Every failed task printed the same reason, because the blanket catch in
complete_all_tasks treated NoSuchElementException and TimeoutException as one
finding. A wait that expires says nothing about whether the section was there,
since wait_for_element swallowed whatever the getter raised and returned False.
A market that does not ship visual search and a points panel that was on screen
and slow both arrived as TimeoutException
wait_for_element now keeps the last reason the getter gave and raises
ElementNeverAppeared when nothing was in the DOM for the whole wait.
_container_by_id raises ElementNotReady when the id is present but no visible
copy has content yet, which is the hydrating case it used to report as missing.
Both subclass what they refine, so claim_bonus_points and
complete_bing_daily_set keep working untouched
A task that was reached and then ran out of time is now a [FAIL] rather than a
[SKIP], because it may have left points behind. Absence stays a [SKIP]
56 tests pass, 20 of them new. A mutation check backs them, forcing every
expired wait to count as absence fails 2, and collapsing the two branches back
into one [SKIP] fails 4
Refs #52
Matching on bing.com/search alone was too narrow. Turn referrals into rewards is a real daily set activity that awards points and it points at a rewards url, so the filter skipped it and the run came up short without saying why.
Three shapes now count, bing.com/search, bing.com/rewards and rewards.bing.com. The bing app promo from #45 sits on bingapp.microsoft.com and stays out of all three, which is what keeps that fix intact. Five tests, one per shape plus a mixed panel.
The positional fallback returned whatever sat at index 3 of the streaks section. On a partially rendered page that is not the daily set: observed live returning 'Mobile App | Check-in: 0/1', and clicking it opens the app store page instead of the panel, which is what #45 and #46 describe. It now checks the label before handing the button back and raises otherwise, so the task is skipped rather than the wrong streak clicked.
read_search_points waited for the close button before reading anything, so a panel that rendered its content but not its button failed the whole search task while the number was already on screen. Traced to that wait with a stacktrace. Closing is best effort now.
The panel hydrates progressively, so waiting for the section only tells you it opened, not that it filled. A check running on the first non-empty state reports whatever happened to be rendered at that moment, which is why the daily set came out differently run to run. Same wait as complete_bing_daily_set: hold out for the full set, report what is there if it never arrives.
Four bugs that reached real runs are covered: a point value the parser could not read, the wrong row of the breakdown panel, a container that looks right but is empty, and a label that matches two different buttons.
Stdlib unittest with small fakes for the selenium calls the selectors make, so this adds no dependency and needs no browser. Run with python -m unittest discover -s tests.
get_daily_set_elements returned everything after the first link in the panel. The panel also carries promotional links, so position hands one back as an activity. Clicking it leaves rewards.bing.com and every element captured before that goes stale, which is the exception reported in #45.
Activities always point at a Bing search, so match on that. When nothing matches it returns nothing rather than falling back to position, since clicking a promo is worse than skipping the task and complete_bing_daily_set already reports the shortfall.
complete_bing_daily_set indexed [0] [1] [2] on whatever wait_for_element returned first. The panel hydrates progressively, so that can be a single activity, and the task died with IndexError before touching the other two. Wait for the full set, fall back to what is there, and re-read per index since a click can re-render the panel.
read_search_points ran into the default 10s timeout because it starts after the earlier tasks navigated away, so the earn page re-renders from scratch first. That skipped the whole search task while points were still available.
The card loop fired fixed 100px scroll events back to back with no pauses, which is the jumpy scrolling, and its while-not-in-viewport loop was unbounded, so a card that never fits the viewport completely would hang the run forever. The way back up unwound a counted number of steps, which lands wrong when the page height changes while cards update.
Scrolling is now wheel input with varying step sizes and short pauses, bounded, aimed at centering the target. The return reads the actual scroll position instead of counting.
searches_needed was computed once as (max - earned) // 5 and never re-checked. Two assumptions fail in practice: some markets award 3 points per search rather than 5, and the daily maximum itself is not stable, observed as 15, 30 and 60 on one account within a day with the counter resetting. The run therefore stopped around 18/30 and still reported success.
Search in rounds instead: measure, run a batch sized on the lower known rate, measure again, stop when the quota is full or a round gains nothing, and warn instead of claiming success when it is not filled.
Also give the ollama client a timeout and bound the empty-response retry, since both were unbounded and an unattended run hung for 14 minutes with 2.3 CPU-seconds. The bare while-not-response loop spins forever on empty responses.
scrollIntoView re-centered every element unconditionally and instantly, which is the page jump between the daily set and explore on bing. The pointer move now scrolls only when the element is actually outside the viewport, smoothly, and polls the rect until the async scroll settles before computing the path.
Visual search matches the full 'visual search streak' label. Plain 'visual search' can also hit a dashboard element that goes stale mid-interaction, and the guessed 'image search' needle is gone.
check_selectors.py walks every selector and prints what resolved, what is absent and what broke, along with browser, driver, page language and the earn section ids. Absent is a normal result for a task a variant does not ship. It completes no activities and claims nothing, so it is safe to run for a bug report.
complete_explore_on_bing_tasks now raises when the section is missing instead of returning quietly, which made complete_all_tasks print [OK] for a task that never ran.
The visible labels the lookups match on are collected in one Labels class. The selectors are market independent but still language dependent, and this makes that explicit and fixable in one place.
The pointer is moved to viewport coordinates, so an element below the fold produced a target outside the window and the driver raised MoveTargetOutOfBoundsException. Observed on the daily set opener during a scheduled run. Also clamps the bezier path to the viewport.
The daily set opener lives in the streaks section, not in a section of its own, so scoping the lookup to a dedicated section disabled the task on every layout. Match on 'daily set streak' instead, since 'daily set' also hits the level up entry.
Visual search keeps a positional fallback inside streaks because its label is not confirmed on layouts that ship it. Claim now prefers an exact label match so the 'Ready to claim' heading cannot win.
Absolute XPaths break outside en-US, where an extra exploreonbing section shifts every positional section index, so every task failed before it started.
Select by visible text, id suffix and visibility instead of position, take the visible copy of ids that are emitted twice for responsive layout, and raise NoSuchElementException for tasks a variant does not ship so the run skips them instead of aborting.
mardausdennis