Add basic indentation for languages without treesitter-based indentation rules (always use the indent of the current line for a new line). (#1341)

Fix several bugs in the treesitter indentation calculation.

Co-authored-by: Triton171 <triton0171@gmail.com>

helix-core/src/indent.rs

@@ -1,6 +1,5 @@
 use crate::{
     chars::{char_is_line_ending, char_is_whitespace},
-    find_first_non_whitespace_char,
     syntax::{IndentQuery, LanguageConfiguration, Syntax},
     tree_sitter::Node,
     Rope, RopeSlice,
@@ -174,8 +173,7 @@ pub fn auto_detect_indent_style(document_text: &Rope) -> Option<IndentStyle> {
 
 /// To determine indentation of a newly inserted line, figure out the indentation at the last col
 /// of the previous line.
-#[allow(dead_code)]
-fn indent_level_for_line(line: RopeSlice, tab_width: usize) -> usize {
+pub fn indent_level_for_line(line: RopeSlice, tab_width: usize) -> usize {
     let mut len = 0;
     for ch in line.chars() {
         match ch {
@@ -210,10 +208,15 @@ fn get_highest_syntax_node_at_bytepos(syntax: &Syntax, pos: usize) -> Option<Nod
     Some(node)
 }
 
-fn calculate_indentation(query: &IndentQuery, node: Option<Node>, newline: bool) -> usize {
-    // NOTE: can't use contains() on query because of comparing Vec<String> and &str
-    // https://doc.rust-lang.org/std/vec/struct.Vec.html#method.contains
-
+/// Calculate the indentation at a given treesitter node.
+/// If newline is false, then any "indent" nodes on the line are ignored ("outdent" still applies).
+/// This is because the indentation is only increased starting at the second line of the node.
+fn calculate_indentation(
+    query: &IndentQuery,
+    node: Option<Node>,
+    line: usize,
+    newline: bool,
+) -> usize {
     let mut increment: isize = 0;
 
     let mut node = match node {
@@ -221,70 +224,45 @@ fn calculate_indentation(query: &IndentQuery, node: Option<Node>, newline: bool)
         None => return 0,
     };
 
-    let mut prev_start = node.start_position().row;
-
-    // if we're calculating indentation for a brand new line then the current node will become the
-    // parent node. We need to take it's indentation level into account too.
-    let node_kind = node.kind();
-    if newline && query.indent.contains(node_kind) {
-        increment += 1;
-    }
-
-    while let Some(parent) = node.parent() {
-        let parent_kind = parent.kind();
-        let start = parent.start_position().row;
-
-        // detect deeply nested indents in the same line
-        // .map(|a| {       <-- ({ is two scopes
-        //     let len = 1; <-- indents one level
-        // })               <-- }) is two scopes
-        let starts_same_line = start == prev_start;
-
-        if query.outdent.contains(node.kind()) && !starts_same_line {
-            // we outdent by skipping the rules for the current level and jumping up
-            // node = parent;
-            increment -= 1;
-            // continue;
+    let mut current_line = line;
+    let mut consider_indent = newline;
+    let mut increment_from_line: isize = 0;
+
+    loop {
+        let node_kind = node.kind();
+        let start = node.start_position().row;
+        if current_line != start {
+            // Indent/dedent by at most one per line:
+            // .map(|a| {       <-- ({ is two scopes
+            //     let len = 1; <-- indents one level
+            // })               <-- }) is two scopes
+            if consider_indent || increment_from_line < 0 {
+                increment += increment_from_line.signum();
+            }
+            increment_from_line = 0;
+            current_line = start;
+            consider_indent = true;
         }
 
-        if query.indent.contains(parent_kind) // && not_first_or_last_sibling
-            && !starts_same_line
-        {
-            // println!("is_scope {}", parent_kind);
-            prev_start = start;
-            increment += 1
+        if query.outdent.contains(node_kind) {
+            increment_from_line -= 1;
+        }
+        if query.indent.contains(node_kind) {
+            increment_from_line += 1;
         }
 
