Merge pull request #5 from cdfmlr/#4_file_decode_error

resolve #4 open file decode error

pyflowchart/__main__.py

@@ -7,21 +7,65 @@
 """
 
 import sys
+import chardet
 import argparse
 
 from pyflowchart.flowchart import Flowchart
 
 
+def detect_decode(file_content: bytes) -> str:
+    """detect_decode detect the encoding of file_content,
+     then decode file_content on the detected encoding.
+
+    If the confidence of detect result is less then 0.9,
+    the UTF-8 will be used to decode. PyFlowchart is
+    designed to convert Python 3 codes into flowcharts.
+    And Python 3 is coding in UTF-8 in default. So only
+    if we can make sure the file is not UTF-8 encoded (
+    i.e. confidence > 0.9) than we will use that no
+    default encoding to decoded it.
+
+    Args:
+        file_content: bytes: binary file content to decode
+
+    Returns:
+        str: decoded content
+    """
+    # detect encoding
+    detect_result = chardet.detect(file_content)
+    # print("DEBUG detect_result =", detect_result)
+
+    encoding = detect_result.get("encoding")
+    confidence = detect_result.get("confidence")
+
+    if confidence < 0.9:
+        encoding = "UTF-8"
+
+    # decode file content by detected encoding
+    try:
+        content = file_content.decode(encoding=encoding)
+    except TypeError:  # TypeError: decode() argument 1 must be str, not None
+        content = file_content.decode()
+
+    return content
+
+
 def main(code_file, field, inner):
-    code = code_file.read()
+    # read file content: binary
+    file_content: bytes = code_file.read()
+    # detect encoding and decode file content by detected encoding
+    code = detect_decode(file_content)
+
     flowchart = Flowchart.from_code(code, field=field, inner=inner)
     print(flowchart.flowchart())
 
 
 if __name__ == '__main__':
     parser = argparse.ArgumentParser(description='Python code to flowchart.')
 
-    parser.add_argument('code_file',  type=argparse.FileType('r'))
+    # code_file: open as binary, detect encoding and decode in main later
+    parser.add_argument('code_file', type=argparse.FileType('rb'))
+
     parser.add_argument('-f', '--field', default="", type=str, help="field to draw flowchart. (e.g. Class.method)")
     parser.add_argument('-i', '--inner', action="store_true", help="parse the body of field")
 

setup.py

@@ -5,7 +5,7 @@
 
 setuptools.setup(
     name='pyflowchart',
-    version='0.1.0',
+    version='0.1.1',
     url='https://github.com/cdfmlr/pyflowchart',
     license='MIT',
     author='CDFMLR',
@@ -23,5 +23,5 @@
         "Topic :: Utilities",
     ],
     python_requires='>=3.6',
-    install_requires=['astunparse'],
+    install_requires=['astunparse', 'chardet'],
 )