⭐️⭐️⭐️ Open-source, self-hosted alternative to Linear and Plane. Project management and issue tracking where teams and AI agents work side by side to plan and ship products.
TypeScript agenticagileai-agentsbug-reportingbug-trackerganttissue-trackerjirajira-alternativekanbanlinearlinear-alternativeopen-sourceproduct-managementproject-managementroadmapself-hostedsprint-planningteam-collaborationtrello-alternative
import { useState } from 'react';
import type { FeedItem } from '@/lib/api/endpoints/activity';
import { useSession } from '@/lib/auth-client';
import { usePermissions } from '@/hooks/usePermissions';
import CommentItem from './CommentItem';
import CommentComposer, { type ComposerContext } from './CommentComposer';
// A comment and everything written under it: each reply a step further in than what
// it answers, and the reply box opens at the bottom.
// How deep a reply keeps stepping in. Past it the thread still nests, but on the same
// indent, so a long exchange does not squeeze the text off the screen.
const MAX_INDENT_DEPTH = 3;
// One row of the card: a comment and how far under the root it hangs.
interface Row {
item: FeedItem;
depth: number;
}
export default function CommentThread({
root,
repliesByParent,
imageByUserId,
composer,
}: {
root: FeedItem;
repliesByParent: Map<number, FeedItem[]>;
// Uploaded avatar per actor id (a feed entry stores the name, not the picture).
imageByUserId: Map<string, string | null>;
composer?: ComposerContext;
}) {
// The row the open box answers, null while it is closed.
const [replyTo, setReplyTo] = useState<Row | null>(null);
// Edit and delete join the reply button where the feed is interactive. The same
// rule the API asserts: the author with work_items edit, or a project owner.
const { data: session } = useSession();
const { can, isOwner } = usePermissions();
const canEditItems = composer != null && can('work_items', 'edit');
const manages = (item: FeedItem) =>
canEditItems && (isOwner || item.actorUserId === session?.user.id);
const rows: Row[] = [];
const collect = (item: FeedItem, depth: number) => {
rows.push({ item, depth });
for (const reply of repliesByParent.get(item.id) ?? []) collect(reply, depth + 1);
};
collect(root, 0);
return (
<li className="flex flex-col gap-2 py-1.5">
{rows.map((row) => (
<div key={row.item.id} className={indentClass(row.depth)}>
<CommentItem
item={row.item}
image={(row.item.actorUserId && imageByUserId.get(row.item.actorUserId)) ?? null}
onReply={composer ? () => setReplyTo(row) : undefined}
canEdit={manages(row.item)}
canDelete={manages(row.item)}
composer={composer}
/>
</div>
))}
{replyTo && composer && (
<div className={indentClass(replyTo.depth + 1)}>
<CommentComposer
{...composer}
replyToId={replyTo.item.id}
replyToName={replyTo.item.actorName}
onClose={() => setReplyTo(null)}
/>
</div>
)}
</li>
);
}
// Spelled out rather than computed: Tailwind generates only the classes it finds in
// the source.
function indentClass(depth: number): string {
const step = Math.min(depth, MAX_INDENT_DEPTH);
return ['', 'ps-8.5', 'ps-12.5', 'ps-16.5'][step];
}