🎉 allow trailing options, get_data(...keep_trailing_empty_cells=True). fix #86

Author: chfw

changelog.yml

@@ -2,6 +2,9 @@ name: pyexcel-io
 organisation: pyexcel
 releases:
 - changes:
+  - action: added
+    details:
+    - "`#86`: allow trailing options, get_data(...keep_trailing_empty_cells=True)."
   - action: fixed
     details:
     - "`#74`: handle zip files which contain non-UTF-8 encoded files."

docs/source/index.rst

@@ -91,6 +91,7 @@ get_data(.., library='pyexcel-ods')
    csvz
    sqlalchemy
    django
+   options
    extensions
 
 

docs/source/options.rst

@@ -0,0 +1,11 @@
+Options
+======================
+
+Here is the documentation on the keyword options for get_data.
+
+keep_trailing_empty_cells
+------------------------------
+
+default: False
+
+If turned on, the return data will contain trailing empty cells.

pyexcel_io/reader.py

@@ -17,6 +17,7 @@ def clean_keywords(keywords):
         "skip_row_func",
         "skip_empty_rows",
         "row_renderer",
+        "keep_trailing_empty_cells",
     ]
     for arg in keywords:
         if arg in args_list:

pyexcel_io/sheet.py

@@ -38,6 +38,7 @@ def __init__(
         skip_column_func=None,
         skip_empty_rows=False,
         row_renderer=None,
+        keep_trailing_empty_cells=False,
         **deprecated_use_of_keywords_here
     ):
         self._native_sheet = sheet
@@ -51,6 +52,7 @@ def __init__(
         self._skip_column = _index_filter
         self._skip_empty_rows = skip_empty_rows
         self._row_renderer = row_renderer
+        self.keep_trailing_empty_cells = keep_trailing_empty_cells
 
         if skip_row_func:
             self._skip_row = skip_row_func
@@ -84,10 +86,13 @@ def to_array(self):
                 elif column_position == constants.STOP_ITERATION:
                     break
 
-                tmp_row.append(cell_value)
-                if cell_value is not None and cell_value != "":
-                    return_row += tmp_row
-                    tmp_row = []
+                if self.keep_trailing_empty_cells:
+                    return_row.append(cell_value)
+                else:
+                    tmp_row.append(cell_value)
+                    if cell_value is not None and cell_value != "":
+                        return_row += tmp_row
+                        tmp_row = []
             if self._skip_empty_rows and len(return_row) < 1:
                 # we by-pass next yeild here
                 # because it is an empty row

tests/test_csv_book.py

@@ -3,6 +3,7 @@
 from unittest import TestCase
 
 import pyexcel_io.manager as manager
+from pyexcel_io import get_data
 from pyexcel_io.sheet import NamedContent
 from pyexcel_io.reader import EncapsulatedSheetReader
 from pyexcel_io._compact import BytesIO, StringIO
@@ -115,6 +116,18 @@ def test_sheet_file_reader(self):
         result = list(r.to_array())
         self.assertEqual(result, [[1], [4, 5, 6], ["", 7]])
 
+    def test_sheet_file_reader_with_trailing_empty_cells(self):
+        r = EncapsulatedSheetReader(
+            CSVFileReader(NamedContent(self.file_type, self.test_file)),
+            keep_trailing_empty_cells=True,
+        )
+        result = list(r.to_array())
+        self.assertEqual(result, [[1], [4, 5, 6, "", ""], ["", 7]])
+
+    def test_get_data_with_trailing_empty_cells(self):
+        result = get_data(self.test_file, keep_trailing_empty_cells=True)
+        self.assertEqual(result[self.test_file], [[1], [4, 5, 6, "", ""], ["", 7]])
+
     def tearDown(self):
         os.unlink(self.test_file)