Sdch view for net-internals

net/base/sdch_manager.h

 // Copyright (c) 2011 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.
 
 // Provides global database of differential decompression dictionaries for the
 // SDCH filter (processes sdch enconded content).
 
 // Exactly one instance of SdchManager is built, and all references are made
 // into that collection.
 //
 // The SdchManager maintains a collection of memory resident dictionaries.  It
 // can find a dictionary (based on a server specification of a hash), store a
 // dictionary, and make judgements about what URLs can use, set, etc. a
 // dictionary.
 
 // These dictionaries are acquired over the net, and include a header
 // (containing metadata) as well as a VCDIFF dictionary (for use by a VCDIFF
 // module) to decompress data.
 
 #ifndef NET_BASE_SDCH_MANAGER_H_
 #define NET_BASE_SDCH_MANAGER_H_
 
 #include <map>
 #include <set>
 #include <string>
+#include <vector>

[Randy Smith (Not in Mondays), 2014/08/13 17:35:46]

Is this still needed?

[baranovich, 2014/08/13 19:13:48]

Done.

 
 #include "base/gtest_prod_util.h"
 #include "base/memory/ref_counted.h"
 #include "base/memory/scoped_ptr.h"
 #include "base/threading/non_thread_safe.h"
 #include "base/time/time.h"
 #include "net/base/net_export.h"
+#include "net/base/net_log.h"

[Randy Smith (Not in Mondays), 2014/08/13 17:35:45]

Can't this be in the cc file?  Maybe with a forward declaration here, though I'm
not seeing the type use that would require it.

[baranovich, 2014/08/13 19:13:48]

Done.

 #include "url/gurl.h"
 
+namespace base {
+class Value;
+class DictionaryValue;

[Randy Smith (Not in Mondays), 2014/08/13 17:35:45]

Is this still needed?

[baranovich, 2014/08/13 19:13:48]

Done.

+}
+
 namespace net {
 
 //------------------------------------------------------------------------------
 // Create a public interface to help us load SDCH dictionaries.
 // The SdchManager class allows registration to support this interface.
 // A browser may register a fetcher that is used by the dictionary managers to
 // get data from a specified URL.  This allows us to use very high level browser
 // functionality in this base (when the functionaity can be provided).
 class NET_EXPORT SdchFetcher {
  public:
@@ skipping 13 matching lines @@
   DISALLOW_COPY_AND_ASSIGN(SdchFetcher);
 };
 
 //------------------------------------------------------------------------------
 
 class NET_EXPORT SdchManager : public NON_EXPORTED_BASE(base::NonThreadSafe) {
  public:
   // A list of errors that appeared and were either resolved, or used to turn
   // off sdch encoding.
   enum ProblemCodes {
-    MIN_PROBLEM_CODE,
+    PROBLEM_CODE_OK = 0,
 
