feat: do not collapse consecutive spaces in markup by default

Author: QuadnucYard

README.md

@@ -53,10 +53,10 @@ Options:
   -V, --version  Print version
 
 Format Configuration:
-  -l, --line-width <LINE_WIDTH>      The column width of the output [default: 80] [aliases: column] [short aliases: c]
-  -t, --indent-width <INDENT_WIDTH>  Spaces per level of indentation in the output [default: 2] [aliases: tab-width]
-      --no-reorder-import-items      Whether not to reorder import items
-      --wrap-text                    Whether to wrap texts in the markup
+  -l, --line-width <LINE_WIDTH>      Maximum width of each line [default: 80] [aliases: column] [short aliases: c]
+  -t, --indent-width <INDENT_WIDTH>  Number of spaces per indentation level [default: 2] [aliases: tab-width]
+      --no-reorder-import-items      Disable alphabetical reordering of import items
+      --wrap-text                    Wrap text in markup to fit within the line width. Implies `--collapse-spaces`
 
 Debug Options:
   -a, --ast         Print the AST of the input file

crates/typstyle-core/src/config.rs

@@ -2,16 +2,18 @@
 #[derive(Debug, Clone, PartialEq, Eq, Hash)]
 #[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
 pub struct Config {
-    /// Number of spaces per tab
+    /// Number of spaces to use for each indentation level.
     pub tab_spaces: usize,
     /// Maximum width of each line.
     pub max_width: usize,
-    /// Maximum number of blank lines which can be put between items.
+    /// Maximum number of consecutive blank lines allowed between code items.
     pub blank_lines_upper_bound: usize,
-    /// Whether to reorder import items.
-    /// When enabled, import items will be sorted alphabetically.
+    /// When `true`, consecutive whitespace in markup is collapsed into a single space.
+    pub collapse_markup_spaces: bool,
+    /// When `true`, import items are sorted alphabetically.
     pub reorder_import_items: bool,
-    /// Whether to wrap texts in the markup.
+    /// When `true`, text in markup will be wrapped to fit within `max_width`.
+    /// Implies `collapse_markup_spaces`.
     pub wrap_text: bool,
 }
 
@@ -22,6 +24,7 @@ impl Default for Config {
             max_width: 80,
             blank_lines_upper_bound: 2,
             reorder_import_items: true,
+            collapse_markup_spaces: false,
             wrap_text: false,
         }
     }

crates/typstyle-core/src/pretty/markup.rs

@@ -2,10 +2,7 @@ use smallvec::SmallVec;
 use typst_syntax::{ast::*, SyntaxKind, SyntaxNode};
 
 use super::{
-    layout::flow::FlowItem,
-    prelude::*,
-    util::{is_comment_node, is_only_one_and},
-    Context, Mode, PrettyPrinter,
+    layout::flow::FlowItem, prelude::*, util::is_comment_node, Context, Mode, PrettyPrinter,
 };
 use crate::ext::StrExt;
 
@@ -55,7 +52,7 @@ impl<'a> PrettyPrinter<'a> {
         body.enclose("_", "_")
     }
 
-    pub(super) fn convert_raw(&'a self, _ctx: Context, raw: Raw<'a>) -> ArenaDoc<'a> {
+    pub(super) fn convert_raw(&'a self, ctx: Context, raw: Raw<'a>) -> ArenaDoc<'a> {
         // no format multiline single backtick raw block
         if !raw.block() && raw.lines().nth(1).is_some() {
             return self.convert_verbatim(raw);
@@ -70,7 +67,7 @@ impl<'a> PrettyPrinter<'a> {
             } else if let Some(text) = child.cast::<Text>() {
                 doc += self.convert_text(text);
             } else if child.kind() == SyntaxKind::RawTrimmed {
-                doc += self.convert_space_untyped(child);
+                doc += self.convert_space_untyped(ctx, child);
             }
         }
         doc
@@ -187,10 +184,12 @@ impl<'a> PrettyPrinter<'a> {
     ) -> ArenaDoc<'a> {
         let ctx = ctx.with_mode(Mode::Markup);
 
