Support bech32m spec

azuchi · Feb 5, 2021 · 8713f1a · 8713f1a
1 parent 935391e
commit 8713f1a
Show file tree

Hide file tree

Showing 4 changed files with 136 additions and 49 deletions.
diff --git a/README.md b/README.md
@@ -1,11 +1,15 @@
 # Bech32 [![Build Status](https://travis-ci.org/azuchi/bech32rb.svg?branch=master)](https://travis-ci.org/azuchi/bech32rb) [![Gem Version](https://badge.fury.io/rb/bech32.svg)](https://badge.fury.io/rb/bech32) [![MIT License](http://img.shields.io/badge/license-MIT-blue.svg?style=flat)](LICENSE) <img src="http://segwit.co/static/public/images/logo.png" width="100">
 
-The implementation of the Bech32 encoder and decoder for Ruby.
+The implementation of the Bech32/Bech32m encoder and decoder for Ruby.
 
 Bech32 is checksummed base32 format that is used in following Bitcoin address format.
 
 https://github.com/bitcoin/bips/blob/master/bip-0173.mediawiki
 
+Bech32m is checksummed base32m format that is used in following Bitcoin address format.
+
+https://github.com/bitcoin/bips/blob/master/bip-0350.mediawiki
+
 ## Installation
 
 Add this line to your application's Gemfile:
@@ -35,13 +39,15 @@ require 'bech32'
 Decode Bech32-encoded data into hrp part and data part.
 
 ```ruby
-hrp, data = Bech32.decode('BC1QW508D6QEJXTDG4Y5R3ZARVARY0C5XW7KV8F3T4')
+hrp, data, spec = Bech32.decode('BC1QW508D6QEJXTDG4Y5R3ZARVARY0C5XW7KV8F3T4')
 
 # hrp is human-readable part of Bech32 format
 'bc'
 
 # data is data part of Bech32 format
 [0, 14, 20, 15, 7, 13, 26, 0, 25, 18, 6, 11, 13, 8, 21, 4, 20, 3, 17, 2, 29, 3, 12, 29, 3, 4, 15, 24, 20, 6, 14, 30, 22]
+
+# spec is whether Bech32::Encoding::BECH32 or Bech32::Encoding::BECH32M
 ```
 
 Decode Bech32-encoded Segwit address into `Bech32::SegwitAddr` instance.
@@ -77,7 +83,7 @@ Encode Bech32 human-readable part and data part into Bech32 string.
 hrp = 'bc'
 data = [0, 14, 20, 15, 7, 13, 26, 0, 25, 18, 6, 11, 13, 8, 21, 4, 20, 3, 17, 2, 29, 3, 12, 29, 3, 4, 15, 24, 20, 6, 14, 30, 22]
 
