Use original text to compute pep8 lines #181

jupytext/cell_reader.py

@@ -95,6 +95,7 @@ def __init__(self, fmt=None, default_language=None):
         self.default_language = default_language or _SCRIPT_EXTENSIONS.get(self.ext, {}).get('language', 'python')
         self.comment_magics = fmt['comment_magics'] if 'comment_magics' in fmt else self.default_comment_magics
         self.metadata = None
+        self.org_content = []
         self.content = []
         self.explicit_eoc = None
         self.cell_type = None
@@ -124,11 +125,8 @@ def read(self, lines):
         if not self.metadata:
             self.metadata = {}
 
-        org_lines = self.content
-        if self.ext == '.py' and self.cell_type != 'code' and self.content:
-            org_lines = ['#']  # cell was originally commented
         if self.ext == '.py' and not self.explicit_eoc:
-            expected_blank_lines = pep8_lines_between_cells(org_lines or [''], lines[pos_next_cell:], self.ext)
+            expected_blank_lines = pep8_lines_between_cells(self.org_content or [''], lines[pos_next_cell:], self.ext)
         else:
             expected_blank_lines = 1
 
@@ -215,6 +213,7 @@ def find_cell_content(self, lines):
 
         # Cell content
         source = lines[cell_start:cell_end_marker]
+        self.org_content = [line for line in source]
 
         # Exactly two empty lines at the end of cell (caused by PEP8)?
         if self.ext == '.py' and self.explicit_eoc:
@@ -467,6 +466,7 @@ def find_cell_content(self, lines):
 
         # Cell content
         source = lines[cell_start:cell_end_marker]
+        self.org_content = [line for line in source]
 
         if self.cell_type != 'code' or (self.metadata and not is_active('py', self.metadata)) \
                 or (self.language is not None and self.language != self.default_language):
@@ -623,6 +623,7 @@ def find_cell_content(self, lines):
 
         # Cell content
         source = lines[cell_start:cell_end_marker]
+        self.org_content = [line for line in source]
 
         if self.cell_type == 'code' and self.comment_magics:
             uncomment_magic(source, self.language or self.default_language)

tests/test_black.py

@@ -24,8 +24,10 @@ def test_apply_black_on_python_notebooks(nb_file, tmpdir):
 
     nb1 = readf(nb_file)
     nb2 = readf(tmp_ipynb)
+    nb3 = readf(tmp_py)
 
     assert len(nb1.cells) == len(nb2.cells)
+    assert len(nb1.cells) == len(nb3.cells)
     for c1, c2 in zip(nb1.cells, nb2.cells):
         # same content (almost)
         assert black_invariant(c1.source) == black_invariant(c2.source)

tests/test_read_simple_python.py

@@ -602,6 +602,22 @@ def test_notebook_with_magic_and_bash_cells(script="""# This is a test for issue
     compare(script, script2)
 
 
+def test_notebook_no_line_to_next_cell(nb=new_notebook(
+    cells=[new_markdown_cell('Markdown cell #1'),
+           new_code_cell('%load_ext line_profiler'),
+           new_markdown_cell('Markdown cell #2'),
+           new_code_cell('%lprun -f ...'),
+           new_markdown_cell('Markdown cell #3'),
+           new_code_cell('# And a function!\n'
+                         'def f(x):\n'
+                         '    return 5')])):
+    script = jupytext.writes(nb, 'py')
+    nb2 = jupytext.reads(script, 'py')
+    nb2.metadata.pop('jupytext')
+
+    compare(nb, nb2)
+
+
 def test_notebook_one_blank_line_before_first_markdown_cell(script="""
 # This is a markdown cell