-        if is_only_one_and(markup.to_untyped().children(), |node| {
-            node.kind() == SyntaxKind::Space
-        }) {
-            return self.arena.space();
+        // If the markup only contains one space, simply convert it.
+        let children = markup.to_untyped().children().as_slice();
+        if children.len() == 1 {
+            if let Some(space) = children[0].cast::<Space>() {
+                return self.convert_space(ctx, space);
+            }
         }
 
         let repr = collect_markup_repr(markup);
@@ -249,7 +248,7 @@ impl<'a> PrettyPrinter<'a> {
         {
             for node in nodes.iter() {
                 doc += if node.kind() == SyntaxKind::Space {
-                    self.arena.space()
+                    self.convert_space_untyped(ctx, node)
                 } else if let Some(text) = node.cast::<Text>() {
                     self.convert_text(text)
                 } else if let Some(expr) = node.cast::<Expr>() {

crates/typstyle-core/src/pretty/math.rs

@@ -94,7 +94,7 @@ impl<'a> PrettyPrinter<'a> {
                 let expr_doc = self.convert_expr(ctx, expr);
                 doc += expr_doc;
             } else if let Some(space) = node.cast::<Space>() {
-                doc += self.convert_space(space);
+                doc += self.convert_space(ctx, space);
             } else if node.kind() == SyntaxKind::Hash {
                 doc += self.arena.text("#");
                 peek_hash = true;
@@ -137,7 +137,7 @@ impl<'a> PrettyPrinter<'a> {
             if last.kind() == SyntaxKind::Space {
                 has_close_space = true;
                 inner_nodes = rest;
-                self.convert_space_untyped(last)
+                self.convert_space_untyped(ctx, last)
             } else {
                 self.arena.nil()
             }
@@ -154,7 +154,7 @@ impl<'a> PrettyPrinter<'a> {
                 FlowItem::tight(self.convert_math(ctx, math))
             } else if node.kind() == SyntaxKind::Space {
                 // We can not arbitrarily break line here, as it may become ugly.
-                FlowItem::tight(self.convert_space_untyped(node))
+                FlowItem::tight(self.convert_space_untyped(ctx, node))
             } else {
                 FlowItem::none()
             }
@@ -209,7 +209,7 @@ impl<'a> PrettyPrinter<'a> {
             if let Some(expr) = node.cast::<Expr>() {
                 FlowItem::spaced(self.convert_expr(ctx, expr))
             } else if let Some(space) = node.cast::<Space>() {
-                FlowItem::tight(self.convert_space(space))
+                FlowItem::tight(self.convert_space(ctx, space))
             } else {
                 FlowItem::tight(self.convert_trivia_untyped(node))
             }

crates/typstyle-core/src/pretty/mod.rs

@@ -100,7 +100,7 @@ impl<'a> PrettyPrinter<'a> {
     fn convert_expr_impl(&'a self, ctx: Context, expr: Expr<'a>) -> ArenaDoc<'a> {
         match expr {
             Expr::Text(t) => self.convert_text(t),
-            Expr::Space(s) => self.convert_space(s),
+            Expr::Space(s) => self.convert_space(ctx, s),
             Expr::Linebreak(b) => self.convert_trivia(b),
             Expr::Parbreak(b) => self.convert_parbreak(b),
             Expr::Escape(e) => self.convert_trivia(e),

crates/typstyle-core/src/pretty/text.rs

@@ -1,6 +1,6 @@
 use typst_syntax::{ast::*, SyntaxNode};
 
-use super::{prelude::*, PrettyPrinter};
+use super::{prelude::*, Context, PrettyPrinter};
 use crate::ext::StrExt;
 
 impl<'a> PrettyPrinter<'a> {
@@ -13,13 +13,19 @@ impl<'a> PrettyPrinter<'a> {
         wrap_text(&self.arena, text.get())
     }
 
