Project

General

Profile

1 2211 aaronmk
# SQL code generation
2
3 2748 aaronmk
import copy
4 2276 aaronmk
import operator
5 2568 aaronmk
import re
6 2653 aaronmk
import UserDict
7 2276 aaronmk
8 2667 aaronmk
import dicts
9 2701 aaronmk
import iters
10
import lists
11 2360 aaronmk
import objects
12 2222 aaronmk
import strings
13 2227 aaronmk
import util
14 2211 aaronmk
15 2587 aaronmk
##### Names
16 2499 aaronmk
17 2608 aaronmk
identifier_max_len = 63 # works for both PostgreSQL and MySQL
18 2587 aaronmk
19
def add_suffix(str_, suffix):
20 2609 aaronmk
    '''Preserves version so that it won't be truncated off the string, leading
21
    to collisions.'''
22 2613 aaronmk
    # Preserve version
23
    before, sep, version = str_.rpartition('#')
24
    if sep != '': # found match
25
        str_ = before
26 2609 aaronmk
        suffix = sep+version+suffix
27 2613 aaronmk
28 2587 aaronmk
    return strings.add_suffix(str_, suffix, identifier_max_len)
29
30 2575 aaronmk
def is_safe_name(name):
31 2583 aaronmk
    '''A name is safe *and unambiguous* if it:
32
    * contains only *lowercase* word (\w) characters
33
    * doesn't start with a digit
34
    * contains "_", so that it's not a keyword
35
    '''
36
    return re.match(r'^(?=.*_)(?!\d)[^\WA-Z]+$', name)
37 2568 aaronmk
38 2499 aaronmk
def esc_name(name, quote='"'):
39
    return quote + name.replace(quote, quote+quote) + quote
40
        # doubling an embedded quote escapes it in both PostgreSQL and MySQL
41
42 2513 aaronmk
def clean_name(name): return name.replace('"', '').replace('`', '')
43
44 2659 aaronmk
##### General SQL code objects
45 2219 aaronmk
46 2349 aaronmk
class MockDb:
47 2503 aaronmk
    def esc_value(self, value): return strings.repr_no_u(value)
48 2349 aaronmk
49 2499 aaronmk
    def esc_name(self, name): return esc_name(name)
50 2349 aaronmk
mockDb = MockDb()
51
52 2514 aaronmk
class BasicObject(objects.BasicObject):
53
    def __init__(self, value): self.value = value
54
55
    def __str__(self): return clean_name(strings.repr_no_u(self))
56
57 2659 aaronmk
##### Unparameterized code objects
58
59 2514 aaronmk
class Code(BasicObject):
60 2658 aaronmk
    def to_str(self, db): raise NotImplementedError()
61 2349 aaronmk
62 2514 aaronmk
    def __repr__(self): return self.to_str(mockDb)
63 2211 aaronmk
64 2269 aaronmk
class CustomCode(Code):
65 2256 aaronmk
    def __init__(self, str_): self.str_ = str_
66
67
    def to_str(self, db): return self.str_
68
69 2815 aaronmk
def as_Code(value, db=None):
70
    '''
71
    @param db If set, runs db.std_code() on the value.
72
    '''
73
    if util.is_str(value):
74
        if db != None: value = db.std_code(value)
75
        return CustomCode(value)
76 2659 aaronmk
    else: return Literal(value)
77
78 2540 aaronmk
class Expr(Code):
79
    def __init__(self, expr): self.expr = expr
80
81
    def to_str(self, db): return '('+self.expr.to_str(db)+')'
82
83 2335 aaronmk
##### Literal values
84
85 2216 aaronmk
class Literal(Code):
86 2211 aaronmk
    def __init__(self, value): self.value = value
87 2213 aaronmk
88
    def to_str(self, db): return db.esc_value(self.value)
89 2211 aaronmk
90 2400 aaronmk
def as_Value(value):
91
    if isinstance(value, Code): return value
92
    else: return Literal(value)
93
94 2216 aaronmk
def is_null(value): return isinstance(value, Literal) and value.value == None
95
96 2711 aaronmk
##### Derived elements
97
98
src_self = object() # tells Col that it is its own source column
99
100
class Derived(Code):
101
    def __init__(self, srcs):
102 2712 aaronmk
        '''An element which was derived from some other element(s).
