sqruff_lib_core/parser/
segments.rs

1pub mod bracketed;
2pub mod file;
3pub mod fix;
4pub mod from;
5pub mod generator;
6pub mod join;
7pub mod meta;
8pub mod object_reference;
9pub mod select;
10pub mod test_functions;
11
12use std::cell::{Cell, OnceCell};
13use std::fmt::Debug;
14use std::hash::{Hash, Hasher};
15use std::rc::Rc;
16
17use itertools::enumerate;
18use rustc_hash::FxHashMap;
19use smol_str::SmolStr;
20
21use crate::dialects::init::DialectKind;
22use crate::dialects::syntax::{SyntaxKind, SyntaxSet};
23use crate::lint_fix::LintFix;
24use crate::parser::markers::PositionMarker;
25use crate::parser::segments::fix::{FixPatch, SourceFix};
26use crate::parser::segments::object_reference::{ObjectReferenceKind, ObjectReferenceSegment};
27use crate::segments::AnchorEditInfo;
28use crate::templaters::TemplatedFile;
29
30pub struct SegmentBuilder {
31    node_or_token: NodeOrToken,
32}
33
34impl SegmentBuilder {
35    pub fn whitespace(id: u32, raw: &str) -> ErasedSegment {
36        SegmentBuilder::token(id, raw, SyntaxKind::Whitespace).finish()
37    }
38
39    pub fn newline(id: u32, raw: &str) -> ErasedSegment {
40        SegmentBuilder::token(id, raw, SyntaxKind::Newline).finish()
41    }
42
43    pub fn keyword(id: u32, raw: &str) -> ErasedSegment {
44        SegmentBuilder::token(id, raw, SyntaxKind::Keyword).finish()
45    }
46
47    pub fn comma(id: u32) -> ErasedSegment {
48        SegmentBuilder::token(id, ",", SyntaxKind::Comma).finish()
49    }
50
51    pub fn symbol(id: u32, raw: &str) -> ErasedSegment {
52        SegmentBuilder::token(id, raw, SyntaxKind::Symbol).finish()
53    }
54
55    pub fn node(
56        id: u32,
57        syntax_kind: SyntaxKind,
58        dialect: DialectKind,
59        segments: Vec<ErasedSegment>,
60    ) -> Self {
61        SegmentBuilder {
62            node_or_token: NodeOrToken {
63                id,
64                syntax_kind,
65                class_types: class_types(syntax_kind),
66                position_marker: None,
67                code_idx: OnceCell::new(),
68                kind: NodeOrTokenKind::Node(NodeData {
69                    dialect,
70                    segments,
71                    raw: Default::default(),
72                    source_fixes: vec![],
73                    descendant_type_set: Default::default(),
74                    raw_segments_with_ancestors: Default::default(),
75                }),
76                hash: OnceCell::new(),
77            },
78        }
79    }
80
81    pub fn token(id: u32, raw: &str, syntax_kind: SyntaxKind) -> Self {
82        SegmentBuilder {
83            node_or_token: NodeOrToken {
84                id,
85                syntax_kind,
86                code_idx: OnceCell::new(),
87                class_types: class_types(syntax_kind),
88                position_marker: None,
89                kind: NodeOrTokenKind::Token(TokenData { raw: raw.into() }),
90                hash: OnceCell::new(),
91            },
92        }
93    }
94
95    pub fn position_from_segments(mut self) -> Self {
96        let segments = match &self.node_or_token.kind {
97            NodeOrTokenKind::Node(node) => &node.segments[..],
98            NodeOrTokenKind::Token(_) => &[],
99        };
100
101        self.node_or_token.position_marker = pos_marker(segments).into();
102        self
103    }
104
105    pub fn with_position(mut self, position: PositionMarker) -> Self {
106        self.node_or_token.position_marker = Some(position);
107        self
108    }
109
110    pub fn with_source_fixes(mut self, source_fixes: Vec<SourceFix>) -> Self {
111        if let NodeOrTokenKind::Node(ref mut node) = self.node_or_token.kind {
112            node.source_fixes = source_fixes;
113        }
114        self
115    }
116
117    pub fn finish(self) -> ErasedSegment {
118        ErasedSegment {
119            value: Rc::new(self.node_or_token),
120        }
121    }
122}
123
124#[derive(Debug, Default)]
125pub struct Tables {
126    counter: Cell<u32>,
127}
128
129impl Tables {
130    pub fn next_id(&self) -> u32 {
131        let id = self.counter.get();
132        self.counter.set(id + 1);
133        id
134    }
135}
136
137#[derive(Debug, Clone)]
138pub struct ErasedSegment {
139    pub(crate) value: Rc<NodeOrToken>,
140}
141
142impl Hash for ErasedSegment {
143    fn hash<H: Hasher>(&self, state: &mut H) {
144        self.hash_value().hash(state);
145    }
146}
147
148impl Eq for ErasedSegment {}
149
150impl ErasedSegment {
151    pub fn raw(&self) -> &SmolStr {
152        match &self.value.kind {
153            NodeOrTokenKind::Node(node) => node.raw.get_or_init(|| {
154                SmolStr::from_iter(self.segments().iter().map(|segment| segment.raw().as_str()))
155            }),
156            NodeOrTokenKind::Token(token) => &token.raw,
157        }
158    }
159
160    pub fn segments(&self) -> &[ErasedSegment] {
161        match &self.value.kind {
162            NodeOrTokenKind::Node(node) => &node.segments,
