feat: added initial HTML support

Cargo.toml

@@ -1,5 +1,5 @@
 [workspace]
-members = [ "harper-cli", "harper-core", "harper-ls", "harper-comments", "harper-wasm", "harper-tree-sitter"]
+members = [ "harper-cli", "harper-core", "harper-ls", "harper-comments", "harper-wasm", "harper-tree-sitter", "harper-html"]
 resolver = "2"
 
 [profile.release]

harper-cli/src/main.rs

@@ -17,13 +17,13 @@ enum Args {
         /// Whether to merely print out the number of errors encountered,
         /// without further details.
         #[arg(short, long)]
-        count: bool
+        count: bool,
     },
     /// Parse a provided document and print the detected symbols.
     Parse {
         /// The file you wish to parse.
-        file: PathBuf
-    }
+        file: PathBuf,
+    },
 }
 
 fn main() -> anyhow::Result<()> {
@@ -35,7 +35,7 @@ fn main() -> anyhow::Result<()> {
 
             let mut linter = LintGroup::new(
                 LintGroupConfig::default(),
-                FullDictionary::create_from_curated()
+                FullDictionary::create_from_curated(),
             );
             let mut lints = linter.lint(&doc);
 
@@ -64,7 +64,7 @@ fn main() -> anyhow::Result<()> {
                 report_builder = report_builder.with_label(
                     Label::new((&filename, lint.span.into()))
                         .with_message(lint.message)
-                        .with_color(primary_color)
+                        .with_color(primary_color),
                 );
             }
 

harper-comments/README.md

@@ -2,4 +2,4 @@
 
 This crate holds a number of functions, but it is primarily a wrapper around `tree-sitter` that allows Harper to locate the comments of a wide variety of programming languages.
 It also has purpose-built parsers for the structured comments of a number of languages, including Go.
-These additional parsers are avaiable through the `TreeSitterParser` and are enabled automatically through there.
+These additional parsers are available through the `CommentParser` and are enabled automatically through there.

harper-html/Cargo.toml

@@ -0,0 +1,14 @@
+[package]
+name = "harper-html"
+version = "0.8.8"
+edition = "2021"
+description = "The language checker for developers."
+license = "Apache-2.0"
+readme = "README.md"
+repository = "https://github.com/elijah-potter/harper"
+
+[dependencies]
+harper-core = { path = "../harper-core", version = "0.8.0" }
+harper-tree-sitter = { path = "../harper-tree-sitter", version = "0.8.0" }
+tree-sitter-html = "0.19.0"
+tree-sitter = "0.20.10"

harper-html/src/lib.rs

@@ -0,0 +1,29 @@
+use harper_core::parsers::{self, Parser, PlainEnglish};
+use harper_core::Token;
+use harper_tree_sitter::TreeSitterMasker;
+use tree_sitter::Node;
+
+pub struct HtmlParser {
+    inner: parsers::Mask<TreeSitterMasker, PlainEnglish>
+}
+
+impl HtmlParser {
+    pub fn new() -> Self {
+        Self {
+            inner: parsers::Mask::new(
+                TreeSitterMasker::new(tree_sitter_html::language(), Self::node_condition),
+                PlainEnglish
+            )
+        }
+    }
+
+    fn node_condition(n: &Node) -> bool {
+        n.kind() == "text"
+    }
+}
+
+impl Parser for HtmlParser {
+    fn parse(&mut self, source: &[char]) -> Vec<Token> {
+        self.inner.parse(source)
+    }
+}

harper-ls/Cargo.toml

@@ -10,6 +10,7 @@ repository = "https://github.com/elijah-potter/harper"
 [dependencies]
 harper-core = { path = "../harper-core", version = "0.8.0", features = ["concurrent"] }
 harper-comments = { path = "../harper-comments", version = "0.8.0" }
+harper-html = { path = "../harper-html", version = "0.8.0" }
 tower-lsp = "0.20.0"
 tokio = { version = "1.36.0", features = ["fs", "rt", "rt-multi-thread", "macros", "io-std", "io-util", "net"] }
 clap = { version = "4.5.1", features = ["derive"] }

