parseMail: fix support for UTF-8 strings in input, accept Uint8Arra…

…y for other encodings The input data was treated as a 'binary string' internally, meaning that if the data was a native string that included non-ASCII chars, these might fail to be decoded. This commit changes the `parseMail` interface so that: - if a string input is passed, it is treated as UTF-8 (not a 'binary string') - Uint8Array inputs are now accepted, and should be used for other encodings
ProtonMail · Jul 3, 2023 · a904720 · a904720
1 parent 2d9638e
commit a904720
Show file tree

Hide file tree

Showing 5 changed files with 46 additions and 15 deletions.
diff --git a/index.d.ts b/index.d.ts
@@ -30,4 +30,4 @@ export interface ParsedMessage {
   'reply-to'?: Address
 }
 
-export function parseMail(message: string): ParsedMessage;
+export function parseMail(message: string | Uint8Array): ParsedMessage;
diff --git a/index.js b/index.js
@@ -1 +1 @@
-export { parseMail } from './lib/mailParser';
+export { parseMail } from './lib/mailParser';
diff --git a/lib/mailParser.js b/lib/mailParser.js
@@ -318,7 +318,7 @@ export const MimeParser = {
    * The input is a string that is immediately parsed, calling all functions on
    * the emitter before this function returns.
    *
-   * @param input   A string or input stream of text to parse.
+   * @param {BinaryString} input   A string or input stream of text to parse.
    * @param emitter The emitter to receive callbacks on.
    * @param opts    A set of options for the parser.
    */
@@ -401,7 +401,7 @@ export const MimeParser = {
    *
    * The input is any type of input that would be accepted by parseSync.
    *
-   * @param input   A string of text to parse.
+   * @param {BinaryString} input   A string of text to parse.
    */
   extractMimeMsg(input, options = {}) {
     var emitter = Object.create(ExtractMimeMsgEmitter);
@@ -548,11 +548,13 @@ export const MimeParser = {
 
 /**
  * Parse MIME message
- * @param {String} data - MIME message to parse
+ * @param {String|Uint8Array} data - MIME message to parse
  * @returns {Object} parsed content (see TS definitions for more details)
  */
 export function parseMail(data) {
-  const { headers, allAttachments, bodyParts } = MimeParser.extractMimeMsg(data);
+  const encoded = (typeof data === 'string') ? new TextEncoder().encode(data) : data;
+
+  const { headers, allAttachments, bodyParts } = MimeParser.extractMimeMsg(uint8ArrayToString(encoded));
   // these fields can only contain a single value
   const singleKeys = new Set([
     'message-id',

diff --git a/test/test_mail_parser.ts b/test/test_mail_parser.ts
@@ -1,13 +1,13 @@
 import { expect } from "chai";
-import { read_file } from "./utils";
+import { read_file_raw } from "./utils";
 import { parseMail } from "../";
 import { uint8ArrayToString } from "../lib/utils";
 
 const toBase64 = uInt8Array => btoa(uint8ArrayToString(uInt8Array));
 
 describe('mail parser', () => {
   it('correctly parses multipart message with both HTML and plain text data', async () => {
-    const eml = await read_file("multipart-complex1");
+    const eml = await read_file_raw("multipart-complex1");
     const { body, attachments } = parseMail(eml);
 
     expect(body.html).to.equal('<html><head>This part should be returned.</head></html>\n');
@@ -22,11 +22,29 @@ describe('mail parser', () => {
     expect(attachments[1].fileName).to.equal('');
   });
 
