#! /usr/bin/gnu/bash #### lynxfilter.sh #### # *** # *** CONFIGURATION # *** # export all variable setting set -a ############# Section you need to change ################### # I (dd) had to fully specify these two, maybe a bug in 2.8.2 # make them point to your own lynx binary and config file LYNX=/u/www47/0/w3c/danield/tmp/lynx2-8-2/lynx LYNX_FLAGS="-cfg=/u/www47/0/w3c/danield/tmp/lynx2-8-2/lynx.cfg" # flag for the filter itself # tablin is a simple script that does "java Linearize $*" # (it's provided with the tablin source archive) FILTER=/u/www47/0/w3c/danield/tmp/Tablin/tablin FILTER_FLAGS= # where you put the Tablin java classes root CLASSPATH=/u/www47/0/w3c/danield/tmp/Tablin # if you need a local version of java, like I do PATH=/u/tarantula/0/w3c/ylafon/jdk1.1.7A/bin/:$PATH ############# End section you need to change ################### ############# DOCUMENTATION ################################### # This script can be used within lynx, in a CGI-like environment, # to run a given filter program on a given URL document and display # the output instead of the original. # This filter must read HTML text from stdin and produces HTML text on stdout. # # A child lynx process is used to retrieve the HTML document (one # could use wget or libwww-perl instead). No temporary files needed. # # Required: # o Lynx on a Unix system, configured and compiled with LYNCGI_LINKS # support (only tested with 2.8.2) # o Bourne shell for the script (only tested with bash) # o A working binary of the filter you want to use (this script is # tailored for stdin/stdout-like filters) # o sed (optional: Perl) # # # TEST: # Instructions for use with a test filter called 'aafilter': # o Put the line "sed s/a/aa/g" into a file called 'aafilter' # somewhere on your system (this doubles all 'a' in its input) # chmod +x on this file (for unix user) # Put location of this executable on the line below that starts # with FILTER= # o Change other paths and flags in configuration section below # as needed. # o Make sure Lynx is compiled with LYNXCGI support. Make lynxfilter.sh # is executable with chmod. Put it in a location where your lynx.cfg # allows lynxcgi scripts. (Read the relevant comments in lynx.cfg # if you are not sure, search for CGI.) # o Define the following environment variables *outside of lynx*, # before starting lynx. Putting them in lynx.cfg *does not work*. # export xhttp_proxy='lynxcgi://localhost/PATH/TO/lynxfilter.sh?/' # (Note the extra ?/ at the end) # # Usage: # o When you view a Page in lynx that you want to run through the filter, # - press 'G' (note capital) to start editing current document's URL # - prefix URL with an x, so that "http:" becomes "xhttp:", and # press ENTER. # In other words, the 4 key sequence 'G', Ctrl-A, 'x', ENTER. # - you should see the same page with 'aa' instead of 'a' everywhere. # # Debug: # Note that you may get an empty page if the filter fails to produce # anything but error messages. # o Test whether you can access the following URL with Lynx: # # Of course, change PATH/TO, to wherever you put this script. # o When you also want to see the messages from the filter together with # the rendered document, use xx instead of one x. In other words, define # xxhttp_proxy='lynxcgi://localhost/PATH/TO/lynxfilter.sh?