From fc51e33cebf0c01aae6b0ea3d9dd6becab9950d5 Mon Sep 17 00:00:00 2001
From: dtonon <git@dtonon.com>
Date: Mon, 1 Jun 2026 20:45:02 +0100
Subject: [PATCH] Link nostr entities and URLs in chat messages

Image URLs included
---
 src/lib/components/ChatContent.svelte |  73 +++-------
 src/lib/linkify.ts                    | 195 ++++++++++++++++++++++++++
 2 files changed, 214 insertions(+), 54 deletions(-)
 create mode 100644 src/lib/linkify.ts

diff --git a/src/lib/components/ChatContent.svelte b/src/lib/components/ChatContent.svelte
index 0d68766..5c8f5c0 100644
--- a/src/lib/components/ChatContent.svelte
+++ b/src/lib/components/ChatContent.svelte
@@ -1,6 +1,6 @@
 <script lang="ts">
-  import * as nip19 from "@nostr/tools/nip19";
   import { loadNostrUser, type NostrUser } from "@nostr/gadgets/metadata";
+  import { tokenizeChat } from "$lib/linkify";
 
   type Props = {
     content: string;
@@ -9,53 +9,7 @@
 
   let { content, profiles = {} }: Props = $props();
 
-  type Token =
-    | { type: "text"; value: string }
-    | { type: "mention"; pubkey: string; entity: string; fallback: string };
-
-  const MENTION_RE = /nostr:(npub1[a-z0-9]+|nprofile1[a-z0-9]+)/gi;
-
-  function shortEntity(entity: string): string {
-    const m = entity.match(/^(npub|nprofile)1/);
-    if (!m) return entity;
-    const prefix = m[0];
-    const rest = entity.slice(prefix.length);
-    if (rest.length <= 12) return entity;
-    return `${prefix}${rest.slice(0, 6)}…${rest.slice(-4)}`;
-  }
-
-  const tokens = $derived.by<Token[]>(() => {
-    const out: Token[] = [];
-    let last = 0;
-    for (const m of content.matchAll(MENTION_RE)) {
-      const start = m.index ?? 0;
-      const entity = m[1].toLowerCase();
-      let pubkey: string | null = null;
-      try {
-        const decoded = nip19.decode(entity);
-        if (decoded.type === "npub") pubkey = decoded.data;
-        else if (decoded.type === "nprofile") pubkey = decoded.data.pubkey;
-      } catch {
-        // Invalid bech32, fall through to text
-      }
-      if (start > last)
-        out.push({ type: "text", value: content.slice(last, start) });
-      if (pubkey) {
-        out.push({
-          type: "mention",
-          pubkey,
-          entity,
-          fallback: shortEntity(entity),
-        });
-      } else {
-        out.push({ type: "text", value: m[0] });
-      }
-      last = start + m[0].length;
-    }
-    if (last < content.length)
-      out.push({ type: "text", value: content.slice(last) });
-    return out;
-  });
+  const tokens = $derived(tokenizeChat(content));
 
   let resolvedUsers = $state<Record<string, NostrUser>>({});
 