163            NodeOrTokenKind::Token(_) => &[],
164        }
165    }
166
167    pub fn get_type(&self) -> SyntaxKind {
168        self.value.syntax_kind
169    }
170
171    pub fn is_type(&self, kind: SyntaxKind) -> bool {
172        self.get_type() == kind
173    }
174
175    pub fn is_meta(&self) -> bool {
176        matches!(
177            self.value.syntax_kind,
178            SyntaxKind::Indent | SyntaxKind::Implicit | SyntaxKind::Dedent | SyntaxKind::EndOfFile
179        )
180    }
181
182    pub fn is_code(&self) -> bool {
183        match &self.value.kind {
184            NodeOrTokenKind::Node(node) => node.segments.iter().any(|s| s.is_code()),
185            NodeOrTokenKind::Token(_) => {
186                !self.is_comment() && !self.is_whitespace() && !self.is_meta()
187            }
188        }
189    }
190
191    pub fn get_raw_segments(&self) -> Vec<ErasedSegment> {
192        self.recursive_crawl_all(false)
193            .into_iter()
194            .filter(|it| it.segments().is_empty())
195            .collect()
196    }
197
198    #[cfg(feature = "stringify")]
199    pub fn stringify(&self, code_only: bool) -> String {
200        serde_yaml::to_string(&self.to_serialised(code_only, true)).unwrap()
201    }
202
203    pub fn child(&self, seg_types: &SyntaxSet) -> Option<ErasedSegment> {
204        self.segments()
205            .iter()
206            .find(|seg| seg_types.contains(seg.get_type()))
207            .cloned()
208    }
209
210    pub fn recursive_crawl(
211        &self,
212        types: &SyntaxSet,
213        recurse_into: bool,
214        no_recursive_types: &SyntaxSet,
215        allow_self: bool,
216    ) -> Vec<ErasedSegment> {
217        let mut acc = Vec::new();
218
219        let matches = allow_self && self.class_types().intersects(types);
220        if matches {
221            acc.push(self.clone());
222        }
223
224        if !self.descendant_type_set().intersects(types) {
225            return acc;
226        }
227
228        if recurse_into || !matches {
229            for seg in self.segments() {
230                if no_recursive_types.is_empty() || !no_recursive_types.contains(seg.get_type()) {
231                    let segments =
232                        seg.recursive_crawl(types, recurse_into, no_recursive_types, true);
233                    acc.extend(segments);
234                }
235            }
236        }
237
238        acc
239    }
240}
241
242impl ErasedSegment {
243    #[track_caller]
244    pub fn new(&self, segments: Vec<ErasedSegment>) -> ErasedSegment {
245        match &self.value.kind {
246            NodeOrTokenKind::Node(node) => {
247                let mut builder = SegmentBuilder::node(
248                    self.value.id,
249                    self.value.syntax_kind,
250                    node.dialect,
251                    segments,
252                )
253                .with_position(self.get_position_marker().unwrap().clone());
254                // Preserve source_fixes during tree rebuilds
255                if !node.source_fixes.is_empty() {
256                    builder = builder.with_source_fixes(node.source_fixes.clone());
257                }
258                builder.finish()
259            }
260            NodeOrTokenKind::Token(_) => self.deep_clone(),
261        }
262    }
263
264    fn change_segments(&self, segments: Vec<ErasedSegment>) -> ErasedSegment {
265        let NodeOrTokenKind::Node(node) = &self.value.kind else {
266            unimplemented!()
267        };
268
269        ErasedSegment {
270            value: Rc::new(NodeOrToken {
271                id: self.value.id,
272                syntax_kind: self.value.syntax_kind,
273                class_types: self.value.class_types.clone(),
274                position_marker: None,
275                code_idx: OnceCell::new(),
276                kind: NodeOrTokenKind::Node(NodeData {
277                    dialect: node.dialect,
278                    segments,
279                    raw: node.raw.clone(),
280                    source_fixes: node.source_fixes.clone(),
281                    descendant_type_set: node.descendant_type_set.clone(),
282                    raw_segments_with_ancestors: node.raw_segments_with_ancestors.clone(),
283                }),
284                hash: OnceCell::new(),
285            }),
286        }
287    }
288
289    pub fn indent_val(&self) -> i8 {
290        self.value.syntax_kind.indent_val()
291    }
292
293    pub fn can_start_end_non_code(&self) -> bool {
294        matches!(
295            self.value.syntax_kind,
296            SyntaxKind::File | SyntaxKind::Unparsable
297        )
298    }
299
300    pub(crate) fn dialect(&self) -> DialectKind {
301        match &self.value.kind {
302            NodeOrTokenKind::Node(node) => node.dialect,
303            NodeOrTokenKind::Token(_) => todo!(),
304        }
305    }
306
307    pub fn get_start_loc(&self) -> (usize, usize) {
