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 | SyntaxKind::InlineComment | SyntaxKind::BlockComment
510        )
511    }
512
513    pub fn is_whitespace(&self) -> bool {
514        matches!(
515            self.value.syntax_kind,
516            SyntaxKind::Whitespace | SyntaxKind::Newline
517        )
518    }
519
520    pub fn is_indent(&self) -> bool {
521        matches!(
522            self.value.syntax_kind,
523            SyntaxKind::Indent | SyntaxKind::Implicit | SyntaxKind::Dedent
524        )
525    }
526
527    pub fn get_position_marker(&self) -> Option<&PositionMarker> {
528        self.value.position_marker.as_ref()
529    }
530
531    pub(crate) fn iter_source_fix_patches(&self, templated_file: &TemplatedFile) -> Vec<FixPatch> {
532        let source_fixes = self.get_source_fixes();
533        let mut patches = Vec::with_capacity(source_fixes.len());
534
535        for source_fix in &source_fixes {
536            patches.push(FixPatch::new(
537                source_fix.templated_slice.clone(),
538                source_fix.edit.clone(),
539                // String::from("source"),
540                source_fix.source_slice.clone(),
541                templated_file.templated_str.clone().unwrap()[source_fix.templated_slice.clone()]
542                    .to_string(),
543                templated_file.source_str[source_fix.source_slice.clone()].to_string(),
544            ));
545        }
546
547        patches
548    }
549
550    pub fn id(&self) -> u32 {
551        self.value.id
552    }
553
554    /// Return any source fixes as list.
555    pub fn get_source_fixes(&self) -> Vec<SourceFix> {
556        match &self.value.kind {
557            NodeOrTokenKind::Node(node) => node.source_fixes.clone(),
558            NodeOrTokenKind::Token(_) => Vec::new(),
559        }
560    }
561
562    /// Return all source fixes from this segment and all its descendants.
563    pub fn get_all_source_fixes(&self) -> Vec<SourceFix> {
564        let mut fixes = self.get_source_fixes();
565        for segment in self.segments() {
566            fixes.extend(segment.get_all_source_fixes());
567        }
568        fixes
569    }
570
571    pub fn edit(
572        &self,
573        id: u32,
574        raw: Option<String>,
575        _source_fixes: Option<Vec<SourceFix>>,
576    ) -> ErasedSegment {
577        match &self.value.kind {
578            NodeOrTokenKind::Node(_node) => {
579                todo!()
580            }
581            NodeOrTokenKind::Token(token) => {
582                let raw = raw.as_deref().unwrap_or(token.raw.as_ref());
583                SegmentBuilder::token(id, raw, self.value.syntax_kind)
584                    .with_position(self.get_position_marker().unwrap().clone())
585                    .finish()
586            }
587        }
588    }
589
590    pub fn class_types(&self) -> &SyntaxSet {
591        &self.value.class_types
592    }
593
594    pub(crate) fn first_non_whitespace_segment_raw_upper(&self) -> Option<String> {
595        for seg in self.get_raw_segments() {
596            if !seg.raw().is_empty() {
597                return Some(seg.raw().to_uppercase());
598            }
599        }
600        None
601    }
602
603    pub fn is(&self, other: &ErasedSegment) -> bool {
604        Rc::ptr_eq(&self.value, &other.value)
605    }
606
607    pub fn addr(&self) -> usize {
608        Rc::as_ptr(&self.value).addr()
609    }
610
611    pub fn direct_descendant_type_set(&self) -> SyntaxSet {
612        self.segments()
613            .iter()
614            .fold(SyntaxSet::EMPTY, |set, it| set.union(it.class_types()))
615    }
616
617    pub fn is_keyword(&self, p0: &str) -> bool {
618        self.is_type(SyntaxKind::Keyword) && self.raw().eq_ignore_ascii_case(p0)
619    }
620
621    pub fn hash_value(&self) -> u64 {
622        *self.value.hash.get_or_init(|| {
623            let mut hasher = ahash::AHasher::default();
624            self.get_type().hash(&mut hasher);
625            self.raw().hash(&mut hasher);
626
627            if let Some(marker) = &self.get_position_marker() {
628                marker.source_position().hash(&mut hasher);
629            } else {
630                None::<usize>.hash(&mut hasher);
631            }
632
633            hasher.finish()
634        })
635    }
636
637    pub fn deep_clone(&self) -> Self {
638        Self {
639            value: Rc::new(self.value.as_ref().clone()),
640        }
641    }
642
643    #[track_caller]
644    pub(crate) fn get_mut(&mut self) -> &mut NodeOrToken {
