HNHacker News
TopNewBestAskShowJobs

janober

6,861 karma · joined February 1, 2016

[ my public key: https://keybase.io/janober; my proof: https://keybase.io/janober/sigs/abucrq_wLDw_fvAmyjmCymIOmY88QNhuacZtUFPApKM ]

Founder of https://n8n.io

submissionscomments
janober··on Show HN: Link.fish – API to extract data from websites as JSON
No is not just bad luck. Currently did not concentrate so much on "text pages" like blogs or articles yet. Mainly on pages which contain more data like prices, geo coordinates, social media profiles, ... That said support for the mentioned pages can simply be added via our point and click GUI by any user. Do sadly not have time right now, but can add support for this pages by tomorrow.
janober··on Show HN: Link.fish – API to extract data from websites as JSON
Hi just launched this API. So would love to get feedback like what could be improved or what is missing. Is a first version so any kind comments are welcome!
janober··on Show HN: Page.REST – An API to fetch details from a web page as JSON
I can maybe help you with that. If interested you can contact me via info@link.fish .
janober··on Show HN: Page.REST – An API to fetch details from a web page as JSON
If you are interested the same is possible with https://link.fish/hn-early-access/ which has a Zapier integration. It allows to send an URL and it then returns its data. However, you would have to send me an email to info@link.fish because the integration is still in beta mode and so only available after invitation.
janober··on Show HN: Link.fish – Boomarking meets web-parsing with archiving
Strange. Would be great if you can send me one or two examples that I can check that everything is working as expected. Thanks!
janober··on Show HN: Link.fish – Boomarking meets web-parsing with archiving
Ah yes. Sadly did not realize it in time to be able to change it.
janober··on Show HN: Link.fish – Boomarking meets web-parsing with archiving
Thanks, great to hear! Still working on the Pro Version. Planned features include for example the creation of HTML snapshots (to also still have the links available after a page gets deleted), priority rendering (when the queue is full pro users will get priority) and some additional views and functionality. But that will evolve over time. What I am also very excited about is the possibility that people will be able to create own plugins which extend the possibilities of link.fish. Already now all the views on link.fish are built as plugins. That plugins could add custom views for specific use-cases (like for example a spreadsheet view of the entries or mind map style for tags) or additional functionality (like voting on links).
janober··on Show HN: Link.fish – Boomarking meets web-parsing with archiving
Probably depends on what you are doing. If you only need simple bookmarking with some notes it is maybe really not worth the effort. However if you for example research something like an apartment, job, car, vacation. Or want to save movies, recipes, restaurants, ... and you want to sort/filter by any of the properties or you want to display pages with location information on a map, there is not much of an alternative right now. Sure, except if you count extracting the information manually, copy it into excel and setting pins by yourself on Google Maps. That takes literally hours without link.fish and a minute with.

Yes, the image situation is not great right now. Looking for a permanent solution is on my to-do-list.

janober··on Show HN: Link.fish – Boomarking meets web-parsing with archiving
What do you mean with ignoring? That is more or less what it should do already. If a page returns 404 or 403 it should display that the page could not be added but still add all other ones. What happens for you and what would you expect to happen? Thanks for your help!
janober··on Show HN: Link.fish – Boomarking meets web-parsing with archiving
Ah yes, that is possible. Simply export your existing Chrome bookmarks as HTML. How it can be done is described here: https://support.google.com/chrome/answer/96816

On link.fish you can then import them by clicking on the three orange dots right of the collection name. In the appearing menu choose the option "Upload Bookmark File". It will probably take a while depending on how many bookmarks you have because it has to visit each page. But after it is done you will receive an email.

janober··on Show HN: Link.fish – Boomarking meets web-parsing with archiving
Every collection can get exported as HTML bookmark file (the same Chrome uses) but that will only contain URLs, titles and tags. It is also possible to export all the URLs with all saved data as JSON or CSV. There is sadly no way yet to export the website snapshots (which are PNG). But that will also be possible in the future.
janober··on Show HN: Link.fish – Boomarking meets web-parsing with archiving
Thanks for your feedback!

link.fish is not about extracting a "part" of the website (in the sense of an image what I guess Safari does), it is really about the information (data). The idea is to make all web-information easily usable for everybody. So depending on what you bookmark different information should get extracted. So if you look for example for an apartment and bookmark one, it will get bedrooms, bathrooms, rent, image, location and so on. There are currently two main problems I want to solve. First bookmark managers do only display the title of the page and some kind of image and not what people actually want to see. And the second, that each website lives in its own world and the data from one page can not be used together with another one. Simple example. I go on vacation and want to display my room from Airbnb on a map with the sights I want to visit from Wikipedia or Lonely Planet and the restaurants from Yelp. Currently, that is totally impossible unless you have a lot of time and a CS degree.

