squawk_ide/
expand_selection.rs

1// via https://github.com/rust-lang/rust-analyzer/blob/8d75311400a108d7ffe17dc9c38182c566952e6e/crates/ide/src/extend_selection.rs#L1C1-L1C1
2//
3// Permission is hereby granted, free of charge, to any
4// person obtaining a copy of this software and associated
5// documentation files (the "Software"), to deal in the
6// Software without restriction, including without
7// limitation the rights to use, copy, modify, merge,
8// publish, distribute, sublicense, and/or sell copies of
9// the Software, and to permit persons to whom the Software
10// is furnished to do so, subject to the following
11// conditions:
12//
13// The above copyright notice and this permission notice
14// shall be included in all copies or substantial portions
15// of the Software.
16//
17// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF
18// ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED
19// TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A
20// PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT
21// SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
22// CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
23// OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR
24// IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
25// DEALINGS IN THE SOFTWARE.
26
27// NOTE: this is pretty much copied as is from rust analyzer with some
28// simplifications. I imagine there's more we can do to adapt it for SQL.
29
30use rowan::{Direction, NodeOrToken, TextRange, TextSize};
31use squawk_syntax::{
32    SyntaxKind, SyntaxNode, SyntaxToken,
33    ast::{self, AstToken},
34};
35
36const DELIMITED_LIST_KINDS: &[SyntaxKind] = &[
37    SyntaxKind::ALTER_OPTION_LIST,
38    SyntaxKind::ARG_LIST,
39    SyntaxKind::ATTRIBUTE_LIST,
40    SyntaxKind::BEGIN_FUNC_OPTION_LIST,
41    SyntaxKind::COLUMN_LIST,
42    SyntaxKind::CONFLICT_INDEX_ITEM_LIST,
43    SyntaxKind::CONSTRAINT_EXCLUSION_LIST,
44    SyntaxKind::COPY_OPTION_LIST,
45    SyntaxKind::DROP_OP_CLASS_OPTION_LIST,
46    SyntaxKind::FDW_OPTION_LIST,
47    SyntaxKind::FUNCTION_SIG_LIST,
48    SyntaxKind::GROUP_BY_LIST,
49    SyntaxKind::JSON_TABLE_COLUMN_LIST,
50    SyntaxKind::OPERATOR_CLASS_OPTION_LIST,
51    SyntaxKind::OPTION_ITEM_LIST,
52    SyntaxKind::OP_SIG_LIST,
53    SyntaxKind::PARAM_LIST,
54    SyntaxKind::PARTITION_ITEM_LIST,
55    SyntaxKind::PARTITION_LIST,
56    SyntaxKind::RETURNING_OPTION_LIST,
57    SyntaxKind::REVOKE_COMMAND_LIST,
58    SyntaxKind::ROLE_LIST,
59    SyntaxKind::ROW_LIST,
60    SyntaxKind::XML_ATTRIBUTE_LIST,
61    SyntaxKind::XML_NAMESPACE_LIST,
62    SyntaxKind::SET_COLUMN_LIST,
63    SyntaxKind::SET_EXPR_LIST,
64    SyntaxKind::SET_OPTIONS_LIST,
65    SyntaxKind::SORT_BY_LIST,
66    SyntaxKind::TABLE_AND_COLUMNS_LIST,
67    SyntaxKind::TABLE_ARG_LIST,
68    SyntaxKind::TABLE_LIST,
69    SyntaxKind::TARGET_LIST,
70    SyntaxKind::TRANSACTION_MODE_LIST,
71    SyntaxKind::VACUUM_OPTION_LIST,
72    SyntaxKind::VARIANT_LIST,
73    SyntaxKind::XML_TABLE_COLUMN_LIST,
74];
75
76pub fn extend_selection(root: &SyntaxNode, range: TextRange) -> TextRange {
77    try_extend_selection(root, range).unwrap_or(range)
78}
79
80fn try_extend_selection(root: &SyntaxNode, range: TextRange) -> Option<TextRange> {
81    let string_kinds = [
82        SyntaxKind::COMMENT,
