sqlglot.planner

  1from __future__ import annotations
  2
  3import math
  4import typing as t
  5
  6from sqlglot import alias, exp
  7from sqlglot.helper import name_sequence
  8from sqlglot.optimizer.eliminate_joins import join_condition
  9
 10
 11class Plan:
 12    def __init__(self, expression: exp.Expression) -> None:
 13        self.expression = expression.copy()
 14        self.root = Step.from_expression(self.expression)
 15        self._dag: t.Dict[Step, t.Set[Step]] = {}
 16
 17    @property
 18    def dag(self) -> t.Dict[Step, t.Set[Step]]:
 19        if not self._dag:
 20            dag: t.Dict[Step, t.Set[Step]] = {}
 21            nodes = {self.root}
 22
 23            while nodes:
 24                node = nodes.pop()
 25                dag[node] = set()
 26
 27                for dep in node.dependencies:
 28                    dag[node].add(dep)
 29                    nodes.add(dep)
 30
 31            self._dag = dag
 32
 33        return self._dag
 34
 35    @property
 36    def leaves(self) -> t.Iterator[Step]:
 37        return (node for node, deps in self.dag.items() if not deps)
 38
 39    def __repr__(self) -> str:
 40        return f"Plan\n----\n{repr(self.root)}"
 41
 42
 43class Step:
 44    @classmethod
 45    def from_expression(
 46        cls, expression: exp.Expression, ctes: t.Optional[t.Dict[str, Step]] = None
 47    ) -> Step:
 48        """
 49        Builds a DAG of Steps from a SQL expression so that it's easier to execute in an engine.
 50        Note: the expression's tables and subqueries must be aliased for this method to work. For
 51        example, given the following expression:
 52
 53        SELECT
 54          x.a,
 55          SUM(x.b)
 56        FROM x AS x
 57        JOIN y AS y
 58          ON x.a = y.a
 59        GROUP BY x.a
 60
 61        the following DAG is produced (the expression IDs might differ per execution):
 62
 63        - Aggregate: x (4347984624)
 64            Context:
 65              Aggregations:
 66                - SUM(x.b)
 67              Group:
 68                - x.a
 69            Projections:
 70              - x.a
 71              - "x".""
 72            Dependencies:
 73            - Join: x (4347985296)
 74              Context:
 75                y:
 76                On: x.a = y.a
 77              Projections:
 78              Dependencies:
 79              - Scan: x (4347983136)
 80                Context:
 81                  Source: x AS x
 82                Projections:
 83              - Scan: y (4343416624)
 84                Context:
 85                  Source: y AS y
 86                Projections:
 87
 88        Args:
 89            expression: the expression to build the DAG from.
 90            ctes: a dictionary that maps CTEs to their corresponding Step DAG by name.
 91
 92        Returns:
 93            A Step DAG corresponding to `expression`.
 94        """
 95        ctes = ctes or {}
 96        expression = expression.unnest()
 97        with_ = expression.args.get("with")
 98
 99        # CTEs break the mold of scope and introduce themselves to all in the context.
100        if with_:
101            ctes = ctes.copy()
102            for cte in with_.expressions:
103                step = Step.from_expression(cte.this, ctes)
104                step.name = cte.alias
105                ctes[step.name] = step  # type: ignore
106
107        from_ = expression.args.get("from")
108
109        if isinstance(expression, exp.Select) and from_:
110            step = Scan.from_expression(from_.this, ctes)
111        elif isinstance(expression, exp.Union):
112            step = SetOperation.from_expression(expression, ctes)
113        else:
114            step = Scan()
115
116        joins = expression.args.get("joins")
117
118        if joins:
119            join = Join.from_joins(joins, ctes)
120            join.name = step.name
121            join.source_name = step.name
122            join.add_dependency(step)
123            step = join
124
125        projections = []  # final selects in this chain of steps representing a select
126        operands = {}  # intermediate computations of agg funcs eg x + 1 in SUM(x + 1)
127        aggregations = set()
128        next_operand_name = name_sequence("_a_")
129
130        def extract_agg_operands(expression):
131            agg_funcs = tuple(expression.find_all(exp.AggFunc))
132            if agg_funcs:
133                aggregations.add(expression)
134
135            for agg in agg_funcs:
136                for operand in agg.unnest_operands():
137                    if isinstance(operand, exp.Column):
138                        continue
139                    if operand not in operands:
140                        operands[operand] = next_operand_name()
141
142                    operand.replace(exp.column(operands[operand], quoted=True))
143
144            return bool(agg_funcs)
145
146        def set_ops_and_aggs(step):
147            step.operands = tuple(alias(operand, alias_) for operand, alias_ in operands.items())
148            step.aggregations = list(aggregations)
149
150        for e in expression.expressions:
151            if e.find(exp.AggFunc):
152                projections.append(exp.column(e.alias_or_name, step.name, quoted=True))
153                extract_agg_operands(e)
