Project

General

Profile

1 2211 aaronmk
# SQL code generation
2
3 2276 aaronmk
import operator
4 2568 aaronmk
import re
5 2653 aaronmk
import UserDict
6 2276 aaronmk
7 2667 aaronmk
import dicts
8 2701 aaronmk
import iters
9
import lists
10 2360 aaronmk
import objects
11 2222 aaronmk
import strings
12 2227 aaronmk
import util
13 2211 aaronmk
14 2587 aaronmk
##### Names
15 2499 aaronmk
16 2608 aaronmk
identifier_max_len = 63 # works for both PostgreSQL and MySQL
17 2587 aaronmk
18
def add_suffix(str_, suffix):
19 2609 aaronmk
    '''Preserves version so that it won't be truncated off the string, leading
20
    to collisions.'''
21 2613 aaronmk
    # Preserve version
22
    before, sep, version = str_.rpartition('#')
23
    if sep != '': # found match
24
        str_ = before
25 2609 aaronmk
        suffix = sep+version+suffix
26 2613 aaronmk
27 2587 aaronmk
    return strings.add_suffix(str_, suffix, identifier_max_len)
28
29 2575 aaronmk
def is_safe_name(name):
30 2583 aaronmk
    '''A name is safe *and unambiguous* if it:
31
    * contains only *lowercase* word (\w) characters
32
    * doesn't start with a digit
33
    * contains "_", so that it's not a keyword
34
    '''
35
    return re.match(r'^(?=.*_)(?!\d)[^\WA-Z]+$', name)
36 2568 aaronmk
37 2499 aaronmk
def esc_name(name, quote='"'):
38
    return quote + name.replace(quote, quote+quote) + quote
39
        # doubling an embedded quote escapes it in both PostgreSQL and MySQL
40
41 2513 aaronmk
def clean_name(name): return name.replace('"', '').replace('`', '')
42
43 2659 aaronmk
##### General SQL code objects
44 2219 aaronmk
45 2349 aaronmk
class MockDb:
46 2503 aaronmk
    def esc_value(self, value): return strings.repr_no_u(value)
47 2349 aaronmk
48 2499 aaronmk
    def esc_name(self, name): return esc_name(name)
49 2349 aaronmk
mockDb = MockDb()
50
51 2514 aaronmk
class BasicObject(objects.BasicObject):
52
    def __init__(self, value): self.value = value
53
54
    def __str__(self): return clean_name(strings.repr_no_u(self))
55
56 2659 aaronmk
##### Unparameterized code objects
57
58 2514 aaronmk
class Code(BasicObject):
59 2658 aaronmk
    def to_str(self, db): raise NotImplementedError()
60 2349 aaronmk
61 2514 aaronmk
    def __repr__(self): return self.to_str(mockDb)
62 2211 aaronmk
63 2269 aaronmk
class CustomCode(Code):
64 2256 aaronmk
    def __init__(self, str_): self.str_ = str_
65
66
    def to_str(self, db): return self.str_
67
68 2659 aaronmk
def as_Code(value):
69
    if util.is_str(value): return CustomCode(value)
70
    else: return Literal(value)
71
72 2540 aaronmk
class Expr(Code):
73
    def __init__(self, expr): self.expr = expr
74
75
    def to_str(self, db): return '('+self.expr.to_str(db)+')'
76
77 2335 aaronmk
##### Literal values
78
79 2216 aaronmk
class Literal(Code):
80 2211 aaronmk
    def __init__(self, value): self.value = value
81 2213 aaronmk
82
    def to_str(self, db): return db.esc_value(self.value)
83 2211 aaronmk
84 2400 aaronmk
def as_Value(value):
85
    if isinstance(value, Code): return value
86
    else: return Literal(value)
87
88 2216 aaronmk
def is_null(value): return isinstance(value, Literal) and value.value == None
89
90 2711 aaronmk
##### Derived elements
91
92
src_self = object() # tells Col that it is its own source column
93
94
class Derived(Code):
95
    def __init__(self, srcs):
96 2712 aaronmk
        '''An element which was derived from some other element(s).