308        match self.get_position_marker() {
309            Some(pos_marker) => pos_marker.working_loc(),
310            None => unreachable!("{self:?} has no PositionMarker"),
311        }
312    }
313
314    pub fn get_end_loc(&self) -> (usize, usize) {
315        match self.get_position_marker() {
316            Some(pos_marker) => pos_marker.working_loc_after(self.raw()),
317            None => {
318                unreachable!("{self:?} has no PositionMarker")
319            }
320        }
321    }
322
323    pub fn is_templated(&self) -> bool {
324        if let Some(pos_marker) = self.get_position_marker() {
325            pos_marker.source_slice.start != pos_marker.source_slice.end && !pos_marker.is_literal()
326        } else {
327            panic!("PosMarker must be set");
328        }
329    }
330
331    pub fn iter_segments(&self, expanding: &SyntaxSet, pass_through: bool) -> Vec<ErasedSegment> {
332        let capacity = if expanding.is_empty() {
333            self.segments().len()
334        } else {
335            0
336        };
337        let mut result = Vec::with_capacity(capacity);
338        for segment in self.segments() {
339            if expanding.contains(segment.get_type()) {
340                let expanding = if pass_through {
341                    expanding
342                } else {
343                    &SyntaxSet::EMPTY
344                };
345                result.append(&mut segment.iter_segments(expanding, false));
346            } else {
347                result.push(segment.clone());
348            }
349        }
350        result
351    }
352
353    pub(crate) fn code_indices(&self) -> Rc<Vec<usize>> {
354        self.value
355            .code_idx
356            .get_or_init(|| {
357                Rc::from(
358                    self.segments()
359                        .iter()
360                        .enumerate()
361                        .filter(|(_, seg)| seg.is_code())
362                        .map(|(idx, _)| idx)
363                        .collect::<Vec<_>>(),
364                )
365            })
366            .clone()
367    }
368
369    pub fn children(
370        &self,
371        seg_types: &'static SyntaxSet,
372    ) -> impl Iterator<Item = &ErasedSegment> + '_ {
373        self.segments()
374            .iter()
375            .filter(move |seg| seg_types.contains(seg.get_type()))
376    }
377
378    pub fn iter_patches(&self, templated_file: &TemplatedFile) -> Vec<FixPatch> {
379        let mut acc = Vec::new();
380
381        let templated_raw = &templated_file.templated_str.as_ref().unwrap()
382            [self.get_position_marker().unwrap().templated_slice.clone()];
383
384        // Always collect source fixes from this segment first
385        acc.extend(self.iter_source_fix_patches(templated_file));
386
387        // Check if any descendants have source_fixes
388        let has_descendant_source_fixes = self
389            .recursive_crawl_all(false)
390            .iter()
391            .any(|s| !s.get_source_fixes().is_empty());
392
393        if self.raw() == templated_raw && !has_descendant_source_fixes {
394            // Already collected source_fix_patches above, just return
395            return acc;
396        }
397        // If there are descendant source_fixes, continue to iterate over children
398
399        if self.get_position_marker().is_none() {
400            return Vec::new();
401        }
402
403        let pos_marker = self.get_position_marker().unwrap();
404        if pos_marker.is_literal() && !has_descendant_source_fixes {
405            acc.extend(self.iter_source_fix_patches(templated_file));
406            acc.push(FixPatch::new(
407                pos_marker.templated_slice.clone(),
408                self.raw().clone(),
409                // SyntaxKind::Literal.into(),
410                pos_marker.source_slice.clone(),
411                templated_file.templated_str.as_ref().unwrap()[pos_marker.templated_slice.clone()]
412                    .to_string(),
413                templated_file.source_str[pos_marker.source_slice.clone()].to_string(),
414            ));
415        } else if self.segments().is_empty() {
416            return acc;
417        } else {
418            let mut segments = self.segments();
419
420            while !segments.is_empty()
421                && matches!(
422                    segments.last().unwrap().get_type(),
423                    SyntaxKind::EndOfFile
424                        | SyntaxKind::Indent
425                        | SyntaxKind::Dedent
426                        | SyntaxKind::Implicit
427                )
428            {
429                segments = &segments[..segments.len() - 1];
430            }
431
432            let pos = self.get_position_marker().unwrap();
433            let mut source_idx = pos.source_slice.start;
434            let mut templated_idx = pos.templated_slice.start;
435            let mut insert_buff = String::new();
