Make base32 handling less forgiving

[Pisces] / binary.js

diff --git a/binary.js b/binary.js
index 362161bb8e83bc4268d7cfcf654ccb0600dc15f0..5ee2bda874ae5fcf84a577064c0f8e0c41770895 100644 (file)
--- a/binary.js
+++ b/binary.js
@@ -106,6 +106,174 @@ const bufferFromArgs = ($, $s) =>
           : `${$}`,
       );
 
           : `${$}`,
       );
 

+/**
+ * Returns the result of decoding the provided base16 string into an
+ * ArrayBuffer.
+ *
+ * ※ This function is not exposed.
+ */
+const decodeBase16 = (source) => {
+  const u4s = map(
+    source,
+    (ucsCharacter) => {
+      const code = getCodeUnit(ucsCharacter, 0);
+      const result = code >= 0x30 && code <= 0x39
+        ? code - 48
+        : code >= 0x41 && code <= 0x46
+        ? code - 55
+        : code >= 0x61 && code <= 0x66
+        ? code - 87
+        : -1;
+      if (result < 0) {
+        throw new RangeError(
+          `Piscēs: Invalid character in Base64: ${ucsCharacter}.`,
+        );
+      } else {
+        return result;
+      }
+    },
+  );
+  const { length } = u4s;
+  if (length % 2 == 1) {
+    // The length is such that an entire letter would be dropped during
+    // a forgiving decode.
+    throw new RangeError(
+      `Piscēs: Base16 string has invalid length: ${source}.`,
+    );
+  } else {
+    // Every letter contributes at least some bits to the result.
+    const dataView = new View(new Buffer(floor(length / 2)));
+    for (let index = 0; index < length - 1;) {
+      call(viewSetUint8, dataView, [
+        floor(index / 2),
+        (u4s[index] << 4) | u4s[++index, index++],
+      ]);
+    }
+    return call(getViewBuffer, dataView, []);
+  }
+};
+
+/**
+ * Returns the result of decoding the provided base32 string into an
+ * ArrayBuffer.
+ *
+ * If the second argument is truthy, uses Crockford’s encoding rather
+ * than the RFC’s (see <https://www.crockford.com/base32.html>). This
+ * is more human‐friendly and tolerant. Check digits are not supported.
+ *
+ * ※ This function is not exposed.
+ */
+const decodeBase32 = (source, wrmg) => {
+  const u5s = map(
+    wrmg
+      ? stringReplace(source, /-/gu, "")
+      : source.length % 8 == 0
+      ? stringReplace(source, /(?:=|={3,4}|={6})$/u, "")
+      : source,
+    (ucsCharacter) => {
+      const code = getCodeUnit(ucsCharacter, 0);
+      const result = wrmg
+        ? code >= 0x30 && code <= 0x39
+          ? code - 48
+          : code >= 0x41 && code <= 0x48
+          ? code - 55
+          : code == 0x49
+          ? 1 // I
+          : code >= 0x4A && code <= 0x4B
+          ? code - 56
+          : code == 0x4C
+          ? 1 // L
+          : code >= 0x4D && code <= 0x4E
+          ? code - 57
+          : code == 0x4F
+          ? 0 // O
+          : code >= 0x50 && code <= 0x54
+          ? code - 58
+          // U is skipped
+          : code >= 0x56 && code <= 0x5A
+          ? code - 59
+          : code >= 0x61 && code <= 0x68
+          ? code - 87
+          : code == 0x69
+          ? 1 // i
+          : code >= 0x6A && code <= 0x6B
+          ? code - 88
+          : code == 0x6C
+          ? 1 // l
+          : code >= 0x6D && code <= 0x6E
+          ? code - 89
+          : code == 0x6F
+          ? 0 // o
+          : code >= 0x70 && code <= 0x74
+          ? code - 90
+          // u is skipped
+          : code >= 0x76 && code <= 0x7A
+          ? code - 91
+          : -1
+        : code >= 0x41 && code <= 0x5A
+        ? code - 65
+        : code >= 0x61 && code <= 0x7A
+        ? code - 97 // same result as above; case insensitive
+        : code >= 0x32 && code <= 0x37
+        ? code - 24 // digits 2–7 map to 26–31
+        : -1;
+      if (result < 0) {
+        throw new RangeError(
+          `Piscēs: Invalid character in Base32: ${ucsCharacter}.`,
+        );
+      } else {
+        return result;
+      }
+    },
+  );
+  const { length } = u5s;
+  const lengthMod8 = length % 8;
+  if (lengthMod8 == 1 || lengthMod8 == 3 || lengthMod8 == 6) {
+    // The length is such that an entire letter would be dropped during
+    // a forgiving decode.
+    throw new RangeError(
+      `Piscēs: Base32 string has invalid length: ${source}.`,
+    );
+  } else {
+    // Every letter contributes at least some bits to the result.
+    const dataView = new View(new Buffer(floor(length * 5 / 8)));
+    for (let index = 0; index < length - 1;) {
+      // The final index is not handled; if the string is not divisible
+      // by 8, some bits might be dropped. This matches the “forgiving
+      // decode” behaviour specified by WhatW·G for base64.
+      const dataIndex = ceil(index * 5 / 8);
+      const remainder = index % 8;
+      if (remainder == 0) {
+        call(viewSetUint8, dataView, [
+          dataIndex,
+          u5s[index] << 3 | u5s[++index] >> 2,
+        ]);
+      } else if (remainder == 1) {
+        call(viewSetUint8, dataView, [
+          dataIndex,
+          u5s[index] << 6 | u5s[++index] << 1 | u5s[++index] >> 4,
+        ]);
+      } else if (remainder == 3) {
+        call(viewSetUint8, dataView, [
+          dataIndex,
+          u5s[index] << 4 | u5s[++index] >> 1,
+        ]);
+      } else if (remainder == 4) {
+        call(viewSetUint8, dataView, [
+          dataIndex,
+          u5s[index] << 7 | u5s[++index] << 2 | u5s[++index] >> 3,
+        ]);
+      } else { // remainder == 6
+        call(viewSetUint8, dataView, [
+          dataIndex,
+          u5s[index] << 5 | u5s[++index, index++],
+        ]);
+      }
+    }
+    return call(getViewBuffer, dataView, []);
+  }
+};
+

