From 70b4e18b583ba1cf3060f0cdda68deabfb8cda27 Mon Sep 17 00:00:00 2001
From: "Sascha L. Teichmann" <sascha.teichmann@intevation.de>
Date: Fri, 26 Aug 2022 16:53:18 +0200
Subject: [PATCH] Treat invalid PMDs as error and stop processing if needed

*  Improve how PMD validation result is used when searching for a valid PMD.
   We now stop if no PMD can be validated against the json schema.
---
 cmd/csaf_aggregator/processor.go  |  4 ++--
 cmd/csaf_checker/processor.go     |  2 +-
 cmd/csaf_downloader/downloader.go |  4 ++--
 csaf/util.go                      | 25 ++++++++++++++++++-------
 4 files changed, 23 insertions(+), 12 deletions(-)

diff --git a/cmd/csaf_aggregator/processor.go b/cmd/csaf_aggregator/processor.go
index 5ae4f667..f08436d8 100644
--- a/cmd/csaf_aggregator/processor.go
+++ b/cmd/csaf_aggregator/processor.go
@@ -85,8 +85,8 @@ func (w *worker) locateProviderMetadata(domain string) error {
 				"Looking for provider-metadata.json of '"+domain+"': "+format+"\n", args...)
 		})
 
-	if lpmd == nil {
-		return fmt.Errorf("no provider-metadata.json found for '%s'", domain)
+	if !lpmd.Valid() {
+		return fmt.Errorf("no valid provider-metadata.json found for '%s'", domain)
 	}
 
 	w.metadataProvider = lpmd.Document
diff --git a/cmd/csaf_checker/processor.go b/cmd/csaf_checker/processor.go
index b66fd517..53993c04 100644
--- a/cmd/csaf_checker/processor.go
+++ b/cmd/csaf_checker/processor.go
@@ -1062,7 +1062,7 @@ func (p *processor) checkProviderMetadata(domain string) error {
 
 	lpmd := csaf.LoadProviderMetadataForDomain(client, domain, p.badProviderMetadata.warn)
 
-	if lpmd == nil {
+	if !lpmd.Valid() {
 		p.badProviderMetadata.error("No valid provider-metadata.json found.")
 		p.badProviderMetadata.error("STOPPING here - cannot perform other checks.")
 		return errStop
diff --git a/cmd/csaf_downloader/downloader.go b/cmd/csaf_downloader/downloader.go
index 69e3eb1e..eee9879e 100644
--- a/cmd/csaf_downloader/downloader.go
+++ b/cmd/csaf_downloader/downloader.go
@@ -93,8 +93,8 @@ func (d *downloader) download(domain string) error {
 				"Looking for provider-metadata.json of '"+domain+"': "+format+"\n", args...)
 		})
 
-	if lpmd == nil {
-		return fmt.Errorf("no provider-metadata.json found for '%s'", domain)
+	if !lpmd.Valid() {
+		return fmt.Errorf("no valid provider-metadata.json found for '%s'", domain)
 	}
 
 	base, err := url.Parse(lpmd.URL)
diff --git a/csaf/util.go b/csaf/util.go
index c66c0210..adf41151 100644
--- a/csaf/util.go
+++ b/csaf/util.go
@@ -34,6 +34,11 @@ type LoadedProviderMetadata struct {
 	Messages []string
 }
 
+// Valid returns true if the loaded document is valid.
+func (lpm *LoadedProviderMetadata) Valid() bool {
+	return lpm != nil && lpm.Document != nil && lpm.Hash != nil
+}
+
 // defaultLogging generates a logging function if given is nil.
 func defaultLogging(
 	logging func(format string, args ...interface{}),
@@ -80,15 +85,17 @@ func LoadProviderMetadataFromURL(
 
 	tee := io.TeeReader(res.Body, hash)
 
-	err = json.NewDecoder(tee).Decode(&result.Document)
+	var doc interface{}
+
+	err = json.NewDecoder(tee).Decode(&doc)
 	// Before checking the err lets check if we had the same
 	// document before. If so it will have failed parsing before.
 
-	result.Hash = hash.Sum(nil)
+	sum := hash.Sum(nil)
 
 	var key string
 	if already != nil {
-		key = string(result.Hash)
+		key = string(sum)
 		if r, ok := already[key]; ok {
 			return r
 		}
@@ -108,7 +115,7 @@ func LoadProviderMetadataFromURL(
 		return &result
 	}
 
-	switch errors, err := ValidateProviderMetadata(result.Document); {
+	switch errors, err := ValidateProviderMetadata(doc); {
 	case err != nil:
 		result.Messages = []string{
 			fmt.Sprintf("%s: Validating against JSON schema failed: %v", url, err)}
@@ -119,6 +126,10 @@ func LoadProviderMetadataFromURL(
 		for _, msg := range errors {
 			result.Messages = append(result.Messages, strings.ReplaceAll(msg, `%`, `%%`))
 		}
+	default:
+		// Only store in result if validation passed.
+		result.Document = doc
+		result.Hash = sum
 	}
 
 	storeLoaded()
@@ -164,7 +175,7 @@ func LoadProviderMetadatasFromSecurity(
 	for _, url := range urls {
 		if result := LoadProviderMetadataFromURL(
 			client, url, already, logging,
-		); result != nil {
+		); result.Valid() {
 			results = append(results, result)
 		}
 	}
@@ -224,7 +235,7 @@ func LoadProviderMetadataForDomain(
 	lg(wellknownResult, wellknownURL)
 
 	// We have a candidate.
-	if wellknownResult != nil {
+	if wellknownResult.Valid() {
 		wellknownGood = wellknownResult
 	}
 
@@ -233,7 +244,7 @@ func LoadProviderMetadataForDomain(
 	secResults := LoadProviderMetadatasFromSecurity(
 		client, secURL, already, logging)
 
-	if secResults == nil {
+	if len(secResults) == 0 {
 		logging("%s failed to load.", secURL)
 	} else {
 		// Filter out the results which are valid.