83        SyntaxKind::STRING,
84        SyntaxKind::BYTE_STRING,
85        SyntaxKind::BIT_STRING,
86        SyntaxKind::DOLLAR_QUOTED_STRING,
87        SyntaxKind::ESC_STRING,
88    ];
89
90    if range.is_empty() {
91        let offset = range.start();
92        let mut leaves = root.token_at_offset(offset);
93        // Make sure that if we're on the whitespace at the start of a line, we
94        // expand to the node on that line instead of the previous one
95        if leaves.clone().all(|it| it.kind() == SyntaxKind::WHITESPACE) {
96            return Some(extend_ws(root, leaves.next()?, offset));
97        }
98        let leaf_range = match root.token_at_offset(offset) {
99            rowan::TokenAtOffset::None => return None,
100            rowan::TokenAtOffset::Single(l) => {
101                if string_kinds.contains(&l.kind()) {
102                    extend_single_word_in_comment_or_string(&l, offset)
103                        .unwrap_or_else(|| l.text_range())
104                } else {
105                    l.text_range()
106                }
107            }
108            rowan::TokenAtOffset::Between(l, r) => pick_best(l, r).text_range(),
109        };
110        return Some(leaf_range);
111    }
112
113    let node = match root.covering_element(range) {
114        NodeOrToken::Token(token) => {
115            if token.text_range() != range {
116                return Some(token.text_range());
117            }
118            if let Some(comment) = ast::Comment::cast(token.clone())
119                && let Some(range) = extend_comments(comment)
120            {
121                return Some(range);
122            }
123            token.parent()?
124        }
125        NodeOrToken::Node(node) => node,
126    };
127
128    if node.text_range() != range {
129        return Some(node.text_range());
130    }
131
132    let node = shallowest_node(&node);
133
134    if node
135        .parent()
136        .is_some_and(|n| DELIMITED_LIST_KINDS.contains(&n.kind()))
137    {
138        if let Some(range) = extend_list_item(&node) {
139            return Some(range);
140        }
141    }
142
143    node.parent().map(|it| it.text_range())
144}
145
146/// Find the shallowest node with same range, which allows us to traverse siblings.
147fn shallowest_node(node: &SyntaxNode) -> SyntaxNode {
148    node.ancestors()
149        .take_while(|n| n.text_range() == node.text_range())
150        .last()
151        .unwrap()
152}
153
154/// Expand to the current word instead the full text range of the node.
155fn extend_single_word_in_comment_or_string(
156    leaf: &SyntaxToken,
157    offset: TextSize,
158) -> Option<TextRange> {
159    let text: &str = leaf.text();
160    let cursor_position: u32 = (offset - leaf.text_range().start()).into();
161
162    let (before, after) = text.split_at(cursor_position as usize);
163
164    fn non_word_char(c: char) -> bool {
165        !(c.is_alphanumeric() || c == '_')
166    }
167
168    let start_idx = before.rfind(non_word_char)? as u32;
169    let end_idx = after.find(non_word_char).unwrap_or(after.len()) as u32;
170
171    // FIXME: use `ceil_char_boundary` from `std::str` when it gets stable
172    // https://github.com/rust-lang/rust/issues/93743
173    fn ceil_char_boundary(text: &str, index: u32) -> u32 {
174        (index..)
