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
36use crate::tokens::is_string_or_comment;
37
38const DELIMITED_LIST_KINDS: &[SyntaxKind] = &[
39    SyntaxKind::ALTER_OPTION_LIST,
40    SyntaxKind::ARG_LIST,
41    SyntaxKind::ATTRIBUTE_LIST,
42    SyntaxKind::BEGIN_FUNC_OPTION_LIST,
43    SyntaxKind::COLUMN_LIST,
44    SyntaxKind::CONFLICT_INDEX_ITEM_LIST,
45    SyntaxKind::CONSTRAINT_EXCLUSION_LIST,
46    SyntaxKind::COPY_OPTION_LIST,
47    SyntaxKind::DROP_OP_CLASS_OPTION_LIST,
48    SyntaxKind::FDW_OPTION_LIST,
49    SyntaxKind::FUNCTION_SIG_LIST,
50    SyntaxKind::GROUP_BY_LIST,
51    SyntaxKind::JSON_TABLE_COLUMN_LIST,
52    SyntaxKind::OPERATOR_CLASS_OPTION_LIST,
53    SyntaxKind::OPTION_ITEM_LIST,
54    SyntaxKind::OP_SIG_LIST,
55    SyntaxKind::PARAM_LIST,
56    SyntaxKind::PARTITION_ITEM_LIST,
57    SyntaxKind::PARTITION_LIST,
58    SyntaxKind::RETURNING_OPTION_LIST,
59    SyntaxKind::REVOKE_COMMAND_LIST,
60    SyntaxKind::ROLE_REF_LIST,
61    SyntaxKind::ROW_LIST,
62    SyntaxKind::EXPR_AS_NAME_LIST,
63    SyntaxKind::XML_NAMESPACE_LIST,
64    SyntaxKind::SET_COLUMN_LIST,
65    SyntaxKind::SET_EXPR_LIST,
66    SyntaxKind::SET_OPTIONS_LIST,
67    SyntaxKind::SORT_BY_LIST,
68    SyntaxKind::TABLE_AND_COLUMNS_LIST,
69    SyntaxKind::TABLE_ARG_LIST,
70    SyntaxKind::TABLE_LIST,
71    SyntaxKind::TARGET_LIST,
72    SyntaxKind::TRANSACTION_MODE_LIST,
73    SyntaxKind::VACUUM_OPTION_LIST,
74    SyntaxKind::VARIANT_LIST,
75    SyntaxKind::XML_TABLE_COLUMN_LIST,
76    SyntaxKind::PATH_PATTERN_LIST,
77];
78
79pub fn extend_selection(root: &SyntaxNode, range: TextRange) -> TextRange {
80    try_extend_selection(root, range).unwrap_or(range)
81}
82
83fn try_extend_selection(root: &SyntaxNode, range: TextRange) -> Option<TextRange> {
84    if range.is_empty() {
85        let offset = range.start();
86        let mut leaves = root.token_at_offset(offset);
87        // Make sure that if we're on the whitespace at the start of a line, we
88        // expand to the node on that line instead of the previous one
89        if leaves.clone().all(|it| it.kind() == SyntaxKind::WHITESPACE) {
90            return Some(extend_ws(root, leaves.next()?, offset));
91        }
92        let leaf_range = match root.token_at_offset(offset) {
93            rowan::TokenAtOffset::None => return None,
94            rowan::TokenAtOffset::Single(l) => {
95                if is_string_or_comment(l.kind()) {
96                    extend_single_word_in_comment_or_string(&l, offset)
97                        .unwrap_or_else(|| l.text_range())
98                } else {
99                    l.text_range()
100                }
101            }
102            rowan::TokenAtOffset::Between(l, r) => pick_best(l, r).text_range(),
103        };
104        return Some(leaf_range);
105    }
106
107    let node = match root.covering_element(range) {
108        NodeOrToken::Token(token) => {
109            if token.text_range() != range {
110                return Some(token.text_range());
111            }
112            if let Some(comment) = ast::Comment::cast(token.clone())
113                && let Some(range) = extend_comments(comment)
114            {
115                return Some(range);
116            }
117            token.parent()?
118        }
119        NodeOrToken::Node(node) => node,
120    };
121
122    if node.text_range() != range {
123        return Some(node.text_range());
124    }
125
126    let node = shallowest_node(&node);
127
128    if node
129        .parent()
130        .is_some_and(|n| DELIMITED_LIST_KINDS.contains(&n.kind()))
131    {
132        if let Some(range) = extend_list_item(&node) {
133            return Some(range);
134        }
135    }
136
137    node.parent().map(|it| it.text_range())
138}
139
140/// Find the shallowest node with same range, which allows us to traverse siblings.
