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 2832 aaronmk
def is_underlying_table(table): return table.to_Table() is table
146
147 2776 aaronmk
def as_Table(table, schema=None):
148 2270 aaronmk
    if table == None or isinstance(table, Code): return table
149 2776 aaronmk
    else: return Table(table, schema)
150 2219 aaronmk
151 2707 aaronmk
def suffixed_table(table, suffix): return Table(table.name+suffix, table.schema)
152
153 2336 aaronmk
class NamedTable(Table):
154
    def __init__(self, name, code, cols=None):
155
        Table.__init__(self, name)
156
157
        if not isinstance(code, Code): code = Table(code)
158 2741 aaronmk
        if not isinstance(code, (Table, FunctionCall, Expr)): code = Expr(code)
159 2742 aaronmk
        if cols != None: cols = map(to_name_only_col, cols)
160 2336 aaronmk
161
        self.code = code
162
        self.cols = cols
163
164
    def to_str(self, db):
165 2467 aaronmk
        str_ = self.code.to_str(db)+'\nAS '+Table.to_str(self, db)
166 2742 aaronmk
        if self.cols != None:
167
            str_ += ' ('+(', '.join((c.to_str(db) for c in self.cols)))+')'
168 2336 aaronmk
        return str_
169
170
    def to_Table(self): return Table(self.name)
171
172 2753 aaronmk
def remove_table_rename(table):
173
    if isinstance(table, NamedTable): table = table.code
174
    return table
175
176 2335 aaronmk
##### Columns
177
178 2711 aaronmk
class Col(Derived):
179 2701 aaronmk
    def __init__(self, name, table=None, srcs=()):
180 2211 aaronmk
        '''
181
        @param table Table|None (for no table)
182 2711 aaronmk
        @param srcs (Col...)|src_self See Derived.set_srcs()
183 2211 aaronmk
        '''
184 2711 aaronmk
        Derived.__init__(self, srcs)
185
186 2241 aaronmk
        if util.is_str(table): table = Table(table)
187 2211 aaronmk
        assert table == None or isinstance(table, Table)
188
189
        self.name = name
190
        self.table = table
191
192
    def to_str(self, db):
193
        str_ = ''
194 2827 aaronmk
        if self.table != None: str_ += self.table.to_Table().to_str(db)+'.'
195 2348 aaronmk
        str_ += db.esc_name(self.name)
196 2211 aaronmk
        return str_
197 2314 aaronmk
198
    def to_Col(self): return self
199 2211 aaronmk
200 2767 aaronmk
def is_table_col(col): return isinstance(col, Col) and col.table != None
201 2393 aaronmk
202 2563 aaronmk
def as_Col(col, table=None, name=None):
203
    '''
204
    @param name If not None, any non-Col input will be renamed using NamedCol.
205
    '''
206
    if name != None:
207
        col = as_Value(col)
208
        if not isinstance(col, Col): col = NamedCol(name, col)
209 2333 aaronmk
210
    if isinstance(col, Code): return col
211 2260 aaronmk
    else: return Col(col, table)
212
213 2750 aaronmk
def with_default_table(col, table, overwrite=False):
214 2747 aaronmk
    col = as_Col(col)
215 2750 aaronmk
    if not isinstance(col, NamedCol) and (overwrite or col.table == None):
216 2748 aaronmk
        col = copy.copy(col) # don't modify input!
217
        col.table = table
218 2747 aaronmk
    return col
219
220 2744 aaronmk
def set_cols_table(table, cols):
221
    table = as_Table(table)
222
223
    for i, col in enumerate(cols):
224
        col = cols[i] = as_Col(col)
225
        col.table = table
226
227 2401 aaronmk
def to_name_only_col(col, check_table=None):
228
    col = as_Col(col)
229 2579 aaronmk
    if not isinstance(col, Col): return col
230 2401 aaronmk
231
    if check_table != None:
232
        table = col.table
233
        assert table == None or table == check_table
234
    return Col(col.name)
235
236 2323 aaronmk
class NamedCol(Col):
237 2229 aaronmk
    def __init__(self, name, code):
238 2310 aaronmk
        Col.__init__(self, name)
239
240 2229 aaronmk
        if not isinstance(code, Code): code = Literal(code)
241
242
        self.code = code
243
244
    def to_str(self, db):
245 2310 aaronmk
        return self.code.to_str(db)+' AS '+Col.to_str(self, db)
246 2314 aaronmk
247
    def to_Col(self): return Col(self.name)
248 2229 aaronmk
249 2462 aaronmk
def remove_col_rename(col):
250
    if isinstance(col, NamedCol): col = col.code
251
    return col
252
253 2833 aaronmk
class NoUnderlyingTableException(Exception): pass
254
255 2830 aaronmk
def underlying_col(col):
256
    col = remove_col_rename(col)
257 2833 aaronmk
    table = remove_table_rename(col.table)
258
    if not is_underlying_table(table): raise NoUnderlyingTableException
259
    return Col(col.name, table, col.srcs)
260 2830 aaronmk
261 2703 aaronmk
def wrap(wrap_func, value):
262
    '''Wraps a value, propagating any column renaming to the returned value.'''