-    // Content-encoding correction problems.
-    ADDED_CONTENT_ENCODING = 1,
-    FIXED_CONTENT_ENCODING = 2,
-    FIXED_CONTENT_ENCODINGS = 3,
-
-    // Content decoding errors.
-    DECODE_HEADER_ERROR = 4,
-    DECODE_BODY_ERROR = 5,
-
-    // More content-encoding correction problems.
-    OPTIONAL_GUNZIP_ENCODING_ADDED = 6,
-
-    // Content encoding correction when we're not even tagged as HTML!?!
-    BINARY_ADDED_CONTENT_ENCODING = 7,
-    BINARY_FIXED_CONTENT_ENCODING = 8,
-    BINARY_FIXED_CONTENT_ENCODINGS = 9,
-
-    // Dictionary selection for use problems.
-    DICTIONARY_FOUND_HAS_WRONG_DOMAIN = 10,
-    DICTIONARY_FOUND_HAS_WRONG_PORT_LIST = 11,
-    DICTIONARY_FOUND_HAS_WRONG_PATH = 12,
-    DICTIONARY_FOUND_HAS_WRONG_SCHEME = 13,
-    DICTIONARY_HASH_NOT_FOUND = 14,
-    DICTIONARY_HASH_MALFORMED = 15,
-
-    // Dictionary saving problems.
-    DICTIONARY_HAS_NO_HEADER = 20,
-    DICTIONARY_HEADER_LINE_MISSING_COLON = 21,
-    DICTIONARY_MISSING_DOMAIN_SPECIFIER = 22,
-    DICTIONARY_SPECIFIES_TOP_LEVEL_DOMAIN = 23,
-    DICTIONARY_DOMAIN_NOT_MATCHING_SOURCE_URL = 24,
-    DICTIONARY_PORT_NOT_MATCHING_SOURCE_URL = 25,
-    DICTIONARY_HAS_NO_TEXT = 26,
-    DICTIONARY_REFERER_URL_HAS_DOT_IN_PREFIX = 27,
-
-    // Dictionary loading problems.
-    DICTIONARY_LOAD_ATTEMPT_FROM_DIFFERENT_HOST = 30,
-    DICTIONARY_SELECTED_FOR_SSL = 31,
-    DICTIONARY_ALREADY_LOADED = 32,
-    DICTIONARY_SELECTED_FROM_NON_HTTP = 33,
-    DICTIONARY_IS_TOO_LARGE= 34,
-    DICTIONARY_COUNT_EXCEEDED = 35,
-    DICTIONARY_ALREADY_SCHEDULED_TO_DOWNLOAD = 36,
-    DICTIONARY_ALREADY_TRIED_TO_DOWNLOAD = 37,
-
-    // Failsafe hack.
-    ATTEMPT_TO_DECODE_NON_HTTP_DATA = 40,
-
-
-    // Content-Encoding problems detected, with no action taken.
-    MULTIENCODING_FOR_NON_SDCH_REQUEST = 50,
-    SDCH_CONTENT_ENCODE_FOR_NON_SDCH_REQUEST = 51,
-
-    // Dictionary manager issues.
-    DOMAIN_BLACKLIST_INCLUDES_TARGET = 61,
-
-    // Problematic decode recovery methods.
-    META_REFRESH_RECOVERY = 70,            // Dictionary not found.
-    // defunct =  71, // Almost the same as META_REFRESH_UNSUPPORTED.
-    // defunct = 72,  // Almost the same as CACHED_META_REFRESH_UNSUPPORTED.
-    // defunct = 73,  // PASSING_THROUGH_NON_SDCH plus DISCARD_TENTATIVE_SDCH.
-    META_REFRESH_UNSUPPORTED = 74,         // Unrecoverable error.
-    CACHED_META_REFRESH_UNSUPPORTED = 75,  // As above, but pulled from cache.
-    PASSING_THROUGH_NON_SDCH = 76,  // Tagged sdch but missing dictionary-hash.
-    INCOMPLETE_SDCH_CONTENT = 77,   // Last window was not completely decoded.
-    PASS_THROUGH_404_CODE = 78,     // URL not found message passing through.
-
-    // This next report is very common, and not really an error scenario, but
-    // it exercises the error recovery logic.
-    PASS_THROUGH_OLD_CACHED = 79,   // Back button got pre-SDCH cached content.
-
-    // Common decoded recovery methods.
-    META_REFRESH_CACHED_RECOVERY = 80,  // Probably startup tab loading.
-    DISCARD_TENTATIVE_SDCH = 81,        // Server decided not to use sdch.
-
-    // Non SDCH problems, only accounted for to make stat counting complete
-    // (i.e., be able to be sure all dictionary advertisements are accounted
-    // for).
-
-    UNFLUSHED_CONTENT = 90,    // Possible error in filter chaining.
-    // defunct = 91,           // MISSING_TIME_STATS (Should never happen.)
-    CACHE_DECODED = 92,        // No timing stats recorded.
-    // defunct = 93,           // OVER_10_MINUTES (No timing stats recorded.)
-    UNINITIALIZED = 94,        // Filter never even got initialized.
-    PRIOR_TO_DICTIONARY = 95,  // We hadn't even parsed a dictionary selector.
-    DECODE_ERROR = 96,         // Something went wrong during decode.
-
-    // Problem during the latency test.
-    LATENCY_TEST_DISALLOWED = 100,  // SDCH now failing, but it worked before!
-
+#define SDCH_PROBLEM_CODE(label, value) label = value,
+#include "net/base/sdch_problem_code_list.h"
+#undef SDCH_PROBLEM_CODE
     MAX_PROBLEM_CODE  // Used to bound histogram.
   };
 