@@ -72,10 +26,21 @@
 
 <span class="break-words whitespace-pre-wrap"
   >{#each tokens as t (t)}{#if t.type === "mention"}{@const u =
-        profiles[t.pubkey] ?? resolvedUsers[t.pubkey]}<a
-        href="https://njump.me/{t.entity}"
-        target="_blank"
-        rel="noopener noreferrer"
-        class="text-brand hover:underline">@{u?.shortName ?? t.fallback}</a
-      >{:else}{t.value}{/if}{/each}</span
+      profiles[t.pubkey] ?? resolvedUsers[t.pubkey]}<a
+      href="https://njump.me/{t.entity}"
+      target="_blank"
+      rel="noopener noreferrer"
+      class="text-brand hover:underline">@{u?.shortName ?? t.fallback}</a
+    >{:else if t.type === "entity"}<a
+      href={t.href}
+      target="_blank"
+      rel="noopener noreferrer"
+      class="text-brand hover:underline">{t.label}</a
+    >{:else if t.type === "link"}<a
+      href={t.href}
+      title={t.href}
+      target="_blank"
+      rel="noopener noreferrer"
+      class="text-brand hover:underline">{t.label}</a
+    >{:else}{t.value}{/if}{/each}</span
 >
diff --git a/src/lib/linkify.ts b/src/lib/linkify.ts
new file mode 100644
index 0000000..01b1c5c
--- /dev/null
+++ b/src/lib/linkify.ts
@@ -0,0 +1,195 @@
+import * as nip19 from "@nostr/tools/nip19";
+
+// Curated TLD list: gTLDs, popular new gTLDs, common ccTLDs
+const TLDS = [
+  "com",
+  "org",
+  "net",
+  "edu",
+  "gov",
+  "mil",
+  "int",
+  "info",
+  "biz",
+  "name",
+  "pro",
+  "io",
+  "co",
+  "app",
+  "dev",
+  "ai",
+  "sh",
+  "me",
+  "ly",
+  "tv",
+  "fm",
+  "lol",
+  "club",
+  "online",
+  "site",
+  "store",
+  "blog",
+  "tech",
+  "xyz",
+  "art",
+  "design",
+  "news",
+  "media",
+  "page",
+  "link",
+  "fun",
+  "gg",
+  "gl",
+  "st",
+  "to",
+  "run",
+  "life",
+  "world",
+  "space",
+  "cloud",
+  "email",
+  "social",
+  "chat",
+  "wtf",
+  "cafe",
+  "zone",
+  "studio",
+  "us",
+  "uk",
+  "de",
+  "fr",
+  "it",
+  "es",
+  "nl",
+  "ru",
+  "jp",
+  "cn",
+  "ca",
+  "au",
+  "br",
+  "in",
+  "mx",
+  "se",
+  "no",
+  "fi",
+  "dk",
+  "ch",
+  "at",
+  "be",
+  "pl",
+  "pt",
+  "gr",
+  "cz",
+  "ie",
+  "nz",
+  "kr",
+  "sg",
+  "hk",
+  "tw",
+  "za",
+  "cc",
+  "ws",
+  "eu",
+  "tr",
+  "ua",
+  "il",
+  "ar",
+  "cl",
+  "pe",
+].join("|");
+
+// Matches nostr entities, http(s) URLs, and bare domains
+const URL_SOURCE = `nostr:(?:npub1|nprofile1|note1|nevent1|naddr1)[a-z0-9]+|https?:\\/\\/[^\\s<>"']+|(?<![\\w@.\\/])(?:[a-z0-9](?:[a-z0-9-]*[a-z0-9])?\\.)+(?:${TLDS})\\b(?:[\\/?#][^\\s<>"']*)?`;
+// Trailing punctuation that is usually not part of the URL
+const TRAILING_PUNCT_RE = /[).,;:!?'"]+$/;
+
+export type ChatToken =
+  | { type: "text"; value: string }
+  | { type: "link"; href: string; label: string }
+  | { type: "mention"; pubkey: string; entity: string; fallback: string }
+  | { type: "entity"; entity: string; href: string; label: string };
+
+// Shorten a nostr entity to xxxxxxxx...xxxx
+export function shortNostrEntity(entity: string): string {
+  if (entity.length <= 12) return entity;
+  return `${entity.slice(0, 8)}...${entity.slice(-4)}`;
+}
+
+// Shorten a URL to domain.tld/...last-10-chars-of-the-final-part
+export function shortUrl(href: string): string {
+  const rest = href.replace(/^https?:\/\//i, "").replace(/\/+$/, "");
+  const slash = rest.indexOf("/");
+  if (slash === -1) return rest;
+  const host = rest.slice(0, slash);
+  const path = rest.slice(slash + 1);
+  if (path === "") return host;
+  const finalPart = path.slice(path.lastIndexOf("/") + 1);
+  const tail = finalPart.length > 10 ? finalPart.slice(-10) : finalPart;
+  return `${host}/...${tail}`;
+}
+
+function decodeEntity(entity: string): ChatToken | null {
+  try {
+    const decoded = nip19.decode(entity);
+    if (decoded.type === "npub")
+      return {
+        type: "mention",
+        pubkey: decoded.data,
+        entity,
+        fallback: shortNostrEntity(entity),
+      };
+    if (decoded.type === "nprofile")
+      return {
+        type: "mention",
+        pubkey: decoded.data.pubkey,
+        entity,
+        fallback: shortNostrEntity(entity),
+      };
+    if (decoded.type === "note" || decoded.type === "nevent" || decoded.type === "naddr")
+      return {
+        type: "entity",
+        entity,
+        href: `https://njump.me/${entity}`,
+        label: shortNostrEntity(entity),
+      };
+  } catch {
+    // Invalid bech32, fall through to text
+  }
+  return null;
+}
+
+// Tokenize a plain-text chat message into text, links, mentions and entities.
+// Image URLs are treated as plain links (never embedded).
+export function tokenizeChat(content: string): ChatToken[] {
+  const out: ChatToken[] = [];
+  let last = 0;
+  const re = new RegExp(URL_SOURCE, "gi");
+  for (const m of content.matchAll(re)) {
+    const start = m.index ?? 0;
+    let raw = m[0];
+    const isNostr = /^nostr:/i.test(raw);
+    let trailing = "";
+    if (!isNostr) {
+      const trail = raw.match(TRAILING_PUNCT_RE);
+      if (trail) {
+        trailing = trail[0];
+        raw = raw.slice(0, -trailing.length);
+      }
+    }
+    if (start > last)
+      out.push({ type: "text", value: content.slice(last, start) });
+    if (isNostr) {
+      const entity = raw.slice(6).toLowerCase();
+      const token = decodeEntity(entity);
+      out.push(token ?? { type: "text", value: m[0] });
+    } else {
+      const href = /^https?:\/\//i.test(raw) ? raw : `https://${raw}`;
+      out.push({ type: "link", href, label: shortUrl(href) });
+      if (trailing) out.push({ type: "text", value: trailing });
+    }
+    last = start + m[0].length;
+  }
+  if (last < content.length)
+    out.push({ type: "text", value: content.slice(last) });
+  return out;
+}