WIP

Author: AugustinMauroy

utils/src/ast-grep/shebang.test.ts

@@ -18,12 +18,12 @@ describe('shebang', () => {
 			assert.equal(shebang?.text(), '#!/usr/bin/env node');
 		});
 
-		it('should take the last shebang line if multiple exist on top of the code', () => {
+		it('should take the first shebang line if multiple exist on top of the code', () => {
 			const code = dedent`
-				#!/usr/bin/env node 1
-				#!/usr/bin/env node 2
-				console.log("Hello, world!");
-			`;
+			#!/usr/bin/env node 1
+			#!/usr/bin/env node 2
+			console.log("Hello, world!");
+		`;
 			const ast = astGrep.parse(astGrep.Lang.JavaScript, code);
 
 			const shebang = getShebang(ast);

utils/src/ast-grep/shebang.ts

@@ -4,24 +4,47 @@ const REGEX_ESCAPE_PATTERN = /[.*+?^${}()|[\]\\]/g;
 
 /**
  * Get the shebang line from the root.
+ * According to ECMAScript spec, shebangs (InputElementHashbangOrRegExp) are only
+ * valid at the start of a Script or Module. We find hash_bang_lines that appear
+ * at the beginning before any actual code. When multiple consecutive shebangs exist at the top,
+ * we return the last one as it would be the effective shebang used.
  * @param root The root node to search.
  * @returns The shebang line if found, otherwise null.
  */
-export const getShebang = (root: SgRoot) =>
-	root.root().find({
+export const getShebang = (root: SgRoot) => {
+	const allShebangs = root.root().findAll({
 		rule: {
 			kind: 'hash_bang_line',
 			regex: '\\bnode(\\.exe)?\\b',
-			not: {
-				// tree-sitter wraps hash-bang in Error node
-				// when it's not in the top of program node
-				inside: {
-					kind: 'ERROR',
-				},
-			},
 		},
 	});
 
+	if (!allShebangs.length) return null;
+
+	// Check if first shebang is at line 0 (start of file)
+	const firstShebang = allShebangs[0];
+	if (firstShebang.range().start.line !== 0) {
+		return null; // Shebang not at start of file
+	}
+
+	// Collect all consecutive shebangs from the start
+	const validShebangs = [firstShebang];
+	for (let i = 1; i < allShebangs.length; i++) {
+		const prevLine = allShebangs[i - 1].range().end.line;
+		const currentLine = allShebangs[i].range().start.line;
+
+		// Check if this shebang is on the next consecutive line
+		if (currentLine === prevLine || currentLine === prevLine + 1) {
+			validShebangs.push(allShebangs[i]);
+		} else {
+			break; // Stop at first non-consecutive shebang
+		}
+	}
+
+	// Return the last consecutive shebang from the start
+	return validShebangs[validShebangs.length - 1];
+};
+
 /**
  * Replace Node.js arguments in the shebang line.
  * @param root The root node to search.