toggling of block comments (#4718)

5 files changed, 337 insertions, 10 deletions

diff --git a/helix-core/src/comment.rs b/helix-core/src/comment.rs
index 9c7e50f3..536b710a 100644
--- a/helix-core/src/comment.rs
+++ b/helix-core/src/comment.rs
@@ -1,9 +1,12 @@
 //! This module contains the functionality toggle comments on lines over the selection
 //! using the comment character defined in the user's `languages.toml`
 
+use smallvec::SmallVec;
+
 use crate::{
-    find_first_non_whitespace_char, Change, Rope, RopeSlice, Selection, Tendril, Transaction,
+    syntax::BlockCommentToken, Change, Range, Rope, RopeSlice, Selection, Tendril, Transaction,
 };
+use helix_stdx::rope::RopeSliceExt;
 use std::borrow::Cow;
 
 /// Given text, a comment token, and a set of line indices, returns the following:
@@ -22,12 +25,12 @@ fn find_line_comment(
 ) -> (bool, Vec<usize>, usize, usize) {
     let mut commented = true;
     let mut to_change = Vec::new();
-    let mut min = usize::MAX; // minimum col for find_first_non_whitespace_char
+    let mut min = usize::MAX; // minimum col for first_non_whitespace_char
     let mut margin = 1;
     let token_len = token.chars().count();
     for line in lines {
         let line_slice = text.line(line);
-        if let Some(pos) = find_first_non_whitespace_char(line_slice) {
+        if let Some(pos) = line_slice.first_non_whitespace_char() {
             let len = line_slice.len_chars();
 
             if pos < min {
@@ -94,6 +97,222 @@ pub fn toggle_line_comments(doc: &Rope, selection: &Selection, token: Option<&st
     Transaction::change(doc, changes.into_iter())
 }
 