-bech = Bech32.encode(hrp, data)
+bech = Bech32.encode(hrp, data, Bech32::Encoding::BECH32)
 => bc1qw508d6qejxtdg4y5r3zarvary0c5xw7kv8f3t4
 ```
 

diff --git a/lib/bech32.rb b/lib/bech32.rb
@@ -1,24 +1,31 @@
 module Bech32
 
+  module Encoding
+    BECH32 = 1
+    BECH32M = 2
+  end
+
   autoload :SegwitAddr, 'bech32/segwit_addr'
 
   SEPARATOR = '1'
 
   CHARSET = %w(q p z r y 9 x 8 g f 2 t v d w 0 s 3 j n 5 4 k h c e 6 m u a 7 l)
 
+  BECH32M_CONST = 0x2bc830a3
+
   module_function
 
   # Returns the encoded Bech32 string.
   #
   #   require 'bech32'
   #
-  #   bech = Bech32.encode('bc', [])
+  #   bech = Bech32.encode('bc', [], Bech32::Encoding::BECH32)
   #
   #   <i>Generates:</i>
   #   'BC1QW508D6QEJXTDG4Y5R3ZARVARY0C5XW7KV8F3T4' # bech
   #
-  def encode(hrp, data)
-    checksummed = data + create_checksum(hrp, data)
+  def encode(hrp, data, spec)
+    checksummed = data + create_checksum(hrp, data, spec)
     hrp + SEPARATOR + checksummed.map{|i|CHARSET[i]}.join
   end
 
@@ -27,11 +34,12 @@ def encode(hrp, data)
   #   require 'bech32'
   #
   #   addr = 'BC1QW508D6QEJXTDG4Y5R3ZARVARY0C5XW7KV8F3T4'
-  #   hrp, data = Bech32.decode(addr)
+  #   hrp, data, spec = Bech32.decode(addr)
   #
   #   <i>Generates:</i>
   #   'bc' # hrp
   #   [0, 14, 20, 15, 7, 13, 26, 0, 25, 18, 6, 11, 13, 8, 21, 4, 20, 3, 17, 2, 29, 3, 12, 29, 3, 4, 15, 24, 20, 6, 14, 30, 22] # data
+  #   1 # spec see Bech32::Encoding
   #
   def decode(bech, max_length = 90)
     # check uppercase/lowercase
@@ -47,20 +55,30 @@ def decode(bech, max_length = 90)
     hrp = bech[0..pos-1]
     data = bech[pos+1..-1].each_char.map{|c|CHARSET.index(c)}
     # check checksum
-    return nil unless verify_checksum(hrp, data)
-    [hrp, data[0..-7]]
+    spec = verify_checksum(hrp, data)
+    spec ? [hrp, data[0..-7], spec] : nil
   end
 
   # Returns computed checksum values of +hrp+ and +data+
-  def create_checksum(hrp, data)
+  def create_checksum(hrp, data, spec)
     values = expand_hrp(hrp) + data
-    polymod = polymod(values + [0, 0, 0, 0, 0, 0]) ^ 1
+    const = (spec == Bech32::Encoding::BECH32M ? Bech32::BECH32M_CONST : 1)
+    polymod = polymod(values + [0, 0, 0, 0, 0, 0]) ^ const
     (0..5).map{|i|(polymod >> 5 * (5 - i)) & 31}
   end
 
   # Verify a checksum given Bech32 string
+  # @param [String] hrp hrp part.
+  # @param [Array[Integer]] data data array.
+  # @return [Integer] spec
   def verify_checksum(hrp, data)
-    polymod(expand_hrp(hrp) + data) == 1
+    const = polymod(expand_hrp(hrp) + data)
+    case const
+    when 1
+      Encoding::BECH32
+    when BECH32M_CONST
+      Encoding::BECH32M
+    end
   end
 
   # Expand the hrp into values for checksum computation.

diff --git a/lib/bech32/segwit_addr.rb b/lib/bech32/segwit_addr.rb
@@ -30,19 +30,21 @@ def script_pubkey=(script_pubkey)
 
