fix: case insensitive headers

Author: erikbrinkman

src/headers.ts

@@ -57,65 +57,69 @@ export interface MhtmlHeaders extends Iterable<[string, string]> {
 }
 
 export class Headers implements MhtmlHeaders {
-  #raw = new Map<string, string[]>();
+  // keyed by lowercased name; the original name casing (first seen) is
+  // preserved for iteration while lookups are case-insensitive, per RFC 5322
+  #raw = new Map<string, { key: string; values: string[] }>();
 
   [Symbol.iterator](): Iterator<[string, string]> {
     return this.entries();
   }
 
   append(key: string, value: string): void {
-    const list = this.#raw.get(key);
-    if (list === undefined) {
-      this.#raw.set(key, [value]);
+    const entry = this.#raw.get(key.toLowerCase());
+    if (entry === undefined) {
+      this.#raw.set(key.toLowerCase(), { key, values: [value] });
     } else {
-      list.push(value);
+      entry.values.push(value);
     }
   }
 
   *entries(delim: string = ", "): IterableIterator<[string, string]> {
-    for (const [key, values] of this.#raw.entries()) {
+    for (const { key, values } of this.#raw.values()) {
       yield [key, values.join(delim)];
     }
   }
 
   *entriesAll(): IterableIterator<[string, string]> {
-    for (const [key, values] of this.#raw.entries()) {
+    for (const { key, values } of this.#raw.values()) {
       for (const val of values) {
         yield [key, val];
       }
     }
   }
 
   get(key: string, delim: string = ", "): string | null {
-    const vals = this.#raw.get(key);
-    if (vals === undefined) {
+    const entry = this.#raw.get(key.toLowerCase());
+    if (entry === undefined) {
       return null;
     } else {
-      return vals.join(delim);
+      return entry.values.join(delim);
     }
   }
 
   getAll(key: string): string[] {
-    return this.#raw.get(key) ?? [];
+    return this.#raw.get(key.toLowerCase())?.values ?? [];
   }
 
   has(key: string): boolean {
-    return this.#raw.has(key);
+    return this.#raw.has(key.toLowerCase());
   }
 
-  keys(): Iterable<string> {
-    return this.#raw.keys();
+  *keys(): IterableIterator<string> {
+    for (const { key } of this.#raw.values()) {
+      yield key;
+    }
   }
 
   *values(delim: string = ", "): IterableIterator<string> {
-    for (const vals of this.#raw.values()) {
-      yield vals.join(delim);
+    for (const { values } of this.#raw.values()) {
+      yield values.join(delim);
     }
   }
 
   *valuesAll(): IterableIterator<string> {
-    for (const vals of this.#raw.values()) {
-      yield* vals;
+    for (const { values } of this.#raw.values()) {
+      yield* values;
     }
   }
 }

src/index.spec.ts

@@ -155,6 +155,35 @@ Subject: =?iso-8859-1?Q?=A1Hola,\xffse=F1or!?=
     );
   });
 
+  test("matches header names and encodings case-insensitively", async () => {
+    const content = `mime-version: 1.0
+content-type: MULTIPART/mixed; BOUNDARY=frontier
+
+--frontier
+content-type: application/octet-stream
+content-transfer-encoding: Base64
+
+aGVsbG8gd29ybGQ=
+--frontier--
+`;
+    const files = [];
+    for await (const file of parseMhtml(stringToStream(content))) {
+      files.push(file);
+    }
+    const headers = files.map(({ headers }) => Object.fromEntries(headers));
+    expect(headers).toEqual([
+      {
+        "mime-version": "1.0",
+        "content-type": "MULTIPART/mixed; BOUNDARY=frontier",
+      },
+      {
+        "content-type": "application/octet-stream",
+        "content-transfer-encoding": "Base64",
+      },
+    ]);
+    expect(decoder.decode(files[1]!.content)).toStrictEqual("hello world");
+  });
+
   test("accepts headers without a space after the colon", async () => {
     const content = `MIME-Version:1.0
 Content-Type:multipart/mixed; boundary=frontier

src/index.ts

@@ -161,9 +161,12 @@ function getBoundary(headers: MhtmlHeaders): [Uint8Array, Uint8Array] {
   let bound: string | undefined;
   let multipart = false;
   for (const field of contentType.split(/;\s*/)) {
-    if (field.startsWith("multipart/")) {
+    // the media type and the boundary parameter name are case-insensitive, but
+    // the boundary value itself is not, so match on a lowercased copy
+    const lower = field.toLowerCase();
+    if (lower.startsWith("multipart/")) {
       multipart = true;
-    } else if (field.startsWith("boundary=")) {
+    } else if (lower.startsWith("boundary=")) {
       bound = field.slice(9);
       // TODO handling of quoted fields is not great
       if (bound.startsWith('"') && bound.endsWith('"')) {
@@ -216,7 +219,10 @@ export async function* parseMhtml(
     const headers = await parseHeaders(lines);
     const [boundary, terminus] = bound ?? (bound = getBoundary(headers));
 
-    const encoding = headers.get("Content-Transfer-Encoding") ?? "7bit";
+    // Content-Transfer-Encoding token values are case-insensitive (RFC 2045)
+    const encoding = (
+      headers.get("Content-Transfer-Encoding") ?? "7bit"
+    ).toLowerCase();
     const decode = decoders.get(encoding);
     if (decode === undefined) {
       throw new Error(`unhandled encoding type: ${encoding}`);