141fn shallowest_node(node: &SyntaxNode) -> SyntaxNode {
142    node.ancestors()
143        .take_while(|n| n.text_range() == node.text_range())
144        .last()
145        .unwrap()
146}
147
148/// Expand to the current word instead the full text range of the node.
149fn extend_single_word_in_comment_or_string(
150    leaf: &SyntaxToken,
151    offset: TextSize,
152) -> Option<TextRange> {
153    let text: &str = leaf.text();
154    let cursor_position: u32 = (offset - leaf.text_range().start()).into();
155
156    let (before, after) = text.split_at(cursor_position as usize);
157
158    fn non_word_char(c: char) -> bool {
159        !(c.is_alphanumeric() || c == '_')
160    }
161
162    let start_idx = before.rfind(non_word_char)? as u32;
163    let end_idx = after.find(non_word_char).unwrap_or(after.len()) as u32;
164
165    // FIXME: use `ceil_char_boundary` from `std::str` when it gets stable
166    // https://github.com/rust-lang/rust/issues/93743
167    fn ceil_char_boundary(text: &str, index: u32) -> u32 {
168        (index..)
169            .find(|&index| text.is_char_boundary(index as usize))
170            .unwrap_or(text.len() as u32)
171    }
172
173    let from: TextSize = ceil_char_boundary(text, start_idx + 1).into();
174    let to: TextSize = (cursor_position + end_idx).into();
175
176    let range = TextRange::new(from, to);
177    if range.is_empty() {
178        None
179    } else {
180        Some(range + leaf.text_range().start())
181    }
182}
183
184fn extend_comments(comment: ast::Comment) -> Option<TextRange> {
185    let prev = adj_comments(&comment, Direction::Prev);
186    let next = adj_comments(&comment, Direction::Next);
187    if prev != next {
188        Some(TextRange::new(
189            prev.syntax().text_range().start(),
190            next.syntax().text_range().end(),
191        ))
192    } else {
193        None
194    }
195}
196
197fn adj_comments(comment: &ast::Comment, dir: Direction) -> ast::Comment {
198    let mut res = comment.clone();
199    for element in comment.syntax().siblings_with_tokens(dir) {
200        let Some(token) = element.as_token() else {
201            break;
202        };
203        if let Some(c) = ast::Comment::cast(token.clone()) {
204            res = c
205        } else if token.kind() != SyntaxKind::WHITESPACE || token.text().contains("\n\n") {
206            break;
207        }
208    }
209    res
210}
211
212fn extend_ws(root: &SyntaxNode, ws: SyntaxToken, offset: TextSize) -> TextRange {
213    let ws_text = ws.text();
214    let suffix = TextRange::new(offset, ws.text_range().end()) - ws.text_range().start();
215    let prefix = TextRange::new(ws.text_range().start(), offset) - ws.text_range().start();
216    let ws_suffix = &ws_text[suffix];
217    let ws_prefix = &ws_text[prefix];
218    if ws_text.contains('\n')
219        && !ws_suffix.contains('\n')
220        && let Some(node) = ws.next_sibling_or_token()
221    {
222        let start = match ws_prefix.rfind('\n') {
223            Some(idx) => ws.text_range().start() + TextSize::from((idx + 1) as u32),
224            None => node.text_range().start(),
225        };
226        let end = if root.text().char_at(node.text_range().end()) == Some('\n') {
227            node.text_range().end() + TextSize::of('\n')
228        } else {
229            node.text_range().end()
230        };
231        return TextRange::new(start, end);
232    }
233    ws.text_range()
234}
235
236fn pick_best(l: SyntaxToken, r: SyntaxToken) -> SyntaxToken {
237    return if priority(&r) > priority(&l) { r } else { l };
238    fn priority(n: &SyntaxToken) -> usize {
239        match n.kind() {
240            SyntaxKind::WHITESPACE => 0,
241            // TODO: we can probably include more here, rust analyzer includes a
242            // handful of keywords
243            SyntaxKind::IDENT => 2,
244            _ => 1,
245        }
246    }
247}
248
249/// Extend list item selection to include nearby delimiter and whitespace.
