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.add_dependency(step)
122            step = join
123
124        projections = []  # final selects in this chain of steps representing a select
125        operands = {}  # intermediate computations of agg funcs eg x + 1 in SUM(x + 1)
126        aggregations = set()
127        next_operand_name = name_sequence("_a_")
128
129        def extract_agg_operands(expression):
130            agg_funcs = tuple(expression.find_all(exp.AggFunc))
131            if agg_funcs:
132                aggregations.add(expression)
133
134            for agg in agg_funcs:
135                for operand in agg.unnest_operands():
136                    if isinstance(operand, exp.Column):
137                        continue
138                    if operand not in operands:
139                        operands[operand] = next_operand_name()
140
141                    operand.replace(exp.column(operands[operand], quoted=True))
142
143            return bool(agg_funcs)
144
145        def set_ops_and_aggs(step):
146            step.operands = tuple(alias(operand, alias_) for operand, alias_ in operands.items())
147            step.aggregations = list(aggregations)
148
149        for e in expression.expressions:
150            if e.find(exp.AggFunc):
151                projections.append(exp.column(e.alias_or_name, step.name, quoted=True))
152                extract_agg_operands(e)
153            else:
154                projections.append(e)
155
156        where = expression.args.get("where")
157
158        if where:
159            step.condition = where.this
160
161        group = expression.args.get("group")
162
163        if group or aggregations:
164            aggregate = Aggregate()
165            aggregate.source = step.name
166            aggregate.name = step.name
167
168            having = expression.args.get("having")
169
170            if having:
171                if extract_agg_operands(exp.alias_(having.this, "_h", quoted=True)):
172                    aggregate.condition = exp.column("_h", step.name, quoted=True)
173                else:
174                    aggregate.condition = having.this
175
176            set_ops_and_aggs(aggregate)
177
178            # give aggregates names and replace projections with references to them
179            aggregate.group = {
180                f"_g{i}": e for i, e in enumerate(group.expressions if group else [])
181            }
182
183            intermediate: t.Dict[str | exp.Expression, str] = {}
184            for k, v in aggregate.group.items():
185                intermediate[v] = k
186                if isinstance(v, exp.Column):
187                    intermediate[v.name] = k
188
189            for projection in projections:
190                for node, *_ in projection.walk():
191                    name = intermediate.get(node)
192                    if name:
193                        node.replace(exp.column(name, step.name))
194
195            if aggregate.condition:
196                for node, *_ in aggregate.condition.walk():
197                    name = intermediate.get(node) or intermediate.get(node.name)
198                    if name:
199                        node.replace(exp.column(name, step.name))
200
201            aggregate.add_dependency(step)
202            step = aggregate
203
204        order = expression.args.get("order")
205
206        if order:
207            if isinstance(step, Aggregate):
208                for i, ordered in enumerate(order.expressions):
209                    if extract_agg_operands(exp.alias_(ordered.this, f"_o_{i}", quoted=True)):
210                        ordered.this.replace(exp.column(f"_o_{i}", step.name, quoted=True))
211
212                set_ops_and_aggs(aggregate)
213
214            sort = Sort()
215            sort.name = step.name
216            sort.key = order.expressions
217            sort.add_dependency(step)
218            step = sort
219
220        step.projections = projections
221
222        if isinstance(expression, exp.Select) and expression.args.get("distinct"):
223            distinct = Aggregate()
224            distinct.source = step.name
225            distinct.name = step.name
226            distinct.group = {
227                e.alias_or_name: exp.column(col=e.alias_or_name, table=step.name)
228                for e in projections or expression.expressions
229            }
230            distinct.add_dependency(step)
231            step = distinct
232
233        limit = expression.args.get("limit")
234
235        if limit:
236            step.limit = int(limit.text("expression"))
237
238        return step
239
240    def __init__(self) -> None:
241        self.name: t.Optional[str] = None
242        self.dependencies: t.Set[Step] = set()
243        self.dependents: t.Set[Step] = set()
244        self.projections: t.Sequence[exp.Expression] = []
245        self.limit: float = math.inf
246        self.condition: t.Optional[exp.Expression] = None
247
248    def add_dependency(self, dependency: Step) -> None:
249        self.dependencies.add(dependency)
250        dependency.dependents.add(self)
251
252    def __repr__(self) -> str:
253        return self.to_s()
254
255    def to_s(self, level: int = 0) -> str:
256        indent = "  " * level
257        nested = f"{indent}    "
258
259        context = self._to_s(f"{nested}  ")
260
261        if context:
262            context = [f"{nested}Context:"] + context
263
264        lines = [
265            f"{indent}- {self.id}",
266            *context,
267            f"{nested}Projections:",
268        ]
269
270        for expression in self.projections:
271            lines.append(f"{nested}  - {expression.sql()}")
272
273        if self.condition:
274            lines.append(f"{nested}Condition: {self.condition.sql()}")