   // Use the following static limits to block DOS attacks until we implement
   // a cached dictionary evicition strategy.
   static const size_t kMaxDictionarySize;
   static const size_t kMaxDictionaryCount;
 
   // There is one instance of |Dictionary| for each memory-cached SDCH
   // dictionary.
@@ skipping 15 matching lines @@
                const std::string& client_hash,
                const GURL& url,
                const std::string& domain,
                const std::string& path,
                const base::Time& expiration,
                const std::set<int>& ports);
     ~Dictionary();
 
     const GURL& url() const { return url_; }
     const std::string& client_hash() const { return client_hash_; }
+    const std::string& domain() const { return domain_; }
+    const std::string& path() const { return path_; }
+    const base::Time& expiration() const { return expiration_; }
+    const std::set<int>& ports() const { return ports_; }
 
     // Security method to check if we can advertise this dictionary for use
     // if the |target_url| returns SDCH compressed data.
-    bool CanAdvertise(const GURL& target_url);
+    bool CanAdvertise(const GURL& target_url) const;
 
     // Security methods to check if we can establish a new dictionary with the
     // given data, that arrived in response to get of dictionary_url.

[Randy Smith (Not in Mondays), 2014/08/13 17:35:46]

Specify return value interpretation?  (And below.)

It's a bit weird that you're specifying one method (AddSdchDictionary) that
returns a bool + writes a pointer with a problem code, but returning problem
codes from all the rest of these routines.  Is there some reason you'd rather
they didn't have the same signature pattern?

[baranovich, 2014/08/13 18:38:58]

If SDCH is disabled (or SDCH over https is disabled, or format-version != 1.0
(but this should be a problem I believe)), then AddSdchDictionary should return
false (dictionary not added), although there are no such problem codes. As we
discussed earlier, three options were available:
1) Add new problem codes (means messing up with histograms)
2) Create new enum to return from AddSdchDictionary. This enum would be a subset
of SdchProblems (~10 values) + 3 codes.
3) Return boolean meaning addition success/failure and problem as an addition to
it.
I decided to go with the last option.

On 2014/08/13 17:35:46, rdsmith wrote:

[Randy Smith (Not in Mondays), 2014/08/19 19:00:53]

Right, thank you for the summary.  Thinking this through more carefully, I think
I'm good with adding new problem codes without messing with the histograms; as
long as we just add (and don't reuse old values, which the comments in the enum
should prevent) I think we're good keeping the same histogram.  And I think that
allows us to clean up the interface (i.e. make a general rule that, other than
the simplest boolean probes, every routine returns a problem code).

I think I'd like to make CanAdvertise consistent with the other Can* routines,
though.
 

-    static bool CanSet(const std::string& domain, const std::string& path,
-                       const std::set<int>& ports, const GURL& dictionary_url);
+    static ProblemCodes CanSet(const std::string& domain,
+                               const std::string& path,
+                               const std::set<int>& ports,
+                               const GURL& dictionary_url);
 
     // Security method to check if we can use a dictionary to decompress a
     // target that arrived with a reference to this dictionary.
-    bool CanUse(const GURL& referring_url);
+    ProblemCodes CanUse(const GURL& referring_url) const;
 
     // Compare paths to see if they "match" for dictionary use.
     static bool PathMatch(const std::string& path,
                           const std::string& restriction);
 
     // Compare domains to see if the "match" for dictionary use.
     static bool DomainMatch(const GURL& url, const std::string& restriction);
 
-
     // The actual text of the dictionary.
     std::string text_;
 
     // Part of the hash of text_ that the client uses to advertise the fact that
     // it has a specific dictionary pre-cached.
     std::string client_hash_;
 
     // The GURL that arrived with the text_ in a URL request to specify where
     // this dictionary may be used.
     const GURL url_;
@@ skipping 60 matching lines @@
   // Check to see if SDCH is enabled (globally), and the given URL is in a
   // supported domain (i.e., not blacklisted, and either the specific supported
   // domain, or all domains were assumed supported).  If it is blacklist, reduce
   // by 1 the number of times it will be reported as blacklisted.
   bool IsInSupportedDomain(const GURL& url);
 
