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::COLUMN_LIST,
41    SyntaxKind::CONFLICT_INDEX_ITEM_LIST,
42    SyntaxKind::CONSTRAINT_EXCLUSION_LIST,
43    SyntaxKind::DROP_OP_CLASS_OPTION_LIST,
44    SyntaxKind::FDW_OPTION_LIST,
45    SyntaxKind::FUNCTION_SIG_LIST,
46    SyntaxKind::GROUP_BY_LIST,
47    SyntaxKind::JSON_TABLE_COLUMN_LIST,
48    SyntaxKind::OPERATOR_CLASS_OPTION_LIST,
49    SyntaxKind::OPTION_ITEM_LIST,
50    SyntaxKind::OP_SIG_LIST,
51    SyntaxKind::PARAM_LIST,
52    SyntaxKind::PARTITION_ITEM_LIST,
53    SyntaxKind::RETURNING_OPTION_LIST,
54    SyntaxKind::REVOKE_COMMAND_LIST,
55    SyntaxKind::ROLE_LIST,
56    SyntaxKind::ROW_LIST,
57    SyntaxKind::XML_ATTRIBUTE_LIST,
58    SyntaxKind::XML_NAMESPACE_LIST,
59    SyntaxKind::SET_COLUMN_LIST,
60    SyntaxKind::SET_EXPR_LIST,
61    SyntaxKind::SET_OPTIONS_LIST,
62    SyntaxKind::SORT_BY_LIST,
63    SyntaxKind::TABLE_ARG_LIST,
64    SyntaxKind::TABLE_LIST,
65    SyntaxKind::TARGET_LIST,
66    SyntaxKind::TRANSACTION_MODE_LIST,
67    SyntaxKind::VACUUM_OPTION_LIST,
68    SyntaxKind::VARIANT_LIST,
69    SyntaxKind::XML_TABLE_COLUMN_LIST,
70];
71
72pub fn extend_selection(root: &SyntaxNode, range: TextRange) -> TextRange {
73    try_extend_selection(root, range).unwrap_or(range)
74}
75
76fn try_extend_selection(root: &SyntaxNode, range: TextRange) -> Option<TextRange> {
77    let string_kinds = [
78        SyntaxKind::COMMENT,
79        SyntaxKind::STRING,
80        SyntaxKind::BYTE_STRING,
81        SyntaxKind::BIT_STRING,
82        SyntaxKind::DOLLAR_QUOTED_STRING,
83        SyntaxKind::ESC_STRING,
84    ];
85
86    if range.is_empty() {
87        let offset = range.start();
88        let mut leaves = root.token_at_offset(offset);
89        // Make sure that if we're on the whitespace at the start of a line, we
90        // expand to the node on that line instead of the previous one
91        if leaves.clone().all(|it| it.kind() == SyntaxKind::WHITESPACE) {
92            return Some(extend_ws(root, leaves.next()?, offset));
93        }
94        let leaf_range = match root.token_at_offset(offset) {
95            rowan::TokenAtOffset::None => return None,
96            rowan::TokenAtOffset::Single(l) => {
97                if string_kinds.contains(&l.kind()) {
98                    extend_single_word_in_comment_or_string(&l, offset)
99                        .unwrap_or_else(|| l.text_range())
100                } else {
101                    l.text_range()
102                }
103            }
104            rowan::TokenAtOffset::Between(l, r) => pick_best(l, r).text_range(),
105        };
106        return Some(leaf_range);
107    }
108
109    let node = match root.covering_element(range) {
110        NodeOrToken::Token(token) => {
111            if token.text_range() != range {
112                return Some(token.text_range());
113            }
114            if let Some(comment) = ast::Comment::cast(token.clone())
115                && let Some(range) = extend_comments(comment)
116            {
117                return Some(range);
118            }
119            token.parent()?
120        }
121        NodeOrToken::Node(node) => node,
122    };
123
124    if node.text_range() != range {
125        return Some(node.text_range());
126    }
127
128    let node = shallowest_node(&node);
129
130    if node
131        .parent()
132        .is_some_and(|n| DELIMITED_LIST_KINDS.contains(&n.kind()))
133    {
134        if let Some(range) = extend_list_item(&node) {
135            return Some(range);
136        }
137    }
138
139    node.parent().map(|it| it.text_range())
140}
141
142/// Find the shallowest node with same range, which allows us to traverse siblings.
