Switch Safe Browsing's metadata from string to struct.

components/safe_browsing_db/util.h

 // Copyright (c) 2015 The Chromium Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 //
 // Utilities for the SafeBrowsing DB code.
 
 #ifndef COMPONENTS_SAFE_BROWSING_DB_UTIL_H_
 #define COMPONENTS_SAFE_BROWSING_DB_UTIL_H_
 
 #include <stdint.h>
@@ skipping 35 matching lines @@
 
   // Url detected by the client-side malware IP list. This IP list is part
   // of the client side detection model.
   SB_THREAT_TYPE_CLIENT_SIDE_MALWARE_URL,
 
   // Url leads to a blacklisted resource script. Note that no warnings should be
   // shown on this threat type, but an incident report might be sent.
   SB_THREAT_TYPE_BLACKLISTED_RESOURCE,
 };
 
+// Metadata that indicates what kind of URL match this is.
+enum class ThreatPatternType {
+  NONE,          // Pattern type didn't appear in the metadata
+  LANDING,       // The match is a landing page
+  DISTRIBUTION,  // The match is a distribution page
+};
+
+// Metadata that was returned by a GetFullHash call. This is the parsed version
+// of the PB (from Pver3, or Pver4 local) or JSON (from Pver4 via GMSCore).
+// Some fields are only applicable to certain lists.
+struct ThreatMetadata {

[mattm, 2016/02/26 02:32:23]

Should SBThreatType be in here too?

[Nathan Parker, 2016/02/26 18:54:48]

I considered that but since that would add SBThreatType to SBFullHashResult, it
would be a larger refactoring.  Some places that create SBFullHashResult's don't
have a SBThreatType to add to it, so we'd need to ensure that users of it
weren't expecting the value to be populated.

I think that's a good idea for simplicity, but should be a separate CL.

[mattm, 2016/02/26 23:23:19]

sgtm. I'm fine having that in a separate CL. Or not doing it if it turns out to
make some things confusing/error prone.

+  ThreatMetadata();
+  ~ThreatMetadata();
+
+  // Type of blacklisted page. Used on malware and UwS lists.
+  // This will be NONE if it wasn't present in the reponse.
+  ThreatPatternType threat_pattern_type;
+
+  // List of permissions blocked. Used with threat_type API_ABUSE.
+  // This will be empty if it wasn't present in the response.
+  std::vector<std::string> api_permissions;
+
+  // Opaque base64 string used for user-population experiments in pver4.
+  // This will be empty if it wasn't present in the response.
+  std::string population_id;
+
+  // This is the only field currently populated, and it'll be removed
+  // when the others are used.
+  // TODO(nparker): Remove this as part of crbug/589610.
+  std::string raw_metadata;
+
+};
+
 

[mattm, 2016/02/26 02:32:23]

nit: extra newline

[Nathan Parker, 2016/02/26 18:54:48]

Done.

 // A truncated hash's type.
 typedef uint32_t SBPrefix;
 
 // A full hash.
 union SBFullHash {
   char full_hash[32];
   SBPrefix prefix;
 };
 
 // Used when we get a gethash response.
 struct SBFullHashResult {
   SBFullHash hash;
   // TODO(shess): Refactor to allow ListType here.
   int list_id;
-  std::string metadata;
+  ThreatMetadata metadata;
   // Used only for V4 results. The cache lifetime for this result. The response
   // must not be cached for more than this duration to avoid false positives.
   base::TimeDelta cache_duration;
 };
 
 // Caches individual response from GETHASH request.
 struct SBCachedFullHashResult {
   SBCachedFullHashResult();
   explicit SBCachedFullHashResult(const base::Time& in_expire_after);
   SBCachedFullHashResult(const SBCachedFullHashResult& other);
@@ skipping 97 matching lines @@
 
 // Given a URL, returns all the paths we need to check.
 void GeneratePathsToCheck(const GURL& url, std::vector<std::string>* paths);
 
 // Given a URL, returns all the patterns we need to check.
 void GeneratePatternsToCheck(const GURL& url, std::vector<std::string>* urls);
 
 }  // namespace safe_browsing
 
 #endif  // COMPONENTS_SAFE_BROWSING_DB_UTIL_H_