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::{BuildHasher, Hash, Hasher};
15use std::rc::Rc;
16
17use hashbrown::{DefaultHashBuilder, HashMap};
18use itertools::enumerate;
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 {
394            if has_descendant_source_fixes {
395                // Tree raw hasn't changed - only source fix patches are needed.
396                // Avoid generating gap patches that could span template boundaries.
397                // This matches SQLFluff's behavior in _iter_templated_patches.
398                for descendant in self.recursive_crawl_all(false).into_iter().skip(1) {
399                    acc.extend(descendant.iter_source_fix_patches(templated_file));
400                }
401            }
402            return acc;
403        }
404
405        if self.get_position_marker().is_none() {
406            return Vec::new();
407        }
408
409        let pos_marker = self.get_position_marker().unwrap();
410        if pos_marker.is_literal() && !has_descendant_source_fixes {
411            acc.extend(self.iter_source_fix_patches(templated_file));
412            acc.push(FixPatch::new(
413                pos_marker.templated_slice.clone(),
414                self.raw().clone(),
415                // SyntaxKind::Literal.into(),
416                pos_marker.source_slice.clone(),
417                templated_file.templated_str.as_ref().unwrap()[pos_marker.templated_slice.clone()]
418                    .to_string(),
419                templated_file.source_str[pos_marker.source_slice.clone()].to_string(),
420            ));
421        } else if self.segments().is_empty() {
422            return acc;
423        } else {
424            let mut segments = self.segments();
425
426            while !segments.is_empty()
427                && matches!(
428                    segments.last().unwrap().get_type(),
429                    SyntaxKind::EndOfFile
430                        | SyntaxKind::Indent
431                        | SyntaxKind::Dedent
432                        | SyntaxKind::Implicit
433                )
434            {
435                segments = &segments[..segments.len() - 1];
436            }
437
438            let pos = self.get_position_marker().unwrap();
439            let mut source_idx = pos.source_slice.start;
440            let mut templated_idx = pos.templated_slice.start;
441            let mut insert_buff = String::new();
442
443            for segment in segments {
444                let pos_marker = segment.get_position_marker().unwrap();
445                if !segment.raw().is_empty() && pos_marker.is_point() {
446                    insert_buff.push_str(segment.raw().as_ref());
447                    continue;
448                }
449
450                let start_diff = pos_marker.templated_slice.start - templated_idx;
451
452                if start_diff > 0 || !insert_buff.is_empty() {
453                    let fixed_raw = std::mem::take(&mut insert_buff);
454                    let raw_segments = segment.get_raw_segments();
455                    let first_segment_pos = raw_segments[0].get_position_marker().unwrap();
456
457                    acc.push(FixPatch::new(
458                        templated_idx..first_segment_pos.templated_slice.start,
459                        fixed_raw.into(),
460                        source_idx..first_segment_pos.source_slice.start,
461                        String::new(),
462                        String::new(),
463                    ));
464                }
465
466                acc.extend(segment.iter_patches(templated_file));
467
468                source_idx = pos_marker.source_slice.end;
469                templated_idx = pos_marker.templated_slice.end;
470            }
471
472            let end_diff = pos.templated_slice.end - templated_idx;
473            if end_diff != 0 || !insert_buff.is_empty() {
474                let source_slice = source_idx..pos.source_slice.end;
475                let templated_slice = templated_idx..pos.templated_slice.end;
476
477                let templated_str = templated_file.templated_str.as_ref().unwrap()