-        // if last_scope && increment > 0 && ...{ ignore }
-
-        node = parent;
+        if let Some(parent) = node.parent() {
+            node = parent;
+        } else {
+            break;
+        }
+    }
+    if consider_indent || increment_from_line < 0 {
+        increment += increment_from_line.signum();
     }
-
     increment.max(0) as usize
 }
 
-#[allow(dead_code)]
-fn suggested_indent_for_line(
-    language_config: &LanguageConfiguration,
-    syntax: Option<&Syntax>,
-    text: RopeSlice,
-    line_num: usize,
-    _tab_width: usize,
-) -> usize {
-    if let Some(start) = find_first_non_whitespace_char(text.line(line_num)) {
-        return suggested_indent_for_pos(
-            Some(language_config),
-            syntax,
-            text,
-            start + text.line_to_char(line_num),
-            false,
-        );
-    };
-
-    // if the line is blank, indent should be zero
-    0
-}
-
 // TODO: two usecases: if we are triggering this for a new, blank line:
 // - it should return 0 when mass indenting stuff
 // - it should look up the wrapper node and count it too when we press o/O
@@ -293,23 +271,20 @@ pub fn suggested_indent_for_pos(
     syntax: Option<&Syntax>,
     text: RopeSlice,
     pos: usize,
+    line: usize,
     new_line: bool,
-) -> usize {
+) -> Option<usize> {
     if let (Some(query), Some(syntax)) = (
         language_config.and_then(|config| config.indent_query()),
         syntax,
     ) {
         let byte_start = text.char_to_byte(pos);
         let node = get_highest_syntax_node_at_bytepos(syntax, byte_start);
-
-        // let config = load indentation query config from Syntax(should contain language_config)
-
         // TODO: special case for comments
         // TODO: if preserve_leading_whitespace
-        calculate_indentation(query, node, new_line)
+        Some(calculate_indentation(query, node, line, new_line))
     } else {
-        // TODO: heuristics for non-tree sitter grammars
-        0
+        None
     }
 }
 
@@ -484,14 +459,23 @@ where
 
         for i in 0..doc.len_lines() {
             let line = text.line(i);
-            let indent = indent_level_for_line(line, tab_width);
-            assert_eq!(
-                suggested_indent_for_line(&language_config, Some(&syntax), text, i, tab_width),
-                indent,
-                "line {}: {}",
-                i,
-                line
-            );
+            if let Some(pos) = crate::find_first_non_whitespace_char(line) {
+                let indent = indent_level_for_line(line, tab_width);
+                assert_eq!(
+                    suggested_indent_for_pos(
+                        Some(&language_config),
+                        Some(&syntax),
+                        text,
+                        text.line_to_char(i) + pos,
+                        i,
+                        false
+                    ),
+                    Some(indent),
+                    "line {}: \"{}\"",
+                    i,
+                    line
+                );
+            }
         }
     }
 }

helix-term/src/commands.rs

@@ -3689,22 +3689,22 @@ fn open(cx: &mut Context, open: Open) {
     let mut offs = 0;
 
     let mut transaction = Transaction::change_by_selection(contents, selection, |range| {
-        let line = range.cursor_line(text);
+        let cursor_line = range.cursor_line(text);
 
-        let line = match open {
+        let new_line = match open {
             // adjust position to the end of the line (next line - 1)
-            Open::Below => line + 1,
+            Open::Below => cursor_line + 1,
             // adjust position to the end of the previous line (current line - 1)
-            Open::Above => line,
+            Open::Above => cursor_line,
         };
 
