DateTime parsing / support about opendirectorydownloader HOT 12 OPEN

Looks good 👍

I also like the idea of using the website meta data for choosing the "culture". Not sure if all servers provide info about the locale though...

from opendirectorydownloader.

Help is needed for a library which can automagically parse a lot of date formats.

from opendirectorydownloader.

This is still a very hard problem. Have tried to find a good datetime string parsing thing, but haven't found anything.

from opendirectorydownloader.

Could you share a few examples of dates you need to parse? :)

from opendirectorydownloader.

All of the formats in open directories.

I would've thought there would be at least one library which could do that. But couldn't really find one, tried multiple times.

Of course it can be written by hand, but preferring a library 😅

from opendirectorydownloader.

Maybe we could start a list of formats we encounter here in this thread? :)

So we can use the list to test different libraries or our custom code?

from opendirectorydownloader.

I'll start (you should be able to edit my comment to add new formats):

• 2019-12-08 01:19
• 2020-08-06 14:43:00
• 2020-08-6 14:43:00
• 26-Apr-2020 17:31
• 2019-Mar-17 03:43:25
• 2 months ago (with tooltip 2020-08-6 14:43:00)
• March 28 2013 at 11:21

Went through the last month and top posts on /r/OpenDirectories...

from opendirectorydownloader.

Wow. Well done. Looks like these are all parseble quite good. The problem starts with the American dates 😅

Looks like I prefer Japanese dates 👍👍👍 Works very good on file systems.

Maybe later I'll create an option like the current --exact-file-size thing. Which also takes the date from the response, and probably also the ETag, and both can go into #20 😇

from opendirectorydownloader.

That sounds good! I'll keep looking for obscure date formats, but I believe these cover >90% of all submissions...
About parsing, I don't know about C#/.NET, but running all of the above dates throw JavaScript's new Date() works for all except for 2 months ago and March 28 2013 at 11:21 (although simply removing the at did the trick). American dates (using / as separator) were recognized correctly as well.
So if a plain built-in parser can handle this, it shouldn't be a huge problem :)

It's called japanese logic? Never knew that! I know it's the superior data format for everything and I try to use it wherever I can, but good to know there are actually people using this in day-to-day life :D

from opendirectorydownloader.

https://docs.microsoft.com/en-us/dotnet/api/system.datetime.tryparse?view=net-5.0#System_DateTime_TryParse_System_String_System_DateTime__

Seems like .NET should be able to parse most of these formats.
We should use a static "culture" format tough, so the app behaves consistently...

from opendirectorydownloader.

You are right. The CultureInfo (language) could maybe even provided/read by the website headers which makes it even better.

See fiddle for example of the parsing.

https://dotnetfiddle.net/HmZvn9

from opendirectorydownloader.

https://code-maze.com/csharp-convert-string-to-datetime/

from opendirectorydownloader.