645        Rc::get_mut(&mut self.value).unwrap()
646    }
647
648    #[track_caller]
649    pub(crate) fn make_mut(&mut self) -> &mut NodeOrToken {
650        Rc::make_mut(&mut self.value)
651    }
652
653    pub fn reference(&self) -> ObjectReferenceSegment {
654        ObjectReferenceSegment(
655            self.clone(),
656            match self.get_type() {
657                SyntaxKind::TableReference => ObjectReferenceKind::Table,
658                SyntaxKind::WildcardIdentifier => ObjectReferenceKind::WildcardIdentifier,
659                _ => ObjectReferenceKind::Object,
660            },
661        )
662    }
663
664    pub fn recursive_crawl_all(&self, reverse: bool) -> Vec<ErasedSegment> {
665        let mut result = Vec::with_capacity(self.segments().len() + 1);
666
667        if reverse {
668            for seg in self.segments().iter().rev() {
669                result.append(&mut seg.recursive_crawl_all(reverse));
670            }
671            result.push(self.clone());
672        } else {
673            result.push(self.clone());
674            for seg in self.segments() {
675                result.append(&mut seg.recursive_crawl_all(reverse));
676            }
677        }
678
679        result
680    }
681
682    pub fn raw_segments_with_ancestors(&self) -> &[(ErasedSegment, Vec<PathStep>)] {
683        match &self.value.kind {
684            NodeOrTokenKind::Node(node) => node.raw_segments_with_ancestors.get_or_init(|| {
685                let mut buffer: Vec<(ErasedSegment, Vec<PathStep>)> =
686                    Vec::with_capacity(self.segments().len());
687                let code_idxs = self.code_indices();
688
689                for (idx, seg) in self.segments().iter().enumerate() {
690                    let new_step = vec![PathStep {
691                        segment: self.clone(),
692                        idx,
693                        len: self.segments().len(),
694                        code_idxs: code_idxs.clone(),
695                    }];
696
697                    // Use seg.get_segments().is_empty() as a workaround to check if the segment is
698                    // a SyntaxKind::Raw type. In the original Python code, this was achieved
699                    // using seg.is_type(SyntaxKind::Raw). Here, we assume that a SyntaxKind::Raw
700                    // segment is characterized by having no sub-segments.
701
702                    if seg.segments().is_empty() {
703                        buffer.push((seg.clone(), new_step));
704                    } else {
705                        let extended =
706                            seg.raw_segments_with_ancestors()
707                                .iter()
708                                .map(|(raw_seg, stack)| {
709                                    let mut new_step = new_step.clone();
710                                    new_step.extend_from_slice(stack);
711                                    (raw_seg.clone(), new_step)
712                                });
713
714                        buffer.extend(extended);
715                    }
716                }
717
718                buffer
719            }),
720            NodeOrTokenKind::Token(_) => &[],
721        }
722    }
723
724    pub fn path_to(&self, other: &ErasedSegment) -> Vec<PathStep> {
725        let midpoint = other;
726
727        for (idx, seg) in enumerate(self.segments()) {
728            let mut steps = vec![PathStep {
729                segment: self.clone(),
730                idx,
731                len: self.segments().len(),
732                code_idxs: self.code_indices(),
733            }];
734
735            if seg.eq(midpoint) {
736                return steps;
737            }
738
739            let res = seg.path_to(midpoint);
740
741            if !res.is_empty() {
742                steps.extend(res);
743                return steps;
744            }
745        }
746
747        Vec::new()
748    }
749
750    pub fn apply_fixes(
751        &self,
752        fixes: &mut FxHashMap<u32, AnchorEditInfo>,
753    ) -> (ErasedSegment, Vec<ErasedSegment>, Vec<ErasedSegment>) {
754        if fixes.is_empty() || self.segments().is_empty() {
755            return (self.clone(), Vec::new(), Vec::new());
756        }
757
758        let mut seg_buffer = Vec::new();
759        let mut has_applied_fixes = false;
760        let mut _requires_validate = false;
761
762        for seg in self.segments() {
763            // Look for uuid match.
