Support metadata for markdown cells in 'light' format #66

jupytext/cell_reader.py

@@ -140,7 +140,10 @@ def options_to_metadata(self, options):
     def find_code_cell_end(self, lines):
         """Given that this is a code cell, return position of
         end of cell marker, and position of next cell start"""
-        self.cell_type = 'code'
+        if self.metadata and 'cell_type' in self.metadata:
+            self.cell_type = self.metadata.pop('cell_type')
+        else:
+            self.cell_type = 'code'
         parser = StringParser(self.language or self.default_language)
         for i, line in enumerate(lines):
             # skip cell header

jupytext/cell_to_text.py

@@ -27,7 +27,7 @@ def comment(lines, prefix):
     return [prefix + ' ' + line if line else prefix for line in lines]
 
 
-class BaseCellExporter:
+class BaseCellExporter(object):
     """A class that represent a notebook cell as text"""
     prefix = None
 
@@ -143,6 +143,13 @@ class LightScriptCellExporter(BaseCellExporter):
     """A class that represent a notebook cell as a Python or Julia script"""
     prefix = '#'
 
+    def is_code(self):
+        # Treat markdown cells with metadata as code cells (#66)
+        if self.cell_type == 'markdown' and self.metadata:
+            self.metadata['cell_type'] = self.cell_type
+            return True
+        return super(LightScriptCellExporter, self).is_code()
+
     def code_to_text(self):
         """Return the text representation of a code cell"""
         active = is_active(self.ext, self.metadata)

jupytext/formats.py

@@ -67,12 +67,14 @@ def __init__(self,
             header_prefix='#',
             cell_reader_class=LightScriptCellReader,
             cell_exporter_class=LightScriptCellExporter,
+            # Version 1.3 on 2018-09-19 - jupytext v0.7.0 : Metadata are
+            # allowed for all cell types (and then include 'cell_type')
             # Version 1.2 on 2018-09-05 - jupytext v0.6.3 : Metadata bracket
             # can be omitted when empty, if previous line is empty #57
             # Version 1.1 on 2018-08-25 - jupytext v0.6.0 : Cells separated
             # with one blank line #38
             # Version 1.0 on 2018-08-22 - jupytext v0.5.2 : Initial version
-            current_version_number='1.2',
+            current_version_number='1.3',
             min_readable_version_number='1.1') for ext in ['.jl', '.py']] + \
     [
         NotebookFormatDescription(
@@ -118,6 +120,7 @@ def guess_format(text, ext):
     if ext in ['.jl', '.py', '.R']:
         twenty_dash = ''.join(['#'] * 20)
         double_percent = '# %%'
+        double_percent_and_space = double_percent + ' '
         twenty_dash_count = 0
         double_percent_count = 0
 
@@ -127,7 +130,10 @@ def guess_format(text, ext):
             if parser.is_quoted():
                 continue
 
-            if line.startswith(double_percent):
+            # Don't count escaped Jupyter magics (no space between
+            # %% and command) as cells
+            if line == double_percent or \
+                    line.startswith(double_percent_and_space):
                 double_percent_count += 1
 
             if line.startswith(twenty_dash):