echodeps

#!/bin/sh

#+
#  Name:
#     echodeps

#  Purpose:
#     Report the packages that a package directly depends on, given
#     the package ".properties" file.

#  Description:
#     Extracts the packages that a package depends on by looking in
#     a given ".properties" file for a jar.class.path property.
#     This property contains a list of the jar files that a package
#     depends on and consequently is also an enumeration of the
#     packages that a package depends on. 
#
#     The format of property files means a jar.class.path will
#     probably look something like:
#
#     jar.class.path=bsh-1.2b7.jar splat_colt.jar jdom.jar jhall.jar \
#     splat_help.jar jsci_core.jar jsci_wavelet.jar \
#     ../astgui/astgui.jar ../ndx/ndx.jar ../hdx/hdx.jar ../array/array.jar \
#     ../tamfits/tamfits.jar ../jniast/jniast.jar \
#     ../soapserver/soapserver.jar ../fits/fits.jar ../hds/hds.jar \
#     ../treeview/treeview.jar ../pal/pal.jar ../diva/diva.jar
#
#     So this script does the job of joining the continuation parts,
#     extracting the jar.class.path property and then reporting the 
#     package part of each jar file (the xxx part of ../xxx/some.jar).
#
#     Note this must be run from the "source" directory.

#  Type of Module:
#     Shell script.

#  Copyright:
#     Copyright (C) 2003 Central Laboratory of the Research Councils

#  Authors:
#     PWD: P.W. Draper (Starlink, Durham University)
#     {enter_new_authors_here}

#  History:
#     30-SEP-2003 (PWD):
#        Original version.
#     {enter_further_changes_here}

#  Bugs:
#     {note_any_bugs_here}

#-

if test "$1" = ""; then 
   echo "Usage: ./scripts/echodeps <.properties file>"
   exit
fi

#  Get sed to join continuation lines (ending in a \) and then extract
#  the line starting "jar.class.path=". Strip off any
#  .jar extensions (needed for ant which is ../ant.jar).  Spaces in
#  that line are then replaced by a newline so that each word becomes
#  a line. All lines containing "../" are then resolved by grep (lines
#  that do not start with "../" are local jar files). Finally awk uses
#  '/' to separate the lines into words of which the second should be
#  the package name!

cat $1 | \
sed -n \
   -e :a \
      -e '/\\$/N; s/\\\n//; ta' \
   -e "/jar\.class\.path=/ s/jar\.class\.path=//gp" | \
sed -e 's/\.jar//g' | \
tr " " "\n" | \
grep '\.\./' | \
awk -F '/' '{print $2}'