Update (#908)

* re-enable Js build/test
  * improve decoder performance
  * rewrite dictionary data in Java/Js to a shorter uncompressed form
  * improve dictionary generation tool

c/dec/bit_reader.h

@@ -108,45 +108,55 @@ static BROTLI_INLINE void BrotliFillBitWindow(
     BrotliBitReader* const br, uint32_t n_bits) {
 #if (BROTLI_64_BITS)
   if (!BROTLI_ALIGNED_READ && BROTLI_IS_CONSTANT(n_bits) && (n_bits <= 8)) {
-    if (br->bit_pos_ >= 56) {
+    uint32_t bit_pos = br->bit_pos_;
-      br->val_ >>= 56;
+    if (bit_pos >= 56) {
-      br->bit_pos_ ^= 56;  /* here same as -= 56 because of the if condition */
+      br->val_ =
-      br->val_ |= BROTLI_UNALIGNED_LOAD64LE(br->next_in) << 8;
+          (br->val_ >> 56) | (BROTLI_UNALIGNED_LOAD64LE(br->next_in) << 8);
+      br->bit_pos_ =
+          bit_pos ^ 56; /* here same as -= 56 because of the if condition */
       br->avail_in -= 7;
       br->next_in += 7;
     }
   } else if (
       !BROTLI_ALIGNED_READ && BROTLI_IS_CONSTANT(n_bits) && (n_bits <= 16)) {
-    if (br->bit_pos_ >= 48) {
+    uint32_t bit_pos = br->bit_pos_;
-      br->val_ >>= 48;
+    if (bit_pos >= 48) {
-      br->bit_pos_ ^= 48;  /* here same as -= 48 because of the if condition */
+      br->val_ =
-      br->val_ |= BROTLI_UNALIGNED_LOAD64LE(br->next_in) << 16;
+          (br->val_ >> 48) | (BROTLI_UNALIGNED_LOAD64LE(br->next_in) << 16);
+      br->bit_pos_ =
+          bit_pos ^ 48; /* here same as -= 48 because of the if condition */
       br->avail_in -= 6;
       br->next_in += 6;
     }
   } else {
-    if (br->bit_pos_ >= 32) {
+    uint32_t bit_pos = br->bit_pos_;
-      br->val_ >>= 32;
+    if (bit_pos >= 32) {
-      br->bit_pos_ ^= 32;  /* here same as -= 32 because of the if condition */
+      br->val_ = (br->val_ >> 32) |
-      br->val_ |= ((uint64_t)BROTLI_UNALIGNED_LOAD32LE(br->next_in)) << 32;
+                 (((uint64_t)BROTLI_UNALIGNED_LOAD32LE(br->next_in)) << 32);
+      br->bit_pos_ =
+          bit_pos ^ 32; /* here same as -= 32 because of the if condition */
       br->avail_in -= BROTLI_SHORT_FILL_BIT_WINDOW_READ;
       br->next_in += BROTLI_SHORT_FILL_BIT_WINDOW_READ;
     }
   }
 #else
   if (!BROTLI_ALIGNED_READ && BROTLI_IS_CONSTANT(n_bits) && (n_bits <= 8)) {
-    if (br->bit_pos_ >= 24) {
+    uint32_t bit_pos = br->bit_pos_;
-      br->val_ >>= 24;
+    if (bit_pos >= 24) {
-      br->bit_pos_ ^= 24;  /* here same as -= 24 because of the if condition */
+      br->val_ =
-      br->val_ |= BROTLI_UNALIGNED_LOAD32LE(br->next_in) << 8;
+          (br->val_ >> 24) | (BROTLI_UNALIGNED_LOAD32LE(br->next_in) << 8);
+      br->bit_pos_ =
+          bit_pos ^ 24; /* here same as -= 24 because of the if condition */
       br->avail_in -= 3;
       br->next_in += 3;
     }
   } else {
-    if (br->bit_pos_ >= 16) {
+    uint32_t bit_pos = br->bit_pos_;
-      br->val_ >>= 16;
+    if (bit_pos >= 16) {
-      br->bit_pos_ ^= 16;  /* here same as -= 16 because of the if condition */
+      br->val_ = (br->val_ >> 16) |
-      br->val_ |= ((uint32_t)BROTLI_UNALIGNED_LOAD16LE(br->next_in)) << 16;
+                 (((uint32_t)BROTLI_UNALIGNED_LOAD16LE(br->next_in)) << 16);
+      br->bit_pos_ =
+          bit_pos ^ 16; /* here same as -= 16 because of the if condition */
       br->avail_in -= BROTLI_SHORT_FILL_BIT_WINDOW_READ;
       br->next_in += BROTLI_SHORT_FILL_BIT_WINDOW_READ;
     }

java/org/brotli/dec/CompoundDictionaryTest.java

@@ -1,4 +1,4 @@
-/* Copyright 2020 Google Inc. All Rights Reserved.
+/* Copyright 2016 Google Inc. All Rights Reserved.
 
