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 2335 aaronmk
##### Tables
91
92 2211 aaronmk
class Table(Code):
93
    def __init__(self, name, schema=None):
94
        '''
95
        @param schema str|None (for no schema)
96
        '''
97
        self.name = name
98
        self.schema = schema
99
100 2348 aaronmk
    def to_str(self, db):
101
        str_ = ''
102
        if self.schema != None: str_ += db.esc_name(self.schema)+'.'
103
        str_ += db.esc_name(self.name)
104
        return str_
105 2336 aaronmk
106
    def to_Table(self): return self
107 2211 aaronmk
108 2219 aaronmk
def as_Table(table):
109 2270 aaronmk
    if table == None or isinstance(table, Code): return table
110 2219 aaronmk
    else: return Table(table)
111
112 2707 aaronmk
def suffixed_table(table, suffix): return Table(table.name+suffix, table.schema)
113
114 2336 aaronmk
class NamedTable(Table):
115
    def __init__(self, name, code, cols=None):
116
        Table.__init__(self, name)
117
118
        if not isinstance(code, Code): code = Table(code)
119
120
        self.code = code
121
        self.cols = cols
122
123
    def to_str(self, db):
124 2467 aaronmk
        str_ = self.code.to_str(db)+'\nAS '+Table.to_str(self, db)
125 2336 aaronmk
        if self.cols != None: str_ += ' ('+(', '.join(self.cols))+')'
126
        return str_
127
128
    def to_Table(self): return Table(self.name)
129
130 2335 aaronmk
##### Columns
131
132 2701 aaronmk
src_self = object() # tells Col that it is its own source column
133
134 2211 aaronmk
class Col(Code):
135 2701 aaronmk
    def __init__(self, name, table=None, srcs=()):
136 2211 aaronmk
        '''
137
        @param table Table|None (for no table)
138 2701 aaronmk
        @param srcs See self.set_srcs()
139 2211 aaronmk
        '''
140 2241 aaronmk
        if util.is_str(table): table = Table(table)
141 2211 aaronmk
        assert table == None or isinstance(table, Table)
142
143
        self.name = name
144
        self.table = table
145 2701 aaronmk
        self.set_srcs(srcs)
146 2211 aaronmk
147
    def to_str(self, db):
148
        str_ = ''
149
        if self.table != None: str_ += self.table.to_str(db)+'.'
150 2348 aaronmk
        str_ += db.esc_name(self.name)
151 2211 aaronmk
        return str_
152 2314 aaronmk
153
    def to_Col(self): return self
154 2701 aaronmk
155
    def set_srcs(self, srcs):
156
        '''
157
        @param srcs (Col...)|src_self The column(s) this column is derived from
158
        '''
159
        if srcs == src_self: srcs = (self,)
160
        srcs = tuple(srcs) # make Col hashable
161
        self.srcs = srcs
162
163
    def _compare_on(self):
164
        compare_on = self.__dict__.copy()
165
        del compare_on['srcs'] # ignore
166
        return compare_on
167 2211 aaronmk
168 2393 aaronmk
def is_table_col(col): return col.table != None
169
170 2701 aaronmk
def cols_srcs(cols): return lists.uniqify(iters.flatten((v.srcs for v in cols)))
171
172 2563 aaronmk
def as_Col(col, table=None, name=None):
173
    '''
174
    @param name If not None, any non-Col input will be renamed using NamedCol.