103 2711 aaronmk
        @param srcs See self.set_srcs()
104
        '''
105
        self.set_srcs(srcs)
106
107 2713 aaronmk
    def set_srcs(self, srcs, overwrite=True):
108 2711 aaronmk
        '''
109
        @param srcs (self_type...)|src_self The element(s) this is derived from
110
        '''
111 2713 aaronmk
        if not overwrite and self.srcs != (): return # already set
112
113 2711 aaronmk
        if srcs == src_self: srcs = (self,)
114
        srcs = tuple(srcs) # make Col hashable
115
        self.srcs = srcs
116
117
    def _compare_on(self):
118
        compare_on = self.__dict__.copy()
119
        del compare_on['srcs'] # ignore
120
        return compare_on
121
122
def cols_srcs(cols): return lists.uniqify(iters.flatten((v.srcs for v in cols)))
123
124 2335 aaronmk
##### Tables
125
126 2712 aaronmk
class Table(Derived):
127
    def __init__(self, name, schema=None, srcs=()):
128 2211 aaronmk
        '''
129
        @param schema str|None (for no schema)
130 2712 aaronmk
        @param srcs (Table...)|src_self See Derived.set_srcs()
131 2211 aaronmk
        '''
132 2712 aaronmk
        Derived.__init__(self, srcs)
133
134 2211 aaronmk
        self.name = name
135
        self.schema = schema
136
137 2348 aaronmk
    def to_str(self, db):
138
        str_ = ''
139
        if self.schema != None: str_ += db.esc_name(self.schema)+'.'
140
        str_ += db.esc_name(self.name)
141
        return str_
142 2336 aaronmk
143
    def to_Table(self): return self
144 2211 aaronmk
145 2776 aaronmk
def as_Table(table, schema=None):
146 2270 aaronmk
    if table == None or isinstance(table, Code): return table
147 2776 aaronmk
    else: return Table(table, schema)
148 2219 aaronmk
149 2707 aaronmk
def suffixed_table(table, suffix): return Table(table.name+suffix, table.schema)
150
151 2336 aaronmk
class NamedTable(Table):
152
    def __init__(self, name, code, cols=None):
153
        Table.__init__(self, name)
154
155
        if not isinstance(code, Code): code = Table(code)
156 2741 aaronmk
        if not isinstance(code, (Table, FunctionCall, Expr)): code = Expr(code)
157 2742 aaronmk
        if cols != None: cols = map(to_name_only_col, cols)
158 2336 aaronmk
159
        self.code = code
160
        self.cols = cols
161
162
    def to_str(self, db):
163 2467 aaronmk
        str_ = self.code.to_str(db)+'\nAS '+Table.to_str(self, db)
164 2742 aaronmk
        if self.cols != None:
165
            str_ += ' ('+(', '.join((c.to_str(db) for c in self.cols)))+')'
166 2336 aaronmk
        return str_
167
168
    def to_Table(self): return Table(self.name)
169
170 2753 aaronmk
def remove_table_rename(table):
171
    if isinstance(table, NamedTable): table = table.code
172
    return table
173
174 2335 aaronmk
##### Columns
175
176 2711 aaronmk
class Col(Derived):
177 2701 aaronmk
    def __init__(self, name, table=None, srcs=()):
178 2211 aaronmk
        '''
179
        @param table Table|None (for no table)
180 2711 aaronmk
        @param srcs (Col...)|src_self See Derived.set_srcs()
181 2211 aaronmk
        '''
182 2711 aaronmk
        Derived.__init__(self, srcs)
183
184 2241 aaronmk
        if util.is_str(table): table = Table(table)
185 2211 aaronmk
        assert table == None or isinstance(table, Table)
186
187
        self.name = name
188
        self.table = table
189
190
    def to_str(self, db):
191
        str_ = ''
192 2827 aaronmk
        if self.table != None: str_ += self.table.to_Table().to_str(db)+'.'
