Load Multiple Blacklists

Add the ability to load multiple text blacklists into the binary blacklist representation.

NOTE: Included Binary file means this change expects to fail on the trybots
in blacklist_io_test.cc:103. All other checks must pass.

BUG=16932
TEST=BlacklistIO*

Review URL: http://codereview.chromium.org/159199

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@21841 0039d316-1c4b-4281-b951-d872f2087c98

9 files changed, 134 insertions, 38 deletions

diff --git a/chrome/browser/privacy_blacklist/blacklist.cc b/chrome/browser/privacy_blacklist/blacklist.cc
index a234bf2..0be3931 100644
--- a/chrome/browser/privacy_blacklist/blacklist.cc
+++ b/chrome/browser/privacy_blacklist/blacklist.cc
@@ -22,6 +22,23 @@ const char* const cookie_headers[2] = { "cookie", "set-cookie" };
 
 }  // namespace
 
+const unsigned int Blacklist::kBlockAll = 1;
+const unsigned int Blacklist::kDontSendCookies = 1 << 1;
+const unsigned int Blacklist::kDontStoreCookies = 1 << 2;
+const unsigned int Blacklist::kDontPersistCookies = 1 << 3;
+const unsigned int Blacklist::kDontSendReferrer = 1 << 4;
+const unsigned int Blacklist::kDontSendUserAgent = 1 << 5;
+const unsigned int Blacklist::kBlockByType = 1 << 6;
+const unsigned int Blacklist::kBlockUnsecure = 1 << 7;
+const unsigned int Blacklist::kBlockRequest = kBlockAll | kBlockUnsecure;
+const unsigned int Blacklist::kBlockResponse = kBlockByType;
+const unsigned int Blacklist::kModifySentHeaders =
+    kDontSendCookies | kDontSendUserAgent | kDontSendReferrer;
+const unsigned int Blacklist::kModifyReceivedHeaders =
+    kDontPersistCookies | kDontStoreCookies;
+const unsigned int Blacklist::kFilterByHeaders =
+    kModifyReceivedHeaders | kBlockByType;
+
 // Value is not important, here just that the object has an address.
 const void* const Blacklist::kRequestDataKey = 0;
 
diff --git a/chrome/browser/privacy_blacklist/blacklist.h b/chrome/browser/privacy_blacklist/blacklist.h
index 4c36d08..a4b6b9ba 100644
--- a/chrome/browser/privacy_blacklist/blacklist.h
+++ b/chrome/browser/privacy_blacklist/blacklist.h
@@ -32,24 +32,21 @@ class FilePath;
 class Blacklist {
  public:
   // Filter attributes (more to come):
-  static const unsigned int kBlockAll = 1;
-  static const unsigned int kDontSendCookies = 1 << 1;
-  static const unsigned int kDontStoreCookies = 1 << 2;
-  static const unsigned int kDontPersistCookies = 1 << 3;
-  static const unsigned int kDontSendReferrer = 1 << 4;
-  static const unsigned int kDontSendUserAgent = 1 << 5;
-  static const unsigned int kBlockByType = 1 << 6;
-  static const unsigned int kBlockUnsecure = 1 << 7;
+  static const unsigned int kBlockAll;
+  static const unsigned int kDontSendCookies;
+  static const unsigned int kDontStoreCookies;
+  static const unsigned int kDontPersistCookies;
+  static const unsigned int kDontSendReferrer;
+  static const unsigned int kDontSendUserAgent;
+  static const unsigned int kBlockByType;
+  static const unsigned int kBlockUnsecure;
 
   // Aggregate filter types:
-  static const unsigned int kBlockRequest = kBlockAll | kBlockUnsecure;
-  static const unsigned int kBlockResponse = kBlockByType;
-  static const unsigned int kModifySentHeaders =
-      kDontSendCookies | kDontSendUserAgent | kDontSendReferrer;
-  static const unsigned int kModifyReceivedHeaders =
-      kDontPersistCookies | kDontStoreCookies;
-  static const unsigned int kFilterByHeaders = kModifyReceivedHeaders |
-      kBlockByType;
+  static const unsigned int kBlockRequest;
+  static const unsigned int kBlockResponse;
+  static const unsigned int kModifySentHeaders;
+  static const unsigned int kModifyReceivedHeaders;
+  static const unsigned int kFilterByHeaders;
 
   // Key used to access data attached to URLRequest objects.
   static const void* const kRequestDataKey;
diff --git a/chrome/browser/privacy_blacklist/blacklist_io.h b/chrome/browser/privacy_blacklist/blacklist_io.h
index fd19ae6..4971d35 100644
--- a/chrome/browser/privacy_blacklist/blacklist_io.h
+++ b/chrome/browser/privacy_blacklist/blacklist_io.h
@@ -37,6 +37,7 @@ class BlacklistIO {
   std::list<Blacklist::Provider*> providers_;
 
