add readme. remove changelog

CHANGELOG.md

@@ -1,5 +0,0 @@
-# Revision history for panguFilter
-
-## 0.1.0.0 -- YYYY-mm-dd
-
-* First version. Released on an unsuspecting world.

README.md

@@ -0,0 +1,11 @@
+# pangu.hs
+
+Insert whitespace between CJK and half-width characters.
+
+This haskell version implements a subset of spacing rules in [pangu.py](https://github.com/vinta/pangu.py).
+
+## test
+
+```sh
+cabal test
+```

pangu.py

@@ -1,191 +0,0 @@
-#!/usr/bin/env python
-# coding: utf-8
-"""
-Paranoid text spacing for good readability, to automatically insert whitespace between CJK (Chinese, Japanese, Korean) and half-width characters (alphabetical letters, numerical digits and symbols).
-
->>> import pangu
->>> nwe_text = pangu.spacing_text('當你凝視著bug，bug也凝視著你')
->>> print(nwe_text)
-'當你凝視著 bug，bug 也凝視著你'
->>> nwe_content = pangu.spacing_file('path/to/file.txt')
->>> print(nwe_content)
-'與 PM 戰鬥的人，應當小心自己不要成為 PM'
-"""
-
-import argparse
-import os
-import re
-import sys
-
-__version__ = '4.0.6.1'
-__all__ = ['spacing_text', 'spacing_file', 'spacing', 'cli']
-
-CJK = r'\u2e80-\u2eff\u2f00-\u2fdf\u3040-\u309f\u30a0-\u30fa\u30fc-\u30ff\u3100-\u312f\u3200-\u32ff\u3400-\u4dbf\u4e00-\u9fff\uf900-\ufaff'
-
-ANY_CJK = re.compile(r'[{CJK}]'.format(CJK=CJK))
-
-CONVERT_TO_FULLWIDTH_CJK_SYMBOLS_CJK = re.compile('([{CJK}])([ ]*(?:[\\:]+|\\.)[ ]*)([{CJK}])'.format(CJK=CJK))  # there is an extra non-capturing group compared to JavaScript version
-CONVERT_TO_FULLWIDTH_CJK_SYMBOLS = re.compile('([{CJK}])[ ]*([~\\!;,\\?]+)[ ]*'.format(CJK=CJK))
-DOTS_CJK = re.compile('([\\.]{{2,}}|\u2026)([{CJK}])'.format(CJK=CJK))  # need to escape { }
-FIX_CJK_COLON_ANS = re.compile('([{CJK}])\\:([A-Z0-9\\(\\)])'.format(CJK=CJK))
-
-CJK_QUOTE = re.compile('([{CJK}])([`"\u05f4])'.format(CJK=CJK))  # no need to escape `
-QUOTE_CJK = re.compile('([`"\u05f4])([{CJK}])'.format(CJK=CJK))  # no need to escape `
-FIX_QUOTE_ANY_QUOTE = re.compile(r'([`"\u05f4]+)(\s*)(.+?)(\s*)([`"\u05f4]+)')
-
-CJK_SINGLE_QUOTE_BUT_POSSESSIVE = re.compile("([{CJK}])('[^s])".format(CJK=CJK))
-SINGLE_QUOTE_CJK = re.compile("(')([{CJK}])".format(CJK=CJK))
-FIX_POSSESSIVE_SINGLE_QUOTE = re.compile("([{CJK}A-Za-z0-9])( )('s)".format(CJK=CJK))
-
-HASH_ANS_CJK_HASH = re.compile('([{CJK}])(#)([{CJK}]+)(#)([{CJK}])'.format(CJK=CJK))
-CJK_HASH = re.compile('([{CJK}])(#([^ ]))'.format(CJK=CJK))
-HASH_CJK = re.compile('(([^ ])#)([{CJK}])'.format(CJK=CJK))
-
-CJK_OPERATOR_ANS = re.compile('([{CJK}])([\\+\\-\\*\\/=&\\|<>])([A-Za-z0-9])'.format(CJK=CJK))
-ANS_OPERATOR_CJK = re.compile('([A-Za-z0-9])([\\+\\-\\*\\/=&\\|<>])([{CJK}])'.format(CJK=CJK))
-
-FIX_SLASH_AS = re.compile(r'([/]) ([a-z\-_\./]+)')
-FIX_SLASH_AS_SLASH = re.compile(r'([/\.])([A-Za-z\-_\./]+) ([/])')
-
-CJK_LEFT_BRACKET = re.compile('([{CJK}])([\\(\\[\\{{<>\u201c])'.format(CJK=CJK))  # need to escape {
