Short Drama Trends — Cross-Platform Rankings & Daily Movers

One leaderboard for the two biggest vertical short-drama apps. This Actor collects the DramaBox and ReelShort catalogs, normalizes them into a single schema, ranks them together, and — from the second run onwards — tells you what moved: views gained, followers gained, episodes added, and which titles are brand new.

It reads each site's server-rendered JSON directly. No browser, no API key, no login.

What you get

One dataset item per title, identical shape on both platforms:

Field	Notes
rank, platformRank, sortedBy	Position in the combined leaderboard and within its own platform
platform, id, title, url	platform + id is the stable join key across runs
views, follows, episodes	The three counters. follows is DramaBox followers / ReelShort collects
viewsDelta, followsDelta, episodesDelta	Change since your previous run — the actual product
isNew, comparedTo	isNew flags a title absent from the last run; comparedTo timestamps the baseline
genres, tags, rating	Each platform's own taxonomy, plus tropes/themes
cover, description, releaseDate, lastUpdatedAt
durationSec, imdbUrl, cast	ReelShort only — DramaBox no longer publishes cast names
snapshotDate, scrapedAt	Group by snapshotDate to build a time series

A run summary (totals per platform, new titles, top ten movers) is saved to the key-value store as SUMMARY.

Getting movement data

Deltas need a baseline, so the first run reports isNew: true and null deltas for everything — that run is the baseline. Every run afterwards compares against it.

Point an Apify Schedule at this Actor (daily is the natural cadence) and each day's dataset is both a snapshot and a diff. The counters live in a named key-value store (short-drama-trends-state by default); use a different name per ranking if you track several genres separately, so they don't overwrite each other.

Input

Input	Default	What it does
Platforms	Both	Rank DramaBox, ReelShort, or one merged leaderboard
Max titles per platform	50	Collected per platform before filtering and ranking
Rank by	Views	Views, followers, episodes, biggest view gain, or newest release
Genre / theme filter	—	e.g. Romance, Werewolves, Second Chance, Time Travel
Minimum views	—	Drop titles below a view threshold
Exact view counts & extra details	On	Visit each title's page (see below)
Track day-over-day changes	On	Remember counters so the next run can diff
Proxy	—	Optional; see Rate limits

Filters run after collection, so a narrow genre or a high minViews can return fewer titles than maxItemsPerPlatform. Raise the cap to widen the net.

Exact view counts

With this on (the default), each title's own page is fetched, which is the only place DramaBox publishes a view count. The figure on its listing rows looks like one and isn't: the same title reads 19.4K in the listing and 6,041,483 views on its own page.

• On — DramaBox view counts are exact; ReelShort adds runtime, IMDb link, cast and release date.
• Off — faster and cheaper. ReelShort counts stay exact (its listings publish real integers, views and collects alike). DramaBox reports views: null rather than a number that would understate a hit by three orders of magnitude — the site's own string is kept verbatim in viewsDisplay — while its followers and episode counts stay accurate, so rank by followers on these runs.

Rate limits

Both sites are happy with a normal run and need no proxy. Under sustained hammering ReelShort starts answering HTTP 202 with an empty body; the Actor treats that as a block, backs off exponentially and retries on a fresh session.

If you do enable a proxy, note that DramaBox rejects Apify's datacenter IPs with a 403 while serving a direct connection normally. Rather than hand you an empty DramaBox half, the Actor crawls DramaBox directly whenever the proxy is datacenter-only, and says so in the log. Pick the residential group if DramaBox traffic must go through a proxy. ReelShort works through either.

Pricing

Pay per result: $0.004 per title stored, and nothing else. A title is only stored when it carries an identity and at least one real counter, so pages that fail, empty listings and filtered-out titles are never charged. A search that matches nothing costs $0.00.

Notes & limits

• Ranking is within what you collected. Both catalogs run to thousands of titles; maxItemsPerPlatform decides how deep the sample goes. A catalog run seeds from each site's merchandised front page, so a small run is biased toward what the platforms are currently pushing — raise the cap for a truer top-N.
• View counts are platform-reported and not comparable in an accounting sense across the two apps; they are comparable to themselves over time, which is what the deltas use.
• A genre is resolved against each platform's own taxonomy where one exists — plurals are handled, so Werewolves finds ReelShort's Werewolf tag. Where it can't be resolved (a typo, or one of the few ReelShort tags whose page 404s), the Actor falls back to a catalog crawl matched against each listing's own themes. That fallback sees only listing-level themes, so it returns noticeably fewer titles — the log says when it happens.