/' # and then use the five keys 'G', Ctrl-A, 'x', 'x', ENTER. # Note that messages are intermixed with the filtered document, # so the combination fed to lynx see is most likely invalid HTML; # lynx may complain about it or produce strange rendering sometimes. # # Tips: # o If enabled below with ALLOW_FILTER_PARAM_FLAGS, additional flags can # be passed to the filter by appending ";opt<=some_flags>" to the x-URL. # Use '+' characters for spaces, for example # For example "xhttp://some.server/something.html;opt=-e+-utf8". # o Settings for no_proxy (from lynx.cfg or environment) still apply # for our fake lynxcgi proxy, so beware. For example xfile_proxy # will not work for xfile://localhost/ if localhost is listed in # no_proxy. # o If you use a (real) proxy, you should export the relevant variables # to this script, i.e. add LYNXCGI_ENVIRONMENT:http_proxy etc. # o You may want to create a separate lynx.cfg file for the child lynx # process (see LYNX_FLAGS below) and take out unnecessary options. # Setting GLOBAL_MAILCAP and PERSONAL_MAILCAP to /dev/null should # make the child lynx start up faster, especially if you have lots # of tests in the mailcap files. # o Add equivalent variables xftp_proxy, xxftp_proxy etc. if you want, # or even xfile_proxy, xxfile_proxy for local files. # *** Any non-empty value means YES for boolean settings. # Should we use perl? USE_PERL=YES # Preserve HTTP headers, needs perl, ignored if USE_PERL is unset. USE_MIME_HEADER=YES PERL=perl SED=sed ALLOW_FILTER_PARAM_FLAGS=YES # look for filter options in URL param DEBUG= # Set this to get a lot of debugging info instead of rendered doc straceprefix='/usr/bin/strace -o strace.log' # if you have strace, for DEBUG traceflags='-trace -tlog' # for the child lynx process, with DEBUG # *** # *** end of configuration # *** if [ "$DEBUG" ]; then echo "content-type: text/plain" echo set -vx # /usr/sbin/lsof -p $$ # if you have it... debugflags="$traceflags" prefix="$straceprefix" fi if [ ! -x "$LYNX" -o ! -x "$FILTER" ]; then echo "content-type: text/plain" echo echo "$LYNX or $FILTER not found, or cannot be executed." exit fi if [ "$USE_MIME_HEADER" ]; then if [ "${QUERY_STRING#/xhttp}" = "$QUERY_STRING" -a \ "${QUERY_STRING#/xxhttp}" = "$QUERY_STRING" ]; then USE_MIME_HEADER="" # -mime_header works only for http URLs fi fi # It's finally time to look at the URL... if [ "${QUERY_STRING#/xx}" = "$QUERY_STRING" ]; then IGNORE_STDERR=YES # if QUERY_STRING does not start with "/xx" real_url="${QUERY_STRING#/x}" else IGNORE_STDERR="" # QUERY_STRING should start with "/xx" real_url="${QUERY_STRING#/xx}" fi if [ "$ALLOW_FILTER_PARAM_FLAGS" ]; then if [ "${QUERY_STRING%;opt=*}" != "$QUERY_STRING" ]; then FILTER_PARAM_FLAGS="${QUERY_STRING##*;opt=}" FILTER_PARAM_FLAGS="${FILTER_PARAM_FLAGS//+/ }" real_url="${real_url%;opt=*}" fi fi if [ "$USE_MIME_HEADER" ]; then lynxmainflag="-mime_header" else echo "content-type: text/html" echo lynxmainflag="-source" fi FILTERCOMMAND="$FILTER $FILTER_FLAGS $FILTER_PARAM_FLAGS" if [ "$USE_PERL" ]; then if [ "$USE_MIME_HEADER" ];then PERL_COMMANDS='$|=1; while (<>) { if (($. == 1 && (/^HTTP/))../^\s*$/) {print $_;next;} if (!$first++) { open(P,"|'$FILTERCOMMAND'") || die "could not start the filter"; select P;$|=1; } print P; } close P or die "could not close pipe"' else # USE_PERL but not MIME_HEDER, not very useful PERL_COMMANDS='$|=1; while (<>) { if (!$first++) { open(P,"|'$FILTERCOMMAND'") || die "could not start the filter"; select P;$|=1; } print P; } close P or die "could not close pipe"' fi fi LYNXCOMMAND="$prefix $LYNX $real_url $LYNX_FLAGS $lynxmainflag $debugflags" if [ "$USE_PERL" ]; then if [ "$IGNORE_STDERR" ]; then $LYNXCOMMAND /dev/null else # intermix stderr with output $LYNXCOMMAND &2 2>&1 1>&3 \ | $SED -e 's/\&/\&/g' \ -e 's/filter: \1\ <\/EM><\/PRE>/' fi else # not USE_PERL if [ "$IGNORE_STDERR" ]; then $LYNXCOMMAND \ | $FILTERCOMMAND 2>/dev/null else # intermix stderr with output $LYNXCOMMAND \ | $FILTERCOMMAND 3>&2 2>&1 1>&3 \ | $SED -e 's/\&/\&/g' \ -e 's/filter: \1\ <\/EM><\/PRE>/' fi fi