From 21e7e35e7356be6cc7a7c870c7a0bbec294d7fa3 Mon Sep 17 00:00:00 2001 From: Max Brunsfeld Date: Thu, 13 Jul 2023 18:38:29 -0700 Subject: [PATCH] Include newlines in between combined injection ranges on different lines --- crates/language/src/syntax_map.rs | 90 ++++++++++++- .../src/syntax_map/syntax_map_tests.rs | 120 ++++++++++++++++-- crates/zed/src/languages/heex/config.toml | 2 +- crates/zed/src/languages/heex/highlights.scm | 6 +- 4 files changed, 201 insertions(+), 17 deletions(-) diff --git a/crates/language/src/syntax_map.rs b/crates/language/src/syntax_map.rs index 526f99ec95..1590294b1a 100644 --- a/crates/language/src/syntax_map.rs +++ b/crates/language/src/syntax_map.rs @@ -569,11 +569,19 @@ impl SyntaxSnapshot { range.end = range.end.saturating_sub(step_start_byte); } - included_ranges = splice_included_ranges( + let changed_indices; + (included_ranges, changed_indices) = splice_included_ranges( old_tree.included_ranges(), &parent_layer_changed_ranges, &included_ranges, ); + insert_newlines_between_ranges( + changed_indices, + &mut included_ranges, + &text, + step_start_byte, + step_start_point, + ); } if included_ranges.is_empty() { @@ -586,7 +594,7 @@ impl SyntaxSnapshot { } log::trace!( - "update layer. language:{}, start:{:?}, ranges:{:?}", + "update layer. language:{}, start:{:?}, included_ranges:{:?}", language.name(), LogAnchorRange(&step.range, text), LogIncludedRanges(&included_ranges), @@ -608,6 +616,16 @@ impl SyntaxSnapshot { }), ); } else { + if matches!(step.mode, ParseMode::Combined { .. }) { + insert_newlines_between_ranges( + 0..included_ranges.len(), + &mut included_ranges, + text, + step_start_byte, + step_start_point, + ); + } + if included_ranges.is_empty() { included_ranges.push(tree_sitter::Range { start_byte: 0, @@ -1275,14 +1293,20 @@ fn get_injections( } } +/// Update the given list of included `ranges`, removing any ranges that intersect +/// `removed_ranges`, and inserting the given `new_ranges`. +/// +/// Returns a new vector of ranges, and the range of the vector that was changed, +/// from the previous `ranges` vector. pub(crate) fn splice_included_ranges( mut ranges: Vec, removed_ranges: &[Range], new_ranges: &[tree_sitter::Range], -) -> Vec { +) -> (Vec, Range) { let mut removed_ranges = removed_ranges.iter().cloned().peekable(); let mut new_ranges = new_ranges.into_iter().cloned().peekable(); let mut ranges_ix = 0; + let mut changed_portion = usize::MAX..0; loop { let next_new_range = new_ranges.peek(); let next_removed_range = removed_ranges.peek(); @@ -1344,11 +1368,69 @@ pub(crate) fn splice_included_ranges( } } + changed_portion.start = changed_portion.start.min(start_ix); + changed_portion.end = changed_portion.end.max(if insert.is_some() { + start_ix + 1 + } else { + start_ix + }); + ranges.splice(start_ix..end_ix, insert); ranges_ix = start_ix; } - ranges + if changed_portion.end < changed_portion.start { + changed_portion = 0..0; + } + + (ranges, changed_portion) +} + +/// Ensure there are newline ranges in between content range that appear on +/// different lines. For performance, only iterate through the given range of +/// indices. All of the ranges in the array are relative to a given start byte +/// and point. +fn insert_newlines_between_ranges( + indices: Range, + ranges: &mut Vec, + text: &text::BufferSnapshot, + start_byte: usize, + start_point: Point, +) { + let mut ix = indices.end + 1; + while ix > indices.start { + ix -= 1; + if 0 == ix || ix == ranges.len() { + continue; + } + + let range_b = ranges[ix].clone(); + let range_a = &mut ranges[ix - 1]; + if range_a.end_point.column == 0 { + continue; + } + + if range_a.end_point.row < range_b.start_point.row { + let end_point = start_point + Point::from_ts_point(range_a.end_point); + let line_end = Point::new(end_point.row, text.line_len(end_point.row)); + if end_point.column as u32 >= line_end.column { + range_a.end_byte += 1; + range_a.end_point.row += 1; + range_a.end_point.column = 0; + } else { + let newline_offset = text.point_to_offset(line_end); + ranges.insert( + ix, + tree_sitter::Range { + start_byte: newline_offset - start_byte, + end_byte: newline_offset - start_byte + 1, + start_point: (line_end - start_point).to_ts_point(), + end_point: ((line_end - start_point) + Point::new(1, 0)).to_ts_point(), + }, + ) + } + } + } } impl OwnedSyntaxLayerInfo { diff --git a/crates/language/src/syntax_map/syntax_map_tests.rs b/crates/language/src/syntax_map/syntax_map_tests.rs index 272501f2d0..a6d35d0e2f 100644 --- a/crates/language/src/syntax_map/syntax_map_tests.rs +++ b/crates/language/src/syntax_map/syntax_map_tests.rs @@ -11,7 +11,7 @@ use util::test::marked_text_ranges; fn test_splice_included_ranges() { let ranges = vec![ts_range(20..30), ts_range(50..60), ts_range(80..90)]; - let new_ranges = splice_included_ranges( + let (new_ranges, change) = splice_included_ranges( ranges.clone(), &[54..56, 58..68], &[ts_range(50..54), ts_range(59..67)], @@ -25,14 +25,16 @@ fn test_splice_included_ranges() { ts_range(80..90), ] ); + assert_eq!(change, 1..3); - let new_ranges = splice_included_ranges(ranges.clone(), &[70..71, 91..100], &[]); + let (new_ranges, change) = splice_included_ranges(ranges.clone(), &[70..71, 91..100], &[]); assert_eq!( new_ranges, &[ts_range(20..30), ts_range(50..60), ts_range(80..90)] ); + assert_eq!(change, 2..3); - let new_ranges = + let (new_ranges, change) = splice_included_ranges(ranges.clone(), &[], &[ts_range(0..2), ts_range(70..75)]); assert_eq!( new_ranges, @@ -44,16 +46,21 @@ fn test_splice_included_ranges() { ts_range(80..90) ] ); + assert_eq!(change, 0..4); - let new_ranges = splice_included_ranges(ranges.clone(), &[30..50], &[ts_range(25..55)]); + let (new_ranges, change) = + splice_included_ranges(ranges.clone(), &[30..50], &[ts_range(25..55)]); assert_eq!(new_ranges, &[ts_range(25..55), ts_range(80..90)]); + assert_eq!(change, 0..1); // does not create overlapping ranges - let new_ranges = splice_included_ranges(ranges.clone(), &[0..18], &[ts_range(20..32)]); + let (new_ranges, change) = + splice_included_ranges(ranges.clone(), &[0..18], &[ts_range(20..32)]); assert_eq!( new_ranges, &[ts_range(20..32), ts_range(50..60), ts_range(80..90)] ); + assert_eq!(change, 0..1); fn ts_range(range: Range) -> tree_sitter::Range { tree_sitter::Range { @@ -511,7 +518,7 @@ fn test_removing_injection_by_replacing_across_boundary() { } #[gpui::test] -fn test_combined_injections() { +fn test_combined_injections_simple() { let (buffer, syntax_map) = test_edit_sequence( "ERB", &[ @@ -653,33 +660,78 @@ fn test_combined_injections_editing_after_last_injection() { #[gpui::test] fn test_combined_injections_inside_injections() { - let (_buffer, _syntax_map) = test_edit_sequence( + let (buffer, syntax_map) = test_edit_sequence( "Markdown", &[ r#" - here is some ERB code: + here is + some + ERB code: ```erb
