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 concat(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.concat(str_, suffix, identifier_max_len)
29

    
30
def truncate(str_): return concat(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, use_concat=False):
210
        str_ = db.esc_name(self.name)
211
        if self.table != None:
212
            str_ = '.'+str_
213
            table_str = self.table.to_Table().to_str(db)
214
            if use_concat: str_ = concat(table_str, str_)
215
            else: str_ = table_str+str_
216
        return str_
217
    
218
    def __str__(self): return clean_name(self.to_str(mockDb, use_concat=True))
219
    
220
    def to_Col(self): return self
221

    
222
def is_table_col(col): return isinstance(col, Col) and col.table != None
223

    
224
def as_Col(col, table=None, name=None):
225
    '''
226
    @param name If not None, any non-Col input will be renamed using NamedCol.
227
    '''
228
    if name != None:
229
        col = as_Value(col)
230
        if not isinstance(col, Col): col = NamedCol(name, col)
231
    
232
    if isinstance(col, Code): return col
233
    else: return Col(col, table)
234

    
235
def with_default_table(col, table, overwrite=False):
236
    col = as_Col(col)
237
    if not isinstance(col, NamedCol) and (overwrite or col.table == None):
238
        col = copy.copy(col) # don't modify input!
239
        col.table = table
240
    return col
241

    
242
def set_cols_table(table, cols):
243
    table = as_Table(table)
244
    
245
    for i, col in enumerate(cols):
246
        col = cols[i] = as_Col(col)
247
        col.table = table
248

    
249
def to_name_only_col(col, check_table=None):
250
    col = as_Col(col)
251
    if not isinstance(col, Col): return col
252
    
253
    if check_table != None:
254
        table = col.table
255
        assert table == None or table == check_table
256
    return Col(col.name)
257

    
258
class NamedCol(Col):
259
    def __init__(self, name, code):
260
        Col.__init__(self, name)
261
        
262
        if not isinstance(code, Code): code = Literal(code)
263
        
264
        self.code = code
265
    
266
    def to_str(self, db):
267
        return self.code.to_str(db)+' AS '+Col.to_str(self, db)
268
    
269
    def to_Col(self): return Col(self.name)
270

    
271
def remove_col_rename(col):
272
    if isinstance(col, NamedCol): col = col.code
273
    return col
274

    
275
def underlying_col(col):
276
    col = remove_col_rename(col)
277
    if not isinstance(col, Col): raise NoUnderlyingTableException
278
    
279
    return Col(col.name, underlying_table(col.table), col.srcs)
280

    
281
def wrap(wrap_func, value):
282
    '''Wraps a value, propagating any column renaming to the returned value.'''
283
    if isinstance(value, NamedCol):
284
        return NamedCol(value.name, wrap_func(value.code))
285
    else: return wrap_func(value)
286

    
287
class ColDict(dicts.DictProxy):
288
    '''A dict that automatically makes inserted entries Col objects'''
289
    
290
    def __init__(self, db, keys_table, dict_={}):
291
        dicts.DictProxy.__init__(self, {})
292
        
293
        keys_table = as_Table(keys_table)
294
        
295
        self.db = db
296
        self.table = keys_table
297
        self.update(dict_) # after setting vars because __setitem__() needs them
298
    
299
    def copy(self): return ColDict(self.db, self.table, self.inner.copy())
300
    
301
    def __getitem__(self, key):
302
        return dicts.DictProxy.__getitem__(self, self._key(key))
303
    
304
    def __setitem__(self, key, value):
305
        key = self._key(key)
306
        if value == None: value = self.db.col_info(key).default
307
        dicts.DictProxy.__setitem__(self, key, as_Col(value, name=key.name))
308
    
309
    def _key(self, key): return as_Col(key, self.table)
310

    
311
##### Functions
312

    
313
Function = Table
314
as_Function = as_Table
315

    
316
class InternalFunction(CustomCode): pass
317

    
318
class FunctionCall(Code):
319
    def __init__(self, function, *args):
320
        '''
321
        @param args [Code|literal-value...] The function's arguments
322
        '''
323
        if not isinstance(function, Code): function = Function(function)
324
        args = map(remove_col_rename, map(as_Value, args))
325
        
326
        self.function = function
327
        self.args = args
328
    
329
    def to_str(self, db):
330
        args_str = ', '.join((v.to_str(db) for v in self.args))
331
        return self.function.to_str(db)+'('+args_str+')'
332

    
333
def wrap_in_func(function, value):
334
    '''Wraps a value inside a function call.
335
    Propagates any column renaming to the returned value.
336
    '''
337
    return wrap(lambda v: FunctionCall(function, v), value)
338

    
339
def unwrap_func_call(func_call, check_name=None):
340
    '''Unwraps any function call to its first argument.
341
    Also removes any column renaming.
342
    '''
343
    func_call = remove_col_rename(func_call)
344
    if not isinstance(func_call, FunctionCall): return func_call
345
    
