Improve tracker source and first-party cleanup

Surface every CookieSource value in the console: the trackers page
filter was missing the HTTP option and the source badge helper had no
EXTENSION case, so HTTP-sourced rows could not be filtered and
extension-sourced rows rendered the raw enum string.

On the backend, the mark-first-party verdict now blanks the stale
description on both the catalog row and its uncategorised org tracker
patterns. A terminal non-third-party row keeps no vendor link, so a
description naming the (now-cleared) vendor would be misleading; the
mapping worker only copies descriptions into empty rows and never
clears them, so clearing is done explicitly here.

Signed-off-by: Émile Ré <emile@probo.com>

apps/console/src/pages/organizations/cookie-banners/configuration/trackers/CookieBannerTrackersPage.tsx

@@ -244,6 +244,7 @@ export default function CookieBannerTrackersPage({
           <Option value="ALL">{__("All sources")}</Option>
           <Option value="SCRIPT">{__("Script")}</Option>
           <Option value="PRE_EXISTING">{__("Pre-existing")}</Option>
+          <Option value="HTTP">{__("HTTP")}</Option>
           <Option value="EXTENSION">{__("Extension")}</Option>
         </Select>
         <Select

packages/helpers/src/tracker.ts

@@ -43,6 +43,7 @@ export function getTrackerSourceBadge(source: string, __: Translator): Badge {
     case "SCRIPT": return { label: __("Script"), variant: "info" };
     case "PRE_EXISTING": return { label: __("Pre-existing"), variant: "outline" };
     case "HTTP": return { label: __("HTTP"), variant: "neutral" };
+    case "EXTENSION": return { label: __("Extension"), variant: "warning" };
     default: return { label: source, variant: "neutral" };
   }
 }

pkg/coredata/common_tracker_pattern.go

@@ -1015,6 +1015,38 @@ WHERE
 	return result.RowsAffected(), nil
 }
 
+// ClearDescriptionByIDs blanks the researched description on the given
+// catalog rows without re-arming enrichment or touching the enrichment
+// payload. It backs the first-party verdict: a terminal non-third-party
+// row keeps no vendor link, so a description that named the (now-cleared)
+// vendor would be stale. The verdict is terminal, so rather than re-derive
+// a description - which would re-run the mapping agent and re-link a
+// vendor - the description simply returns to empty. Returns the number of
+// rows updated.
+func (ps *CommonTrackerPatterns) ClearDescriptionByIDs(
+	ctx context.Context,
+	tx pg.Tx,
+	ids []gid.GID,
+) (int64, error) {
+	q := `
+UPDATE common_tracker_patterns
+SET
+    description = '',
+    updated_at = NOW()
+WHERE
+    id = ANY(@ids)
+`
+
+	args := pgx.StrictNamedArgs{"ids": ids}
+
+	result, err := tx.Exec(ctx, q, args)
+	if err != nil {
+		return 0, fmt.Errorf("cannot clear common tracker pattern description: %w", err)
+	}
+
+	return result.RowsAffected(), nil
+}
+
 // RequestEnrichmentByIDs arms enrichment on the given common tracker
 // patterns by stamping enrichment_requested_at, which is the only column
 // the enrichment worker claims on. It resets enrichment_attempts to 0 so