97 2711 aaronmk
        @param srcs See self.set_srcs()
98
        '''
99
        self.set_srcs(srcs)
100
101 2713 aaronmk
    def set_srcs(self, srcs, overwrite=True):
102 2711 aaronmk
        '''
103
        @param srcs (self_type...)|src_self The element(s) this is derived from
104
        '''
105 2713 aaronmk
        if not overwrite and self.srcs != (): return # already set
106
107 2711 aaronmk
        if srcs == src_self: srcs = (self,)
108
        srcs = tuple(srcs) # make Col hashable
109
        self.srcs = srcs
110
111
    def _compare_on(self):
112
        compare_on = self.__dict__.copy()
113
        del compare_on['srcs'] # ignore
114
        return compare_on
115
116
def cols_srcs(cols): return lists.uniqify(iters.flatten((v.srcs for v in cols)))
117
118 2335 aaronmk
##### Tables
119
120 2712 aaronmk
class Table(Derived):
121
    def __init__(self, name, schema=None, srcs=()):
122 2211 aaronmk
        '''
123
        @param schema str|None (for no schema)
124 2712 aaronmk
        @param srcs (Table...)|src_self See Derived.set_srcs()
125 2211 aaronmk
        '''
126 2712 aaronmk
        Derived.__init__(self, srcs)
127
128 2211 aaronmk
        self.name = name
129
        self.schema = schema
130
131 2348 aaronmk
    def to_str(self, db):
132
        str_ = ''
133
        if self.schema != None: str_ += db.esc_name(self.schema)+'.'
134
        str_ += db.esc_name(self.name)
135
        return str_
136 2336 aaronmk
137
    def to_Table(self): return self
138 2211 aaronmk
139 2219 aaronmk
def as_Table(table):
140 2270 aaronmk
    if table == None or isinstance(table, Code): return table
141 2219 aaronmk
    else: return Table(table)
142
143 2707 aaronmk
def suffixed_table(table, suffix): return Table(table.name+suffix, table.schema)
144
145 2336 aaronmk
class NamedTable(Table):
146
    def __init__(self, name, code, cols=None):
147
        Table.__init__(self, name)
148
149
        if not isinstance(code, Code): code = Table(code)
150
151
        self.code = code
152
        self.cols = cols
153
154
    def to_str(self, db):
155 2467 aaronmk
        str_ = self.code.to_str(db)+'\nAS '+Table.to_str(self, db)
156 2336 aaronmk
        if self.cols != None: str_ += ' ('+(', '.join(self.cols))+')'
157
        return str_
158
159
    def to_Table(self): return Table(self.name)
160
161 2335 aaronmk
##### Columns
162
163 2711 aaronmk
class Col(Derived):
164 2701 aaronmk
    def __init__(self, name, table=None, srcs=()):
165 2211 aaronmk
        '''
166
        @param table Table|None (for no table)
167 2711 aaronmk
        @param srcs (Col...)|src_self See Derived.set_srcs()
168 2211 aaronmk
        '''
169 2711 aaronmk
        Derived.__init__(self, srcs)
170
171 2241 aaronmk
        if util.is_str(table): table = Table(table)
172 2211 aaronmk
        assert table == None or isinstance(table, Table)
173
174
        self.name = name
175
        self.table = table
176
177
    def to_str(self, db):
178
        str_ = ''
179
        if self.table != None: str_ += self.table.to_str(db)+'.'
180 2348 aaronmk
        str_ += db.esc_name(self.name)
181 2211 aaronmk
        return str_
182 2314 aaronmk
183
    def to_Col(self): return self
184 2211 aaronmk
185 2393 aaronmk
def is_table_col(col): return col.table != None
186
187 2563 aaronmk
def as_Col(col, table=None, name=None):
188
    '''
189
    @param name If not None, any non-Col input will be renamed using NamedCol.
