Match Facebook's before-snapshot scroll depth to the verification loop

Codex-reported: the before-submit article snapshot only covered the
initially-loaded viewport, but the post-submit verification loop
scrolls (VERIFY_SCROLL_ATTEMPTS x VERIFY_SCROLL_PIXELS) to reveal more
posts. An older post sitting below the fold at snapshot time -- absent
from that snapshot -- would get revealed by the verification loop's
own scrolling and wrongly look "new", the same class of false positive
the snapshot was added to prevent.

Extracted _snapshot_existing_articles(), which scrolls through the
exact same depth as the verification loop and accumulates every
article fingerprint seen at each step (not just the final scroll
position, since Facebook's feed can remove off-screen articles from
the DOM as new ones load in). Both the before-snapshot and the
verification loop now share VERIFY_SCROLL_ATTEMPTS/VERIFY_SCROLL_PIXELS
constants so they can't drift apart again the way this bug happened in
the first place.

No live TikTok/Facebook API calls, no account access. 149 tests pass,
compileall clean, git diff --check clean.

Author: tradewithmeai

CHANGELOG.md

@@ -35,7 +35,7 @@ that will firm up once it leaves beta. Dates are when a release was tagged, not
     safe-by-default test suite; added a dedicated TikTok OAuth check group to `doctor.py`
     (`python doctor.py --tiktok`). Still safe by default either way -- `--confirm-publish` is
     required for a real publish on both, exactly as for every other platform.
-  - **Nine further issues found by code review before any live retest, none of them guessed:**
+  - **Ten further issues found by code review before any live retest, none of them guessed:**
     (1) `auth/publish_facebook.py` never forced its browser context to English like every other
     publisher does -- fixed with the same `FORCE_ENGLISH_LOCALE` constant. (2) The documented
     `--client-secrets path/to/tiktok_client.json` setup form silently broke every token refresh
@@ -77,7 +77,14 @@ that will firm up once it leaves beta. Dates are when a release was tagged, not
     round) narrowed false positives from the profile bio/nav text, but could still match an
     *older* post with similar leading text (a repeated caption, or a common generic opening).
     Fixed by snapshotting the profile timeline's existing posts before submitting anything, then
-    only counting a match in a post that's genuinely new since that snapshot.
+    only counting a match in a post that's genuinely new since that snapshot. (10) That
+    before-snapshot only covered the initially-loaded viewport, but the post-submit verification
+    loop scrolls to reveal more posts -- so an OLDER post sitting below the fold at snapshot time
+    would get revealed by the verification loop's own scrolling and wrongly look "new" too. Fixed
+    by having the before-snapshot scroll through the exact same depth
+    (`VERIFY_SCROLL_ATTEMPTS` x `VERIFY_SCROLL_PIXELS`) the verification loop does, accumulating
+    every article seen at each step (Facebook's feed can remove off-screen articles from the DOM
+    as new ones load in, so a single final-position snapshot isn't enough either).
 - **YouTube's upload category is no longer hardcoded.** `auth/publish_youtube.py` gained
   `--category-id` (default unchanged: `22`, People & Blogs) -- closes a previously-documented
   defect in the upload path.

auth/publish_facebook.py

@@ -52,6 +52,15 @@
 VIDEO_PROCESSING_WAIT_ITERATIONS = 150
 VIDEO_PROCESSING_WAIT_INTERVAL_MS = 2000
 
+# How far the post-submit verification step scrolls, and in how many steps, looking for the new
+# post. The pre-submit snapshot (_snapshot_existing_articles) scrolls through exactly the same
+# depth -- Codex-reported: an earlier version only snapshotted the initially-loaded viewport, so
+# an OLDER post sitting below the fold at snapshot time (absent from that snapshot) but revealed
+# by the verification loop's own scrolling would wrongly look "new". Both loops must cover the
+# same ground for the before/after comparison to mean anything.
+VERIFY_SCROLL_ATTEMPTS = 3
+VERIFY_SCROLL_PIXELS = 4000
+
 
 def _save_debug_screenshot(page) -> Path | None:
     """Best-effort screenshot saved BEFORE the calling code raises -- the browser context is
@@ -111,6 +120,21 @@ def _article_fingerprints(page) -> list[str]:
     )
 
