apunkt/rowboat
1
0
Fork 0
mirror of https://github.com/rowboatlabs/rowboat.git synced 2026-05-25 18:55:19 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions

fixed multiple recepients / attachments in email issue

Browse source
This commit is contained in:
Arjun 2026-05-25 20:04:52 +05:30
parent c0493e8b9f
commit 83389e93fc
3 changed files with 27 additions and 9 deletions
Download patch file Download diff file Expand all files Collapse all files

30
apps/x/packages/core/src/knowledge/sync_gmail.ts
View file

@ -31,9 +31,16 @@ const MAX_THREADS_IN_DIGEST = 10;
const RECENT_BACKFILL_INTERVAL_MS = 15 * 60 * 1000;
const nhm = new NodeHtmlMarkdown();
// Bump whenever snapshot-building logic changes in a way that should invalidate
// previously cached snapshots (e.g. attachment / recipient parsing fixes). The
// short-circuit in buildAndCacheSnapshot only reuses a cache whose version matches,
// so stale entries are transparently rebuilt on the next sync.
const SNAPSHOT_PARSER_VERSION = 2;
interface SnapshotCacheEntry {
historyId: string;
fetchedAt: string;
parserVersion?: number;
snapshot: GmailThreadSnapshot;
}
@ -56,6 +63,7 @@ function writeCachedSnapshot(threadId: string, historyId: string, snapshot: Gmai
const entry: SnapshotCacheEntry = {
historyId,
fetchedAt: new Date().toISOString(),
parserVersion: SNAPSHOT_PARSER_VERSION,
snapshot,
};
fs.writeFileSync(cachePath(threadId), JSON.stringify(entry), 'utf-8');
@ -308,19 +316,24 @@ interface ExtractedAttachment {
* saveAttachment / processThread, so the renderer can hand them to
* shell.openPath via the existing IPC.
*/
function extractAttachments(msgId: string, payload: gmail.Schema$MessagePart): ExtractedAttachment[] {
function extractAttachments(msgId: string, payload: gmail.Schema$MessagePart, html?: string): ExtractedAttachment[] {
const out: ExtractedAttachment[] = [];
const walk = (part: gmail.Schema$MessagePart): void => {
const filename = part.filename;
const attId = part.body?.attachmentId;
if (filename && attId) {
// Exclude only true inline images (image/* with a Content-ID, which
// get baked into bodyHtml as data URLs by inlineCidImages). Other
// parts with Content-ID — PDFs, .log files, .ics, etc. — are real
// attachments; Gmail just stamps Content-ID on most parts.
const cid = part.headers?.find(h => h.name?.toLowerCase() === 'content-id')?.value;
// Exclude only images that are genuinely inline — i.e. their Content-ID
// is actually referenced via `cid:` in the HTML body, so inlineCidImages
// already baked them in as data URLs. Gmail stamps a Content-ID on most
// parts (including real, separately-attached images like screenshots or
// scanned docs), so a Content-ID alone must NOT exclude an attachment;
// otherwise attached images silently disappear from the thread view.
const cidRaw = part.headers?.find(h => h.name?.toLowerCase() === 'content-id')?.value;
const cid = cidRaw?.replace(/^<|>$/g, '').trim();
const mime = part.mimeType || '';
const isInlineImage = !!cid && mime.startsWith('image/');
const referencedInHtml = !!cid && !!html
&& new RegExp(`cid:${cid.replace(/[.*+?^${}()|[\]\\]/g, '\\$&')}`, 'i').test(html);
const isInlineImage = mime.startsWith('image/') && referencedInHtml;
if (!isInlineImage) {
const safeName = `${msgId}_${cleanFilename(filename)}`;
out.push({
@ -577,6 +590,7 @@ async function buildAndCacheSnapshot(
threadData.historyId &&
cached &&
cached.historyId === threadData.historyId &&
cached.parserVersion === SNAPSHOT_PARSER_VERSION &&
cached.snapshot.importance
) {
return cached.snapshot;
@ -602,7 +616,7 @@ async function buildAndCacheSnapshot(
}
}
const isDraft = msg.labelIds?.includes('DRAFT') ?? false;
const attachments = msg.payload && msg.id ? extractAttachments(msg.id, msg.payload) : [];
const attachments = msg.payload && msg.id ? extractAttachments(msg.id, msg.payload, parts.html) : [];
return {
id: msg.id || undefined,
from: headerValue(headers, 'From') || 'Unknown',