250fn extend_list_item(node: &SyntaxNode) -> Option<TextRange> {
251    fn is_single_line_ws(node: &SyntaxToken) -> bool {
252        node.kind() == SyntaxKind::WHITESPACE && !node.text().contains('\n')
253    }
254
255    fn nearby_comma(node: &SyntaxNode, dir: Direction) -> Option<SyntaxToken> {
256        node.siblings_with_tokens(dir)
257            .skip(1)
258            .find(|node| match node {
259                NodeOrToken::Node(_) => true,
260                NodeOrToken::Token(it) => !is_single_line_ws(it),
261            })
262            .and_then(|it| it.into_token())
263            .filter(|node| node.kind() == SyntaxKind::COMMA)
264    }
265
266    if let Some(comma) = nearby_comma(node, Direction::Next) {
267        // Include any following whitespace when delimiter is after list item.
268        let final_node = comma
269            .next_sibling_or_token()
270            .and_then(|n| n.into_token())
271            .filter(is_single_line_ws)
272            .unwrap_or(comma);
273
274        return Some(TextRange::new(
275            node.text_range().start(),
276            final_node.text_range().end(),
277        ));
278    }
279
280    if let Some(comma) = nearby_comma(node, Direction::Prev) {
281        return Some(TextRange::new(
282            comma.text_range().start(),
283            node.text_range().end(),
284        ));
285    }
286
287    None
288}
289
290#[cfg(test)]
291mod tests {
292    use super::*;
293    use crate::test_utils::fixture;
294    use insta::assert_debug_snapshot;
295    use squawk_syntax::{SourceFile, ast::AstNode};
296
297    fn expand(sql: &str) -> Vec<String> {
298        let (offset, sql) = fixture(sql);
299        let parse = SourceFile::parse(&sql);
300        let file = parse.tree();
301        let root = file.syntax();
302
303        let mut range = TextRange::empty(offset);
304        let mut results = vec![];
305
306        for _ in 0..20 {
307            let new_range = extend_selection(root, range);
308            if new_range == range {
309                break;
310            }
311            range = new_range;
312            results.push(sql[range].to_string());
313        }
314
315        results
316    }
317
318    #[test]
319    fn simple() {
320        assert_debug_snapshot!(expand(r#"select $01 + 1"#), @r#"
321        [
322            "1",
323            "1 + 1",
324            "select 1 + 1",
325        ]
326        "#);
327    }
328
329    #[test]
330    fn word_in_string_string() {
331        assert_debug_snapshot!(expand(r"
332select 'some stret$0ched out words in a string'
333"), @r#"
334        [
335            "stretched",
336            "'some stretched out words in a string'",
337            "select 'some stretched out words in a string'",
338            "\nselect 'some stretched out words in a string'\n",
339        ]
340        "#);
341    }
342
343    #[test]
344    fn string() {
345        assert_debug_snapshot!(expand(r"
346select b'foo$0 bar'
347'buzz';
348"), @r#"
349        [
350            "foo",
351            "b'foo bar'",
352            "b'foo bar'\n'buzz'",
353            "select b'foo bar'\n'buzz'",
354            "\nselect b'foo bar'\n'buzz';\n",
355        ]
356        "#);
357    }
358
359    #[test]
360    fn dollar_string() {
361        assert_debug_snapshot!(expand(r"
362select $$foo$0 bar$$;
363"), @r#"
364        [
365            "foo",
366            "$$foo bar$$",
367            "select $$foo bar$$",
368            "\nselect $$foo bar$$;\n",
369        ]
370        "#);
371    }
372
373    #[test]
374    fn comment_muli_line() {
375        assert_debug_snapshot!(expand(r"
376-- foo bar
377-- buzz$0
378-- boo
379select 1
380"), @r#"
381        [
382            "-- buzz",
383            "-- foo bar\n-- buzz\n-- boo",
384            "\n-- foo bar\n-- buzz\n-- boo\nselect 1\n",
385        ]
386        "#);
387    }
388
389    #[test]
390    fn comment() {
391        assert_debug_snapshot!(expand(r"
392-- foo bar$0
393select 1
394"), @r#"
395        [
396            "-- foo bar",
397            "\n-- foo bar\nselect 1\n",
398        ]
399        "#);
400
401        assert_debug_snapshot!(expand(r"
402/* foo bar$0 */
403select 1
404"), @r#"
405        [
406            "bar",
407            "/* foo bar */",
408            "\n/* foo bar */\nselect 1\n",
409        ]
410        "#);
