Refactor quote handling and add new rules for possessive quotes

src/MyLib.hs

@@ -96,14 +96,14 @@ dotsCJK = do
 fixCJKcolAN :: Rule
 fixCJKcolAN = do
   cjk <- cjkChar
-  _ <- chunk ":"
+  _ <- char ':'
   an <- alphaNumChar
   return $ T.pack $ [cjk] ++ "：" ++ [an]
 
 -- quotes
 -- seems confusing ...
 quotesym :: [Char]
-quotesym = "\x05f4\"\'`"
+quotesym = "'`\x05f4\""
 
 cjkquote :: Rule
 cjkquote = do
@@ -125,7 +125,37 @@ fixQuote = do
   closeQuotes <- T.pack <$> some (oneOf quotesym)
   return $ openQuotes <> T.strip content <> closeQuotes
 
--- the rule set
+cjkpossessivequote :: Rule
+cjkpossessivequote = do
+  cjk <- cjkChar
+  _ <- char '\''
+  _ <- lookAhead $ anySingleBut 's'
+  return $ T.pack $ cjk : " '"
+
+-- This singlequoteCJK rule will turn '你好' into ' 你好'
+-- which seems not desirable...
+-- however, the behavior is aligned with python version
+singlequoteCJK :: Rule
+singlequoteCJK = do
+  _ <- char '\''
+  cjk <- cjkChar
+  return $ T.pack $ "' " ++ [cjk]
+
+fixPossessivequote :: Rule
+fixPossessivequote = do
+  pre <- cjkChar <|> alphaNumChar
+  _ <- some spaceChar
+  _ <- chunk "'s"
+  return $ T.pack $ pre : "'s"
+
+-- hash
+-- hashANSCJKhash :: Rule
+-- hashANSCJKhash = do
+--   cjk1 <- cjkChar
+--   _ <- char '#'
+
+
+-- rule set, the order matters
 myRules :: RuleSet
 myRules =
   [ fullwidthCJKsymCJK,
@@ -134,5 +164,9 @@ myRules =
     fixCJKcolAN,
     cjkquote,
     quoteCJK,
-    fixQuote
+    fixQuote,
+    cjkpossessivequote,
+    -- singlequoteCJK,
+    fixPossessivequote,
+    empty -- a dummy rule
   ]

test/Main.hs

@@ -15,7 +15,8 @@ main = hspec $ do
       applyRules myRules "你…好" `shouldBe` "你… 好"
       applyRules myRules "你...好" `shouldBe` "你... 好"
       applyRules myRules "你:0" `shouldBe` "你：0"
-      applyRules myRules "我说:\" 他说:\'你好\'\"" `shouldBe` "我说：\"他说：\'你好\'\""
+    it "fixes quotes" $ do
-    it "adds spaces" $ do
+      applyRules myRules "我说:\" 他说:'你好'\"" `shouldBe` "我说:\"他说:' 你好 '\""
-      applyRules myRules "\'你好\'" `shouldBe` "\'你好\'"
+      -- applyRules myRules "'你好'" `shouldBe` "' 你好'"  -- strange behavior
-      applyRules myRules "你\'hello\'" `shouldBe` "你 \'hello\'"
+      applyRules myRules "你'hello'" `shouldBe` "你 'hello'"
+      applyRules myRules "我 's " `shouldBe` "我's "