refactor: style loading logic in DataFrameHtmlFormatter (apache#1177)

Author: kosiew

python/datafusion/dataframe_formatter.py

@@ -135,9 +135,6 @@ class DataFrameHtmlFormatter:
           session
     """
 
-    # Class variable to track if styles have been loaded in the notebook
-    _styles_loaded = False
-
     def __init__(
         self,
         max_cell_length: int = 25,
@@ -260,23 +257,6 @@ def set_custom_header_builder(self, builder: Callable[[Any], str]) -> None:
         """
         self._custom_header_builder = builder
 
-    @classmethod
-    def is_styles_loaded(cls) -> bool:
-        """Check if HTML styles have been loaded in the current session.
-
-        This method is primarily intended for debugging UI rendering issues
-        related to style loading.
-
-        Returns:
-            True if styles have been loaded, False otherwise
-
-        Example:
-            >>> from datafusion.dataframe_formatter import DataFrameHtmlFormatter
-            >>> DataFrameHtmlFormatter.is_styles_loaded()
-            False
-        """
-        return cls._styles_loaded
-
     def format_html(
         self,
         batches: list,
@@ -315,18 +295,7 @@ def format_html(
         # Build HTML components
         html = []
 
-        # Only include styles and scripts if:
-        # 1. Not using shared styles, OR
-        # 2. Using shared styles but they haven't been loaded yet
-        include_styles = (
-            not self.use_shared_styles or not DataFrameHtmlFormatter._styles_loaded
-        )
-
-        if include_styles:
-            html.extend(self._build_html_header())
-            # If we're using shared styles, mark them as loaded
-            if self.use_shared_styles:
-                DataFrameHtmlFormatter._styles_loaded = True
+        html.extend(self._build_html_header())
 
         html.extend(self._build_table_container_start())
 
@@ -338,7 +307,7 @@ def format_html(
         html.append("</div>")
 
         # Add footer (JavaScript and messages)
-        if include_styles and self.enable_cell_expansion:
+        if self.enable_cell_expansion:
             html.append(self._get_javascript())
 
         # Always add truncation message if needed (independent of styles)
@@ -375,14 +344,20 @@ def format_str(
 
     def _build_html_header(self) -> list[str]:
         """Build the HTML header with CSS styles."""
-        html = []
-        html.append("<style>")
-        # Only include expandable CSS if cell expansion is enabled
-        if self.enable_cell_expansion:
-            html.append(self._get_default_css())
+        default_css = self._get_default_css() if self.enable_cell_expansion else ""
+        script = f"""
+<script>
+if (!document.getElementById('df-styles')) {{
+  const style = document.createElement('style');
+  style.id = 'df-styles';
+  style.textContent = `{default_css}`;
+  document.head.appendChild(style);
+}}
+</script>
+"""
+        html = [script]
         if self.custom_css:
-            html.append(self.custom_css)
-        html.append("</style>")
+            html.append(f"<style>{self.custom_css}</style>")
         return html
 
     def _build_table_container_start(self) -> list[str]:
@@ -570,28 +545,31 @@ def _get_default_css(self) -> str:
     def _get_javascript(self) -> str:
         """Get JavaScript code for interactive elements."""
         return """
-            <script>
-            function toggleDataFrameCellText(table_uuid, row, col) {
-                var shortText = document.getElementById(
-                    table_uuid + "-min-text-" + row + "-" + col
-                );
-                var fullText = document.getElementById(
-                    table_uuid + "-full-text-" + row + "-" + col
-                );
-                var button = event.target;
-
-                if (fullText.style.display === "none") {
-                    shortText.style.display = "none";
-                    fullText.style.display = "inline";
-                    button.textContent = "(less)";
-                } else {
-                    shortText.style.display = "inline";
-                    fullText.style.display = "none";
-                    button.textContent = "...";
-                }
-            }
-            </script>
-        """
+<script>
+if (!window.__df_formatter_js_loaded__) {
+  window.__df_formatter_js_loaded__ = true;
+  window.toggleDataFrameCellText = function (table_uuid, row, col) {
+    var shortText = document.getElementById(
+      table_uuid + "-min-text-" + row + "-" + col
+    );
+    var fullText = document.getElementById(
+      table_uuid + "-full-text-" + row + "-" + col
+    );
+    var button = event.target;
+
+    if (fullText.style.display === "none") {
+      shortText.style.display = "none";
+      fullText.style.display = "inline";
+      button.textContent = "(less)";
+    } else {
+      shortText.style.display = "inline";
+      fullText.style.display = "none";
+      button.textContent = "...";
+    }
+  };
+}
+</script>
+"""
 
 
 class FormatterManager:
@@ -712,24 +690,9 @@ def reset_formatter() -> None:
         >>> reset_formatter()  # Reset formatter to default settings
     """
     formatter = DataFrameHtmlFormatter()
-    # Reset the styles_loaded flag to ensure styles will be reloaded
-    DataFrameHtmlFormatter._styles_loaded = False
     set_formatter(formatter)
 