143fn shallowest_node(node: &SyntaxNode) -> SyntaxNode {
144    node.ancestors()
145        .take_while(|n| n.text_range() == node.text_range())
146        .last()
147        .unwrap()
148}
149
150/// Expand to the current word instead the full text range of the node.
151fn extend_single_word_in_comment_or_string(
152    leaf: &SyntaxToken,
153    offset: TextSize,
154) -> Option<TextRange> {
155    let text: &str = leaf.text();
156    let cursor_position: u32 = (offset - leaf.text_range().start()).into();
157
158    let (before, after) = text.split_at(cursor_position as usize);
159
160    fn non_word_char(c: char) -> bool {
161        !(c.is_alphanumeric() || c == '_')
162    }
163
164    let start_idx = before.rfind(non_word_char)? as u32;
165    let end_idx = after.find(non_word_char).unwrap_or(after.len()) as u32;
166
167    // FIXME: use `ceil_char_boundary` from `std::str` when it gets stable
168    // https://github.com/rust-lang/rust/issues/93743
169    fn ceil_char_boundary(text: &str, index: u32) -> u32 {
170        (index..)
171            .find(|&index| text.is_char_boundary(index as usize))
172            .unwrap_or(text.len() as u32)
173    }
174
175    let from: TextSize = ceil_char_boundary(text, start_idx + 1).into();
176    let to: TextSize = (cursor_position + end_idx).into();
177
178    let range = TextRange::new(from, to);
179    if range.is_empty() {
180        None
181    } else {
182        Some(range + leaf.text_range().start())
183    }
184}
185
186fn extend_comments(comment: ast::Comment) -> Option<TextRange> {
187    let prev = adj_comments(&comment, Direction::Prev);
188    let next = adj_comments(&comment, Direction::Next);
189    if prev != next {
190        Some(TextRange::new(
191            prev.syntax().text_range().start(),
192            next.syntax().text_range().end(),
193        ))
194    } else {
195        None
196    }
197}
198
199fn adj_comments(comment: &ast::Comment, dir: Direction) -> ast::Comment {
200    let mut res = comment.clone();
201    for element in comment.syntax().siblings_with_tokens(dir) {
202        let Some(token) = element.as_token() else {
203            break;
204        };
205        if let Some(c) = ast::Comment::cast(token.clone()) {
206            res = c
207        } else if token.kind() != SyntaxKind::WHITESPACE || token.text().contains("\n\n") {
208            break;
209        }
210    }
211    res
212}
213
214fn extend_ws(root: &SyntaxNode, ws: SyntaxToken, offset: TextSize) -> TextRange {
215    let ws_text = ws.text();
216    let suffix = TextRange::new(offset, ws.text_range().end()) - ws.text_range().start();
217    let prefix = TextRange::new(ws.text_range().start(), offset) - ws.text_range().start();
218    let ws_suffix = &ws_text[suffix];
219    let ws_prefix = &ws_text[prefix];
220    if ws_text.contains('\n')
221        && !ws_suffix.contains('\n')
222        && let Some(node) = ws.next_sibling_or_token()
223    {
224        let start = match ws_prefix.rfind('\n') {
225            Some(idx) => ws.text_range().start() + TextSize::from((idx + 1) as u32),
226            None => node.text_range().start(),
227        };
228        let end = if root.text().char_at(node.text_range().end()) == Some('\n') {
229            node.text_range().end() + TextSize::of('\n')
230        } else {
231            node.text_range().end()
232        };
233        return TextRange::new(start, end);
234    }
235    ws.text_range()
236}
237
238fn pick_best(l: SyntaxToken, r: SyntaxToken) -> SyntaxToken {
239    return if priority(&r) > priority(&l) { r } else { l };
240    fn priority(n: &SyntaxToken) -> usize {
241        match n.kind() {
242            SyntaxKind::WHITESPACE => 0,
243            // TODO: we can probably include more here, rust analyzer includes a
244            // handful of keywords
245            SyntaxKind::IDENT => 2,
246            _ => 1,
247        }
248    }
249}
250
251/// Extend list item selection to include nearby delimiter and whitespace.