190
    '''
191
    if name != None:
192
        col = as_Value(col)
193
        if not isinstance(col, Col): col = NamedCol(name, col)
194 2333 aaronmk
195
    if isinstance(col, Code): return col
196 2260 aaronmk
    else: return Col(col, table)
197
198 2401 aaronmk
def to_name_only_col(col, check_table=None):
199
    col = as_Col(col)
200 2579 aaronmk
    if not isinstance(col, Col): return col
201 2401 aaronmk
202
    if check_table != None:
203
        table = col.table
204
        assert table == None or table == check_table
205
    return Col(col.name)
206
207 2323 aaronmk
class NamedCol(Col):
208 2229 aaronmk
    def __init__(self, name, code):
209 2310 aaronmk
        Col.__init__(self, name)
210
211 2229 aaronmk
        if not isinstance(code, Code): code = Literal(code)
212
213
        self.code = code
214
215
    def to_str(self, db):
216 2310 aaronmk
        return self.code.to_str(db)+' AS '+Col.to_str(self, db)
217 2314 aaronmk
218
    def to_Col(self): return Col(self.name)
219 2229 aaronmk
220 2462 aaronmk
def remove_col_rename(col):
221
    if isinstance(col, NamedCol): col = col.code
222
    return col
223
224 2703 aaronmk
def wrap(wrap_func, value):
225
    '''Wraps a value, propagating any column renaming to the returned value.'''
226
    if isinstance(value, NamedCol):
227
        return NamedCol(value.name, wrap_func(value.code))
228
    else: return wrap_func(value)
229
230 2667 aaronmk
class ColDict(dicts.DictProxy):
231 2564 aaronmk
    '''A dict that automatically makes inserted entries Col objects'''
232
233 2645 aaronmk
    def __init__(self, db, keys_table, dict_={}):
234 2667 aaronmk
        dicts.DictProxy.__init__(self, {})
235
236 2645 aaronmk
        keys_table = as_Table(keys_table)
237
238 2642 aaronmk
        self.db = db
239 2641 aaronmk
        self.table = keys_table
240 2653 aaronmk
        self.update(dict_) # after setting vars because __setitem__() needs them
241 2641 aaronmk
242 2667 aaronmk
    def copy(self): return ColDict(self.db, self.table, self.inner.copy())
243 2655 aaronmk
244 2667 aaronmk
    def __getitem__(self, key):
245
        return dicts.DictProxy.__getitem__(self, self._key(key))
246 2653 aaronmk
247 2564 aaronmk
    def __setitem__(self, key, value):
248 2642 aaronmk
        key = self._key(key)
249 2661 aaronmk
        if value == None: value = self.db.col_default(key)
250 2667 aaronmk
        dicts.DictProxy.__setitem__(self, key, as_Col(value, name=key.name))
251 2564 aaronmk
252 2641 aaronmk
    def _key(self, key): return as_Col(key, self.table)
253 2564 aaronmk
254 2524 aaronmk
##### Functions
255
256
class Function(Table): pass
257
258 2697 aaronmk
def TempFunction(name, autocommit):
259
    schema = None
260
    if not autocommit: schema = 'pg_temp'
261
    return Function(name, schema)
262
263 2691 aaronmk
class InternalFunction(CustomCode): pass
264
265 2524 aaronmk
class FunctionCall(Code):
266
    def __init__(self, function, *args):
267
        '''
268 2690 aaronmk
        @param args [Code|literal-value...] The function's arguments
269 2524 aaronmk
        '''
270
        if not isinstance(function, Code): function = Function(function)
271 2690 aaronmk
        args = map(remove_col_rename, map(as_Value, args))
272 2524 aaronmk
273
        self.function = function
274
        self.args = args
275
276
    def to_str(self, db):
277
        args_str = ', '.join((v.to_str(db) for v in self.args))
278
        return self.function.to_str(db)+'('+args_str+')'
279
280 2533 aaronmk
def wrap_in_func(function, value):
281
    '''Wraps a value inside a function call.
282
    Propagates any column renaming to the returned value.
283
    '''
284 2703 aaronmk
    return wrap(lambda v: FunctionCall(function, v), value)
285 2533 aaronmk
286 2561 aaronmk
def unwrap_func_call(func_call, check_name=None):
287
    '''Unwraps any function call to its first argument.
288
    Also removes any column renaming.
