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

#!/mod/bin/jimsh

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

proc loadseries {dir} {
	global seriesmap
	set seriesmap [dict create]

	if {![file exists "$dir/series.info"]} { return }

	if {[catch {set fd [open "$dir/series.info" "r"]} msg]} {
		puts "Error opening series.info: $msg"
		return
	}

	foreach line [split [read $fd] "\n\r"] {
		set i [string first " ==> " $line]
		if {$i == -1} { continue }
		set name [string range $line 0 $($i - 1)]
		set prefix [string range $line $($i + 5) end]
		if {$prefix eq ""} { continue }
		dict set seriesmap $name $prefix
		regsub -all -- {[[:space:]]+} [string tolower $name] "" name
		dict set seriesmap $name $prefix
	}
	$fd close
}

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

	# 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 {[dict exists $seriesmap $title]} {
		set ntitle "$seriesmap($title): $title"
		set title $ntitle
	} else {
		# Try lower case without spaces
		regsub -all -- {[[:space:]]+} [string tolower $title] "" ntitle
		if {[dict exists $seriesmap $ntitle]} {
			set title "$seriesmap($ntitle): $title"
		}
	}

	# 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
}