478                    [templated_slice.clone()]
479                .to_owned();
480                let source_str = templated_file.source_str[source_slice.clone()].to_owned();
481
482                acc.push(FixPatch::new(
483                    templated_slice,
484                    insert_buff.into(),
485                    source_slice,
486                    templated_str,
487                    source_str,
488                ));
489            }
490        }
491
492        acc
493    }
494
495    pub fn descendant_type_set(&self) -> &SyntaxSet {
496        match &self.value.kind {
497            NodeOrTokenKind::Node(node) => node.descendant_type_set.get_or_init(|| {
498                self.segments()
499                    .iter()
500                    .flat_map(|segment| {
501                        segment
502                            .descendant_type_set()
503                            .clone()
504                            .union(segment.class_types())
505                    })
506                    .collect()
507            }),
508            NodeOrTokenKind::Token(_) => const { &SyntaxSet::EMPTY },
509        }
510    }
511
512    pub fn is_comment(&self) -> bool {
513        matches!(
514            self.value.syntax_kind,
515            SyntaxKind::Comment
516                | SyntaxKind::InlineComment
517                | SyntaxKind::BlockComment
518                | SyntaxKind::NotebookStart
519        )
520    }
521
522    pub fn is_whitespace(&self) -> bool {
523        matches!(
524            self.value.syntax_kind,
525            SyntaxKind::Whitespace | SyntaxKind::Newline
526        )
527    }
528
529    pub fn is_indent(&self) -> bool {
530        matches!(
531            self.value.syntax_kind,
532            SyntaxKind::Indent | SyntaxKind::Implicit | SyntaxKind::Dedent
533        )
534    }
535
536    pub fn get_position_marker(&self) -> Option<&PositionMarker> {
537        self.value.position_marker.as_ref()
538    }
539
540    pub(crate) fn iter_source_fix_patches(&self, templated_file: &TemplatedFile) -> Vec<FixPatch> {
541        let source_fixes = self.get_source_fixes();
542        let mut patches = Vec::with_capacity(source_fixes.len());
543
544        for source_fix in &source_fixes {
545            patches.push(FixPatch::new(
546                source_fix.templated_slice.clone(),
547                source_fix.edit.clone(),
548                // String::from("source"),
549                source_fix.source_slice.clone(),
550                templated_file.templated_str.clone().unwrap()[source_fix.templated_slice.clone()]
551                    .to_string(),
552                templated_file.source_str[source_fix.source_slice.clone()].to_string(),
553            ));
554        }
555
556        patches
557    }
558
559    pub fn id(&self) -> u32 {
560        self.value.id
561    }
562
563    /// Return any source fixes as list.
564    pub fn get_source_fixes(&self) -> Vec<SourceFix> {
565        match &self.value.kind {
566            NodeOrTokenKind::Node(node) => node.source_fixes.clone(),
567            NodeOrTokenKind::Token(_) => Vec::new(),
568        }
569    }
570
571    /// Return all source fixes from this segment and all its descendants.
572    pub fn get_all_source_fixes(&self) -> Vec<SourceFix> {
573        let mut fixes = self.get_source_fixes();
574        for segment in self.segments() {
575            fixes.extend(segment.get_all_source_fixes());
576        }
577        fixes
578    }
579
580    pub fn edit(
581        &self,
582        id: u32,
583        raw: Option<String>,
584        _source_fixes: Option<Vec<SourceFix>>,
585    ) -> ErasedSegment {
586        match &self.value.kind {
587            NodeOrTokenKind::Node(_node) => {
588                todo!()
589            }
590            NodeOrTokenKind::Token(token) => {
591                let raw = raw.as_deref().unwrap_or(token.raw.as_ref());
592                SegmentBuilder::token(id, raw, self.value.syntax_kind)
593                    .with_position(self.get_position_marker().unwrap().clone())
594                    .finish()
595            }
596        }
