internal/postgres: allow matching exact prefix with exact:

Currently, the package exclusion process is case insensitive, because certain hosts such as GitHub are case insensitive.

The only way to exclude a case sensitive version without also excluding all case insensitive matches is to add the prefix to caseSensitiveExcludedPaths in internal/frontend/details.go.

This CL implements a new approach to excluding case sensitive paths: it enables specifying an "exact:" prefix in excluded.txt that will force matching the pattern exactly, instead of with case insensitivity.

For golang/go#77822

Change-Id: I123ad2e13ff6673084006c6d51377308d95230ce
Reviewed-on: https://go-review.googlesource.com/c/pkgsite/+/761521
kokoro-CI: kokoro <noreply+kokoro@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Ethan Lee <ethanalee@google.com>

3 files changed, 29 insertions, 12 deletions

diff --git a/internal/frontend/details.go b/internal/frontend/details.go
index 27b40f05..f1438de4 100644
--- a/internal/frontend/details.go
+++ b/internal/frontend/details.go
@@ -103,6 +103,8 @@ func checkExcluded(ctx context.Context, ds internal.DataSource, fullPath, versio
 
 // Paths to exclude if they match exactly.
 // These are very rare, so it's simpler to hardcode them rather than use the DB.
+// TODO(mkalil): Remove once go.dev/cl/761521, which provides an alternative
+// implementation for case sensitive exclusions, is merged.
 var caseSensitiveExcludedPaths = map[string]bool{
 	"github.com/ibm/sarama":      true, // https://go.dev/issue/71342
 	"github.com/burntsushi/toml": true, // https://go.dev/issue/68357
diff --git a/internal/postgres/excluded.go b/internal/postgres/excluded.go
index 176b9c7e..6225c6b3 100644
--- a/internal/postgres/excluded.go
+++ b/internal/postgres/excluded.go
@@ -31,22 +31,28 @@ func (db *DB) IsExcluded(ctx context.Context, path, version string) bool {
 
 func excludes(pattern, path, version string) bool {
 	// Certain hosts (such as GitHub) are case insensitive.
-	// Therefore, we err on the side of insensitive exclusions.
+	// By default, we err on the side of insensitive exclusions.
+	// If the pattern starts with "exact:", we match exactly (case sensitive).
+	isExact := strings.HasPrefix(pattern, "exact:")
+	if isExact {
+		pattern = pattern[len("exact:"):]
+	}
 
-	// Patterns with "@" must match the full path (case insensitively).
 	mod, ver, found := strings.Cut(pattern, "@")
 	if found {
+		if isExact {
+			return mod == path && ver == version
+		}
 		return strings.EqualFold(mod, path) && ver == version
 	}
-	// Patterns without "@" can match the full path or be a componentwise prefix
-	// (case insensitively).
-	if strings.EqualFold(pattern, path) {
-		return true
-	}
-	if !strings.HasSuffix(pattern, "/") {
-		pattern += "/"
+
+	pattern = strings.TrimSuffix(pattern, "/")
+
+	if isExact {
+		return pattern == path || strings.HasPrefix(path, pattern+"/")
 	}
-	return strings.HasPrefix(strings.ToLower(path), strings.ToLower(pattern))
+	return strings.EqualFold(pattern, path) ||
+		strings.HasPrefix(strings.ToLower(path), strings.ToLower(pattern)+"/")
 }
 
 // InsertExcludedPattern inserts pattern into the excluded_prefixes table.
diff --git a/internal/postgres/excluded_test.go b/internal/postgres/excluded_test.go
index 13206a85..e8050aee 100644
--- a/internal/postgres/excluded_test.go
+++ b/internal/postgres/excluded_test.go
@@ -15,7 +15,7 @@ func TestIsExcluded(t *testing.T) {
 	defer release()
 	ctx := context.Background()
 
-	for _, pat := range []string{"bad", "badslash/", "baddy@v1.2.3", "github.com/bad"} {
+	for _, pat := range []string{"bad", "badslash/", "baddy@v1.2.3", "github.com/bad", "exact:github.com/CASE/go", "exact:github.com/PACKAGE@v1.0.0"} {
 		if err := testDB.InsertExcludedPattern(ctx, pat, "someone", "because"); err != nil {
 			t.Fatal(err)
 		}
@@ -31,7 +31,7 @@ func TestIsExcluded(t *testing.T) {
 		{"badness", "", false},
 		{"bad/ness", "", true},
 		{"bad.com/foo", "", false},
-		{"badslash", "", false},
+		{"badslash", "", true},
 		{"badslash/more", "", true},
 		{"badslash/more", "v1.2.3", true},
 		{"baddys", "v1.2.3", false},
@@ -48,6 +48,15 @@ func TestIsExcluded(t *testing.T) {
 		{"github.com/bad/repo", "", true},
 		{"github.com/bad/Repo", "", true},
 		{"github.com/Bad/repo", "", true},
+
+		// tests for exact: prefix
+		{"github.com/CASE/go", "", true},
+		{"github.com/case/go", "", false},
+		{"github.com/CASE/go/sub", "", true},
+		{"github.com/case/go/sub", "", false},
+		{"github.com/PACKAGE", "v1.0.0", true},
+		{"github.com/package", "v1.0.0", false},
+		{"github.com/PACKAGE", "v1.0.1", false},
 	} {
 		got := testDB.IsExcluded(ctx, test.path, test.version)
 		if got != test.want {