sqruff_lib_core/utils/analysis/
query.rs

1use std::cell::RefCell;
2use std::rc::Rc;
3
4use smol_str::{SmolStr, StrExt, ToSmolStr};
5
6use super::select::SelectStatementColumnsAndTables;
7use crate::dialects::Dialect;
8use crate::dialects::common::AliasInfo;
9use crate::dialects::syntax::{SyntaxKind, SyntaxSet};
10use crate::helpers::IndexMap;
11use crate::parser::segments::ErasedSegment;
12use crate::utils::analysis::select::get_select_statement_info;
13use crate::utils::functional::segments::Segments;
14
15const SELECTABLE_TYPES: SyntaxSet = SyntaxSet::new(&[
16    SyntaxKind::WithCompoundStatement,
17    SyntaxKind::SetExpression,
18    SyntaxKind::SelectStatement,
19]);
20
21const SUBSELECT_TYPES: SyntaxSet = SyntaxSet::new(&[
22    SyntaxKind::MergeStatement,
23    SyntaxKind::UpdateStatement,
24    SyntaxKind::DeleteStatement,
25    // NOTE: Values clauses won't have sub selects, but it's
26    // also harmless to look, and they may appear in similar
27    // locations. We include them here because they come through
28    // the same code paths - although are likely to return nothing.
29    SyntaxKind::ValuesClause,
30]);
31
32#[derive(Debug, Clone, Copy)]
33pub enum QueryType {
34    Simple,
35    WithCompound,
36}
37
38pub struct WildcardInfo {
39    pub segment: ErasedSegment,
40    pub tables: Vec<SmolStr>,
41}
42
43#[derive(Debug, Clone)]
44pub struct Selectable<'me> {
45    pub selectable: ErasedSegment,
46    pub dialect: &'me Dialect,
47}
48
49impl Selectable<'_> {
50    pub fn find_alias(&self, table: &str) -> Option<AliasInfo> {
51        self.select_info()
52            .as_ref()?
53            .table_aliases
54            .iter()
55            .find(|&t| t.aliased && t.ref_str == table)
56            .cloned()
57    }
58}
59
60impl Selectable<'_> {
61    pub fn wildcard_info(&self) -> Vec<WildcardInfo> {
62        let Some(select_info) = self.select_info() else {
63            return Vec::new();
64        };
65
66        let mut buff = Vec::new();
67        for seg in select_info.select_targets {
68            if seg
69                .0
70                .child(const { &SyntaxSet::new(&[SyntaxKind::WildcardExpression]) })
71                .is_some()
72            {
73                if seg.0.raw().contains('.') {
74                    let table = seg
75                        .0
76                        .raw()
77                        .rsplit_once('.')
78                        .map(|x| x.0)
79                        .unwrap_or_default()
80                        .to_smolstr();
81                    buff.push(WildcardInfo {
82                        segment: seg.0.clone(),
83                        tables: vec![table],
84                    });
85                } else {
86                    let tables = select_info
87                        .table_aliases
88                        .iter()
89                        .filter(|it| !it.ref_str.is_empty())
90                        .map(|it| {
91                            if it.aliased {
92                                it.ref_str.clone()
93                            } else {
94                                it.from_expression_element.raw().clone()
95                            }
96                        })
97                        .collect();
98                    buff.push(WildcardInfo {
99                        segment: seg.0.clone(),
100                        tables,
101                    });
102                }
103            }
104        }
105
106        buff
107    }
108}
109
110impl Selectable<'_> {
111    pub fn select_info(&self) -> Option<SelectStatementColumnsAndTables> {
112        if self.selectable.is_type(SyntaxKind::SelectStatement) {
113            return get_select_statement_info(&self.selectable, self.dialect.into(), false);
114        }
115
116        let values = Segments::new(self.selectable.clone(), None);
117        let alias_expression = values
118            .children_all()
119            .find_first_where(|it: &ErasedSegment| it.is_type(SyntaxKind::AliasExpression));
120        let name = alias_expression
121            .children_all()
122            .find_first_where(|it: &ErasedSegment| {
123                matches!(
124                    it.get_type(),
125                    SyntaxKind::NakedIdentifier | SyntaxKind::QuotedIdentifier,
126                )
127            });
128
129        let alias_info = AliasInfo {
130            ref_str: if name.is_empty() {
131                SmolStr::new_static("")
132            } else {
133                name.first().unwrap().raw().clone()
