socialmag/platforms/threads.py

"""Threads posting and notifications flow (2026-09 inline composer).

    post             — post via inline composer
    check_login      — verify signed in
    get_notifications — scrape the Activity page
"""

import asyncio

from ..cdp import human_delay, CDPError
from . import guard, require_text, validate_media
from ..selectors import THREADS as S

PLATFORM = "threads"


async def post(cdp_client, text, media_path=None, options=None):
    options = options or {}
    require_text(PLATFORM, text)
    validate_media(media_path)

    tab, reused = await cdp_client.get_or_open_tab(S["url"], "threads.com")
    if reused:
        await tab.eval("window.onbeforeunload = null")
        await tab.navigate(S["url"])
    return await guard(PLATFORM, tab, _flow(tab, text, media_path))


async def _flow(tab, text, media_path):
    # Wait for the inline composer to appear
    await _wait_for_composer(tab)
    await human_delay()

    # Click the composer input area
    await tab.click_js(S["composer_trigger_js"])
    await human_delay(0.5, 1.0)

    # Type into whichever input is now focused — could be a contenteditable
    # or a textbox that appeared after clicking
    textbox = await _find_active_input(tab)
    await tab.type_text(textbox, text)

    if media_path:
        if not await tab.exists(S["file_input"]):
            # Look for an attach media button near the composer
            await human_delay()
        if await tab.exists(S["file_input"]):
            await tab.set_file_input(S["file_input"], media_path)
            await human_delay(1.5, 3.0)

    await human_delay(0.5, 1.0)
    await tab.click_js(S["post_button_js"])

    # Success = the composer returns to its empty placeholder state
    await _wait_for_post_complete(tab)


async def _wait_for_composer(tab, timeout=20.0):
    deadline = asyncio.get_running_loop().time() + timeout
    while asyncio.get_running_loop().time() < deadline:
        has_input = await tab.eval(
            "!!(document.querySelector('[placeholder=\"What\\'s new?\"]')"
            " || [...document.querySelectorAll('span,div')]"
            ".find(e => e.innerText && e.innerText.trim() === 'What\\'s new?'))"
        )
        if has_input:
            return
        await asyncio.sleep(0.5)
    from ..cdp import CDPError
    raise CDPError("threads: composer ('What's new?') not found after 20s")


async def _find_active_input(tab):
    """After clicking the composer, find the actual text input element."""
    # Check for contenteditable first (Threads often uses these)
    for sel in ('[contenteditable="true"]', '[role="textbox"]', 'textarea',
                '[placeholder="What\'s new?"]'):
        if await tab.exists(sel):
            return sel
    from ..cdp import CDPError
    raise CDPError("threads: no text input found after clicking composer")


async def _wait_for_post_complete(tab, timeout=30.0):
    """Wait for the post to go through — composer empties or a toast appears."""
    deadline = asyncio.get_running_loop().time() + timeout
    while asyncio.get_running_loop().time() < deadline:
        # Check if the composer is back to empty placeholder state
        empty = await tab.eval(
            "(() => {"
            " const ce = document.querySelector('[contenteditable=\"true\"]');"
            " if (ce && ce.innerText.trim() === '') return true;"
            " const tb = document.querySelector('[role=\"textbox\"]');"
            " if (tb && tb.innerText.trim() === '') return true;"
            " return false;"
            "})()"
        )
        if empty:
            return
        await asyncio.sleep(0.5)
    from ..cdp import CDPError
    raise CDPError("threads: composer still has text 30s after clicking Post")


async def check_login(cdp_client):
    tab, reused = await cdp_client.get_or_open_tab(S["url"], "threads.com", activate=False)
    if reused:
        await tab.eval("window.onbeforeunload = null")
        await tab.navigate(S["url"])
    if await tab.exists(S["login_marker"]):
        return False
    # JS state, not the "What's new?" composer: that composer never mounts in a
    # backgrounded tab, so a signed-in account read as signed out (measured
    # 2026-09-12 — DOM signal False while the session cookie was right there).
    # Threads runs on the Instagram session, hence ds_user_id.
    state = await tab.eval(
        "(() => { const m = document.cookie.match(/(?:^|;\\s*)ds_user_id=(\\d+)/);"
        " return m ? 'in:' + m[1] : (document.cookie ? 'out' : 'unavailable'); })()"
    )
    if state.startswith("in:"):
        return True
    if state == "out":
        return False
    raise CDPError("threads: no cookies readable on threads.com — cannot tell signed-in from signed-out")


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 []