263
    if isinstance(value, NamedCol):
264
        return NamedCol(value.name, wrap_func(value.code))
265
    else: return wrap_func(value)
266
267 2667 aaronmk
class ColDict(dicts.DictProxy):
268 2564 aaronmk
    '''A dict that automatically makes inserted entries Col objects'''
269
270 2645 aaronmk
    def __init__(self, db, keys_table, dict_={}):
271 2667 aaronmk
        dicts.DictProxy.__init__(self, {})
272
273 2645 aaronmk
        keys_table = as_Table(keys_table)
274
275 2642 aaronmk
        self.db = db
276 2641 aaronmk
        self.table = keys_table
277 2653 aaronmk
        self.update(dict_) # after setting vars because __setitem__() needs them
278 2641 aaronmk
279 2667 aaronmk
    def copy(self): return ColDict(self.db, self.table, self.inner.copy())
280 2655 aaronmk
281 2667 aaronmk
    def __getitem__(self, key):
282
        return dicts.DictProxy.__getitem__(self, self._key(key))
283 2653 aaronmk
284 2564 aaronmk
    def __setitem__(self, key, value):
285 2642 aaronmk
        key = self._key(key)
286 2819 aaronmk
        if value == None: value = self.db.col_info(key).default
287 2667 aaronmk
        dicts.DictProxy.__setitem__(self, key, as_Col(value, name=key.name))
288 2564 aaronmk
289 2641 aaronmk
    def _key(self, key): return as_Col(key, self.table)
290 2564 aaronmk
291 2524 aaronmk
##### Functions
292
293
class Function(Table): pass
294
295 2697 aaronmk
def TempFunction(name, autocommit):
296
    schema = None
297
    if not autocommit: schema = 'pg_temp'
298
    return Function(name, schema)
299
300 2691 aaronmk
class InternalFunction(CustomCode): pass
301
302 2524 aaronmk
class FunctionCall(Code):
303
    def __init__(self, function, *args):
304
        '''
305 2690 aaronmk
        @param args [Code|literal-value...] The function's arguments
306 2524 aaronmk
        '''
307
        if not isinstance(function, Code): function = Function(function)
308 2690 aaronmk
        args = map(remove_col_rename, map(as_Value, args))
309 2524 aaronmk
310
        self.function = function
311
        self.args = args
312
313
    def to_str(self, db):
314
        args_str = ', '.join((v.to_str(db) for v in self.args))
315
        return self.function.to_str(db)+'('+args_str+')'
316
317 2533 aaronmk
def wrap_in_func(function, value):
318
    '''Wraps a value inside a function call.
319
    Propagates any column renaming to the returned value.
320
    '''
321 2703 aaronmk
    return wrap(lambda v: FunctionCall(function, v), value)
322 2533 aaronmk
323 2561 aaronmk
def unwrap_func_call(func_call, check_name=None):
324
    '''Unwraps any function call to its first argument.
325
    Also removes any column renaming.
326
    '''
327
    func_call = remove_col_rename(func_call)
328
    if not isinstance(func_call, FunctionCall): return func_call
329
330
    if check_name != None:
331
        name = func_call.function.name
332
        assert name == None or name == check_name
333
    return func_call.args[0]
334
335 2335 aaronmk
##### Conditions
336 2259 aaronmk
337 2398 aaronmk
class ColValueCond(Code):
338
    def __init__(self, col, value):
339
        value = as_ValueCond(value)
340
341
        self.col = col
342
        self.value = value
343
344
    def to_str(self, db): return self.value.to_str(db, self.col)
345
346 2577 aaronmk
def combine_conds(conds, keyword=None):
347
    '''
348
    @param keyword The keyword to add before the conditions, if any
349
    '''
350
    str_ = ''
351
    if keyword != None:
352
        if conds == []: whitespace = ''
353
        elif len(conds) == 1: whitespace = ' '
354
        else: whitespace = '\n'
355
        str_ += keyword+whitespace
356
357
    str_ += '\nAND '.join(conds)
358
    return str_
359
360 2398 aaronmk
##### Condition column comparisons
361
362 2514 aaronmk
class ValueCond(BasicObject):
363 2213 aaronmk
    def __init__(self, value):
364 2225 aaronmk
        if not isinstance(value, Code): value = Literal(value)
365 2462 aaronmk
        value = remove_col_rename(value)
366 2213 aaronmk
367
        self.value = value
368 2214 aaronmk
369 2216 aaronmk
    def to_str(self, db, left_value):
370 2214 aaronmk
        '''
371 2216 aaronmk
        @param left_value The Code object that the condition is being applied on
372 2214 aaronmk
        '''
373
        raise NotImplemented()
374 2228 aaronmk
375 2514 aaronmk
    def __repr__(self): return self.to_str(mockDb, '<left_value>')
376 2211 aaronmk
377
class CompareCond(ValueCond):
378
    def __init__(self, value, operator='='):
379 2222 aaronmk
        '''
380
        @param operator By default, compares NULL values literally. Use '~=' or
381
            '~!=' to pass NULLs through.