 /**
  * Returns the result of decoding the provided base64 string into an
  * ArrayBuffer.
 /**
  * Returns the result of decoding the provided base64 string into an
  * ArrayBuffer.


@@ -141,10 +309,13 @@ const decodeBase64 = (source, safe = false) => {
   );
   const { length } = u6s;
   if (length % 4 == 1) {
   );
   const { length } = u6s;
   if (length % 4 == 1) {

+    // The length is such that an entire letter would be dropped during
+    // a forgiving decode.

     throw new RangeError(
       `Piscēs: Base64 string has invalid length: ${source}.`,
     );
   } else {
     throw new RangeError(
       `Piscēs: Base64 string has invalid length: ${source}.`,
     );
   } else {

+    // Every letter contributes at least some bits to the result.

     const dataView = new View(new Buffer(floor(length * 3 / 4)));
     for (let index = 0; index < length - 1;) {
       // The final index is not handled; if the string is not divisible
     const dataView = new View(new Buffer(floor(length * 3 / 4)));
     for (let index = 0; index < length - 1;) {
       // The final index is not handled; if the string is not divisible


@@ -173,6 +344,124 @@ const decodeBase64 = (source, safe = false) => {
   }
 };
 
   }
 };
 

+/**
+ * Returns the result of encoding the provided ArrayBuffer into a
+ * base16 string.
+ *
+ * ※ This function is not exposed.
+ */
+const encodeBase16 = (buffer) => {
+  const dataView = new View(buffer);
+  const byteLength = call(getBufferByteLength, buffer, []);
+  const minimumLengthOfResults = byteLength * 2;
+  const resultingCodeUnits = fill(
+    objectCreate(
+      binaryCodeUnitIterablePrototype,
+      { length: { value: minimumLengthOfResults } },
+    ),
+    0x3D,
+  );
+  for (let index = 0; index < byteLength;) {
+    const codeUnitIndex = index * 2;
+    const datum = call(viewGetUint8, dataView, [index++]);
+    const u4s = [datum >> 4, datum & 0xF];
+    for (let u4i = 0; u4i < 2; ++u4i) {
+      const u4 = u4s[u4i];
+      const result = u4 < 10 ? u4 + 48 : u4 < 16 ? u4 + 55 : -1;
+      if (result < 0) {
+        throw new RangeError(
+          `Piscēs: Unexpected Base16 value: ${u4}.`,
+        );
+      } else {
+        resultingCodeUnits[codeUnitIndex + u4i] = result;
+      }
+    }
+  }
+  return stringFromCodeUnits(...resultingCodeUnits);
+};
+
+/**
+ * Returns the result of encoding the provided ArrayBuffer into a
+ * base32 string.
+ *
+ * ※ This function is not exposed.
+ */
+const encodeBase32 = (buffer, wrmg = false) => {
+  const dataView = new View(buffer);
+  const byteLength = call(getBufferByteLength, buffer, []);
+  const minimumLengthOfResults = ceil(byteLength * 8 / 5);
+  const fillByte = wrmg ? 0x2D : 0x3D;
+  const resultingCodeUnits = fill(
+    objectCreate(
+      binaryCodeUnitIterablePrototype,
+      {
+        length: {
+          value: minimumLengthOfResults +
+            (8 - (minimumLengthOfResults % 8)) % 8,
+        },
+      },
+    ),
+    fillByte,
+  );
+  for (let index = 0; index < byteLength;) {
+    const codeUnitIndex = ceil(index * 8 / 5);
+    const currentIndex = codeUnitIndex + +(
+      0b01011 & 1 << index % 5 &&
+      resultingCodeUnits[codeUnitIndex] != fillByte
+    ); // bytes 0, 1 & 3 handle two letters; this is for the second
