jason/ThunderAI
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

improved multipart html body decoding

Browse source
This commit is contained in:
Mic 2026-04-12 22:10:00 +02:00
parent b235005911
commit 8522f0b7f6
2 changed files with 31 additions and 17 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

46
js/mzta-utils.js
View file

@ -145,38 +145,52 @@ export async function getMailSubject(tab){
} }
function extractTextParts(fullMessage) { function extractTextParts(fullMessage) {
const textParts = []; const textParts = []
function walkParts(parts) { function walkParts(parts) {
for (const part of parts) { for (const part of parts) {
if (part.parts && part.parts.length > 0) { if (part.parts && part.parts.length > 0) {
// Recursively walk through sub-parts walkParts(part.parts)
walkParts(part.parts); }
} else { // console.log(">>>>>>>>>>>> extractTextParts: part.contentType: " + part.contentType + ", part.decryptionStatus: " + part.decryptionStatus + ", part.body: " + part.body);
// Check if contentType starts with "text/" if (part.contentType && part.contentType.startsWith('text/')) {
if (part.contentType && part.contentType.startsWith("text/")) { textParts.push(part)
textParts.push(part);
} }
} }
} }
}
if (fullMessage.parts && fullMessage.parts.length > 0) { if (fullMessage.parts && fullMessage.parts.length > 0) {
walkParts(fullMessage.parts); walkParts(fullMessage.parts)
} }
return textParts
return textParts;
} }
export function getMailBody(fullMessage){ function smartDecode(buf) {
try {
return new TextDecoder('utf-8', { fatal: true }).decode(buf);
} catch (e) {
return new TextDecoder('windows-1252').decode(buf);
}
}
export async function getMailBody(fullMessage, messageId) {
const textParts = extractTextParts(fullMessage); const textParts = extractTextParts(fullMessage);
let text = ""; let text = "";
let html = ""; let html = "";
// console.log(">>>>>>>>>>>>>> getMailBody: textParts: " + JSON.stringify(textParts));
// console.log(">>>>>>>>>>>>>> getMailBody: fullMessage: " + JSON.stringify(fullMessage));
for (const part of textParts) { for (const part of textParts) {
let body = part.body;
if ((body === undefined || body === "") && messageId && part.partName) {
const file = await browser.messages.getAttachmentFile(messageId, part.partName);
const buf = await file.arrayBuffer();
//const buf = new TextDecoder('utf-8').decode(buf);
body = smartDecode(buf);
}
if (part.contentType === "text/plain") { if (part.contentType === "text/plain") {
text += part.body; // console.log(">>>>>>>>>>>>>> getMailBody: part.body (TEXT): " + body);
text += body ?? "";
} else if (part.contentType === "text/html") { } else if (part.contentType === "text/html") {
html += part.body; // console.log(">>>>>>>>>>>>>> getMailBody: part.body (HTML): " + (body ? body.substring(0, 80) : body));
html += body ?? "";
} }
} }
if(html === "") { if(html === "") {

4
mzta-background.js
View file

@ -1113,7 +1113,7 @@ async function processEmails(args) {
if (addTagsAuto || spamFilter) { if (addTagsAuto || spamFilter) {
curr_fullMessage = await browser.messages.getFull(message.id); curr_fullMessage = await browser.messages.getFull(message.id);
msg_text = getMailBody(curr_fullMessage); msg_text = await getMailBody(curr_fullMessage);
taLog.log("Starting from the HTML body if present and converting to plain text..."); taLog.log("Starting from the HTML body if present and converting to plain text...");
body_text = htmlBodyToPlainText(msg_text.html); body_text = htmlBodyToPlainText(msg_text.html);
if( body_text.length == 0 ){ if( body_text.length == 0 ){
@ -1283,7 +1283,7 @@ async function processEmails(args) {
// extract body of current message as text // extract body of current message as text
const curr_message_full = await browser.messages.getFull(curr_message.id); const curr_message_full = await browser.messages.getFull(curr_message.id);
const curr_body_full_html = getMailBody(curr_message_full); const curr_body_full_html = await getMailBody(curr_message_full);
let curr_body_full_text = htmlBodyToPlainText(curr_body_full_html.html); let curr_body_full_text = htmlBodyToPlainText(curr_body_full_html.html);
if( curr_body_full_text.length === 0) { if( curr_body_full_text.length === 0) {
taLog.log("No HTML found in the message body, using plain text..."); taLog.log("No HTML found in the message body, using plain text...");