289
    '''
290
    func_call = remove_col_rename(func_call)
291
    if not isinstance(func_call, FunctionCall): return func_call
292
293
    if check_name != None:
294
        name = func_call.function.name
295
        assert name == None or name == check_name
296
    return func_call.args[0]
297
298 2335 aaronmk
##### Conditions
299 2259 aaronmk
300 2398 aaronmk
class ColValueCond(Code):
301
    def __init__(self, col, value):
302
        value = as_ValueCond(value)
303
304
        self.col = col
305
        self.value = value
306
307
    def to_str(self, db): return self.value.to_str(db, self.col)
308
309 2577 aaronmk
def combine_conds(conds, keyword=None):
310
    '''
311
    @param keyword The keyword to add before the conditions, if any
312
    '''
313
    str_ = ''
314
    if keyword != None:
315
        if conds == []: whitespace = ''
316
        elif len(conds) == 1: whitespace = ' '
317
        else: whitespace = '\n'
318
        str_ += keyword+whitespace
319
320
    str_ += '\nAND '.join(conds)
321
    return str_
322
323 2398 aaronmk
##### Condition column comparisons
324
325 2514 aaronmk
class ValueCond(BasicObject):
326 2213 aaronmk
    def __init__(self, value):
327 2225 aaronmk
        if not isinstance(value, Code): value = Literal(value)
328 2462 aaronmk
        value = remove_col_rename(value)
329 2213 aaronmk
330
        self.value = value
331 2214 aaronmk
332 2216 aaronmk
    def to_str(self, db, left_value):
333 2214 aaronmk
        '''
334 2216 aaronmk
        @param left_value The Code object that the condition is being applied on
335 2214 aaronmk
        '''
336
        raise NotImplemented()
337 2228 aaronmk
338 2514 aaronmk
    def __repr__(self): return self.to_str(mockDb, '<left_value>')
339 2211 aaronmk
340
class CompareCond(ValueCond):
341
    def __init__(self, value, operator='='):
342 2222 aaronmk
        '''
343
        @param operator By default, compares NULL values literally. Use '~=' or
344
            '~!=' to pass NULLs through.
345
        '''
346 2211 aaronmk
        ValueCond.__init__(self, value)
347
        self.operator = operator
348
349 2216 aaronmk
    def to_str(self, db, left_value):
350
        if not isinstance(left_value, Code): left_value = Col(left_value)
351 2462 aaronmk
        left_value = remove_col_rename(left_value)
352 2216 aaronmk
353 2222 aaronmk
        right_value = self.value
354
        left = left_value.to_str(db)
355
        right = right_value.to_str(db)
356
357
        # Parse operator
358 2216 aaronmk
        operator = self.operator
359 2222 aaronmk
        passthru_null_ref = [False]
360
        operator = strings.remove_prefix('~', operator, passthru_null_ref)
361
        neg_ref = [False]
362
        operator = strings.remove_prefix('!', operator, neg_ref)
363
        equals = operator.endswith('=')
364
        if equals and is_null(self.value): operator = 'IS'
365
366
        # Create str
367
        str_ = left+' '+operator+' '+right
368
        if equals and not passthru_null_ref[0] and isinstance(right_value, Col):
369 2578 aaronmk
            str_ = '('+str_+' OR ('+left+' IS NULL AND '+right+' IS NULL))'
370
        if neg_ref[0]: str_ = 'NOT '+str_
371 2222 aaronmk
        return str_
372 2216 aaronmk
373 2260 aaronmk
# Tells as_ValueCond() to assume a non-ValueCond is a literal value
374
assume_literal = object()
375
376
def as_ValueCond(value, default_table=assume_literal):
377
    if not isinstance(value, ValueCond):
378
        if default_table is not assume_literal:
379
            value = as_Col(value, default_table)
380
        return CompareCond(value)
381 2216 aaronmk
    else: return value
382 2219 aaronmk
383 2335 aaronmk
##### Joins
384
385 2352 aaronmk
join_same = object() # tells Join the left and right columns have the same name
386 2260 aaronmk
387 2353 aaronmk
# Tells Join the left and right columns have the same name and are never NULL
388
join_same_not_null = object()
389
390 2260 aaronmk
filter_out = object() # tells Join to filter out rows that match the join
391
392 2514 aaronmk
class Join(BasicObject):
393 2260 aaronmk
    def __init__(self, table, mapping, type_=None):
394
        '''
395
        @param mapping dict(right_table_col=left_table_col, ...)
