Merge pull request #68 from DHI/bug/allow_all_python_identifiers_as_v…

…ariable_name

Bug/allow all python identifiers as variable name

mikeio1d/result_network/various.py

@@ -4,23 +4,51 @@
 from html import escape
 
 
-def make_proper_variable_name(string, extra_string_before_digit="_"):
+class ValidPythonIdentifierTranslatorTable:
+    """
+    Translator table for replacing invalid python characters with a valid identifier.
+
+    Used as argument to Python's built-in str.translate() method.
+
+    Parameters
+    ----------
+    replacement : str, default = "_"
+       The character to replace invalid characters with.
+    """
+
+    def __init__(self, replacement="_"):
+        self.replacement = replacement
+
+    def __getitem__(self, key):
+        key = chr(key)
+        if key.isdigit() or key.isidentifier():
+            return key
+        else:
+            return self.replacement
+
+
+_translator_table = ValidPythonIdentifierTranslatorTable()
+
+
+def make_proper_variable_name(string: str, extra_string_before_digit="_"):
     """
     Makes a more proper variable name.
 
     It is assumed that the input string never is or after manipulations
     becomes an '_' or an empty string.
     """
-    # Replace all non alpha numeric characters by an underscore.
-    string = re.sub(r"[^a-zA-Z0-9]", "_", string)
+    # Replace all characters that are not valid python identifier by an underscore.
+    string = string.translate(_translator_table)
     # Replace more than two underscores with a single underscore.
     string = re.sub(r"_{2,}", "_", string)
-    # Add an extra string is the string starts with a number.
-    string = extra_string_before_digit + string if string and string[0].isdigit() else string
     # Remove a starting underscore
-    string = string[1:] if string[0] == "_" else string
+    if len(string) > 1:
+        string = string[1:] if string[0] == "_" else string
     # Remove a trailing underscore
-    string = string[:-1] if string[-1] == "_" else string
+    if len(string) > 1:
+        string = string[:-1] if string[-1] == "_" else string
+    # Add an extra string if the string starts with a number.
+    string = extra_string_before_digit + string if string and string[0].isdigit() else string
     return string
 
 

tests/test_various.py

@@ -0,0 +1,40 @@
+import os
+import pytest
+
+from mikeio1d import Res1D
+from mikeio1d.result_network.various import make_proper_variable_name
+
+
+@pytest.fixture
+def test_file_path():
+    test_folder_path = os.path.dirname(os.path.abspath(__file__))
+    # Original file name was Exam6Base.res1d
+    return os.path.join(test_folder_path, "testdata", "Network_chinese.res1d")
+
+
+def test_make_proper_variable_name():
+    mpvn = make_proper_variable_name  # alias
+    assert mpvn("a") == "a"
+    assert mpvn("1") == "_1"
+    assert mpvn("1a") == "_1a"
+    assert mpvn("a1") == "a1"
+    assert mpvn("a1a") == "a1a"
+    assert mpvn("myname??++") == "myname"
+    assert mpvn("myname??++something") == "myname_something"
+    assert mpvn("你好") == "你好"
+    assert mpvn("123你好") == "_123你好"
+
+
+def test_network_attributes_allow_chinese_characters(test_file_path):
+    """Test that network attributes allow chinese characters."""
+    res = Res1D(test_file_path)
+    assert res.nodes.风景
+    assert res.nodes.电脑
+    assert res.nodes.餐厅
+    assert res.nodes.爱情
+    assert res.nodes.音乐
+    assert res.nodes.星期
+    assert res.nodes.医生
+    assert res.nodes.美丽
+    assert res.nodes.学生
+    assert res.nodes.蛋糕

tests/testdata/Network_chinese_source.txt

@@ -0,0 +1,5 @@
+Base model: MIKE+ Examples - Sirius - DWF_Network Scenario
+Updates:
+1. Change the simulation duration to 1 hour.
+2. Change the saving frequency to 10 minutes.
+3. Update the IDs of 10 links to use chinese words.

tests/testdata/__init__.py

@@ -24,6 +24,8 @@ class testdata:
 
     Network_res1d: str = format_path("Network.res1d")
     """A basic urban network file."""
+    Network_res1d_chinese: str = format_path("Network_chinese.res1d")
+    """A basic urban network file with chinese characters for some links."""
     NetworkRiver_res1d: str = format_path("NetworkRiver.res1d")
     """A basic river network file."""
     Catchments_res1d: str = format_path("Catchments.res1d")