 
-def reset_styles_loaded_state() -> None:
-    """Reset the styles loaded state to force reloading of styles.
-
-    This can be useful when switching between notebook sessions or
-    when styles need to be refreshed.
-
-    Example:
-        >>> from datafusion.html_formatter import reset_styles_loaded_state
-        >>> reset_styles_loaded_state()  # Force styles to reload in next render
-    """
-    DataFrameHtmlFormatter._styles_loaded = False
-
-
 def _refresh_formatter_reference() -> None:
     """Refresh formatter reference in any modules using it.
 

python/tests/test_dataframe.py

@@ -42,7 +42,6 @@
     configure_formatter,
     get_formatter,
     reset_formatter,
-    reset_styles_loaded_state,
 )
 from datafusion.expr import Window
 from pyarrow.csv import write_csv
@@ -2177,27 +2176,15 @@ def test_html_formatter_shared_styles(df, clean_formatter_state):
     # First, ensure we're using shared styles
     configure_formatter(use_shared_styles=True)
 
-    # Get HTML output for first table - should include styles
     html_first = df._repr_html_()
-
-    # Verify styles are included in first render
-    assert "<style>" in html_first
-    assert ".expandable-container" in html_first
-
-    # Get HTML output for second table - should NOT include styles
     html_second = df._repr_html_()
 
-    # Verify styles are NOT included in second render
+    assert "<script>" in html_first
+    assert "df-styles" in html_first
+    assert "<script>" in html_second
+    assert "df-styles" in html_second
+    assert "<style>" not in html_first
     assert "<style>" not in html_second
-    assert ".expandable-container" not in html_second
-
-    # Reset the styles loaded state and verify styles are included again
-    reset_styles_loaded_state()
-    html_after_reset = df._repr_html_()
-
-    # Verify styles are included after reset
-    assert "<style>" in html_after_reset
-    assert ".expandable-container" in html_after_reset
 
 
 def test_html_formatter_no_shared_styles(df, clean_formatter_state):
@@ -2206,15 +2193,15 @@ def test_html_formatter_no_shared_styles(df, clean_formatter_state):
     # Configure formatter to NOT use shared styles
     configure_formatter(use_shared_styles=False)
 
-    # Generate HTML multiple times
     html_first = df._repr_html_()
     html_second = df._repr_html_()
 
-    # Verify styles are included in both renders
-    assert "<style>" in html_first
-    assert "<style>" in html_second
-    assert ".expandable-container" in html_first
-    assert ".expandable-container" in html_second
+    assert "<script>" in html_first
+    assert "<script>" in html_second
+    assert "df-styles" in html_first
+    assert "df-styles" in html_second
+    assert "<style>" not in html_first
+    assert "<style>" not in html_second
 
 
 def test_html_formatter_manual_format_html(clean_formatter_state):
@@ -2228,20 +2215,15 @@ def test_html_formatter_manual_format_html(clean_formatter_state):
 
     formatter = get_formatter()
 
-    # First call should include styles
     html_first = formatter.format_html([batch], batch.schema)
-    assert "<style>" in html_first
-
-    # Second call should not include styles (using shared styles by default)
     html_second = formatter.format_html([batch], batch.schema)
-    assert "<style>" not in html_second
 
-    # Reset loaded state
-    reset_styles_loaded_state()
-
-    # After reset, styles should be included again
-    html_reset = formatter.format_html([batch], batch.schema)
-    assert "<style>" in html_reset
+    assert "<script>" in html_first
+    assert "<script>" in html_second
+    assert "df-styles" in html_first
+    assert "df-styles" in html_second
+    assert "<style>" not in html_first
+    assert "<style>" not in html_second
 
     # Create a new formatter with shared_styles=False
     local_formatter = DataFrameHtmlFormatter(use_shared_styles=False)
@@ -2250,8 +2232,12 @@ def test_html_formatter_manual_format_html(clean_formatter_state):
     local_html_1 = local_formatter.format_html([batch], batch.schema)
     local_html_2 = local_formatter.format_html([batch], batch.schema)
 
-    assert "<style>" in local_html_1
-    assert "<style>" in local_html_2
+    assert "<script>" in local_html_1
+    assert "<script>" in local_html_2
+    assert "df-styles" in local_html_1
+    assert "df-styles" in local_html_2
+    assert "<style>" not in local_html_1
+    assert "<style>" not in local_html_2
 
 
 def test_fill_null_basic(null_df):