Merge pull request #228 from chinapnr/v1.1.10_develop

V1.1.10 develop

docs/change_log.rst

@@ -1,5 +1,10 @@
 更新记录
 ===========================
+2019.4.30 v1.1.10
+---------------------------
+* `#226 <https://github.com/chinapnr/fishbase/issues/226>`_, file, add function :meth:`fish_file.get_file_encoding`, doc and unittest;
+
+
 2019.4.15 v1.1.9
 ---------------------------
 * `#222 <https://github.com/chinapnr/fishbase/issues/222>`_, common, edit function :meth:`fish_logger.conf_as_dict`, optimize

docs/conf.py

@@ -58,9 +58,9 @@
 # built documents.
 #
 # The short X.Y version.
-version = '1.1.9'
+version = '1.1.10'
 # The full version, including alpha/beta/rc tags.
-release = '1.1.9'
+release = '1.1.10'
 
 # The language for content autogenerated by Sphinx. Refer to documentation
 # for a list of supported languages.

docs/fish_file.rst

@@ -4,6 +4,7 @@
 .. autosummary::
     fish_file.check_sub_path_create
     fish_file.get_abs_filename_with_sub_path
+    fish_file.get_file_encoding
 
 .. automodule:: fish_file
    :members:

fishbase/__init__.py

@@ -27,4 +27,4 @@
 from .fish_project import *
 from .fish_random import *
 
-__version__ = '1.1.9'  # type: str
+__version__ = '1.1.10'  # type: str

fishbase/fish_file.py

@@ -9,6 +9,7 @@
 
 # 2017.1.8 v1.0.9 created
 
+import chardet
 import pathlib
 
 
@@ -140,6 +141,50 @@ def check_sub_path_create(sub_path):
         return False, True
 
 
+# 2019.4.1 v1.1.10 edit by Hu Jun, #226
+def get_file_encoding(file_path):
+
+    """
+    获取给定文件的编码；
+
+    :param:
+        * file_path: (string) 文件的完整路径
+    :return:
+        * file_encoding (string)，文件编码
+
+    举例如下::
+
+        print('--- get_file_encoding demo ---')
+        result = get_file_encoding(__file__)
+        print(result)
+        print('---')
+
+    输出结果::
+
+        --- get_file_encoding demo ---
+        utf-8
+        ---
+
+        """
+    # 判断给定路径是否是一个文件
+    flag = pathlib.Path(file_path).is_file()
+    if not flag:
+        raise RuntimeError('file {}, does not exist'.format(file_path))
+
+    # 读入文件
+    with open(file_path, 'rb') as f:
+        buf = f.read()
+        # 获取文件信息
+        file_info = chardet.detect(buf)
+
+        encoding = file_info['encoding']
+        if encoding.startswith(('utf-8', 'UTF-8')):
+            file_encoding = encoding
+        else:
+            file_encoding = 'GB2312'
+        return file_encoding
+
+
 # 生成使用模块时的下一级路径某文件的完整文件名
 # 2016.5.18 create by David Yi
 # 2017.1.8 v1.0.9 edit the function name, #19004

requirements.txt

@@ -4,3 +4,4 @@ pytest==4.0.2
 pytest-cov
 coveralls
 pyyaml
+chardet

test/test_file.py

@@ -3,10 +3,13 @@
 # 2018.5.28 create by David Yi
 
 import os
+import pytest
 from fishbase.fish_file import *
 
 # 定义当前路径
 current_path = os.path.dirname(os.path.abspath(__file__))
+# 定义配置文件名
+test_filename = os.path.join(current_path, 'test_gb2312_encoding.txt')
 
 
 # 2018.5.28 v1.0.13 #19040,#19042, create by David Yi, fish_file unittest
@@ -67,3 +70,14 @@ def test_check_sub_path_create_01(self):
 
         # 删除临时文件
         os.rmdir(abs_path)
+
+    # tc for get_file_encoding()
+    def test_get_file_encoding_01(self):
+        assert get_file_encoding(__file__) == 'utf-8'
+        # 应该读不到返回的 dict 内容
+        with pytest.raises(RuntimeError):
+            temp_file = __file__ + '.for_test'
+            get_file_encoding(temp_file)
+
+        gbk_file_path = test_filename
+        assert get_file_encoding(gbk_file_path) == 'GB2312'

test/test_gb2312_encoding.txt

@@ -0,0 +1 @@
+this is a test file for fish_file.get_file_encoding