597    }
598
599    pub fn class_types(&self) -> &SyntaxSet {
600        &self.value.class_types
601    }
602
603    pub(crate) fn first_non_whitespace_segment_raw_upper(&self) -> Option<String> {
604        for seg in self.get_raw_segments() {
605            if !seg.raw().is_empty() {
606                return Some(seg.raw().to_uppercase());
607            }
608        }
609        None
610    }
611
612    pub fn is(&self, other: &ErasedSegment) -> bool {
613        Rc::ptr_eq(&self.value, &other.value)
614    }
615
616    pub fn addr(&self) -> usize {
617        Rc::as_ptr(&self.value).addr()
618    }
619
620    pub fn direct_descendant_type_set(&self) -> SyntaxSet {
621        self.segments()
622            .iter()
623            .fold(SyntaxSet::EMPTY, |set, it| set.union(it.class_types()))
624    }
625
626    pub fn is_keyword(&self, p0: &str) -> bool {
627        self.is_type(SyntaxKind::Keyword) && self.raw().eq_ignore_ascii_case(p0)
628    }
629
630    pub fn hash_value(&self) -> u64 {
631        *self.value.hash.get_or_init(|| {
632            let mut hasher = DefaultHashBuilder::default().build_hasher();
633            self.get_type().hash(&mut hasher);
634            self.raw().hash(&mut hasher);
635
636            if let Some(marker) = &self.get_position_marker() {
637                marker.source_position().hash(&mut hasher);
638            } else {
639                None::<usize>.hash(&mut hasher);
640            }
641
642            hasher.finish()
643        })
644    }
645
646    pub fn deep_clone(&self) -> Self {
647        Self {
648            value: Rc::new(self.value.as_ref().clone()),
649        }
650    }
651
652    #[track_caller]
653    pub(crate) fn get_mut(&mut self) -> &mut NodeOrToken {
654        Rc::get_mut(&mut self.value).unwrap()
655    }
656
657    #[track_caller]
658    pub(crate) fn make_mut(&mut self) -> &mut NodeOrToken {
659        Rc::make_mut(&mut self.value)
660    }
661
662    pub fn reference(&self) -> ObjectReferenceSegment {
663        ObjectReferenceSegment(
664            self.clone(),
665            match self.get_type() {
666                SyntaxKind::TableReference => ObjectReferenceKind::Table,
667                SyntaxKind::WildcardIdentifier => ObjectReferenceKind::WildcardIdentifier,
668                _ => ObjectReferenceKind::Object,
669            },
670        )
671    }
672
673    pub fn recursive_crawl_all(&self, reverse: bool) -> Vec<ErasedSegment> {
674        let mut result = Vec::with_capacity(self.segments().len() + 1);
675
676        if reverse {
677            for seg in self.segments().iter().rev() {
678                result.append(&mut seg.recursive_crawl_all(reverse));
679            }
680            result.push(self.clone());
681        } else {
682            result.push(self.clone());
683            for seg in self.segments() {
684                result.append(&mut seg.recursive_crawl_all(reverse));
685            }
686        }
687
688        result
689    }
690
691    pub fn raw_segments_with_ancestors(&self) -> &[(ErasedSegment, Vec<PathStep>)] {
692        match &self.value.kind {
693            NodeOrTokenKind::Node(node) => node.raw_segments_with_ancestors.get_or_init(|| {
694                let mut buffer: Vec<(ErasedSegment, Vec<PathStep>)> =
695                    Vec::with_capacity(self.segments().len());
696                let code_idxs = self.code_indices();
697
698                for (idx, seg) in self.segments().iter().enumerate() {
699                    let new_step = vec![PathStep {
700                        segment: self.clone(),
701                        idx,
702                        len: self.segments().len(),
703                        code_idxs: code_idxs.clone(),
704                    }];
705
706                    // Use seg.get_segments().is_empty() as a workaround to check if the segment is
707                    // a SyntaxKind::Raw type. In the original Python code, this was achieved
708                    // using seg.is_type(SyntaxKind::Raw). Here, we assume that a SyntaxKind::Raw
709                    // segment is characterized by having no sub-segments.