175
    '''
176
    if name != None:
177
        col = as_Value(col)
178
        if not isinstance(col, Col): col = NamedCol(name, col)
179 2333 aaronmk
180
    if isinstance(col, Code): return col
181 2260 aaronmk
    else: return Col(col, table)
182
183 2401 aaronmk
def to_name_only_col(col, check_table=None):
184
    col = as_Col(col)
185 2579 aaronmk
    if not isinstance(col, Col): return col
186 2401 aaronmk
187
    if check_table != None:
188
        table = col.table
189
        assert table == None or table == check_table
190
    return Col(col.name)
191
192 2323 aaronmk
class NamedCol(Col):
193 2229 aaronmk
    def __init__(self, name, code):
194 2310 aaronmk
        Col.__init__(self, name)
195
196 2229 aaronmk
        if not isinstance(code, Code): code = Literal(code)
197
198
        self.code = code
199
200
    def to_str(self, db):
201 2310 aaronmk
        return self.code.to_str(db)+' AS '+Col.to_str(self, db)
202 2314 aaronmk
203
    def to_Col(self): return Col(self.name)
204 2229 aaronmk
205 2462 aaronmk
def remove_col_rename(col):
206
    if isinstance(col, NamedCol): col = col.code
207
    return col
208
209 2703 aaronmk
def wrap(wrap_func, value):
210
    '''Wraps a value, propagating any column renaming to the returned value.'''
211
    if isinstance(value, NamedCol):
212
        return NamedCol(value.name, wrap_func(value.code))
213
    else: return wrap_func(value)
214
215 2667 aaronmk
class ColDict(dicts.DictProxy):
216 2564 aaronmk
    '''A dict that automatically makes inserted entries Col objects'''
217
218 2645 aaronmk
    def __init__(self, db, keys_table, dict_={}):
219 2667 aaronmk
        dicts.DictProxy.__init__(self, {})
220
221 2645 aaronmk
        keys_table = as_Table(keys_table)
222
223 2642 aaronmk
        self.db = db
224 2641 aaronmk
        self.table = keys_table
225 2653 aaronmk
        self.update(dict_) # after setting vars because __setitem__() needs them
226 2641 aaronmk
227 2667 aaronmk
    def copy(self): return ColDict(self.db, self.table, self.inner.copy())
228 2655 aaronmk
229 2667 aaronmk
    def __getitem__(self, key):
230
        return dicts.DictProxy.__getitem__(self, self._key(key))
231 2653 aaronmk
232 2564 aaronmk
    def __setitem__(self, key, value):
233 2642 aaronmk
        key = self._key(key)
234 2661 aaronmk
        if value == None: value = self.db.col_default(key)
235 2667 aaronmk
        dicts.DictProxy.__setitem__(self, key, as_Col(value, name=key.name))
236 2564 aaronmk
237 2641 aaronmk
    def _key(self, key): return as_Col(key, self.table)
238 2564 aaronmk
239 2524 aaronmk
##### Functions
240
241
class Function(Table): pass
242
243 2697 aaronmk
def TempFunction(name, autocommit):
244
    schema = None
245
    if not autocommit: schema = 'pg_temp'
246
    return Function(name, schema)
247
248 2691 aaronmk
class InternalFunction(CustomCode): pass
249
250 2524 aaronmk
class FunctionCall(Code):
251
    def __init__(self, function, *args):
252
        '''
253 2690 aaronmk
        @param args [Code|literal-value...] The function's arguments
254 2524 aaronmk
        '''
255
        if not isinstance(function, Code): function = Function(function)
256 2690 aaronmk
        args = map(remove_col_rename, map(as_Value, args))
257 2524 aaronmk
258
        self.function = function
259
        self.args = args
260
261
    def to_str(self, db):
262
        args_str = ', '.join((v.to_str(db) for v in self.args))
263
        return self.function.to_str(db)+'('+args_str+')'
264
265 2533 aaronmk
def wrap_in_func(function, value):
266
    '''Wraps a value inside a function call.
267
    Propagates any column renaming to the returned value.
268
    '''
269 2703 aaronmk
    return wrap(lambda v: FunctionCall(function, v), value)
270 2533 aaronmk
271 2561 aaronmk
def unwrap_func_call(func_call, check_name=None):
272
    '''Unwraps any function call to its first argument.
273
    Also removes any column renaming.
