#!/bin/ksh
#
# $Id: srch,v 1.9 2008/10/01 18:39:18 vogelke Exp $
#
# NAME:
#    srch
#
# SYNOPSIS:
#    srch [-ev] [-cfhlmnrstuw] pattern
#
# DESCRIPTION:
#    Look through all Estraier DBs for "pattern".
#    Default behavior (no options) is to search all DBs.
#
# OPTIONS:
#    -e    exact phrase search: don't insert AND between words
#    -v    print the version and exit
#
#    -c    search source-code index
#    -f    search index of mail headers received FROM people
#    -h    search home index
#    -l    search logs index
#    -m    search mail-unread index
#    -n    search notebook index
#    -p    search PDF index
#    -r    search root index
#    -s    search mail-saved index
#    -t    search index of mail headers sent TO people
#    -u    search usr index
#    -w    search web index
#
# AUTHOR:
#    Karl Vogel <vogelke@pobox.com>
#    Sumaria Systems, Inc.

lrevno='$RCSfile: srch,v $ $Revision: 1.9 $'
lrevdate='$Date: 2008/10/01 18:39:18 $'

. /usr/local/lib/ksh/path.ksh
. /usr/local/lib/ksh/die.ksh
. /usr/local/lib/ksh/usage.ksh
. /usr/local/lib/ksh/version.ksh

# list just the folders in a directory.

folders () {
    dir=$1
    ls -l $dir | grep '^d' | awk '{print $9}' | grep -v '[A-Z]'
}

#
# Command-line options.
#

sect=
exact='no'

while getopts ":echlmnprstuvw" opt; do
    case $opt in
        c) sect="$sect src" ;;
        e) exact='yes' ;;
        h) sect="$sect home" ;;
        l) sect="$sect logs" ;;
        m) sect="$sect mail-unread" ;;
        n) sect="$sect notebook" ;;
        p) sect="$sect pdf" ;;
        r) sect="$sect root" ;;
        s) sect="$sect mail-saved" ;;
        t) sect="$sect sent" ;;
        u) sect="$sect usr" ;;
        w) sect="$sect web" ;;
        v) version; exit 0 ;;
        \?) usage "-$OPTARG: invalid option"; exit 1 ;;
    esac
done
shift $(($OPTIND - 1))

#
# Sanity checks.
#

top='/search/estraier'
test -d "$top" || die "$top: not found"

#
# If exact search specified, search for that phrase.
# If not, create search pattern: "word1 AND word2 AND word3 ..."
# Run the search, 40 hits maximum.
# Ignore password-checker stuff; that holds dictionaries.
#

case "$@" in
    "") usage "no pattern to search for" ;;
    *)  ;;
esac

case "$exact" in
    yes) pattern="$@" ;;
    no)  pattern=`echo $@ | sed -e 's/ / AND /g'` ;;
esac

echo "Searching for: $pattern"

# top-level directories are sections.
case "$sect" in
    "") sect=`folders $top` ;;
    *)  ;;
esac

for db in $sect; do
    if test -d "$top/$db/srch"; then
        echo
        echo "=== $db"

        estcmd search -sf -vu -max 40 $top/$db/srch "$pattern" |
            grep -v /src/security/password-checker/ |
            grep 'file://' |
            sed -e 's!^.*file://! !'
    fi
done

exit 0