Can't run plugin anytotx......

neil.munro · Post by **neil.munro** » Thu Nov 22, 2001 7:52 am

Hi, I think that I am doing something wrong somewhere along the line but I can't for the life of me figure what is going on.
I am trying to index pdf files (among other things but I will just talk about pdf for the moment).
Everytime and every combination so far has resulted in:
018 Can't run plugin anytotx
error message.
So I will start with my specifications and what I am doing:
Webinator version: Commercial Webinator 2.5
(though when I run texis -version, I get :
Texis Web Script (Vortex) Copyright (c) 1996-2000 Thunderstone - EPI, Inc.
Free Webinator Version 3.0.955383945 of Apr 10, 2000 (sparc-sun-solaris2.5.1)

Operating System: SunSolaris 8

I have put the anytotx file into a directory within the PATH. I have put the .so files in a directory in the LD_LIBRARY_PATH.
It runs well from the command line eg anytotx blah blah
It has the same owner and group as texis/gw/webserver.
But it does not function when I include it in a gw command.
I have tried the following:
gw -dmydatabase -napplication/pdf,pdf,anytotx http://mywebsite.com
and
gw -dmydatabase -napplication/pdf,pdf,/the/full/path/to/anytotx http://mywebsite.com
as well as trying it with the -fpdf option.
Everytime with the same error message....
018 Can't run plugin /usr/local/bin/anytotx
Do you have any idea of what is causing this? Security? Can't find it? Other configuration? or just plain doesn't like me?

Any help will be grateful...

regards,

Neil.

bart · Post by **bart** » Thu Nov 22, 2001 7:54 am

Check to see that there's execute permissions on anytotx. Also try running it by itself. Its a stdin/stdout program.

neil.munro · Post by **neil.munro** » Thu Nov 22, 2001 8:00 am

Sorry I didn't include that info before....
I checked the permission on anytotx and they looked OK. Just in ase, I set them to chmod 777.
Still no better...
I CAN run it from the command line eg anytotx my.pdf
works as it should..
but still not from with the gw command line.....
...still baffled...
Neil.

Post by **John** » Thu Nov 22, 2001 9:48 am

If gw is setuid, which it would be with a default install, the OS may not respect LD_LIBRARY_PATH for security reasons. Either copy the .so files to /usr/lib or other standard location, or run gw not setuid, but as the appropriate user.

neil.munro · Post by **neil.munro** » Fri Nov 23, 2001 4:10 am

Ok...Firstly, I found out that the version was not the commercial version but just the free version...the wrong CD was installed...
So I have corrected that now, so when gw -version and texis -version is run, then
Webinator WWW Site Indexer Version 2.56 (Commercial) and
Texis Web Script (Vortex) Copyright (c) 1996-2000 Thunderstone - EPI, Inc.
Commercial Webinator Version 3.0.955383945 of Apr 10, 2000 (sparc-sun-solaris2.5.1)
are returned.

I have since changed gw to run not as setuid but as webserver..... no change

My .so files are in the directory /usr/lib
but I don't think that that is the problem as anytotx runs if I enter it by the command line.

I tried the command line:
gw -dmydatabase -napplication/pdf,pdf,cat http://myurl.com
just to see if it would at least run 'cat', but the same error message.

I have tried full path to anytotx but the error message just returns with
018 Can't run /full/path/to/anytotx

Do you have a full checklist of EVERYTHING that I can check? Permissions for all files required..?
Does anyone else running Webinator 2.56 have this problem??
This is getting increasingly urgent...

Regards,
Neil.

Post by **mark** » Mon Nov 26, 2001 10:32 am

It sounds like you have the wrong version of the plugin for your version of Webinator. Newer versions of Webinator expect the plugin to respond correctly to the "--identify" option. Please open a ticket at
http://www.thunderstone.com/texis/site/ ... pport.html
with your DR number so that we can get this resolved.
Include the output of gw -version and anytotx -h in the ticket.

stan_jones · Post by **stan_jones** » Mon Nov 26, 2001 8:36 pm

After struggling for months after others here have given up, I have made it past the "no documents found"
to here with 018 .... msg. It is not likely you. It is more likely an install that didn't install right. I am now
at the same point with the same conclusion. There appears from the documentation that we have been
provided with that there is no way to know what the successful install looks like. This is very frustrating.