382
        '''
383 2211 aaronmk
        ValueCond.__init__(self, value)
384
        self.operator = operator
385
386 2216 aaronmk
    def to_str(self, db, left_value):
387
        if not isinstance(left_value, Code): left_value = Col(left_value)
388 2462 aaronmk
        left_value = remove_col_rename(left_value)
389 2216 aaronmk
390 2222 aaronmk
        right_value = self.value
391
392
        # Parse operator
393 2216 aaronmk
        operator = self.operator
394 2222 aaronmk
        passthru_null_ref = [False]
395
        operator = strings.remove_prefix('~', operator, passthru_null_ref)
396
        neg_ref = [False]
397
        operator = strings.remove_prefix('!', operator, neg_ref)
398
        equals = operator.endswith('=')
399 2825 aaronmk
        if equals and is_null(right_value): operator = 'IS'
400 2222 aaronmk
401 2825 aaronmk
        # Handle nullable columns
402
        check_null = False
403
        if equals and not passthru_null_ref[0] and isinstance(right_value, Col):
404
            check_null = True
405
406
        left = left_value.to_str(db)
407
        right = right_value.to_str(db)
408
409 2222 aaronmk
        # Create str
410
        str_ = left+' '+operator+' '+right
411 2825 aaronmk
        if check_null:
412 2578 aaronmk
            str_ = '('+str_+' OR ('+left+' IS NULL AND '+right+' IS NULL))'
413
        if neg_ref[0]: str_ = 'NOT '+str_
414 2222 aaronmk
        return str_
415 2216 aaronmk
416 2260 aaronmk
# Tells as_ValueCond() to assume a non-ValueCond is a literal value
417
assume_literal = object()
418
419
def as_ValueCond(value, default_table=assume_literal):
420
    if not isinstance(value, ValueCond):
421
        if default_table is not assume_literal:
422 2748 aaronmk
            value = with_default_table(value, default_table)
423 2260 aaronmk
        return CompareCond(value)
424 2216 aaronmk
    else: return value
425 2219 aaronmk
426 2335 aaronmk
##### Joins
427
428 2352 aaronmk
join_same = object() # tells Join the left and right columns have the same name
429 2260 aaronmk
430 2353 aaronmk
# Tells Join the left and right columns have the same name and are never NULL
431
join_same_not_null = object()
432
433 2260 aaronmk
filter_out = object() # tells Join to filter out rows that match the join
434
435 2514 aaronmk
class Join(BasicObject):
436 2746 aaronmk
    def __init__(self, table, mapping={}, type_=None):
437 2260 aaronmk
        '''
438
        @param mapping dict(right_table_col=left_table_col, ...)
439 2352 aaronmk
            * if left_table_col is join_same: left_table_col = right_table_col
440 2353 aaronmk
              * Note that right_table_col must be a string
441
            * if left_table_col is join_same_not_null:
442
              left_table_col = right_table_col and both have NOT NULL constraint
443
              * Note that right_table_col must be a string
444 2260 aaronmk
        @param type_ None (for plain join)|str (e.g. 'LEFT')|filter_out
445
            * filter_out: equivalent to 'LEFT' with the query filtered by
446
              `table_pkey IS NULL` (indicating no match)
447
        '''
448
        if util.is_str(table): table = Table(table)
449
        assert type_ == None or util.is_str(type_) or type_ is filter_out
450
451
        self.table = table
452
        self.mapping = mapping
453
        self.type_ = type_
454
455 2749 aaronmk
    def to_str(self, db, left_table_):
456 2260 aaronmk
        def join(entry):
457
            '''Parses non-USING joins'''
458
            right_table_col, left_table_col = entry
459
460 2353 aaronmk
            # Switch order (right_table_col is on the left in the comparison)
461
            left = right_table_col
462
            right = left_table_col
463 2749 aaronmk
            left_table = self.table
464
            right_table = left_table_
465 2353 aaronmk
466 2747 aaronmk
            # Parse left side
467 2748 aaronmk
            left = with_default_table(left, left_table)
468 2747 aaronmk
469 2260 aaronmk
            # Parse special values
470 2747 aaronmk
            left_on_right = Col(left.name, right_table)
471
            if right is join_same: right = left_on_right
472 2353 aaronmk
            elif right is join_same_not_null:
