BibGlimpse.SETUP

#!/bin/bash

# Script configures BibGlimpse for
# running WebGlimpse installations
# 
# (c) Boku Bioinformatics,
#     Thomas Tuechler, 2006

# Output verbosity: '' or '-v'
V='';

echo "";
echo "WELCOME TO BIBGLIMPSE INSTALLATION SCRIPT";
echo "----------------------------------------------------------------";
echo "";
echo "";


# 0) Glimpse installation
# -----------------------------------------------------------------------------
# make sure that Glimpse is configured with the following two flags:
echo "Are you sure that glimpse was configured with the following two flags:";
echo "   ./configure --with-file-end-mark='\t'";
echo "               --enable-structured-queries;";
echo "If not, please recompile glimpse first, else press enter to continue...";
read goon;


# 1) getting the paths
# -----------------------------------------------------------------------------

# getting the WG2LIB installation path
while [ ! -s "$WG2LIB"/lib/wrMedline.pl ];
do
  echo "Please specify the path of the Webglimpse installation (/usr/local/lib/wg2):";
  read WG2LIB;
  
  if [ ! -s "$WG2LIB"/lib/wrMedline.pl ]; then
      echo "'$WG2LIB' does not contain the BibGlimpse modules!";
  fi
done;


# getting the WG2CGI path
while [ ! -s "$WG2CGI"/wrrepos.cgi ];
do
  echo "Please specify the cgi-bin path (/usr/local/lib/apache/cgi-bin):";
  read WG2CGI;

  if [ ! -s "$WG2CGI"/wrrepos.cgi ]; then
      echo "'$WG2CGI' does not contain the BibGlimpse modules!";
  fi
done;


# getting the WG2ARC path
while [ ! -d "$WG2ARC" ] && [ -z "$WG2ARC" ];
do
  echo "Please specify the archive path (/usr/local/lib/apache/wg2/archives):";
  read WG2ARC;

  if [ ! -d "$WG2ARC" ]; then
      echo "'$WG2ARC' is not a directory!";
  fi
done;


# getting the USRBIN path
while [ ! -d "$USRBIN" ] && [ -z "$USRBIN" ];
do
  echo "Please specify the archive path (/usr/local/bin):";
  read USRBIN;

  if [ ! -d "$USRBIN" ]; then
      echo "'$USRBIN' is not a directory!";
  fi
done;


# 2) $WG/lib/(wr)usexpdf.sh
# -----------------------------------------------------------------------------
# use the advanced pdf filtering tool wrusexpdf.sh instead of the old one
# and include the correct binaries path for pdftotext, md5sum etc.
echo "";
echo "Including the special PDF filter 'wrusexpdf.sh'...";

mv $V "$WG2LIB"/lib/usexpdf.sh "$WG2LIB"/lib/old.usexpdf.sh;

sed "s^export PATH\=\/usr\/local\/bin\:\$PATH\;^export PATH\=$USRBIN\:\$PATH\;^" < "$WG2LIB"/lib/wrusexpdf.sh > "$WG2LIB"/lib/usexpdf.sh;

[ -s "$WG2LIB"/lib/usexpdf.sh ] && rm $V "$WG2LIB"/lib/wrusexpdf.sh;

chmod +x "$WG2LIB"/lib/usexpdf.sh;


# 3) $WG/templates/.glimpse_filters
# -----------------------------------------------------------------------------
# add the paths to usexpdf.sh to the .glimpse_filters template. this will
# make the program index pdf files as well.
echo "Setting the '.glimpse_filters' to use 'usexpdf.sh'...";

echo -e "*.pdf\t|WGHOME|/lib/usexpdf.sh\n*.PDF\t|WGHOME|/lib/usexpdf.sh" >> "$WG2LIB"/templates/.glimpse_filters;