764            // This handles potential positioning ambiguity.
765
766            let Some(mut anchor_info) = fixes.remove(&seg.id()) else {
767                seg_buffer.push(seg.clone());
768                continue;
769            };
770
771            if anchor_info.fixes.len() == 2
772                && matches!(anchor_info.fixes[0], LintFix::CreateAfter { .. })
773            {
774                anchor_info.fixes.reverse();
775            }
776
777            let fixes_count = anchor_info.fixes.len();
778            for lint_fix in anchor_info.fixes {
779                has_applied_fixes = true;
780
781                // Deletes are easy.
782                if matches!(lint_fix, LintFix::Delete { .. }) {
783                    // We're just getting rid of this segment.
784                    _requires_validate = true;
785                    // NOTE: We don't add the segment in this case.
786                    continue;
787                }
788
789                // Otherwise it must be a replace or a create.
790                assert!(matches!(
791                    lint_fix,
792                    LintFix::Replace { .. }
793                        | LintFix::CreateBefore { .. }
794                        | LintFix::CreateAfter { .. }
795                ));
796
797                match lint_fix {
798                    LintFix::CreateAfter { edit, .. } => {
799                        if fixes_count == 1 {
800                            // In the case of a creation after that is not part
801                            // of a create_before/create_after pair, also add
802                            // this segment before the edit.
803                            seg_buffer.push(seg.clone());
804                        }
805                        for s in edit {
806                            seg_buffer.push(s);
807                        }
808                        _requires_validate = true;
809                    }
810                    LintFix::CreateBefore { edit, .. } => {
811                        for s in edit {
812                            seg_buffer.push(s);
813                        }
814                        seg_buffer.push(seg.clone());
815                        _requires_validate = true;
816                    }
817                    LintFix::Replace { edit, .. } => {
818                        let mut consumed_pos = false;
819                        let is_single_same_type =
820                            edit.len() == 1 && edit[0].class_types() == seg.class_types();
821
822                        for mut s in edit {
823                            if !consumed_pos && s.raw() == seg.raw() {
824                                consumed_pos = true;
825                                s.make_mut()
826                                    .set_position_marker(seg.get_position_marker().cloned());
827                            }
828                            seg_buffer.push(s);
829                        }
830
831                        if !is_single_same_type {
832                            _requires_validate = true;
833                        }
834                    }
835                    LintFix::Delete { .. } => {
836                        // Already handled above
837                        unreachable!()
838                    }
839                }
840            }
841        }
842
843        if has_applied_fixes {
844            seg_buffer =
845                position_segments(&seg_buffer, self.get_position_marker().as_ref().unwrap());
846        }
847
848        let seg_queue = seg_buffer;
849        let mut seg_buffer = Vec::new();
850        for seg in seg_queue {
851            let (mid, pre, post) = seg.apply_fixes(fixes);
852
853            seg_buffer.extend(pre);
854            seg_buffer.push(mid);
855            seg_buffer.extend(post);
856        }
857
858        let seg_buffer =
859            position_segments(&seg_buffer, self.get_position_marker().as_ref().unwrap());
860        (self.new(seg_buffer), Vec::new(), Vec::new())
861    }
862}
863
864#[cfg(any(test, feature = "serde"))]
865pub mod serde {
