Optional end of cell marker for python scripts #31

nbrmd/cells.py

@@ -54,10 +54,26 @@ def code_to_text(self,
                 if options != '':
                     lines.append('#+ ' + options)
             else:
+                # Issue #31:  does the cell ends with a blank line?
+                # Do we find two blank lines in the cell? In that case
+                # we add an end-of-cell marker
+                if (len(source) and _BLANK_LINE.match(source[-1])) or \
+                        code_to_cell(self, source, False)[1] != len(source):
+                    cellend = '-'
+                    while True:
+                        cellend_re = re.compile('^#( )' + cellend + '\s*$')
+                        if len(list(filter(cellend_re.match, source))):
+                            cellend = cellend + '-'
+                        else:
+                            break
+                    metadata['cellend'] = cellend
+
                 options = metadata_to_json_options(metadata)
                 if options != '{}':
                     lines.append('# + ' + options)
             lines.extend(source)
+            if 'cellend' in metadata:
+                lines.append('# ' + metadata['cellend'])
         else:
             lines.extend(self.markdown_escape(
                 code_to_rmd(source, metadata, default_language)))
@@ -232,10 +248,18 @@ def code_to_cell(self, lines, parse_opt):
     else:
         metadata = {}
 
-    # Find end of cell and return
     if self.ext == '.Rmd':
+        end_cell_re = _END_CODE_MD
+    elif 'cellend' in metadata:
+        end_cell_re = re.compile('^#( )' + metadata['cellend'] + '\s*$')
+        del metadata['cellend']
+    else:
+        end_cell_re = None
+
+    # Find end of cell and return
+    if end_cell_re:
         for pos, line in enumerate(lines):
-            if pos > 0 and _END_CODE_MD.match(line):
+            if pos > 0 and end_cell_re.match(line):
                 next_line_blank = pos + 1 == len(lines) or \
                                   _BLANK_LINE.match(lines[pos + 1])
                 if next_line_blank and pos + 2 != len(lines):

tests/python_notebook_sample.py

@@ -21,17 +21,24 @@
 # - two blank lines if followed by an other code cell
 # - one blank line if followed by a markdown cell
 
-# Code cells can have blank lines, but no two consecutive blank lines (that's
-# a cell break!). Below we have a cell with multiple instructions:
+# Code cells can have blank lines, but no two consecutive blank lines (unless
+# a specific cell end is specified in the cell options, see below).
+# Now we have a cell with multiple instructions
 
 a = 3
 
 a + 1
 
-# ## Metadata in code cells
+# And a cell with an arbitrary count of blank lines. This last example
+# is also an instance of a cell with metadata information in json format,
+# escaped with '#+' or '# +'
 
-# In case a code cell has metadata information, it
-# is represented in json format, escaped with '#+' or '# +'
+# + {"endcell": "-"}
+def f(x):
+    return x + 1
 
-# + {"scrolled": true}
-a + 2
+
+def g(x):
+    return x + 2
+
+# -

tests/test_read_simple_python.py

@@ -104,6 +104,30 @@ def f(x):
     compare(pynb, pynb2)
 
 
+def test_read_cell_two_blank_lines(pynb="""# ---
+# title: cell with two consecutive blank lines
+# ---
+
+# + {"cellend": "-"}
+a = 1
+
+
+a + 2
+# -
+"""):
+    nb = nbrmd.reads(pynb, ext='.py')
+
+    assert len(nb.cells) == 2
+    assert nb.cells[0].cell_type == 'raw'
+    assert nb.cells[0].source == '---\ntitle: cell with two ' \
+                                 'consecutive blank lines\n---'
+    assert nb.cells[1].cell_type == 'code'
+    assert nb.cells[1].source == 'a = 1\n\n\na + 2'
+
+    pynb2 = nbrmd.writes(nb, ext='.py')
+    compare(pynb, pynb2)
+
+
 def test_read_multiline_comment(pynb="""'''This is a multiline
 comment with "quotes", 'single quotes'
 # and comments