   FRIEND_TEST(BlacklistIOTest, Generic);
+  FRIEND_TEST(BlacklistIOTest, Combine);
   DISALLOW_COPY_AND_ASSIGN(BlacklistIO);
 };
 
diff --git a/chrome/browser/privacy_blacklist/blacklist_io_unittest.cc b/chrome/browser/privacy_blacklist/blacklist_io_unittest.cc
index 6899321..cd43449 100644
--- a/chrome/browser/privacy_blacklist/blacklist_io_unittest.cc
+++ b/chrome/browser/privacy_blacklist/blacklist_io_unittest.cc
@@ -13,17 +13,17 @@
 
 TEST(BlacklistIOTest, Generic) {
   // Testing data path.
-  std::wstring data_dir;
+  FilePath data_dir;
   PathService::Get(chrome::DIR_TEST_DATA, &data_dir);
 
-  std::wstring input(data_dir);
-  file_util::AppendToPath(&input, L"blacklist_small.pbl");
+  FilePath input =
+      data_dir.Append(FilePath::FromWStringHack(L"blacklist_small.pbl"));
 
-  std::wstring expected(data_dir);
-  file_util::AppendToPath(&expected, L"blacklist_small.pbr");
+  FilePath expected =
+      data_dir.Append(FilePath::FromWStringHack(L"blacklist_small.pbr"));
 
   BlacklistIO io;
-  EXPECT_TRUE(io.Read(FilePath::FromWStringHack(input)));
+  EXPECT_TRUE(io.Read(input));
   const std::list<Blacklist::Entry*>& blacklist = io.blacklist();
   EXPECT_EQ(5U, blacklist.size());
 
@@ -38,10 +38,68 @@ TEST(BlacklistIOTest, Generic) {
   EXPECT_EQ("Sample", io.providers().front()->name());
   EXPECT_EQ("http://www.google.com", io.providers().front()->url());
 