866    use serde::ser::SerializeMap;
867    use serde::{Deserialize, Serialize};
868
869    use crate::parser::segments::ErasedSegment;
870
871    #[derive(Serialize, Deserialize)]
872    #[serde(untagged)]
873    pub enum SerialisedSegmentValue {
874        Single(String),
875        Nested(Vec<TupleSerialisedSegment>),
876    }
877
878    #[derive(Deserialize)]
879    pub struct TupleSerialisedSegment(String, SerialisedSegmentValue);
880
881    impl Serialize for TupleSerialisedSegment {
882        fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
883        where
884            S: serde::Serializer,
885        {
886            let mut map = serializer.serialize_map(None)?;
887            map.serialize_key(&self.0)?;
888            map.serialize_value(&self.1)?;
889            map.end()
890        }
891    }
892
893    impl TupleSerialisedSegment {
894        pub fn sinlge(key: String, value: String) -> Self {
895            Self(key, SerialisedSegmentValue::Single(value))
896        }
897
898        pub fn nested(key: String, segments: Vec<TupleSerialisedSegment>) -> Self {
899            Self(key, SerialisedSegmentValue::Nested(segments))
900        }
901    }
902
903    impl ErasedSegment {
904        pub fn to_serialised(&self, code_only: bool, show_raw: bool) -> TupleSerialisedSegment {
905            if show_raw && self.segments().is_empty() {
906                TupleSerialisedSegment::sinlge(
907                    self.get_type().as_str().to_string(),
908                    self.raw().to_string(),
909                )
910            } else if code_only {
911                let segments = self
912                    .segments()
913                    .iter()
914                    .filter(|seg| seg.is_code() && !seg.is_meta())
915                    .map(|seg| seg.to_serialised(code_only, show_raw))
916                    .collect::<Vec<_>>();
917
918                TupleSerialisedSegment::nested(self.get_type().as_str().to_string(), segments)
919            } else {
920                let segments = self
921                    .segments()
922                    .iter()
923                    .map(|seg| seg.to_serialised(code_only, show_raw))
924                    .collect::<Vec<_>>();
925
926                TupleSerialisedSegment::nested(self.get_type().as_str().to_string(), segments)
927            }
928        }
929    }
930}
931
932impl PartialEq for ErasedSegment {
933    fn eq(&self, other: &Self) -> bool {
934        if self.id() == other.id() {
935            return true;
936        }
937
938        let pos_self = self.get_position_marker();
939        let pos_other = other.get_position_marker();
940        if let Some((pos_self, pos_other)) = pos_self.zip(pos_other) {
941            self.get_type() == other.get_type()
942                && pos_self.working_loc() == pos_other.working_loc()
943                && self.raw() == other.raw()
944        } else {
945            false
946        }
947    }
948}
949
950pub fn position_segments(
951    segments: &[ErasedSegment],
952    parent_pos: &PositionMarker,
953) -> Vec<ErasedSegment> {
954    if segments.is_empty() {
955        return Vec::new();
956    }
957
958    let (mut line_no, mut line_pos) = { (parent_pos.working_line_no, parent_pos.working_line_pos) };
959
960    let mut segment_buffer: Vec<ErasedSegment> = Vec::new();
961    for (idx, segment) in enumerate(segments) {
962        let old_position = segment.get_position_marker();
963
964        let mut new_position = match old_position {
965            Some(pos_marker) => pos_marker.clone(),
966            None => {
967                let start_point = if idx > 0 {
968                    let prev_seg = segment_buffer[idx - 1].clone();
969                    Some(prev_seg.get_position_marker().unwrap().end_point_marker())
970                } else {
971                    Some(parent_pos.start_point_marker())
972                };
973
974                let mut end_point = None;
975                for fwd_seg in &segments[idx + 1..] {
976                    if fwd_seg.get_position_marker().is_some() {
977                        end_point = Some(