175            .find(|&index| text.is_char_boundary(index as usize))
176            .unwrap_or(text.len() as u32)
177    }
178
179    let from: TextSize = ceil_char_boundary(text, start_idx + 1).into();
180    let to: TextSize = (cursor_position + end_idx).into();
181
182    let range = TextRange::new(from, to);
183    if range.is_empty() {
184        None
185    } else {
186        Some(range + leaf.text_range().start())
187    }
188}
189
190fn extend_comments(comment: ast::Comment) -> Option<TextRange> {
191    let prev = adj_comments(&comment, Direction::Prev);
192    let next = adj_comments(&comment, Direction::Next);
193    if prev != next {
194        Some(TextRange::new(
195            prev.syntax().text_range().start(),
196            next.syntax().text_range().end(),
197        ))
198    } else {
199        None
200    }
201}
202
203fn adj_comments(comment: &ast::Comment, dir: Direction) -> ast::Comment {
204    let mut res = comment.clone();
205    for element in comment.syntax().siblings_with_tokens(dir) {
206        let Some(token) = element.as_token() else {
207            break;
208        };
209        if let Some(c) = ast::Comment::cast(token.clone()) {
210            res = c
211        } else if token.kind() != SyntaxKind::WHITESPACE || token.text().contains("\n\n") {
212            break;
213        }
214    }
215    res
216}
217
218fn extend_ws(root: &SyntaxNode, ws: SyntaxToken, offset: TextSize) -> TextRange {
219    let ws_text = ws.text();
220    let suffix = TextRange::new(offset, ws.text_range().end()) - ws.text_range().start();
221    let prefix = TextRange::new(ws.text_range().start(), offset) - ws.text_range().start();
222    let ws_suffix = &ws_text[suffix];
223    let ws_prefix = &ws_text[prefix];
224    if ws_text.contains('\n')
225        && !ws_suffix.contains('\n')
226        && let Some(node) = ws.next_sibling_or_token()
227    {
228        let start = match ws_prefix.rfind('\n') {
229            Some(idx) => ws.text_range().start() + TextSize::from((idx + 1) as u32),
230            None => node.text_range().start(),
231        };
232        let end = if root.text().char_at(node.text_range().end()) == Some('\n') {
233            node.text_range().end() + TextSize::of('\n')
234        } else {
235            node.text_range().end()
236        };
237        return TextRange::new(start, end);
238    }
239    ws.text_range()
240}
241
242fn pick_best(l: SyntaxToken, r: SyntaxToken) -> SyntaxToken {
243    return if priority(&r) > priority(&l) { r } else { l };
244    fn priority(n: &SyntaxToken) -> usize {
245        match n.kind() {
246            SyntaxKind::WHITESPACE => 0,
247            // TODO: we can probably include more here, rust analyzer includes a
248            // handful of keywords
249            SyntaxKind::IDENT => 2,
250            _ => 1,
251        }
252    }
253}
254
255/// Extend list item selection to include nearby delimiter and whitespace.
256fn extend_list_item(node: &SyntaxNode) -> Option<TextRange> {
257    fn is_single_line_ws(node: &SyntaxToken) -> bool {
258        node.kind() == SyntaxKind::WHITESPACE && !node.text().contains('\n')
259    }
260
261    fn nearby_comma(node: &SyntaxNode, dir: Direction) -> Option<SyntaxToken> {
262        node.siblings_with_tokens(dir)
263            .skip(1)
264            .find(|node| match node {
265                NodeOrToken::Node(_) => true,
266                NodeOrToken::Token(it) => !is_single_line_ws(it),
267            })
268            .and_then(|it| it.into_token())
269            .filter(|node| node.kind() == SyntaxKind::COMMA)
270    }
271
272    if let Some(comma) = nearby_comma(node, Direction::Next) {
273        // Include any following whitespace when delimiter is after list item.