-RIGHT_BRACKET_CJK = re.compile('([\\)\\]\\}}<>\u201d])([{CJK}])'.format(CJK=CJK))  # need to escape }
-FIX_LEFT_BRACKET_ANY_RIGHT_BRACKET = re.compile(r'([\(\[\{<\u201c]+)(\s*)(.+?)(\s*)([\)\]\}>\u201d]+)')  # need to escape { }
-ANS_CJK_LEFT_BRACKET_ANY_RIGHT_BRACKET = re.compile('([A-Za-z0-9{CJK}])[ ]*([\u201c])([A-Za-z0-9{CJK}\\-_ ]+)([\u201d])'.format(CJK=CJK))
-LEFT_BRACKET_ANY_RIGHT_BRACKET_ANS_CJK = re.compile('([\u201c])([A-Za-z0-9{CJK}\\-_ ]+)([\u201d])[ ]*([A-Za-z0-9{CJK}])'.format(CJK=CJK))
-
-AN_LEFT_BRACKET = re.compile(r'([A-Za-z0-9])([\(\[\{])')
-RIGHT_BRACKET_AN = re.compile(r'([\)\]\}])([A-Za-z0-9])')
-
-CJK_ANS = re.compile('([{CJK}])([A-Za-z\u0370-\u03ff0-9@\\$%\\^&\\*\\-\\+\\\\=\\|/\u00a1-\u00ff\u2150-\u218f\u2700—\u27bf])'.format(CJK=CJK))
-ANS_CJK = re.compile('([A-Za-z\u0370-\u03ff0-9~\\!\\$%\\^&\\*\\-\\+\\\\=\\|;:,\\./\\?\u00a1-\u00ff\u2150-\u218f\u2700—\u27bf])([{CJK}])'.format(CJK=CJK))
-
-S_A = re.compile(r'(%)([A-Za-z])')
-
-MIDDLE_DOT = re.compile(r'([ ]*)([\u00b7\u2022\u2027])([ ]*)')
-
-# Python version only
-TILDES = re.compile(r'~+')
-EXCLAMATION_MARKS = re.compile(r'!+')
-SEMICOLONS = re.compile(r';+')
-COLONS = re.compile(r':+')
-COMMAS = re.compile(r',+')
-PERIODS = re.compile(r'\.+')
-QUESTION_MARKS = re.compile(r'\?+')
-
-
-def convert_to_fullwidth(symbols):
-    symbols = TILDES.sub('～', symbols)
-    symbols = EXCLAMATION_MARKS.sub('！', symbols)
-    symbols = SEMICOLONS.sub('；', symbols)
-    symbols = COLONS.sub('：', symbols)
-    symbols = COMMAS.sub('，', symbols)
-    symbols = PERIODS.sub('。', symbols)
-    symbols = QUESTION_MARKS.sub('？', symbols)
-    return symbols.strip()
-
-
-def spacing(text):
-    """
-    Perform paranoid text spacing on text.
-    """
-    if len(text) <= 1 or not ANY_CJK.search(text):
-        return text
-
-    new_text = text
-
-    # TODO: refactoring
-    matched = CONVERT_TO_FULLWIDTH_CJK_SYMBOLS_CJK.search(new_text)
-    while matched:
-        start, end = matched.span()
-        new_text = ''.join((new_text[:start + 1], convert_to_fullwidth(new_text[start + 1:end - 1]), new_text[end - 1:]))
-        matched = CONVERT_TO_FULLWIDTH_CJK_SYMBOLS_CJK.search(new_text)
-
-    matched = CONVERT_TO_FULLWIDTH_CJK_SYMBOLS.search(new_text)
-    while matched:
-        start, end = matched.span()
-        new_text = ''.join((new_text[:start + 1].strip(), convert_to_fullwidth(new_text[start + 1:end]), new_text[end:].strip()))
-        matched = CONVERT_TO_FULLWIDTH_CJK_SYMBOLS.search(new_text)
-
-    new_text = DOTS_CJK.sub(r'\1 \2', new_text)
-    new_text = FIX_CJK_COLON_ANS.sub(r'\1：\2', new_text)
-
-    new_text = CJK_QUOTE.sub(r'\1 \2', new_text)
-    new_text = QUOTE_CJK.sub(r'\1 \2', new_text)
-    new_text = FIX_QUOTE_ANY_QUOTE.sub(r'\1\3\5', new_text)
-
-    new_text = CJK_SINGLE_QUOTE_BUT_POSSESSIVE.sub(r'\1 \2', new_text)