    Distributed under MIT license.
    See file LICENSE for detail or copy at https://opensource.org/licenses/MIT

java/org/brotli/dec/Decode.java

@@ -1127,13 +1127,13 @@ final class Decode {
           s.distanceCode = CMD_LOOKUP[cmdCode + 3];
           BitReader.fillBitWindow(s);
           {
-            int extraBits = insertAndCopyExtraBits & 0xFF;
+            int insertLengthExtraBits = insertAndCopyExtraBits & 0xFF;
-            s.insertLength = insertLengthOffset + BitReader.readBits(s, extraBits);
+            s.insertLength = insertLengthOffset + BitReader.readBits(s, insertLengthExtraBits);
           }
           BitReader.fillBitWindow(s);
           {
-            int extraBits = insertAndCopyExtraBits >> 8;
+            int copyLengthExtraBits = insertAndCopyExtraBits >> 8;
-            s.copyLength = copyLengthOffset + BitReader.readBits(s, extraBits);
+            s.copyLength = copyLengthOffset + BitReader.readBits(s, copyLengthExtraBits);
           }
 
           s.j = 0;

java/org/brotli/dec/SynthTest.java

@@ -2915,7 +2915,7 @@ public class SynthTest {
       (byte) 0x12, (byte) 0x00, (byte) 0x00, (byte) 0x77, (byte) 0xda, (byte) 0xcc, (byte) 0xe1,
       (byte) 0x7b, (byte) 0xfa, (byte) 0x0f
     };
-    /* This lines is added manually. */
+    /* This line is added manually. */
     char[] expected = new char[16777216]; Arrays.fill(expected, '*');
     checkSynth(
     /*

js/BUILD

@@ -23,11 +23,7 @@ closure_js_library(
 closure_js_library(
     name = "decode",
     srcs = ["decode.js"],
-    suppress = [
+    suppress = ["JSC_USELESS_BLOCK"],
-        "JSC_DUP_VAR_DECLARATION",
-        "JSC_INVALID_OCTAL_LITERAL",
-        "JSC_USELESS_BLOCK",
-    ],
     deps = [":polyfill"],
 )
 
@@ -36,6 +32,8 @@ load("@io_bazel_rules_closure//closure:defs.bzl", "closure_js_test")
 closure_js_test(
     name = "all_tests",
     srcs = ["decode_test.js"],
+    entry_points = ["decode_test.js"],
+    suppress = ["moduleLoad"],
     deps = [
         ":decode",
         ":polyfill",

js/decode_test.js

@@ -1,5 +1,11 @@
+/* Copyright 2017 Google Inc. All Rights Reserved.
+
+   Distributed under MIT license.
+   See file LICENSE for detail or copy at https://opensource.org/licenses/MIT
+*/
+import {BrotliDecode} from "./decode.js";
+const testSuite = goog.require('goog.testing.testSuite');
 goog.require('goog.testing.asserts');
-goog.require('goog.testing.jsunit');
 
 /**
  * @param {!Int8Array} bytes
@@ -19,79 +25,17 @@ function stringToBytes(str) {
   return out;
 }
 
-function testMetadata() {
+testSuite({
+  testMetadata() {
     assertEquals("", bytesToString(BrotliDecode(Int8Array.from([1, 11, 0, 42, 3]))));
-}
+  },
 