193 2348 aaronmk
        str_ += db.esc_name(self.name)
194 2211 aaronmk
        return str_
195 2314 aaronmk
196
    def to_Col(self): return self
197 2211 aaronmk
198 2767 aaronmk
def is_table_col(col): return isinstance(col, Col) and col.table != None
199 2393 aaronmk
200 2563 aaronmk
def as_Col(col, table=None, name=None):
201
    '''
202
    @param name If not None, any non-Col input will be renamed using NamedCol.
203
    '''
204
    if name != None:
205
        col = as_Value(col)
206
        if not isinstance(col, Col): col = NamedCol(name, col)
207 2333 aaronmk
208
    if isinstance(col, Code): return col
209 2260 aaronmk
    else: return Col(col, table)
210
211 2750 aaronmk
def with_default_table(col, table, overwrite=False):
212 2747 aaronmk
    col = as_Col(col)
213 2750 aaronmk
    if not isinstance(col, NamedCol) and (overwrite or col.table == None):
214 2748 aaronmk
        col = copy.copy(col) # don't modify input!
215
        col.table = table
216 2747 aaronmk
    return col
217
218 2744 aaronmk
def set_cols_table(table, cols):
219
    table = as_Table(table)
220
221
    for i, col in enumerate(cols):
222
        col = cols[i] = as_Col(col)
223
        col.table = table
224
225 2401 aaronmk
def to_name_only_col(col, check_table=None):
226
    col = as_Col(col)
227 2579 aaronmk
    if not isinstance(col, Col): return col
228 2401 aaronmk
229
    if check_table != None:
230
        table = col.table
231
        assert table == None or table == check_table
232
    return Col(col.name)
233
234 2323 aaronmk
class NamedCol(Col):
235 2229 aaronmk
    def __init__(self, name, code):
236 2310 aaronmk
        Col.__init__(self, name)
237
238 2229 aaronmk
        if not isinstance(code, Code): code = Literal(code)
239
240
        self.code = code
241
242
    def to_str(self, db):
243 2310 aaronmk
        return self.code.to_str(db)+' AS '+Col.to_str(self, db)
244 2314 aaronmk
245
    def to_Col(self): return Col(self.name)
246 2229 aaronmk
247 2462 aaronmk
def remove_col_rename(col):
248
    if isinstance(col, NamedCol): col = col.code
249
    return col
250
251 2703 aaronmk
def wrap(wrap_func, value):
252
    '''Wraps a value, propagating any column renaming to the returned value.'''
253
    if isinstance(value, NamedCol):
254
        return NamedCol(value.name, wrap_func(value.code))
255
    else: return wrap_func(value)
256
257 2667 aaronmk
class ColDict(dicts.DictProxy):
258 2564 aaronmk
    '''A dict that automatically makes inserted entries Col objects'''
259
260 2645 aaronmk
    def __init__(self, db, keys_table, dict_={}):
261 2667 aaronmk
        dicts.DictProxy.__init__(self, {})
262
263 2645 aaronmk
        keys_table = as_Table(keys_table)
264
265 2642 aaronmk
        self.db = db
266 2641 aaronmk
        self.table = keys_table
267 2653 aaronmk
        self.update(dict_) # after setting vars because __setitem__() needs them
268 2641 aaronmk
269 2667 aaronmk
    def copy(self): return ColDict(self.db, self.table, self.inner.copy())
270 2655 aaronmk
271 2667 aaronmk
    def __getitem__(self, key):
272
        return dicts.DictProxy.__getitem__(self, self._key(key))
273 2653 aaronmk
274 2564 aaronmk
    def __setitem__(self, key, value):
275 2642 aaronmk
        key = self._key(key)
276 2819 aaronmk
        if value == None: value = self.db.col_info(key).default
277 2667 aaronmk
        dicts.DictProxy.__setitem__(self, key, as_Col(value, name=key.name))
278 2564 aaronmk
279 2641 aaronmk
    def _key(self, key): return as_Col(key, self.table)
280 2564 aaronmk
281 2524 aaronmk
##### Functions
282
283
class Function(Table): pass
284
285 2697 aaronmk
def TempFunction(name, autocommit):
286
    schema = None
287
    if not autocommit: schema = 'pg_temp'
288
    return Function(name, schema)
289
290 2691 aaronmk
class InternalFunction(CustomCode): pass
291
292 2524 aaronmk
class FunctionCall(Code):
293
    def __init__(self, function, *args):
294
        '''
295 2690 aaronmk
        @param args [Code|literal-value...] The function's arguments
296 2524 aaronmk
        '''
297
        if not isinstance(function, Code): function = Function(function)
298 2690 aaronmk
        args = map(remove_col_rename, map(as_Value, args))
299 2524 aaronmk
300
        self.function = function
301
        self.args = args
302
303
    def to_str(self, db):
304
        args_str = ', '.join((v.to_str(db) for v in self.args))
305
        return self.function.to_str(db)+'('+args_str+')'
306
307 2533 aaronmk
def wrap_in_func(function, value):
308
    '''Wraps a value inside a function call.