     # Returns segwit address string which generated from hrp, witness version and witness program.
     def addr
-      Bech32.encode(hrp, [ver] + convert_bits(prog, 8, 5))
+      spec = (ver == 0 ? Bech32::Encoding::BECH32 : Bech32::Encoding::BECH32M)
+      Bech32.encode(hrp, [ver] + convert_bits(prog, 8, 5), spec)
     end
 
     private
 
     def parse_addr(addr)
-      @hrp, data = Bech32.decode(addr)
+      @hrp, data, spec = Bech32.decode(addr)
       raise 'Invalid address.' if hrp.nil? || data[0].nil? || ![HRP_MAINNET, HRP_TESTNET, HRP_REGTEST].include?(hrp)
       @ver = data[0]
       raise 'Invalid witness version' if @ver > 16
       @prog = convert_bits(data[1..-1], 5, 8, false)
       raise 'Invalid witness program' if @prog.nil? || @prog.length < 2 || @prog.length > 40
       raise 'Invalid witness program with version 0' if @ver == 0 && (@prog.length != 20 && @prog.length != 32)
+      raise 'Witness version and encoding spec do not match' if (@ver == 0 && spec != Bech32::Encoding::BECH32) || (@ver != 0 && spec != Bech32::Encoding::BECH32M)
     end
 
     def convert_bits(data, from, to, padding=true)

diff --git a/spec/bech32_spec.rb b/spec/bech32_spec.rb
@@ -2,7 +2,7 @@
 
 describe Bech32 do
 
-  VALID_CHECKSUM = [
+  VALID_BECH32 = [
       "A12UEL5L",
       "a12uel5l",
       "an83characterlonghumanreadablepartthatcontainsthenumber1andtheexcludedcharactersbio1tt5tgs",
@@ -12,7 +12,17 @@
       "?1ezyfcl"
   ]
 
-  INVALID_CHECKSUM = [
+  VALID_BECH32M = [
+    "A1LQFN3A",
+    "a1lqfn3a",
+    "an83characterlonghumanreadablepartthatcontainsthetheexcludedcharactersbioandnumber11sg7hg6",
+    "abcdef1l7aum6echk45nj3s0wdvt2fg8x9yrzpqzd3ryx",
+    "11llllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllludsr8",
+    "split1checkupstagehandshakeupstreamerranterredcaperredlc445v",
+    "?1v759aa",
+  ]
+
+  INVALID_BECH32 = [
       " 1nwldj5",         # HRP character out of range
       "\x7F" + "1axkwrx", # HRP character out of range
       "\x80" + "1eym55h", # HRP character out of range
@@ -22,54 +32,105 @@
       "1pzry9x0s0muk",    # Empty HRP
       "x1b4n0q5v",        # Invalid data character
       "li1dgmt3",         # Too short checksum
-      "de1lg7wt\xff",     # Invalid character in checksum
+      "de1lg7wt" + "\xff",# Invalid character in checksum
       "A1G7SGD8",         # checksum calculated with uppercase form of HRP
       "10a06t8",          # empty HRP
       "1qzzfhee",         # empty HRP
   ]
 
+  INVALID_BECH32M = [
+    " 1xj0phk",          # HRP character out of range
+    "\x7F" + "1g6xzxy",  # HRP character out of range
+    "\x80" + "1vctc34",  # HRP character out of range
+    # overall max length exceeded
+    "an84characterslonghumanreadablepartthatcontainsthetheexcludedcharactersbioandnumber11d6pts4",
+    "qyrz8wqd2c9m",      # No separator character
+    "1qyrz8wqd2c9m",     # Empty HRP