+#[derive(Debug, PartialEq, Eq)]
+pub enum CommentChange {
+    Commented {
+        range: Range,
+        start_pos: usize,
+        end_pos: usize,
+        start_margin: bool,
+        end_margin: bool,
+        start_token: String,
+        end_token: String,
+    },
+    Uncommented {
+        range: Range,
+        start_pos: usize,
+        end_pos: usize,
+        start_token: String,
+        end_token: String,
+    },
+    Whitespace {
+        range: Range,
+    },
+}
+
+pub fn find_block_comments(
+    tokens: &[BlockCommentToken],
+    text: RopeSlice,
+    selection: &Selection,
+) -> (bool, Vec<CommentChange>) {
+    let mut commented = true;
+    let mut only_whitespace = true;
+    let mut comment_changes = Vec::with_capacity(selection.len());
+    let default_tokens = tokens.first().cloned().unwrap_or_default();
+    // TODO: check if this can be removed on MSRV bump
+    #[allow(clippy::redundant_clone)]
+    let mut start_token = default_tokens.start.clone();
+    #[allow(clippy::redundant_clone)]
+    let mut end_token = default_tokens.end.clone();
+
+    let mut tokens = tokens.to_vec();
+    // sort the tokens by length, so longer tokens will match first
+    tokens.sort_by(|a, b| {
+        if a.start.len() == b.start.len() {
+            b.end.len().cmp(&a.end.len())
+        } else {
+            b.start.len().cmp(&a.start.len())
+        }
+    });
+    for range in selection {
+        let selection_slice = range.slice(text);
+        if let (Some(start_pos), Some(end_pos)) = (
+            selection_slice.first_non_whitespace_char(),
+            selection_slice.last_non_whitespace_char(),
+        ) {
+            let mut line_commented = false;
+            let mut after_start = 0;
+            let mut before_end = 0;
+            let len = (end_pos + 1) - start_pos;
+
+            for BlockCommentToken { start, end } in &tokens {
+                let start_len = start.chars().count();
+                let end_len = end.chars().count();
+                after_start = start_pos + start_len;
+                before_end = end_pos.saturating_sub(end_len);
+
+                if len >= start_len + end_len {
+                    let start_fragment = selection_slice.slice(start_pos..after_start);
+                    let end_fragment = selection_slice.slice(before_end + 1..end_pos + 1);
+
+                    // block commented with these tokens
+                    if start_fragment == start.as_str() && end_fragment == end.as_str() {
+                        start_token = start.to_string();
+                        end_token = end.to_string();
+                        line_commented = true;
+                        break;
+                    }
+                }
+            }
+
+            if !line_commented {
+                comment_changes.push(CommentChange::Uncommented {
+                    range: *range,
+                    start_pos,
+                    end_pos,
+                    start_token: default_tokens.start.clone(),
+                    end_token: default_tokens.end.clone(),
+                });
+                commented = false;
+            } else {
+                comment_changes.push(CommentChange::Commented {
+                    range: *range,
+                    start_pos,
+                    end_pos,
+                    start_margin: selection_slice
+                        .get_char(after_start)
+                        .map_or(false, |c| c == ' '),
+                    end_margin: after_start != before_end
+                        && selection_slice
+                            .get_char(before_end)
+                            .map_or(false, |c| c == ' '),
+                    start_token: start_token.to_string(),
+                    end_token: end_token.to_string(),
+                });
+            }
+            only_whitespace = false;
+        } else {
+            comment_changes.push(CommentChange::Whitespace { range: *range });
+        }
+    }
+    if only_whitespace {
+        commented = false;
+    }
+    (commented, comment_changes)
+}
+
+#[must_use]
+pub fn create_block_comment_transaction(
+    doc: &Rope,
+    selection: &Selection,
+    commented: bool,
+    comment_changes: Vec<CommentChange>,
+) -> (Transaction, SmallVec<[Range; 1]>) {
+    let mut changes: Vec<Change> = Vec::with_capacity(selection.len() * 2);
+    let mut ranges: SmallVec<[Range; 1]> = SmallVec::with_capacity(selection.len());
+    let mut offs = 0;
+    for change in comment_changes {
+        if commented {
+            if let CommentChange::Commented {
+                range,
+                start_pos,
+                end_pos,
+                start_token,
+                end_token,
+                start_margin,
+                end_margin,
+            } = change
+            {
+                let from = range.from();
+                changes.push((
+                    from + start_pos,
+                    from + start_pos + start_token.len() + start_margin as usize,
+                    None,
+                ));
+                changes.push((
+                    from + end_pos - end_token.len() - end_margin as usize + 1,
+                    from + end_pos + 1,
+                    None,
+                ));
+            }
+        } else {
+            // uncommented so manually map ranges through changes
+            match change {
+                CommentChange::Uncommented {
+                    range,
+                    start_pos,
+                    end_pos,
+                    start_token,
+                    end_token,
+                } => {
+                    let from = range.from();
+                    changes.push((
+                        from + start_pos,
+                        from + start_pos,
+                        Some(Tendril::from(format!("{} ", start_token))),
+                    ));
+                    changes.push((
+                        from + end_pos + 1,
+                        from + end_pos + 1,
+                        Some(Tendril::from(format!(" {}", end_token))),
+                    ));
+
+                    let offset = start_token.chars().count() + end_token.chars().count() + 2;
+                    ranges.push(
+                        Range::new(from + offs, from + offs + end_pos + 1 + offset)
+                            .with_direction(range.direction()),
+                    );
+                    offs += offset;
+                }
+                CommentChange::Commented { range, .. } | CommentChange::Whitespace { range } => {
+                    ranges.push(Range::new(range.from() + offs, range.to() + offs));
+                }
+            }
+        }
+    }
+    (Transaction::change(doc, changes.into_iter()), ranges)
+}
+
+#[must_use]
+pub fn toggle_block_comments(
+    doc: &Rope,
+    selection: &Selection,
+    tokens: &[BlockCommentToken],
+) -> Transaction {
+    let text = doc.slice(..);
+    let (commented, comment_changes) = find_block_comments(tokens, text, selection);
+    let (mut transaction, ranges) =
+        create_block_comment_transaction(doc, selection, commented, comment_changes);
+    if !commented {
+        transaction = transaction.with_selection(Selection::new(ranges, selection.primary_index()));
+    }
+    transaction
+}
+
+pub fn split_lines_of_selection(text: RopeSlice, selection: &Selection) -> Selection {
+    let mut ranges = SmallVec::new();
+    for range in selection.ranges() {
+        let (line_start, line_end) = range.line_range(text.slice(..));
+        let mut pos = text.line_to_char(line_start);
+        for line in text.slice(pos..text.line_to_char(line_end + 1)).lines() {
+            let start = pos;
+            pos += line.len_chars();
+            ranges.push(Range::new(start, pos));
+        }
+    }
+    Selection::new(ranges, 0)
+}
+
 #[cfg(test)]
 mod test {
     use super::*;
@@ -149,4 +368,49 @@ mod test {
 
