasicObject):
378
    def __init__(self, value):
379
        value = remove_col_rename(as_Value(value))
380
        
381
        self.value = value
382
    
383
    def to_str(self, db, left_value):
384
        '''
385
        @param left_value The Code object that the condition is being applied on
386
        '''
387
        raise NotImplemented()
388
    
389
    def __repr__(self): return self.to_str(mockDb, '<left_value>')
390

    
391
class CompareCond(ValueCond):
392
    def __init__(self, value, operator='='):
393
        '''
394
        @param operator By default, compares NULL values literally. Use '~=' or
395
            '~!=' to pass NULLs through.
396
        '''
397
        ValueCond.__init__(self, value)
398
        self.operator = operator
399
    
400
    def to_str(self, db, left_value):
401
        left_value = remove_col_rename(as_Col(left_value))
402
        
403
        right_value = self.value
404
        
405
        # Parse operator
406
        operator = self.operator
407
        passthru_null_ref = [False]
408
        operator = strings.remove_prefix('~', operator, passthru_null_ref)
409
        neg_ref = [False]
410
        operator = strings.remove_prefix('!', operator, neg_ref)
411
        equals = operator.endswith('=') # also includes <=, >=
412
        
413
        # Handle nullable columns
414
        check_null = False
415
        if not passthru_null_ref[0]: # NULLs compare equal
416
            try: left_value = ensure_not_null(db, left_value)
417
            except ensure_not_null_excs: # fall back to alternate method
418
                check_null = equals and isinstance(right_value, Col)
419
            else:
420
                if isinstance(left_value, EnsureNotNull):
421
                    right_value = ensure_not_null(db, right_value,
422
                        left_value.type) # apply same function to both sides
423
        
424
        if equals and is_null(right_value): operator = 'IS'
425
        
426
        left = left_value.to_str(db)
427
        right = right_value.to_str(db)
428
        
429
        # Create str
430
        str_ = left+' '+operator+' '+right
431
        if check_null:
432
            str_ = '('+str_+' OR ('+left+' IS NULL AND '+right+' IS NULL))'
433
        if neg_ref[0]: str_ = 'NOT '+str_
434
        return str_
435

    
436
# Tells as_ValueCond() to assume a non-ValueCond is a literal value
437
assume_literal = object()
438

    
439
def as_ValueCond(value, default_table=assume_literal):
440
    if not isinstance(value, ValueCond):
441
        if default_table is not assume_literal:
442
            value = with_default_table(value, default_table)
443
        return CompareCond(value)
444
    else: return value
445

    
446
##### Joins
447

    
448
join_same = object() # tells Join the left and right columns have the same name
449

    
450
# Tells Join the left and right columns have the same name and are never NULL
451
join_same_not_null = object()
452

    
453
filter_out = object() # tells Join to filter out rows that match the join
454

    
455
class Join(BasicObject):
456
    def __init__(self, table, mapping={}, type_=None):
457
        '''
458
        @param mapping dict(right_table_col=left_table_col, ...)
459
            * if left_table_col is join_same: left_table_col = right_table_col
460
              * Note that right_table_col must be a string
461
            * if left_table_col is join_same_not_null:
462
              left_table_col = right_table_col and both have NOT NULL constraint
463
              * Note that right_table_col must be a string
464
        @param type_ None (for plain join)|str (e.g. 'LEFT')|filter_out
465
            * filter_out: equivalent to 'LEFT' with the query filtered by
466
              `table_pkey IS NULL` (indicating no match)
467
        '''
468
        if util.is_str(table): table = Table(table)
469
        assert type_ == None or util.is_str(type_) or type_ is filter_out
470
        
471
        self.table = table
472
        self.mapping = mapping
473
        self.type_ = type_
474
    
475
    def to_str(self, db, left_table_):
476
        def join(entry):
477
            '''Parses non-USING joins'''
478
            right_table_col, left_table_col = entry
479
            
480
            # Switch order (right_table_col is on the left in the comparison)
481
            left = right_table_col
482
            right = left_table_col
483
            left_table = self.table
484
            right_table = left_table_
485
            
486
            # Parse left side
487
            left = with_default_table(left, left_table)
488
            
489
            # Parse special values
490
            left_on_right = Col(left.name, right_table)
491
            if right is join_same: right = left_on_right
492
            elif right is join_same_not_null:
493
                right = CompareCond(left_on_right, '~=')
494
            