I, too, put the libs in /usr/lib (on IRIX 6.5) and that only served to provide me with a good response to
--identify command option. I am testing my install against a site http://www.dot.ca.gov/stan/index.html
that has only a few html files and one pdf. Providing the full path to anytotx does not help me either.

I would appreciate knowing what files are needed for anytotx to be used with -n option of gw and what
permissions and directory locations can be for it to index this pdf. I would be willing to make my directories
anything you want if I could only get some success before all my customers depart for some other world.

I have tried the install as root, as webinator user, as web server user and have, like the previous person
gave all the files 777.

-stan-

Post by **mark** » Tue Nov 27, 2001 11:06 am

Ideally webinator 2.x would be installed as the web server user. All of it's files and directories would be owned and writable by that user. Any shared libs that come with the pdf plugin should be in the same directory as anytotx or in the system library directory. If gw is setuid the system may ignore libs that are anywhere but in the system library directory.

Gw releases later than 19990602 will test the plugin on startup by giving it the --identify option. If this fails it will say "Can't run plugin". Plugin releases later than 19990301 understand --identify.

If the walk starts, the test passed. During the walk, if the plugin fails it will say "Plugin Failed".

You can test the plugin on individual files by hand with something like:
anytotx -mTitle <YOURPDFFILE.pdf
It should give back the text of the document.

The pdf file on your example site has no text. Only a scanned picture of text. Your paper to pdf converter needs to perform OCR on the scanned image to create text to search.

stan_jones · Post by **stan_jones** » Thu Nov 29, 2001 5:50 pm

Thanks Mark -----

The pdf files on our site appear to be of the "scanned" type. I found someone who created for me
a "Test.pdf" file with Acrobat.

When I executed "bin/anytotx -mTitle < Test.pdf" it responded by typing out the text in the pdf.

When I executed "bin/gw -n"application,pdf,/www/pub/webinator/bin/anytotx" -dtest2 http://www.dot.ca.gov/stan.html the response was

$ bin/gw -n"application/pdf,pdf,/www/pub/webinator/bin/anytotx" -dtest2 http://www.dot.ca.gov/stan/index.html
018 Can't run plugin /www/pub/webinator/bin/anytotx

$ ls -l bin
total 26176
-rw-rw-r-- 1 webinatr webteam 2078 Mar 26 1999 Readme.txt
-rwsrwxrwx 1 webinatr webteam 341728 Nov 26 01:52 anytotx
-rwxrwxr-x 1 webinatr webteam 36248 Jun 29 2000 geturl
-rwsrwxr-x 1 webinatr webteam 4206192 Nov 26 01:52 gw
-rw-rw-r-- 1 webinatr webteam 209842 Sep 25 1998 gw25.lj
-rw-rw-r-- 1 webinatr webteam 198258 Sep 25 1998 gw25.ps
-rwxrwxr-x 1 webinatr webteam 49472 Jun 29 2000 gwpatch
-r-xr-xr-x 1 webinatr webteam 2346280 Nov 26 01:52 libpdftoolkit.so.3.0
-r-xr-xr-x 1 webinatr webteam 1244336 Nov 26 01:52 libpfs.so
-rw-rw-r-- 1 webinatr webteam 1206 Mar 31 1999 sandr
-rw-r--r-- 1 webinatr webteam 3592 Nov 26 01:52 sandr.vtx
-rwsrwxr-x 1 webinatr webteam 4735856 Jun 29 2000 texis
-rwxrwxr-x 1 webinatr webteam 456 Nov 26 01:52 webinator.sh
$ echo @#
$ echo $LIBPATH

$ echo $PATH
/usr/sbin:/usr/bsd:/sbin:/usr/bin:/etc:/usr/etc:/usr/bin/X11
$ PATH=$PATH:.
$ PATH=$PATH:/www/pub/webinator/bin
$ export PATH
$ echo $PATH
/usr/sbin:/usr/bsd:/sbin:/usr/bin:/etc:/usr/etc:/usr/bin/X11:.:/www/pub/webinator/bin

$

Post by **mark** » Thu Nov 29, 2001 6:26 pm

Your OS probably won't let setuid programs use shared libs from anywhere but the /lib or /usr/lib. One solution is to place the *.so* files into one of those directories. The other solution is to remove the setuid bit from gw and anytotx (chmod u-s gw anytotx).

BTW, the LIBPATH variable is not what you want. On Solaris you want LD_LIBRARY_PATH.