Handle \bibitem when checking undefined references (#1244)

Add a bibitem list to the latex Semantics struct, bibitem to CommandName and SyntaxKind enums, and handlers for processing bibitem nodes in a tex document.
Added checks against the bibitem list for undefined references and unused references.

crates/base-db/src/semantics/tex.rs

@@ -38,6 +38,7 @@ pub struct Semantics {
     pub can_be_root: bool,
     pub can_be_compiled: bool,
     pub diagnostic_suppressions: Vec<TextRange>,
+    pub bibitems: FxHashSet<Span>,
 }
 
 impl Semantics {
@@ -49,7 +50,7 @@ impl Semantics {
                 }
                 latex::SyntaxElement::Token(token) => match token.kind() {
                     latex::COMMAND_NAME => {
-                        self.commands.push(Span::command(&token));
+                            self.commands.push(Span::command(&token));
                     }
                     latex::COMMENT if token.text().contains("texlab: ignore") => {
                         self.diagnostic_suppressions.push(token.text_range());
@@ -85,6 +86,8 @@ impl Semantics {
             self.process_theorem_definition(theorem_def);
         } else if let Some(graphics_path) = latex::GraphicsPath::cast(node.clone()) {
             self.process_graphics_path(graphics_path);
+        } else if let Some(bibitem) = latex::BibItem::cast(node.clone()) {
+            self.process_bibitem(bibitem);
         }
     }
 
@@ -334,6 +337,16 @@ impl Semantics {
             self.graphics_paths.insert(path.to_string());
         }
     }
+
+    fn process_bibitem(&mut self, bibitem: latex::BibItem) {
+        if let Some(name) = bibitem.name() {
+            if let Some(key) = name.key() {
+                self.bibitems.insert(
+                    Span::from(&key)
+                );
+            }
+        }
+    }
 }
 
 #[derive(Debug, PartialEq, Eq, PartialOrd, Ord, Clone, Copy, Hash)]

crates/diagnostics/src/citations.rs

@@ -18,13 +18,17 @@ pub fn detect_undefined_citations<'a>(
 ) -> Option<()> {
     let data = document.data.as_tex()?;
 
+    let bibitems: FxHashSet<&str> = data.semantics.bibitems.iter()
+        .map(|bib| bib.text.as_str())
+        .collect();
+
     let entries: FxHashSet<&str> = Entry::find_all(project)
         .map(|(_, entry)| entry.name_text())
         .collect();
 
     for citation in &data.semantics.citations {
         let name = citation.name_text();
-        if name != "*" && !entries.contains(name) {
+        if name != "*" && !entries.contains(name) && !bibitems.contains(name) {
             let diagnostic = Diagnostic::Tex(citation.name.range, TexError::UndefinedCitation);
             results
                 .entry(document.uri.clone())
@@ -41,24 +45,38 @@ pub fn detect_unused_entries<'a>(
     document: &'a Document,
     results: &mut FxHashMap<Url, Vec<Diagnostic>>,
 ) -> Option<()> {
-    let data = document.data.as_bib()?;
-
-    // If this is a huge bibliography, then don't bother checking for unused entries.
-    if data.semantics.entries.len() > MAX_UNUSED_ENTRIES {
-        return None;
-    }
 
     let citations: FxHashSet<&str> = Citation::find_all(project)
         .map(|(_, citation)| citation.name_text())
         .collect();
 
-    for entry in &data.semantics.entries {
-        if !citations.contains(entry.name.text.as_str()) {
-            let diagnostic = Diagnostic::Bib(entry.name.range, BibError::UnusedEntry);
-            results
-                .entry(document.uri.clone())
-                .or_default()
-                .push(diagnostic);
+    if let Some(data) = document.data.as_bib()
+    {
+        // If this is a huge bibliography, then don't bother checking for unused entries.
+        if data.semantics.entries.len() > MAX_UNUSED_ENTRIES {
+            return None;
+        }
+
+        for entry in &data.semantics.entries {
+            if !citations.contains(entry.name.text.as_str()) {
+                let diagnostic = Diagnostic::Bib(entry.name.range, BibError::UnusedEntry);
+                results
+                    .entry(document.uri.clone())
+                    .or_default()
+                    .push(diagnostic);
+            }
+        }
+    };
+
+    if let Some(tex_data) = document.data.as_tex(){
+        for bibitem in &tex_data.semantics.bibitems {
+            if !citations.contains(bibitem.text.as_str()) {
+                let diagnostic = Diagnostic::Bib(bibitem.range, BibError::UnusedEntry);
+                results
+                    .entry(document.uri.clone())
+                    .or_default()
+                    .push(diagnostic);
+            }
         }
     }
 

crates/parser/src/latex.rs

@@ -158,6 +158,7 @@ impl<'a> Parser<'a> {
                 CommandName::EndBlockComment => self.generic_command(),
                 CommandName::VerbatimBlock => self.verbatim_block(),
                 CommandName::GraphicsPath => self.graphics_path(),
+                CommandName::BibItem => self.bibitem(),
             },
         }
     }
@@ -1235,6 +1236,18 @@ impl<'a> Parser<'a> {
             }
         }
     }
+
+    fn bibitem(&mut self) {
+        self.builder.start_node(BIBITEM.into());
+        self.eat();
+        self.trivia();
+
+        if self.lexer.peek() == Some(Token::LCurly) {
+            self.curly_group_word();
+        }
+
+        self.builder.finish_node();
+    }
 }
 
 pub fn parse_latex(text: &str, config: &SyntaxConfig) -> GreenNode {

crates/parser/src/latex/lexer/commands.rs

@@ -96,6 +96,7 @@ pub fn classify(name: &str, config: &SyntaxConfig) -> CommandName {
         "iffalse" => CommandName::BeginBlockComment,
         "fi" => CommandName::EndBlockComment,
         "verb" => CommandName::VerbatimBlock,
+        "bibitem" => CommandName::BibItem,
 
         _ if config.citation_commands.contains(name) => CommandName::Citation,
         _ if config.label_definition_commands.contains(name) => CommandName::LabelDefinition,

crates/parser/src/latex/lexer/types.rs

@@ -95,6 +95,7 @@ pub enum CommandName {
     BeginBlockComment,
     EndBlockComment,
     VerbatimBlock,
+    BibItem,
 }
 
 #[derive(Debug, PartialEq, Eq, PartialOrd, Ord, Clone, Copy, Hash)]

crates/syntax/src/latex/cst.rs

@@ -750,3 +750,15 @@ impl GraphicsPath {
         self.syntax().descendants().filter_map(CurlyGroupWord::cast)
     }
 }
+
+cst_node!(BibItem, BIBITEM);
+
+impl BibItem {
+    pub fn command(&self) -> Option<SyntaxToken> {
+        self.syntax().first_token()
+    }
+
+    pub fn name(&self) -> Option<CurlyGroupWord> {
+        self.syntax().children().find_map(CurlyGroupWord::cast)
+    }
+}

crates/syntax/src/latex/kind.rs

@@ -82,6 +82,7 @@ pub enum SyntaxKind {
     ENVIRONMENT_DEFINITION,
     GRAPHICS_PATH,
     BLOCK_COMMENT,
+    BIBITEM,
     ROOT,
 }