-    new_text = SINGLE_QUOTE_CJK.sub(r'\1 \2', new_text)
-    new_text = FIX_POSSESSIVE_SINGLE_QUOTE.sub(r"\1's", new_text)
-
-    new_text = HASH_ANS_CJK_HASH.sub(r'\1 \2\3\4 \5', new_text)
-    new_text = CJK_HASH.sub(r'\1 \2', new_text)
-    new_text = HASH_CJK.sub(r'\1 \3', new_text)
-
-    new_text = CJK_OPERATOR_ANS.sub(r'\1 \2 \3', new_text)
-    new_text = ANS_OPERATOR_CJK.sub(r'\1 \2 \3', new_text)
-
-    new_text = FIX_SLASH_AS.sub(r'\1\2', new_text)
-    new_text = FIX_SLASH_AS_SLASH.sub(r'\1\2\3', new_text)
-
-    new_text = CJK_LEFT_BRACKET.sub(r'\1 \2', new_text)
-    new_text = RIGHT_BRACKET_CJK.sub(r'\1 \2', new_text)
-    new_text = FIX_LEFT_BRACKET_ANY_RIGHT_BRACKET.sub(r'\1\3\5', new_text)
-    new_text = ANS_CJK_LEFT_BRACKET_ANY_RIGHT_BRACKET.sub(r'\1 \2\3\4', new_text)
-    new_text = LEFT_BRACKET_ANY_RIGHT_BRACKET_ANS_CJK.sub(r'\1\2\3 \4', new_text)
-
-    new_text = AN_LEFT_BRACKET.sub(r'\1 \2', new_text)
-    new_text = RIGHT_BRACKET_AN.sub(r'\1 \2', new_text)
-
-    new_text = CJK_ANS.sub(r'\1 \2', new_text)
-    new_text = ANS_CJK.sub(r'\1 \2', new_text)
-
-    new_text = S_A.sub(r'\1 \2', new_text)
-
-    new_text = MIDDLE_DOT.sub('・', new_text)
-
-    return new_text.strip()
-
-
-def spacing_text(text):
-    """
-    Perform paranoid text spacing on text. An alias of `spacing()`.
-    """
-    return spacing(text)
-
-
-def spacing_file(path):
-    """
-    Perform paranoid text spacing from file.
-    """
-    # TODO: read line by line
-    with open(os.path.abspath(path)) as f:
-        return spacing_text(f.read())
-
-
-def cli(args=None):
-    if not args:
-        args = sys.argv[1:]
-
-    parser = argparse.ArgumentParser(
-        prog='pangu',
-        description='pangu.py -- Paranoid text spacing for good readability, to automatically insert whitespace between CJK and half-width characters (alphabetical letters, numerical digits and symbols).',
-    )
-    parser.add_argument('-v', '--version', action='version', version=__version__)
-    parser.add_argument('-t', '--text', action='store_true', dest='is_text', required=False, help='specify the input value is a text')
-    parser.add_argument('-f', '--file', action='store_true', dest='is_file', required=False, help='specify the input value is a file path')
-    parser.add_argument('text_or_path', action='store', type=str, help='the text or file path to apply spacing')
-
-    if not sys.stdin.isatty():
-        print(spacing_text(sys.stdin.read()))  # noqa: T003
-    else:
-        args = parser.parse_args(args)
-        if args.is_text:
-            print(spacing_text(args.text_or_path))  # noqa: T003
-        elif args.is_file:
-            print(spacing_file(args.text_or_path))  # noqa: T003
-        else:
-            print(spacing_text(args.text_or_path))  # noqa: T003
-
-
-if __name__ == '__main__':
-    cli()