473 2747 aaronmk
                right = CompareCond(left_on_right, '~=')
474 2260 aaronmk
475 2747 aaronmk
            # Parse right side
476 2353 aaronmk
            right = as_ValueCond(right, right_table)
477 2747 aaronmk
478
            return right.to_str(db, left)
479 2260 aaronmk
480 2265 aaronmk
        # Create join condition
481
        type_ = self.type_
482 2276 aaronmk
        joins = self.mapping
483 2746 aaronmk
        if joins == {}: join_cond = None
484
        elif type_ is not filter_out and reduce(operator.and_,
485 2460 aaronmk
            (v is join_same_not_null for v in joins.itervalues())):
486 2260 aaronmk
            # all cols w/ USING, so can use simpler USING syntax
487 2747 aaronmk
            cols = map(to_name_only_col, joins.iterkeys())
488
            join_cond = 'USING ('+(', '.join((c.to_str(db) for c in cols)))+')'
489 2757 aaronmk
        else: join_cond = combine_conds(map(join, joins.iteritems()), 'ON')
490 2260 aaronmk
491 2757 aaronmk
        if isinstance(self.table, NamedTable): whitespace = '\n'
492
        else: whitespace = ' '
493
494 2260 aaronmk
        # Create join
495
        if type_ is filter_out: type_ = 'LEFT'
496 2266 aaronmk
        str_ = ''
497
        if type_ != None: str_ += type_+' '
498 2757 aaronmk
        str_ += 'JOIN'+whitespace+self.table.to_str(db)
499
        if join_cond != None: str_ += whitespace+join_cond
500 2266 aaronmk
        return str_
501 2349 aaronmk
502 2514 aaronmk
    def __repr__(self): return self.to_str(mockDb, '<left_table>')
503 2424 aaronmk
504
##### Value exprs
505
506 2737 aaronmk
default = CustomCode('DEFAULT')
507
508 2424 aaronmk
row_count = CustomCode('count(*)')
509 2674 aaronmk
510 2692 aaronmk
def EnsureNotNull(value, null=r'\N'):
511 2694 aaronmk
    return FunctionCall(InternalFunction('coalesce'), as_Col(value), null)
512 2692 aaronmk
513 2737 aaronmk
##### Table exprs
514
515
class Values(Code):
516
    def __init__(self, values):
517 2739 aaronmk
        '''
518
        @param values [...]|[[...], ...] Can be one or multiple rows.
519
        '''
520
        rows = values
521
        if len(values) >= 1 and not lists.is_seq(values[0]): # only one row
522
            rows = [values]
523
        for i, row in enumerate(rows):
524
            rows[i] = map(remove_col_rename, map(as_Value, row))
525 2737 aaronmk
526 2739 aaronmk
        self.rows = rows
527 2737 aaronmk
528
    def to_str(self, db):
529 2739 aaronmk
        def row_str(row):
530
            return '('+(', '.join((v.to_str(db) for v in row)))+')'
531
        return 'VALUES '+(', '.join(map(row_str, self.rows)))
532 2737 aaronmk
533 2740 aaronmk
def NamedValues(name, cols, values):
534 2745 aaronmk
    '''
535
    @post `cols` will be changed to Col objects with the table set to `name`.
536
    '''
537
    set_cols_table(name, cols)
538 2740 aaronmk
    return NamedTable(name, Values(values), cols)
539
540 2674 aaronmk
##### Database structure
541
542
class TypedCol(Col):
543 2818 aaronmk
    def __init__(self, name, type_, default=None, nullable=True):
544
        assert default == None or isinstance(default, Code)
545
546 2674 aaronmk
        Col.__init__(self, name)
547
548
        self.type = type_
549 2818 aaronmk
        self.default = default
550
        self.nullable = nullable
551 2674 aaronmk
552 2818 aaronmk
    def to_str(self, db):
553
        str_ = Col.to_str(self, db)+' '+self.type
554
        if not self.nullable: str_ += ' NOT NULL'
555
        if self.default != None: str_ += ' DEFAULT '+self.default.to_str(db)
556
        return str_
557 2674 aaronmk
558
    def to_Col(self): return Col(self.name)
559 2822 aaronmk
560
null_sentinels = {'text': r'\N'}
561
562 2826 aaronmk
def ensure_not_null(db, value, ignore_unknown_type=True):
563 2822 aaronmk
    if is_table_col(value):
564 2831 aaronmk
        typed_col = db.col_info(underlying_col(value))
565 2822 aaronmk
        if typed_col.nullable:
566
            try: null = null_sentinels[typed_col.type]
567 2826 aaronmk
            except KeyError:
568
                if not ignore_unknown_type: raise
569
            else: value = EnsureNotNull(value, null)
570 2822 aaronmk
    return value