978                            fwd_seg.get_raw_segments()[0]
979                                .get_position_marker()
980                                .unwrap()
981                                .start_point_marker(),
982                        );
983                        break;
984                    }
985                }
986
987                if let Some((start_point, end_point)) = start_point
988                    .as_ref()
989                    .zip(end_point.as_ref())
990                    .filter(|(start_point, end_point)| start_point != end_point)
991                {
992                    PositionMarker::from_points(start_point, end_point)
993                } else if let Some(start_point) = start_point.as_ref() {
994                    start_point.clone()
995                } else if let Some(end_point) = end_point.as_ref() {
996                    end_point.clone()
997                } else {
998                    unimplemented!("Unable to position new segment")
999                }
1000            }
1001        };
1002
1003        new_position = new_position.with_working_position(line_no, line_pos);
1004        (line_no, line_pos) = PositionMarker::infer_next_position(segment.raw(), line_no, line_pos);
1005
1006        let mut new_seg = if !segment.segments().is_empty() && old_position != Some(&new_position) {
1007            let child_segments = position_segments(segment.segments(), &new_position);
1008            segment.change_segments(child_segments)
1009        } else {
1010            segment.deep_clone()
1011        };
1012
1013        new_seg.get_mut().set_position_marker(new_position.into());
1014        segment_buffer.push(new_seg);
1015    }
1016
1017    segment_buffer
1018}
1019
1020#[derive(Debug, Clone)]
1021pub struct NodeOrToken {
1022    id: u32,
1023    syntax_kind: SyntaxKind,
1024    class_types: SyntaxSet,
1025    position_marker: Option<PositionMarker>,
1026    kind: NodeOrTokenKind,
1027    code_idx: OnceCell<Rc<Vec<usize>>>,
1028    hash: OnceCell<u64>,
1029}
1030
1031#[derive(Debug, Clone)]
1032pub enum NodeOrTokenKind {
1033    Node(NodeData),
1034    Token(TokenData),
1035}
1036
1037impl NodeOrToken {
1038    pub fn set_position_marker(&mut self, position_marker: Option<PositionMarker>) {
1039        self.position_marker = position_marker;
1040    }
1041
1042    pub fn set_id(&mut self, id: u32) {
1043        self.id = id;
1044    }
1045}
1046
1047#[derive(Debug, Clone)]
1048pub struct NodeData {
1049    dialect: DialectKind,
1050    segments: Vec<ErasedSegment>,
1051    raw: OnceCell<SmolStr>,
1052    source_fixes: Vec<SourceFix>,
1053    descendant_type_set: OnceCell<SyntaxSet>,
1054    raw_segments_with_ancestors: OnceCell<Vec<(ErasedSegment, Vec<PathStep>)>>,
1055}
1056
1057#[derive(Debug, Clone, PartialEq)]
1058pub struct TokenData {
1059    raw: SmolStr,
1060}
1061
1062#[track_caller]
1063pub fn pos_marker(segments: &[ErasedSegment]) -> PositionMarker {
1064    let markers = segments.iter().filter_map(|seg| seg.get_position_marker());
1065
1066    PositionMarker::from_child_markers(markers)
1067}
1068
1069#[derive(Debug, Clone)]
1070pub struct PathStep {
1071    pub segment: ErasedSegment,
1072    pub idx: usize,
1073    pub len: usize,
1074    pub code_idxs: Rc<Vec<usize>>,
1075}
1076
1077fn class_types(syntax_kind: SyntaxKind) -> SyntaxSet {
1078    match syntax_kind {
1079        SyntaxKind::ColumnReference => SyntaxSet::new(&[SyntaxKind::ObjectReference, syntax_kind]),
1080        SyntaxKind::WildcardIdentifier => {
1081            SyntaxSet::new(&[SyntaxKind::WildcardIdentifier, SyntaxKind::ObjectReference])
1082        }
1083        SyntaxKind::TableReference => SyntaxSet::new(&[SyntaxKind::ObjectReference, syntax_kind]),
1084        _ => SyntaxSet::single(syntax_kind),
1085    }
1086}
1087
1088#[cfg(test)]
1089mod tests {
1090    use super::*;
1091    use crate::lint_fix::LintFix;
1092    use crate::linter::compute_anchor_edit_info;
1093    use crate::parser::segments::test_functions::{raw_seg, raw_segments};
1094
1095    #[test]
1096    /// Test comparison of raw segments.