309
    Propagates any column renaming to the returned value.
310
    '''
311 2703 aaronmk
    return wrap(lambda v: FunctionCall(function, v), value)
312 2533 aaronmk
313 2561 aaronmk
def unwrap_func_call(func_call, check_name=None):
314
    '''Unwraps any function call to its first argument.
315
    Also removes any column renaming.
316
    '''
317
    func_call = remove_col_rename(func_call)
318
    if not isinstance(func_call, FunctionCall): return func_call
319
320
    if check_name != None:
321
        name = func_call.function.name
322
        assert name == None or name == check_name
323
    return func_call.args[0]
324
325 2335 aaronmk
##### Conditions
326 2259 aaronmk
327 2398 aaronmk
class ColValueCond(Code):
328
    def __init__(self, col, value):
329
        value = as_ValueCond(value)
330
331
        self.col = col
332
        self.value = value
333
334
    def to_str(self, db): return self.value.to_str(db, self.col)
335
336 2577 aaronmk
def combine_conds(conds, keyword=None):
337
    '''
338
    @param keyword The keyword to add before the conditions, if any
339
    '''
340
    str_ = ''
341
    if keyword != None:
342
        if conds == []: whitespace = ''
343
        elif len(conds) == 1: whitespace = ' '
344
        else: whitespace = '\n'
345
        str_ += keyword+whitespace
346
347
    str_ += '\nAND '.join(conds)
348
    return str_
349
350 2398 aaronmk
##### Condition column comparisons
351
352 2514 aaronmk
class ValueCond(BasicObject):
353 2213 aaronmk
    def __init__(self, value):
354 2225 aaronmk
        if not isinstance(value, Code): value = Literal(value)
355 2462 aaronmk
        value = remove_col_rename(value)
356 2213 aaronmk
357
        self.value = value
358 2214 aaronmk
359 2216 aaronmk
    def to_str(self, db, left_value):
360 2214 aaronmk
        '''
361 2216 aaronmk
        @param left_value The Code object that the condition is being applied on
362 2214 aaronmk
        '''
363
        raise NotImplemented()
364 2228 aaronmk
365 2514 aaronmk
    def __repr__(self): return self.to_str(mockDb, '<left_value>')
366 2211 aaronmk
367
class CompareCond(ValueCond):
368
    def __init__(self, value, operator='='):
369 2222 aaronmk
        '''
370
        @param operator By default, compares NULL values literally. Use '~=' or
371
            '~!=' to pass NULLs through.