   // Schedule the URL fetching to load a dictionary. This will always return
   // before the dictionary is actually loaded and added.
   // After the implied task does completes, the dictionary will have been
   // cached in memory.
-  void FetchDictionary(const GURL& request_url, const GURL& dictionary_url);
+  ProblemCodes FetchDictionary(const GURL& request_url,
+                               const GURL& dictionary_url);
 
   // Security test function used before initiating a FetchDictionary.
-  // Return true if fetch is legal.
-  bool CanFetchDictionary(const GURL& referring_url,
-                          const GURL& dictionary_url) const;
+  // Return PROBLEM_CODE_OK if fetch is legal.
+  ProblemCodes CanFetchDictionary(const GURL& referring_url,
+                                  const GURL& dictionary_url) const;
 
   // Add an SDCH dictionary to our list of availible dictionaries. This addition
   // will fail (return false) if addition is illegal (data in the dictionary is
   // not acceptable from the dictionary_url; dictionary already added, etc.).
+  // |problem| will contain related problem code.
   bool AddSdchDictionary(const std::string& dictionary_text,
-                         const GURL& dictionary_url);
+                         const GURL& dictionary_url,
+                         ProblemCodes* problem);
 
   // Find the vcdiff dictionary (the body of the sdch dictionary that appears
   // after the meta-data headers like Domain:...) with the given |server_hash|
   // to use to decompreses data that arrived as SDCH encoded content.  Check to
   // be sure the returned |dictionary| can be used for decoding content supplied
   // in response to a request for |referring_url|.
   // Return null in |dictionary| if there is no matching legal dictionary.
-  void GetVcdiffDictionary(const std::string& server_hash,
-                           const GURL& referring_url,
-                           scoped_refptr<Dictionary>* dictionary);
+  // Returns PROBLEM_CODE_OK if dictionary is not found, SDCH(-over-https) is
+  // disabled, or if matching legal dictionary exists. Otherwise returns the
+  // corresponding problem code.
+  ProblemCodes GetVcdiffDictionary(const std::string& server_hash,
+                                   const GURL& referring_url,
+                                   scoped_refptr<Dictionary>* dictionary);
 
   // Get list of available (pre-cached) dictionaries that we have already loaded
   // into memory.  The list is a comma separated list of (client) hashes per
   // the SDCH spec.
   void GetAvailDictionaryList(const GURL& target_url, std::string* list);
 
   // Construct the pair of hashes for client and server to identify an SDCH
   // dictionary.  This is only made public to facilitate unit testing, but is
   // otherwise private
   static void GenerateHash(const std::string& dictionary_text,
                            std::string* client_hash, std::string* server_hash);
 
   // For Latency testing only, we need to know if we've succeeded in doing a
   // round trip before starting our comparative tests.  If ever we encounter
   // problems with SDCH, we opt-out of the test unless/until we perform a
   // complete SDCH decoding.
   bool AllowLatencyExperiment(const GURL& url) const;
 
   void SetAllowLatencyExperiment(const GURL& url, bool enable);
 
+  base::Value* SdchInfoToValue() const;
+
  private:
+  bool IsInBlacklistedDomain(const GURL& url);
+
+  static bool IsSdchEnabledForUrl(const GURL& url);
+
   typedef std::map<std::string, int> DomainCounter;
   typedef std::set<std::string> ExperimentSet;
 
   // A map of dictionaries info indexed by the hash that the server provides.
   typedef std::map<std::string, scoped_refptr<Dictionary> > DictionaryMap;
 
   // Support SDCH compression, by advertising in headers.
   static bool g_sdch_enabled_;
 
   // Support SDCH compression for HTTPS requests and responses. When supported,
@@ skipping 19 matching lines @@
   // List of hostnames for which a latency experiment is allowed (because a
   // round trip test has recently passed).
   ExperimentSet allow_latency_experiment_;
 
   DISALLOW_COPY_AND_ASSIGN(SdchManager);
 };
 
 }  // namespace net
 
 #endif  // NET_BASE_SDCH_MANAGER_H_