Split domains to parts

src/classification/basic.sh

@@ -16,22 +16,22 @@ def isSecure:
 def classifyUrl(origin):
 	origin as $origin
 	| {
-		isSameDomain: (.domain | isSameDomain($origin.domain)),
-		isSubdomain: (.domain | isSubdomain($origin.domain)),
+		# TODO: work on .domain.parts, not .domain.original?
+		isSameDomain: (.domain.original | isSameDomain($origin.domain.original)),
+		isSubdomain: (.domain.original | isSubdomain($origin.domain.original)),
 		isSecure: (.protocol | isSecure)
 	};
 
 def mangle(origin):
 	origin as $origin
-	| .url as $urlParts
 	| . + {
-		classification : $urlParts | classifyUrl($origin)
+		classification : .url | classifyUrl($origin)
 	};
 
 .origin.url as $origin
 | {
 	origin: .origin | mangle($origin),
-	requestedUrls: .requestedUrls | map(mangle(($origin)))
+	requestedUrls: .requestedUrls | map(mangle($origin))
 }
 EOF
 

src/classification/disconnect/add.sh

@@ -35,13 +35,8 @@ def deleteEmptyArrayKey(key):
 def matchDisconnect:
 	# Match the domain to disconnect's list.
 	# If the domain is a subdomain of a domain in disconnect's list, include it too.
-	. as $domain
-	| ($domain | split(".")) as $domainParts
-	# Negative range to build the domain from parts from the right.
-	| [ range((($domainParts | length) * -1); -1) ]
-	| map(
-		# Assemble the domain, longest domain combination first.
-		($domainParts[.:] | join(".")) as $subdomain
+	map(
+		. as $subdomain
 		| if $disconnect | has($subdomain) then
 			(
 				# Inject the matched service domain into the returned object.
@@ -59,7 +54,7 @@ def matchDisconnect:
 
 def mangle:
 	.blocks += ({
-			disconnect: .url.domain | matchDisconnect
+			disconnect: .url.domain.parts | matchDisconnect
 		}
 		| deleteEmptyArrayKey("disconnect"))
 	| deleteNullKey("blocks");

src/extract/request/expand-parts.sh

@@ -2,6 +2,24 @@
 set -e
 
 read -d '' expandParts <<-'EOF' || true
+def splitDomainToPartsArray:
+	split(".") as $domainParts
+	# Negative range to build the domain from parts from the right.
+	| [ range((($domainParts | length) * -1); 0) ]
+	| map(
+		# Assemble the domain, longest domain combination first.
+		$domainParts[.:] | join(".")
+	);
+
+def splitDomainToParts:
+	. as $domain
+	| splitDomainToPartsArray as $domainParts
+	| {
+		original: $domain,
+		parts: $domainParts,
+		tld: $domainParts[-1:][0]
+	};
+
 def splitUrlToParts:
 	split("://") as $protocolParts
 	| if ($protocolParts | length) == 1 then
@@ -12,18 +30,10 @@ def splitUrlToParts:
 		{
 			original: .,
 			protocol: $protocolParts[0],
-			domain: ($protocolParts[1] | split("/")[0])
+			domain: ($protocolParts[1] | split("/")[0] | splitDomainToParts)
 		}
 	end;
 
-def classifyUrl(origin):
-	origin as $origin
-	| {
-		isSameDomain: (.domain == $origin.domain),
-		isSubdomain: ((.domain // "") | endswith("." + $origin.domain)),
-		isSecure: (.protocol == "https")
-	};
-
 def trim(str):
 	str as $str
 	| ltrimstr($str) | rtrimstr($str);
@@ -52,9 +62,8 @@ def splitMime:
 	}
 	| deleteNullKeys;
 
-def mangle(origin):
-	origin as $origin
-	| (.url | splitUrlToParts) as $urlParts
+def mangle:
+	(.url | splitUrlToParts) as $urlParts
 	| {
 		url: $urlParts,
 		status: .status,
@@ -64,10 +73,9 @@ def mangle(origin):
 	}
 	| deleteNullKeys;
 
-(.origin.url | splitUrlToParts) as $origin
-| {
-	origin: .origin | mangle($origin),
-	requestedUrls: .requestedUrls | map(mangle(($origin)))
+{
+	origin: .origin | mangle,
+	requestedUrls: .requestedUrls | map(mangle)
 }
 EOF