:
# mkcategtab - make HTML table from command list, grouped by categories
#
# Author:	Heiner Steven (heiner.steven@odn.de), 1/00
# Category:	HTML
#
# Thanks to Jean-Pierre Pilote for pointing out a off-by-one bug, causing
# the program to use wrong directory names for generated URL.
#
# $Id: mkcategtab,v 1.9 2014/09/08 18:46:26 heiner Exp $

PN=`basename "$0"`

PATH=$PATH:../../bin export PATH

: ${AWK:=awk}

# Ignore all commands ending in ".txt"
Args=
while [ $# -gt 0 ]
do
    if [ -d "$1" ]
    then
    	: 				# ignore directory
    else
	case "$1" in
	    *.txt)	;;			# Ignore
	    *)	Args="$Args $1";;
	esac
    fi
    shift
done
set -- $Args

Categories=${TMPDIR:=/tmp}/mkcmd$$
NewFiles=$TMPDIR/mkcmd$$.new
trap 'rm -f "$Categories" "$NewFiles"' 0
trap "exit 2" 1 2 3 13 15

# Create a list of new files
lastmonth=`fdate +"%Y-%m-01" -1 month`
lsrecent -d $lastmonth "$@" > "$NewFiles"

# Search all "Category:" header lines, and write them to a file.
# All lines *must* be sorted in dictionary order for "join" to work.
# File format:
#	filename<TAB>category text
egrep -i '^#[ 	]+(category|categories|kategorie|kategorien)[. 	]*:[ 	]*' "$@" |
	awk -F: '!L[$1]++' |
	sed -e 's/[ 	][ 	]*/ /g' \
	    -e 's/:#[ 	]*[cC][aA][tT][eE][gG][oO][rR][yY][. 	]*:[ 	]*/	/' \
	    -e 's/:#[ 	]*[cC][aA][tT][eE][gG][oO][rR][iI][eE][sS][. 	]*:[ 	]*/	/' \
	    -e 's/:#[ 	]*[kK][aA][tT][eE][gG][oO][rR][iI][eE][. 	]*:[ 	]*/	/' \
	    -e 's/:#[ 	]*[kK][aA][tT][eE][gG][oO][rR][iI][eE][nN][. 	]*:[ 	]*/	/' |
	sort -k1,1 > "$Categories" || exit

{
    # Format: "filename:# Shellscript: name - descriptive text"
    egrep '^# (Shellscript|Title[ 	]*):[ 	]' "$@" |
	sed -e 's/[ 	][ 	]*/ /g'			\
	    -e 's/:# Shellscript:[ 	]*/	/' 	\
	    -e 's/:# Title[ 	]*:[ 	]*/	/' 

    # Format: "filename:# name - descriptive text"
    egrep '^#[ 	][ 	]*[a-z0-9][a-z0-9]* - ' "$@" |
	sed 's/[ 	][ 	]*/ /g' |
	sed 's/:\#[ 	]*/	/'

    # Output format:
    #	filename<TAB>descriptive text
} |
    awk -F"	"  '!L[$1]++' |		# each file can appear only once
    sort -f -k1,1 |
    # Input must be sorted on the first field for join to work
    join -t '	' - $Categories |
    # Output format (sorted by command name, case ignored):
    #	filename<TAB>name - descriptive text<TAB>category
    sort -k2 |
    $AWK -F'	' '
        BEGIN {
	    ChangeText ["new"]    = "${NEWTAG}"
	    ChangeText ["update"] = "${UPDATETAG}"
	    NewFiles = "'"$NewFiles"'"
	    # Input example:
	    #	test.sh	update	2000-02-02	1.2
	    #	testing	new	2000-02-02	1.2
	    while ( getline < NewFiles ) {
	    	IsNewFile [$1] = $2
	    }
	    close (NewFiles)
	}
	{
	    filename    = $1
	    description = $2
	    categories  = $3
	    haschanged  = ChangeText [IsNewFile [filename]]

	    if ( filename ~ /\// ) {
		for ( i=length (filename); i>0; --i ) {
		    if ( substr (filename, i, 1) == "/" ) break
		}
		if ( i > 0 ) {
		    dirname = substr (filename, 1, i-1) "/"
		    basename = substr (filename, i+1, length (filename) - i)
		} else {
		    dirname = ""
		    basename = filename
		}
	    }