346
    if check_name != None:
347
        name = func_call.function.name
348
        assert name == None or name == check_name
349
    return func_call.args[0]
350

    
351
##### Conditions
352

    
353
class ColValueCond(Code):
354
    def __init__(self, col, value):
355
        value = as_ValueCond(value)
356
        
357
        self.col = col
358
        self.value = value
359
    
360
    def to_str(self, db): return self.value.to_str(db, self.col)
361

    
362
def combine_conds(conds, keyword=None):
363
    '''
364
    @param keyword The keyword to add before the conditions, if any
365
    '''
366
    str_ = ''
367
    if keyword != None:
368
        if conds == []: whitespace = ''
369
        elif len(conds) == 1: whitespace = ' '
370
        else: whitespace = '\n'
371
        str_ += keyword+whitespace
372
    
373
    str_ += '\nAND '.join(conds)
374
    return str_
375

    
376
##### Condition column comparisons
377

    
378
class ValueCond(BasicObject):
379
    def __init__(self, value):
380
        value = remove_col_rename(as_Value(value))
381
        
382
        self.value = value
383
    
384
    def to_str(self, db, left_value):
385
        '''
386
        @param left_value The Code object that the condition is being applied on
387
        '''
388
        raise NotImplemented()
389
    
390
    def __repr__(self): return self.to_str(mockDb, '<left_value>')
391

    
392
class CompareCond(ValueCond):
393
    def __init__(self, value, operator='='):
394
        '''
395
        @param operator By default, compares NULL values literally. Use '~=' or
396
            '~!=' to pass NULLs through.
397
        '''
398
        ValueCond.__init__(self, value)
399
        self.operator = operator
400
    
401
    def to_str(self, db, left_value):
402
        left_value = remove_col_rename(as_Col(left_value))
403
        
404
        right_value = self.value
405
        
406
        # Parse operator
407
        operator = self.operator
408
        passthru_null_ref = [False]
409
        operator = strings.remove_prefix('~', operator, passthru_null_ref)
410
        neg_ref = [False]
411
        operator = strings.remove_prefix('!', operator, neg_ref)
412
        equals = operator.endswith('=') # also includes <=, >=
413
        
414
        # Handle nullable columns
415
        check_null = False
416
        if not passthru_null_ref[0]: # NULLs compare equal
417
            try: left_value = ensure_not_null(db, left_value)
418
            except ensure_not_null_excs: # fall back to alternate method
419
                check_null = equals and isinstance(right_value, Col)
420
            else:
421
                if isinstance(left_value, EnsureNotNull):
422
                    right_value = ensure_not_null(db, right_value,
423
                        left_value.type) # apply same function to both sides
424
        
425
        if equals and is_null(right_value): operator = 'IS'
426
        
427
        left = left_value.to_str(db)
428
        right = right_value.to_str(db)
429
        
430
        # Create str
431
        str_ = left+' '+operator+' '+right
432
        if check_null:
433
            str_ = '('+str_+' OR ('+left+' IS NULL AND '+right+' IS NULL))'
434
        if neg_ref[0]: str_ = 'NOT '+str_
435
        return str_
436

    
437
# Tells as_ValueCond() to assume a non-ValueCond is a literal value
438
assume_literal = object()
439

    
440
def as_ValueCond(value, default_table=assume_literal):
441
    if not isinstance(value, ValueCond):
442
        if default_table is not assume_literal:
443
            value = with_default_table(value, default_table)
444
        return CompareCond(value)
445
    else: return value
446

    
447
##### Joins
448

    
449
join_same = object() # tells Join the left and right columns have the same name
450

    
451
# Tells Join the left and right columns have the same name and are never NULL
452
join_same_not_null = object()
453

    
454
filter_out = object() # tells Join to filter out rows that match the join
455

    
456
class Join(BasicObject):
457
    def __init__(self, table, mapping={}, type_=None):
458
        '''
459
        @param mapping dict(right_table_col=left_table_col, ...)
460
            * if left_table_col is join_same: left_table_col = right_table_col
461
              * Note that right_table_col must be a string
462
            * if left_table_col is join_same_not_null:
463
              left_table_col = right_table_col and both have NOT NULL constraint
464
              * Note that right_table_col must be a string
465
        @param type_ None (for plain join)|str (e.g. 'LEFT')|filter_out
466
            * filter_out: equivalent to 'LEFT' with the query filtered by
467
              `table_pkey IS NULL` (indicating no match)
468
        '''
469
        if util.is_str(table): table = Table(table)
470
        assert type_ == None or util.is_str(type_) or type_ is filter_out
471
        
472
        self.table = table
473
        self.mapping = mapping
474
        self.type_ = type_
475
    
476
    def to_str(self, db, left_table_):
477
        def join(entry):
478
            '''Parses non-USING joins'''
479
            right_table_col, left_table_col = entry
480
            
