WIP | Fix handling for table_style to prevent repeated re-splitting

html4docx/h4d.py

@@ -58,6 +58,28 @@ def __init__(self):
         ]
         self.table_style = DEFAULT_TABLE_STYLE
 
+    def set_table_style(self, new_table_style: str) -> None:
+        """Set table style for all tables in the final DOCX file. The full list of style options can be found at
+        https://python-docx.readthedocs.io/en/latest/user/styles-understanding.html#table-styles-in-default-template
+
+        Args:
+            new_table_style (str): DOCX-supported table style string, i.e. "Colorful Grid Accent 2".
+
+        Raises:
+            ValueError: Thrown if input string does not contain spaces but fails the regex for splitting by capitals.
+        """
+        # If the input table style string does not contain spaces, i.e. "TableGrid", split with spaces by capitals.
+        if " " not in new_table_style:
+            try:
+                # Fixed 'style lookup by style_id is deprecated.'
+                # https://stackoverflow.com/a/29567907/17274446
+                self.table_style = ' '.join(re.findall(r'[A-Z][^A-Z]*', new_table_style))
+                return
+            except KeyError as key_error:
+                raise ValueError(f"Unable to apply style {self.table_style}.") from key_error
+        # Otherwise, save without reformatting input.
+        self.table_style = new_table_style
+
     def set_initial_attrs(self, document=None):
         self.tags = {
             'span': [],
@@ -330,12 +352,7 @@ def handle_table(self, current_attrs):
         self.table = self.doc.add_table(rows, cols)
 
         if self.table_style:
-            try:
-                # Fixed 'style lookup by style_id is deprecated.'
-                # https://stackoverflow.com/a/29567907/17274446
-                self.table_style = ' '.join(re.findall(r'[A-Z][^A-Z]*', self.table_style))
-            except KeyError as e:
-                raise ValueError(f"Unable to apply style {self.table_style}.") from e
+            self.table.style = self.table_style
 
         rows = self.get_table_rows(table_soup)
         cell_row = 0

setup.py

@@ -40,5 +40,9 @@
     install_requires = [
         'python-docx>=1.1.0',
         'beautifulsoup4>=4.12.2'
+    ],
+    tests_require = [
+        'pytest>=8.1.1',
+        'pytest-cov>=4.1.0'
     ]
 )

tests/unittests/test_h4d.py

@@ -0,0 +1,44 @@
+"""Unit test suite that covers html4docx/h4d.py."""
+
+import os
+import pathlib
+from html4docx.h4d import HtmlToDocx
+from docx import Document
+
+
+def test_h4d_set_table_style() -> None:
+    """Test that covers set_table_style() in h4d.py."""
+    test_docx_filename = "test_h4d_set_table_style.docx"
+
+    with open(f"{str(pathlib.Path(__file__).parent.resolve())}/heyo.html", "r", encoding="utf-8") as heyo_html:
+        test_html_string = heyo_html.read()
+
+    # Test where table_style and table.style are both None.
+    parser = HtmlToDocx()
+    document = Document()
+    parser.add_html_to_document(test_html_string, document)
+    document.save(test_docx_filename)
+    assert not parser.table_style
+    assert hasattr(parser, "table")
+    assert not parser.table
+
+    # Test where table_style is set to valid table Style without spaces but parsed to correct syntax.
+    parser = HtmlToDocx()
+    parser.set_table_style("TableGrid")
+    assert parser.table_style == "Table Grid"
+    document = Document()
+    parser.add_html_to_document(test_html_string, document)
+    document.save(test_docx_filename)
+    assert parser.table_style == "Table Grid"
+
+    # Test where table_style is set to valid table Style with spaces.
+    parser = HtmlToDocx()
+    parser.set_table_style("Colorful Grid Accent 2")
+    assert parser.table_style == "Colorful Grid Accent 2"
+    document = Document()
+    parser.add_html_to_document(test_html_string, document)
+    document.save(test_docx_filename)
+    assert parser.table_style == "Colorful Grid Accent 2"
+
+    # Cleanup test file.
+    os.remove(test_docx_filename)