	    # The description has the format
	    #	name - descriptive text [...]
	    n = split (description, d, "[ 	]")
	    cmdname = ""
	    for ( i=1; i<=n; i++ ) {
	    	if ( d [i] == "-" ) {
		    ++i; break
		}
		if ( i>1 ) cmdname = cmdname " "
		cmdname = cmdname "" d [i]
	    }

	    descr = ""
	    for ( j=i; j<=n; j++ ) {
	    	if ( j > i ) descr = descr " "
		descr = descr d [j]
	    }

	    if ( basename "" != cmdname &&
		    length (basename) >= length (cmdname) ) {
		print "WARNING: file " filename " is called " cmdname | "cat >&2"
	    }

	    # We should take "basename" instead of "cmdname", but
	    # with DOS the name would be truncated to 8 characters
	    fileurl = dirname "" cmdname

	    if ( fileurl == "" ) fileurl = "&nbsp;"
	    if ( descr == "" ) descr = "&nbsp;"
	    url [cmdname]   = fileurl
	    ChangeNote [cmdname] = haschanged
	    descriptions [cmdname] = descr

	    # "category" may contain more than one category,
	    # separated by ";" or ","
	    n = split (categories, c, "[,;]")
	    for ( i=1; i<=n; i++ ) {
	    	category = c [i]
		if ( category == "" ) continue
		# Remove leading and trailing whitespace
	    	gsub (/^[ 	]*/, "", category)
	    	gsub (/[ 	]*$/, "", category)
		# Substitute blanks with "non breaking spaces"
		gsub (/[ 	][ 	]*/, "\\&nbsp;", category)
		if ( cmds [category] != "" )
		    cmds [category] = cmds [category] "	"
		cmds [category] = cmds [category] "" cmdname
	    }
	}
	END {
	    # Now write the HTML output.
	    # We will take care to write the tables in a way
	    # that older browsers may display the result, too.

	    # Copy all category texts to the vector categ[]. This
	    # way we can sort the vectory on the category name.
	    ncat = 0
	    for ( category in cmds ) {
	    	categ [++ncat] = category
	    }

	    # Sort categories in categ[] using "Insertion Sort"
	    for ( i=1; i<=ncat; i++ ) {
	        smallest = i;
	    	for ( j=i+1; j<=ncat; j++ ) {
		    if ( categ [j] < categ [smallest] )
		    	smallest = j;	# find position of smallest element
		}
		tmp = categ [i]
		categ [i] = categ [smallest]
		categ [smallest] = tmp
	    }

	    # At first print a list of links to directly jump to a category
	    print "<DIV ALIGN=CENTER>"
	    print "[ <SMALL>"
	    for ( i=1; i<=ncat; i++ ) {
		category = categ [i]
	        L = "<A HREF=\"#category" i "\">" category "</A>"
		if ( i < ncat ) L = L " | "
		print L
	    }
	    print "</SMALL> ]"
	    print "</DIV>\n"

	    # Now write all categories, using a table for the
	    # commands contained within.
	    for ( i=1; i<=ncat; i++ ) {
	    	category = categ [i]
		n = split (cmds [category], c, "[	]")
		# c[1..n] contains the command names

	    	print "<P>${SSH}<A NAME=\"category" i "\">${SFIXEDFONT}" category "${EFIXEDFONT}</A>${ESH}</P>"
		#L = "<SMALL>"
		#L = L "<A HREF=\"#top\">Top</A></SMALL>"
		#print L
		print "<TABLE BORDER=\"0\" WIDTH=\"100%\">"
		for ( j=1; j<=n; j++ ) {
		    cmd = c [j]
		    viewurl = url [cmd] ".txt"
		    L = "  <TR><TH ALIGN=\"RIGHT\">"
		    L = L ChangeNote [cmd]
		    L = L "<A HREF=\"" viewurl "\">" cmd "</A></TH>"
		    L = L " <TD><A HREF=\"" url [cmd] "\">&nbsp;D&nbsp;</A></TD>"
		    L = L " <TD>" descriptions [cmd]
		    # Special handling: terminate line for older browsers
		    L = L "<BR></TD></TR>"
		    print L
		}
		print "</TABLE>\n"
	    }

	    print "<CENTER><P><SMALL>This table was generated ${DATE} using the script \"'"$PN"'\"</SMALL></P></CENTER>"
	}
    '