Adding encoding specification parameter when opening yaml file

docs/readers.md

@@ -80,6 +80,8 @@ Example:
 
     {{ read_yaml('assets/tables/yaml_table.yml') | add_indentation(spaces=4) }}
 
+The file is read as UTF-8. Use the `encoding` argument for files in another encoding, for example {% raw %}`{{ read_yaml('assets/tables/yaml_table.yml', encoding='cp1251') }}`{% endraw %}.
+
 
 ### `read_table`
 
@@ -210,6 +212,8 @@ Example:
 
     {{ read_raw('assets/tables/markdown_table.md')  | add_indentation(spaces=4) }}
 
+The file is read as UTF-8. Use the `encoding` argument for files in another encoding, for example {% raw %}`{{ read_raw('assets/tables/markdown_table.md', encoding='cp1251') }}`{% endraw %}.
+
 
 ## Macros
 
@@ -269,6 +273,8 @@ Example:
 
     {{ pd_read_yaml('assets/tables/yaml_table.yml').to_markdown(tablefmt="pipe", index=False) | add_indentation(spaces=4) }}
 
+The file is read as UTF-8. Use the `encoding` argument for files in another encoding, for example {% raw %}`{{ pd_read_yaml('assets/tables/yaml_table.yml', encoding='cp1251') }}`{% endraw %}.
+
 
 ### `pd_read_table`
 

src/mkdocs_table_reader_plugin/readers.py

@@ -133,15 +133,17 @@ def read_excel(*args, **kwargs) -> str:
 
 @ParseArgs
 def pd_read_yaml(*args, **kwargs) -> str:
+    encoding = kwargs.pop("encoding", "utf-8")
     json_kwargs = kwargs_in_func(kwargs, pd.json_normalize)
-    with open(args[0]) as f:
+    with open(args[0], encoding=encoding) as f:
         df = pd.json_normalize(yaml.safe_load(f), **json_kwargs)
     return df
 
 @ParseArgs
 def read_yaml(*args, **kwargs) -> str:
+    encoding = kwargs.pop("encoding", "utf-8")
     json_kwargs = kwargs_in_func(kwargs, pd.json_normalize)
-    with open(args[0]) as f:
+    with open(args[0], encoding=encoding) as f:
         df = pd.json_normalize(yaml.safe_load(f), **json_kwargs)
 
     markdown_kwargs = kwargs_not_in_func(kwargs, pd.json_normalize)
@@ -170,7 +172,8 @@ def read_raw(*args, **kwargs) -> str:
     Returns:
         str: file contents
     """
-    with open(args[0]) as f:
+    encoding = kwargs.pop("encoding", "utf-8")
+    with open(args[0], encoding=encoding) as f:
         return f.read()
 
 
@@ -194,4 +197,4 @@ def read_raw(*args, **kwargs) -> str:
     "pd_read_json": pd_read_json,
     "pd_read_feather": pd_read_feather,
 }
-MACROS = {**READERS, **MACRO_ONLY}
+MACROS = {**READERS, **MACRO_ONLY}

tests/fixtures/encoding/assets/tables/cp1251_table.md

@@ -0,0 +1,3 @@
+| Òîâàð  | Öåíà |
+|--------|-----:|
+| Ñûð    |  539956 |

tests/fixtures/encoding/assets/tables/cp1251_table.yml

@@ -0,0 +1,4 @@
+- Òîâàð: Õëåá
+  Öåíà: 531456
+- Òîâàð: Ìîëîêî
+  Öåíà: 80

tests/fixtures/encoding/docs/index.md

@@ -0,0 +1,11 @@
+# Test page
+
+Files that are not UTF-8 encoded can be read by specifying the `encoding`:
+
+## read_yaml
+
+{{ read_yaml('assets/tables/cp1251_table.yml', encoding='cp1251') }}
+
+## read_raw
+
+{{ read_raw('assets/tables/cp1251_table.md', encoding='cp1251') }}

tests/fixtures/encoding/mkdocs.yml

@@ -0,0 +1,6 @@
+site_name: test git_table_reader site
+use_directory_urls: false
+
+plugins:
+    - search
+    - table-reader

tests/test_build.py

@@ -387,3 +387,24 @@ def test_macros_jinja2_syntax(tmp_path):
     contents = page_with_tag.read_text()
     assert re.search(r"531456", contents)
 
+
+def test_non_utf8_encoding(tmp_path):
+    """
+    A project where files are not UTF-8 encoded, and 'encoding' is specified.
+    """
+
+    tmp_proj = setup_clean_mkdocs_folder(
+        "tests/fixtures/encoding/mkdocs.yml", tmp_path
+    )
+
+    result = build_docs_setup(tmp_proj)
+    assert result.exit_code == 0, "'mkdocs build' command failed"
+
+    page_with_tag = tmp_proj / "site/index.html"
+    contents = page_with_tag.read_text(encoding="utf-8")
+    # read_yaml() inserted the cp1251 encoded yaml file
+    assert re.search(r"531456", contents)
+    assert re.search(r"Хлеб", contents)
+    # read_raw() inserted the cp1251 encoded markdown file
+    assert re.search(r"539956", contents)
+    assert re.search(r"Сыр", contents)