mnemonic: implement Wordlist class

Wordlist subclasses 'tuple' so it can be transparently used.
'in' and '.index()' are fast.
Use Wordlist in bip39_is_checksum_valid, which makes that faster.

electrum/gui/kivy/uix/dialogs/installwizard.py

@@ -859,7 +859,7 @@ class RestoreSeedDialog(WizardDialog):
         super(RestoreSeedDialog, self).__init__(wizard, **kwargs)
         self._test = kwargs['test']
         from electrum.mnemonic import Mnemonic
-        from electrum.old_mnemonic import words as old_wordlist
+        from electrum.old_mnemonic import wordlist as old_wordlist
         self.words = set(Mnemonic('en').wordlist).union(set(old_wordlist))
         self.ids.text_input_seed.text = test_seed if is_test else ''
         self.message = _('Please type your seed phrase using the virtual keyboard.')

electrum/gui/qt/seed_dialog.py

@@ -30,7 +30,7 @@ from PyQt5.QtWidgets import (QVBoxLayout, QCheckBox, QHBoxLayout, QLineEdit,
 
 from electrum.i18n import _
 from electrum.mnemonic import Mnemonic, seed_type
-import electrum.old_mnemonic
+from electrum import old_mnemonic
 
 from .util import (Buttons, OkButton, WWLabel, ButtonsTextEdit, icon_path,
                    EnterButton, CloseButton, WindowModalDialog, ColorScheme)
@@ -150,7 +150,7 @@ class SeedLayout(QVBoxLayout):
 
     def initialize_completer(self):
         bip39_english_list = Mnemonic('en').wordlist
-        old_list = electrum.old_mnemonic.words
+        old_list = old_mnemonic.wordlist
         only_old_list = set(old_list) - set(bip39_english_list)
         self.wordlist = list(bip39_english_list) + list(only_old_list)  # concat both lists
         self.wordlist.sort()

electrum/keystore.py

@@ -42,7 +42,7 @@ from .crypto import (pw_decode, pw_encode, sha256, sha256d, PW_HASH_VERSION_LATE
                      SUPPORTED_PW_HASH_VERSIONS, UnsupportedPasswordHashVersion, hash_160)
 from .util import (InvalidPassword, WalletFileException,
                    BitcoinException, bh2u, bfh, inv_dict, is_hex_str)
-from .mnemonic import Mnemonic, load_wordlist, seed_type, is_seed
+from .mnemonic import Mnemonic, Wordlist, seed_type, is_seed
 from .plugin import run_hook
 from .logging import Logger
 
@@ -811,7 +811,7 @@ def bip39_is_checksum_valid(mnemonic: str) -> Tuple[bool, bool]:
     """
     words = [ normalize('NFKD', word) for word in mnemonic.split() ]
     words_len = len(words)
-    wordlist = load_wordlist("english.txt")
+    wordlist = Wordlist.from_file("english.txt")
     n = len(wordlist)
     i = 0
     words.reverse()

electrum/mnemonic.py

@@ -27,6 +27,8 @@ import math
 import hashlib
 import unicodedata
 import string
+from typing import Sequence, Dict
+from types import MappingProxyType
 
 from .util import resource_path, bfh, bh2u, randrange
 from .crypto import hmac_oneshot
@@ -88,28 +90,48 @@ def normalize_text(seed: str) -> str:
     return seed
 
 
-_WORDLIST_CACHE = {}
+_WORDLIST_CACHE = {}  # type: Dict[str, Wordlist]
 
 
-def load_wordlist(filename) -> tuple:
+class Wordlist(tuple):
-    path = resource_path('wordlist', filename)
-    if path not in _WORDLIST_CACHE:
-        with open(path, 'r', encoding='utf-8') as f:
-            s = f.read().strip()
-        s = unicodedata.normalize('NFKD', s)
-        lines = s.split('\n')
-        wordlist = []
-        for line in lines:
-            line = line.split('#')[0]
-            line = line.strip(' \r')
-            assert ' ' not in line
-            if line:
-                wordlist.append(line)
 