panguFilter.cabal

@@ -46,7 +46,7 @@ category:           Text
 build-type:         Simple
 
 -- Extra doc files to be distributed with the package, such as a CHANGELOG or a README.
-extra-doc-files:    CHANGELOG.md
+-- extra-doc-files:    CHANGELOG.md
 
 -- Extra source files to be distributed with the package, such as examples, or a tutorial module.
 -- extra-source-files:
@@ -59,7 +59,7 @@ library
     import:           warnings
 
     -- Modules exported by the library.
-    exposed-modules:  MyLib
+    exposed-modules:  Pangu
 
     -- Modules included in this library but not exported.
     -- other-modules:

src/MyLib.hs

@@ -1,138 +0,0 @@
-{-# LANGUAGE OverloadedStrings #-}
-
-module MyLib where
-
-import Data.Function (fix)
-import Data.Text (Text)
-import qualified Data.Text as T
-import Data.Void (Void)
-import Replace.Megaparsec (streamEdit)
-import Text.Megaparsec
-import Text.Megaparsec.Char
-
--------------------------------------------------------------------------------
-type Parser = Parsec Void Text
-
-type Rule = Parser Text
-
-type RuleSet = [Rule]
-
-applyUntilFixed :: Rule -> Text -> Text
-applyUntilFixed rule =
-  fix
-    ( \loop current ->
-        let next = streamEdit rule id current
-         in if next == current then next else loop next
-    )
-
-applyRules :: RuleSet -> Text -> Text
-applyRules rules input = foldl (flip applyUntilFixed) input rules
-
--------------------------------------------------------------------------------
--- rules for pangu
-
--- | Check if a character falls within the CJK ranges provided
-isCJK :: Char -> Bool
-isCJK c = any (\(start, end) -> c >= start && c <= end) cjkRanges
-  where
-    cjkRanges =
-      [ ('\x2e80', '\x2eff'),
-        ('\x2f00', '\x2fdf'),
-        ('\x3040', '\x309f'),
-        ('\x30a0', '\x30fa'),
-        ('\x30fc', '\x30ff'),
-        ('\x3100', '\x312f'),
-        ('\x3200', '\x32ff'),
-        ('\x3400', '\x4dbf'),
-        ('\x4e00', '\x9fff'),
-        ('\xf900', '\xfaff')
-      ]
-
-convertToFullwidth :: Char -> Char
-convertToFullwidth c =
-  case c of
-    ':' -> '：'
-    '.' -> '。'
-    '~' -> '～'
-    '!' -> '！'
-    '?' -> '？'
-    ',' -> '，'
-    ';' -> '；'
-    '\"' -> '”'
-    '\'' -> '’'
-    _ -> c
-
--- A parser that matches a single CJK character
-cjkChar :: Parser Char
-cjkChar = satisfy isCJK
-
--- use python.py as reference for these rules
-
-fullwidthCJKsymCJK :: Rule
-fullwidthCJKsymCJK = do
-  lcjk <- cjkChar
-  _ <- many (char ' ')
-  sym <- try (some (char ':')) <|> count 1 (char '.')
-  _ <- many (char ' ')
-  rcjk <- cjkChar
-  let transformedsym =  map convertToFullwidth sym
-  return $ T.pack $ [lcjk] ++ transformedsym ++ [rcjk]
-
-fullwidthCJKsym :: Rule
-fullwidthCJKsym = do
-  cjk <- cjkChar
-  _ <- many (char ' ')
-  sym <- some $ oneOf ("~!?,;" :: [Char])
-  _ <- many (char ' ')
-  let transformedsym = T.pack $ map convertToFullwidth sym
-  return $ T.pack [cjk] <> transformedsym
-
-dotsCJK :: Rule
-dotsCJK = do
-  dots <- chunk "..." <|> chunk "…"
-  cjk <- cjkChar
-  return $ dots <> T.pack (" " ++ [cjk])
-
-fixCJKcolAN :: Rule
-fixCJKcolAN = do
-  cjk <- cjkChar
-  _ <- chunk ":"
-  an <- alphaNumChar
-  return $ T.pack $ [cjk] ++ "：" ++ [an]
-
--- quotes
--- seems confusing ...
-quotesym :: [Char]
-quotesym = "\x05f4\"\'`"
-
-cjkquote :: Rule
-cjkquote = do
-  cjk <- cjkChar
-  quote <- oneOf quotesym
-  return $ T.pack $ [cjk] ++ " " ++ [quote]
-
-quoteCJK :: Rule
-quoteCJK = do
-  quote <- oneOf quotesym
-  cjk <- cjkChar
-  return $ T.pack $ [quote] ++ " " ++ [cjk]
-
-fixQuote :: Rule
-fixQuote = do
-  openQuotes <- T.pack <$> some (oneOf quotesym)
-  _ <- many spaceChar
-  content <- T.pack <$> someTill anySingle (lookAhead $ some (oneOf quotesym))
-  closeQuotes <- T.pack  <$> some (oneOf quotesym)
-  return $ openQuotes <> T.strip content <> closeQuotes
-
--- the rule set
-myRules :: RuleSet
-myRules =
-  [ fullwidthCJKsymCJK,
-    fullwidthCJKsym,
-    dotsCJK,
-    fixCJKcolAN,
-    cjkquote,
-    quoteCJK,
-    fixQuote
-  ]