252fn extend_list_item(node: &SyntaxNode) -> Option<TextRange> {
253    fn is_single_line_ws(node: &SyntaxToken) -> bool {
254        node.kind() == SyntaxKind::WHITESPACE && !node.text().contains('\n')
255    }
256
257    fn nearby_comma(node: &SyntaxNode, dir: Direction) -> Option<SyntaxToken> {
258        node.siblings_with_tokens(dir)
259            .skip(1)
260            .find(|node| match node {
261                NodeOrToken::Node(_) => true,
262                NodeOrToken::Token(it) => !is_single_line_ws(it),
263            })
264            .and_then(|it| it.into_token())
265            .filter(|node| node.kind() == SyntaxKind::COMMA)
266    }
267
268    if let Some(comma) = nearby_comma(node, Direction::Next) {
269        // Include any following whitespace when delimiter is after list item.
270        let final_node = comma
271            .next_sibling_or_token()
272            .and_then(|n| n.into_token())
273            .filter(is_single_line_ws)
274            .unwrap_or(comma);
275
276        return Some(TextRange::new(
277            node.text_range().start(),
278            final_node.text_range().end(),
279        ));
280    }
281
282    if let Some(comma) = nearby_comma(node, Direction::Prev) {
283        return Some(TextRange::new(
284            comma.text_range().start(),
285            node.text_range().end(),
286        ));
287    }
288
289    None
290}
291
292#[cfg(test)]
293mod tests {
294    use super::*;
295    use crate::test_utils::fixture;
296    use insta::assert_debug_snapshot;
297    use squawk_syntax::{SourceFile, ast::AstNode};
298
299    fn expand(sql: &str) -> Vec<String> {
300        let (offset, sql) = fixture(sql);
301        let parse = SourceFile::parse(&sql);
302        let file = parse.tree();
303        let root = file.syntax();
304
305        let mut range = TextRange::empty(offset);
306        let mut results = Vec::new();
307
308        for _ in 0..20 {
309            let new_range = extend_selection(root, range);
310            if new_range == range {
311                break;
312            }
313            range = new_range;
314            results.push(sql[range].to_string());
315        }
316
317        results
318    }
319
320    #[test]
321    fn simple() {
322        assert_debug_snapshot!(expand(r#"select $01 + 1"#), @r#"
323        [
324            "1",
325            "1 + 1",
326            "select 1 + 1",
327        ]
328        "#);
329    }
330
331    #[test]
332    fn word_in_string_string() {
333        assert_debug_snapshot!(expand(r"
334select 'some stret$0ched out words in a string'
335"), @r#"
336        [
337            "stretched",
338            "'some stretched out words in a string'",
339            "select 'some stretched out words in a string'",
340            "\nselect 'some stretched out words in a string'\n",
341        ]
342        "#);
343    }
344
345    #[test]
346    fn string() {
347        assert_debug_snapshot!(expand(r"
348select b'foo$0 bar'
349'buzz';
350"), @r#"
351        [
352            "foo",
353            "b'foo bar'",
354            "b'foo bar'\n'buzz'",
355            "select b'foo bar'\n'buzz'",
356            "\nselect b'foo bar'\n'buzz';\n",
357        ]
358        "#);
359    }
360
361    #[test]
362    fn dollar_string() {
363        assert_debug_snapshot!(expand(r"
364select $$foo$0 bar$$;
365"), @r#"
366        [
367            "foo",
368            "$$foo bar$$",
369            "select $$foo bar$$",
370            "\nselect $$foo bar$$;\n",
371        ]
372        "#);
373    }
374
375    #[test]
376    fn comment_muli_line() {
377        assert_debug_snapshot!(expand(r"
378-- foo bar
379-- buzz$0
380-- boo
381select 1
382"), @r#"
383        [
384            "-- buzz",
385            "-- foo bar\n-- buzz\n-- boo",
386            "\n-- foo bar\n-- buzz\n-- boo\nselect 1\n",
387        ]
388        "#);
389    }
390
391    #[test]
392    fn comment() {
393        assert_debug_snapshot!(expand(r"
394-- foo bar$0
395select 1
396"), @r#"
397        [
398            "-- foo bar",
399            "\n-- foo bar\nselect 1\n",
400        ]
401        "#);
402
403        assert_debug_snapshot!(expand(r"
404/* foo bar$0 */
405select 1
406"), @r#"
407        [
408            "bar",
409            "/* foo bar */",
410            "\n/* foo bar */\nselect 1\n",