-    pub(super) fn convert_space(&'a self, space: Space) -> ArenaDoc<'a> {
-        self.convert_space_untyped(space.to_untyped())
+    pub(super) fn convert_space(&'a self, ctx: Context, space: Space<'a>) -> ArenaDoc<'a> {
+        self.convert_space_untyped(ctx, space.to_untyped())
     }
 
-    pub(super) fn convert_space_untyped(&'a self, node: &SyntaxNode) -> ArenaDoc<'a> {
+    pub(super) fn convert_space_untyped(
+        &'a self,
+        ctx: Context,
+        node: &'a SyntaxNode,
+    ) -> ArenaDoc<'a> {
         if node.text().has_linebreak() {
             self.arena.hardline()
+        } else if ctx.mode.is_markup() && !self.config.collapse_markup_spaces {
+            self.arena.text(node.text().as_str())
         } else {
             self.arena.space()
         }

crates/typstyle/src/cli.rs

@@ -66,7 +66,7 @@ pub enum Command {
 
 #[derive(Args)]
 pub struct StyleArgs {
-    /// The column width of the output
+    /// Maximum width of each line.
     #[arg(
         short = 'l',
         long,
@@ -77,7 +77,7 @@ pub struct StyleArgs {
     )]
     pub line_width: usize,
 
-    /// Spaces per level of indentation in the output
+    /// Number of spaces per indentation level.
     #[arg(
         short = 't',
         long,
@@ -87,11 +87,11 @@ pub struct StyleArgs {
     )]
     pub indent_width: usize,
 
-    /// Whether not to reorder import items.
+    /// Disable alphabetical reordering of import items.
     #[arg(long, default_value_t = false, global = true)]
     pub no_reorder_import_items: bool,
 
-    /// Whether to wrap texts in the markup.
+    /// Wrap text in markup to fit within the line width. Implies `--collapse-spaces`.
     #[arg(long, default_value_t = false, global = true)]
     pub wrap_text: bool,
 }

crates/typstyle/tests/test_style_args.rs

@@ -54,7 +54,7 @@ fn test_reorder_import_items() {
 fn test_wrap_text() {
     let space = Workspace::new();
 
-    let stdin = "lorem ipsum dolor sit amet, consectetur adipiscing elit.";
+    let stdin = "lorem  ipsum   dolor sit amet, consectetur   adipiscing elit.";
 
     typstyle_cmd_snapshot!(space.cli().args(["-c=20", "--wrap-text"]).pass_stdin(stdin), @r"
     success: true

tests/fixtures/articles/snap/undergraduate-math.typ-0.snap

@@ -585,7 +585,7 @@ Comment on an expression as here (there is also `overbrace(..)`).
 
 = Dots
 Use low dots in a list ${0, 1, 2, ...}$, entered as `{0, 1, 2, ...}`.
-Use centered dots in a sum or product $1 + dots.h.c + 100$, entered as `1 + dots.h.c + 100`.
+Use centered dots in a sum or product $1 + dots.h.c + 100$, entered as  `1 + dots.h.c + 100`.
 You can also get vertical dots `dots.v`, diagonal dots `dots.down` and anti-diagonal dots `dots.up`.
 
 = Roman names

tests/fixtures/articles/snap/undergraduate-math.typ-120.snap

@@ -238,7 +238,7 @@ Comment on an expression as here (there is also `overbrace(..)`).
 
 = Dots
 Use low dots in a list ${0, 1, 2, ...}$, entered as `{0, 1, 2, ...}`.
-Use centered dots in a sum or product $1 + dots.h.c + 100$, entered as `1 + dots.h.c + 100`.
+Use centered dots in a sum or product $1 + dots.h.c + 100$, entered as  `1 + dots.h.c + 100`.
 You can also get vertical dots `dots.v`, diagonal dots `dots.down` and anti-diagonal dots `dots.up`.
 
 = Roman names