    <% people.each do |person| %>
  • <%= person.name %>
    • +
  • <%= person.age %>
    • <% end %>
``` "#, r#" - here is some ERB code: + here is + some + ERB code: ```erb
    <% people«2».each do |person| %>
  • <%= person.name %>
    • +
  • <%= person.age %>
    • + <% end %> +
+ ``` + "#, + // Inserting a comment character inside one code directive + // does not cause the other code directive to become a comment, + // because newlines are included in between each injection range. + r#" + here is + some + ERB code: + + ```erb +
    + <% people2.each do |person| %> +
  • <%= «# »person.name %>
    • +
  • <%= person.age %>
    • <% end %>
``` "#, ], ); + + // Check that the code directive below the ruby comment is + // not parsed as a comment. + assert_capture_ranges( + &syntax_map, + &buffer, + &["method"], + " + here is + some + ERB code: + + ```erb +
    + <% people2.«each» do |person| %> +
  • <%= # person.name %>
    • +
  • <%= person.«age» %>
    • + <% end %> +
+ ``` + ", + ); } #[gpui::test] @@ -984,11 +1036,14 @@ fn check_interpolation( fn test_edit_sequence(language_name: &str, steps: &[&str]) -> (Buffer, SyntaxMap) { let registry = Arc::new(LanguageRegistry::test()); + registry.add(Arc::new(elixir_lang())); + registry.add(Arc::new(heex_lang())); registry.add(Arc::new(rust_lang())); registry.add(Arc::new(ruby_lang())); registry.add(Arc::new(html_lang())); registry.add(Arc::new(erb_lang())); registry.add(Arc::new(markdown_lang())); + let language = registry .language_for_name(language_name) .now_or_never() @@ -1074,6 +1129,7 @@ fn ruby_lang() -> Language { r#" ["if" "do" "else" "end"] @keyword (instance_variable) @ivar + (call method: (identifier) @method) "#, ) .unwrap() @@ -1158,6 +1214,52 @@ fn markdown_lang() -> Language { .unwrap() } +fn elixir_lang() -> Language { + Language::new( + LanguageConfig { + name: "Elixir".into(), + path_suffixes: vec!["ex".into()], + ..Default::default() + }, + Some(tree_sitter_elixir::language()), + ) + .with_highlights_query( + r#" + + "#, + ) + .unwrap() +} + +fn heex_lang() -> Language { + Language::new( + LanguageConfig { + name: "HEEx".into(), + path_suffixes: vec!["heex".into()], + ..Default::default() + }, + Some(tree_sitter_heex::language()), + ) + .with_injection_query( + r#" + ( + (directive + [ + (partial_expression_value) + (expression_value) + (ending_expression_value) + ] @content) + (#set! language "elixir") + (#set! combined) + ) + + ((expression (expression_value) @content) + (#set! language "elixir")) + "#, + ) + .unwrap() +} + fn range_for_text(buffer: &Buffer, text: &str) -> Range { let start = buffer.as_rope().to_string().find(text).unwrap(); start..start + text.len() diff --git a/crates/zed/src/languages/heex/config.toml b/crates/zed/src/languages/heex/config.toml index b5d74f80f3..c9f952ee3c 100644 --- a/crates/zed/src/languages/heex/config.toml +++ b/crates/zed/src/languages/heex/config.toml @@ -4,4 +4,4 @@ autoclose_before = ">})" brackets = [ { start = "<", end = ">", close = true, newline = true }, ] -block_comment = ["<%!--", "--%>"] +block_comment = ["<%!-- ", " --%>"] diff --git a/crates/zed/src/languages/heex/highlights.scm b/crates/zed/src/languages/heex/highlights.scm index 8728110d58..5252b71fac 100644 --- a/crates/zed/src/languages/heex/highlights.scm +++ b/crates/zed/src/languages/heex/highlights.scm @@ -1,10 +1,7 @@ ; HEEx delimiters [ - "--%>" - "-->" "/>" "" + "--%>" + "-->" + "