411    }
412
413    #[test]
414    fn create_table_with_comment() {
415        assert_debug_snapshot!(expand(r"
416-- foo bar buzz
417create table t(
418  x int$0,
419  y text
420);
421"), @r#"
422        [
423            "int",
424            "x int",
425            "x int,",
426            "(\n  x int,\n  y text\n)",
427            "-- foo bar buzz\ncreate table t(\n  x int,\n  y text\n)",
428            "\n-- foo bar buzz\ncreate table t(\n  x int,\n  y text\n);\n",
429        ]
430        "#);
431    }
432
433    #[test]
434    fn column_list() {
435        assert_debug_snapshot!(expand(r#"create table t($0x int)"#), @r#"
436        [
437            "x",
438            "x int",
439            "(x int)",
440            "create table t(x int)",
441        ]
442        "#);
443
444        assert_debug_snapshot!(expand(r#"create table t($0x int, y int)"#), @r#"
445        [
446            "x",
447            "x int",
448            "x int, ",
449            "(x int, y int)",
450            "create table t(x int, y int)",
451        ]
452        "#);
453
454        assert_debug_snapshot!(expand(r#"create table t(x int, $0y int)"#), @r#"
455        [
456            "y",
457            "y int",
458            ", y int",
459            "(x int, y int)",
460            "create table t(x int, y int)",
461        ]
462        "#);
463    }
464
465    #[test]
466    fn start_of_line_whitespace_select() {
467        assert_debug_snapshot!(expand(r#"    
468select 1;
469
470$0    select 2;"#), @r#"
471        [
472            "    select 2",
473            "    \nselect 1;\n\n    select 2;",
474        ]
475        "#);
476    }
477
478    #[test]
479    fn select_list() {
480        assert_debug_snapshot!(expand(r#"select x$0, y from t"#), @r#"
481        [
482            "x",
483            "x, ",
484            "x, y",
485            "select x, y",
486            "select x, y from t",
487        ]
488        "#);
489
490        assert_debug_snapshot!(expand(r#"select x, y$0 from t"#), @r#"
491        [
492            "y",
493            ", y",
494            "x, y",
495            "select x, y",
496            "select x, y from t",
497        ]
498        "#);
499    }
500
501    #[test]
502    fn expand_whitespace() {
503        assert_debug_snapshot!(expand(r#"select 1 + 
504$0
5051;"#), @r#"
506        [
507            " \n\n",
508            "1 + \n\n1",
509            "select 1 + \n\n1",
510            "select 1 + \n\n1;",
511        ]
512        "#);
513    }
514
515    #[test]
516    fn function_args() {
517        assert_debug_snapshot!(expand(r#"select f(1$0, 2)"#), @r#"
518        [
519            "1",
520            "1, ",
521            "(1, 2)",
522            "f(1, 2)",
523            "select f(1, 2)",
524        ]
525        "#);
526    }
527
528    #[test]
529    fn prefer_idents() {
530        assert_debug_snapshot!(expand(r#"select foo$0+bar"#), @r#"
531        [
532            "foo",
533            "foo+bar",
534            "select foo+bar",
535        ]
536        "#);
537
538        assert_debug_snapshot!(expand(r#"select foo+$0bar"#), @r#"
539        [
540            "bar",
541            "foo+bar",
542            "select foo+bar",
543        ]
544        "#);
545    }
546
547    #[test]
548    fn list_variants() {
549        let delimited_ws_list_kinds = &[
550            SyntaxKind::CREATE_DATABASE_OPTION_LIST,
551            SyntaxKind::FUNC_OPTION_LIST,
552            SyntaxKind::ROLE_OPTION_LIST,
553            SyntaxKind::SEQUENCE_OPTION_LIST,
554            SyntaxKind::TRIGGER_EVENT_LIST,
555            SyntaxKind::XML_COLUMN_OPTION_LIST,
556            SyntaxKind::WHEN_CLAUSE_LIST,
557            SyntaxKind::LABEL_AND_PROPERTIES_LIST,
558        ];
559
560        let unhandled_list_kinds = (0..SyntaxKind::__LAST as u16)
561            .map(SyntaxKind::from)
562            .filter(|kind| {
563                format!("{:?}", kind).ends_with("_LIST") && !delimited_ws_list_kinds.contains(kind)
564            })
565            .filter(|kind| !DELIMITED_LIST_KINDS.contains(kind))
566            .collect::<Vec<_>>();
567
568        assert_eq!(
569            unhandled_list_kinds,
570            vec![],
571            "We shouldn't have any unhandled list kinds"
572        )
573    }
574}
squawk_ide/expand_selection.rs

squawk_ide/
expand_selection.rs