-        # wordlists shouldn't be mutated, but just in case,
+    def __init__(self, words: Sequence[str]):
-        # convert it to a tuple
+        super().__init__()
-        _WORDLIST_CACHE[path] = tuple(wordlist)
+        index_from_word = {w: i for i, w in enumerate(words)}
-    return _WORDLIST_CACHE[path]
+        self._index_from_word = MappingProxyType(index_from_word)  # no mutation
+
+    def index(self, word, start=None, stop=None) -> int:
+        try:
+            return self._index_from_word[word]
+        except KeyError as e:
+            raise ValueError from e
+
+    def __contains__(self, word) -> bool:
+        try:
+            self.index(word)
+        except ValueError:
+            return False
+        else:
+            return True
+
+    @classmethod
+    def from_file(cls, filename) -> 'Wordlist':
+        path = resource_path('wordlist', filename)
+        if path not in _WORDLIST_CACHE:
+            with open(path, 'r', encoding='utf-8') as f:
+                s = f.read().strip()
+            s = unicodedata.normalize('NFKD', s)
+            lines = s.split('\n')
+            words = []
+            for line in lines:
+                line = line.split('#')[0]
+                line = line.strip(' \r')
+                assert ' ' not in line
+                if line:
+                    words.append(line)
+
+            _WORDLIST_CACHE[path] = Wordlist(words)
+        return _WORDLIST_CACHE[path]
 
 
 filenames = {
@@ -130,8 +152,7 @@ class Mnemonic(Logger):
         lang = lang or 'en'
         self.logger.info(f'language {lang}')
         filename = filenames.get(lang[0:2], 'english.txt')
-        self.wordlist = load_wordlist(filename)
+        self.wordlist = Wordlist.from_file(filename)
-        self.wordlist_indexes = {w: i for i, w in enumerate(self.wordlist)}
         self.logger.info(f"wordlist has {len(self.wordlist)} words")
 
     @classmethod
@@ -162,11 +183,11 @@ class Mnemonic(Logger):
         i = 0
         while words:
             w = words.pop()
-            k = self.wordlist_indexes[w]
+            k = self.wordlist.index(w)
             i = i*n + k
         return i
 
-    def make_seed(self, seed_type=None, *, num_bits=132):
+    def make_seed(self, seed_type=None, *, num_bits=132) -> str:
         if seed_type is None:
             seed_type = 'segwit'
         prefix = version.seed_prefix(seed_type)

electrum/old_mnemonic.py

@@ -23,10 +23,12 @@
 # CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 # SOFTWARE.
 
+from .mnemonic import Wordlist
+
 
 # list of words from http://en.wiktionary.org/wiki/Wiktionary:Frequency_lists/Contemporary_poetry
 
-words = (
+_words = (
 "like",
 "just",
 "love",
@@ -1654,9 +1656,10 @@ words = (
 "weapon",
 "weary",
 )
-_words_indexes = {w: i for i, w in enumerate(words)}
 
-n = len(words)
+wordlist = Wordlist(_words)
+
+n = len(wordlist)
 assert n == 1626
 
 
@@ -1672,7 +1675,7 @@ def mn_encode( message ):
         w1 = (x%n)
         w2 = ((x//n) + w1)%n
         w3 = ((x//n//n) + w2)%n
-        out += [ words[w1], words[w2], words[w3] ]
+        out += [ wordlist[w1], wordlist[w2], wordlist[w3] ]
     return out
 
 
@@ -1680,9 +1683,9 @@ def mn_decode( wlist ):
     out = ''
     for i in range(len(wlist)//3):
         word1, word2, word3 = wlist[3*i:3*i+3]
-        w1 =  _words_indexes[word1]
+        w1 =  wordlist.index(word1)
-        w2 = (_words_indexes[word2]) % n
+        w2 = (wordlist.index(word2)) % n
-        w3 = (_words_indexes[word3]) % n
+        w3 = (wordlist.index(word3)) % n
         x = w1 +n*((w2-w1)%n) +n*n*((w3-w2)%n)
         out += '%08x'%x
     return out