Skip to main content

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            table_reference_buffer: Vec::new(),
151        }
152        .into()
153    }
154}
155
156#[derive(Debug, Clone)]
157pub struct Query<'me> {
158    pub inner: Rc<RefCell<QueryInner<'me>>>,
159}
160
161#[derive(Debug, Clone)]
162pub struct QueryInner<'me> {
163    pub query_type: QueryType,
164    pub dialect: &'me Dialect,
165    pub selectables: Vec<Selectable<'me>>,
166    pub ctes: IndexMap<SmolStr, Query<'me>>,
167    pub parent: Option<Query<'me>>,
168    pub subqueries: Vec<Query<'me>>,
169    pub cte_definition_segment: Option<ErasedSegment>,
170    pub cte_name_segment: Option<ErasedSegment>,
171}
172
173impl<'me> Query<'me> {
174    pub fn crawl_sources(
175        &self,
176        segment: ErasedSegment,
177        pop: bool,
178        lookup_cte: bool,
179    ) -> Vec<Source<'me>> {
180        let mut acc = Vec::new();
181
182        for seg in segment.recursive_crawl(
183            const {
184                &SyntaxSet::new(&[
185                    SyntaxKind::TableReference,
186                    SyntaxKind::SetExpression,
187                    SyntaxKind::SelectStatement,
188                    SyntaxKind::ValuesClause,
189                ])
190            },
191            false,
192            &SyntaxSet::EMPTY,
193            false,
194        ) {
195            if seg.is_type(SyntaxKind::TableReference) {
196                let _seg = seg.reference();
197                if !_seg.is_qualified()
198                    && lookup_cte
199                    && let Some(cte) = self.lookup_cte(seg.raw().as_ref(), pop)
200                {
201                    acc.push(Source::Query(cte));
202                }
203                acc.push(Source::TableReference(seg.raw().clone()));
204            } else {
205                acc.push(Source::Query(Query::from_segment(
206                    &seg,
207                    self.inner.borrow().dialect,
208                    Some(self.clone()),
209                )))
210            }
211        }
212
213        if acc.is_empty()
214            && let Some(table_expr) =
215                segment.child(const { &SyntaxSet::new(&[SyntaxKind::TableExpression]) })
216        {
217            return vec![Source::TableReference(table_expr.raw().to_smolstr())];
218        }
219
220        acc
221    }
222
223    #[track_caller]
224    pub fn lookup_cte(&self, name: &str, pop: bool) -> Option<Query<'me>> {
225        let cte = if pop {
226            self.inner
227                .borrow_mut()
228                .ctes
229                .shift_remove(&name.to_uppercase_smolstr())
230        } else {
231            self.inner
232                .borrow()
233                .ctes
234                .get(&name.to_uppercase_smolstr())
235                .cloned()
236        };
237
238        cte.or_else(move || {
239            self.inner
240                .borrow_mut()
241                .parent
242                .as_mut()
243                .and_then(|it| it.lookup_cte(name, pop))
244        })
245    }
246
247    pub fn id(&self) -> *const RefCell<QueryInner<'me>> {
248        Rc::as_ptr(&self.inner)
249    }
250
251    fn post_init(&self) {
252        let this = self.clone();
253
254        for subquery in &RefCell::borrow(&self.inner).subqueries {
255            RefCell::borrow_mut(&subquery.inner).parent = this.clone().into();
256        }
257
258        for cte in RefCell::borrow(&self.inner).ctes.values().cloned() {
259            RefCell::borrow_mut(&cte.inner).parent = this.clone().into();
260        }
261    }
262}
263
264impl<'me> Query<'me> {
265    pub fn children(&self) -> Vec<Self> {
266        self.inner
267            .borrow()
268            .ctes
269            .values()
270            .chain(self.inner.borrow().subqueries.iter())
271            .cloned()
272            .collect()
273    }
274
275    fn extract_subqueries<'a>(selectable: &Selectable, dialect: &'a Dialect) -> Vec<Query<'a>> {
276        let mut acc = Vec::new();
277
278        for subselect in selectable.selectable.recursive_crawl(
279            &SELECTABLE_TYPES,
280            false,
281            &SyntaxSet::EMPTY,
282            false,
283        ) {
