Ajout d'une correction orthographique

1 files changed, 84 insertions, 6 deletions

diff --git a/dfr.py b/dfr.py
index e664264..1ed5d40 100755
--- a/dfr.py
+++ b/dfr.py
@@ -11,6 +11,7 @@ Maybe extended to other languages later.
 import sys
 import argparse
 import msgpack
+import gzip
 import sqlite3
 from os.path import exists
 import os
@@ -27,6 +28,60 @@ else:
 
 dico = 'dfr.db'
 
+def initWordList():
+    if (not arg.wordList) or (arg.wordList == f'{DIR_PATH}/assets/wordList' and not exists(f'{DIR_PATH}/assets/wordList')):
+        create_wordlist()
+        arg.wordList = f'{DIR_PATH}/assets/wordList'
+
+
+
+def didYouMean(word):
+    A = "abcdefghijklmnopqrstuvwxyzéèïœä"
+    def var(word_i, i=1, alpha=A, memo={}):
+
+        if word_i == "":
+            for c in A:
+                memo[c] = True
+                yield c
+
+        if word_i in memo:
+            return
+
+        if i == 1:
+            yield word_i
+            return
+
+        for word in var(word_i, i - 1, alpha, memo={}):
+            if word not in memo:
+                memo[word] = True
+                yield word
+            for i in range(len(word)):
+                for c in A:
+                    # Replace
+                    if c != word[i]:
+                        if word[:i] + c + word[i + 1:] not in memo:
+                            memo[word[:i] + c + word[i + 1:]] = True
+                            yield word[:i] + c + word[i + 1:]
+                    # Insert
+                    if word[:i] + c + word[i:] not in memo:
+                        memo[word[:i] + c + word[i:]] = True
+                        yield word[:i] + c + word[i:]
+
+            for i in range(len(word)):
+                if word[:i] + word[i + 1:] not in memo:
+                    memo[word[:i] + word[i + 1:]] = True
+                    yield word[:i] + word[i + 1:]
+
+
+    with gzip.open(arg.wordList, 'r') as f:
+        wl = f.read().decode()
+        listWord = {w for w in wl.split('\n')}
+        ret = [w for w in var(word, 2, memo={}) if w in listWord]
+        if ret == []:
+            return [w for w in var(word, 3, memo={}) if w in listWord]
+        else:
+            return ret
+
 
 def get_def_sql(word):
     with sqlite3.connect(dico) as con:
@@ -45,6 +100,16 @@ def get_def_sql(word):
                                    }, res))
 
 
+def create_wordlist():
+    with sqlite3.connect(dico) as con:
+        cur = con.cursor()
+        cur.execute('''SELECT word FROM entry''')
+
+        with gzip.open(f"{DIR_PATH}/assets/wordList", 'w') as f:
+            to_w = list({f[0] for f in cur.fetchall()})
+            f.write(str.encode("\n".join(to_w)))
+
+
 def get_def_sql_reg(word):
     with sqlite3.connect(dico) as con:
         cur = con.cursor()
@@ -73,9 +138,9 @@ def matching(word):
         print(f'Error: Word list {arg.wordList} not found', file=sys.stderr)
         return
 
-    with open(arg.wordList, 'rb') as f:
-        msgpackList = f.read()
-        listWord = msgpack.unpackb(msgpackList, raw=False)
+    with gzip.open(arg.wordList, 'r') as f:
+        msgpackList = f.read().decode()
+        listWord = msgpackList.split('\n')
 
         if word[0] != '/':
             for w in listWord:
@@ -113,7 +178,7 @@ if __name__ == '__main__':
                         help='the word or the pattern to match')
     parser.add_argument('-l', '--word-list', dest='wordList', action='store',
                         help='the filename of the word list',
-                        default=None)
+                        default=f'{DIR_PATH}/assets/wordList')
     parser.add_argument('-d', '--dico', dest='dico', action='store',
                         help='the filename of the dictionnary',
                         default='dfr.db')
@@ -149,6 +214,8 @@ if __name__ == '__main__':
             dico = f'{DIR_PATH}/assets/{dico}'
 
     if arg.matching:
+        initWordList()
+
         ret = matching(arg.word)
         for word in ret:
             print(word)
@@ -157,11 +224,22 @@ if __name__ == '__main__':
     else:
         if arg.first:
             a = arg.action(arg.word)
-            if a :
+            if a:
                 if a[0]['def']:
                     a[0]['def'] = [a[0]['def'][0]]
                     a[0]['def'][0]['ex'] = []
                 ui.show_terminal(a[0])
         else:
-            for w in arg.action(arg.word):
+            res = arg.action(arg.word)
+            for w in res:
                 ui.show_terminal(w)
+            if not res:
+                print("Word not found")
+
+                initWordList()
+                listVar = didYouMean(arg.word)
+
+                if len(listVar) == 1:
+                    print("Did you mean  ", listVar[0])
+                elif len(listVar) > 1:
+                    print("Did you mean  ", ", ".join(listVar[:-1]), 'or', listVar[-1], '?')