134            },
135            segment: name.first().cloned(),
136            aliased: !name.is_empty(),
137            from_expression_element: self.selectable.clone(),
138            alias_expression: alias_expression.first().cloned(),
139            object_reference: None,
140        };
141
142        SelectStatementColumnsAndTables {
143            select_statement: self.selectable.clone(),
144            table_aliases: vec![alias_info],
145            standalone_aliases: Vec::new(),
146            reference_buffer: Vec::new(),
147            select_targets: Vec::new(),
148            col_aliases: Vec::new(),
149            using_cols: Vec::new(),
150        }
151        .into()
152    }
153}
154
155#[derive(Debug, Clone)]
156pub struct Query<'me> {
157    pub inner: Rc<RefCell<QueryInner<'me>>>,
158}
159
160#[derive(Debug, Clone)]
161pub struct QueryInner<'me> {
162    pub query_type: QueryType,
163    pub dialect: &'me Dialect,
164    pub selectables: Vec<Selectable<'me>>,
165    pub ctes: IndexMap<SmolStr, Query<'me>>,
166    pub parent: Option<Query<'me>>,
167    pub subqueries: Vec<Query<'me>>,
168    pub cte_definition_segment: Option<ErasedSegment>,
169    pub cte_name_segment: Option<ErasedSegment>,
170}
171
172impl<'me> Query<'me> {
173    pub fn crawl_sources(
174        &self,
175        segment: ErasedSegment,
176        pop: bool,
177        lookup_cte: bool,
178    ) -> Vec<Source<'me>> {
179        let mut acc = Vec::new();
180
181        for seg in segment.recursive_crawl(
182            const {
183                &SyntaxSet::new(&[
184                    SyntaxKind::TableReference,
185                    SyntaxKind::SetExpression,
186                    SyntaxKind::SelectStatement,
187                    SyntaxKind::ValuesClause,
188                ])
189            },
190            false,
191            &SyntaxSet::EMPTY,
192            false,
193        ) {
194            if seg.is_type(SyntaxKind::TableReference) {
195                let _seg = seg.reference();
196                if !_seg.is_qualified()
197                    && lookup_cte
198                    && let Some(cte) = self.lookup_cte(seg.raw().as_ref(), pop)
199                {
200                    acc.push(Source::Query(cte));
201                }
202                acc.push(Source::TableReference(seg.raw().clone()));
203            } else {
204                acc.push(Source::Query(Query::from_segment(
205                    &seg,
206                    self.inner.borrow().dialect,
207                    Some(self.clone()),
208                )))
209            }
210        }
211
212        if acc.is_empty()
213            && let Some(table_expr) =
214                segment.child(const { &SyntaxSet::new(&[SyntaxKind::TableExpression]) })
215        {
216            return vec![Source::TableReference(table_expr.raw().to_smolstr())];
217        }
218
219        acc
220    }
221
222    #[track_caller]
223    pub fn lookup_cte(&self, name: &str, pop: bool) -> Option<Query<'me>> {
224        let cte = if pop {
225            self.inner
226                .borrow_mut()
227                .ctes
228                .shift_remove(&name.to_uppercase_smolstr())
229        } else {
230            self.inner
231                .borrow()
232                .ctes
233                .get(&name.to_uppercase_smolstr())
234                .cloned()
235        };
236
237        cte.or_else(move || {
238            self.inner
239                .borrow_mut()
240                .parent
241                .as_mut()
242                .and_then(|it| it.lookup_cte(name, pop))
243        })
244    }
245
246    pub fn id(&self) -> *const RefCell<QueryInner<'me>> {
247        Rc::as_ptr(&self.inner)
248    }
249
250    fn post_init(&self) {
251        let this = self.clone();
252
253        for subquery in &RefCell::borrow(&self.inner).subqueries {
254            RefCell::borrow_mut(&subquery.inner).parent = this.clone().into();
255        }
256
257        for cte in RefCell::borrow(&self.inner).ctes.values().cloned() {
258            RefCell::borrow_mut(&cte.inner).parent = this.clone().into();
259        }
260    }
261}
262
263impl<'me> Query<'me> {
264    pub fn children(&self) -> Vec<Self> {
265        self.inner
266            .borrow()
267            .ctes
268            .values()
269            .chain(self.inner.borrow().subqueries.iter())
270            .cloned()
271            .collect()
272    }
273
274    fn extract_subqueries<'a>(selectable: &Selectable, dialect: &'a Dialect) -> Vec<Query<'a>> {