436
437            for segment in segments {
438                let pos_marker = segment.get_position_marker().unwrap();
439                if !segment.raw().is_empty() && pos_marker.is_point() {
440                    insert_buff.push_str(segment.raw().as_ref());
441                    continue;
442                }
443
444                let start_diff = pos_marker.templated_slice.start - templated_idx;
445
446                if start_diff > 0 || !insert_buff.is_empty() {
447                    let fixed_raw = std::mem::take(&mut insert_buff);
448                    let raw_segments = segment.get_raw_segments();
449                    let first_segment_pos = raw_segments[0].get_position_marker().unwrap();
450
451                    acc.push(FixPatch::new(
452                        templated_idx..first_segment_pos.templated_slice.start,
453                        fixed_raw.into(),
454                        source_idx..first_segment_pos.source_slice.start,
455                        String::new(),
456                        String::new(),
457                    ));
458                }
459
460                acc.extend(segment.iter_patches(templated_file));
461
462                source_idx = pos_marker.source_slice.end;
463                templated_idx = pos_marker.templated_slice.end;
464            }
465
466            let end_diff = pos.templated_slice.end - templated_idx;
467            if end_diff != 0 || !insert_buff.is_empty() {
468                let source_slice = source_idx..pos.source_slice.end;
469                let templated_slice = templated_idx..pos.templated_slice.end;
470
471                let templated_str = templated_file.templated_str.as_ref().unwrap()
472                    [templated_slice.clone()]
473                .to_owned();
474                let source_str = templated_file.source_str[source_slice.clone()].to_owned();
475
476                acc.push(FixPatch::new(
477                    templated_slice,
478                    insert_buff.into(),
479                    source_slice,
480                    templated_str,
481                    source_str,
482                ));
483            }
484        }
485
486        acc
487    }
488
489    pub fn descendant_type_set(&self) -> &SyntaxSet {
490        match &self.value.kind {
491            NodeOrTokenKind::Node(node) => node.descendant_type_set.get_or_init(|| {
492                self.segments()
493                    .iter()
494                    .flat_map(|segment| {
495                        segment
496                            .descendant_type_set()
497                            .clone()
498                            .union(segment.class_types())
499                    })
500                    .collect()
501            }),
502            NodeOrTokenKind::Token(_) => const { &SyntaxSet::EMPTY },
503        }
504    }
505
506    pub fn is_comment(&self) -> bool {
507        matches!(
508            self.value.syntax_kind,
509            SyntaxKind::Comment
510                | SyntaxKind::InlineComment
511                | SyntaxKind::BlockComment
512                | SyntaxKind::NotebookStart
513        )
514    }
515
516    pub fn is_whitespace(&self) -> bool {
517        matches!(
518            self.value.syntax_kind,
519            SyntaxKind::Whitespace | SyntaxKind::Newline
520        )
521    }
522
523    pub fn is_indent(&self) -> bool {
524        matches!(
525            self.value.syntax_kind,
526            SyntaxKind::Indent | SyntaxKind::Implicit | SyntaxKind::Dedent
527        )
528    }
529
530    pub fn get_position_marker(&self) -> Option<&PositionMarker> {
531        self.value.position_marker.as_ref()
532    }
533
534    pub(crate) fn iter_source_fix_patches(&self, templated_file: &TemplatedFile) -> Vec<FixPatch> {
535        let source_fixes = self.get_source_fixes();
536        let mut patches = Vec::with_capacity(source_fixes.len());
537
538        for source_fix in &source_fixes {
539            patches.push(FixPatch::new(
540                source_fix.templated_slice.clone(),
541                source_fix.edit.clone(),
542                // String::from("source"),
543                source_fix.source_slice.clone(),
544                templated_file.templated_str.clone().unwrap()[source_fix.templated_slice.clone()]
545                    .to_string(),
546                templated_file.source_str[source_fix.source_slice.clone()].to_string(),
547            ));
548        }
549
550        patches
551    }
552
553    pub fn id(&self) -> u32 {
554        self.value.id
555    }
556
557    /// Return any source fixes as list.
558    pub fn get_source_fixes(&self) -> Vec<SourceFix> {
559        match &self.value.kind {
560            NodeOrTokenKind::Node(node) => node.source_fixes.clone(),
561            NodeOrTokenKind::Token(_) => Vec::new(),
562        }
563    }
564
565    /// Return all source fixes from this segment and all its descendants.