         // TODO: account for uncommenting with uneven comment indentation
     }
+
+    #[test]
+    fn test_find_block_comments() {
+        // three lines 5 characters.
+        let mut doc = Rope::from("1\n2\n3");
+        // select whole document
+        let selection = Selection::single(0, doc.len_chars());
+
+        let text = doc.slice(..);
+
+        let res = find_block_comments(&[BlockCommentToken::default()], text, &selection);
+
+        assert_eq!(
+            res,
+            (
+                false,
+                vec![CommentChange::Uncommented {
+                    range: Range::new(0, 5),
+                    start_pos: 0,
+                    end_pos: 4,
+                    start_token: "/*".to_string(),
+                    end_token: "*/".to_string(),
+                }]
+            )
+        );
+
+        // comment
+        let transaction = toggle_block_comments(&doc, &selection, &[BlockCommentToken::default()]);
+        transaction.apply(&mut doc);
+
+        assert_eq!(doc, "/* 1\n2\n3 */");
+
+        // uncomment
+        let selection = Selection::single(0, doc.len_chars());
+        let transaction = toggle_block_comments(&doc, &selection, &[BlockCommentToken::default()]);
+        transaction.apply(&mut doc);
+        assert_eq!(doc, "1\n2\n3");
+
+        // don't panic when there is just a space in comment
+        doc = Rope::from("/* */");
+        let selection = Selection::single(0, doc.len_chars());
+        let transaction = toggle_block_comments(&doc, &selection, &[BlockCommentToken::default()]);
+        transaction.apply(&mut doc);
+        assert_eq!(doc, "");
+    }
 }
diff --git a/helix-core/src/indent.rs b/helix-core/src/indent.rs
index c29bb3a0..2a0a3876 100644
--- a/helix-core/src/indent.rs
+++ b/helix-core/src/indent.rs
@@ -1,10 +1,10 @@
 use std::{borrow::Cow, collections::HashMap};
 
+use helix_stdx::rope::RopeSliceExt;
 use tree_sitter::{Query, QueryCursor, QueryPredicateArg};
 
 use crate::{
     chars::{char_is_line_ending, char_is_whitespace},
-    find_first_non_whitespace_char,
     graphemes::{grapheme_width, tab_width_at},
     syntax::{IndentationHeuristic, LanguageConfiguration, RopeProvider, Syntax},
     tree_sitter::Node,
@@ -970,7 +970,7 @@ pub fn indent_for_newline(
                 let mut num_attempts = 0;
                 for line_idx in (0..=line_before).rev() {
                     let line = text.line(line_idx);
-                    let first_non_whitespace_char = match find_first_non_whitespace_char(line) {
+                    let first_non_whitespace_char = match line.first_non_whitespace_char() {
                         Some(i) => i,
                         None => {
                             continue;
diff --git a/helix-core/src/lib.rs b/helix-core/src/lib.rs
index 94802eba..1abd90d1 100644
--- a/helix-core/src/lib.rs
+++ b/helix-core/src/lib.rs
@@ -37,9 +37,6 @@ pub mod unicode {
 
 pub use helix_loader::find_workspace;
 
