Added BOM detection and BOM marker removal

If a source code uses encoding with BOM marker, this marker becomes a part of a prefix in a JS.CompilationUnit. During the equality check on the Java side (org.openrewrite.Parser.requirePrintEqualsInput) between the source code and the code generated based on the LST, Java trims the BOM marker. So this causes an idempotency error. Therefore, we also detect and trim the BOM marker on a JS/TS parser side from JS.CompilationUnit prefix.
openrewrite · Jan 8, 2025 · 779c678 · 779c678
1 parent 2d8ec7c
commit 779c678
Showing 1 changed file with 33 additions and 3 deletions.
diff --git a/openrewrite/src/javascript/parser.ts b/openrewrite/src/javascript/parser.ts
@@ -210,15 +210,45 @@ export class JavaScriptParserVisitor {
         return this.visit(node) as T;
     }
 
+    detectBOMAndTextEncoding(content: String) : { hasBom: boolean; encoding: string | null } {
+        const BOM_UTF8 = "\uFEFF"; // BOM for UTF-8
+        const BOM_UTF16_LE = [0xFF, 0xFE]; // BOM for UTF-16 Little Endian
+
+        // Detect BOM
+        const hasUtf8Bom = content.startsWith(BOM_UTF8);
+        const hasUtf16LeBom = content.charCodeAt(0) === BOM_UTF16_LE[0] && content.charCodeAt(1) === BOM_UTF16_LE[1];
+
+        if (hasUtf8Bom) {
+            return { hasBom: true, encoding: 'utf8' };
+        } else if (hasUtf16LeBom) {
+            return { hasBom: true, encoding: 'utf16le' };
+        }
+
+        return { hasBom: false, encoding: null };
+    }
+
     visitSourceFile(node: ts.SourceFile): JS.CompilationUnit {
+
+        let bomAndTextEncoding = this.detectBOMAndTextEncoding(node.getFullText());
+
+        let prefix = this.prefix(node);
+        if (bomAndTextEncoding.hasBom) {
+            // If a node full text has a BOM marker, it becomes a part of the prefix, so we remove it
+            if (bomAndTextEncoding.encoding === 'utf8') {
+                prefix = prefix.withWhitespace(prefix.whitespace!.slice(1));
+            } else if (bomAndTextEncoding.encoding === 'utf16le') {
+                prefix = prefix.withWhitespace(prefix.whitespace!.slice(2));
+            }
+        }
+
         return new JS.CompilationUnit(
             randomId(),
-            this.prefix(node),
+            prefix,
             Markers.EMPTY,
             this.sourceFile.fileName,
             null,
-            null,
-            false,
+            bomAndTextEncoding.encoding,
+            bomAndTextEncoding.hasBom,
             null,
             [],
             this.semicolonPaddedStatementList(node.statements),