+    const remainder = currentIndex % 8;
+    const currentByte = call(viewGetUint8, dataView, [index]);
+    const nextByte =
+      0b01011010 & 1 << remainder && ++index < byteLength
+        // digits 1, 3, 4 & 6 span multiple bytes
+        ? call(viewGetUint8, dataView, [index])
+        : 0;
+    const u5 = remainder == 0
+      ? currentByte >> 3
+      : remainder == 1
+      ? (currentByte & 0b00000111) << 2 | nextByte >> 6
+      : remainder == 2
+      ? (currentByte & 0b00111111) >> 1
+      : remainder == 3
+      ? (currentByte & 0b00000001) << 4 | nextByte >> 4
+      : remainder == 4
+      ? (currentByte & 0b00001111) << 1 | nextByte >> 7
+      : remainder == 5
+      ? (currentByte & 0b01111111) >> 2
+      : remainder == 6
+      ? (currentByte & 0b00000011) << 3 | nextByte >> 5
+      : (++index, currentByte & 0b00011111); // remainder == 7
+    const result = wrmg
+      ? u5 < 10 ? u5 + 48 : u5 < 18
+        ? u5 + 55
+        // skip I
+        : u5 < 20
+        ? u5 + 56
+        // skip L
+        : u5 < 22
+        ? u5 + 57
+        // skip O
+        : u5 < 27
+        ? u5 + 58
+        // skip U
+        : u5 < 32
+        ? u5 + 59
+        : -1
+      : u5 < 26
+      ? u5 + 65
+      : u5 < 32
+      ? u5 + 24
+      : -1;
+    if (result < 0) {
+      throw new RangeError(`Piscēs: Unexpected Base32 value: ${u5}.`);
+    } else {
+      resultingCodeUnits[currentIndex] = result;
+    }
+  }
+  const answer = stringFromCodeUnits(...resultingCodeUnits);
+  return wrmg ? answer.replace(/-+$/u, "") : answer;
+};
+

 /**
  * Returns the result of encoding the provided ArrayBuffer into a
  * base64 string.
 /**
  * Returns the result of encoding the provided ArrayBuffer into a
  * base64 string.


@@ -253,6 +542,50 @@ const sourceFromArgs = ($, $s) =>
     "",
   );
 
     "",
   );
 

+/**
+ * Returns an ArrayBuffer generated from the provided base16 string.
+ *
+ * This function can also be used as a tag for a template literal. The
+ * literal will be interpreted akin to `String.raw`.
+ *
+ * ☡ This function throws if the provided string is not a valid base16
+ * string.
+ */
+export const base16Binary = ($, ...$s) =>
+  decodeBase16(sourceFromArgs($, $s));
+
+/**
+ * Returns a (big‐endian) base16 string created from the provided typed
+ * array, buffer, or (16‐bit) string.
+ *
+ * This function can also be used as a tag for a template literal. The
+ * literal will be interpreted akin to `String.raw`.
+ */
+export const base16String = ($, ...$s) =>
+  encodeBase16(bufferFromArgs($, $s));
+
+/**
+ * Returns an ArrayBuffer generated from the provided base32 string.
+ *
+ * This function can also be used as a tag for a template literal. The
+ * literal will be interpreted akin to `String.raw`.
+ *
+ * ☡ This function throws if the provided string is not a valid base32
+ * string.
+ */
+export const base32Binary = ($, ...$s) =>
+  decodeBase32(sourceFromArgs($, $s));
+
+/**
+ * Returns a (big‐endian) base32 string created from the provided typed
+ * array, buffer, or (16‐bit) string.
+ *
+ * This function can also be used as a tag for a template literal. The
+ * literal will be interpreted akin to `String.raw`.
+ */
+export const base32String = ($, ...$s) =>
+  encodeBase32(bufferFromArgs($, $s));
+