274
    '''
275
    func_call = remove_col_rename(func_call)
276
    if not isinstance(func_call, FunctionCall): return func_call
277
278
    if check_name != None:
279
        name = func_call.function.name
280
        assert name == None or name == check_name
281
    return func_call.args[0]
282
283 2335 aaronmk
##### Conditions
284 2259 aaronmk
285 2398 aaronmk
class ColValueCond(Code):
286
    def __init__(self, col, value):
287
        value = as_ValueCond(value)
288
289
        self.col = col
290
        self.value = value
291
292
    def to_str(self, db): return self.value.to_str(db, self.col)
293
294 2577 aaronmk
def combine_conds(conds, keyword=None):
295
    '''
296
    @param keyword The keyword to add before the conditions, if any
297
    '''
298
    str_ = ''
299
    if keyword != None:
300
        if conds == []: whitespace = ''
301
        elif len(conds) == 1: whitespace = ' '
302
        else: whitespace = '\n'
303
        str_ += keyword+whitespace
304
305
    str_ += '\nAND '.join(conds)
306
    return str_
307
308 2398 aaronmk
##### Condition column comparisons
309
310 2514 aaronmk
class ValueCond(BasicObject):
311 2213 aaronmk
    def __init__(self, value):
312 2225 aaronmk
        if not isinstance(value, Code): value = Literal(value)
313 2462 aaronmk
        value = remove_col_rename(value)
314 2213 aaronmk
315
        self.value = value
316 2214 aaronmk
317 2216 aaronmk
    def to_str(self, db, left_value):
318 2214 aaronmk
        '''
319 2216 aaronmk
        @param left_value The Code object that the condition is being applied on
320 2214 aaronmk
        '''
321
        raise NotImplemented()
322 2228 aaronmk
323 2514 aaronmk
    def __repr__(self): return self.to_str(mockDb, '<left_value>')
324 2211 aaronmk
325
class CompareCond(ValueCond):
326
    def __init__(self, value, operator='='):
327 2222 aaronmk
        '''
328
        @param operator By default, compares NULL values literally. Use '~=' or
329
            '~!=' to pass NULLs through.
330
        '''
331 2211 aaronmk
        ValueCond.__init__(self, value)
332
        self.operator = operator
333
334 2216 aaronmk
    def to_str(self, db, left_value):
335
        if not isinstance(left_value, Code): left_value = Col(left_value)
336 2462 aaronmk
        left_value = remove_col_rename(left_value)
337 2216 aaronmk
338 2222 aaronmk
        right_value = self.value
339
        left = left_value.to_str(db)
340
        right = right_value.to_str(db)
341
342
        # Parse operator
343 2216 aaronmk
        operator = self.operator
344 2222 aaronmk
        passthru_null_ref = [False]
345
        operator = strings.remove_prefix('~', operator, passthru_null_ref)
346
        neg_ref = [False]
347
        operator = strings.remove_prefix('!', operator, neg_ref)
348
        equals = operator.endswith('=')
349
        if equals and is_null(self.value): operator = 'IS'
350
351
        # Create str
352
        str_ = left+' '+operator+' '+right
353
        if equals and not passthru_null_ref[0] and isinstance(right_value, Col):
354 2578 aaronmk
            str_ = '('+str_+' OR ('+left+' IS NULL AND '+right+' IS NULL))'
355
        if neg_ref[0]: str_ = 'NOT '+str_
356 2222 aaronmk
        return str_
357 2216 aaronmk
358 2260 aaronmk
# Tells as_ValueCond() to assume a non-ValueCond is a literal value
359
assume_literal = object()
360
361
def as_ValueCond(value, default_table=assume_literal):
362
    if not isinstance(value, ValueCond):
363
        if default_table is not assume_literal:
364
            value = as_Col(value, default_table)
365
        return CompareCond(value)
366 2216 aaronmk
    else: return value
367 2219 aaronmk
368 2335 aaronmk
##### Joins
369
370 2352 aaronmk
join_same = object() # tells Join the left and right columns have the same name
371 2260 aaronmk
372 2353 aaronmk
# Tells Join the left and right columns have the same name and are never NULL
373
join_same_not_null = object()
374
375 2260 aaronmk
filter_out = object() # tells Join to filter out rows that match the join
376
377 2514 aaronmk
class Join(BasicObject):
378 2260 aaronmk
    def __init__(self, table, mapping, type_=None):
379
        '''
380
        @param mapping dict(right_table_col=left_table_col, ...)
