Add EasyPrivacy

ad_block_client.cc

@@ -799,19 +799,20 @@ bool AdBlockClient::parse(const char *input) {
   // Otherwise it needs to be done manually via initBloomFilter and
   // initExceptionBloomFilter
   if (!bloomFilter) {
-    bloomFilter = new BloomFilter(15, 40000);
+    bloomFilter = new BloomFilter(15, 80000);
   }
   if (!exceptionBloomFilter) {
-    exceptionBloomFilter = new BloomFilter(10, 10000);
+    exceptionBloomFilter = new BloomFilter(10, 20000);
   }
   if (!hostAnchoredHashSet) {
-    // Optimized to be 1:1 with the easylist number of host anchored hosts
-    hostAnchoredHashSet = new HashSet<Filter>(5395);
+    // Optimized to be 1:1 with the easylist / easyprivacy
+    // number of host anchored hosts.
+    hostAnchoredHashSet = new HashSet<Filter>(18000);
   }
   if (!hostAnchoredExceptionHashSet) {
-    // Optimized to be 1:1 with the easylist number of host anchored
-    // exception hosts
-    hostAnchoredExceptionHashSet = new HashSet<Filter>(428);
+    // Optimized to be 1:1 with the easylist / easyprivacy
+    // number of host anchored exception hosts.
+    hostAnchoredExceptionHashSet = new HashSet<Filter>(2000);
   }
 
   const char *p = input;

lists/default.h

@@ -16,6 +16,12 @@ const std::vector<FilterList> default_lists = {
     "EasyList",
     {},
     "https://easylist.to/"
+  }, { 
+    "48010209-AD34-4DF5-A80C-3D2A7C3920C0",
+    "https://easylist.to/easylist/easyprivacy.txt",
+    "EasyPrivacy",
+    {},
+    "https://easylist.to/"
   }, {
     "200392E7-9A0F-40DF-86EB-6AF7E4071322",
     "https://raw.githubusercontent.com/brave/adblock-lists/master/ublock-unbreak.txt", // NOLINT

main.cc

@@ -88,6 +88,8 @@ int checkForClient(AdBlockClient *pClient, const char *outputPath,
 
 int main(int argc, char**argv) {
   std::string && easyListTxt = getFileContents("./test/data/easylist.txt");
+  std::string && easyPrivacyTxt =
+    getFileContents("./test/data/easyprivacy.txt");
   std::string && ublockUnblockTxt =
     getFileContents("./test/data/ublock-unbreak.txt");
   std::string && braveUnblockTxt =
@@ -100,6 +102,7 @@ int main(int argc, char**argv) {
   // Parse filter lists for adblock
   AdBlockClient adBlockClient;
   adBlockClient.parse(easyListTxt.c_str());
+  adBlockClient.parse(easyPrivacyTxt.c_str());
   adBlockClient.parse(ublockUnblockTxt.c_str());
   adBlockClient.parse(braveUnblockTxt.c_str());
   std::vector<std::string> checkVector;

perf.cc

@@ -70,6 +70,8 @@ void doSiteList(AdBlockClient *pClient, bool outputPerf) {
 int main(int argc, char**argv) {
   std::string && easyListTxt =
     getFileContents("./test/data/easylist.txt");
+  std::string && easyPrivacyTxt =
+    getFileContents("./test/data/easyprivacy.txt");
   std::string && braveUnblockTxt =
     getFileContents("./test/data/brave-unbreak.txt");
   std::string && ublockUnblockTxt =
@@ -87,6 +89,7 @@ int main(int argc, char**argv) {
 
   AdBlockClient adBlockClient;
   adBlockClient.parse(easyListTxt.c_str());
+  adBlockClient.parse(easyPrivacyTxt.c_str());
   adBlockClient.parse(ublockUnblockTxt.c_str());
   adBlockClient.parse(braveUnblockTxt.c_str());
   doSiteList(&adBlockClient, true);
@@ -110,6 +113,7 @@ int main(int argc, char**argv) {
   AdBlockClient allClient;
   allClient.enableBadFingerprintDetection();
   allClient.parse(easyListTxt.c_str());
+  allClient.parse(easyPrivacyTxt.c_str());
   allClient.parse(ublockUnblockTxt.c_str());
   allClient.parse(braveUnblockTxt.c_str());
   allClient.parse(spam404MainBlacklistTxt.c_str());

scripts/check.js

@@ -20,6 +20,8 @@
  *   node scripts/check.js  --host www.cnet.com --list ./test/data/sitelist.txt
  * Checking a list of URLS with discovery:
  *  node scripts/check.js  --host www.cnet.com --list ./test/data/sitelist.txt --discover
+ * Get stats for a particular adblock list:
+ *   node scripts/check.js  --uuid 67F880F5-7602-4042-8A3D-01481FD7437A --stats
 */
 const commander = require('commander')
 const {makeAdBlockClientFromListUUID, makeAdBlockClientFromDATFile, makeAdBlockClientFromListURL, makeAdBlockClientFromString, makeAdBlockClientFromFilePath, readSiteList} = require('../lib/util')
@@ -38,13 +40,14 @@ commander
   .option('-o, --output [output]', 'Optionally saves a DAT file')
   .option('-L --list [list]', 'Filename for list of sites to check')
   .option('-D --discover', 'If speciied does filter discovery for matched filter')
+  .option('-s --stats', 'If speciied outputs parsing stats')
   .option('-C, --cache', 'Optionally cache results and use cached results')
   .option('-O, --filter-option [filterOption]', 'Filter option to use', filterStringToFilterOption, FilterOptions.noFilterOption)
   .parse(process.argv)
 
 let p = Promise.reject('Usage: node check.js --location <location> --host <host> [--uuid <uuid>]')
 
-if (commander.host && (commander.location || commander.list)) {
+if (commander.host && (commander.location || commander.list) || commander.stats) {
   p.catch(() => {})
   if (commander.uuid) {
     p = makeAdBlockClientFromListUUID(commander.uuid)
@@ -64,7 +67,10 @@ if (commander.host && (commander.location || commander.list)) {
 }
 
 p.then((adBlockClient) => {
-  console.log('Parsing stats:', adBlockClient.getParsingStats())
+  if (commander.stats) {
+    console.log('Parsing stats:', adBlockClient.getParsingStats())
+    return
+  }
   if (commander.location) {
     console.log('params:', commander.location, commander.filterOption, commander.host)
     if (commander.discover) {

test/js/filterListTest.js

@@ -9,7 +9,7 @@ const {adBlockLists} = require('../..')
 describe('adBlockLists', function () {
   describe('default', function () {
     it('contains 4 default lists', function () {
-      assert.equal(adBlockLists.default.length, 4)
+      assert.equal(adBlockLists.default.length, 5)
     })
     it('has uuid property', function () {
       adBlockLists.default.forEach((list) => {