274        let final_node = comma
275            .next_sibling_or_token()
276            .and_then(|n| n.into_token())
277            .filter(is_single_line_ws)
278            .unwrap_or(comma);
279
280        return Some(TextRange::new(
281            node.text_range().start(),
282            final_node.text_range().end(),
283        ));
284    }
285
286    if let Some(comma) = nearby_comma(node, Direction::Prev) {
287        return Some(TextRange::new(
288            comma.text_range().start(),
289            node.text_range().end(),
290        ));
291    }
292
293    None
294}
295
296#[cfg(test)]
297mod tests {
298    use super::*;
299    use crate::test_utils::fixture;
300    use insta::assert_debug_snapshot;
301    use squawk_syntax::{SourceFile, ast::AstNode};
302
303    fn expand(sql: &str) -> Vec<String> {
304        let (offset, sql) = fixture(sql);
305        let parse = SourceFile::parse(&sql);
306        let file = parse.tree();
307        let root = file.syntax();
308
309        let mut range = TextRange::empty(offset);
310        let mut results = Vec::new();
311
312        for _ in 0..20 {
313            let new_range = extend_selection(root, range);
314            if new_range == range {
315                break;
316            }
317            range = new_range;
318            results.push(sql[range].to_string());
319        }
320
321        results
322    }
323
324    #[test]
325    fn simple() {
326        assert_debug_snapshot!(expand(r#"select $01 + 1"#), @r#"
327        [
328            "1",
329            "1 + 1",
330            "select 1 + 1",
331        ]
332        "#);
333    }
334
335    #[test]
336    fn word_in_string_string() {
337        assert_debug_snapshot!(expand(r"
338select 'some stret$0ched out words in a string'
339"), @r#"
340        [
341            "stretched",
342            "'some stretched out words in a string'",
343            "select 'some stretched out words in a string'",
344            "\nselect 'some stretched out words in a string'\n",
345        ]
346        "#);
347    }
348
349    #[test]
350    fn string() {
351        assert_debug_snapshot!(expand(r"
352select b'foo$0 bar'
353'buzz';
354"), @r#"
355        [
356            "foo",
357            "b'foo bar'",
358            "b'foo bar'\n'buzz'",
359            "select b'foo bar'\n'buzz'",
360            "\nselect b'foo bar'\n'buzz';\n",
361        ]
362        "#);
363    }
364
365    #[test]
366    fn dollar_string() {
367        assert_debug_snapshot!(expand(r"
368select $$foo$0 bar$$;
369"), @r#"
370        [
371            "foo",
372            "$$foo bar$$",
373            "select $$foo bar$$",
374            "\nselect $$foo bar$$;\n",
375        ]
376        "#);
377    }
378
379    #[test]
380    fn comment_muli_line() {
381        assert_debug_snapshot!(expand(r"
382-- foo bar
383-- buzz$0
384-- boo
385select 1
386"), @r#"
387        [
388            "-- buzz",
389            "-- foo bar\n-- buzz\n-- boo",
390            "\n-- foo bar\n-- buzz\n-- boo\nselect 1\n",
391        ]
392        "#);
393    }
394
395    #[test]
396    fn comment() {
397        assert_debug_snapshot!(expand(r"
398-- foo bar$0
399select 1
400"), @r#"
401        [
402            "-- foo bar",
403            "\n-- foo bar\nselect 1\n",
404        ]
405        "#);
406
407        assert_debug_snapshot!(expand(r"
408/* foo bar$0 */
409select 1
410"), @r#"
411        [
412            "bar",
413            "/* foo bar */",
414            "\n/* foo bar */\nselect 1\n",
415        ]
416        "#);
417    }
418
419    #[test]
420    fn create_table_with_comment() {
421        assert_debug_snapshot!(expand(r"
422-- foo bar buzz
423create table t(
424  x int$0,
425  y text
426);
427"), @r#"
428        [
429            "int",
430            "x int",
431            "x int,",
432            "(\n  x int,\n  y text\n)",
433            "-- foo bar buzz\ncreate table t(\n  x int,\n  y text\n)",