# 4) $WG/dist/wgfilter-index
# -----------------------------------------------------------------------------
# adapt this template file such that it accepts *.pdf's by replacing and adding
echo "Setting the 'wgfilter-index' to allow PDFs...";

mv $V "$WG2LIB"/dist/wgfilter-index "$WG2LIB"/dist/old.wgfilter-index;
sed -e "s^Deny \\\\.pdf\\$^Allow \\\\.pdf\\$\nAllow \\\\.PDF\\$^" -e "s^Allow \\\\..?html?\\$^Deny \\\\.s?html?\\$^" < "$WG2LIB"/dist/old.wgfilter-index > "$WG2LIB"/dist/wgfilter-index;

echo -e "Deny \.pl$\nDeny \.sh$\nDeny \.tmp$\nDeny \.ag\.\nDeny \.bib\$\nDeny \.medline\$\nDeny \.medl\$\nDeny \.suppl\$\nDeny \.pubmed\$\nDeny \.anno\$\nDeny \.htaccess\nDeny README\.txt" >> "$WG2LIB"/dist/wgfilter-index;

# 5) $WG/templates/wgreindex
# -----------------------------------------------------------------------------
# enable the glimpseindex building with structured fields with -s and -z flag:
echo "Enabling indexbuilding with structured queries in 'wgreindex'...";

mv $V "$WG2LIB"/templates/wgreindex "$WG2LIB"/templates/old.wgreindex;
sed -e "s^-H |INDEXDIR| -t -o^-H |INDEXDIR| -s -z -t -o^g" < "$WG2LIB"/templates/old.wgreindex > "$WG2LIB"/templates/wgreindex;


# 6) $WG/lib/wgHeader.pm
# -----------------------------------------------------------------------------
# Change the $FILE_END_MARK=" " definition in the $WG/lib/wgHeader.pm file
# to "\t". in accordance with the --file-end-mark='\t' flag used in the
# Glimpse installation. This will allow filenames with spaces.
echo "Enabling the BibGlimpse features in 'wgHeader.pm'...";

mv $V "$WG2LIB"/lib/wgHeader.pm "$WG2LIB"/lib/old.wgHeader.pm;
sed -e "s^\$FILE_END_MARK = \" \"^\$FILE_END_MARK = \"\\\t\"^" -e "s^\$WRREPOS = 0^\$WRREPOS = 1^" < "$WG2LIB"/lib/old.wgHeader.pm > "$WG2LIB"/lib/wgHeader.pm;


# 7) $CGI/wrrepos.cgi
#    $CGI/wrsearch.cgi
# -----------------------------------------------------------------------------
# Specify path to webglimpse libraries (eg. to '/bi/arch/lib/wg2'):
echo "Linking 'wrrepos.cgi' and 'wrsearch.cgi' to the WebGlimpse libraries...";

sed "s^\$WEBGLIMPSE_LIB='';^\$WEBGLIMPSE_LIB='$WG2LIB/lib';^" < "$WG2CGI"/wrrepos.cgi > "$WG2CGI"/new.wrrepos.cgi;
mv $V "$WG2CGI"/new.wrrepos.cgi "$WG2CGI"/wrrepos.cgi;
chmod +x "$WG2CGI"/wrrepos.cgi;

sed "s^\$WEBGLIMPSE_LIB='';^\$WEBGLIMPSE_LIB='$WG2LIB/lib';^" < "$WG2CGI"/wrsearch.cgi > "$WG2CGI"/new.wrsearch.cgi;
mv $V "$WG2CGI"/new.wrsearch.cgi "$WG2CGI"/wrsearch.cgi;
chmod +x "$WG2CGI"/wrsearch.cgi;

# 8) $CGI/logout/logout.cgi
# -----------------------------------------------------------------------------
# If you want to restrict access to your repository by means of .htaccess,
# create a logout directory within $CGI, that has a separate .htaccess
# file containing:
#
#  AuthName "Logout"
#  AuthUserFile /your/path/to/.htpasswd 
#  Require user logout
#
# which allows for the user 'logout' only. Do not forget to provide
# a user 'logout' with password logout in your .htpasswd file!
# All this provides a basic logout option using .htaccess user control.

echo "";
echo "If you want to have a logout button for .htaccess access";
echo "control included on the web, please consult the corresponding";
echo "instructions in the BibGlimpse.README. Press enter to";
echo "continue standard installation... ";
read goon;

mkdir $WG2CGI/logout;
cat > "$WG2CGI"/logout/logout.cgi << EOF
#!/usr/bin/perl -w
use strict;
use CGI;
use CGI::Carp qw(fatalsToBrowser);
my \$q = new CGI;
print \$q->header();
my $user = \$ENV\{REMOTE_USER\};
print
    "You are now logged out!<br />",
    "Current REMOTE_USER is set to \$user.<br />",
    "<br />",
    "(On some browsers you still need to close the browser,<br />",
    " to ensure that the previous pages can not be reloaded!)";
EOF
chmod +x "$WG2CGI"/logout/logout.cgi


