Rakuma (Fril) Item Search
Pricing
from $0.70 / 1,000 item scrapeds
Rakuma (Fril) Item Search
Search Rakuma - Rakuten's flea market, Mercari's sibling - and get one flat row per item: item id, name, price in JPY, sold flag, brand, thumbnail and the item URL.
Pricing
from $0.70 / 1,000 item scrapeds
Rating
0.0
(0)
Developer
Superslow Sloth
Maintained by CommunityActor stats
0
Bookmarked
2
Total users
1
Monthly active users
4 days ago
Last modified
Categories
Share
Search Rakuma - Rakuten's flea market, Mercari's sibling - and get one flat row per item: no nested objects, no HTML to strip yourself. Built for a reseller's new-listing watch or a price check: give it one or more keywords, get back every matching item's price, sold status, brand and photo.
Input
| field | meaning |
|---|---|
keywords | One or more searches, exactly as typed into Rakuma's own search box. Every keyword shares the maxResults budget below, in the order given |
sort | relevance (default), newest, most_liked, price_asc, price_desc - Rakuma's own four sort options plus its default |
maxResults | Stop after this many items in total. Rakuma serves 40 per page, so anything up to 40 per keyword is a single page request |
proxyConfiguration | Residential by default - see Proxy below |
price band and on-sale-only are not input fields, on purpose
Measured 2026-09-28, live against https://fril.jp/s: none of price_min,
price_max, min_price/max_price, price_from/price_to,
price_gte/price_lte changed the result set at all, and no value of
status (on_sale, selling, sold_out, 1 through 5, ...) produced a
clean on-sale-only or sold-out-only split - some landed on Rakuma's plain
homepage instead of search results, others changed the sold/on-sale mix in a
way with no consistent pattern. The checkbox and price-band control visible on
the page are a client-side React component with no URL round trip in the
page's own markup, so there is nothing this actor's input can wire up. Every
row already carries the sold flag read straight off the page - filter on
that column after the run instead.
Output
| field | meaning |
|---|---|
item_id | The 32-character id in the item's own permalink (item.fril.jp/<id>) |
name | Listing title |
url | Canonical item.fril.jp/<id> link |
price_jpy | Price in yen, always JPY - Rakuma quotes nothing else |
sold | true when the card carries Rakuma's "SOLD OUT" ribbon |
brand_id, brand_name | Rakuma's brand tag, when the listing carries one |
size | Always null - never printed on the search page for any category tried, clothing included |
like_count | Always null - the heart button carries no number for a logged-out session |
shipping_included | Always null - no shipping badge or caption is printed on the search page at all |
thumbnail_url | The real listing photo (Rakuma's lazy-load placeholder sprite is filtered out) |
query | Which keyword this row came from |
Field names follow Mercari Japan's actor wherever the two sites share a
concept (query, item_id, name, url, price_jpy, sold, brand_id,
brand_name, thumbnail_url), so a Rakuma run and a Mercari run union into
one price-watch table.
null is not zero
size, like_count and shipping_included are always null because Rakuma's
search results page never prints them - not because this actor failed to read
them. A 0 in any of those columns would read as a measurement ("no likes",
"free shipping") when the truth is that the page said nothing at all.
Paging and sort
Rakuma pages with a single page number and orders results with two
parameters together: sort names the field, order names the direction,
because Rakuma reuses the field name sell_price for both "cheapest first"
and "priciest first" and only order tells them apart. newest maps to
sort=created_at&order=desc - the option a new-listing watch wants.
Proxy
Residential is the default. Measured from a running Actor on 2026-09-28:
Apify's datacenter proxy got HTTP 403 from fril.jp on every exit tried (4 of
4, rotating between each), while RESIDENTIAL - unpinned, and pinned to
apifyProxyCountry: "JP" - returned a full page of results first time. You can
switch to datacenter to cut proxy cost, but expect the run to end with
refusals in the log and no rows.
Billing
Pay per event, all-in: $0.0007 per item, plus $0.002 actor-start per run. Platform usage - compute, bandwidth, proxy - is included; nothing is billed to you beyond the two event prices above, and nothing is charged twice for the same item id even when two keywords both turn it up.
For comparison, on the Apify Store: jungle_synthesizer's Rakuma/Fril listings
scraper charges $0.001 per item plus a $0.10 actor-start fee; youfuxu's
charges $0.003 per item. This actor's $0.0007 per item is exactly 30% under the cheaper of the two
($0.001), and its own actor-start fee is $0.002 per run rather than the $0.10
jungle_synthesizer charges on top of its per-item price.