The .fish domain is a thing which I get very different responses to. Some people really love it and some people not so much. One day I really have to look into that some more (when I have time and money). Right now the most important thing was that it is rememberable, short and cheap.

Depends on what part you mean is not new. Bookmarking for sure is not. ;-) The combination with the data extraction should be because did not find it anywhere else and that is why I created it. However totally agree on the rest! Will do that once I got some more feedback from users and I have money to pay somebody.

With the video agree. That was actually exactly the idea to seem needy and state something obvious. Should make a quite boring video a little bit funnier and different but did obviously not work that well ;-)

Yes is sadly always way more work than expected. So good luck on your project & launch. Thanks again for your help!

janober··on Show HN: Link.fish – Boomarking meets web-parsing with archiving
Sorry, have to improve the landing page if that is not clear without video. The main idea about link.fish is that it extracts the data of website and lets you work with it (sort and filter). So if you bookmark an apartment you get bedrooms, bathroom, rent, ... you can display them as list or map and that across different websites. For movies you get actors, director, ... for recipes calories, ingredients, ... and so on. Additionally, can you collaborate with other people in real-time and it creates snapshots so that even if websites get deleted you still have the page saved.

To extract the data it uses the same what search engines use (schema.org) and for pages which do not support it own rules can be defined. That rules are not hard coded they can be created by any user. If a page does not have any data there is the button "Add Domain Support". There you can then select what kind of data should be extracted (like for example a "Product") and what properties (like name, price, image, ...) and then select the data on the website. That has to be done for all not yet supported websites but then it works for all such pages on that website for all users. Not technical people can also simply press "Data Missing?" then we get a message and add support ourself.

janober··on Show HN: Link.fish – Boomarking meets web-parsing with archiving
Sorry for not answering the comments. Currently get always the message that I should slow down when I try to post. Will answer all as soon as I can.
janober··on Show HN: Link.fish – Boomarking meets web-parsing with archiving
Thanks for your feedback!

About the disappearing window in the chrome extension I can sadly not do anything as far as I know. Is build in chrome. However if you want something that does not disappear you can right click on the fish-icon and select options. But normally logging into the extension is not really needed. If you are logged into link.fish the chrome-extension should find the data by itself and log itself in when you click the first time on the fish-icon.

Yes sorry for copy and pasting there is currently a limit of 250 URLs at a time. If you want to add more at once you would have to do that by uploading an HTML-bookmark file (like chrome exports to).

The "No entries found" thing is something I have on my to-do-list. Is really kind of confusing to see that message while the entries load. Sorry.

