Fix parsing \verb command

See #828.
latex-lsp · Apr 29, 2023 · 6188ddf · 6188ddf
1 parent b61313e
commit 6188ddf
Show file tree

Hide file tree

Showing 7 changed files with 66 additions and 6 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -11,6 +11,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 - Fix spurious completion results when completing environments ([#883](https://github.com/latex-lsp/texlab/issues/883))
 - Fix regression when guessing cursor position after formatting ([#880](https://github.com/latex-lsp/texlab/issues/880))
+- Fix parsing `\verb` command ([#828](https://github.com/latex-lsp/texlab/issues/828))
 
 ## [5.5.0] - 2023-04-16
 

diff --git a/crates/parser/src/latex.rs b/crates/parser/src/latex.rs
@@ -44,6 +44,11 @@ impl<'a> Parser<'a> {
         self.builder.token(kind.into(), text);
     }
 
+    fn eat_remap(&mut self, kind: SyntaxKind) {
+        let (_, text) = self.lexer.eat().unwrap();
+        self.builder.token(kind.into(), text);
+    }
+
     fn peek(&self) -> Option<Token> {
         self.lexer.peek()
     }
@@ -98,7 +103,7 @@ impl<'a> Parser<'a> {
                 self.eat();
                 self.builder.finish_node();
             }
-            Token::Word | Token::Comma => self.text(context),
+            Token::Pipe | Token::Word | Token::Comma => self.text(context),
             Token::Eq => self.eat(),
             Token::Dollar => self.formula(),
             Token::CommandName(name) => match name {
@@ -141,6 +146,7 @@ impl<'a> Parser<'a> {
                 CommandName::EnvironmentDefinition => self.environment_definition(),
                 CommandName::BeginBlockComment => self.block_comment(),
                 CommandName::EndBlockComment => self.generic_command(),
+                CommandName::VerbatimBlock => self.verbatim_block(),
                 CommandName::GraphicsPath => self.graphics_path(),
             },
         }
@@ -158,6 +164,7 @@ impl<'a> Parser<'a> {
                         | Token::Whitespace
                         | Token::LineComment
                         | Token::Word
+                        | Token::Pipe
                         | Token::Comma
                 ) && (context.allow_comma || kind != Token::Comma)
             })
@@ -219,7 +226,7 @@ impl<'a> Parser<'a> {
         self.eat();
         self.trivia();
         match self.peek() {
-            Some(Token::Word) => {
+            Some(Token::Word | Token::Pipe) => {
                 self.key();
             }
             Some(Token::CommandName(_)) => {
@@ -244,6 +251,7 @@ impl<'a> Parser<'a> {
                         | Token::Whitespace
                         | Token::LineComment
                         | Token::Word
+                        | Token::Pipe
                         | Token::Comma
                 )
             })
@@ -298,7 +306,7 @@ impl<'a> Parser<'a> {
         self.eat();
         self.trivia();
         match self.peek() {
-            Some(Token::Word) => {
+            Some(Token::Word | Token::Pipe) => {
                 self.key();
             }
             Some(_) | None => {}
@@ -334,7 +342,12 @@ impl<'a> Parser<'a> {
         self.eat();
         while self
             .peek()
-            .filter(|&kind| matches!(kind, Token::Whitespace | Token::LineComment | Token::Word))
+            .filter(|&kind| {
+                matches!(
+                    kind,
+                    Token::Whitespace | Token::LineComment | Token::Word | Token::Pipe
+                )
+            })
             .is_some()
         {
             self.eat();
@@ -390,7 +403,7 @@ impl<'a> Parser<'a> {
         while let Some(kind) = self.peek() {
             match kind {
                 Token::LineBreak | Token::Whitespace | Token::LineComment => self.eat(),
-                Token::Word => {
+                Token::Word | Token::Pipe => {
                     self.key_value_pair();
                     if self.peek() == Some(Token::Comma) {
                         self.eat();
@@ -1116,6 +1129,23 @@ impl<'a> Parser<'a> {
 