harper-ls/src/backend.rs

@@ -5,45 +5,21 @@ use std::sync::Arc;
 use harper_comments::CommentParser;
 use harper_core::parsers::{Markdown, PlainEnglish};
 use harper_core::{
-    Dictionary,
-    Document,
-    FullDictionary,
-    LintGroup,
-    Linter,
-    MergedDictionary,
-    Token,
-    TokenKind
+    Dictionary, Document, FullDictionary, LintGroup, Linter, MergedDictionary, Token, TokenKind,
 };
+use harper_html::HtmlParser;
 use serde_json::Value;
 use tokio::sync::{Mutex, RwLock};
 use tower_lsp::jsonrpc::Result;
 use tower_lsp::lsp_types::notification::PublishDiagnostics;
 use tower_lsp::lsp_types::{
-    CodeActionOrCommand,
-    CodeActionParams,
-    CodeActionProviderCapability,
-    CodeActionResponse,
-    Command,
-    Diagnostic,
-    DidChangeConfigurationParams,
-    DidChangeTextDocumentParams,
-    DidCloseTextDocumentParams,
-    DidOpenTextDocumentParams,
-    DidSaveTextDocumentParams,
-    ExecuteCommandOptions,
-    ExecuteCommandParams,
-    InitializeParams,
-    InitializeResult,
-    InitializedParams,
-    MessageType,
-    PublishDiagnosticsParams,
-    Range,
-    ServerCapabilities,
-    TextDocumentSyncCapability,
-    TextDocumentSyncKind,
-    TextDocumentSyncOptions,
-    TextDocumentSyncSaveOptions,
-    Url
+    CodeActionOrCommand, CodeActionParams, CodeActionProviderCapability, CodeActionResponse,
+    Command, Diagnostic, DidChangeConfigurationParams, DidChangeTextDocumentParams,
+    DidCloseTextDocumentParams, DidOpenTextDocumentParams, DidSaveTextDocumentParams,
+    ExecuteCommandOptions, ExecuteCommandParams, InitializeParams, InitializeResult,
+    InitializedParams, MessageType, PublishDiagnosticsParams, Range, ServerCapabilities,
+    TextDocumentSyncCapability, TextDocumentSyncKind, TextDocumentSyncOptions,
+    TextDocumentSyncSaveOptions, Url,
 };
 use tower_lsp::{Client, LanguageServer};
 use tracing::{error, info, instrument};
@@ -59,15 +35,15 @@ struct DocumentState {
     document: Document,
     ident_dict: Arc<FullDictionary>,
     linter: LintGroup<MergedDictionary<FullDictionary>>,
-    language_id: Option<String>
+    language_id: Option<String>,
 }
 
 /// Deallocate
 pub struct Backend {
     client: Client,
     static_dictionary: Arc<FullDictionary>,
     config: RwLock<Config>,
-    doc_state: Mutex<HashMap<Url, DocumentState>>
+    doc_state: Mutex<HashMap<Url, DocumentState>>,
 }
 
 impl Backend {
@@ -78,7 +54,7 @@ impl Backend {
             client,
             static_dictionary: dictionary.into(),
             doc_state: Mutex::new(HashMap::new()),
-            config: RwLock::new(config)
+            config: RwLock::new(config),
         }
     }
 