src/Pangu.hs

@@ -0,0 +1,242 @@
+{-# LANGUAGE OverloadedStrings #-}
+
+module Pangu (pangu, isCJK) where
+
+import Data.Function (fix)
+import Data.Text (Text)
+import qualified Data.Text as T
+import Data.Void (Void)
+import Replace.Megaparsec (streamEdit)
+import Text.Megaparsec
+import Text.Megaparsec.Char
+
+-------------------------------------------------------------------------------
+type Parser = Parsec Void Text
+
+type Rule = Parser Text
+
+type RuleSet = [Rule]
+
+applyUntilFixed :: Rule -> Text -> Text
+applyUntilFixed rule =
+  fix
+    ( \loop current ->
+        let next = streamEdit (try rule) id current
+         in if next == current then next else loop next
+    )
+
+applyRulesRecursively :: RuleSet -> Text -> Text
+applyRulesRecursively rules input = foldl (flip applyUntilFixed) input rules
+
+applyRules :: RuleSet -> Text -> Text
+applyRules rules input = foldl (flip applyOnce) input rules
+  where
+    applyOnce rule = streamEdit (try rule) id
+
+-------------------------------------------------------------------------------
+-- rules for pangu
+
+-- alphaNumChar from megaparsec matches CJK chars...
+-- need to implement a new one
+alphanumericChar :: Parser Char
+alphanumericChar = satisfy $ \c ->
+  (c >= 'a' && c <= 'z')
+    || (c >= 'A' && c <= 'Z')
+    || (c >= '0' && c <= '9')
+
+-- | Check if a character falls within the CJK ranges provided
+isCJK :: Char -> Bool
+isCJK c = any (\(start, end) -> c >= start && c <= end) cjkRanges
+  where
+    cjkRanges =
+      [ ('\x2e80', '\x2eff'),
+        ('\x2f00', '\x2fdf'),
+        ('\x3040', '\x309f'),
+        ('\x30a0', '\x30fa'),
+        ('\x30fc', '\x30ff'),
+        ('\x3100', '\x312f'),
+        ('\x3200', '\x32ff'),
+        ('\x3400', '\x4dbf'),
+        ('\x4e00', '\x9fff'),
+        ('\xf900', '\xfaff')
+      ]
+
+convertToFullwidth :: Char -> Char
+convertToFullwidth c =
+  case c of
+    ':' -> '：'
+    '.' -> '。'
+    '~' -> '～'
+    '!' -> '！'
+    '?' -> '？'
+    ',' -> '，'
+    ';' -> '；'
+    '\"' -> '”'
+    '\'' -> '’'
+    _ -> c
+
+-- A parser that matches a single CJK character
+cjkChar :: Parser Char
+cjkChar = satisfy isCJK
+
+-- use python.py as reference for these rules
+
+fullwidthCJKsymCJK :: Rule
+fullwidthCJKsymCJK = do
+  lcjk <- cjkChar
+  _ <- many (char ' ')
+  sym <- try (some (char ':')) <|> count 1 (char '.')
+  _ <- many (char ' ')
+  rcjk <- cjkChar
+  let transformedsym = map convertToFullwidth sym
+  return $ T.pack $ [lcjk] ++ transformedsym ++ [rcjk]
+
+fullwidthCJKsym :: Rule
+fullwidthCJKsym = do
+  cjk <- cjkChar
+  _ <- many (char ' ')
+  sym <- some $ oneOf ("~!?,;" :: [Char])
+  _ <- many (char ' ')
+  let transformedsym = T.pack $ map convertToFullwidth sym
+  return $ T.pack [cjk] <> transformedsym
+
+dotsCJK :: Rule