+    "y1b0jsk6g",         # Invalid data character
+    "lt1igcx5c0",        # Invalid data character
+    "in1muywd",          # Too short checksum
+    "mm1crxm3i",         # Invalid character in checksum
+    "au1s5cgom",         # Invalid character in checksum
+    "M1VUXWEZ",          # Checksum calculated with uppercase form of HRP
+    "16plkw9",           # Empty HRP
+    "1p2gdwpf",          # Empty HRP
+  ]
+
   VALID_ADDRESS = [
-      ["BC1QW508D6QEJXTDG4Y5R3ZARVARY0C5XW7KV8F3T4", "0014751e76e8199196d454941c45d1b3a323f1433bd6"],
-      ["tb1qrp33g0q5c5txsp9arysrx4k6zdkfs4nce4xj0gdcccefvpysxf3q0sl5k7",
-       "00201863143c14c5166804bd19203356da136c985678cd4d27a1b8c6329604903262"],
-      ["bc1pw508d6qejxtdg4y5r3zarvary0c5xw7kw508d6qejxtdg4y5r3zarvary0c5xw7k7grplx",
-       "5128751e76e8199196d454941c45d1b3a323f1433bd6751e76e8199196d454941c45d1b3a323f1433bd6"],
-      ["BC1SW50QA3JX3S", "6002751e"],
-      ["bc1zw508d6qejxtdg4y5r3zarvaryvg6kdaj", "5210751e76e8199196d454941c45d1b3a323"],
-      ["tb1qqqqqp399et2xygdj5xreqhjjvcmzhxw4aywxecjdzew6hylgvsesrxh6hy",
-       "0020000000c4a5cad46221b2a187905e5266362b99d5e91c6ce24d165dab93e86433"],
-      ["bcrt1q8w9txuafw0q3lvhqrdsl49fea53h7rfwulh46h", "00143b8ab373a973c11fb2e01b61fa9539ed237f0d2e"]
+    ["BC1QW508D6QEJXTDG4Y5R3ZARVARY0C5XW7KV8F3T4", "0014751e76e8199196d454941c45d1b3a323f1433bd6"],
+    ["tb1qrp33g0q5c5txsp9arysrx4k6zdkfs4nce4xj0gdcccefvpysxf3q0sl5k7",
+     "00201863143c14c5166804bd19203356da136c985678cd4d27a1b8c6329604903262"],
+    ["bc1pw508d6qejxtdg4y5r3zarvary0c5xw7kw508d6qejxtdg4y5r3zarvary0c5xw7kt5nd6y",
+     "5128751e76e8199196d454941c45d1b3a323f1433bd6751e76e8199196d454941c45d1b3a323f1433bd6"],
+    ["BC1SW50QGDZ25J", "6002751e"],
+    ["bc1zw508d6qejxtdg4y5r3zarvaryvaxxpcs", "5210751e76e8199196d454941c45d1b3a323"],
+    ["tb1qqqqqp399et2xygdj5xreqhjjvcmzhxw4aywxecjdzew6hylgvsesrxh6hy",
+     "0020000000c4a5cad46221b2a187905e5266362b99d5e91c6ce24d165dab93e86433"],
+    ["tb1pqqqqp399et2xygdj5xreqhjjvcmzhxw4aywxecjdzew6hylgvsesf3hn0c",
+     "5120000000c4a5cad46221b2a187905e5266362b99d5e91c6ce24d165dab93e86433"],
+    ["bc1p0xlxvlhemja6c4dqv22uapctqupfhlxm9h8z3k2e72q4k9hcz7vqzk5jj0",
+     "512079be667ef9dcbbac55a06295ce870b07029bfcdb2dce28d959f2815b16f81798"]
   ]
 
   INVALID_ADDRESS = [
-      "tc1qw508d6qejxtdg4y5r3zarvary0c5xw7kg3g4ty",
-      "bc1qw508d6qejxtdg4y5r3zarvary0c5xw7kv8f3t5",
-      "BC13W508D6QEJXTDG4Y5R3ZARVARY0C5XW7KN40WF2",
-      "bc1rw5uspcuh",
-      "bc10w508d6qejxtdg4y5r3zarvary0c5xw7kw508d6qejxtdg4y5r3zarvary0c5xw7kw5rljs90",
-      "bca0w508d6qejxtdg4y5r3zarvary0c5xw7kw508d6qejxtdg4y5r3zarvary0c5xw7kw5rljs90234567789035",
-      "BC1QR508D6QEJXTDG4Y5R3ZARVARYV98GJ9P",
-      "tb1qrp33g0q5c5txsp9arysrx4k6zdkfs4nce4xj0gdcccefvpysxf3q0sL5k7",
-      "bc1zw508d6qejxtdg4y5r3zarvaryvqyzf3du",
-      "tb1qrp33g0q5c5txsp9arysrx4k6zdkfs4nce4xj0gdcccefvpysxf3pjxtptv",