566    pub fn get_all_source_fixes(&self) -> Vec<SourceFix> {
567        let mut fixes = self.get_source_fixes();
568        for segment in self.segments() {
569            fixes.extend(segment.get_all_source_fixes());
570        }
571        fixes
572    }
573
574    pub fn edit(
575        &self,
576        id: u32,
577        raw: Option<String>,
578        _source_fixes: Option<Vec<SourceFix>>,
579    ) -> ErasedSegment {
580        match &self.value.kind {
581            NodeOrTokenKind::Node(_node) => {
582                todo!()
583            }
584            NodeOrTokenKind::Token(token) => {
585                let raw = raw.as_deref().unwrap_or(token.raw.as_ref());
586                SegmentBuilder::token(id, raw, self.value.syntax_kind)
587                    .with_position(self.get_position_marker().unwrap().clone())
588                    .finish()
589            }
590        }
591    }
592
593    pub fn class_types(&self) -> &SyntaxSet {
594        &self.value.class_types
595    }
596
597    pub(crate) fn first_non_whitespace_segment_raw_upper(&self) -> Option<String> {
598        for seg in self.get_raw_segments() {
599            if !seg.raw().is_empty() {
600                return Some(seg.raw().to_uppercase());
601            }
602        }
603        None
604    }
605
606    pub fn is(&self, other: &ErasedSegment) -> bool {
607        Rc::ptr_eq(&self.value, &other.value)
608    }
609
610    pub fn addr(&self) -> usize {
611        Rc::as_ptr(&self.value).addr()
612    }
613
614    pub fn direct_descendant_type_set(&self) -> SyntaxSet {
615        self.segments()
616            .iter()
617            .fold(SyntaxSet::EMPTY, |set, it| set.union(it.class_types()))
618    }
619
620    pub fn is_keyword(&self, p0: &str) -> bool {
621        self.is_type(SyntaxKind::Keyword) && self.raw().eq_ignore_ascii_case(p0)
622    }
623
624    pub fn hash_value(&self) -> u64 {
625        *self.value.hash.get_or_init(|| {
626            let mut hasher = ahash::AHasher::default();
627            self.get_type().hash(&mut hasher);
628            self.raw().hash(&mut hasher);
629
630            if let Some(marker) = &self.get_position_marker() {
631                marker.source_position().hash(&mut hasher);
632            } else {
633                None::<usize>.hash(&mut hasher);
634            }
635
636            hasher.finish()
637        })
638    }
639
640    pub fn deep_clone(&self) -> Self {
641        Self {
642            value: Rc::new(self.value.as_ref().clone()),
643        }
644    }
645
646    #[track_caller]
647    pub(crate) fn get_mut(&mut self) -> &mut NodeOrToken {
648        Rc::get_mut(&mut self.value).unwrap()
649    }
650
651    #[track_caller]
652    pub(crate) fn make_mut(&mut self) -> &mut NodeOrToken {
653        Rc::make_mut(&mut self.value)
654    }
655
656    pub fn reference(&self) -> ObjectReferenceSegment {
657        ObjectReferenceSegment(
658            self.clone(),
659            match self.get_type() {
660                SyntaxKind::TableReference => ObjectReferenceKind::Table,
661                SyntaxKind::WildcardIdentifier => ObjectReferenceKind::WildcardIdentifier,
662                _ => ObjectReferenceKind::Object,
663            },
664        )
665    }
666
667    pub fn recursive_crawl_all(&self, reverse: bool) -> Vec<ErasedSegment> {
668        let mut result = Vec::with_capacity(self.segments().len() + 1);
669
670        if reverse {
671            for seg in self.segments().iter().rev() {
672                result.append(&mut seg.recursive_crawl_all(reverse));
673            }
674            result.push(self.clone());
675        } else {
676            result.push(self.clone());
677            for seg in self.segments() {
678                result.append(&mut seg.recursive_crawl_all(reverse));
679            }
680        }
681
682        result
683    }
684
685    pub fn raw_segments_with_ancestors(&self) -> &[(ErasedSegment, Vec<PathStep>)] {
686        match &self.value.kind {
687            NodeOrTokenKind::Node(node) => node.raw_segments_with_ancestors.get_or_init(|| {
688                let mut buffer: Vec<(ErasedSegment, Vec<PathStep>)> =
689                    Vec::with_capacity(self.segments().len());
690                let code_idxs = self.code_indices();
691
692                for (idx, seg) in self.segments().iter().enumerate() {
693                    let new_step = vec![PathStep {
694                        segment: self.clone(),
695                        idx,
696                        len: self.segments().len(),
697                        code_idxs: code_idxs.clone(),
698                    }];
699
700                    // Use seg.get_segments().is_empty() as a workaround to check if the segment is
701                    // a SyntaxKind::Raw type. In the original Python code, this was achieved
702                    // using seg.is_type(SyntaxKind::Raw). Here, we assume that a SyntaxKind::Raw
703                    // segment is characterized by having no sub-segments.
704
705                    if seg.segments().is_empty() {
706                        buffer.push((seg.clone(), new_step));
707                    } else {
708                        let extended =
709                            seg.raw_segments_with_ancestors()
710                                .iter()
711                                .map(|(raw_seg, stack)| {
712                                    let mut new_step = new_step.clone();
713                                    new_step.extend_from_slice(stack);
714                                    (raw_seg.clone(), new_step)
715                                });
716
717                        buffer.extend(extended);
718                    }
719                }
720
721                buffer
722            }),
723            NodeOrTokenKind::Token(_) => &[],
724        }
725    }
726
727    pub fn path_to(&self, other: &ErasedSegment) -> Vec<PathStep> {
728        let midpoint = other;
729
730        for (idx, seg) in enumerate(self.segments()) {
731            let mut steps = vec![PathStep {
732                segment: self.clone(),
733                idx,
734                len: self.segments().len(),
735                code_idxs: self.code_indices(),
736            }];
737
738            if seg.eq(midpoint) {
739                return steps;
740            }
741
742            let res = seg.path_to(midpoint);
743
744            if !res.is_empty() {
745                steps.extend(res);
746                return steps;
747            }
748        }
749
750        Vec::new()
751    }
752
753    pub fn apply_fixes(
754        &self,
755        fixes: &mut FxHashMap<u32, AnchorEditInfo>,
756    ) -> (ErasedSegment, Vec<ErasedSegment>, Vec<ErasedSegment>) {
757        if fixes.is_empty() || self.segments().is_empty() {
758            return (self.clone(), Vec::new(), Vec::new());
759        }
760
761        let mut seg_buffer = Vec::new();
762        let mut has_applied_fixes = false;
763        let mut _requires_validate = false;
764
765        for seg in self.segments() {
766            // Look for uuid match.