 
+def _snapshot_existing_articles(page) -> set[str]:
+    """Fingerprint every article currently reachable by scrolling through
+    VERIFY_SCROLL_ATTEMPTS steps of VERIFY_SCROLL_PIXELS each -- the exact same depth the
+    post-submit verification loop scrolls through. Accumulates the union seen at every step
+    (not just the final one) since Facebook's feed can virtualize/remove off-screen articles from
+    the DOM as new ones load in, so an article visible at step 1 might not still be present once
+    the page has scrolled to step 3."""
+    fingerprints: set[str] = set(_article_fingerprints(page))
+    for _ in range(VERIFY_SCROLL_ATTEMPTS):
+        page.mouse.wheel(0, VERIFY_SCROLL_PIXELS)
+        page.wait_for_timeout(1500)
+        fingerprints.update(_article_fingerprints(page))
+    return fingerprints
+
+
 def _open_composer(page) -> bool:
     """Click the "What's on your mind?" opener on the home feed. The visible text includes the
     logged-in user's own first name (e.g. "What's on your mind, Alex?"), so this matches on the
@@ -236,7 +260,7 @@ def publish_facebook(
             if profile_url:
                 page.goto(profile_url, timeout=STEP_TIMEOUT_MS)
                 page.wait_for_timeout(3000)
-                existing_article_fingerprints = set(_article_fingerprints(page))
+                existing_article_fingerprints = _snapshot_existing_articles(page)
                 page.goto("https://www.facebook.com/", timeout=STEP_TIMEOUT_MS)
                 page.wait_for_timeout(2000)
 
@@ -338,27 +362,32 @@ def publish_facebook(
                 snippet = text[:40]
                 page.goto(profile_url, timeout=STEP_TIMEOUT_MS)
                 page.wait_for_timeout(4000)
-                for attempt in range(3):
-                    # Codex-reported, in two parts. First: a bare
+                for attempt in range(VERIFY_SCROLL_ATTEMPTS):
+                    # Codex-reported, in three parts so far. First: a bare
                     # document.body.innerText.includes() check would also match the snippet in
                     # the profile bio or nav text -- fixed by scoping to `[role="article"]`
                     # (Facebook's own ARIA role for a feed post). Second: scoping to articles
                     # alone still doesn't prove the match is the post just made, not an older one
                     # starting with similar text (a repeated caption, or a common generic
                     # opening) -- fixed by comparing against the before-snapshot taken earlier
-                    # and only counting a match in an article that's NEW since then. Still
-                    # unverified against a live account like everything else in this file -- this
-                    # narrows the false-positive surface further, it doesn't eliminate every way
-                    # this could be wrong (e.g. two genuinely new posts with identical leading
-                    # text in the same run).
+                    # and only counting a match in an article that's NEW since then. Third: the
+                    # before-snapshot only covered the initially-loaded viewport, so an older post
+                    # sitting below the fold at snapshot time -- but revealed by THIS loop's own
+                    # scrolling -- would wrongly look "new" too. Fixed by having
+                    # _snapshot_existing_articles scroll through the exact same
+                    # VERIFY_SCROLL_ATTEMPTS x VERIFY_SCROLL_PIXELS depth up front, so both passes
+                    # cover identical ground. Still unverified against a live account like
+                    # everything else in this file -- this narrows the false-positive surface
+                    # further, it doesn't eliminate every way this could be wrong (e.g. two
+                    # genuinely new posts with identical leading text in the same run).
                     current_fingerprints = _article_fingerprints(page)
                     new_fingerprints = [
                         fp for fp in current_fingerprints if fp not in existing_article_fingerprints
                     ]
                     if any(snippet in fp for fp in new_fingerprints):
                         verified = True
                         break
-                    page.mouse.wheel(0, 4000)
+                    page.mouse.wheel(0, VERIFY_SCROLL_PIXELS)
                     page.wait_for_timeout(3000)
 
             result = {