+dotsCJK = do
+  dots <- chunk "..." <|> chunk "…"
+  cjk <- cjkChar
+  return $ dots <> T.pack (" " ++ [cjk])
+
+fixCJKcolAN :: Rule
+fixCJKcolAN = do
+  cjk <- cjkChar
+  _ <- char ':'
+  an <- alphanumericChar
+  return $ T.pack $ [cjk] ++ "：" ++ [an]
+
+-- quotes
+-- seems confusing ...
+quotesym :: [Char]
+quotesym = "'`\x05f4\""
+
+cjkquote :: Rule
+cjkquote = do
+  cjk <- cjkChar
+  quote <- oneOf quotesym
+  return $ T.pack $ [cjk] ++ " " ++ [quote]
+
+quoteCJK :: Rule
+quoteCJK = do
+  quote <- oneOf quotesym
+  cjk <- cjkChar
+  return $ T.pack $ [quote] ++ " " ++ [cjk]
+
+fixQuote :: Rule
+fixQuote = do
+  openQuotes <- T.pack <$> some (oneOf quotesym)
+  _ <- many spaceChar
+  content <- T.pack <$> someTill anySingle (lookAhead $ some (oneOf quotesym))
+  closeQuotes <- T.pack <$> some (oneOf quotesym)
+  return $ openQuotes <> T.strip content <> closeQuotes
+
+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 <|> alphanumericChar
+  _ <- some spaceChar
+  _ <- chunk "'s"
+  return $ T.pack $ pre : "'s"
+
+-- hash
+hashANSCJKhash :: Rule
+hashANSCJKhash = do
+  cjk1 <- cjkChar
+  _ <- char '#'
+  mid <- some cjkChar
+  _ <- char '#'
+  cjk2 <- cjkChar
+  return $ T.pack $ [cjk1] ++ " #" ++ mid ++ "# " ++ [cjk2]
+
+cjkhash :: Rule
+cjkhash = do
+  cjk <- cjkChar
+  _ <- char '#'
+  _ <- lookAhead $ anySingleBut ' '
+  return $ T.pack $ cjk : " #"
+
+hashcjk :: Rule
+hashcjk = do
+  _ <- char '#'
+  _ <- lookAhead $ anySingleBut ' '
+  cjk <- cjkChar
+  return $ T.pack $ "# " ++ [cjk]
+
+-- operators
+cjkOPTan :: Rule
+cjkOPTan = do
+  cjk <- cjkChar
+  opt <- oneOf ("+-=*/&|<>%" :: [Char])
+  an <- alphanumericChar
+  return $ T.pack [cjk, ' ', opt, ' ', an]
+
+anOPTcjk :: Rule
+anOPTcjk = do
+  an <- alphanumericChar
+  opt <- oneOf ("+-=*/&|<>%" :: [Char])
+  cjk <- cjkChar
+  return $ T.pack [an, ' ', opt, ' ', cjk]
+
+-- slash/bracket rules are not implemented
+
+-- CJK and alphanumeric without space
+
+cjkans :: Rule
+cjkans = do
+  cjk <- cjkChar
+  _ <- lookAhead (alphanumericChar <|> oneOf ("@$%^&*-+\\=|/" :: [Char]))
+  return $ T.pack [cjk, ' ']
+
+anscjk :: Rule
+anscjk = do
+  an <- alphanumericChar <|> oneOf ("~!$%^&*-+\\=|;:,./?" :: [Char])
+  _ <- lookAhead cjkChar
+  return $ T.pack [an, ' ']
+
+-- rule set, the order matters
+recursiveRules :: RuleSet
+recursiveRules = [fullwidthCJKsymCJK, fullwidthCJKsym]
+
+onepassRules :: RuleSet
+onepassRules =
+  [ dotsCJK,
+    fixCJKcolAN,
+    cjkquote,
+    quoteCJK,
+    fixQuote,
+    cjkpossessivequote,
+    -- singlequoteCJK,
+    fixPossessivequote,
+    hashANSCJKhash,
+    cjkhash,
+    -- hashcjk,
+    anscjk,
+    cjkans,
+    empty -- a dummy rule
+  ]
+
+pangu :: Text -> Text
+pangu input = applyRules onepassRules $ applyRulesRecursively recursiveRules input