154            else:
155                projections.append(e)
156
157        where = expression.args.get("where")
158
159        if where:
160            step.condition = where.this
161
162        group = expression.args.get("group")
163
164        if group or aggregations:
165            aggregate = Aggregate()
166            aggregate.source = step.name
167            aggregate.name = step.name
168
169            having = expression.args.get("having")
170
171            if having:
172                if extract_agg_operands(exp.alias_(having.this, "_h", quoted=True)):
173                    aggregate.condition = exp.column("_h", step.name, quoted=True)
174                else:
175                    aggregate.condition = having.this
176
177            set_ops_and_aggs(aggregate)
178
179            # give aggregates names and replace projections with references to them
180            aggregate.group = {
181                f"_g{i}": e for i, e in enumerate(group.expressions if group else [])
182            }
183
184            intermediate: t.Dict[str | exp.Expression, str] = {}
185            for k, v in aggregate.group.items():
186                intermediate[v] = k
187                if isinstance(v, exp.Column):
188                    intermediate[v.name] = k
189
190            for projection in projections:
191                for node in projection.walk():
192                    name = intermediate.get(node)
193                    if name:
194                        node.replace(exp.column(name, step.name))
195
196            if aggregate.condition:
197                for node in aggregate.condition.walk():
198                    name = intermediate.get(node) or intermediate.get(node.name)
199                    if name:
200                        node.replace(exp.column(name, step.name))
201
202            aggregate.add_dependency(step)
203            step = aggregate
204
205        order = expression.args.get("order")
206
207        if order:
208            if isinstance(step, Aggregate):
209                for i, ordered in enumerate(order.expressions):
210                    if extract_agg_operands(exp.alias_(ordered.this, f"_o_{i}", quoted=True)):
211                        ordered.this.replace(exp.column(f"_o_{i}", step.name, quoted=True))
212
213                set_ops_and_aggs(aggregate)
214
215            sort = Sort()
216            sort.name = step.name
217            sort.key = order.expressions
218            sort.add_dependency(step)
219            step = sort
220
221        step.projections = projections
222
223        if isinstance(expression, exp.Select) and expression.args.get("distinct"):
224            distinct = Aggregate()
225            distinct.source = step.name
226            distinct.name = step.name
227            distinct.group = {
228                e.alias_or_name: exp.column(col=e.alias_or_name, table=step.name)
229                for e in projections or expression.expressions
230            }
231            distinct.add_dependency(step)
232            step = distinct
233
234        limit = expression.args.get("limit")
235
236        if limit:
237            step.limit = int(limit.text("expression"))
238
239        return step
240
241    def __init__(self) -> None:
242        self.name: t.Optional[str] = None
243        self.dependencies: t.Set[Step] = set()
244        self.dependents: t.Set[Step] = set()
245        self.projections: t.Sequence[exp.Expression] = []
246        self.limit: float = math.inf
247        self.condition: t.Optional[exp.Expression] = None
248
249    def add_dependency(self, dependency: Step) -> None:
250        self.dependencies.add(dependency)
251        dependency.dependents.add(self)
252
253    def __repr__(self) -> str:
254        return self.to_s()
255
256    def to_s(self, level: int = 0) -> str:
257        indent = "  " * level
258        nested = f"{indent}    "
259
260        context = self._to_s(f"{nested}  ")
261
262        if context:
263            context = [f"{nested}Context:"] + context
264
265        lines = [
266            f"{indent}- {self.id}",
267            *context,
268            f"{nested}Projections:",
269        ]
270
271        for expression in self.projections:
272            lines.append(f"{nested}  - {expression.sql()}")
273
274        if self.condition:
275            lines.append(f"{nested}Condition: {self.condition.sql()}")
276
277        if self.limit is not math.inf:
278            lines.append(f"{nested}Limit: {self.limit}")
279
280        if self.dependencies:
281            lines.append(f"{nested}Dependencies:")
282            for dependency in self.dependencies:
283                lines.append("  " + dependency.to_s(level + 1))
284
285        return "\n".join(lines)
286
287    @property
288    def type_name(self) -> str:
289        return self.__class__.__name__
290
291    @property
292    def id(self) -> str:
293        name = self.name
294        name = f" {name}" if name else ""
295        return f"{self.type_name}:{name} ({id(self)})"
296
297    def _to_s(self, _indent: str) -> t.List[str]:
298        return []
299
300
301class Scan(Step):
302    @classmethod
303    def from_expression(
304        cls, expression: exp.Expression, ctes: t.Optional[t.Dict[str, Step]] = None
305    ) -> Step:
