webif/var/mongoose/html/dedup/normalise.jim

#!/mod/bin/jimsh

set dedup_prefixes {
	{^new series\.* *}
	{^cbeebies\.* *}
	{^cbbc\.* *}
	{^brand new series *-* *}
	{^\.+}
}

proc dedupnormalise {title {reserve ""}} {
	global dedup_prefixes

	# Strip common prefixes
	foreach prefix $dedup_prefixes {
		regsub -nocase -all -- $prefix $title "" title
	}

	# Strip anything following a colon.
	regsub -all -- { *[:].*$} $title "" title

	# If the resulting string is longer than 40 characters then
	# split around . and take the left hand side if appropriate.
	if {[string length $title] > 40} {
		lassign [split $title "."] v w
		set title $v
		if {[string length $title] < 6 && [string length $w] < 6} {
			append title "_$w"
		}
	}

	# if still short, add the reserve string.
	if {[string length $title] < 10} {
		if {[string match "${title}*" $reserve]} {
			set title $reserve
		} else {
			append title " $reserve"
		}
	}

	# Shorten if too long.
	if {[string length $title] > 40} {
		set title [string range $title 0 39]
	}

	return $title
}