Newbie question: remote files via http?

[eriera1]

If I have several files in an external web server. From unix shell it is posible to have access to this files as if there were local files?

Something like that:
more http://mydomain/test.txt

and ths contents of the external text.txt file would be displayed...

Thanks,

[Torsten. · Accepted solution]

For scripting you need something like

http://hpux.connect.org.uk/hppd/hpux/Networking/WWW/curl-7.20.1/

or a perl or tcl script.

[Matti_Kurkela]

In your example command:
> more http://mydomain/test.txt

the shell will just relay the string "http://mydomain/test.txt" as a parameter to /usr/bin/more.

Because HP-UX /usr/bin/more does not support HTTP, you will get an error message:

http://mydomain/test.txt: No such file or directory

Your desired functionality cannot be easily achieved by just adding features to Unix shell. To get what you want, you would need one of the following:

A) support for HTTP protocol added to all commands you wish to use with an external web server

or

B) a "HTTP filesystem" which would intercept all file requests within its mountpoint and download the appropriate file from the web server.
(e.g.:
mount -t httpfs dummy_parameter /web
more /web/mydomain/test.txt
or something like that.)

Unfortunately, as far as I know, B) is not available for HP-UX and A) would require replacing standard system commands with customized versions.

In Linux, a sub-system called FUSE allows developers an easy way to create pseudo-filesystems like this, so someone has already implemented it:
http://httpfs.sourceforge.net/

If someone would first implement a FUSE-like interface for HP-UX kernel, then it might be relatively simple to port this httpfs to HP-UX too.

MK

[James R. Ferguson]

Hi:

And exactly what do you want to do with these files? You could obtain a textual representation with 'lynx' as for example:

# lynx -dump http://forums.itrc.hp.com/service/forums/questionanswer.do?threadId=1432950

This utility is available for HP-uX from the HP-UX Porting Centre if you are interested.

Regards!

...JRF...

[Steven Schweda]

> For scripting you need something like [...]

Or you could use wget.

http://www.gnu.org/software/wget/

> [...] have access [...]

Details depend on exactly what "have access"
means.

[eriera1]

Thanks everybody!

I'm using a software called Adobe Output Designer to generate pdf's. This software has a built-in instruction called ^graph that receive as parameter the full path to an image file and render the image in the generated pdf. Like ^graph \abc\myimage.jpg

I was wondering if it was posible to use http stored images like option B (FUSE). :) I mean ACO software would still think the file is a local file but instead a wrapper would get the file via http.

Anyway because before an external process would need to retrieve the list of urls where the images exists, he can do the work of downloading and copying into a local file too (using wget or curl or whatever).