396 2352 aaronmk
            * if left_table_col is join_same: left_table_col = right_table_col
397 2353 aaronmk
              * Note that right_table_col must be a string
398
            * if left_table_col is join_same_not_null:
399
              left_table_col = right_table_col and both have NOT NULL constraint
400
              * Note that right_table_col must be a string
401 2260 aaronmk
        @param type_ None (for plain join)|str (e.g. 'LEFT')|filter_out
402
            * filter_out: equivalent to 'LEFT' with the query filtered by
403
              `table_pkey IS NULL` (indicating no match)
404
        '''
405
        if util.is_str(table): table = Table(table)
406
        assert type_ == None or util.is_str(type_) or type_ is filter_out
407
408
        self.table = table
409
        self.mapping = mapping
410
        self.type_ = type_
411
412
    def to_str(self, db, left_table):
413 2353 aaronmk
        # Switch order (left_table is on the right in the comparison)
414
        right_table = left_table
415
        left_table = self.table # note left_table is reassigned
416
417 2260 aaronmk
        def join(entry):
418
            '''Parses non-USING joins'''
419
            right_table_col, left_table_col = entry
420
421 2353 aaronmk
            # Switch order (right_table_col is on the left in the comparison)
422
            left = right_table_col
423
            right = left_table_col
424
425 2260 aaronmk
            # Parse special values
426 2353 aaronmk
            if right is join_same: right = left
427
            elif right is join_same_not_null:
428
                right = CompareCond(as_Col(left, right_table), '~=')
429 2260 aaronmk
430 2353 aaronmk
            right = as_ValueCond(right, right_table)
431 2578 aaronmk
            return right.to_str(db, as_Col(left, left_table))
432 2260 aaronmk
433 2265 aaronmk
        # Create join condition
434
        type_ = self.type_
435 2276 aaronmk
        joins = self.mapping
436 2265 aaronmk
        if type_ is not filter_out and reduce(operator.and_,
437 2460 aaronmk
            (v is join_same_not_null for v in joins.itervalues())):
438 2260 aaronmk
            # all cols w/ USING, so can use simpler USING syntax
439 2298 aaronmk
            cols = (as_Col(v).to_str(db) for v in joins.iterkeys())
440
            join_cond = 'USING ('+(', '.join(cols))+')'
441 2576 aaronmk
        else:
442
            if len(joins) == 1: whitespace = ' '
443
            else: whitespace = '\n'
444 2577 aaronmk
            join_cond = combine_conds(map(join, joins.iteritems()), 'ON')
445 2260 aaronmk
446
        # Create join
447
        if type_ is filter_out: type_ = 'LEFT'
448 2266 aaronmk
        str_ = ''
449
        if type_ != None: str_ += type_+' '
450 2736 aaronmk
        str_ += 'JOIN '+left_table.to_str(db)
451
        if type_ != 'CROSS': str_ += ' '+join_cond
452 2266 aaronmk
        return str_
453 2349 aaronmk
454 2514 aaronmk
    def __repr__(self): return self.to_str(mockDb, '<left_table>')
455 2424 aaronmk
456
##### Value exprs
457
458 2737 aaronmk
default = CustomCode('DEFAULT')
459
460 2424 aaronmk
row_count = CustomCode('count(*)')
461 2674 aaronmk
462 2692 aaronmk
def EnsureNotNull(value, null=r'\N'):
463 2694 aaronmk
    return FunctionCall(InternalFunction('coalesce'), as_Col(value), null)
464 2692 aaronmk
465 2737 aaronmk
##### Table exprs
466
467
class Values(Code):
468
    def __init__(self, values):
469 2739 aaronmk
        '''
470
        @param values [...]|[[...], ...] Can be one or multiple rows.
471
        '''
472
        rows = values
473
        if len(values) >= 1 and not lists.is_seq(values[0]): # only one row
474
            rows = [values]
475
        for i, row in enumerate(rows):
476
            rows[i] = map(remove_col_rename, map(as_Value, row))
477 2737 aaronmk
478 2739 aaronmk
        self.rows = rows
479 2737 aaronmk
480
    def to_str(self, db):
481 2739 aaronmk
        def row_str(row):
482
            return '('+(', '.join((v.to_str(db) for v in row)))+')'
483
        return 'VALUES '+(', '.join(map(row_str, self.rows)))
484 2737 aaronmk
485 2674 aaronmk
##### Database structure
486
487
class TypedCol(Col):
488
    def __init__(self, name, type_):
489
        Col.__init__(self, name)
490
491
        self.type = type_
492
493
    def to_str(self, db): return Col.to_str(self, db)+' '+self.type
494
495
    def to_Col(self): return Col(self.name)