4a9d116 · helix · git.iko.soy

git.iko.soy/helix

Hacky way to specify indent scopes per language via toml configs.

Can't do it via a scm query nicely because it returns an iterator over all the matches, whereas we want to traverse the tree ourselves. Can't extract the pattern data from a parsed query either. Oh well, toml files for now.

commit4a9d1163e04ba0fc29f8532afd35381b7df55e48 authorBlaž Hrastnik <blaz@mxxn.io> dateFri 14 May 2021, 19:21 (5 years ago) parent7260720

7 files changed+100−45

compare with master

unifiedsplit

Cargo.lock+1 helix-core/Cargo.toml+1 helix-core/src/indent.rs+20−41 helix-core/src/syntax.rs+29−1 helix-term/src/commands.rs+14−3 helix-view/src/document.rs+5 runtime/queries/rust/indents.toml+30

Cargo.lock+1

@@ -291,6 +291,7 @@ dependencies = [

291291 "serde",

292292 "smallvec",

293293 "tendril",

294+ "toml",

294295 "tree-sitter",

295296 "unicode-segmentation",

296297 "unicode-width",

helix-core/Cargo.toml+1

@@ -22,5 +22,6 @@ once_cell = "1.4"

2222regex = "1"

2323

2424serde = { version = "1.0", features = ["derive"] }

25+toml = "0.5"

2526

2627etcetera = "0.3"

helix-core/src/indent.rs+20−41

@@ -1,6 +1,6 @@

11use crate::{

22 find_first_non_whitespace_char,

3− syntax::Syntax,

3+ syntax::{IndentQuery, LanguageConfiguration, Syntax},

44 tree_sitter::{Node, Tree},

55 Rope, RopeSlice,

66};

@@ -43,40 +43,11 @@ fn get_highest_syntax_node_at_bytepos(syntax: &Syntax, pos: usize) -> Option<Nod

4343 Some(node)

4444}

4545

46−fn calculate_indentation(node: Option<Node>, newline: bool) -> usize {

47− let mut increment: i32 = 0;

48−

49− // Hardcoded for rust for now

50− let indent_scopes = &[

51− "while_expression",

52− "for_expression",

53− "loop_expression",

54− "if_expression",

55− "if_let_expression",

56− // "match_expression",

57− // "match_arm",

58− "tuple_expression",

59− "array_expression",

60− // indent_except_first_scopes

61− "use_list",

62− "block",

63− "match_block",

64− "arguments",

65− "parameters",

66− "declaration_list",

67− "field_declaration_list",

68− "field_initializer_list",

69− "struct_pattern",

70− "tuple_pattern",

71− "enum_variant_list",

72− // "function_item",

73− // "closure_expression",

74− "binary_expression",

75− "field_expression",

76− "where_clause",

77− ];

46+fn calculate_indentation(query: &IndentQuery, node: Option<Node>, newline: bool) -> usize {

47+ // NOTE: can't use contains() on query because of comparing Vec<String> and &str

48+ // https://doc.rust-lang.org/std/vec/struct.Vec.html#method.contains

7849

79− let outdent = &["where", "}", "]", ")"];

50+ let mut increment: i32 = 0;

8051

8152 let mut node = match node {

8253 Some(node) => node,

@@ -88,7 +59,7 @@ fn calculate_indentation(node: Option<Node>, newline: bool) -> usize {

8859 // if we're calculating indentation for a brand new line then the current node will become the

8960 // parent node. We need to take it's indentation level into account too.

