feat: hash uniformity for base digests

subzey · web-flow · commit 451858b0bb33 · 2021-11-11T18:37:22.000+03:00
diff --git a/lib/getHashDigest.js b/lib/getHashDigest.js
@@ -11,34 +11,46 @@ const baseEncodeTables = {
   64: "0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ-_",
 };
 
-function encodeBufferToBase(buffer, base) {
+/**
+ * @param {Uint32Array} uint32Array Treated as a long base-0x100000000 number, little endian
+ * @param {number} divisor The divisor
+ * @return {number} Modulo (remainder) of the division
+ */
+function divmod32(uint32Array, divisor) {
+  let carry = 0;
+  for (let i = uint32Array.length - 1; i >= 0; i--) {
+    const value = carry * 0x100000000 + uint32Array[i];
+    carry = value % divisor;
+    uint32Array[i] = Math.floor(value / divisor);
+  }
+  return carry;
+}
+
+function encodeBufferToBase(buffer, base, length) {
   const encodeTable = baseEncodeTables[base];
 
   if (!encodeTable) {
     throw new Error("Unknown encoding base" + base);
   }
 
-  const readLength = buffer.length;
-  const Big = require("big.js");
+  // Input bits are only enough to generate this many characters
+  const limit = Math.ceil((buffer.length * 8) / Math.log2(base));
+  length = Math.min(length, limit);
 
-  Big.RM = Big.DP = 0;
+  // Most of the crypto digests (if not all) has length a multiple of 4 bytes.
+  // Fewer numbers in the array means faster math.
+  const uint32Array = new Uint32Array(Math.ceil(buffer.length / 4));
 
-  let b = new Big(0);
-
-  for (let i = readLength - 1; i >= 0; i--) {
-    b = b.times(256).plus(buffer[i]);
-  }
+  // Make sure the input buffer data is copied and is not mutated by reference.
+  // divmod32() would corrupt the BulkUpdateDecorator cache otherwise.
+  buffer.copy(Buffer.from(uint32Array.buffer));
 
   let output = "";
 
-  while (b.gt(0)) {
-    output = encodeTable[b.mod(base)] + output;
-    b = b.div(base);
+  for (let i = 0; i < length; i++) {
+    output = encodeTable[divmod32(uint32Array, base)] + output;
   }
 
-  Big.DP = 20;
-  Big.RM = 1;
-
   return output;
 }
 
@@ -110,10 +122,7 @@ function getHashDigest(buffer, algorithm, digestType, maxLength) {
     digestType === "base58" ||
     digestType === "base62"
   ) {
-    return encodeBufferToBase(hash.digest(), digestType.substr(4)).substr(
-      0,
-      maxLength
-    );
+    return encodeBufferToBase(hash.digest(), digestType.substr(4), maxLength);
   } else {
     return hash.digest(digestType || "hex").substr(0, maxLength);
   }
diff --git a/package.json b/package.json
@@ -3,9 +3,7 @@
   "version": "3.1.3",
   "author": "Tobias Koppers @sokra",
   "description": "utils for webpack loaders",
-  "dependencies": {
-    "big.js": "^6.1.1"
-  },
+  "dependencies": {},
   "scripts": {
     "lint": "prettier --list-different . && eslint .",
     "pretest": "yarn lint",
diff --git a/test/getHashDigest.test.js b/test/getHashDigest.test.js
@@ -11,8 +11,8 @@ describe("getHashDigest()", () => {
     ["abc\\0💩", "xxhash64", "hex", undefined, "86733ec125b93904"],
     ["abc\\0💩", "xxhash64", "base64", undefined, "hnM+wSW5OQQ="],
     ["abc\\0♥", "xxhash64", "base64", undefined, "S5o0KX3APSA="],
-    ["abc\\0💩", "xxhash64", "base52", undefined, "cfByjQcJZIU"],
-    ["abc\\0♥", "xxhash64", "base52", undefined, "qdLyAQjLlod"],
+    ["abc\\0💩", "xxhash64", "base52", undefined, "acfByjQcJZIU"],
+    ["abc\\0♥", "xxhash64", "base52", undefined, "aqdLyAQjLlod"],
 