1097    fn test_parser_base_segments_raw_compare() {
1098        let template: TemplatedFile = "foobar".into();
1099        let rs1 = SegmentBuilder::token(0, "foobar", SyntaxKind::Word)
1100            .with_position(PositionMarker::new(
1101                0..6,
1102                0..6,
1103                template.clone(),
1104                None,
1105                None,
1106            ))
1107            .finish();
1108        let rs2 = SegmentBuilder::token(0, "foobar", SyntaxKind::Word)
1109            .with_position(PositionMarker::new(
1110                0..6,
1111                0..6,
1112                template.clone(),
1113                None,
1114                None,
1115            ))
1116            .finish();
1117
1118        assert_eq!(rs1, rs2)
1119    }
1120
1121    #[test]
1122    // TODO Implement
1123    /// Test raw segments behave as expected.
1124    fn test_parser_base_segments_raw() {
1125        let raw_seg = raw_seg();
1126
1127        assert_eq!(raw_seg.raw(), "foobar");
1128    }
1129
1130    #[test]
1131    /// Test BaseSegment.compute_anchor_edit_info().
1132    fn test_parser_base_segments_compute_anchor_edit_info() {
1133        let raw_segs = raw_segments();
1134        let tables = Tables::default();
1135
1136        // Construct a fix buffer, intentionally with:
1137        // - one duplicate.
1138        // - two different incompatible fixes on the same segment.
1139        let fixes = vec![
1140            LintFix::replace(
1141                raw_segs[0].clone(),
1142                vec![raw_segs[0].edit(tables.next_id(), Some("a".to_string()), None)],
1143                None,
1144            ),
1145            LintFix::replace(
1146                raw_segs[0].clone(),
1147                vec![raw_segs[0].edit(tables.next_id(), Some("a".to_string()), None)],
1148                None,
1149            ),
1150            LintFix::replace(
1151                raw_segs[0].clone(),
1152                vec![raw_segs[0].edit(tables.next_id(), Some("b".to_string()), None)],
1153                None,
1154            ),
1155        ];
1156
1157        let mut anchor_edit_info = Default::default();
1158        compute_anchor_edit_info(&mut anchor_edit_info, fixes);
1159
1160        // Check the target segment is the only key we have.
1161        assert_eq!(
1162            anchor_edit_info.keys().collect::<Vec<_>>(),
1163            vec![&raw_segs[0].id()]
1164        );
1165
1166        let anchor_info = anchor_edit_info.get(&raw_segs[0].id()).unwrap();
1167
1168        // Check that the duplicate as been deduplicated i.e. this isn't 3.
1169        assert_eq!(anchor_info.replace, 2);
1170
1171        // Check the fixes themselves.
1172        //   Note: There's no duplicated first fix.
1173        assert_eq!(
1174            anchor_info.fixes[0],
1175            LintFix::replace(
1176                raw_segs[0].clone(),
1177                vec![raw_segs[0].edit(tables.next_id(), Some("a".to_string()), None)],
1178                None,
1179            )
1180        );
1181        assert_eq!(
1182            anchor_info.fixes[1],
1183            LintFix::replace(
1184                raw_segs[0].clone(),
1185                vec![raw_segs[0].edit(tables.next_id(), Some("b".to_string()), None)],
1186                None,
1187            )
1188        );
1189
1190        // Check the first replace
1191        assert_eq!(
1192            anchor_info.fixes[anchor_info.first_replace.unwrap()],
1193            LintFix::replace(
1194                raw_segs[0].clone(),
1195                vec![raw_segs[0].edit(tables.next_id(), Some("a".to_string()), None)],
1196                None,
1197            )
1198        );
1199    }
1200}
sqruff_lib_core/parser/segments.rs

sqruff_lib_core/parser/
segments.rs