275        let mut acc = Vec::new();
276
277        for subselect in selectable.selectable.recursive_crawl(
278            &SELECTABLE_TYPES,
279            false,
280            &SyntaxSet::EMPTY,
281            false,
282        ) {
283            acc.push(Query::from_segment(&subselect, dialect, None));
284        }
285
286        acc
287    }
288
289    pub fn from_root<'a>(root_segment: &ErasedSegment, dialect: &'a Dialect) -> Option<Query<'a>> {
290        let stmts = root_segment.recursive_crawl(
291            &SELECTABLE_TYPES,
292            true,
293            &SyntaxSet::single(SyntaxKind::MergeStatement),
294            true,
295        );
296        let selectable_segment = stmts.first()?;
297
298        Some(Query::from_segment(selectable_segment, dialect, None))
299    }
300
301    pub fn from_segment<'a>(
302        segment: &ErasedSegment,
303        dialect: &'a Dialect,
304        parent: Option<Query<'a>>,
305    ) -> Query<'a> {
306        let mut selectables = Vec::new();
307        let mut subqueries = Vec::new();
308        let mut cte_defs: Vec<ErasedSegment> = Vec::new();
309        let mut query_type = QueryType::Simple;
310
311        if segment.is_type(SyntaxKind::SelectStatement)
312            || SUBSELECT_TYPES.contains(segment.get_type())
313        {
314            selectables.push(Selectable {
315                selectable: segment.clone(),
316                dialect,
317            });
318        } else if segment.is_type(SyntaxKind::SetExpression) {
319            selectables.extend(
320                segment
321                    .children(const { &SyntaxSet::new(&[SyntaxKind::SelectStatement]) })
322                    .cloned()
323                    .map(|selectable| Selectable {
324                        selectable,
325                        dialect,
326                    }),
327            )
328        } else {
329            query_type = QueryType::WithCompound;
330
331            for seg in segment.recursive_crawl(
332                const { &SyntaxSet::new(&[SyntaxKind::SelectStatement]) },
333                false,
334                const { &SyntaxSet::single(SyntaxKind::CommonTableExpression) },
335                true,
336            ) {
337                selectables.push(Selectable {
338                    selectable: seg,
339                    dialect,
340                });
341            }
342
343            for seg in segment.recursive_crawl(
344                const { &SyntaxSet::new(&[SyntaxKind::CommonTableExpression]) },
345                false,
346                const { &SyntaxSet::single(SyntaxKind::WithCompoundStatement) },
347                true,
348            ) {
349                cte_defs.push(seg);
350            }
351        }
352
353        for selectable in &selectables {
354            subqueries.extend(Self::extract_subqueries(selectable, dialect));
355        }
356
357        let outer_query = Query {
358            inner: Rc::new(RefCell::new(QueryInner {
359                query_type,
360                dialect,
361                selectables,
362                ctes: <_>::default(),
363                parent,
364                subqueries,
365                cte_definition_segment: None,
366                cte_name_segment: None,
367            })),
368        };
369
370        outer_query.post_init();
371
372        if cte_defs.is_empty() {
373            return outer_query;
374        }
375
376        let mut ctes = IndexMap::default();
377        for cte in cte_defs {
378            let name_seg = cte.segments()[0].clone();
379            let name = name_seg.raw().to_uppercase_smolstr();
380
381            let queries = cte.recursive_crawl(
382                const { &SELECTABLE_TYPES.union(&SUBSELECT_TYPES) },
383                true,
384                &SyntaxSet::EMPTY,
385                true,
386            );
387
388            if queries.is_empty() {
389                continue;
390            };
391
392            let query = &queries[0];
393            let query = Self::from_segment(query, dialect, outer_query.clone().into());
394
395            RefCell::borrow_mut(&query.inner).cte_definition_segment = cte.into();
396            RefCell::borrow_mut(&query.inner).cte_name_segment = name_seg.into();
397
398            ctes.insert(name, query);
399        }
400
401        RefCell::borrow_mut(&outer_query.inner).ctes = ctes;
402        outer_query
403    }
404}
405
406pub enum Source<'a> {
407    TableReference(SmolStr),
408    Query(Query<'a>),
409}