Preserve the default lexer in jupytext.default_lexer

Warn if there are multiple lexers

Author: mwouts

src/jupytext/jupytext.py

@@ -34,13 +34,7 @@
     set_main_and_cell_language,
 )
 from .metadata_filter import filter_metadata, update_metadata_filters
-from .myst import (
-    MYST_FORMAT_NAME,
-    language_to_lexer,
-    myst_extensions,
-    myst_to_notebook,
-    notebook_to_myst,
-)
+from .myst import MYST_FORMAT_NAME, myst_extensions, myst_to_notebook, notebook_to_myst
 from .pandoc import md_to_notebook, notebook_to_md
 from .pep8 import pep8_lines_between_cells
 from .quarto import notebook_to_qmd, qmd_to_notebook
@@ -242,10 +236,7 @@ def writes(self, nb, metadata=None, **kwargs):
                 "pygments_lexer", None
             )
             if default_lexer is None:
-                language = metadata.get("kernelspec", {}).get(
-                    "language"
-                ) or metadata.get("jupytext", {}).get("main_language")
-                default_lexer = language_to_lexer(language)
+                default_lexer = metadata.get("jupytext", {}).get("default_lexer", None)
             return notebook_to_myst(
                 self.filter_notebook(nb, metadata),
                 default_lexer=default_lexer,

src/jupytext/myst.py

@@ -5,7 +5,6 @@
 import json
 import re
 import warnings
-from collections import Counter
 from textwrap import dedent
 
 import nbformat as nbf
@@ -255,20 +254,6 @@ def read_cell_metadata(token, cell_index):
     return metadata
 
 
-def lexer_to_language(lexer):
-    """Convert a lexer name to a language name."""
-    if lexer == "ipython3":
-        return "python"
-    return None
-
-
-def language_to_lexer(language):
-    """Convert a language name to a lexer name."""
-    if language == "python":
-        return "ipython3"
-    return None
-
-
 def myst_to_notebook(
     text,
     code_directive=CODE_DIRECTIVE,
@@ -293,6 +278,7 @@ def myst_to_notebook(
     tokens = get_parser().parse(text + "\n")
     lines = text.splitlines()
     md_start_line = 0
+    default_lexer = None
 
     # get the document metadata
     metadata_nb = {}
@@ -325,8 +311,6 @@ def _flush_markdown(start_line, token, md_metadata):
     nesting_level = 0
     md_metadata = {}
 
-    languages = Counter()
-
     for token in tokens:
         nesting_level += token.nesting
 
@@ -339,9 +323,14 @@ def _flush_markdown(start_line, token, md_metadata):
             options, body_lines = read_fenced_cell(token, len(notebook.cells), "Code")
             assert token.info.startswith(code_directive)
             lexer = token.info[len(code_directive) :].strip()
-            language = lexer_to_language(lexer)
-            if language:
-                languages[language] += 1
+            if lexer:
+                if not default_lexer:
+                    default_lexer = lexer
+                elif lexer != default_lexer:
+                    warnings.warn(
+                        f"All code cells in a MyST notebook must have the same language: {lexer}!={default_lexer}"
+                    )
+
             meta = nbf.from_dict(options)
             source_map.append(token.map[0] + 1)
             notebook.cells.append(
@@ -371,14 +360,11 @@ def _flush_markdown(start_line, token, md_metadata):
     if add_source_map:
         notebook.metadata["source_map"] = source_map
 
-    if "kernelspec" not in notebook.metadata:
-        for language, count in languages.most_common(1):
-            if count:
-                jupytext_metadata = {"main_language": language}
-                if not notebook.metadata:
-                    jupytext_metadata["notebook_metadata_filter"] = "-all"
-                notebook.metadata["jupytext"] = jupytext_metadata
-                break
+    if "kernelspec" not in notebook.metadata and default_lexer:
+        jupytext_metadata = {"default_lexer": default_lexer}
+        if not notebook.metadata:
+            jupytext_metadata["notebook_metadata_filter"] = "-all"
+        notebook.metadata["jupytext"] = jupytext_metadata
 
     return notebook
 

tests/data/notebooks/outputs/myst_to_ipynb/fenced_code_vs_code_cells.ipynb

@@ -43,7 +43,7 @@
  ],
  "metadata": {
   "jupytext": {
-   "main_language": "python",
+   "default_lexer": "ipython3",
    "notebook_metadata_filter": "-all"
   }
  },

tests/data/notebooks/outputs/myst_to_ipynb/reference_link.ipynb

@@ -35,7 +35,7 @@
  ],
  "metadata": {
   "jupytext": {
-   "main_language": "python",
+   "default_lexer": "ipython3",
    "notebook_metadata_filter": "-all"
   }
  },

tests/functional/simple_notebooks/test_ipynb_to_myst.py

@@ -229,7 +229,10 @@ async def test_myst_representation_same_cli_or_contents_manager(
         line for line in text_api.splitlines() if line.startswith("```{code-cell")
     }
 
-    assert code_cells == {"```{code-cell} ipython3"}
+    if language_info == "std":
+        assert code_cells == {"```{code-cell} ipython3"}
+    else:
+        assert code_cells == {"```{code-cell}"}
 
     # We get the same file with the command line jupytext
     tmpdir.mkdir("cli").join("notebook.ipynb").write(json.dumps(nb))

tests/unit/test_markdown_in_code_cells.py

@@ -56,7 +56,7 @@ def test_triple_backticks_in_code_cell(
 def test_triple_backticks_in_code_cell_myst(
     no_jupytext_version_number,
     nb=new_notebook(
-        metadata={"main_language": "python"},
+        metadata={"jupytext": {"default_lexer": "ipython3"}},
         cells=[
             new_code_cell(
                 '''a = """
@@ -69,7 +69,7 @@ def test_triple_backticks_in_code_cell_myst(
     ),
     text='''---
 jupytext:
-  main_language: python
+  default_lexer: ipython3
 ---
 
 ````{code-cell} ipython3