767            // This handles potential positioning ambiguity.
768
769            let Some(mut anchor_info) = fixes.remove(&seg.id()) else {
770                seg_buffer.push(seg.clone());
771                continue;
772            };
773
774            if anchor_info.fixes.len() == 2
775                && matches!(anchor_info.fixes[0], LintFix::CreateAfter { .. })
776            {
777                anchor_info.fixes.reverse();
778            }
779
780            let fixes_count = anchor_info.fixes.len();
781            for lint_fix in anchor_info.fixes {
782                has_applied_fixes = true;
783
784                // Deletes are easy.
785                if matches!(lint_fix, LintFix::Delete { .. }) {
786                    // We're just getting rid of this segment.
787                    _requires_validate = true;
788                    // NOTE: We don't add the segment in this case.
789                    continue;
790                }
791
792                // Otherwise it must be a replace or a create.
793                assert!(matches!(
794                    lint_fix,
795                    LintFix::Replace { .. }
796                        | LintFix::CreateBefore { .. }
797                        | LintFix::CreateAfter { .. }
798                ));
799
800                match lint_fix {
801                    LintFix::CreateAfter { edit, .. } => {
802                        if fixes_count == 1 {
803                            // In the case of a creation after that is not part
804                            // of a create_before/create_after pair, also add
805                            // this segment before the edit.
806                            seg_buffer.push(seg.clone());
807                        }
808                        for s in edit {
809                            seg_buffer.push(s);
810                        }
811                        _requires_validate = true;
812                    }
813                    LintFix::CreateBefore { edit, .. } => {
814                        for s in edit {
815                            seg_buffer.push(s);
816                        }
817                        seg_buffer.push(seg.clone());
818                        _requires_validate = true;
819                    }
820                    LintFix::Replace { edit, .. } => {
821                        let mut consumed_pos = false;
822                        let is_single_same_type =
823                            edit.len() == 1 && edit[0].class_types() == seg.class_types();
824
825                        for mut s in edit {
826                            if !consumed_pos && s.raw() == seg.raw() {
827                                consumed_pos = true;
828                                s.make_mut()
829                                    .set_position_marker(seg.get_position_marker().cloned());
830                            }
831                            seg_buffer.push(s);
832                        }
833
834                        if !is_single_same_type {
835                            _requires_validate = true;
836                        }
837                    }
838                    LintFix::Delete { .. } => {
839                        // Already handled above
840                        unreachable!()
841                    }
842                }
843            }
844        }
845
846        if has_applied_fixes {
847            seg_buffer =
848                position_segments(&seg_buffer, self.get_position_marker().as_ref().unwrap());
849        }
850
851        let seg_queue = seg_buffer;
852        let mut seg_buffer = Vec::new();
853        for seg in seg_queue {
854            let (mid, pre, post) = seg.apply_fixes(fixes);
855
856            seg_buffer.extend(pre);
857            seg_buffer.push(mid);
858            seg_buffer.extend(post);
859        }
860
861        let seg_buffer =
862            position_segments(&seg_buffer, self.get_position_marker().as_ref().unwrap());
863        (self.new(seg_buffer), Vec::new(), Vec::new())
864    }
865}
866
867#[cfg(any(test, feature = "serde"))]
868pub mod serde {
869    use serde::ser::SerializeMap;
870    use serde::{Deserialize, Serialize};
871
872    use crate::parser::segments::ErasedSegment;
873
874    #[derive(Serialize, Deserialize)]
875    #[serde(untagged)]
876    pub enum SerialisedSegmentValue {
877        Single(String),
878        Nested(Vec<TupleSerialisedSegment>),
879    }
880
881    #[derive(Deserialize)]
882    pub struct TupleSerialisedSegment(String, SerialisedSegmentValue);
883
884    impl Serialize for TupleSerialisedSegment {
885        fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
886        where
887            S: serde::Serializer,
888        {
889            let mut map = serializer.serialize_map(None)?;
890            map.serialize_key(&self.0)?;
891            map.serialize_value(&self.1)?;
892            map.end()
893        }
894    }
895
896    impl TupleSerialisedSegment {
897        pub fn sinlge(key: String, value: String) -> Self {
898            Self(key, SerialisedSegmentValue::Single(value))
899        }
900
901        pub fn nested(key: String, segments: Vec<TupleSerialisedSegment>) -> Self {
902            Self(key, SerialisedSegmentValue::Nested(segments))
903        }
904    }
905
906    impl ErasedSegment {
907        pub fn to_serialised(&self, code_only: bool, show_raw: bool) -> TupleSerialisedSegment {
908            if show_raw && self.segments().is_empty() {
909                TupleSerialisedSegment::sinlge(
910                    self.get_type().as_str().to_string(),
911                    self.raw().to_string(),
912                )
913            } else if code_only {
914                let segments = self
915                    .segments()
916                    .iter()
917                    .filter(|seg| seg.is_code() && !seg.is_meta())
918                    .map(|seg| seg.to_serialised(code_only, show_raw))
919                    .collect::<Vec<_>>();
920
921                TupleSerialisedSegment::nested(self.get_type().as_str().to_string(), segments)
922            } else {
923                let segments = self
924                    .segments()