     ["test string", "md4", "hex", 4, "2e06"],
     ["test string", "md4", "base64", undefined, "Lgbt1PFiMmjFpRcw2KCyrw=="],
@@ -34,7 +34,8 @@ describe("getHashDigest()", () => {
     ],
     ["test string", "md5", "base52", undefined, "dJnldHSAutqUacjgfBQGLQx"],
     ["test string", "md5", "base64", undefined, "b421md6Yb6t6IWJbeRZYnA=="],
-    ["test string", "md5", "base26", 6, "bhtsgu"],
+    ["test string", "md5", "base26", undefined, "bhtsgujtzvmjtgtzlqvubqggbvgx"],
+    ["test string", "md5", "base26", 6, "ggbvgx"],
     ["abc\\0♥", "md5", "hex", undefined, "2e897b64f8050e66aff98d38f7a012c5"],
     ["abc\\0💩", "md5", "hex", undefined, "63ad5b3d675c5890e0c01ed339ba0187"],
     ["abc\\0💩", "md5", "base64", undefined, "Y61bPWdcWJDgwB7TOboBhw=="],
@@ -79,3 +80,46 @@ describe("getHashDigest()", () => {
     );
   });
 });
+
+function testDistribution(digestType, length, tableSize, iterations) {
+  const lowerBound = Math.round(iterations / 2);
+  const upperBound = Math.round(iterations * 2);
+
+  const stats = [];
+  for (let i = tableSize * iterations; i-- > 0; ) {
+    const generatedString = loaderUtils.getHashDigest(
+      `Some input #${i}`,
+      undefined,
+      digestType,
+      length
+    );
+
+    for (let pos = 0; pos < length; pos++) {
+      const char = generatedString[pos];
+      stats[pos] = stats[pos] || {};
+      stats[pos][char] = (stats[pos][char] || 0) + 1;
+    }
+  }
+
+  for (let pos = 0; pos < length; pos++) {
+    const chars = Object.keys(stats[pos]).sort();
+    test(`distinct chars at position ${pos}`, () => {
+      expect(chars.length).toBe(tableSize);
+    });
+    for (const char of chars) {
+      test(`occurences of char "${char}" at position ${pos} should be around ${iterations}`, () => {
+        expect(stats[pos][char]).toBeLessThanOrEqual(upperBound);
+        expect(stats[pos][char]).toBeGreaterThanOrEqual(lowerBound);
+      });
+    }
+  }
+}
+
+describe("getHashDigest() char distribution", () => {
+  describe("should be uniform for base62", () => {
+    testDistribution("base62", 8, 62, 100);
+  });
+  describe("should be uniform for base26", () => {
+    testDistribution("base26", 8, 26, 100);
+  });
+});
diff --git a/yarn.lock b/yarn.lock
@@ -865,11 +865,6 @@ bcrypt-pbkdf@^1.0.0:
   dependencies:
     tweetnacl "^0.14.3"
 
-big.js@^6.1.1:
-  version "6.1.1"
-  resolved "https://registry.yarnpkg.com/big.js/-/big.js-6.1.1.tgz#63b35b19dc9775c94991ee5db7694880655d5537"
-  integrity sha512-1vObw81a8ylZO5ePrtMay0n018TcftpTA5HFKDaSuiUDBo8biRBtjIobw60OpwuvrGk+FsxKamqN4cnmj/eXdg==
-
 brace-expansion@^1.1.7:
   version "1.1.11"
   resolved "https://registry.yarnpkg.com/brace-expansion/-/brace-expansion-1.1.11.tgz#3c7fcbf529d87226f3d2f52b966ff5271eb441dd"
