#
# format a DAX as readable text
#


awkprog1=`mktemp /tmp/awk.XXXXXX`
awkprog2=`mktemp /tmp/awk.XXXXXX`
dax=`mktemp /tmp/dax.XXXXXX`

cat > $dax

#  <job id="ID000001" namespace="air" name="align_warp" level="5" dv-namespace="wilde.ipaw1.wf1" dv-name="align_warp.d001.s0001">
#    <argument><filename file="Data/Raw/reference.img"/> <filename file="Data/Raw/anatomy1.img"/> <filename file="Data/Derived/warp1.warp"/> -m 12 -q</argument>

# First find the max level in the dax

cat <<EOF >$awkprog1

BEGIN { maxlevel = 0; }

/<job / {
	level = \$0
	gsub("\\"", "", level);
	sub("^.*level=", "", level);
	sub(" .*\$", "", level);
	if (level > maxlevel) maxlevel = level;
}

END { print maxlevel }

EOF

MAXLEVEL=`awk -f $awkprog1 $dax`

echo MAXLEVEL=$MAXLEVEL

cat <<EOF >$awkprog2

/<adag / {
	wfname = \$0
	gsub("\\"", "", wfname);
	sub("^.*name=", "", wfname);
	sub(" .*\$", "", wfname);
	printf("Provenance trace for workflow %s\n\n", wfname);
	printf("Files used in workflow:\n\n")
}

/<job / {
	line = \$0
	sub( "^.*<job ", "", line);
	sub( ">$", "", line);
	gsub( "\\"", "", line);
	split(line, nvs, " ");
	for (i in nvs) {
		split(nvs[i], nvp, "=");
		ja[nvp[1]] = nvp[2];
	}
	printf "JOB %s [%d] %s::%s\n", ja["id"], $MAXLEVEL-ja["level"]+1, ja["dv-namespace"], ja["dv-name"]
	print "\n  File usage:\n";

}

/<job /,/job>/ {
	# job = job \$0 "\n";
}

/job>/ {
	print "\n  Runtime provenance:\n";

	sqlcmd = sprintf( "sq -h \"SELECT i.start, i.duration, exitcode, resource, host FROM ptc_invocation i, ptc_job j where wf_label = '%s' and dv_name = '%s' and j.id = i.id;\"", wfname, ja["dv-name"]);
        while ((sqlcmd | getline) > 0)
               printf("    %s\n",\$0);
	close(sqlcmd);
	print "";

	sqlcmd = sprintf( "sq -h \"SELECT r.* FROM ptc_invocation i, ptc_job j, ptc_rusage r where wf_label = '%s' and dv_name = '%s' and i.id = j.id and j.rusage = r.id;\"", wfname, ja["dv-name"]);
        while ((sqlcmd | getline) > 0)
               printf("    %s\n",\$0);
	close(sqlcmd);
	print "";

}

#  <filename file="Data/Derived/atlas-x.pgm" link="inout"/>

/<filename file=.*link=/ {
	fline = \$0
	sub("^.*<filename ", "", fline);
	gsub("\\"", "", fline);
	sub("link=", "usage=", fline);
        sub("/>","",fline);
#	print "    " fline;
}

/<!-- part 2/ {
	print "";
}

# <uses file="Data/Raw/anatomy1.img" link="input" dontRegister="false" dontTransfer="false"/>

/<uses file/ {
	fline = \$0
	sub("^.*<uses ", "", fline);
	gsub("\\"", "", fline);
	sub("link=", "usage=", fline);
        sub(" dont.*/>","",fline);
	print "    " fline;
}
/<child ref=/ {
	child = \$2;
	sub( /^.*ID/,"ID",child);
	sub( /".*/,"",child);
	parents = "";
}

/<parent ref=/ {
	parent = \$2;
	sub( /^.*ID/,"ID",parent);
	sub( /".*/,"",parent);
	parents = parents " " parent;
}

/<\\/child>/ {
	parlist[child] = parents;
#	print "DB: set parents of" child " to " parents;
}

END {

}

EOF

awk -f $awkprog2 $dax 

rm -f $awkprog1 $awkprog2 $dax