rohit/probo
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Improve tracker source and first-party cleanup

Browse Source 
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>
This commit is contained in:
Émile Ré
2026-06-18 19:37:54 +02:00
parent 3b777e985c
commit 265c56d00b
5 changed files with 97 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

27
pkg/proboctl/commontrackerpattern/mark_first_party.go
View File

@@ -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 " +
"linked to them are remapped (org third party cleared, mapping re-armed) so " +
"the pipeline drops the stale vendor; because the verdict is terminal the " +
"mapping worker leaves them 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).",
"Any vendor link is cleared and the now-stale description - which may name " +
"the wrong vendor - is blanked on both the catalog row and the uncategorised " +
"org tracker patterns linked to it. Those org patterns are remapped (org " +
"third party cleared, mapping re-armed) so the pipeline drops the stale " +
"vendor; because the verdict is terminal the mapping worker leaves them " +
"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