test/Main.hs

@@ -1,21 +1,28 @@
 {-# LANGUAGE OverloadedStrings #-}
 module Main (main) where
-import MyLib
+import Pangu
 import Test.Hspec
 
 
 main :: IO ()
 main = hspec $ do
-  describe "MyLib.cjksym(cjk)" $ do
+  describe "Pangu.cjksym(cjk)" $ do
     it "converts symbols to fullwidth" $ do
-      applyRules myRules "你 : 好" `shouldBe` "你：好"
+      pangu "你 : 好" `shouldBe` "你：好"
-      applyRules myRules "你.好" `shouldBe` "你。好"
+      pangu "你.好" `shouldBe` "你。好"
-      applyRules myRules "你:好:他" `shouldBe` "你：好：他"
+      pangu "你:好:他" `shouldBe` "你：好：他"
-      applyRules myRules "你   ? 好" `shouldBe` "你？好"
+      pangu "你   ? 好" `shouldBe` "你？好"
-      applyRules myRules "你…好" `shouldBe` "你… 好"
+      pangu "你…好" `shouldBe` "你… 好"
-      applyRules myRules "你...好" `shouldBe` "你... 好"
+      pangu "你...好" `shouldBe` "你... 好"
-      applyRules myRules "你:0" `shouldBe` "你：0"
+      pangu "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
+      pangu "我说:\" 他说:'你好'\"" `shouldBe` "我说:\"他说:' 你好 '\""
-      applyRules myRules "\'你好\'" `shouldBe` "\'你好\'"
+      -- pangu "'你好'" `shouldBe` "' 你好'"  -- strange behavior
-      applyRules myRules "你\'hello\'" `shouldBe` "你 \'hello\'"
+      pangu "你'hello'" `shouldBe` "你 'hello'"
+      pangu "我 's " `shouldBe` "我's "
+    it "fixes hash" $ do
+      pangu "你好#测试#世界" `shouldBe` "你好 #测试# 世界"
+    it "add spaces" $ do
+      pangu "你好and世界" `shouldBe` "你好 and 世界"
+      pangu "當你凝視著bug，bug也凝視著你" `shouldBe` "當你凝視著 bug，bug 也凝視著你"
+      pangu "與PM戰鬥的人,應當小心自己不要成為PM" `shouldBe` "與 PM 戰鬥的人，應當小心自己不要成為 PM"