372
        '''
373 2211 aaronmk
        ValueCond.__init__(self, value)
374
        self.operator = operator
375
376 2216 aaronmk
    def to_str(self, db, left_value):
377
        if not isinstance(left_value, Code): left_value = Col(left_value)
378 2462 aaronmk
        left_value = remove_col_rename(left_value)
379 2216 aaronmk
380 2222 aaronmk
        right_value = self.value
381
382
        # Parse operator
383 2216 aaronmk
        operator = self.operator
384 2222 aaronmk
        passthru_null_ref = [False]
385
        operator = strings.remove_prefix('~', operator, passthru_null_ref)
386
        neg_ref = [False]
387
        operator = strings.remove_prefix('!', operator, neg_ref)
388
        equals = operator.endswith('=')
389 2825 aaronmk
        if equals and is_null(right_value): operator = 'IS'
390 2222 aaronmk
391 2825 aaronmk
        # Handle nullable columns
392
        check_null = False
393
        if equals and not passthru_null_ref[0] and isinstance(right_value, Col):
394
            check_null = True
395
396
        left = left_value.to_str(db)
397
        right = right_value.to_str(db)
398
399 2222 aaronmk
        # Create str
400
        str_ = left+' '+operator+' '+right
401 2825 aaronmk
        if check_null:
402 2578 aaronmk
            str_ = '('+str_+' OR ('+left+' IS NULL AND '+right+' IS NULL))'
403
        if neg_ref[0]: str_ = 'NOT '+str_
404 2222 aaronmk
        return str_
405 2216 aaronmk
406 2260 aaronmk
# Tells as_ValueCond() to assume a non-ValueCond is a literal value
407
assume_literal = object()
408
409
def as_ValueCond(value, default_table=assume_literal):
410
    if not isinstance(value, ValueCond):
411
        if default_table is not assume_literal:
412 2748 aaronmk
            value = with_default_table(value, default_table)
413 2260 aaronmk
        return CompareCond(value)
414 2216 aaronmk
    else: return value
415 2219 aaronmk
416 2335 aaronmk
##### Joins
417
418 2352 aaronmk
join_same = object() # tells Join the left and right columns have the same name
419 2260 aaronmk
420 2353 aaronmk
# Tells Join the left and right columns have the same name and are never NULL
421
join_same_not_null = object()
422
423 2260 aaronmk
filter_out = object() # tells Join to filter out rows that match the join
424
425 2514 aaronmk
class Join(BasicObject):
426 2746 aaronmk
    def __init__(self, table, mapping={}, type_=None):
427 2260 aaronmk
        '''
428
        @param mapping dict(right_table_col=left_table_col, ...)
429 2352 aaronmk
            * if left_table_col is join_same: left_table_col = right_table_col
430 2353 aaronmk
              * Note that right_table_col must be a string
431
            * if left_table_col is join_same_not_null:
432
              left_table_col = right_table_col and both have NOT NULL constraint
433
              * Note that right_table_col must be a string
434 2260 aaronmk
        @param type_ None (for plain join)|str (e.g. 'LEFT')|filter_out
435
            * filter_out: equivalent to 'LEFT' with the query filtered by
436
              `table_pkey IS NULL` (indicating no match)
437
        '''
438
        if util.is_str(table): table = Table(table)
439
        assert type_ == None or util.is_str(type_) or type_ is filter_out
440
441
        self.table = table
442
        self.mapping = mapping
443
        self.type_ = type_
444
445 2749 aaronmk
    def to_str(self, db, left_table_):
446 2260 aaronmk
        def join(entry):
447
            '''Parses non-USING joins'''
448
            right_table_col, left_table_col = entry
449
450 2353 aaronmk
            # Switch order (right_table_col is on the left in the comparison)
451
            left = right_table_col
452
            right = left_table_col
453 2749 aaronmk
            left_table = self.table
454
            right_table = left_table_
455 2353 aaronmk
456 2747 aaronmk
            # Parse left side
457 2748 aaronmk
            left = with_default_table(left, left_table)
458 2747 aaronmk
459 2260 aaronmk
            # Parse special values
460 2747 aaronmk
            left_on_right = Col(left.name, right_table)
461
            if right is join_same: right = left_on_right
462 2353 aaronmk