janober··on Show HN: Link.fish – Boomarking meets web-parsing with archiving
A lot of bug fixes, improved usability, added archiving feature, ....
janober··on Show HN: Link.fish – Boomarking meets web-parsing with archiving
Thanks! Would love to get some feedback if it solves your problem and if not what needs improvement.
janober··on Show HN: Link.fish – Boomarking meets web-parsing with archiving
Hi, would love to get some feedback for the current beta version. Especially about usability, onboarding and the very new Data-Selector tool which allows tech savvy users to add support for currently not supported pages. The latter one gets currently just activated for HN beta users. Above all negative feedback would be very helpful as it will help me improve the most (as long as it is constructive). That could be generic things or reasons why you would currently not sign up. Thanks!
janober··on Show HN: Link.fish – Boomarking meets web-parsing with archiving
Thanks! Right now mainly working on getting the product totally production ready. No additional features for now. Once everything works perfectly want to add a pro account which offers additional functionality. For example, allows to also create HTML snapshots.
janober··on SoundCloud cuts jobs, closes SF and London offices
Very sorry to hear! Apparently additionally also the engineers in New York: https://twitter.com/pje_txt/status/882977097232338947
janober··on Elon Musk has published an outline of his ambitious plan to colonize Mars
Link is broke (404)
janober··on Show HN: Link.fish – Bookmarking Meets Web-Parsing (HN Exclusive)
Great to hear thanks! About the data. It depends on what you mean. If you mean that you can currently not change the data of a single bookmark that's true. However, if you mean that the data can not be corrected in general then not, but it is not very clear for incomplete data yet. There are two possibilities. None technical people can simply click on the (!) in the top right corner of an entry to inform us about wrong or incomplete data. We will then fix it asap. More tech savvy people can use our "Data Selector" tool. If there is no data at all it will display a large button for it directly on the entry. If the data is incomplete it can be reached via "Plugins -> Data Selector" on the top of the page. It allows to easily define what kind of data should be extracted and how it can be found. That tool is still in an early stage so would definitely love some feedback!
janober··on Show HN: Link.fish – Bookmarking Meets Web-Parsing (HN Exclusive)
Thanks a lot for the feedback. That is very helpful! Could you maybe tell me what parts you are specifically referring to? Because we actually try to collect as less information as possible. So do we for example not even ask for the name of the users. The only thing we ask for is an email address. The tracking (with for example Google Analytics) is used to know how many people are active and how the page is used to know what features to concentrate on to improve link.fish further (surveys are nice but are never as exact as actual usage data). We also do NOT display any ads nor is it every planned (hate them myself). What is maybe planned in the future (and can then also get deactivated for future paid accounts) is to send users special offers in which they could be interested. Like for example if somebody bookmarks a lot of apartments we can assume he wants to move and so he is maybe interested in getting 20% off for a moving service and/or DSL. Because we will not do any of that soon and currently I am only interested in honest user feedback you can simply sign up and write me an email to jan@link.fish . I will then make sure that as soon we have pro accounts you will get a free one. Thanks!
janober··on Show HN: Link.fish – Bookmarking Meets Web-Parsing (HN Exclusive)
Great, thanks a lot! Will check it out and see if it fixes the issue.
janober··on Show HN: Link.fish – Bookmarking Meets Web-Parsing (HN Exclusive)
By default does it use a combination of microdata, metadata and common HTML structures to find information. If nothing gets found that way, people can use a custom build in tool which makes selecting the information very easy. If the structure changes, it does not yet get detected automatically. Rather does the first person which sees that information is not there, would have to load the parser and fix the parts that changed. Normally does not take more than a few minutes.
janober··on Show HN: Link.fish – Bookmarking Meets Web-Parsing (HN Exclusive)
Probably depends on how you define success. Kimonolabs did exit and import.io is still running and how it seems not that bad.

Apart from that are they quite different to link.fish . What we have in common is that we all turning websites into structured data. That is however where it ends. They offer pure data extraction services for professionals and companies for big sets of data for a few predefined websites (B2B). It can be automated, download and then they can do with the raw data whatever they want. link.fish, on the other hand, is a bookmarking service which allows anybody (B2C) to work with every websites without technical knowledge (not totally there yet). They can display anything directly in link.fish like sort their recipes by preparation time, their movies by rating and the sights they want to visit on their vacation on a map.

janober··on Show HN: Link.fish – Bookmarking Meets Web-Parsing (HN Exclusive)
It has to do with IndexedDB which gets used to save some data.
janober··on Show HN: Link.fish – Bookmarking Meets Web-Parsing (HN Exclusive)
Thanks! Will do my best ;-)
janober··on Show HN: Link.fish – Bookmarking Meets Web-Parsing (HN Exclusive)
Thanks a lot! Right now it does not use the clients to scrape the data. But is maybe later planned to be also able to extract data from pages behind logins and/or when there are issues because of the scraping.
janober··on Show HN: Link.fish – Bookmarking Meets Web-Parsing (HN Exclusive)
Ah yes, hope the Safari users make the effort to test it in another browser for now. Will try to find a good workaround for the final version because it does not seem like Apple will ever fix it.
← PreviousPage 5 of 6Next →