File: compiler.py

package info (click to toggle)
python-django-tree-queries 0.20-1
  • links: PTS, VCS
  • area: main
  • in suites: sid
  • size: 388 kB
  • sloc: python: 2,269; makefile: 26; sh: 6
file content (577 lines) | stat: -rw-r--r-- 20,436 bytes parent folder | download
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
import django
from django.db import connections
from django.db.models import Expression, F, QuerySet, Value, Window
from django.db.models.functions import RowNumber
from django.db.models.sql.compiler import SQLCompiler
from django.db.models.sql.query import Query


SEPARATOR = "\x1f"


def _find_tree_model(cls):
    return cls._meta.get_field("parent").model


class TreeQuery(Query):
    def __init__(self, *args, **kwargs):
        super().__init__(*args, **kwargs)
        self._setup_query()

    def _setup_query(self):
        """
        Run on initialization and at the end of chaining. Any attributes that
        would normally be set in __init__() should go here instead.
        """
        # We add the variables for `sibling_order` and `rank_table_query` here so they
        # act as instance variables which do not persist between user queries
        # the way class variables do

        # Only add the sibling_order attribute if the query doesn't already have one to preserve cloning behavior
        if not hasattr(self, "sibling_order"):
            # Add an attribute to control the ordering of siblings within trees
            opts = _find_tree_model(self.model)._meta
            self.sibling_order = opts.ordering if opts.ordering else opts.pk.attname

        # Only add the rank_table_query attribute if the query doesn't already have one to preserve cloning behavior
        if not hasattr(self, "rank_table_query"):
            # Create a default QuerySet for the rank_table to use
            # so we can avoid recursion
            self.rank_table_query = QuerySet(model=_find_tree_model(self.model))

        if not hasattr(self, "tree_fields"):
            self.tree_fields = {}

    def get_compiler(self, using=None, connection=None, **kwargs):
        # Copied from django/db/models/sql/query.py
        if using is None and connection is None:
            raise ValueError("Need either using or connection")
        if using:
            connection = connections[using]
        # Difference: Not connection.ops.compiler, but our own compiler which
        # adds the CTE.

        # **kwargs passes on elide_empty from Django 4.0 onwards
        return TreeCompiler(self, connection, using, **kwargs)

    def get_sibling_order(self):
        return self.sibling_order

    def get_rank_table_query(self):
        return self.rank_table_query

    def get_tree_fields(self):
        return self.tree_fields


class TreeCompiler(SQLCompiler):
    CTE_POSTGRESQL = """
    WITH RECURSIVE __rank_table(
        {tree_fields_columns}
        "{pk}",
        "{parent}",
        "rank_order"
    ) AS (
        {rank_table}
    ),
    __tree (
        {tree_fields_names}
        "tree_depth",
        "tree_path",
        "tree_ordering",
        "tree_pk"
    ) AS (
        SELECT
            {tree_fields_initial}
            0,
            array[T.{pk}],
            array[T.rank_order],
            T."{pk}"
        FROM __rank_table T
        WHERE T."{parent}" IS NULL

        UNION ALL

        SELECT
            {tree_fields_recursive}
            __tree.tree_depth + 1,
            __tree.tree_path || T.{pk},
            __tree.tree_ordering || T.rank_order,
            T."{pk}"
        FROM __rank_table T
        JOIN __tree ON T."{parent}" = __tree.tree_pk
    )
    """

    CTE_MYSQL = """
    WITH RECURSIVE __rank_table(
        {tree_fields_columns}
        {pk},
        {parent},
        rank_order
    ) AS (
        {rank_table}
    ),
    __tree(
        {tree_fields_names}
        tree_depth,
        tree_path,
        tree_ordering,
        tree_pk
    ) AS (
        SELECT
            {tree_fields_initial}
            0,
            -- Limit to max. 50 levels...
            CAST(CONCAT("{sep}", {pk}, "{sep}") AS char(1000)),
            CAST(CONCAT("{sep}", LPAD(CONCAT(T.rank_order, "{sep}"), 20, "0"))
                AS char(1000)),
            T.{pk}
        FROM __rank_table T
        WHERE T.{parent} IS NULL

        UNION ALL

        SELECT
            {tree_fields_recursive}
            __tree.tree_depth + 1,
            CONCAT(__tree.tree_path, T2.{pk}, "{sep}"),
            CONCAT(__tree.tree_ordering, LPAD(CONCAT(T2.rank_order, "{sep}"), 20, "0")),
            T2.{pk}
        FROM __tree, __rank_table T2
        WHERE __tree.tree_pk = T2.{parent}
    )
    """