306        table = expression
307        alias_ = expression.alias_or_name
308
309        if isinstance(expression, exp.Subquery):
310            table = expression.this
311            step = Step.from_expression(table, ctes)
312            step.name = alias_
313            return step
314
315        step = Scan()
316        step.name = alias_
317        step.source = expression
318        if ctes and table.name in ctes:
319            step.add_dependency(ctes[table.name])
320
321        return step
322
323    def __init__(self) -> None:
324        super().__init__()
325        self.source: t.Optional[exp.Expression] = None
326
327    def _to_s(self, indent: str) -> t.List[str]:
328        return [f"{indent}Source: {self.source.sql() if self.source else '-static-'}"]  # type: ignore
329
330
331class Join(Step):
332    @classmethod
333    def from_joins(
334        cls, joins: t.Iterable[exp.Join], ctes: t.Optional[t.Dict[str, Step]] = None
335    ) -> Join:
336        step = Join()
337
338        for join in joins:
339            source_key, join_key, condition = join_condition(join)
340            step.joins[join.alias_or_name] = {
341                "side": join.side,  # type: ignore
342                "join_key": join_key,
343                "source_key": source_key,
344                "condition": condition,
345            }
346
347            step.add_dependency(Scan.from_expression(join.this, ctes))
348
349        return step
350
351    def __init__(self) -> None:
352        super().__init__()
353        self.source_name: t.Optional[str] = None
354        self.joins: t.Dict[str, t.Dict[str, t.List[str] | exp.Expression]] = {}
355
356    def _to_s(self, indent: str) -> t.List[str]:
357        lines = [f"{indent}Source: {self.source_name or self.name}"]
358        for name, join in self.joins.items():
359            lines.append(f"{indent}{name}: {join['side'] or 'INNER'}")
360            join_key = ", ".join(str(key) for key in t.cast(list, join.get("join_key") or []))
361            if join_key:
362                lines.append(f"{indent}Key: {join_key}")
363            if join.get("condition"):
364                lines.append(f"{indent}On: {join['condition'].sql()}")  # type: ignore
365        return lines
366
367
368class Aggregate(Step):
369    def __init__(self) -> None:
370        super().__init__()
371        self.aggregations: t.List[exp.Expression] = []
372        self.operands: t.Tuple[exp.Expression, ...] = ()
373        self.group: t.Dict[str, exp.Expression] = {}
374        self.source: t.Optional[str] = None
375
376    def _to_s(self, indent: str) -> t.List[str]:
377        lines = [f"{indent}Aggregations:"]
378
379        for expression in self.aggregations:
380            lines.append(f"{indent}  - {expression.sql()}")
381
382        if self.group:
383            lines.append(f"{indent}Group:")
384            for expression in self.group.values():
385                lines.append(f"{indent}  - {expression.sql()}")
386        if self.condition:
387            lines.append(f"{indent}Having:")
388            lines.append(f"{indent}  - {self.condition.sql()}")
389        if self.operands:
390            lines.append(f"{indent}Operands:")
391            for expression in self.operands:
392                lines.append(f"{indent}  - {expression.sql()}")
393
394        return lines
395
396
397class Sort(Step):
398    def __init__(self) -> None:
399        super().__init__()
400        self.key = None
401
402    def _to_s(self, indent: str) -> t.List[str]:
403        lines = [f"{indent}Key:"]
404
405        for expression in self.key:  # type: ignore
406            lines.append(f"{indent}  - {expression.sql()}")
407
408        return lines
409
410
411class SetOperation(Step):
412    def __init__(
413        self,
414        op: t.Type[exp.Expression],
415        left: str | None,
416        right: str | None,
417        distinct: bool = False,
418    ) -> None:
419        super().__init__()
420        self.op = op
421        self.left = left
422        self.right = right
423        self.distinct = distinct
424
425    @classmethod
426    def from_expression(
427        cls, expression: exp.Expression, ctes: t.Optional[t.Dict[str, Step]] = None
428    ) -> SetOperation:
429        assert isinstance(expression, exp.Union)
430
431        left = Step.from_expression(expression.left, ctes)
432        # SELECT 1 UNION SELECT 2  <-- these subqueries don't have names
433        left.name = left.name or "left"
434        right = Step.from_expression(expression.right, ctes)
435        right.name = right.name or "right"
436        step = cls(
437            op=expression.__class__,
438            left=left.name,
439            right=right.name,
440            distinct=bool(expression.args.get("distinct")),
441        )
442
443        step.add_dependency(left)
444        step.add_dependency(right)
445
446        limit = expression.args.get("limit")
447
448        if limit:
449            step.limit = int(limit.text("expression"))
450
451        return step
452
453    def _to_s(self, indent: str) -> t.List[str]:
454        lines = []
455        if self.distinct:
456            lines.append(f"{indent}Distinct: {self.distinct}")
457        return lines
458
459    @property
460    def type_name(self) -> str:
461        return self.op.__name__