-      "bc1gmk9yu",
+    # Invalid HRP
+    "tc1p0xlxvlhemja6c4dqv22uapctqupfhlxm9h8z3k2e72q4k9hcz7vq5zuyut",
+    # Invalid checksum algorithm (bech32 instead of bech32m)
+    "bc1p0xlxvlhemja6c4dqv22uapctqupfhlxm9h8z3k2e72q4k9hcz7vqh2y7hd",
+    # Invalid checksum algorithm (bech32 instead of bech32m)
+    "tb1z0xlxvlhemja6c4dqv22uapctqupfhlxm9h8z3k2e72q4k9hcz7vqglt7rf",
+    # Invalid checksum algorithm (bech32 instead of bech32m)
+    "BC1S0XLXVLHEMJA6C4DQV22UAPCTQUPFHLXM9H8Z3K2E72Q4K9HCZ7VQ54WELL",
+    # Invalid checksum algorithm (bech32m instead of bech32)
+    "bc1qw508d6qejxtdg4y5r3zarvary0c5xw7kemeawh",
+    # Invalid checksum algorithm (bech32m instead of bech32)
+    "tb1q0xlxvlhemja6c4dqv22uapctqupfhlxm9h8z3k2e72q4k9hcz7vq24jc47",
+    # Invalid character in checksum
+    "bc1p38j9r5y49hruaue7wxjce0updqjuyyx0kh56v8s25huc6995vvpql3jow4",
+    # Invalid witness version
+    "BC130XLXVLHEMJA6C4DQV22UAPCTQUPFHLXM9H8Z3K2E72Q4K9HCZ7VQ7ZWS8R",
+    # Invalid program length (1 byte)
+    "bc1pw5dgrnzv",
+    # Invalid program length (41 bytes)
+    "bc1p0xlxvlhemja6c4dqv22uapctqupfhlxm9h8z3k2e72q4k9hcz7v8n0nx0muaewav253zgeav",
+    # Invalid program length for witness version 0 (per BIP141)
+    "BC1QR508D6QEJXTDG4Y5R3ZARVARYV98GJ9P",
+    # Mixed case
+    "tb1p0xlxvlhemja6c4dqv22uapctqupfhlxm9h8z3k2e72q4k9hcz7vq47Zagq",
+    # More than 4 padding bits
+    "bc1p0xlxvlhemja6c4dqv22uapctqupfhlxm9h8z3k2e72q4k9hcz7v07qwwzcrf",
+    # Non-zero padding in 8-to-5 conversion
+    "tb1p0xlxvlhemja6c4dqv22uapctqupfhlxm9h8z3k2e72q4k9hcz7vpggkg4j",
+    # Empty data section
+    "bc1gmk9yu"
   ]
 
   it 'valid checksum' do
-    VALID_CHECKSUM.each do |bech|
-      hrp, _ = Bech32.decode(bech)
-      expect(hrp).to be_truthy
-      pos = bech.rindex('1')
-      bech = bech[0..pos] + (bech[pos + 1].ord ^ 1).chr + bech[pos+2..-1]
-      hrp, _ = Bech32.decode(bech)
-      expect(hrp).to be_nil
+    specs = {}
+    specs[Bech32::Encoding::BECH32] = VALID_BECH32
+    specs[Bech32::Encoding::BECH32M] = VALID_BECH32M
+    specs.each do |k, v|
+      v.each do |bech|
+        hrp, _, spec = Bech32.decode(bech)
+        expect(hrp).to be_truthy
+        expect(spec).to eq(k)
+        pos = bech.rindex('1')
+        bech = bech[0..pos] + (bech[pos + 1].ord ^ 1).chr + bech[pos+2..-1]
+        hrp, _ = Bech32.decode(bech)
+        expect(hrp).to be_nil
+      end
     end
   end
 
-  def test_invalid_checksum
-    INVALID_CHECKSUM.each do |bech|
-      hrp, _ = Bech32.decode(bech)
-      assert_nil (hrp)
+  it 'test_invalid_checksum' do
+    specs = {}
+    specs[Bech32::Encoding::BECH32] = INVALID_BECH32
+    specs[Bech32::Encoding::BECH32M] = INVALID_BECH32M
+    specs.each do |k, v|
+      v.each do |bech|
+        hrp, _, spec = Bech32.decode(bech)
+        expect(hrp.nil? || spec != k).to be true
+      end
     end
   end