710
711                    if seg.segments().is_empty() {
712                        buffer.push((seg.clone(), new_step));
713                    } else {
714                        let extended =
715                            seg.raw_segments_with_ancestors()
716                                .iter()
717                                .map(|(raw_seg, stack)| {
718                                    let mut new_step = new_step.clone();
719                                    new_step.extend_from_slice(stack);
720                                    (raw_seg.clone(), new_step)
721                                });
722
723                        buffer.extend(extended);
724                    }
725                }
726
727                buffer
728            }),
729            NodeOrTokenKind::Token(_) => &[],
730        }
731    }
732
733    pub fn path_to(&self, other: &ErasedSegment) -> Vec<PathStep> {
734        let midpoint = other;
735
736        for (idx, seg) in enumerate(self.segments()) {
737            let mut steps = vec![PathStep {
738                segment: self.clone(),
739                idx,
740                len: self.segments().len(),
741                code_idxs: self.code_indices(),
742            }];
743
744            if seg.eq(midpoint) {
745                return steps;
746            }
747
748            let res = seg.path_to(midpoint);
749
750            if !res.is_empty() {
751                steps.extend(res);
752                return steps;
753            }
754        }
755
756        Vec::new()
757    }
758
759    pub fn apply_fixes(
760        &self,
761        fixes: &mut HashMap<u32, AnchorEditInfo>,
762    ) -> (ErasedSegment, Vec<ErasedSegment>, Vec<ErasedSegment>) {
763        if fixes.is_empty() || self.segments().is_empty() {
764            return (self.clone(), Vec::new(), Vec::new());
765        }
766
767        let mut seg_buffer = Vec::new();
768        let mut has_applied_fixes = false;
769        let mut _requires_validate = false;
770
771        for seg in self.segments() {
772            // Look for uuid match.
773            // This handles potential positioning ambiguity.
774
775            let Some(mut anchor_info) = fixes.remove(&seg.id()) else {
776                seg_buffer.push(seg.clone());
777                continue;
778            };
779
780            if anchor_info.fixes.len() == 2
781                && matches!(anchor_info.fixes[0], LintFix::CreateAfter { .. })
782            {
783                anchor_info.fixes.reverse();
784            }
785
786            let fixes_count = anchor_info.fixes.len();
787            for lint_fix in anchor_info.fixes {
788                has_applied_fixes = true;
789
790                // Deletes are easy.
791                if matches!(lint_fix, LintFix::Delete { .. }) {
792                    // We're just getting rid of this segment.
793                    _requires_validate = true;
794                    // NOTE: We don't add the segment in this case.
795                    continue;
796                }
797
798                // Otherwise it must be a replace or a create.
799                assert!(matches!(
800                    lint_fix,
801                    LintFix::Replace { .. }
802                        | LintFix::CreateBefore { .. }
803                        | LintFix::CreateAfter { .. }
804                ));
805
806                match lint_fix {
807                    LintFix::CreateAfter { edit, .. } => {
808                        if fixes_count == 1 {
809                            // In the case of a creation after that is not part
810                            // of a create_before/create_after pair, also add
811                            // this segment before the edit.
812                            seg_buffer.push(seg.clone());
813                        }
814                        for s in edit {
815                            seg_buffer.push(s);
816                        }
817                        _requires_validate = true;
818                    }
819                    LintFix::CreateBefore { edit, .. } => {
820                        for s in edit {
821                            seg_buffer.push(s);
822                        }
823                        seg_buffer.push(seg.clone());
824                        _requires_validate = true;
825                    }
826                    LintFix::Replace { edit, .. } => {
827                        let mut consumed_pos = false;
828                        let is_single_same_type =
829                            edit.len() == 1 && edit[0].class_types() == seg.class_types();
830
831                        for mut s in edit {
832                            if !consumed_pos && s.raw() == seg.raw() {
833                                consumed_pos = true;