925                    .iter()
926                    .map(|seg| seg.to_serialised(code_only, show_raw))
927                    .collect::<Vec<_>>();
928
929                TupleSerialisedSegment::nested(self.get_type().as_str().to_string(), segments)
930            }
931        }
932    }
933}
934
935impl PartialEq for ErasedSegment {
936    fn eq(&self, other: &Self) -> bool {
937        if self.id() == other.id() {
938            return true;
939        }
940
941        let pos_self = self.get_position_marker();
942        let pos_other = other.get_position_marker();
943        if let Some((pos_self, pos_other)) = pos_self.zip(pos_other) {
944            self.get_type() == other.get_type()
945                && pos_self.working_loc() == pos_other.working_loc()
946                && self.raw() == other.raw()
947        } else {
948            false
949        }
950    }
951}
952
953pub fn position_segments(
954    segments: &[ErasedSegment],
955    parent_pos: &PositionMarker,
956) -> Vec<ErasedSegment> {
957    if segments.is_empty() {
958        return Vec::new();
959    }
960
961    let (mut line_no, mut line_pos) = { (parent_pos.working_line_no, parent_pos.working_line_pos) };
962
963    let mut segment_buffer: Vec<ErasedSegment> = Vec::new();
964    for (idx, segment) in enumerate(segments) {
965        let old_position = segment.get_position_marker();
966
967        let mut new_position = match old_position {
968            Some(pos_marker) => pos_marker.clone(),
969            None => {
970                let start_point = if idx > 0 {
971                    let prev_seg = segment_buffer[idx - 1].clone();
972                    Some(prev_seg.get_position_marker().unwrap().end_point_marker())
973                } else {
974                    Some(parent_pos.start_point_marker())
975                };
976
977                let mut end_point = None;
978                for fwd_seg in &segments[idx + 1..] {
979                    if fwd_seg.get_position_marker().is_some() {
980                        end_point = Some(
981                            fwd_seg.get_raw_segments()[0]
982                                .get_position_marker()
983                                .unwrap()
984                                .start_point_marker(),
985                        );
986                        break;
987                    }
988                }
989
990                if let Some((start_point, end_point)) = start_point
991                    .as_ref()
992                    .zip(end_point.as_ref())
993                    .filter(|(start_point, end_point)| start_point != end_point)
994                {
995                    PositionMarker::from_points(start_point, end_point)
996                } else if let Some(start_point) = start_point.as_ref() {
997                    start_point.clone()
998                } else if let Some(end_point) = end_point.as_ref() {
999                    end_point.clone()
1000                } else {
1001                    unimplemented!("Unable to position new segment")
1002                }
1003            }
1004        };
1005
1006        new_position = new_position.with_working_position(line_no, line_pos);
1007        (line_no, line_pos) = PositionMarker::infer_next_position(segment.raw(), line_no, line_pos);
1008
1009        let mut new_seg = if !segment.segments().is_empty() && old_position != Some(&new_position) {
1010            let child_segments = position_segments(segment.segments(), &new_position);
1011            segment.change_segments(child_segments)
1012        } else {
1013            segment.deep_clone()
1014        };
1015
1016        new_seg.get_mut().set_position_marker(new_position.into());
1017        segment_buffer.push(new_seg);
1018    }
1019
1020    segment_buffer
1021}
1022
1023#[derive(Debug, Clone)]
1024pub struct NodeOrToken {
1025    id: u32,
1026    syntax_kind: SyntaxKind,
1027    class_types: SyntaxSet,
1028    position_marker: Option<PositionMarker>,
1029    kind: NodeOrTokenKind,
1030    code_idx: OnceCell<Rc<Vec<usize>>>,
1031    hash: OnceCell<u64>,
1032}
1033
1034#[derive(Debug, Clone)]
1035#[allow(clippy::large_enum_variant)]
1036pub enum NodeOrTokenKind {
1037    Node(NodeData),
1038    Token(TokenData),
1039}
1040
1041impl NodeOrToken {
1042    pub fn set_position_marker(&mut self, position_marker: Option<PositionMarker>) {
1043        self.position_marker = position_marker;
1044    }
1045
1046    pub fn set_id(&mut self, id: u32) {
1047        self.id = id;
1048    }
1049}
1050
1051#[derive(Debug, Clone)]
1052pub struct NodeData {
1053    dialect: DialectKind,
1054    segments: Vec<ErasedSegment>,
1055    raw: OnceCell<SmolStr>,
1056    source_fixes: Vec<SourceFix>,
1057    descendant_type_set: OnceCell<SyntaxSet>,
1058    raw_segments_with_ancestors: OnceCell<Vec<(ErasedSegment, Vec<PathStep>)>>,
1059}
1060
1061#[derive(Debug, Clone, PartialEq)]
1062pub struct TokenData {
1063    raw: SmolStr,
1064}
1065
1066#[track_caller]
1067pub fn pos_marker(segments: &[ErasedSegment]) -> PositionMarker {
1068    let markers = segments.iter().filter_map(|seg| seg.get_position_marker());
1069
1070    PositionMarker::from_child_markers(markers)
1071}
1072
1073#[derive(Debug, Clone)]
1074pub struct PathStep {
1075    pub segment: ErasedSegment,
1076    pub idx: usize,
1077    pub len: usize,
1078    pub code_idxs: Rc<Vec<usize>>,
1079}
1080
1081fn class_types(syntax_kind: SyntaxKind) -> SyntaxSet {
1082    match syntax_kind {
1083        SyntaxKind::ColumnReference => SyntaxSet::new(&[SyntaxKind::ObjectReference, syntax_kind]),
1084        SyntaxKind::WildcardIdentifier => {
1085            SyntaxSet::new(&[SyntaxKind::WildcardIdentifier, SyntaxKind::ObjectReference])
1086        }
1087        SyntaxKind::TableReference => SyntaxSet::new(&[SyntaxKind::ObjectReference, syntax_kind]),
1088        _ => SyntaxSet::single(syntax_kind),
1089    }
1090}
1091
1092#[cfg(test)]
1093mod tests {
1094    use super::*;
1095    use crate::lint_fix::LintFix;
1096    use crate::linter::compute_anchor_edit_info;
1097    use crate::parser::segments::test_functions::{raw_seg, raw_segments};
1098
1099    #[test]
1100    /// Test comparison of raw segments.