# *************************************************************************** #
#  The optional modifications regarding the layout of the search results      #
#  are not implemented within this basic installation script.                 #
# *************************************************************************** #

echo "If you want to customize the format of the search results,";
echo "please refer to the optional points in the BibGlimpse.README.";
echo "Press enter to continue standard installation... ";
read goon;


# *************************************************************************** #
#  The configuration of the archives is best done on the wgarcmin webinter-   #
#  face and is thus to be done there.                                         #
# *************************************************************************** #

# 11) configure archive via webinterface
# -----------------------------------------------------------------------------

echo "To configure the archive it is best to use the webingerface.";
echo "Please open a browser and go the wgarcmin.cgi you have just";
echo "installed on your server.";
echo "Use YOUR settings filling in the forms described below:";
echo "";
echo " NOTE: You need to create two archives: One archive is to be";
echo "       accessed from the web all the time (the 'main' archive),";
echo "       whilst the other one is for stealthly reindexing in background.";
echo "       Check for correct permissions! Webuser must have full access";
echo "       to the directories to index!";
echo ""
echo " Edit domain configureation:";
echo "   Server Name: your.server.com";
echo "   Document Root: /your/servers/htdocs";
echo "   Port: xxxx";
echo "   Protocol: http";
echo "   User Dir: public_html";
echo "   Directory Index Files: index.html index.htm index.shtml index.php";
echo "   ScriptAlias & Extensions: \.php.?(\?|$) \/cgi";
echo "";
echo " Add new Archive";
echo "   Title: Your_prints_repository_main";
echo "   Description: Prints repository";
echo "   Index by Directory";
echo "     Directory Path: /your/servers/htdocs/prints";
echo "     Equivalent URL: http://your.server.com:xxxx/prints";
echo "     Max #Pages: 99999";
echo "";
echo " Add new Archive";
echo "   Title: Your_prints_repository_stealth";
echo "   Description: Prints repository";
echo "   Index by Directory";
echo "     Directory Path: /your/servers/htdocs/prints";
echo "     Equivalent URL: http://your.server.com:xxxx/prints";
echo "     Max #Pages: 99999";
echo "";
echo " Press enter to continue when webconfiguration is done...";
read goon;


# 12) Reindexing in background
# -----------------------------------------------------------------------------
# From the above you should now have two directories in $ARC,
# namely $ARC/1 and $ARC/2.
# To keep the working archive $ARC/1 searchable while reindexing
# wrwgreindex will mirror $ARC/1 to $ARC/2, build the index there
# and then port the resulting index back to $ARC/1.
# To do so, we need to replace the standard wgreindex script in $ARC/1.
# Set $ARC to your full archive path and execute the commands below:
# Note that reindexing can also be called from Crontab, as described
# in the WebGlimpse documentation!
#
# ATTENTION: Take care that sym-links are not interpolated:
#            /your/archives/1 must not become
#	     /your/somewhere/else/because/of/symlink/arch/...
#            or the 'sed' command in wrwgreindex will fail
#            and the abra retrieval in wrrepos.cgi will
#            silently produce crap!
echo "Implementing reindexing in background...";


WG2ARC_MAIN="$WG2ARC/1";
WG2ARC_STEALTH="$WG2ARC/2";

cat > "$WG2ARC_MAIN"/wgreindex << EOF
#!/bin/sh

# Since this is the visible archive we will not do any
# reindexing here. instead we will move the reindexing
# to the stealth twin of this archive.
# 
# (c) Boku Bioinformatics
#     Thomas Tuechler, David Kreil 2006
#

rm -f /tmp/.wg*;
"$WG2ARC_MAIN"/wrwgreindex \$1 "$WG2ARC_STEALTH";
EOF

chmod +x $WG2ARC_MAIN/wgreindex;

echo "Finished installation. Start indexing with:";
echo "$WG2ARC/1/wgreindex;";
echo "";