-pub fn find_first_non_whitespace_char(line: RopeSlice) -> Option<usize> {
-    line.chars().position(|ch| !ch.is_whitespace())
-}
 mod rope_reader;
 
 pub use rope_reader::RopeReader;
diff --git a/helix-core/src/syntax.rs b/helix-core/src/syntax.rs
index 0d8559ca..3b224e1b 100644
--- a/helix-core/src/syntax.rs
+++ b/helix-core/src/syntax.rs
@@ -99,7 +99,19 @@ pub struct LanguageConfiguration {
     pub shebangs: Vec<String>, // interpreter(s) associated with language
     #[serde(default)]
     pub roots: Vec<String>, // these indicate project roots <.git, Cargo.toml>
-    pub comment_token: Option<String>,
+    #[serde(
+        default,
+        skip_serializing,
+        deserialize_with = "from_comment_tokens",
+        alias = "comment-token"
+    )]
+    pub comment_tokens: Option<Vec<String>>,
+    #[serde(
+        default,
+        skip_serializing,
+        deserialize_with = "from_block_comment_tokens"
+    )]
+    pub block_comment_tokens: Option<Vec<BlockCommentToken>>,
     pub text_width: Option<usize>,
     pub soft_wrap: Option<SoftWrap>,
 
@@ -240,6 +252,59 @@ impl<'de> Deserialize<'de> for FileType {
     }
 }
 
+fn from_comment_tokens<'de, D>(deserializer: D) -> Result<Option<Vec<String>>, D::Error>
+where
+    D: serde::Deserializer<'de>,
+{
+    #[derive(Deserialize)]
+    #[serde(untagged)]
+    enum CommentTokens {
+        Multiple(Vec<String>),
+        Single(String),
+    }
+    Ok(
+        Option::<CommentTokens>::deserialize(deserializer)?.map(|tokens| match tokens {
+            CommentTokens::Single(val) => vec![val],
+            CommentTokens::Multiple(vals) => vals,
+        }),
+    )
+}
+
+#[derive(Clone, Debug, Serialize, Deserialize)]
+pub struct BlockCommentToken {
+    pub start: String,
+    pub end: String,
+}
+
+impl Default for BlockCommentToken {
+    fn default() -> Self {
+        BlockCommentToken {
+            start: "/*".to_string(),
+            end: "*/".to_string(),
+        }
+    }
+}
+
+fn from_block_comment_tokens<'de, D>(
+    deserializer: D,
+) -> Result<Option<Vec<BlockCommentToken>>, D::Error>
+where
+    D: serde::Deserializer<'de>,
+{
+    #[derive(Deserialize)]
+    #[serde(untagged)]
+    enum BlockCommentTokens {
+        Multiple(Vec<BlockCommentToken>),
+        Single(BlockCommentToken),
+    }
+    Ok(
+        Option::<BlockCommentTokens>::deserialize(deserializer)?.map(|tokens| match tokens {
+            BlockCommentTokens::Single(val) => vec![val],
+            BlockCommentTokens::Multiple(vals) => vals,
+        }),
+    )
+}
+
 #[derive(Clone, Copy, Debug, Serialize, Deserialize, PartialEq, Eq, Hash)]
 #[serde(rename_all = "kebab-case")]
 pub enum LanguageServerFeature {
diff --git a/helix-core/tests/indent.rs b/helix-core/tests/indent.rs
index 53265e0b..31946c56 100644
--- a/helix-core/tests/indent.rs
+++ b/helix-core/tests/indent.rs
@@ -4,6 +4,7 @@ use helix_core::{
     syntax::{Configuration, Loader},
     Syntax,
 };
+use helix_stdx::rope::RopeSliceExt;
 use ropey::Rope;
 use std::{ops::Range, path::PathBuf, process::Command, sync::Arc};
 
@@ -211,7 +212,7 @@ fn test_treesitter_indent(
         if ignored_lines.iter().any(|range| range.contains(&(i + 1))) {
             continue;
         }
-        if let Some(pos) = helix_core::find_first_non_whitespace_char(line) {
+        if let Some(pos) = line.first_non_whitespace_char() {
             let tab_width: usize = 4;
             let suggested_indent = treesitter_indent_for_pos(
                 indent_query,