From 1f27778ae88947027ec613bb3e4f1dd3aff6351f Mon Sep 17 00:00:00 2001
From: NickOvt <nikolai@zone.ee>
Date: Mon, 8 Apr 2024 09:47:27 +0300
Subject: [PATCH] fix(api-messages-attachment-download): is sendAsString param
 is set, and is set to true then decode the original file and send back as
 UTF-8 ZMS-134 (#655)

* add attachment charset in the indexer

* test adding a field

* test another way

* walk through mimetree to find the attachment charset and decode the message using this charset. Decoding is piped to the output stream

* test

* revert last change

* use regex for check, fix piping

* fixes
---
 lib/api/messages.js | 37 +++++++++++++++++++++++++++++++++++--
 1 file changed, 35 insertions(+), 2 deletions(-)

diff --git a/lib/api/messages.js b/lib/api/messages.js
index 449e8da7..ad0e2105 100644
--- a/lib/api/messages.js
+++ b/lib/api/messages.js
@@ -23,6 +23,7 @@ const TaskHandler = require('../task-handler');
 const { prepareSearchFilter, uidRangeStringToQuery } = require('../prepare-search-filter');
 const { getMongoDBQuery /*, getElasticSearchQuery*/ } = require('../search-query');
 //const { getClient } = require('../elasticsearch');
+let iconv = require('iconv-lite');
 
 const BimiHandler = require('../bimi-handler');
 const {
@@ -1528,7 +1529,11 @@ module.exports = (db, server, messageHandler, userHandler, storageHandler, setti
             summary: 'Download Attachment',
             description: 'This method returns attachment file contents in binary form',
             validationObjs: {
-                queryParams: {},
+                queryParams: {
+                    sendAsString: booleanSchema
+                        .default(false)
+                        .description('If true then sends the original attachment back in string format with correct encoding.')
+                },
                 pathParams: {
                     user: userId,
                     mailbox: mailboxId,
@@ -1600,7 +1605,7 @@ module.exports = (db, server, messageHandler, userHandler, storageHandler, setti
                             _id: true,
                             user: true,
                             attachments: true,
-                            'mimeTree.attachmentMap': true
+                            mimeTree: true
                         }
                     }
                 );
@@ -1637,6 +1642,8 @@ module.exports = (db, server, messageHandler, userHandler, storageHandler, setti
                 });
             }
 
+            let [attachmentCharset] = getAttachmentCharset(messageData.mimeTree, attachment);
+
             res.writeHead(200, {
                 'Content-Type': attachmentData.contentType || 'application/octet-stream'
             });
@@ -1678,6 +1685,10 @@ module.exports = (db, server, messageHandler, userHandler, storageHandler, setti
             } else if (attachmentData.transferEncoding === 'quoted-printable') {
                 attachmentStream.pipe(new libqp.Decoder()).pipe(res);
             } else {
+                if (!/ascii|utf[-_]?8/i.test(attachmentCharset) && result.value.sendAsString) {
+                    attachmentStream.pipe(iconv.decodeStream(attachmentCharset)).pipe(res);
+                    return;
+                }
                 attachmentStream.pipe(res);
             }
         })
@@ -3901,3 +3912,25 @@ function parseAddresses(data) {
     walk([].concat(data || []));
     return Array.from(addresses);
 }
+
+function getAttachmentCharset(mimeTree, attachmentId) {
+    if (mimeTree.attachmentId && mimeTree.attachmentId === attachmentId) {
+        // current mimeTree (sub)object has the attachmentId field, and it is the one we search
+        // get the parsedHeader -> content-type -> params -> charset
+
+        return [mimeTree.parsedHeader['content-type']?.params?.charset || 'UTF-8', true];
+    } else if (mimeTree.childNodes) {
+        // current mimetree (sub)object does not have the attachmentId field and it is not equal to the one we search
+        // loop childNodes
+        let charset;
+        for (const childNode of Object.values(mimeTree.childNodes)) {
+            charset = getAttachmentCharset(childNode, attachmentId);
+            if (charset[1] === true) {
+                // actually found the charset, early return
+                return charset;
+            }
+        }
+    }
+
+    return ['UTF-8', false];
+}