411        ]
412        "#);
413    }
414
415    #[test]
416    fn create_table_with_comment() {
417        assert_debug_snapshot!(expand(r"
418-- foo bar buzz
419create table t(
420  x int$0,
421  y text
422);
423"), @r#"
424        [
425            "int",
426            "x int",
427            "x int,",
428            "(\n  x int,\n  y text\n)",
429            "-- foo bar buzz\ncreate table t(\n  x int,\n  y text\n)",
430            "\n-- foo bar buzz\ncreate table t(\n  x int,\n  y text\n);\n",
431        ]
432        "#);
433    }
434
435    #[test]
436    fn column_list() {
437        assert_debug_snapshot!(expand(r#"create table t($0x int)"#), @r#"
438        [
439            "x",
440            "x int",
441            "(x int)",
442            "create table t(x int)",
443        ]
444        "#);
445
446        assert_debug_snapshot!(expand(r#"create table t($0x int, y int)"#), @r#"
447        [
448            "x",
449            "x int",
450            "x int, ",
451            "(x int, y int)",
452            "create table t(x int, y int)",
453        ]
454        "#);
455
456        assert_debug_snapshot!(expand(r#"create table t(x int, $0y int)"#), @r#"
457        [
458            "y",
459            "y int",
460            ", y int",
461            "(x int, y int)",
462            "create table t(x int, y int)",
463        ]
464        "#);
465    }
466
467    #[test]
468    fn start_of_line_whitespace_select() {
469        assert_debug_snapshot!(expand(r#"    
470select 1;
471
472$0    select 2;"#), @r#"
473        [
474            "    select 2",
475            "    \nselect 1;\n\n    select 2;",
476        ]
477        "#);
478    }
479
480    #[test]
481    fn select_list() {
482        assert_debug_snapshot!(expand(r#"select x$0, y from t"#), @r#"
483        [
484            "x",
485            "x, ",
486            "x, y",
487            "select x, y",
488            "select x, y from t",
489        ]
490        "#);
491
492        assert_debug_snapshot!(expand(r#"select x, y$0 from t"#), @r#"
493        [
494            "y",
495            ", y",
496            "x, y",
497            "select x, y",
498            "select x, y from t",
499        ]
500        "#);
501    }
502
503    #[test]
504    fn expand_whitespace() {
505        assert_debug_snapshot!(expand(r#"select 1 + 
506$0
5071;"#), @r#"
508        [
509            " \n\n",
510            "1 + \n\n1",
511            "select 1 + \n\n1",
512            "select 1 + \n\n1;",
513        ]
514        "#);
515    }
516
517    #[test]
518    fn function_args() {
519        assert_debug_snapshot!(expand(r#"select f(1$0, 2)"#), @r#"
520        [
521            "1",
522            "1, ",
523            "(1, 2)",
524            "f(1, 2)",
525            "select f(1, 2)",
526        ]
527        "#);
528    }
529
530    #[test]
531    fn prefer_idents() {
532        assert_debug_snapshot!(expand(r#"select foo$0+bar"#), @r#"
533        [
534            "foo",
535            "foo+bar",
536            "select foo+bar",
537        ]
538        "#);
539
540        assert_debug_snapshot!(expand(r#"select foo+$0bar"#), @r#"
541        [
542            "bar",
543            "foo+bar",
544            "select foo+bar",
545        ]
546        "#);
547    }
548
549    #[test]
550    fn list_variants() {
551        let delimited_ws_list_kinds = &[
552            SyntaxKind::CREATE_DATABASE_OPTION_LIST,
553            SyntaxKind::FUNC_OPTION_LIST,
554            SyntaxKind::ROLE_OPTION_LIST,
555            SyntaxKind::SEQUENCE_OPTION_LIST,
556            SyntaxKind::TRIGGER_EVENT_LIST,
557            SyntaxKind::XML_COLUMN_OPTION_LIST,
558            SyntaxKind::WHEN_CLAUSE_LIST,
559        ];
560
561        let unhandled_list_kinds = (0..SyntaxKind::__LAST as u16)
562            .map(SyntaxKind::from)
563            .filter(|kind| {
564                format!("{:?}", kind).ends_with("_LIST") && !delimited_ws_list_kinds.contains(kind)
565            })
566            .filter(|kind| !DELIMITED_LIST_KINDS.contains(kind))
567            .collect::<Vec<_>>();
568
569        assert_eq!(
570            unhandled_list_kinds,
571            vec![],
572            "We shouldn't have any unhandled list kinds"
573        )
574    }
575}
squawk_ide/expand_selection.rs

squawk_ide/
expand_selection.rs