834                                s.make_mut()
835                                    .set_position_marker(seg.get_position_marker().cloned());
836                            }
837                            seg_buffer.push(s);
838                        }
839
840                        if !is_single_same_type {
841                            _requires_validate = true;
842                        }
843                    }
844                    LintFix::Delete { .. } => {
845                        // Already handled above
846                        unreachable!()
847                    }
848                }
849            }
850        }
851
852        if has_applied_fixes {
853            seg_buffer =
854                position_segments(&seg_buffer, self.get_position_marker().as_ref().unwrap());
855        }
856
857        let seg_queue = seg_buffer;
858        let mut seg_buffer = Vec::new();
859        for seg in seg_queue {
860            let (mid, pre, post) = seg.apply_fixes(fixes);
861
862            seg_buffer.extend(pre);
863            seg_buffer.push(mid);
864            seg_buffer.extend(post);
865        }
866
867        let seg_buffer =
868            position_segments(&seg_buffer, self.get_position_marker().as_ref().unwrap());
869        (self.new(seg_buffer), Vec::new(), Vec::new())
870    }
871}
872
873#[cfg(any(test, feature = "serde"))]
874pub mod serde {
875    use serde::ser::SerializeMap;
876    use serde::{Deserialize, Serialize};
877
878    use crate::parser::segments::ErasedSegment;
879
880    #[derive(Serialize, Deserialize)]
881    #[serde(untagged)]
882    pub enum SerialisedSegmentValue {
883        Single(String),
884        Nested(Vec<TupleSerialisedSegment>),
885    }
886
887    #[derive(Deserialize)]
888    pub struct TupleSerialisedSegment(String, SerialisedSegmentValue);
889
890    impl Serialize for TupleSerialisedSegment {
891        fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
892        where
893            S: serde::Serializer,
894        {
895            let mut map = serializer.serialize_map(None)?;
896            map.serialize_key(&self.0)?;
897            map.serialize_value(&self.1)?;
898            map.end()
899        }
900    }
901
902    impl TupleSerialisedSegment {
903        pub fn sinlge(key: String, value: String) -> Self {
904            Self(key, SerialisedSegmentValue::Single(value))
905        }
906
907        pub fn nested(key: String, segments: Vec<TupleSerialisedSegment>) -> Self {
908            Self(key, SerialisedSegmentValue::Nested(segments))
909        }
910    }
911
912    impl ErasedSegment {
913        pub fn to_serialised(&self, code_only: bool, show_raw: bool) -> TupleSerialisedSegment {
914            if show_raw && self.segments().is_empty() {
915                TupleSerialisedSegment::sinlge(
916                    self.get_type().as_str().to_string(),
917                    self.raw().to_string(),
918                )
919            } else if code_only {
920                let segments = self
921                    .segments()
922                    .iter()
923                    .filter(|seg| seg.is_code() && !seg.is_meta())
924                    .map(|seg| seg.to_serialised(code_only, show_raw))
925                    .collect::<Vec<_>>();
926
927                TupleSerialisedSegment::nested(self.get_type().as_str().to_string(), segments)
928            } else {
929                let segments = self
930                    .segments()
931                    .iter()
932                    .map(|seg| seg.to_serialised(code_only, show_raw))
933                    .collect::<Vec<_>>();
934
935                TupleSerialisedSegment::nested(self.get_type().as_str().to_string(), segments)
936            }
937        }
938    }
939}
940
941impl PartialEq for ErasedSegment {
942    fn eq(&self, other: &Self) -> bool {
943        if self.id() == other.id() {
944            return true;
945        }
946
947        let pos_self = self.get_position_marker();
948        let pos_other = other.get_position_marker();
949        if let Some((pos_self, pos_other)) = pos_self.zip(pos_other) {
950            self.get_type() == other.get_type()
951                && pos_self.working_loc() == pos_other.working_loc()
952                && self.raw() == other.raw()
953        } else {
954            false
955        }
956    }
957}
958
959pub fn position_segments(
960    segments: &[ErasedSegment],
961    parent_pos: &PositionMarker,
962) -> Vec<ErasedSegment> {
963    if segments.is_empty() {
964        return Vec::new();
965    }
966
967    let (mut line_no, mut line_pos) = { (parent_pos.working_line_no, parent_pos.working_line_pos) };
968
969    let mut segment_buffer: Vec<ErasedSegment> = Vec::new();
970    for (idx, segment) in enumerate(segments) {
971        let old_position = segment.get_position_marker();