9061 let node_kind = node.kind();

91− if newline && indent_scopes.contains(&node_kind) {

62+ if newline && query.indent.contains(node_kind) {

9263 increment += 1;

9364 }

9465

@@ -102,14 +73,14 @@ fn calculate_indentation(node: Option<Node>, newline: bool) -> usize {

10273 // }) <-- }) is two scopes

10374 let starts_same_line = start == prev_start;

10475

105− if outdent.contains(&node.kind()) && !starts_same_line {

76+ if query.outdent.contains(node.kind()) && !starts_same_line {

10677 // we outdent by skipping the rules for the current level and jumping up

10778 // node = parent;

10879 increment -= 1;

10980 // continue;

11081 }

11182

112− if indent_scopes.contains(&parent_kind) // && not_first_or_last_sibling

83+ if query.indent.contains(parent_kind) // && not_first_or_last_sibling

11384 && !starts_same_line

11485 {

11586 // println!("is_scope {}", parent_kind);

@@ -128,6 +99,7 @@ fn calculate_indentation(node: Option<Node>, newline: bool) -> usize {

12899}

129100

130101fn suggested_indent_for_line(

102+ language_config: &LanguageConfiguration,

131103 syntax: Option<&Syntax>,

132104 text: RopeSlice,

133105 line_num: usize,

@@ -137,7 +109,7 @@ fn suggested_indent_for_line(

137109 let current = indent_level_for_line(line, tab_width);

138110

139111 if let Some(start) = find_first_non_whitespace_char(text, line_num) {

140− return suggested_indent_for_pos(syntax, text, start, false);

112+ return suggested_indent_for_pos(Some(language_config), syntax, text, start, false);

141113 };

142114

143115 // if the line is blank, indent should be zero

@@ -148,18 +120,24 @@ fn suggested_indent_for_line(

148120// - it should return 0 when mass indenting stuff

149121// - it should look up the wrapper node and count it too when we press o/O

150122pub fn suggested_indent_for_pos(

123+ language_config: Option<&LanguageConfiguration>,

151124 syntax: Option<&Syntax>,

152125 text: RopeSlice,

153126 pos: usize,

154127 new_line: bool,

155128) -> usize {

156− if let Some(syntax) = syntax {

129+ if let (Some(query), Some(syntax)) = (

130+ language_config.and_then(|config| config.indent_query()),

131+ syntax,

132+ ) {

157133 let byte_start = text.char_to_byte(pos);

158134 let node = get_highest_syntax_node_at_bytepos(syntax, byte_start);

159135

136+ // let config = load indentation query config from Syntax(should contain language_config)

137+

160138 // TODO: special case for comments

161139 // TODO: if preserve_leading_whitespace

162− calculate_indentation(node, new_line)

140+ calculate_indentation(query, node, new_line)

163141 } else {

164142 // TODO: heuristics for non-tree sitter grammars

165143 0

@@ -286,6 +264,7 @@ where

286264 tab_width: 4,

287265 unit: String::from(" "),

288266 }),

267+ indent_query: OnceCell::new(),

289268 }],

290269 });

291270

@@ -304,7 +283,7 @@ where

304283 let line = text.line(i);

305284 let indent = indent_level_for_line(line, tab_width);

306285 assert_eq!(

307− suggested_indent_for_line(Some(&syntax), text, i, tab_width),

286+ suggested_indent_for_line(&language_config, Some(&syntax), text, i, tab_width),

308287 indent,

309288 "line {}: {}",

310289 i,

helix-core/src/syntax.rs+29−1

@@ -4,7 +4,7 @@ pub use helix_syntax::{get_language, get_language_name, Lang};

44use std::{

55 borrow::Cow,

66 cell::RefCell,

7− collections::HashMap,

7+ collections::{HashMap, HashSet},

88 path::{Path, PathBuf},

99 sync::Arc,

1010};

@@ -41,6 +41,9 @@ pub struct LanguageConfiguration {

4141 pub language_server: Option<LanguageServerConfiguration>,

4242 #[serde(skip_serializing_if = "Option::is_none")]

4343 pub indent: Option<IndentationConfiguration>,

44+

45+ #[serde(skip)]

46+ pub(crate) indent_query: OnceCell<Option<IndentQuery>>,

4447}

4548

4649#[derive(Serialize, Deserialize)]

@@ -59,6 +62,17 @@ pub struct IndentationConfiguration {

5962 pub unit: String,

6063}

6164

65+#[derive(Serialize, Deserialize)]

66+#[serde(rename_all = "kebab-case")]