284            acc.push(Query::from_segment(&subselect, dialect, None));
285        }
286
287        acc
288    }
289
290    pub fn from_root<'a>(root_segment: &ErasedSegment, dialect: &'a Dialect) -> Option<Query<'a>> {
291        let stmts = root_segment.recursive_crawl(
292            &SELECTABLE_TYPES,
293            true,
294            &SyntaxSet::single(SyntaxKind::MergeStatement),
295            true,
296        );
297        let selectable_segment = stmts.first()?;
298
299        Some(Query::from_segment(selectable_segment, dialect, None))
300    }
301
302    pub fn from_segment<'a>(
303        segment: &ErasedSegment,
304        dialect: &'a Dialect,
305        parent: Option<Query<'a>>,
306    ) -> Query<'a> {
307        let mut selectables = Vec::new();
308        let mut subqueries = Vec::new();
309        let mut cte_defs: Vec<ErasedSegment> = Vec::new();
310        let mut query_type = QueryType::Simple;
311
312        if segment.is_type(SyntaxKind::SelectStatement)
313            || SUBSELECT_TYPES.contains(segment.get_type())
314        {
315            selectables.push(Selectable {
316                selectable: segment.clone(),
317                dialect,
318            });
319        } else if segment.is_type(SyntaxKind::SetExpression) {
320            selectables.extend(
321                segment
322                    .children(const { &SyntaxSet::new(&[SyntaxKind::SelectStatement]) })
323                    .cloned()
324                    .map(|selectable| Selectable {
325                        selectable,
326                        dialect,
327                    }),
328            )
329        } else {
330            query_type = QueryType::WithCompound;
331
332            for seg in segment.recursive_crawl(
333                const { &SyntaxSet::new(&[SyntaxKind::SelectStatement]) },
334                false,
335                const { &SyntaxSet::single(SyntaxKind::CommonTableExpression) },
336                true,
337            ) {
338                selectables.push(Selectable {
339                    selectable: seg,
340                    dialect,
341                });
342            }
343
344            for seg in segment.recursive_crawl(
345                const { &SyntaxSet::new(&[SyntaxKind::CommonTableExpression]) },
346                false,
347                const { &SyntaxSet::single(SyntaxKind::WithCompoundStatement) },
348                true,
349            ) {
350                cte_defs.push(seg);
351            }
352        }
353
354        for selectable in &selectables {
355            subqueries.extend(Self::extract_subqueries(selectable, dialect));
356        }
357
358        let outer_query = Query {
359            inner: Rc::new(RefCell::new(QueryInner {
360                query_type,
361                dialect,
362                selectables,
363                ctes: <_>::default(),
364                parent,
365                subqueries,
366                cte_definition_segment: None,
367                cte_name_segment: None,
368            })),
369        };
370
371        outer_query.post_init();
372
373        if cte_defs.is_empty() {
374            return outer_query;
375        }
376
377        let mut ctes = IndexMap::default();
378        for cte in cte_defs {
379            let name_seg = cte.segments()[0].clone();
380            let name = name_seg.raw().to_uppercase_smolstr();
381
382            let queries = cte.recursive_crawl(
383                const { &SELECTABLE_TYPES.union(&SUBSELECT_TYPES) },
384                true,
385                &SyntaxSet::EMPTY,
386                true,
387            );
388
389            if queries.is_empty() {
390                continue;
391            };
392
393            let query = &queries[0];
394            let query = Self::from_segment(query, dialect, outer_query.clone().into());
395
396            RefCell::borrow_mut(&query.inner).cte_definition_segment = cte.into();
397            RefCell::borrow_mut(&query.inner).cte_name_segment = name_seg.into();
398
399            ctes.insert(name, query);
400        }
401
402        RefCell::borrow_mut(&outer_query.inner).ctes = ctes;
403        outer_query
404    }
405}
406
407pub enum Source<'a> {
408    TableReference(SmolStr),
409    Query(Query<'a>),
410}