Reputation: 8169

wget or curl from stdin

I'd like to download a web pages while supplying URLs from stdin. Essentially one process continuously produces URLs to stdout/file and I want to pipe them to wget or curl. (Think about it as simple web crawler if you want).

This seems to work fine:

tail 1.log | wget -i - -O - -q

But when I use 'tail -f' and it doesn't work anymore (buffering or wget is waiting for EOF?):

tail -f 1.log | wget -i - -O - -q

Could anybody provide a solution using wget, curl or any other standard Unix tool? Ideally I don't won't want to restart wget in the loop, just keep it running downloading URLs as they come.

Upvotes: 21

Answers (4)

bo0k

Reputation: 153

The effective way is to avoid using xargs, if downloading files from the same web server:

wget -q -N -i - << EOF
http://sitename/dir1/file1
http://sitename/dir2/file2
http://sitename/dir3/file3
EOF

Upvotes: 0

Silas S. Brown

Reputation: 1662

Try piping the tail -f through python -c $'import pycurl;c=pycurl.Curl()\nwhile True: c.setopt(pycurl.URL,raw_input().strip()),c.perform()'

This gets curl (well, you probably meant the command-line curl and I'm calling it as a library from a Python one-liner, but it's still curl) to fetch each URL immediately, while still taking advantage of keeping the socket to the server open if you're requesting multiple URLs from the same server in sequence. It's not completely robust though: if one of your URLs is duff, the whole command will fail (you might want to make it a proper Python script and add try / except to handle this), and there's also the small detail that it will throw EOFError on EOF (but I'm assuming that's not important if you're using tail -f).

Upvotes: -1