495
            # Parse right side
496
            right = as_ValueCond(right, right_table)
497
            
498
            return right.to_str(db, left)
499
        
500
        # Create join condition
501
        type_ = self.type_
502
        joins = self.mapping
503
        if joins == {}: join_cond = None
504
        elif type_ is not filter_out and reduce(operator.and_,
505
            (v is join_same_not_null for v in joins.itervalues())):
506
            # all cols w/ USING, so can use simpler USING syntax
507
            cols = map(to_name_only_col, joins.iterkeys())
508
            join_cond = 'USING ('+(', '.join((c.to_str(db) for c in cols)))+')'
509
        else: join_cond = combine_conds(map(join, joins.iteritems()), 'ON')
510
        
511
        if isinstance(self.table, NamedTable): whitespace = '\n'
512
        else: whitespace = ' '
513
        
514
        # Create join
515
        if type_ is filter_out: type_ = 'LEFT'
516
        str_ = ''
517
        if type_ != None: str_ += type_+' '
518
        str_ += 'JOIN'+whitespace+self.table.to_str(db)
519
        if join_cond != None: str_ += whitespace+join_cond
520
        return str_
521
    
522
    def __repr__(self): return self.to_str(mockDb, '<left_table>')
523

    
524
##### Value exprs
525

    
526
default = CustomCode('DEFAULT')
527

    
528
row_count = CustomCode('count(*)')
529

    
530
# See <http://www.postgresql.org/docs/8.3/static/datatype-numeric.html>
531
null_sentinels = {'integer': 2147483647, 'text': r'\N',
532
    'character varying': r'\N'}
533

    
534
class EnsureNotNull(FunctionCall):
535
    def __init__(self, value, type_):
536
        FunctionCall.__init__(self, InternalFunction('COALESCE'), as_Col(value),
537
            null_sentinels[type_])
538
        
539
        self.type = type_
540

    
541
##### Table exprs
542

    
543
class Values(Code):
544
    def __init__(self, values):
545
        '''
546
        @param values [...]|[[...], ...] Can be one or multiple rows.
547
        '''
548
        rows = values
549
        if len(values) >= 1 and not lists.is_seq(values[0]): # only one row
550
            rows = [values]
551
        for i, row in enumerate(rows):
552
            rows[i] = map(remove_col_rename, map(as_Value, row))
553
        
554
        self.rows = rows
555
    
556
    def to_str(self, db):
557
        def row_str(row):
558
            return '('+(', '.join((v.to_str(db) for v in row)))+')'
559
        return 'VALUES '+(', '.join(map(row_str, self.rows)))
560

    
561
def NamedValues(name, cols, values):
562
    '''
563
    @post `cols` will be changed to Col objects with the table set to `name`.
564
    '''
565
    table = NamedTable(name, Values(values), cols)
566
    set_cols_table(table, cols)
567
    return table
568

    
569
##### Database structure
570

    
571
class TypedCol(Col):
572
    def __init__(self, name, type_, default=None, nullable=True,
573
        constraints=None):
574
        assert default == None or isinstance(default, Code)
575
        
576
        Col.__init__(self, name)
577
        
578
        self.type = type_
579
        self.default = default
580
        self.nullable = nullable
581
        self.constraints = constraints
582
    
583
    def to_str(self, db):
584
        str_ = Col.to_str(self, db)+' '+self.type
585
        if not self.nullable: str_ += ' NOT NULL'
586
        if self.default != None: str_ += ' DEFAULT '+self.default.to_str(db)
587
        if self.constraints != None: str_ += ' '+self.constraints
588
        return str_
589
    
590
    def to_Col(self): return Col(self.name)
591

    
592
ensure_not_null_excs = (NoUnderlyingTableException, KeyError)
593

    
594
def ensure_not_null(db, col, type_=None):
595
    '''
596
    @param col If type_ is not set, must have an underlying column.
597
    @param type_ If set, overrides the underlying column's type.
598
    @return EnsureNotNull|Col
599
    @throws ensure_not_null_excs
600
    '''
601
    nullable = True
602
    try: typed_col = db.col_info(underlying_col(col))
603
    except NoUnderlyingTableException:
604
        if type_ == None: raise
605
    else:
606
        if type_ == None: type_ = typed_col.type
607
        nullable = typed_col.nullable
608
    
609
    if nullable: col = EnsureNotNull(col, type_)
610
    return col
(25-25/36)