    CTE_SQLITE = """
    WITH RECURSIVE __rank_table(
        {tree_fields_columns}
        {pk},
        {parent},
        rank_order
    ) AS (
        {rank_table}
    ),
    __tree(
        {tree_fields_names}
        tree_depth,
        tree_path,
        tree_ordering,
        tree_pk
    ) AS (
        SELECT
            {tree_fields_initial}
            0,
            printf("{sep}%%s{sep}", {pk}),
            printf("{sep}%%020s{sep}", T.rank_order),
            T."{pk}" tree_pk
        FROM __rank_table T
        WHERE T."{parent}" IS NULL

        UNION ALL

        SELECT
            {tree_fields_recursive}
            __tree.tree_depth + 1,
            __tree.tree_path || printf("%%s{sep}", T.{pk}),
            __tree.tree_ordering || printf("%%020s{sep}", T.rank_order),
            T."{pk}"
        FROM __rank_table T
        JOIN __tree ON T."{parent}" = __tree.tree_pk
    )
    """

    # Optimized CTEs without rank table for simple cases
    CTE_POSTGRESQL_SIMPLE = """
    WITH RECURSIVE __tree (
        {tree_fields_names}"tree_depth",
        "tree_path",
        "tree_ordering",
        "tree_pk"
    ) AS (
        SELECT
            {tree_fields_initial}0,
            array[T.{pk}],
            array[T."{order_field}"],
            T.{pk}
        FROM {db_table} T
        WHERE T."{parent}" IS NULL

        UNION ALL

        SELECT
            {tree_fields_recursive}__tree.tree_depth + 1,
            __tree.tree_path || T.{pk},
            __tree.tree_ordering || T."{order_field}",
            T.{pk}
        FROM {db_table} T
        JOIN __tree ON T."{parent}" = __tree.tree_pk
    )
    """

    CTE_MYSQL_SIMPLE = """
    WITH RECURSIVE __tree(
        {tree_fields_names}tree_depth,
        tree_path,
        tree_ordering,
        tree_pk
    ) AS (
        SELECT
            {tree_fields_initial}0,
            CAST(CONCAT("{sep}", T.{pk}, "{sep}") AS char(1000)),
            CAST(CONCAT("{sep}", LPAD(CONCAT(T.`{order_field}`, "{sep}"), 20, "0")) AS char(1000)),
            T.{pk}
        FROM {db_table} T
        WHERE T.`{parent}` IS NULL

        UNION ALL

        SELECT
            {tree_fields_recursive}__tree.tree_depth + 1,
            CONCAT(__tree.tree_path, T.{pk}, "{sep}"),
            CONCAT(__tree.tree_ordering, LPAD(CONCAT(T.`{order_field}`, "{sep}"), 20, "0")),
            T.{pk}
        FROM {db_table} T, __tree
        WHERE __tree.tree_pk = T.`{parent}`
    )
    """

    CTE_SQLITE_SIMPLE = """
    WITH RECURSIVE __tree(
        {tree_fields_names}tree_depth,
        tree_path,
        tree_ordering,
        tree_pk
    ) AS (
        SELECT
            {tree_fields_initial}0,
            "{sep}" || T."{pk}" || "{sep}",
            "{sep}" || printf("%%020s", T."{order_field}") || "{sep}",
            T."{pk}"
        FROM {db_table} T
        WHERE T."{parent}" IS NULL

        UNION ALL

        SELECT
            {tree_fields_recursive}__tree.tree_depth + 1,
            __tree.tree_path || T."{pk}" || "{sep}",
            __tree.tree_ordering || printf("%%020s", T."{order_field}") || "{sep}",
            T."{pk}"
        FROM {db_table} T
        JOIN __tree ON T."{parent}" = __tree.tree_pk
    )
    """

    def _can_skip_rank_table(self):
        """
        Determine if we can skip the rank table optimization.
        We can skip it when:
        1. No tree filters are applied (rank_table_query is unchanged)
        2. Simple ordering (single field, ascending)
        3. No custom tree fields
        """

        # Check if tree filters have been applied
        original_query = QuerySet(model=_find_tree_model(self.query.model))
        if str(self.query.get_rank_table_query().query) != str(original_query.query):
            return False

        # Check if custom tree fields are simple column references
        tree_fields = self.query.get_tree_fields()
        if tree_fields:
            model = _find_tree_model(self.query.model)
            for name, column in tree_fields.items():
                # Only allow simple column names (no complex expressions)
                if not isinstance(column, str):
                    return False
                # Check if it's a valid field on the model
                try:
                    model._meta.get_field(column)
                except FieldDoesNotExist:
                    return False

        # Check for complex ordering
        sibling_order = self.query.get_sibling_order()
        if isinstance(sibling_order, (list, tuple)):
            if len(sibling_order) > 1:
                return False
            order_field = sibling_order[0]
        else:
            order_field = sibling_order