         self.builder.finish_node();
     }
+
+    fn verbatim_block(&mut self) {
+        self.builder.start_node(GENERIC_COMMAND.into());
+        self.eat();
+        self.builder.finish_node();
+        self.trivia();
+
+        if self.peek() == Some(Token::Pipe) {
+            self.eat_remap(SyntaxKind::VERBATIM);
+            while let Some(kind) = self.peek() {
+                self.eat_remap(SyntaxKind::VERBATIM);
+                if kind == Token::Pipe {
+                    break;
+                }
+            }
+        }
+    }
 }
 
 pub fn parse_latex(text: &str, config: &SyntaxConfig) -> GreenNode {

diff --git a/crates/parser/src/latex/lexer.rs b/crates/parser/src/latex/lexer.rs
@@ -38,6 +38,7 @@ impl<'a> Lexer<'a> {
             Token::RParen => SyntaxKind::R_PAREN,
             Token::Comma => SyntaxKind::COMMA,
             Token::Eq => SyntaxKind::EQUALITY_SIGN,
+            Token::Pipe => SyntaxKind::WORD,
             Token::Word => SyntaxKind::WORD,
             Token::Dollar => SyntaxKind::DOLLAR,
             Token::CommandName(_) => SyntaxKind::COMMAND_NAME,

diff --git a/crates/parser/src/latex/lexer/commands.rs b/crates/parser/src/latex/lexer/commands.rs
@@ -80,6 +80,7 @@ pub fn classify(name: &str, config: &SyntaxConfig) -> CommandName {
         "graphicspath" => CommandName::GraphicsPath,
         "iffalse" => CommandName::BeginBlockComment,
         "fi" => CommandName::EndBlockComment,
+        "verb" => CommandName::VerbatimBlock,
         _ if config.citation_commands.contains(name) => CommandName::Citation,
         _ => CommandName::Generic,
     }

diff --git a/crates/parser/src/latex/lexer/types.rs b/crates/parser/src/latex/lexer/types.rs
@@ -35,7 +35,10 @@ pub enum Token {
     #[token("=")]
     Eq,
 
-    #[regex(r"[^\s\\%\{\},\$\[\]\(\)=]+")]
+    #[token("|")]
+    Pipe,
+
+    #[regex(r"[^\s\\%\{\},\$\[\]\(\)=\|]+")]
     Word,
 
     #[regex(r"\$\$?")]
@@ -117,6 +120,7 @@ pub enum CommandName {
     GraphicsPath,
     BeginBlockComment,
     EndBlockComment,
+    VerbatimBlock,
 }
 
 #[derive(Debug, PartialEq, Eq, PartialOrd, Ord, Clone, Copy, Hash)]

diff --git a/crates/parser/src/snapshots/parser__latex__tests__parse@issue_828.txt.snap b/crates/parser/src/snapshots/parser__latex__tests__parse@issue_828.txt.snap
@@ -0,0 +1,22 @@
+---
+source: crates/parser/src/latex.rs
+expression: root
+input_file: crates/parser/src/test_data/latex/issue_828.txt
+---
+ROOT@0..51
+  PREAMBLE@0..51
+    GENERIC_COMMAND@0..5
+      COMMAND_NAME@0..5 "\\verb"
+    VERBATIM@5..6 "|"
+    VERBATIM@6..17 "<STATEMENT>"
+    VERBATIM@17..22 "     "
+    VERBATIM@22..24 "if"
+    VERBATIM@24..25 "("
+    VERBATIM@25..31 "<expr>"
+    VERBATIM@31..32 ")"
+    VERBATIM@32..33 "{"
+    VERBATIM@33..43 "<body>else"
+    VERBATIM@43..44 "{"
+    VERBATIM@44..50 "<body>"
+    VERBATIM@50..51 "|"
+
diff --git a/crates/parser/src/test_data/latex/issue_828.txt b/crates/parser/src/test_data/latex/issue_828.txt
@@ -0,0 +1 @@
+\verb|<STATEMENT>     if(<expr>){<body>else{<body>|