275
276        if self.limit is not math.inf:
277            lines.append(f"{nested}Limit: {self.limit}")
278
279        if self.dependencies:
280            lines.append(f"{nested}Dependencies:")
281            for dependency in self.dependencies:
282                lines.append("  " + dependency.to_s(level + 1))
283
284        return "\n".join(lines)
285
286    @property
287    def type_name(self) -> str:
288        return self.__class__.__name__
289
290    @property
291    def id(self) -> str:
292        name = self.name
293        name = f" {name}" if name else ""
294        return f"{self.type_name}:{name} ({id(self)})"
295
296    def _to_s(self, _indent: str) -> t.List[str]:
297        return []
298
299
300class Scan(Step):
301    @classmethod
302    def from_expression(
303        cls, expression: exp.Expression, ctes: t.Optional[t.Dict[str, Step]] = None
304    ) -> Step:
305        table = expression
306        alias_ = expression.alias_or_name
307
308        if isinstance(expression, exp.Subquery):
309            table = expression.this
310            step = Step.from_expression(table, ctes)
311            step.name = alias_
312            return step
313
314        step = Scan()
315        step.name = alias_
316        step.source = expression
317        if ctes and table.name in ctes:
318            step.add_dependency(ctes[table.name])
319
320        return step
321
322    def __init__(self) -> None:
323        super().__init__()
324        self.source: t.Optional[exp.Expression] = None
325
326    def _to_s(self, indent: str) -> t.List[str]:
327        return [f"{indent}Source: {self.source.sql() if self.source else '-static-'}"]  # type: ignore
328
329
330class Join(Step):
331    @classmethod
332    def from_joins(
333        cls, joins: t.Iterable[exp.Join], ctes: t.Optional[t.Dict[str, Step]] = None
334    ) -> Step:
335        step = Join()
336
337        for join in joins:
338            source_key, join_key, condition = join_condition(join)
339            step.joins[join.alias_or_name] = {
340                "side": join.side,  # type: ignore
341                "join_key": join_key,
342                "source_key": source_key,
343                "condition": condition,
344            }
345
346            step.add_dependency(Scan.from_expression(join.this, ctes))
347
348        return step
349
350    def __init__(self) -> None:
351        super().__init__()
352        self.joins: t.Dict[str, t.Dict[str, t.List[str] | exp.Expression]] = {}
353
354    def _to_s(self, indent: str) -> t.List[str]:
355        lines = []
356        for name, join in self.joins.items():
357            lines.append(f"{indent}{name}: {join['side'] or 'INNER'}")
358            join_key = ", ".join(str(key) for key in t.cast(list, join.get("join_key") or []))
359            if join_key:
360                lines.append(f"{indent}Key: {join_key}")
361            if join.get("condition"):
362                lines.append(f"{indent}On: {join['condition'].sql()}")  # type: ignore
363        return lines
364
365
366class Aggregate(Step):
367    def __init__(self) -> None:
368        super().__init__()
369        self.aggregations: t.List[exp.Expression] = []
370        self.operands: t.Tuple[exp.Expression, ...] = ()
371        self.group: t.Dict[str, exp.Expression] = {}
372        self.source: t.Optional[str] = None
373
374    def _to_s(self, indent: str) -> t.List[str]:
375        lines = [f"{indent}Aggregations:"]
376
377        for expression in self.aggregations:
378            lines.append(f"{indent}  - {expression.sql()}")
379
380        if self.group:
381            lines.append(f"{indent}Group:")
382            for expression in self.group.values():
383                lines.append(f"{indent}  - {expression.sql()}")
384        if self.condition:
385            lines.append(f"{indent}Having:")
386            lines.append(f"{indent}  - {self.condition.sql()}")
387        if self.operands:
388            lines.append(f"{indent}Operands:")
389            for expression in self.operands:
390                lines.append(f"{indent}  - {expression.sql()}")
391
392        return lines
393
394
395class Sort(Step):
396    def __init__(self) -> None:
397        super().__init__()
398        self.key = None
399
400    def _to_s(self, indent: str) -> t.List[str]:
401        lines = [f"{indent}Key:"]
402
403        for expression in self.key:  # type: ignore
404            lines.append(f"{indent}  - {expression.sql()}")
405
406        return lines
407
408
409class SetOperation(Step):
410    def __init__(
411        self,
412        op: t.Type[exp.Expression],
413        left: str | None,
414        right: str | None,
415        distinct: bool = False,
416    ) -> None:
417        super().__init__()
418        self.op = op
419        self.left = left
420        self.right = right
421        self.distinct = distinct
422
423    @classmethod
424    def from_expression(
425        cls, expression: exp.Expression, ctes: t.Optional[t.Dict[str, Step]] = None
426    ) -> Step:
427        assert isinstance(expression, exp.Union)
428        left = Step.from_expression(expression.left, ctes)
429        right = Step.from_expression(expression.right, ctes)
430        step = cls(
431            op=expression.__class__,
432            left=left.name,
433            right=right.name,
434            distinct=bool(expression.args.get("distinct")),
435        )
436        step.add_dependency(left)
437        step.add_dependency(right)
438        return step
439
440    def _to_s(self, indent: str) -> t.List[str]:
441        lines = []
442        if self.distinct:
443            lines.append(f"{indent}Distinct: {self.distinct}")
444        return lines
445
446    @property
447    def type_name(self) -> str:
448        return self.op.__name__