481
            # Switch order (right_table_col is on the left in the comparison)
482
            left = right_table_col
483
            right = left_table_col
484
            left_table = self.table
485
            right_table = left_table_
486
            
487
            # Parse left side
488
            left = with_default_table(left, left_table)
489
            
490
            # Parse special values
491
            left_on_right = Col(left.name, right_table)
492
            if right is join_same: right = left_on_right
493
            elif right is join_same_not_null:
494
                right = CompareCond(left_on_right, '~=')
495
            
496
            # Parse right side
497
            right = as_ValueCond(right, right_table)
498
            
499
            return right.to_str(db, left)
500
        
501
        # Create join condition
502
        type_ = self.type_
503
        joins = self.mapping
504
        if joins == {}: join_cond = None
505
        elif type_ is not filter_out and reduce(operator.and_,
506
            (v is join_same_not_null for v in joins.itervalues())):
507
            # all cols w/ USING, so can use simpler USING syntax
508
            cols = map(to_name_only_col, joins.iterkeys())
509
            join_cond = 'USING ('+(', '.join((c.to_str(db) for c in cols)))+')'
510
        else: join_cond = combine_conds(map(join, joins.iteritems()), 'ON')
511
        
512
        if isinstance(self.table, NamedTable): whitespace = '\n'
513
        else: whitespace = ' '
514
        
515
        # Create join
516
        if type_ is filter_out: type_ = 'LEFT'
517
        str_ = ''
518
        if type_ != None: str_ += type_+' '
519
        str_ += 'JOIN'+whitespace+self.table.to_str(db)
520
        if join_cond != None: str_ += whitespace+join_cond
521
        return str_
522
    
523
    def __repr__(self): return self.to_str(mockDb, '<left_table>')
524

    
525
##### Value exprs
526

    
527
default = CustomCode('DEFAULT')
528

    
529
row_count = CustomCode('count(*)')
530

    
531
# See <http://www.postgresql.org/docs/8.3/static/datatype-numeric.html>
532
null_sentinels = {'integer': 2147483647, 'text': r'\N',
533
    'character varying': r'\N'}
534

    
535
class EnsureNotNull(FunctionCall):
536
    def __init__(self, value, type_):
537
        FunctionCall.__init__(self, InternalFunction('COALESCE'), as_Col(value),
538
            null_sentinels[type_])
539
        
540
        self.type = type_
541

    
542
##### Table exprs
543

    
544
class Values(Code):
545
    def __init__(self, values):
546
        '''
547
        @param values [...]|[[...], ...] Can be one or multiple rows.
548
        '''
549
        rows = values
550
        if len(values) >= 1 and not lists.is_seq(values[0]): # only one row
551
            rows = [values]
552
        for i, row in enumerate(rows):
553
            rows[i] = map(remove_col_rename, map(as_Value, row))
554
        
555
        self.rows = rows
556
    
557
    def to_str(self, db):
558
        def row_str(row):
559
            return '('+(', '.join((v.to_str(db) for v in row)))+')'
560
        return 'VALUES '+(', '.join(map(row_str, self.rows)))
561

    
562
def NamedValues(name, cols, values):
563
    '''
564
    @post `cols` will be changed to Col objects with the table set to `name`.
565
    '''
566
    table = NamedTable(name, Values(values), cols)
567
    set_cols_table(table, cols)
568
    return table
569

    
570
##### Database structure
571

    
572
class TypedCol(Col):
573
    def __init__(self, name, type_, default=None, nullable=True,
574
        constraints=None):
575
        assert default == None or isinstance(default, Code)
576
        
577
        Col.__init__(self, name)
578
        
579
        self.type = type_
580
        self.default = default
581
        self.nullable = nullable
582
        self.constraints = constraints
583
    
584
    def to_str(self, db):
585
        str_ = Col.to_str(self, db)+' '+self.type
586
        if not self.nullable: str_ += ' NOT NULL'
587
        if self.default != None: str_ += ' DEFAULT '+self.default.to_str(db)
588
        if self.constraints != None: str_ += ' '+self.constraints
589
        return str_
590
    
591
    def to_Col(self): return Col(self.name)
592

    
593
ensure_not_null_excs = (NoUnderlyingTableException, KeyError)
594

    
595
def ensure_not_null(db, col, type_=None):
596
    '''
597
    @param col If type_ is not set, must have an underlying column.
598
    @param type_ If set, overrides the underlying column's type.
599
    @return EnsureNotNull|Col
600
    @throws ensure_not_null_excs
601
    '''
602
    nullable = True
603
    try: typed_col = db.col_info(underlying_col(col))
604
    except NoUnderlyingTableException:
605
        if type_ == None: raise
606
    else:
607
        if type_ == None: type_ = typed_col.type
608
        nullable = typed_col.nullable
609
    
610
    if nullable: col = EnsureNotNull(col, type_)
611
    return col
(25-25/36)