972
973        let mut new_position = match old_position {
974            Some(pos_marker) => pos_marker.clone(),
975            None => {
976                let start_point = if idx > 0 {
977                    let prev_seg = segment_buffer[idx - 1].clone();
978                    Some(prev_seg.get_position_marker().unwrap().end_point_marker())
979                } else {
980                    Some(parent_pos.start_point_marker())
981                };
982
983                let mut end_point = None;
984                for fwd_seg in &segments[idx + 1..] {
985                    if fwd_seg.get_position_marker().is_some() {
986                        end_point = Some(
987                            fwd_seg.get_raw_segments()[0]
988                                .get_position_marker()
989                                .unwrap()
990                                .start_point_marker(),
991                        );
992                        break;
993                    }
994                }
995
996                if let Some((start_point, end_point)) = start_point
997                    .as_ref()
998                    .zip(end_point.as_ref())
999                    .filter(|(start_point, end_point)| start_point != end_point)
1000                {
1001                    PositionMarker::from_points(start_point, end_point)
1002                } else if let Some(start_point) = start_point.as_ref() {
1003                    start_point.clone()
1004                } else if let Some(end_point) = end_point.as_ref() {
1005                    end_point.clone()
1006                } else {
1007                    unimplemented!("Unable to position new segment")
1008                }
1009            }
1010        };
1011
1012        new_position = new_position.with_working_position(line_no, line_pos);
1013        (line_no, line_pos) = PositionMarker::infer_next_position(segment.raw(), line_no, line_pos);
1014
1015        let mut new_seg = if !segment.segments().is_empty() && old_position != Some(&new_position) {
1016            let child_segments = position_segments(segment.segments(), &new_position);
1017            segment.change_segments(child_segments)
1018        } else {
1019            segment.deep_clone()
1020        };
1021
1022        new_seg.get_mut().set_position_marker(new_position.into());
1023        segment_buffer.push(new_seg);
1024    }
1025
1026    segment_buffer
1027}
1028
1029#[derive(Debug, Clone)]
1030pub struct NodeOrToken {
1031    id: u32,
1032    syntax_kind: SyntaxKind,
1033    class_types: SyntaxSet,
1034    position_marker: Option<PositionMarker>,
1035    kind: NodeOrTokenKind,
1036    code_idx: OnceCell<Rc<Vec<usize>>>,
1037    hash: OnceCell<u64>,
1038}
1039
1040#[derive(Debug, Clone)]
1041#[allow(clippy::large_enum_variant)]
1042pub enum NodeOrTokenKind {
1043    Node(NodeData),
1044    Token(TokenData),
1045}
1046
1047impl NodeOrToken {
1048    pub fn set_position_marker(&mut self, position_marker: Option<PositionMarker>) {
1049        self.position_marker = position_marker;
1050    }
1051
1052    pub fn set_id(&mut self, id: u32) {
1053        self.id = id;
1054    }
1055}
1056
1057#[derive(Debug, Clone)]
1058pub struct NodeData {
1059    dialect: DialectKind,
1060    segments: Vec<ErasedSegment>,
1061    raw: OnceCell<SmolStr>,
1062    source_fixes: Vec<SourceFix>,
1063    descendant_type_set: OnceCell<SyntaxSet>,
1064    raw_segments_with_ancestors: OnceCell<Vec<(ErasedSegment, Vec<PathStep>)>>,
1065}
1066
1067#[derive(Debug, Clone, PartialEq)]
1068pub struct TokenData {
1069    raw: SmolStr,
1070}
1071
1072#[track_caller]
1073pub fn pos_marker(segments: &[ErasedSegment]) -> PositionMarker {
1074    let markers = segments.iter().filter_map(|seg| seg.get_position_marker());
1075
1076    PositionMarker::from_child_markers(markers)
1077}
1078
1079#[derive(Debug, Clone)]
1080pub struct PathStep {
1081    pub segment: ErasedSegment,
1082    pub idx: usize,
1083    pub len: usize,
1084    pub code_idxs: Rc<Vec<usize>>,
1085}
1086
1087fn class_types(syntax_kind: SyntaxKind) -> SyntaxSet {
1088    match syntax_kind {
1089        SyntaxKind::ColumnReference => SyntaxSet::new(&[SyntaxKind::ObjectReference, syntax_kind]),
1090        SyntaxKind::WildcardIdentifier => {
1091            SyntaxSet::new(&[SyntaxKind::WildcardIdentifier, SyntaxKind::ObjectReference])
1092        }
1093        SyntaxKind::TableReference => SyntaxSet::new(&[SyntaxKind::ObjectReference, syntax_kind]),
1094        _ => SyntaxSet::single(syntax_kind),
1095    }
1096}
1097
1098#[cfg(test)]
1099mod tests {
1100    use super::*;
1101    use crate::lint_fix::LintFix;
1102    use crate::linter::compute_anchor_edit_info;
1103    use crate::parser::segments::test_functions::{raw_seg, raw_segments};
1104
1105    #[test]
1106    /// Test comparison of raw segments.