381 2352 aaronmk
            * if left_table_col is join_same: left_table_col = right_table_col
382 2353 aaronmk
              * Note that right_table_col must be a string
383
            * if left_table_col is join_same_not_null:
384
              left_table_col = right_table_col and both have NOT NULL constraint
385
              * Note that right_table_col must be a string
386 2260 aaronmk
        @param type_ None (for plain join)|str (e.g. 'LEFT')|filter_out
387
            * filter_out: equivalent to 'LEFT' with the query filtered by
388
              `table_pkey IS NULL` (indicating no match)
389
        '''
390
        if util.is_str(table): table = Table(table)
391
        assert type_ == None or util.is_str(type_) or type_ is filter_out
392
393
        self.table = table
394
        self.mapping = mapping
395
        self.type_ = type_
396
397
    def to_str(self, db, left_table):
398 2353 aaronmk
        # Switch order (left_table is on the right in the comparison)
399
        right_table = left_table
400
        left_table = self.table # note left_table is reassigned
401
402 2260 aaronmk
        def join(entry):
403
            '''Parses non-USING joins'''
404
            right_table_col, left_table_col = entry
405
406 2353 aaronmk
            # Switch order (right_table_col is on the left in the comparison)
407
            left = right_table_col
408
            right = left_table_col
409
410 2260 aaronmk
            # Parse special values
411 2353 aaronmk
            if right is join_same: right = left
412
            elif right is join_same_not_null:
413
                right = CompareCond(as_Col(left, right_table), '~=')
414 2260 aaronmk
415 2353 aaronmk
            right = as_ValueCond(right, right_table)
416 2578 aaronmk
            return right.to_str(db, as_Col(left, left_table))
417 2260 aaronmk
418 2265 aaronmk
        # Create join condition
419
        type_ = self.type_
420 2276 aaronmk
        joins = self.mapping
421 2265 aaronmk
        if type_ is not filter_out and reduce(operator.and_,
422 2460 aaronmk
            (v is join_same_not_null for v in joins.itervalues())):
423 2260 aaronmk
            # all cols w/ USING, so can use simpler USING syntax
424 2298 aaronmk
            cols = (as_Col(v).to_str(db) for v in joins.iterkeys())
425
            join_cond = 'USING ('+(', '.join(cols))+')'
426 2576 aaronmk
        else:
427
            if len(joins) == 1: whitespace = ' '
428
            else: whitespace = '\n'
429 2577 aaronmk
            join_cond = combine_conds(map(join, joins.iteritems()), 'ON')
430 2260 aaronmk
431
        # Create join
432
        if type_ is filter_out: type_ = 'LEFT'
433 2266 aaronmk
        str_ = ''
434
        if type_ != None: str_ += type_+' '
435 2353 aaronmk
        str_ += 'JOIN '+left_table.to_str(db)+' '+join_cond
436 2266 aaronmk
        return str_
437 2349 aaronmk
438 2514 aaronmk
    def __repr__(self): return self.to_str(mockDb, '<left_table>')
439 2424 aaronmk
440
##### Value exprs
441
442
row_count = CustomCode('count(*)')
443 2674 aaronmk
444 2692 aaronmk
def EnsureNotNull(value, null=r'\N'):
445 2694 aaronmk
    return FunctionCall(InternalFunction('coalesce'), as_Col(value), null)
446 2692 aaronmk
447 2674 aaronmk
##### Database structure
448
449
class TypedCol(Col):
450
    def __init__(self, name, type_):
451
        Col.__init__(self, name)
452
453
        self.type = type_
454
455
    def to_str(self, db): return Col.to_str(self, db)+' '+self.type
456
457
    def to_Col(self): return Col(self.name)