Show HN: Till – Unblock and scale your web scrapers, with minimal code changes

by paramaw • 5 years ago

209 58 5 years ago

cube002• 5 years ago

Till helps you circumvent detected as a web scraper by identifying your scraper as a real web browser. It does this by generating random user-agent headers and randomizing proxy IPs (that you supply) on every HTTP request.

How did we start with consent using robots.txt and end up here?

A good neighbor doesn't use circumvention.

JulianWasTaken• 5 years ago

And only a few paragraphs down from a heading that says

Till was architected to follow best practices that DataHen has accumulated

I know there are plenty of sites out there out to prevent any scraping whatsoever, or to improperly prevent some situations most of us would agree is reasonable behavior, but this appears to be blatantly hostile to web admins out there.

cube002• 5 years ago

I agree there are cases where entities don't want things scraped that really should be scraped to preserve the history they'd like to rewrite.

My concern is that tools like these that use circumvention by default become the go-to when someone needs scraping and makes life hell for us running sites on hardware that's enough to service our small customer base but not an army of bots.

paco3346• 5 years ago

Admin of a 5000+ website platform (with customer inventory) here. This the exact kind of thing I've been working so hard to block lately.

Since Feb 2021 we've seen a substantial increase in scraping of our customers' inventory to the point we now have more bot traffic than human traffic. Annoyingly, only 5-10 requests come from a single IP so we've had to resort to always challenging requests that come from certain ASNs (typically those owned by datacenters).

This type of project frustrates me because it knowlingly goes against a site's desired bot action (via robots.txt).

realusername• 5 years ago

We ended up here since some people wrongly decided that only Google has a right to scrape.

PaulHoule6• 5 years ago

I've run sites that have a lot of pages where 80%+ of the traffic is web crawlers.

Google sends some traffic so I can afford to let them scrape. Bing crawls almost as much as Google but sends 5% as much traffic. Baidu crawls more than Google and never sends a single visitor.

I hate reinforcing the Google monopoly, but a crawler that doesn't send any traffic is expensive to serve.

1vuio0pswjnm7• 5 years ago

Then Google decided that no one has a right to scrape Google.

The web needs a more efficient system for distributing an index of its content. Having web developers design websites in a gazillion different permutations, when they are all bascially doing the same thing, and then having a handful of companies "scrape" them is neither an efficient nor sensibly-designed system.

The web (more generally, the internet) is a giant copy machine. Google, NSA and others have copies. Yet if we were to allow everyone to have copies by faciltating this through technical means (e.g., Wikipedia-style data dumps), many folks would panic. When Google first started indexing, it was not a business, and many folks did panic and there were many complaints. It's 2021; folks are still spooked by others being able to easily copy what they publish to the web. However it's OK to them if it's Google doing the copying. If there were healthy competition and many search engines to choose from, if one search engine did not have the lion's share of web traffic, it's doubtful Google would be given "exclusive" permission in robots.txt.

ok2938• 5 years ago

We ended up here, because a) the promise of easy to use structured data - that is inherent in data processing systems - has not been fulfilled: you have to scrape HTML to reconstruct some relational table, and b) information, despite being almost free to transmit in large amounts today, is treated as if it were a physical good, and each copy costs money. There is just lots of money in there, because our economic system looks more backwards than forwards.

Nextgrid2• 5 years ago

If a real user can access the content then that user should be able to delegate the work to a machine.

Doctor_Fegg2• 5 years ago

But what when there are separate interfaces specifically designed for real users and machines?

Case in point: OpenStreetMap (run by volunteer sysadmins) provides completely open, machine-readable data dumps. You can use these to set up your own geocoding, rendering or similar service. There is copious documentation, several post-processed dumps for particular purposes, etc. etc.

OSM also provides a friendly, human-scale, user-facing interface for map browsing and search. There are clearly documented limitations/TOUs for automated use of this interface.

Does that prevent people from setting up massive scrapers to scrape the human-facing interface, rather than using the machine-facing data dumps? No, it does not; and the volunteer sysadmins have to spend an inordinate amount of time protecting the service against these people.

DataHen's proudly admitted practices ("No need to worry about IP bans, we auto rotate IPs on any requests that are made."; "our massive pool of auto-rotating proxies, user agents, and 'secret-sauce' helps get around them") is directly antithetical to this sort of scenario. I find this incredibly irresponsible and unethical.

Kiro3• 5 years ago

Says who?

danpalmer2• 5 years ago

I work on a lot of web scraping and we have business agreements with every site that we scrape explicitly allowing us to do so and with pre-approval for the scraping rate (which we carefully control).

None of this gets around over-eager Cloudflare or Akamai rules set up years ago by some contractor that the businesses have no real ability to change.

Havoc• 5 years ago

Why scrape at all if there are agreements in place. Seems like a API task

anigbrowl• 5 years ago

Good neighbors don't track people's every move online either, but here we are.

awestroke• 5 years ago

Looks extremely useful. Starred for if I ever need to do scraping

mellosouls• 5 years ago

This sounds technically interesting but ethically dubious; also requires sign-up - even though free - which (I think?) is discouraged for Show HN.