1107    fn test_parser_base_segments_raw_compare() {
1108        let template: TemplatedFile = "foobar".into();
1109        let rs1 = SegmentBuilder::token(0, "foobar", SyntaxKind::Word)
1110            .with_position(PositionMarker::new(
1111                0..6,
1112                0..6,
1113                template.clone(),
1114                None,
1115                None,
1116            ))
1117            .finish();
1118        let rs2 = SegmentBuilder::token(0, "foobar", SyntaxKind::Word)
1119            .with_position(PositionMarker::new(
1120                0..6,
1121                0..6,
1122                template.clone(),
1123                None,
1124                None,
1125            ))
1126            .finish();
1127
1128        assert_eq!(rs1, rs2)
1129    }
1130
1131    #[test]
1132    // TODO Implement
1133    /// Test raw segments behave as expected.
1134    fn test_parser_base_segments_raw() {
1135        let raw_seg = raw_seg();
1136
1137        assert_eq!(raw_seg.raw(), "foobar");
1138    }
1139
1140    #[test]
1141    /// Test BaseSegment.compute_anchor_edit_info().
1142    fn test_parser_base_segments_compute_anchor_edit_info() {
1143        let raw_segs = raw_segments();
1144        let tables = Tables::default();
1145
1146        // Construct a fix buffer, intentionally with:
1147        // - one duplicate.
1148        // - two different incompatible fixes on the same segment.
1149        let fixes = vec![
1150            LintFix::replace(
1151                raw_segs[0].clone(),
1152                vec![raw_segs[0].edit(tables.next_id(), Some("a".to_string()), None)],
1153                None,
1154            ),
1155            LintFix::replace(
1156                raw_segs[0].clone(),
1157                vec![raw_segs[0].edit(tables.next_id(), Some("a".to_string()), None)],
1158                None,
1159            ),
1160            LintFix::replace(
1161                raw_segs[0].clone(),
1162                vec![raw_segs[0].edit(tables.next_id(), Some("b".to_string()), None)],
1163                None,
1164            ),
1165        ];
1166
1167        let mut anchor_edit_info = Default::default();
1168        compute_anchor_edit_info(&mut anchor_edit_info, fixes);
1169
1170        // Check the target segment is the only key we have.
1171        assert_eq!(
1172            anchor_edit_info.keys().collect::<Vec<_>>(),
1173            vec![&raw_segs[0].id()]
1174        );
1175
1176        let anchor_info = anchor_edit_info.get(&raw_segs[0].id()).unwrap();
1177
1178        // Check that the duplicate as been deduplicated i.e. this isn't 3.
1179        assert_eq!(anchor_info.replace, 2);
1180
1181        // Check the fixes themselves.
1182        //   Note: There's no duplicated first fix.
1183        assert_eq!(
1184            anchor_info.fixes[0],
1185            LintFix::replace(
1186                raw_segs[0].clone(),
1187                vec![raw_segs[0].edit(tables.next_id(), Some("a".to_string()), None)],
1188                None,
1189            )
1190        );
1191        assert_eq!(
1192            anchor_info.fixes[1],
1193            LintFix::replace(
1194                raw_segs[0].clone(),
1195                vec![raw_segs[0].edit(tables.next_id(), Some("b".to_string()), None)],
1196                None,
1197            )
1198        );
1199
1200        // Check the first replace
1201        assert_eq!(
1202            anchor_info.fixes[anchor_info.first_replace.unwrap()],
1203            LintFix::replace(
1204                raw_segs[0].clone(),
1205                vec![raw_segs[0].edit(tables.next_id(), Some("a".to_string()), None)],
1206                None,
1207            )
1208        );
1209    }
1210}
sqruff_lib_core/parser/segments.rs

sqruff_lib_core/parser/
segments.rs