-function testEmpty() {
+  testCompoundDictionary() {
-  assertEquals("", bytesToString(BrotliDecode(Int8Array.from([6]))));
-  assertEquals("", bytesToString(BrotliDecode(Int8Array.from([0x81, 1]))));
-}
-
-function testBaseDictWord() {
-  var input = Int8Array.from([
-    0x1b, 0x03, 0x00, 0x00, 0x00, 0x00, 0x80, 0xe3, 0xb4, 0x0d, 0x00, 0x00,
-    0x07, 0x5b, 0x26, 0x31, 0x40, 0x02, 0x00, 0xe0, 0x4e, 0x1b, 0x41, 0x02
-  ]);
-  /** @type {!Int8Array} */
-  var output = BrotliDecode(input);
-  assertEquals("time", bytesToString(output));
-}
-
-function testBlockCountMessage() {
-  var input = Int8Array.from([
-    0x1b, 0x0b, 0x00, 0x11, 0x01, 0x8c, 0xc1, 0xc5, 0x0d, 0x08, 0x00, 0x22,
-    0x65, 0xe1, 0xfc, 0xfd, 0x22, 0x2c, 0xc4, 0x00, 0x00, 0x38, 0xd8, 0x32,
-    0x89, 0x01, 0x12, 0x00, 0x00, 0x77, 0xda, 0x04, 0x10, 0x42, 0x00, 0x00, 0x00
-  ]);
-  /** @type {!Int8Array} */
-  var output = BrotliDecode(input);
-  assertEquals("aabbaaaaabab", bytesToString(output));
-}
-
-function testCompressedUncompressedShortCompressedSmallWindow() {
-  var input = Int8Array.from([
-    0x21, 0xf4, 0x0f, 0x00, 0x00, 0x00, 0x00, 0x1c, 0xa7, 0x6d, 0x00, 0x00,
-    0x38, 0xd8, 0x32, 0x89, 0x01, 0x12, 0x00, 0x00, 0x77, 0xda, 0x34, 0x7b,
-    0xdb, 0x50, 0x80, 0x02, 0x80, 0x62, 0x62, 0x62, 0x62, 0x62, 0x62, 0x31,
-    0x00, 0x00, 0x00, 0x00, 0x00, 0x38, 0x4e, 0xdb, 0x00, 0x00, 0x70, 0xb0,
-    0x65, 0x12, 0x03, 0x24, 0x00, 0x00, 0xee, 0xb4, 0x11, 0x24, 0x00
-  ]);
-  /** @type {!Int8Array} */
-  var output = BrotliDecode(input);
-  assertEquals(
-    "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa" +
-    "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa" +
-    "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa" +
-    "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa" +
-    "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa" +
-    "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa" +
-    "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa" +
-    "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa" +
-    "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa" +
-    "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa" +
-    "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa" +
-    "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa" +
-    "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa" +
-    "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa" +
-    "aaaaaaaaaaaaaabbbbbbbbbb", bytesToString(output));
-}
-
-function testIntactDistanceRingBuffer0() {
-  var input = Int8Array.from([
-    0x1b, 0x0a, 0x00, 0x00, 0x00, 0x00, 0x80, 0xe3, 0xb4, 0x0d, 0x00, 0x00,
-    0x07, 0x5b, 0x26, 0x31, 0x40, 0x02, 0x00, 0xe0, 0x4e, 0x1b, 0xa1, 0x80,
-    0x20, 0x00
-  ]);
-  /** @type {!Int8Array} */
-  var output = BrotliDecode(input);
-  assertEquals("himselfself", bytesToString(output));
-}
-
-function testCompoundDictionary() {
     var txt = "kot lomom kolol slona\n";
     var dictionary = stringToBytes(txt);
     var compressed = [0xa1, 0xa8, 0x00, 0xc0, 0x2f, 0x01, 0x10, 0xc4, 0x44, 0x09, 0x00];
     assertEquals(txt.length, compressed.length * 2);
-  assertEquals(txt, bytesToString(BrotliDecode(Int8Array.from(compressed), {customDictionary: dictionary})));
+    var options = {"customDictionary": dictionary};
+    assertEquals(txt, bytesToString(BrotliDecode(Int8Array.from(compressed), options)));
   }
+});

js/polyfill.js

@@ -63,3 +63,12 @@ if (!Int8Array.prototype.fill) {
 if (!Int32Array.prototype.fill) {
   Int32Array.prototype.fill = Array.prototype.fill;
 }
+
+if (!Int8Array.prototype.slice) {
+  Object.defineProperty(Int8Array.prototype, 'slice', {
+    value: function (begin, end)
+     {
+        return new Int8Array(Array.prototype.slice.call(this, begin, end));
+     }
+  });
+}

research/dictionary_generator.cc

@@ -3,6 +3,7 @@
 #include <cstdio>
 #include <cstring>
 #include <fstream>
+#include <glob.h>
 #include <vector>
 