            elif right is join_same_not_null:
463 2747 aaronmk
                right = CompareCond(left_on_right, '~=')
464 2260 aaronmk
465 2747 aaronmk
            # Parse right side
466 2353 aaronmk
            right = as_ValueCond(right, right_table)
467 2747 aaronmk
468
            return right.to_str(db, left)
469 2260 aaronmk
470 2265 aaronmk
        # Create join condition
471
        type_ = self.type_
472 2276 aaronmk
        joins = self.mapping
473 2746 aaronmk
        if joins == {}: join_cond = None
474
        elif type_ is not filter_out and reduce(operator.and_,
475 2460 aaronmk
            (v is join_same_not_null for v in joins.itervalues())):
476 2260 aaronmk
            # all cols w/ USING, so can use simpler USING syntax
477 2747 aaronmk
            cols = map(to_name_only_col, joins.iterkeys())
478
            join_cond = 'USING ('+(', '.join((c.to_str(db) for c in cols)))+')'
479 2757 aaronmk
        else: join_cond = combine_conds(map(join, joins.iteritems()), 'ON')
480 2260 aaronmk
481 2757 aaronmk
        if isinstance(self.table, NamedTable): whitespace = '\n'
482
        else: whitespace = ' '
483
484 2260 aaronmk
        # Create join
485
        if type_ is filter_out: type_ = 'LEFT'
486 2266 aaronmk
        str_ = ''
487
        if type_ != None: str_ += type_+' '
488 2757 aaronmk
        str_ += 'JOIN'+whitespace+self.table.to_str(db)
489
        if join_cond != None: str_ += whitespace+join_cond
490 2266 aaronmk
        return str_
491 2349 aaronmk
492 2514 aaronmk
    def __repr__(self): return self.to_str(mockDb, '<left_table>')
493 2424 aaronmk
494
##### Value exprs
495
496 2737 aaronmk
default = CustomCode('DEFAULT')
497
498 2424 aaronmk
row_count = CustomCode('count(*)')
499 2674 aaronmk
500 2692 aaronmk
def EnsureNotNull(value, null=r'\N'):
501 2694 aaronmk
    return FunctionCall(InternalFunction('coalesce'), as_Col(value), null)
502 2692 aaronmk
503 2737 aaronmk
##### Table exprs
504
505
class Values(Code):
506
    def __init__(self, values):
507 2739 aaronmk
        '''
508
        @param values [...]|[[...], ...] Can be one or multiple rows.
509
        '''
510
        rows = values
511
        if len(values) >= 1 and not lists.is_seq(values[0]): # only one row
512
            rows = [values]
513
        for i, row in enumerate(rows):
514
            rows[i] = map(remove_col_rename, map(as_Value, row))
515 2737 aaronmk
516 2739 aaronmk
        self.rows = rows
517 2737 aaronmk
518
    def to_str(self, db):
519 2739 aaronmk
        def row_str(row):
520
            return '('+(', '.join((v.to_str(db) for v in row)))+')'
521
        return 'VALUES '+(', '.join(map(row_str, self.rows)))
522 2737 aaronmk
523 2740 aaronmk
def NamedValues(name, cols, values):
524 2745 aaronmk
    '''
525
    @post `cols` will be changed to Col objects with the table set to `name`.
526
    '''
527
    set_cols_table(name, cols)
528 2740 aaronmk
    return NamedTable(name, Values(values), cols)
529
530 2674 aaronmk
##### Database structure
531
532
class TypedCol(Col):
533 2818 aaronmk
    def __init__(self, name, type_, default=None, nullable=True):
534
        assert default == None or isinstance(default, Code)
535
536 2674 aaronmk
        Col.__init__(self, name)
537
538
        self.type = type_
539 2818 aaronmk
        self.default = default
540
        self.nullable = nullable
541 2674 aaronmk
542 2818 aaronmk
    def to_str(self, db):
543
        str_ = Col.to_str(self, db)+' '+self.type
544
        if not self.nullable: str_ += ' NOT NULL'
545
        if self.default != None: str_ += ' DEFAULT '+self.default.to_str(db)
546
        return str_
547 2674 aaronmk
548
    def to_Col(self): return Col(self.name)
549 2822 aaronmk
550
null_sentinels = {'text': r'\N'}
551
552 2826 aaronmk
def ensure_not_null(db, value, ignore_unknown_type=True):
553 2822 aaronmk
    if is_table_col(value):
554
        typed_col = db.col_info(value)
555
        if typed_col.nullable:
556
            try: null = null_sentinels[typed_col.type]
557 2826 aaronmk
            except KeyError:
558
                if not ignore_unknown_type: raise
559
            else: value = EnsureNotNull(value, null)
560 2822 aaronmk
    return value