@@ -173,7 +149,7 @@ impl Backend {
     #[instrument(skip(self))]
     async fn generate_file_dictionary(
         &self,
-        url: &Url
+        url: &Url,
     ) -> anyhow::Result<MergedDictionary<FullDictionary>> {
         let (global_dictionary, file_dictionary) = tokio::join!(
             self.generate_global_dictionary(),
@@ -190,11 +166,11 @@ impl Backend {
     async fn update_document_from_file(
         &self,
         url: &Url,
-        language_id: Option<&str>
+        language_id: Option<&str>,
     ) -> anyhow::Result<()> {
         let content = match tokio::fs::read_to_string(
             url.to_file_path()
-                .map_err(|_| anyhow::format_err!("Could not extract file path."))?
+                .map_err(|_| anyhow::format_err!("Could not extract file path."))?,
         )
         .await
         {
@@ -213,7 +189,7 @@ impl Backend {
         &self,
         url: &Url,
         text: &str,
-        language_id: Option<&str>
+        language_id: Option<&str>,
     ) -> anyhow::Result<()> {
         let mut doc_lock = self.doc_state.lock().await;
         let config_lock = self.config.read().await;
@@ -225,7 +201,7 @@ impl Backend {
         let mut doc_state = DocumentState {
             linter: LintGroup::new(
                 config_lock.lint_config,
-                self.generate_file_dictionary(url).await?
+                self.generate_file_dictionary(url).await?,
             ),
             language_id: language_id
                 .map(|v| v.to_string())
@@ -260,6 +236,8 @@ impl Backend {
                 Document::new(text, Box::new(Markdown))
             } else if language_id == "gitcommit" {
                 Document::new(text, Box::new(GitCommitParser))
+            } else if language_id == "html" {
+                Document::new(text, Box::new(HtmlParser::new()))
             } else if language_id == "mail" {
                 Document::new(text, Box::new(PlainEnglish))
             } else {
@@ -276,7 +254,7 @@ impl Backend {
     async fn generate_code_actions(
         &self,
         url: &Url,
-        range: Range
+        range: Range,
     ) -> Result<Vec<CodeActionOrCommand>> {
         let (config, mut doc_states) = tokio::join!(self.config.read(), self.doc_state.lock());
         let Some(doc_state) = doc_states.get_mut(url) else {
@@ -308,7 +286,7 @@ impl Backend {
             actions.push(CodeActionOrCommand::Command(Command::new(
                 "Open URL".to_string(),
                 "HarperOpen".to_string(),
-                Some(vec![doc_state.document.get_span_content_str(span).into()])
+                Some(vec![doc_state.document.get_span_content_str(span).into()]),
             )))
         }
 
@@ -328,7 +306,7 @@ impl Backend {
         lints_to_diagnostics(
             doc_state.document.get_full_content(),
             &lints,
-            config.diagnostic_severity
+            config.diagnostic_severity,
         )
     }
 
@@ -339,7 +317,7 @@ impl Backend {
         let result = PublishDiagnosticsParams {
             uri: url.clone(),
             diagnostics,
-            version: None
+            version: None,
         };
 
         self.client
@@ -369,12 +347,11 @@ impl LanguageServer for Backend {
                         change: Some(TextDocumentSyncKind::FULL),
                         will_save: None,
                         will_save_wait_until: None,
-                        save: Some(TextDocumentSyncSaveOptions::Supported(true))
-                    }
+                        save: Some(TextDocumentSyncSaveOptions::Supported(true)),
+                    },
                 )),
-
                 ..Default::default()
-            }
+            },
         })
     }
 
@@ -405,7 +382,7 @@ impl LanguageServer for Backend {
             .update_document(
                 &params.text_document.uri,
                 &params.text_document.text,
-                Some(&params.text_document.language_id)
+                Some(&params.text_document.language_id),
             )
             .await;
 
@@ -494,7 +471,7 @@ impl LanguageServer for Backend {
                     error!("Unable to open URL: {}", err);
                 }
             },
-            _ => ()
+            _ => (),
         }
 
         Ok(None)
@@ -525,7 +502,7 @@ impl LanguageServer for Backend {
         // Update documents with new config
         futures::future::join_all(
             keys.iter()
-                .map(|key| self.update_document_from_file(key, None))
+                .map(|key| self.update_document_from_file(key, None)),
         )
         .await;
     }