Added optional parameter to API to allow for preservation of certain characters.

.gitignore

@@ -1,3 +1,7 @@
+# Custom
+.mypy_cache/*
+.vscode/*
+
 .coverage
 *.sw[op]
 *.~
@@ -9,3 +13,5 @@ docs/_build
 hanziconv.egg-info
 htmlcov
 MANIFEST
+
+.cache/v/cache/lastfailed

README.rst

@@ -98,7 +98,7 @@ Python API
 Example
 *******
 
-.. code-block:: pycon
+.. code-block:: python
 
     >>> from hanziconv import HanziConv
     >>> print(HanziConv.toSimplified('繁簡轉換器'))
@@ -138,4 +138,3 @@ License
 
 The character map used in this module is based on the Multi-function
 Chinese Character Database developed by Chinese University of Hong Kong.
-

hanziconv/charmap.py

@@ -168,3 +168,6 @@
 
 simplified_charmap = cuhk_simplified + extra_simplified
 traditional_charmap = cuhk_traditional + extra_traditional
+
+simp_to_trad = {s: t for s, t in zip(simplified_charmap, traditional_charmap)}
+trad_to_simp = {t: s for s, t in zip(simplified_charmap, traditional_charmap)}

hanziconv/hanziconv.py

@@ -28,8 +28,10 @@ class by doing:
     >>> from hanziconv import HanziConv
 """
 
-import os
-from .charmap import simplified_charmap, traditional_charmap
+from .charmap import (simplified_charmap,
+                      traditional_charmap,
+                      simp_to_trad,
+                      trad_to_simp)
 
 
 class HanziConv(object):
@@ -39,7 +41,7 @@ class HanziConv(object):
     __simplified_charmap = simplified_charmap
 
     @classmethod
-    def __convert(cls, text, toTraditional=True):
+    def __convert(cls, text, toTraditional=True, custom_mapping=None):
         """Convert `text` to Traditional characters if `toTraditional` is
         True, else convert to simplified characters
 
@@ -51,37 +53,41 @@ def __convert(cls, text, toTraditional=True):
         if isinstance(text, bytes):
             text = text.decode('utf-8')
 
-        fromMap = cls.__simplified_charmap
-        toMap = cls.__traditional_charmap
+        mapper = simp_to_trad
         if not toTraditional:
-            fromMap = cls.__traditional_charmap
-            toMap = cls.__simplified_charmap
+            mapper = trad_to_simp
+
+        if custom_mapping:
+            assert isinstance(custom_mapping, dict), \
+                'custom_mapping should be a dictionary'
+            mapper.update(custom_mapping)
 
         final = []
         for c in text:
-            index = fromMap.find(c)
-            if index != -1:
-                final.append(toMap[index])
+            if c in mapper.keys():
+                final.append(mapper[c])
             else:
                 final.append(c)
         return ''.join(final)
 
     @classmethod
-    def toSimplified(cls, text):
+    def toSimplified(cls, text, custom_mapping=None):
         """Convert `text` to simplified character string.  Assuming text is
         traditional character string
 
-        :param text:  text to convert
-        :returns:     converted UTF-8 characters
+        :param text:           text to convert
+        :param custom_mapping: A dictionary of custom mappings to override
+                               hanziconv's defaults.
+        :returns:              converted UTF-8 characters
 
         >>> from hanziconv import HanziConv
         >>> print(HanziConv.toSimplified('繁簡轉換器'))
         繁简转换器
         """
-        return cls.__convert(text, toTraditional=False)
+        return cls.__convert(text, toTraditional=False, custom_mapping=custom_mapping)
 
     @classmethod
-    def toTraditional(cls, text):
+    def toTraditional(cls, text, custom_mapping=None):
         """Convert `text` to traditional character string.  Assuming text is
         simplified character string
 
@@ -91,25 +97,30 @@ def toTraditional(cls, text):
         >>> from hanziconv import HanziConv
         >>> print(HanziConv.toTraditional('繁简转换器'))
         繁簡轉換器
+        >>> print(HanziConv.toTraditional('祢是我的荣耀', custom_mapping={'祢': '祢'}))
+        祢是我的榮耀
         """
-        return cls.__convert(text, toTraditional=True)
+        return cls.__convert(text, toTraditional=True, custom_mapping=custom_mapping)
 
     @classmethod
-    def same(cls, text1, text2):
+    def same(cls, text1, text2, custom_mapping=None):
         """Return True if text1 and text2 meant literally the same, False
         otherwise
 
         :param text1: string to compare to ``text2``
         :param text2: string to compare to ``text1``
-        :returns:     **True**  -- ``text1`` and ``text2`` are the same in meaning,
+        :returns:     **True**  -- ``text1`` and ``text2`` are the same in \
+            meaning,
                       **False** -- otherwise
 
         >>> from hanziconv import HanziConv
         >>> print(HanziConv.same('繁简转换器', '繁簡轉換器'))
         True
+        >>> print(HanziConv.same('祢是我的荣耀', '祢是我的榮耀', custom_mapping={'祢': '祢'}))
+        True
         """
-        t1 = cls.toSimplified(text1)
-        t2 = cls.toSimplified(text2)
+        t1 = cls.toSimplified(text1, custom_mapping=custom_mapping)
+        t2 = cls.toSimplified(text2, custom_mapping=custom_mapping)
         return t1 == t2
 
 

hanziconv/specialnoun.py

@@ -66,4 +66,3 @@
     for data in data1, data2:
         for d in textwrap.dedent(data).strip().split(' '):
             s2t_exceptions[key].append(d)
-

requirements.txt

@@ -0,0 +1,3 @@
+requests==2.18.4
+beautifulsoup4==4.6.0
+setuptools==36.3.0

tests/test_convert.py

@@ -0,0 +1,19 @@
+import pytest
+from hanziconv import HanziConv
+
+
+convert = HanziConv.toSimplified
+
+
+def test_convert_custom_mapping():
+    custom_mapping = {
+        '祢': '祢',
+        '面': '面',
+        '里': '裡',
+        '傢': '家',
+    }
+
+    text = '住在祢傢里面'
+    expected = '住在祢家裡面'
+
+    assert convert(text, custom_mapping=custom_mapping) == expected