tree-sitter-aspx

A tree-sitter grammar for ASP.NET Web Forms markup: .aspx pages, .ascx user controls, .master master pages, and the directive headers of .asax, .asmx, and .ashx files.

Classic ASP (.asp with VBScript) and Razor (.cshtml, .vbhtml) are different languages and are out of scope. Code-behind files (.aspx.cs, .aspx.vb) are ordinary C# or VB.NET and are not parsed by this grammar.

The grammar parses directives, server comments, HTML comments, code nuggets, <script> and <style> elements, and tags with attributes. It does not validate HTML. Void tags (br, img, meta, and the other HTML void elements) do not swallow the following markup. An end tag closes the nearest open element; the names are not required to match. <script> and <style> bodies are raw text, so < inside JavaScript or CSS is not a tag.

Embedded C# and VB.NET

Code nuggets and server-script bodies are opaque ranges. The page Language attribute (C# or VB, any casing) is a language_value node, and the scanner labels each later nugget c_sharp_code or vb_code. When the attribute is absent, the label is c_sharp_code.

| Node | What it is | | --- | --- | | encoded_expression, output_expression, binding_expression | <%: %>, <%= %>, <%# %> | | statement_block | <% %> | | expression_builder / expression_value | <%$ %>, not C# or VB | | server_comment | <%-- --%>, not code | | c_sharp_code, vb_code | The bytes inside a nugget or a <script> body | | handler_code | Everything after a WebHandler or WebService directive that is not a <script> element |

queries/injections.scm maps those code nodes to c_sharp or vb. Statement blocks set injection.combined, so <% if (ready) { %> … <% } %> is one injected document and the markup between them is not part of it. Output, encoded, and binding nuggets are injected one by one. A <script> whose start tag has runat="server" keeps the page language. Any other <script> body is injected as JavaScript, even though the node is still c_sharp_code or vb_code. <style> bodies are injected as CSS.

A consumer that counts executable code should count c_sharp_code and vb_code bytes, and should skip a <script> body whose start tag has no runat="server". Markup, attribute text, style text, expression-builder text, and server comments are not executable code. Server comments are their own node kind.

.ashx and .asmx files are a WebHandler or WebService directive plus either a <script runat="server"> element or the rest of the file as handler_code. .asax stays markup.

Some legacy pages are UTF-16. Decode them to UTF-8 before parsing. tools/corpus-check.mjs does that.

Node package metadata

The default export exposes name, language, and nodeTypeInfo on Linux and Windows. nodeTypeInfo contains the entries from src/node-types.json.

import Parser from 'tree-sitter';
import Aspx from 'tree-sitter-aspx';

const parser = new Parser();
parser.setLanguage(Aspx);
const tree = parser.parse('<%@ Page Language="C#" %>');
console.log(tree.rootNode.toString());

Rust

let mut parser = tree_sitter::Parser::new();
parser
    .set_language(&tree_sitter_aspx::LANGUAGE.into())
    .expect("Error loading ASPX parser");

LANGUAGE is a tree_sitter_language::LanguageFn. The crate's runtime dependency is tree-sitter-language, so the consuming crate's own tree-sitter version is what parses.

Building

npm ci --ignore-scripts
npm rebuild tree-sitter-cli
npx --no-install tree-sitter generate
cargo build

Release jobs require a version tag on main matching the Cargo package version and publish from the commit that passed verification. The npm and GitHub Packages archives carry that same version in both package.json and tree-sitter.json.