1101    fn test_parser_base_segments_raw_compare() {
1102        let template: TemplatedFile = "foobar".into();
1103        let rs1 = SegmentBuilder::token(0, "foobar", SyntaxKind::Word)
1104            .with_position(PositionMarker::new(
1105                0..6,
1106                0..6,
1107                template.clone(),
1108                None,
1109                None,
1110            ))
1111            .finish();
1112        let rs2 = SegmentBuilder::token(0, "foobar", SyntaxKind::Word)
1113            .with_position(PositionMarker::new(
1114                0..6,
1115                0..6,
1116                template.clone(),
1117                None,
1118                None,
1119            ))
1120            .finish();
1121
1122        assert_eq!(rs1, rs2)
1123    }
1124
1125    #[test]
1126    // TODO Implement
1127    /// Test raw segments behave as expected.
1128    fn test_parser_base_segments_raw() {
1129        let raw_seg = raw_seg();
1130
1131        assert_eq!(raw_seg.raw(), "foobar");
1132    }
1133
1134    #[test]
1135    /// Test BaseSegment.compute_anchor_edit_info().
1136    fn test_parser_base_segments_compute_anchor_edit_info() {
1137        let raw_segs = raw_segments();
1138        let tables = Tables::default();
1139
1140        // Construct a fix buffer, intentionally with:
1141        // - one duplicate.
1142        // - two different incompatible fixes on the same segment.
1143        let fixes = vec![
1144            LintFix::replace(
1145                raw_segs[0].clone(),
1146                vec![raw_segs[0].edit(tables.next_id(), Some("a".to_string()), None)],
1147                None,
1148            ),
1149            LintFix::replace(
1150                raw_segs[0].clone(),
1151                vec![raw_segs[0].edit(tables.next_id(), Some("a".to_string()), None)],
1152                None,
1153            ),
1154            LintFix::replace(
1155                raw_segs[0].clone(),
1156                vec![raw_segs[0].edit(tables.next_id(), Some("b".to_string()), None)],
1157                None,
1158            ),
1159        ];
1160
1161        let mut anchor_edit_info = Default::default();
1162        compute_anchor_edit_info(&mut anchor_edit_info, fixes);
1163
1164        // Check the target segment is the only key we have.
1165        assert_eq!(
1166            anchor_edit_info.keys().collect::<Vec<_>>(),
1167            vec![&raw_segs[0].id()]
1168        );
1169
1170        let anchor_info = anchor_edit_info.get(&raw_segs[0].id()).unwrap();
1171
1172        // Check that the duplicate as been deduplicated i.e. this isn't 3.
1173        assert_eq!(anchor_info.replace, 2);
1174
1175        // Check the fixes themselves.
1176        //   Note: There's no duplicated first fix.
1177        assert_eq!(
1178            anchor_info.fixes[0],
1179            LintFix::replace(
1180                raw_segs[0].clone(),
1181                vec![raw_segs[0].edit(tables.next_id(), Some("a".to_string()), None)],
1182                None,
1183            )
1184        );
1185        assert_eq!(
1186            anchor_info.fixes[1],
1187            LintFix::replace(
1188                raw_segs[0].clone(),
1189                vec![raw_segs[0].edit(tables.next_id(), Some("b".to_string()), None)],
1190                None,
1191            )
1192        );
1193
1194        // Check the first replace
1195        assert_eq!(
1196            anchor_info.fixes[anchor_info.first_replace.unwrap()],
1197            LintFix::replace(
1198                raw_segs[0].clone(),
1199                vec![raw_segs[0].edit(tables.next_id(), Some("a".to_string()), None)],
1200                None,
1201            )
1202        );
1203    }
1204}
sqruff_lib_core/parser/segments.rs

sqruff_lib_core/parser/
segments.rs