 #include "./deorummolae.h"
@@ -71,14 +72,13 @@ static void writeFile(const char* file, const std::string& content) {
   outfile.close();
 }
 
-static void writeSamples(char const* argv[], const std::vector<int>& pathArgs,
+static void writeSamples(const std::vector<std::string>& paths,
     const std::vector<size_t>& sizes, const uint8_t* data) {
   size_t offset = 0;
-  for (size_t i = 0; i < pathArgs.size(); ++i) {
+  for (size_t i = 0; i < paths.size(); ++i) {
-    int j = pathArgs[i];
+    const char* path = paths[i].c_str();
-    const char* file = argv[j];
     size_t sampleSize = sizes[i];
-    std::ofstream outfile(file, std::ofstream::binary);
+    std::ofstream outfile(path, std::ofstream::binary);
     outfile.write(reinterpret_cast<const char*>(data + offset),
         static_cast<std::streamsize>(sampleSize));
     outfile.close();
@@ -135,7 +135,7 @@ int main(int argc, char const* argv[]) {
 
   std::vector<uint8_t> data;
   std::vector<size_t> sizes;
-  std::vector<int> pathArgs;
+  std::vector<std::string> paths;
   size_t total = 0;
   for (int i = 1; i < argc; ++i) {
     if (argv[i] == nullptr) {
@@ -217,6 +217,9 @@ int main(int argc, char const* argv[]) {
         }
       } else if (arg1 == 's') {
         sliceLen = readInt(arg2);
+        // TODO: investigate why sliceLen == 4..5 greatly slows down
+        //               durschlag engine, but only from command line;
+        //               durschlag_runner seems to work fine with those.
         if (sliceLen < 4 || sliceLen > 256) {
           printHelp(fileName(argv[0]));
           fprintf(stderr, "Invalid option '%s'\n", argv[i]);
@@ -265,9 +268,14 @@ int main(int argc, char const* argv[]) {
       }
     }
 
-    std::string content = readFile(argv[i]);
+    glob_t resolved_paths;
+    memset(&resolved_paths, 0, sizeof(resolved_paths));
+    bool ok = true;
+    if (glob(argv[i], GLOB_TILDE, NULL, &resolved_paths) == 0) {
+      for(size_t j = 0; j < resolved_paths.gl_pathc; ++j) {
+        std::string content = readFile(resolved_paths.gl_pathv[j]);
         if (chunkLen == 0) {
-      pathArgs.push_back(i);
+          paths.emplace_back(resolved_paths.gl_pathv[j]);
           data.insert(data.end(), content.begin(), content.end());
           total += content.size();
           sizes.push_back(content.size());
@@ -286,6 +294,15 @@ int main(int argc, char const* argv[]) {
           sizes.push_back(chunk.size());
         }
       }
+    } else {
+      ok = false;
+    }
+    globfree(&resolved_paths);
+    if (!ok) exit(1);
+  }
+
+  fprintf(stderr, "Number of chunks: %zu; total size: %zu\n", sizes.size(),
+          total);
 
   bool wantDictionary = (dictionaryArg == -1);
   if (method == METHOD_DISTILL || method == METHOD_PURIFY) {
@@ -313,10 +330,10 @@ int main(int argc, char const* argv[]) {
         targetSize, sliceLen, blockSize, sizes, data.data()));
   } else if (method == METHOD_DISTILL) {
     durchschlag_distill(sliceLen, minimumPopulation, &sizes, data.data());
-    writeSamples(argv, pathArgs, sizes, data.data());
+    writeSamples(paths, sizes, data.data());
   } else if (method == METHOD_PURIFY) {
     durchschlag_purify(sliceLen, minimumPopulation, sizes, data.data());
-    writeSamples(argv, pathArgs, sizes, data.data());
+    writeSamples(paths, sizes, data.data());
   } else {
     printHelp(fileName(argv[0]));
     fprintf(stderr, "Unknown generator\n");

scripts/.travis.sh

@@ -62,8 +62,7 @@ case "$1" in
 		bazel build -c opt ...:all &&
 		cd go && bazel test -c opt ...:all && cd .. &&
 		cd java && bazel test -c opt ...:all && cd .. &&
-		echo "bazelbuild/rules_closure seems to be dead; js build is skipped" &&
+		cd js && bazel test -c opt ...:all && cd .. &&
-		echo "cd js && bazel test -c opt ...:all && cd .." &&
 		cd research && bazel build -c opt ...:all && cd ..
 		;;
 	esac