        # Check for descending order or complex expressions
        if (
            isinstance(order_field, str)
            and order_field.startswith("-")
            or not isinstance(order_field, str)
        ):
            return False

        # Check for related field lookups (contains __)
        if "__" in order_field:
            return False

        # Check if the ordering field is numeric/integer
        # For string fields, the optimization might not preserve correct order
        # because we bypass the ROW_NUMBER() ranking that the complex CTE uses
        field = _find_tree_model(self.query.model)._meta.get_field(order_field)
        if not hasattr(field, "get_internal_type"):
            return False
        field_type = field.get_internal_type()
        if field_type not in (
            "AutoField",
            "BigAutoField",
            "IntegerField",
            "BigIntegerField",
            "PositiveIntegerField",
            "PositiveSmallIntegerField",
            "SmallIntegerField",
        ):
            return False

        return True

    def get_rank_table(self):
        # Get and validate sibling_order
        sibling_order = self.query.get_sibling_order()

        if isinstance(sibling_order, (list, tuple)):
            order_fields = sibling_order
        elif isinstance(sibling_order, str):
            order_fields = [sibling_order]
        else:
            raise ValueError(
                "Sibling order must be a string or a list or tuple of strings."
            )

        # Convert strings to expressions. This is to maintain backwards compatibility
        # with Django versions < 4.1
        if django.VERSION < (4, 1):
            base_order = []
            for field in order_fields:
                if isinstance(field, Expression):
                    base_order.append(field)
                elif isinstance(field, str):
                    if field[0] == "-":
                        base_order.append(F(field[1:]).desc())
                    else:
                        base_order.append(F(field).asc())
            order_fields = base_order

        # Get the rank table query
        rank_table_query = self.query.get_rank_table_query()

        rank_table_query = (
            rank_table_query.order_by()  # Ensure there is no ORDER BY at the end of the SQL
            # Values allows us to both limit and specify the order of
            # the columns selected so that they match the CTE
            .values(
                *self.query.get_tree_fields().values(),
                "pk",
                "parent",
                rank_order=Window(
                    expression=RowNumber(),
                    order_by=order_fields,
                ),
            )
        )

        rank_table_sql, rank_table_params = rank_table_query.query.sql_with_params()

        return rank_table_sql, rank_table_params

    def as_sql(self, *args, **kwargs):
        # Try detecting if we're used in a EXISTS(1 as "a") subquery like
        # Django's sql.Query.exists() generates. If we detect such a query
        # we're skipping the tree generation since it's not necessary in the
        # best case and references unused table aliases (leading to SQL errors)
        # in the worst case. See GitHub issue #63.
        if (
            self.query.subquery
            and (ann := self.query.annotations)
            and ann == {"a": Value(1)}
        ):
            return super().as_sql(*args, **kwargs)

        # The general idea is that if we have a summary query (e.g. .count())
        # then we do not want to ask Django to add the tree fields to the query
        # using .query.add_extra. The way to determine whether we have a
        # summary query on our hands is to check the is_summary attribute of
        # all annotations.
        #
        # A new case appeared in the GitHub issue #26: Queries using
        # .distinct().count() crashed. The reason for this is that Django uses
        # a distinct subquery *without* annotations -- the annotations are kept
        # in the surrounding query. Because of this we look at the distinct and
        # subquery attributes.
        #
        # I am not confident that this is the perfect way to approach this
        # problem but I just gotta stop worrying and trust the testsuite.
        skip_tree_fields = (
            self.query.distinct and self.query.subquery
        ) or any(  # pragma: no branch
            # OK if generator is not consumed completely
            annotation.is_summary
            for alias, annotation in self.query.annotations.items()
        )
        opts = _find_tree_model(self.query.model)._meta

        params = {
            "parent": "parent_id",  # XXX Hardcoded.
            "pk": opts.pk.attname,
            "db_table": opts.db_table,
            "sep": SEPARATOR,
        }

        # Check if we can use the optimized path without rank table
        use_rank_table = not self._can_skip_rank_table()

        if use_rank_table:
            # Get the rank_table SQL and params
            rank_table_sql, rank_table_params = self.get_rank_table()
            params["rank_table"] = rank_table_sql
        else:
            # Use optimized path - get the order field for simple CTE
            sibling_order = self.query.get_sibling_order()
            if isinstance(sibling_order, (list, tuple)):
                order_field = sibling_order[0]
            else:
                order_field = sibling_order
            params["order_field"] = order_field
            rank_table_params = []