 /**
  * Returns an ArrayBuffer generated from the provided base64 string.
  *
 /**
  * Returns an ArrayBuffer generated from the provided base64 string.
  *


@@ -298,6 +631,41 @@ export const filenameSafeBase64Binary = ($, ...$s) =>
 export const filenameSafeBase64String = ($, ...$s) =>
   encodeBase64(bufferFromArgs($, $s), true);
 
 export const filenameSafeBase64String = ($, ...$s) =>
   encodeBase64(bufferFromArgs($, $s), true);
 

+/**
+ * Returns whether the provided value is a base16 string.
+ *
+ * ※ This function returns false if the provided value is not a string
+ * primitive.
+ */
+export const isBase16 = ($) => {
+  if (typeof $ !== "string") {
+    return false;
+  } else {
+    const source = stringReplace($, /[\t\n\f\r ]+/gu, "");
+    return source.length % 2 != 1 &&
+      call(reExec, /[^0-9A-F]/iu, [source]) == null;
+  }
+};
+
+/**
+ * Returns whether the provided value is a base32 string.
+ *
+ * ※ This function returns false if the provided value is not a string
+ * primitive.
+ */
+export const isBase32 = ($) => {
+  if (typeof $ !== "string") {
+    return false;
+  } else {
+    const source = stringReplace($, /[\t\n\f\r ]+/gu, "");
+    const trimmed = source.length % 8 == 0
+      ? stringReplace(source, /(?:=|={3,4}|={6})$/u, "")
+      : source;
+    return trimmed.length % 8 != 1 &&
+      call(reExec, /[^2-7A-Z/]/iu, [trimmed]) == null;
+  }
+};
+

 /**
  * Returns whether the provided value is a Base64 string.
  *
 /**
  * Returns whether the provided value is a Base64 string.
  *


@@ -335,3 +703,44 @@ export const isFilenameSafeBase64 = ($) => {
       call(reExec, /[^0-9A-Za-z_-]/u, [trimmed]) == null;
   }
 };
       call(reExec, /[^0-9A-Za-z_-]/u, [trimmed]) == null;
   }
 };

+
+/**
+ * Returns whether the provided value is a W·R·M·G (Crockford) base32
+ * string. Check digits are not supported.
+ *
+ * ※ This function returns false if the provided value is not a string
+ * primitive.
+ */
+export const isWRMGBase32 = ($) => {
+  if (typeof $ !== "string") {
+    return false;
+  } else {
+    const source = stringReplace($, /[\t\n\f\r ]+/gu, "");
+    const trimmed = stringReplace(source, /-/gu, "");
+    return trimmed.length % 8 != 1 &&
+      call(reExec, /[^0-9A-TV-Z]/iu, [trimmed]) == null;
+  }
+};
+
+/**
+ * Returns an ArrayBuffer generated from the provided W·R·M·G
+ * (Crockford) base32 string.
+ *
+ * This function can also be used as a tag for a template literal. The
+ * literal will be interpreted akin to `String.raw`.
+ *
+ * ☡ This function throws if the provided string is not a valid W·R·M·G
+ * base32 string.
+ */
+export const wrmgBase32Binary = ($, ...$s) =>
+  decodeBase32(sourceFromArgs($, $s), true);
+
+/**
+ * Returns a (big‐endian) W·R·M·G (Crockford) base32 string created
+ * from the provided typed array, buffer, or (16‐bit) string.
+ *
+ * This function can also be used as a tag for a template literal. The
+ * literal will be interpreted akin to `String.raw`.
+ */
+export const wrmgBase32String = ($, ...$s) =>
+  encodeBase32(bufferFromArgs($, $s), true);