[BUG] Fix handling of leading whitespaces in code blocks

jupyter · Jan 6, 2025 · 5f11975 · 5f11975
1 parent 5f508eb
commit 5f11975
Show file tree

Hide file tree

Showing 4 changed files with 38 additions and 3 deletions.
diff --git a/nbconvert/filters/highlight.py b/nbconvert/filters/highlight.py
@@ -179,7 +179,7 @@ def _pygments_highlight(source, output_formatter, language="ipython", metadata=N
 
     if lexer is None:
         try:
-            lexer = get_lexer_by_name(language, stripall=True)
+            lexer = get_lexer_by_name(language, stripall=False)
         except ClassNotFound:
             warn("No lexer found for language %r. Treating as plain text." % language, stacklevel=2)
             from pygments.lexers.special import TextLexer

diff --git a/nbconvert/filters/markdown_mistune.py b/nbconvert/filters/markdown_mistune.py
@@ -317,7 +317,7 @@ def block_code(self, code: str, info: Optional[str] = None) -> str:
             try:
                 if info.strip().split(None, 1):
                     lang = info.strip().split(maxsplit=1)[0]
-                    lexer = get_lexer_by_name(lang, stripall=True)
+                    lexer = get_lexer_by_name(lang, stripall=False)
             except ClassNotFound:
                 code = f"{lang}\n{code}"
                 lang = None

diff --git a/pyproject.toml b/pyproject.toml
@@ -224,7 +224,6 @@ ignore = [
   "T201",   # `print` found
   "RUF012", # Mutable class attributes should be annotated
   "UP031",  # Use format specifiers instead of percent format
-
 ]
 unfixable = [
   "T201",   # Don't touch print statements
@@ -253,6 +252,7 @@ unfixable = [
 "nbconvert/__init__.py" = ["F401", "F403"]
 # PLR2004 Magic value used in comparison
 "nbconvert/filters/ansi.py" = ["PLR2004"]
+"tests/exporters/test_html.py" = ["RUF001"]
 
 [tool.interrogate]
 ignore-init-module=true

diff --git a/tests/exporters/test_html.py b/tests/exporters/test_html.py
@@ -262,3 +262,38 @@ def test_language_code_error(self):
         (output, resources) = exporter.from_filename(self._get_notebook())
 
         assert '<html lang="en">' in output
+
+
+def test_syntax_highlight_leading_whitespace():
+    """Test that syntax highlight doesn't strip leading spaces."""
+    nb = v4.reads(r"""
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "id": "29da71a9-ae40-4098-8c3b-31a98e79fc12",
+   "metadata": {},
+   "source": [
+    "```APL\n",
+    "      1+2×⍳3\n",
+    "3 5 7\n",
+    "```\n",
+    "\n",
+    "```\n",
+    "      1+2×⍳3\n",
+    "3 5 7\n",
+    "```"
+   ]
+  }
+ ],
+ "metadata": {},
+ "nbformat": 4,
+ "nbformat_minor": 5
+}
+    """)
+    output, _ = HTMLExporter().from_notebook_node(nb)
+    # Check that the second code block has the leading spaces
+    assert "<pre><code>      1+2×⍳3\n3 5 7\n</code></pre>" in output
+
+    # Check that the APL-formatted code block has the leading spaces
+    assert '<span class="w">      </span>' in output