         // Index to insert newlines after, as well as the char width
         // to use to compensate for those inserted newlines.
-        let (line_end_index, line_end_offset_width) = if line == 0 {
+        let (line_end_index, line_end_offset_width) = if new_line == 0 {
             (0, 0)
         } else {
             (
-                line_end_char_index(&doc.text().slice(..), line.saturating_sub(1)),
+                line_end_char_index(&doc.text().slice(..), new_line.saturating_sub(1)),
                 doc.line_ending.len_chars(),
             )
         };
@@ -3715,8 +3715,10 @@ fn open(cx: &mut Context, open: Open) {
             doc.syntax(),
             text,
             line_end_index,
+            new_line.saturating_sub(1),
             true,
-        );
+        )
+        .unwrap_or_else(|| indent::indent_level_for_line(text.line(cursor_line), doc.tab_width()));
         let indent = doc.indent_unit().repeat(indent_level);
         let indent_len = indent.len();
         let mut text = String::with_capacity(1 + indent_len);
@@ -4451,48 +4453,48 @@ pub mod insert {
             };
             let curr = contents.get_char(pos).unwrap_or(' ');
 
-            // TODO: offset range.head by 1? when calculating?
+            let current_line = text.char_to_line(pos);
             let indent_level = indent::suggested_indent_for_pos(
                 doc.language_config(),
                 doc.syntax(),
                 text,
-                pos.saturating_sub(1),
+                pos,
+                current_line,
                 true,
-            );
-            let indent = doc.indent_unit().repeat(indent_level);
-            let mut text = String::with_capacity(1 + indent.len());
-            text.push_str(doc.line_ending.as_str());
-            text.push_str(&indent);
+            )
+            .unwrap_or_else(|| {
+                indent::indent_level_for_line(text.line(current_line), doc.tab_width())
+            });
 
-            let head = pos + offs + text.chars().count();
+            let indent = doc.indent_unit().repeat(indent_level);
+            let mut text = String::new();
+            // If we are between pairs (such as brackets), we want to insert an additional line which is indented one level more and place the cursor there
+            let new_head_pos = if helix_core::auto_pairs::PAIRS.contains(&(prev, curr)) {
+                let inner_indent = doc.indent_unit().repeat(indent_level + 1);
+                text.reserve_exact(2 + indent.len() + inner_indent.len());
+                text.push_str(doc.line_ending.as_str());
+                text.push_str(&inner_indent);
+                let new_head_pos = pos + offs + text.chars().count();
+                text.push_str(doc.line_ending.as_str());
+                text.push_str(&indent);
+                new_head_pos
+            } else {
+                text.reserve_exact(1 + indent.len());
+                text.push_str(doc.line_ending.as_str());
+                text.push_str(&indent);
+                pos + offs + text.chars().count()
+            };
 
             // TODO: range replace or extend
             // range.replace(|range| range.is_empty(), head); -> fn extend if cond true, new head pos
             // can be used with cx.mode to do replace or extend on most changes
-            ranges.push(Range::new(
-                if range.is_empty() {
-                    head
-                } else {
-                    range.anchor + offs
-                },
-                head,
-            ));
-
-            // if between a bracket pair
-            if helix_core::auto_pairs::PAIRS.contains(&(prev, curr)) {
-                // another newline, indent the end bracket one level less
-                let indent = doc.indent_unit().repeat(indent_level.saturating_sub(1));
-                text.push_str(doc.line_ending.as_str());
-                text.push_str(&indent);
-            }
-
+            ranges.push(Range::new(new_head_pos, new_head_pos));
             offs += text.chars().count();
 
             (pos, pos, Some(text.into()))
         });
 
         transaction = transaction.with_selection(Selection::new(ranges, selection.primary_index()));
-        //
 
         doc.apply(&transaction, view.id);
     }

runtime/queries/rust/indents.toml

@@ -9,6 +9,7 @@ indent = [
  "field_initializer_list",
  "struct_pattern",
  "tuple_pattern",
+ "unit_expression",
  "enum_variant_list",
  "call_expression",
  "binary_expression",