pkg/coredata/tracker_pattern.go

@@ -1406,6 +1406,49 @@ WHERE
 	return result.RowsAffected(), nil
 }
 
+// ClearDescriptionForUncategorisedByCommonTrackerPatternIDs blanks the
+// description on the uncategorised org tracker patterns linked to the
+// given common tracker patterns. It pairs with the first-party verdict on
+// the catalog row: when the catalog description is cleared because its
+// vendor attribution was wrong, the descriptions fanned out to org
+// patterns named the same stale vendor and must be cleared too - the
+// mapping worker only ever copies a description into an empty org row, it
+// never clears one. Like the mapping re-arm it is a global catalog
+// operation, so it is intentionally not tenant-scoped, and it leaves
+// excluded and user-categorised patterns untouched. The cookie_categories
+// subquery is used only for filtering. Returns the number of org patterns
+// cleared.
+func (tps *TrackerPatterns) ClearDescriptionForUncategorisedByCommonTrackerPatternIDs(
+	ctx context.Context,
+	tx pg.Tx,
+	commonIDs []gid.GID,
+) (int64, error) {
+	q := `
+UPDATE tracker_patterns
+SET
+	description = '',
+	updated_at = NOW()
+WHERE
+	common_tracker_pattern_id = ANY(@common_ids)
+	AND excluded = false
+	AND cookie_category_id IN (
+		SELECT id FROM cookie_categories WHERE kind = @uncategorised_kind
+	)
+`
+
+	args := pgx.StrictNamedArgs{
+		"common_ids":         commonIDs,
+		"uncategorised_kind": CookieCategoryKindUncategorised,
+	}
+
+	result, err := tx.Exec(ctx, q, args)
+	if err != nil {
+		return 0, fmt.Errorf("cannot clear uncategorised tracker pattern descriptions: %w", err)
+	}
+
+	return result.RowsAffected(), nil
+}
+
 // RequestMappingForUnmappedByInitiatorDomains re-arms mapping on the
 // still-unmapped org tracker patterns whose detected trackers share one
 // of the given initiator domains, so the mapping worker re-resolves them

pkg/proboctl/commontrackerpattern/mark_first_party.go

@@ -44,12 +44,14 @@ func newCmdMarkFirstParty(f *cmdutil.Factory) *cobra.Command {
 		Long: "Record the terminal FIRST_PARTY verdict on selected common tracker " +
 			"patterns: the artifact has no third party (it is the scanned site's own, " +
 			"a generic library/log key, or an extension key embedding the site origin). " +
-			"Any vendor link is cleared, and the uncategorised org tracker patterns " +
+			"Any vendor link is cleared and the now-stale description - which may name " +
-			"linked to them are remapped (org third party cleared, mapping re-armed) so " +
+			"the wrong vendor - is blanked on both the catalog row and the uncategorised " +
-			"the pipeline drops the stale vendor; because the verdict is terminal the " +
+			"org tracker patterns linked to it. Those org patterns are remapped (org " +
-			"mapping worker leaves them unattributed. User-categorised and excluded org " +
+			"third party cleared, mapping re-armed) so the pipeline drops the stale " +
-			"patterns are left untouched. Selection mirrors 'reenrich'. To re-attribute " +
+			"vendor; because the verdict is terminal the mapping worker leaves them " +
-			"a row later, use 'link' (which returns it to THIRD_PARTY).",
+			"unattributed. User-categorised and excluded org patterns are left " +
+			"untouched. Selection mirrors 'reenrich'. To re-attribute a row later, use " +
+			"'link' (which returns it to THIRD_PARTY).",
 		Args: cobra.NoArgs,
 	}
 
@@ -109,6 +111,7 @@ func newCmdMarkFirstParty(f *cmdutil.Factory) *cobra.Command {
 		var (
 			marked   int64
 			remapped int64
+			cleared  int64
 		)
 
 		if err := pgClient.WithTx(
@@ -121,6 +124,10 @@ func newCmdMarkFirstParty(f *cmdutil.Factory) *cobra.Command {
 					return err
 				}
 
+				if _, err = ps.ClearDescriptionByIDs(ctx, tx, ids); err != nil {
+					return err
+				}
+
 				var tps coredata.TrackerPatterns
 
 				remapped, err = tps.RequestMappingForUncategorisedByCommonTrackerPatternIDs(ctx, tx, ids)
@@ -128,6 +135,11 @@ func newCmdMarkFirstParty(f *cmdutil.Factory) *cobra.Command {
 					return err
 				}
 
+				cleared, err = tps.ClearDescriptionForUncategorisedByCommonTrackerPatternIDs(ctx, tx, ids)
+				if err != nil {
+					return err
+				}
+
 				return nil
 			},
 		); err != nil {
@@ -136,9 +148,10 @@ func newCmdMarkFirstParty(f *cmdutil.Factory) *cobra.Command {
 
 		_, _ = fmt.Fprintf(
 			out,
-			"Marked %d pattern(s) first-party, remapped %d uncategorised org tracker pattern(s).\n",
+			"Marked %d pattern(s) first-party, remapped %d uncategorised org tracker pattern(s), cleared %d stale org description(s).\n",
 			marked,
 			remapped,
+			cleared,
 		)
 
 		return nil