make docs

README.md

@@ -192,13 +192,26 @@ csv-kwargs:
 
 `pipe_tables`  
 If True, a pipe table will be constructed directly in markdown syntax
-instead of via AST. `markdown` is implied to be True. This trades
-correctness for speed. It won’t be correct if any of the cell is
-multiline for example, resulting in an invalid pipe table. However, it
-is much faster comparing to previous `markdown: True` case because
-previously per cell a subprocess to execute pandoc the parse the
+instead of via AST. `markdown` is implied to be True. `header` will be
+overridden as true because `pipe_tables` must has header in pandoc.
+
+This trades correctness for speed. It won’t be correct if any of the
+cell is multiline for example, resulting in an invalid pipe table.
+However, it is much faster comparing to previous `markdown: True` case
+because previously per cell a subprocess to execute pandoc the parse the
 markdown to AST is needed.
 
+`grid_tables`  
+If True, a grid table will be constructed directly in markdown syntax
+instead of via AST. `markdown` is implied to be True. `header` can be
+used together with this.
+
+This trades correctness for speed. This should be more robust than
+`pipe_tables` since the `grid_tables` syntax supports everything the
+pandoc AST supports. This however depends on an external dependency.
+Install it by either `pip install terminaltables` or
+`conda install terminaltables`.
+
 `raw_markdown`  
 If True, force output the table as a pipe table (which is
 tab-delimited.) This is sometimes useful if pandoc is very stubborn to

README.rst

@@ -165,12 +165,26 @@ keys are recognized:
 
 ``pipe_tables``
    If True, a pipe table will be constructed directly in markdown syntax
-   instead of via AST. ``markdown`` is implied to be True. This trades
-   correctness for speed. It won’t be correct if any of the cell is
-   multiline for example, resulting in an invalid pipe table. However,
-   it is much faster comparing to previous ``markdown: True`` case
-   because previously per cell a subprocess to execute pandoc the parse
-   the markdown to AST is needed.
+   instead of via AST. ``markdown`` is implied to be True. ``header``
+   will be overridden as true because ``pipe_tables`` must has header in
+   pandoc.
+
+   This trades correctness for speed. It won’t be correct if any of the
+   cell is multiline for example, resulting in an invalid pipe table.
+   However, it is much faster comparing to previous ``markdown: True``
+   case because previously per cell a subprocess to execute pandoc the
+   parse the markdown to AST is needed.
+
+``grid_tables``
+   If True, a grid table will be constructed directly in markdown syntax
+   instead of via AST. ``markdown`` is implied to be True. ``header``
+   can be used together with this.
+
+   This trades correctness for speed. This should be more robust than
+   ``pipe_tables`` since the ``grid_tables`` syntax supports everything
+   the pandoc AST supports. This however depends on an external
+   dependency. Install it by either ``pip install terminaltables`` or
+   ``conda install terminaltables``.
 
 ``raw_markdown``
    If True, force output the table as a pipe table (which is

pantable/pantable.py

@@ -168,7 +168,7 @@ def regularize_table_list(raw_table_list):
     for i, (n, row) in enumerate(zip(length_of_rows, raw_table_list)):
         if n != n_col:
             row += [''] * (n_col - n)
-            panflute.debug("pantable: the {}-th row is shorter than the longest row. Empty cells appended.")
+            panflute.debug("pantable: the {}-th row is shorter than the longest row. Empty cells appended.".format(i))
     return n_col