If you are able to show how this service is administered responsibly with due regard for those running websites and blocking for good reason, best of luck with it.

capableweb4• 5 years ago

This sounds technically interesting but ethically dubious
If you are able to show how this service is administered responsibly

Imagine if this was the comments about HTTP when it was first launched (honestly, surely there was, but not as much as praise).

Protocols and tools are not ethically dubious or responsible for anything. The users who use those protocols and tools are. Anything that can be used for good/bad can also be used for bad/good.

OJFord• 5 years ago

also requires sign-up - even though free - which (I think?) is discouraged for Show HN.

I don't doubt it's 'community discouraged' because people don't want to have to, but I think the only rule is that there must be something to try out - no announcement-only, sign up for wait-list, hype-building type thing.

Kiro3• 5 years ago

also requires sign-up - even though free - which (I think?) is discouraged for Show HN.

Look at past Show HNs and tell me how many respect this. It's such an obsolete rule that it's irrelevant and almost OT to even comment on.

dannyw• 5 years ago

Can this bypass cloudflare?

new_guy• 5 years ago

These guys bypass cloudflare https://microlink.io it's open source but it's just more convenient to pay and forget about it.

jonatron• 5 years ago

Cloudflare have a few configuration options for the level of bot protection, so it's probably more a question of what percentage would be blocked for a particular site.

yourad_io2• 5 years ago

This is entirely correct.

It will also depend on the proxy's ASN, the rate of requests, the ability to solve captchas and the presence of an actual js environment in the browser (as opposed to, say, curl - Cloudflare uses a javascript environment check script to detect puppeteer/selenium etc.)

captainmuon• 5 years ago

This looks really interesting, I'm going to bookmark it!

Proxy IP address rotation

I wonder where you would get a decent proxy list nowadays. In the late 90s, you could find list of accidental open proxies that "hackers" collected. But nowadays I've only seen really shady "residential proxies" that are basically private PCs with malware. Is there a decent source for proxies that are not widely blocked and not criminal and not too expensive?

And by the way, while many people here question the morality of scraping, I have at least two legitimate use cases:

1) I built a little "vertical" search engine for a niche, using a combination of scraping and RSS. It doesn't consume much traffic and does really help people discover content in this niche.

2) A friend does research on urban development and rent prices and they scrape rent ads to get their data (or let students type it off manually, I'm not sure...)

showerst• 5 years ago

I run a number of proxies for (legally) scraping government data sites with overly restrictive views/hour policies.

They're just squid boxes spun up on various cheap VPS providers. A $5 digitalocean instance will power a lot of squid traffic, and as long as you stay away from AWS, the IP ranges tend not to be banned.

Just make sure to lock down your allowed source list, so you don't become one of those hacked open proxies!

kordlessagain• 5 years ago

If anyone needs a screenshot bot: https://github.com/kordless/grub-2.0

Still need to do some work on it, but it does do the job!

aftbit• 5 years ago

What's the difference between the linked repo and the hosted service at https://till.datahen.com/api/v1 ? Why do I need to sign up for a token if I am hosting my own server? Or otherwise, why do I need to host a server if I'm just calling your APIs? What sort of scraped data is exposed to the till hosted service?

paramaw2• 5 years ago

We don't have a hosted service at the moment. It's all run on premise. The API link you mentioned is for a Till instance to validate the token and for usage stats reporting for each user. The Auth token validates each user's free and premium features and limits based on usage stats more specifically total requests counts and cache hits counter. We don't record nor track anything related to the requests made on Till.

axiosgunnar• 5 years ago

Why is this on Github if I still need an API and cannot run it locally since the actual code is not public?

Or is this some sort of „source available“ content marketing nonsense?

JeremyBanks• 5 years ago

Till was architected to follow best practices that DataHen has accumulated over the years of scraping at a massive scale.

The best practice is not to build a business that's built on evading the security of the people you're depending on.

Just get into crime if you want to do that.

ChrisArchitect• 5 years ago

what kinds of scraping is being talked about here? As a maintainer of many a website I hate seeing so so much bot traffic these days. So many sketchy foreign things. And then throw in the spam/vulnerability testing bots/random URLs etc. sigh. Filling up the logs..Filling up the logs.

Almost totally dependent on Cloudflare to mitigate these days/as an in-between.

DaveExeter• 5 years ago

I have simple PHP code that adds misbehaving IP address to iptables. Keeps the logs clean, because they can't even connect over port 80.

k1rcher• 5 years ago

Does this have any sort of JavaScript emulation/support? Or is it purely HTTP requests?

The GID signature piece is especially interesting, I have ran into blockers in scaling scrapers in the past and this sort of organizational/tracking platform sounds awesome.

paramaw2• 5 years ago

You can connect your local browser to Till. But, you need to have your browser to accept a custom CA(Certificate Authority). Instructions here: https://till.datahen.com/docs/installation#ca

Show HN: Till – Unblock and scale your web scrapers, with minimal code changes

Related Stories