v0.2: twitter module, feed/notifications/reply/open_link tools, dashboard

platforms/threads.py

@@ -1,15 +1,13 @@
-"""Threads posting flow (2026-09 inline composer).
+"""Threads posting and notifications flow (2026-09 inline composer).
 
-    1. navigate threads.com
-    2. click the inline "What's new?" composer
-    3. type text
-    4. click Post button
-    5. text clearing / post appearing = success
+    post             — post via inline composer
+    check_login      — verify signed in
+    get_notifications — scrape the Activity page
 """
 
 import asyncio
 
-from ..cdp import human_delay
+from ..cdp import human_delay, CDPError
 from . import guard, require_text, validate_media
 from ..selectors import THREADS as S
 
@@ -117,3 +115,68 @@ async def check_login(cdp_client):
         " || [...document.querySelectorAll('span,div')]"
         ".find(e => e.innerText && e.innerText.trim() === 'What\\'s new?'))"
     )
+
+
+async def get_notifications(cdp_client, count=10):
+    """Scrape the Threads Activity page."""
+    tab, reused = await cdp_client.get_or_open_tab(
+        "https://www.threads.com/activity", "threads.com"
+    )
+    if reused:
+        await tab.eval("window.onbeforeunload = null")
+        await tab.navigate("https://www.threads.com/activity")
+
+    await asyncio.sleep(4)
+
+    count = min(count, 20)
+    js = r"""
+        (() => {
+            const region = document.querySelector('[role="region"]');
+            if (!region) return [];
+            const ORIGIN = 'https://www.threads.com';
+            const abs = (h) => !h ? null
+                : (/^https?:/.test(h) ? h : ORIGIN + (h.startsWith('/') ? h : '/' + h));
+
+            // Row detection, from the live DOM (probed 2026-09-11): a
+            // notification row is an element that links somewhere, carries a
+            // relative-time marker ("3h", "6h", "2d"), and holds no smaller
+            // element that does the same. The length floor rejects the bare
+            // "6h" time chip, which is itself a link on some rows.
+            const isRow = (el) => {
+                const t = el.innerText ? el.innerText.trim() : '';
+                return t.length >= 12
+                    && /\b\d+[smhdw]\b/.test(t)
+                    && !!el.querySelector('a[href]');
+            };
+            const cands = [...region.querySelectorAll('*')].filter(isRow);
+            const rows = cands.filter(el => !cands.some(o => o !== el && el.contains(o)));
+
+            const results = [];
+            for (const row of rows) {
+                const text = row.innerText.trim().replace(/\n+/g, ' ');
+                if (text.length < 5) continue;
+                const hrefs = [...row.querySelectorAll('a[href]')]
+                    .map(x => x.getAttribute('href'));
+                // Prefer the post permalink; otherwise whatever the row links to
+                // (settings notices link to /settings/... — a real destination).
+                const post = hrefs.find(h => /\/@[^\/]+\/post\//.test(h || ''));
+                results.push({ text: text, url: abs(post || hrefs[0]) });
+            }
+
+            // Fallback: no anchors at all (layout change, or a notification kind
+            // that doesn't link anywhere). Keep the old text-split rather than
+            // returning an empty list that would read as "no activity".
+            if (!results.length) {
+                const items = region.innerText.trim()
+                    .split(/(?<=\b\d+[hdm]\b)\s*\n\s*/);
+                for (const p of items) {
+                    const t = p.trim().replace(/\n+/g, ' ');
+                    if (t.length < 5) continue;
+                    results.push({ text: t, url: null });
+                }
+            }
+            return results.slice(0, COUNT);
+        })()
+    """.replace("COUNT", str(count))
+
+    return await tab.eval(js) or []