434            "\n-- foo bar buzz\ncreate table t(\n  x int,\n  y text\n);\n",
435        ]
436        "#);
437    }
438
439    #[test]
440    fn column_list() {
441        assert_debug_snapshot!(expand(r#"create table t($0x int)"#), @r#"
442        [
443            "x",
444            "x int",
445            "(x int)",
446            "create table t(x int)",
447        ]
448        "#);
449
450        assert_debug_snapshot!(expand(r#"create table t($0x int, y int)"#), @r#"
451        [
452            "x",
453            "x int",
454            "x int, ",
455            "(x int, y int)",
456            "create table t(x int, y int)",
457        ]
458        "#);
459
460        assert_debug_snapshot!(expand(r#"create table t(x int, $0y int)"#), @r#"
461        [
462            "y",
463            "y int",
464            ", y int",
465            "(x int, y int)",
466            "create table t(x int, y int)",
467        ]
468        "#);
469    }
470
471    #[test]
472    fn start_of_line_whitespace_select() {
473        assert_debug_snapshot!(expand(r#"    
474select 1;
475
476$0    select 2;"#), @r#"
477        [
478            "    select 2",
479            "    \nselect 1;\n\n    select 2;",
480        ]
481        "#);
482    }
483
484    #[test]
485    fn select_list() {
486        assert_debug_snapshot!(expand(r#"select x$0, y from t"#), @r#"
487        [
488            "x",
489            "x, ",
490            "x, y",
491            "select x, y",
492            "select x, y from t",
493        ]
494        "#);
495
496        assert_debug_snapshot!(expand(r#"select x, y$0 from t"#), @r#"
497        [
498            "y",
499            ", y",
500            "x, y",
501            "select x, y",
502            "select x, y from t",
503        ]
504        "#);
505    }
506
507    #[test]
508    fn expand_whitespace() {
509        assert_debug_snapshot!(expand(r#"select 1 + 
510$0
5111;"#), @r#"
512        [
513            " \n\n",
514            "1 + \n\n1",
515            "select 1 + \n\n1",
516            "select 1 + \n\n1;",
517        ]
518        "#);
519    }
520
521    #[test]
522    fn function_args() {
523        assert_debug_snapshot!(expand(r#"select f(1$0, 2)"#), @r#"
524        [
525            "1",
526            "1, ",
527            "(1, 2)",
528            "f(1, 2)",
529            "select f(1, 2)",
530        ]
531        "#);
532    }
533
534    #[test]
535    fn prefer_idents() {
536        assert_debug_snapshot!(expand(r#"select foo$0+bar"#), @r#"
537        [
538            "foo",
539            "foo+bar",
540            "select foo+bar",
541        ]
542        "#);
543
544        assert_debug_snapshot!(expand(r#"select foo+$0bar"#), @r#"
545        [
546            "bar",
547            "foo+bar",
548            "select foo+bar",
549        ]
550        "#);
551    }
552
553    #[test]
554    fn list_variants() {
555        let delimited_ws_list_kinds = &[
556            SyntaxKind::CREATE_DATABASE_OPTION_LIST,
557            SyntaxKind::FUNC_OPTION_LIST,
558            SyntaxKind::ROLE_OPTION_LIST,
559            SyntaxKind::SEQUENCE_OPTION_LIST,
560            SyntaxKind::TRIGGER_EVENT_LIST,
561            SyntaxKind::XML_COLUMN_OPTION_LIST,
562            SyntaxKind::WHEN_CLAUSE_LIST,
563        ];
564
565        let unhandled_list_kinds = (0..SyntaxKind::__LAST as u16)
566            .map(SyntaxKind::from)
567            .filter(|kind| {
568                format!("{:?}", kind).ends_with("_LIST") && !delimited_ws_list_kinds.contains(kind)
569            })
570            .filter(|kind| !DELIMITED_LIST_KINDS.contains(kind))
571            .collect::<Vec<_>>();
572
573        assert_eq!(
574            unhandled_list_kinds,
575            vec![],
576            "We shouldn't have any unhandled list kinds"
577        )
578    }
579}
squawk_ide/expand_selection.rs

squawk_ide/
expand_selection.rs