67+pub struct IndentQuery {

68+ #[serde(default)]

69+ #[serde(skip_serializing_if = "HashSet::is_empty")]

70+ pub indent: HashSet<String>,

71+ #[serde(default)]

72+ #[serde(skip_serializing_if = "HashSet::is_empty")]

73+ pub outdent: HashSet<String>,

74+}

75+

6276fn read_query(language: &str, filename: &str) -> String {

6377 static INHERITS_REGEX: Lazy<Regex> =

6478 Lazy::new(|| Regex::new(r";+\s*inherits\s*:?\s*([a-z_,()]+)\s*").unwrap());

@@ -127,6 +141,20 @@ impl LanguageConfiguration {

127141 .clone()

128142 }

129143

144+ pub fn indent_query(&self) -> Option<&IndentQuery> {

145+ self.indent_query

146+ .get_or_init(|| {

147+ let language = get_language_name(self.language_id).to_ascii_lowercase();

148+

149+ let root = crate::runtime_dir();

150+ let path = root.join("queries").join(language).join("indents.toml");

151+

152+ let toml = std::fs::read(&path).ok()?;

153+ toml::from_slice(&toml).ok()

154+ })

155+ .as_ref()

156+ }

157+

130158 pub fn scope(&self) -> &str {

131159 &self.scope

132160 }

helix-term/src/commands.rs+14−3

@@ -1171,7 +1171,13 @@ fn open(cx: &mut Context, open: Open) {

11711171 let index = doc.text().line_to_char(line).saturating_sub(1);

11721172

11731173 // TODO: share logic with insert_newline for indentation

1174− let indent_level = indent::suggested_indent_for_pos(doc.syntax(), text, index, true);

1174+ let indent_level = indent::suggested_indent_for_pos(

1175+ doc.language_config(),

1176+ doc.syntax(),

1177+ text,

1178+ index,

1179+ true,

1180+ );

11751181 let indent = doc.indent_unit().repeat(indent_level);

11761182 let mut text = String::with_capacity(1 + indent.len());

11771183 text.push('\n');

@@ -1649,8 +1655,13 @@ pub mod insert {

16491655 let curr = contents.char(pos);

16501656

16511657 // TODO: offset range.head by 1? when calculating?

1652− let indent_level =

1653− indent::suggested_indent_for_pos(doc.syntax(), text, pos.saturating_sub(1), true);

1658+ let indent_level = indent::suggested_indent_for_pos(

1659+ doc.language_config(),

1660+ doc.syntax(),

1661+ text,

1662+ pos.saturating_sub(1),

1663+ true,

1664+ );

16541665 let indent = doc.indent_unit().repeat(indent_level);

16551666 let mut text = String::with_capacity(1 + indent.len());

16561667 text.push('\n');

helix-view/src/document.rs+5

@@ -328,6 +328,11 @@ impl Document {

328328 .map(|language| language.scope.as_str())

329329 }

330330

331+ #[inline]

332+ pub fn language_config(&self) -> Option<&LanguageConfiguration> {

333+ self.language.as_deref()

334+ }

335+

331336 #[inline]

332337 /// Current document version, incremented at each change.

333338 pub fn version(&self) -> i32 {

runtime/queries/rust/indents.tomlnew file+30

@@ -0,0 +1,30 @@

1+indent = [

2+ "while_expression",

3+ "for_expression",

4+ "loop_expression",

5+ "if_expression",

6+ "if_let_expression",

7+ "tuple_expression",

8+ "array_expression",

9+ "use_list",

10+ "block",

11+ "match_block",

12+ "arguments",

13+ "parameters",

14+ "declaration_list",

15+ "field_declaration_list",

16+ "field_initializer_list",

17+ "struct_pattern",

18+ "tuple_pattern",

19+ "enum_variant_list",

20+ "binary_expression",

21+ "field_expression",

22+ "where_clause"

23+]

24+

25+outdent = [

26+ "where",

27+ "}",

28+ "]",

29+ ")"

30+]