Fixes https://github.com/camp-zju/geonode/issues/36

geonode/utils.py

@@ -933,8 +933,16 @@ def check_shp_columnnames(layer):
                 charset)
 
             if not a.match(field_name):
-                new_field_name = custom_slugify(field_name)
-
+                # once the field_name contains Chinese, to use slugify_zh
+                has_ch = False
+                for ch in field_name:
+                    if u'\u4e00' <= ch <= u'\u9fff':
+                        has_ch=True
+                        break
+                if has_ch == True:
+                    new_field_name = slugify_zh(field_name, separator='_')
+                else:
+                    new_field_name = custom_slugify(field_name)
                 if not b.match(new_field_name):
                     new_field_name = '_' + new_field_name
                 j = 0
@@ -1302,3 +1310,37 @@ def chmod_tree(dst, permissions=0o777):
         for dirname in dirnames:
             path = os.path.join(dirpath, dirname)
             os.chmod(path, permissions)
+
+
+def slugify_zh(text, separator='_'):
+    """
+    Make a slug from the given text, which is simplified from slugify.
+    It remove the other args and do not convert Chinese into Pinyin
+    :param text (str): initial text
+    :param separator (str): separator between words
+    :return (str):
+    """
+
+    QUOTE_PATTERN = re.compile(r'[\']+')
+    ALLOWED_CHARS_PATTERN = re.compile(u'[^\u4e00-\u9fa5a-z0-9]+')
+    DUPLICATE_DASH_PATTERN = re.compile('-{2,}')
+    NUMBERS_PATTERN = re.compile('(?<=\d),(?=\d)')
+    DEFAULT_SEPARATOR = '-'
+
+    if not isinstance(text, types.UnicodeType):
+        text = unicode(text, 'utf-8', 'ignore')
+    # replace quotes with dashes - pre-process
+    text = QUOTE_PATTERN.sub(DEFAULT_SEPARATOR, text)
+    # make the text lowercase
+    text = text.lower()
+    # remove generated quotes -- post-process
+    text = QUOTE_PATTERN.sub('', text)
+    # cleanup numbers
+    text = NUMBERS_PATTERN.sub('', text)
+    # replace all other unwanted characters
+    text = re.sub(ALLOWED_CHARS_PATTERN, DEFAULT_SEPARATOR, text)
+    # remove redundant
+    text = re.sub(DUPLICATE_DASH_PATTERN, DEFAULT_SEPARATOR, text).strip(DEFAULT_SEPARATOR)
+    if separator != DEFAULT_SEPARATOR:
+        text = text.replace(DEFAULT_SEPARATOR, separator)
+    return text