-  std::wstring output;
+  FilePath output;
   PathService::Get(base::DIR_TEMP, &output);
-  file_util::AppendToPath(&output, L"blacklist_small.pbr");
-  CHECK(io.Write(FilePath::FromWStringHack(output)));
+  output = output.Append(FilePath::FromWStringHack(L"blacklist_small.pbr"));
+  CHECK(io.Write(output));
+  EXPECT_TRUE(file_util::ContentsEqual(output, expected));
+  EXPECT_TRUE(file_util::Delete(output, false));
+}
+
+TEST(BlacklistIOTest, Combine) {
+  // Testing data path.
+  FilePath data_dir;
+  PathService::Get(chrome::DIR_TEST_DATA, &data_dir);
+  data_dir = data_dir.Append(FilePath::FromWStringHack(L"blacklist_samples"));
+
+  FilePath input1 =
+      data_dir.Append(FilePath::FromWStringHack(L"annoying_ads.pbl"));
+
+  FilePath input2 =
+    data_dir.Append(FilePath::FromWStringHack(L"block_flash.pbl"));
+
+  FilePath input3 =
+    data_dir.Append(FilePath::FromWStringHack(L"session_cookies.pbl"));
+
+  BlacklistIO io;
+  EXPECT_TRUE(io.Read(input1));
+  EXPECT_TRUE(io.Read(input2));
+  EXPECT_TRUE(io.Read(input3));
+
+  const std::list<Blacklist::Entry*>& blacklist = io.blacklist();
+  EXPECT_EQ(5U, blacklist.size());
+
+  std::list<Blacklist::Entry*>::const_iterator i = blacklist.begin();
+  EXPECT_EQ(Blacklist::kBlockAll, (*i)->attributes());
+  EXPECT_EQ("annoying.ads.tv/@", (*i++)->pattern());
+  EXPECT_EQ(Blacklist::kBlockAll, (*i)->attributes());
+  EXPECT_EQ("@/annoying/120x600.jpg", (*i++)->pattern());
+  EXPECT_EQ(Blacklist::kBlockAll, (*i)->attributes());
+  EXPECT_EQ("@/annoying_ads/@", (*i++)->pattern());
+  EXPECT_EQ(Blacklist::kBlockByType, (*i)->attributes());
+  EXPECT_EQ("@", (*i++)->pattern());
+  EXPECT_EQ(Blacklist::kDontPersistCookies, (*i)->attributes());
+  EXPECT_EQ("@", (*i++)->pattern());
+
+  const std::list<Blacklist::Provider*>& providers = io.providers();
+  EXPECT_EQ(3U, providers.size());
+
+  std::list<Blacklist::Provider*>::const_iterator j = providers.begin();
+  EXPECT_EQ("AnnoyingAds", (*j)->name());
+  EXPECT_EQ("http://www.ads.tv", (*j++)->url());
+  EXPECT_EQ("BlockFlash", (*j)->name());
+  EXPECT_EQ("http://www.google.com", (*j++)->url());
+  EXPECT_EQ("SessionCookies", (*j)->name());
+  EXPECT_EQ("http://www.google.com", (*j++)->url());
+
+  FilePath output;
+  PathService::Get(base::DIR_TEMP, &output);
+  output = output.Append(FilePath::FromWStringHack(L"combine3.pbr"));
+
+  FilePath expected =
+      data_dir.Append(FilePath::FromWStringHack(L"combine3.pbr"));
+
+  CHECK(io.Write(output));
   EXPECT_TRUE(file_util::ContentsEqual(output, expected));
   EXPECT_TRUE(file_util::Delete(output, false));
 }
diff --git a/chrome/test/data/blacklist_samples/annoying_ads.pbl b/chrome/test/data/blacklist_samples/annoying_ads.pbl
new file mode 100644
index 0000000..b9dc8d2
--- /dev/null
+++ b/chrome/test/data/blacklist_samples/annoying_ads.pbl
@@ -0,0 +1,12 @@
+[Chromium::PrivacyBlacklist]
+|Name: AnnoyingAds
+|URL: http://www.ads.tv
+
+# Block Ads by servers
+annoying.ads.tv/@ => kBlockAll
+
+# Block Ads by name
+@/annoying/120x600.jpg => kBlockAll
+
+# Block Ads by path
+@/annoying_ads/@ => kBlockAll
diff --git a/chrome/test/data/blacklist_samples/block_flash.pbl b/chrome/test/data/blacklist_samples/block_flash.pbl
new file mode 100644
index 0000000..2373f30
--- /dev/null
+++ b/chrome/test/data/blacklist_samples/block_flash.pbl
@@ -0,0 +1,5 @@
+[Chromium::PrivacyBlacklist]
+|Name: BlockFlash
+|URL: http://www.google.com
+
+@ => kBlockByType(application/x-shockwave-flash)
diff --git a/chrome/test/data/blacklist_samples/combine3.pbr b/chrome/test/data/blacklist_samples/combine3.pbr
new file mode 100644
index 0000000..30ef556
--- /dev/null
+++ b/chrome/test/data/blacklist_samples/combine3.pbr
diff --git a/chrome/test/data/blacklist_samples/session_cookies.pbl b/chrome/test/data/blacklist_samples/session_cookies.pbl
new file mode 100644
index 0000000..f83727b
--- /dev/null
+++ b/chrome/test/data/blacklist_samples/session_cookies.pbl
@@ -0,0 +1,6 @@
+[Chromium::PrivacyBlacklist]
+|Name: SessionCookies
+|URL: http://www.google.com
+
+# Block all Flash media everywhere.
+@ =>  kDontPersistCookies
diff --git a/chrome/tools/pbl_tool/pbl_tool.cc b/chrome/tools/pbl_tool/pbl_tool.cc
index 355c7ea..2e8c037 100644
--- a/chrome/tools/pbl_tool/pbl_tool.cc
+++ b/chrome/tools/pbl_tool/pbl_tool.cc
@@ -24,8 +24,8 @@
 namespace {
 
 int PrintUsage(int argc, ICHAR* argv[]) {
-  ICERR << "Usage: " << argv[0] << " <source> <target>\n"
-           "       <source> is the text blacklist (.pbl) to load.\n"
+  ICERR << "Usage: " << argv[0] << " <source>... <target>\n"
+           "       <source> are text blacklists (.pbl) to load.\n"
            "       <target> is the binary output blacklist repository.\n\n"
            "Adds all entries from <source> to <target>.\n"
            "Creates <target> if it does not exist.\n";
@@ -40,18 +40,18 @@ int IMAIN(int argc, ICHAR* argv[]) {
   if (argc < 3)
     return PrintUsage(argc, argv);
 
-  FilePath input(argv[1]);
-  FilePath output(argv[2]);
-
   BlacklistIO io;
-  if (io.Read(input)) {
-    if (io.Write(output)) {
-      return 0;
-    } else {
-      ICERR << "Error writing output file " << argv[2] << "\n";
+  for (int current = 1; current < argc-1; ++current) {
+    FilePath input(argv[current]);
+    if (!io.Read(input)) {
+      ICERR << "Error reading input file " << argv[current] << "\n";
+      return -1;
     }
-  } else {
-    ICERR << "Error reading input file " << argv[1] << "\n";
   }
-  return -1;
+
+  FilePath output(argv[argc-1]);
+  if (!io.Write(output))
+    ICERR << "Error writing output file " << argv[2] << "\n";
+
+  return 0;
 }