+  it('correctly parses UTF-8 string input', async () => {
+    const eml = `Content-Type: multipart/mixed;
+    boundary="------------cJMvmFk1NneB7MT4jwYHY7ap"
+
+This is a multi-part message in MIME format.
+--------------cJMvmFk1NneB7MT4jwYHY7ap
+Content-Type: text/plain; charset=UTF-8;
+Content-Transfer-Encoding: 8bit
+
+Import HTML cöntäct//Subjεέςτ//
+
+--------------cJMvmFk1NneB7MT4jwYHY7ap--`;
+    const { body } = parseMail(eml);
+
+    expect(body.text).to.equal('Import HTML cöntäct//Subjεέςτ//\n');
+  });
+
   it('correctly parses SHIFT-JIS body with png attachment', async () => {
     const expectedText = 'Portable Network Graphics（ポータブル・ネットワーク・グラフィックス、PNG）はコンピュータでビットマップ画像を扱うファイルフォーマットである。圧縮アルゴリズムとしてDeflateを採用している、圧縮による画質の劣化のない可逆圧縮の画像ファイルフォーマットである。\n';
     const expectedAttachmentContent = 'iVBORw0KGgoAAAANSUhEUgAAAIAAAABECAIAAADGJao+AAAAwklEQVR4Xu3UgQbDMBRA0bc03f//b7N0VuqJEmwoc+KqNEkDh9b+2HuJu1KNO4f+AQCAAAAQAAACAEAAAAgAAAEAIAAABACAAAAQAAACAEAAAAgAAAEAIAAAANReamRLlPWYfNH0klxcPs+cP3NxWF+vi3lb7pa2R+vx6tHOtuN1O+a5lY3HzgM5ya/GM5N7ZjfPq7/5yS8IgAAAEAAAAgBAAAAIAAABACAAAAQAgAAAEAAAAgBAAAAIAAABACAAAIw322gDIPvtlmUAAAAASUVORK5CYII=';
 
-    const eml = await read_file("shift-jis-image");
+    const eml = await read_file_raw("shift-jis-image");
+    console.log(eml)
     const { body, subject, headers, attachments: [attachment] } = parseMail(eml);
 
     expect(body.text).to.equal(expectedText);
@@ -39,7 +57,7 @@ describe('mail parser', () => {
   });
 
   it('correctly reads binary attachments', async () => {
-    const eml = await read_file("multipart-binary");
+    const eml = await read_file_raw("multipart-binary");
     const { attachments: [attachment] } = parseMail(eml);
 
     expect(attachment.content).to.deep.equal(new Uint8Array([1, 2, 3]));
@@ -48,7 +66,7 @@ describe('mail parser', () => {
   });
 
   it('includes the content-id and filename for each attachment', async () => {
-    const eml = await read_file("multipart-content-id");
+    const eml = await read_file_raw("multipart-content-id");
     const { attachments: [attachment1, attachment2] } = parseMail(eml);
 
     expect(attachment1.content).to.deep.equal(attachment2.content);
@@ -60,7 +78,7 @@ describe('mail parser', () => {
   });
 
   it('returns an empty array for empty attachment body', async () => {
-    const eml = await read_file("multipart-empty-attachment");
+    const eml = await read_file_raw("multipart-empty-attachment");
     const { attachments: [attachment] } = parseMail(eml);
 
     expect(attachment.content).to.be.instanceOf(Uint8Array);
@@ -70,15 +88,15 @@ describe('mail parser', () => {
   });
 
   it('decodes the subject', async () => {
-    const eml = await read_file("multipart-encrypted-subject-utf8");
+    const eml = await read_file_raw("multipart-encrypted-subject-utf8");
     const { subject, body } = parseMail(eml);
 
     expect(subject).to.equal('subject with emojis 😃😇');
     expect(body.text).to.equal('test utf8 in encrypted subject\n');
   });
 
   it('parses addresses and date', async () => {
-    const eml = await read_file("multipart-addresses");
+    const eml = await read_file_raw("multipart-addresses");
     const { from, to, cc, bcc, date } = parseMail(eml);
 
     expect(from).to.deep.equal({ name: 'Some One', email: 'someone@test.com' });
@@ -89,7 +107,7 @@ describe('mail parser', () => {
   });
 
   it('parses address groups', async () => {
-    const eml = await read_file("multipart-addresses-groups");
+    const eml = await read_file_raw("multipart-addresses-groups");
     const { from, to, cc, bcc } = parseMail(eml);
 
     expect(from).to.deep.equal({ name: 'Some One', email: 'someone@test.com' });

diff --git a/test/utils.js b/test/utils.js
@@ -118,6 +118,17 @@ export function read_file(file, start = undefined, end = undefined) {
   });
 }
 
+export function read_file_raw(file) {
+  return new Promise(function(resolve, reject) {
+      fetch('base/test/data/' + file)
+        .then(response => response.ok ? response.arrayBuffer() : reject(new Error('error fetching file')))
+        .then(buffer => {
+          resolve(new Uint8Array(buffer))
+        })
+        .catch(err => reject(err))
+    });
+}
+
 export function isWebKit() {
   // simply testing the 'webkit' appears in the userAgent is not enough, as that is the case for
   // any browser running on macOS.