        # Set database-specific CTE template and column reference format
        if self.connection.vendor == "postgresql":
            cte = (
                self.CTE_POSTGRESQL_SIMPLE
                if not use_rank_table
                else self.CTE_POSTGRESQL
            )
            cte_initial = "array[{column}]::text[], "
            cte_recursive = "__tree.{name} || {column}::text, "
        elif self.connection.vendor == "sqlite":
            cte = self.CTE_SQLITE_SIMPLE if not use_rank_table else self.CTE_SQLITE
            cte_initial = 'printf("{sep}%%s{sep}", {column}), '
            cte_recursive = '__tree.{name} || printf("%%s{sep}", {column}), '
        elif self.connection.vendor == "mysql":
            cte = self.CTE_MYSQL_SIMPLE if not use_rank_table else self.CTE_MYSQL
            cte_initial = 'CAST(CONCAT("{sep}", {column}, "{sep}") AS char(1000)), '
            cte_recursive = 'CONCAT(__tree.{name}, {column}, "{sep}"), '

        tree_fields = self.query.get_tree_fields()
        qn = self.connection.ops.quote_name

        # Generate tree field parameters using unified templates
        # Set column reference format based on CTE type
        if use_rank_table:
            # Complex CTE uses rank table references
            column_ref_format = "{column}"
            params.update({
                "tree_fields_columns": "".join(
                    f"{qn(column)}, " for column in tree_fields.values()
                ),
            })
        else:
            # Simple CTE uses direct table references
            column_ref_format = "T.{column}"

        # Generate unified tree field parameters
        params.update({
            "tree_fields_names": "".join(f"{qn(name)}, " for name in tree_fields),
            "tree_fields_initial": "".join(
                cte_initial.format(
                    column=column_ref_format.format(column=qn(column)),
                    name=qn(name),
                    sep=SEPARATOR,
                )
                for name, column in tree_fields.items()
            ),
            "tree_fields_recursive": "".join(
                cte_recursive.format(
                    column=column_ref_format.format(column=qn(column)),
                    name=qn(name),
                    sep=SEPARATOR,
                )
                for name, column in tree_fields.items()
            ),
        })

        if "__tree" not in self.query.extra_tables:  # pragma: no branch - unlikely
            tree_params = params.copy()

            # use aliased table name (U0, U1, U2)
            base_table = self.query.__dict__.get("base_table")
            if base_table is not None:
                tree_params["db_table"] = base_table

            # When using tree queries in subqueries our base table may use
            # an alias. Let's hope using the first alias is correct.
            aliases = self.query.table_map.get(tree_params["db_table"])
            if aliases:
                tree_params["db_table"] = aliases[0]

            select = {
                "tree_depth": "__tree.tree_depth",
                "tree_path": "__tree.tree_path",
                "tree_ordering": "__tree.tree_ordering",
            }
            # Add custom tree fields for both simple and complex CTEs
            select.update({name: f"__tree.{name}" for name in tree_fields})
            self.query.add_extra(
                # Do not add extra fields to the select statement when it is a
                # summary query or when using .values() or .values_list()
                select={} if skip_tree_fields or self.query.values_select else select,
                select_params=None,
                where=["__tree.tree_pk = {db_table}.{pk}".format(**tree_params)],
                params=None,
                tables=["__tree"],
                order_by=(
                    []
                    # Do not add ordering for aggregates, or if the ordering
                    # has already been specified using .extra()
                    if skip_tree_fields or self.query.extra_order_by
                    else ["__tree.tree_ordering"]  # DFS is the only true way
                ),
            )

        sql_0, sql_1 = super().as_sql(*args, **kwargs)
        explain = ""
        if sql_0.startswith("EXPLAIN "):
            explain, sql_0 = sql_0.split(" ", 1)

        # Pass any additional rank table sql paramaters so that the db backend can handle them.
        # This only works because we know that the CTE is at the start of the query.
        return (
            "".join([explain, cte.format(**params), sql_0]),
            (*rank_table_params, *sql_1),
        )

    def get_converters(self, expressions):
        converters = super().get_converters(expressions)
        tree_fields = {"__tree.tree_path", "__tree.tree_ordering"} | {
            f"__tree.{name}" for name in self.query.tree_fields
        }
        for i, expression in enumerate(expressions):
            # We care about tree fields and annotations only
            if not hasattr(expression, "sql"):
                continue

            if expression.sql in tree_fields:
                converters[i] = ([converter], expression)
        return converters


def converter(value, expression, connection, context=None):
    # context can be removed as soon as we only support Django>=2.0
    if isinstance(value, str):
        # MySQL/MariaDB and sqlite3 do not support arrays. Split the value on
        # the ASCII unit separator (chr(31)).
        # NOTE: The representation of array is NOT part of the API.
        value = value.split(SEPARATOR)[1:-1]

    try:
        # Either all values are convertible to int or don't bother
        return [int(v) for v in value]  # Maybe Field.to_python()?
    except ValueError:
        return value