Filename extracted as URL

[Larrax]

From the following input (which is a legit archive filename):

PAYMENT EUR 1,420.00.zip

URL is extracted using find_urls:

1,420.00.zip

[voldmar]

.zip is the valid TLD, according to public suffix list. But the comma should not be the the part of domain I guess

[gleb-shnshn]

#47 fixed

[lipoja]

First of all thank you for your time working on the patch. However I am not quite sure about the fix, please have a look on my comment: #47 (comment)

And lets discuss this topic a bit maybe we can agree on something.
Thank you!

[andreys42]

as growth point : you can add some probability in extraction algorith, for example to decrease false positive rate you can use frequency of TLD usage as attribute (weight), so that probability of detected domain (for example Apple.Inc) would be lower than one that uses popular TLD (Apple.com/net/com and so on).
Finally, users can choose threshold of this prob value that fits the best their purposes ...