Fixes #226 - Allow importing ragged CSV files (#456)

HISTORY.md

@@ -9,6 +9,11 @@
   will remain in Tablib 1.x and will be fixed (reversed) in Tablib 2.0.0 (#453). If you
   count on the broken behavior, please update your code when you upgrade to Tablib 2.x.
 
+### Improvements
+
+- Tablib is now able to import CSV content where not all rows have the same
+  length. Missing columns on any line receive the empty string (#226).
+
 ## 1.0.0 (2020-01-13)
 
 ### Breaking changes

src/tablib/formats/_csv.py

@@ -46,6 +46,8 @@ def import_set(cls, dset, in_stream, headers=True, **kwargs):
             if (i == 0) and (headers):
                 dset.headers = row
             elif row:
+                if i > 0 and len(row) < dset.width:
+                    row += [''] * (dset.width - len(row))
                 dset.append(row)
 
     @classmethod

tests/test_tablib.py

@@ -804,6 +804,25 @@ def test_csv_import_set_with_unicode_str(self):
         data.csv = csv_text
         self.assertEqual(data.width, 7)
 
+    def test_csv_import_set_ragged(self):
+        """Import CSV set when not all rows have the same length."""
+        csv_text = (
+            "H1,H2,H3\n"
+            "A,B\n"
+            "C,D,E\n"
+            "\n"
+            "F\n"
+        )
+        dataset = tablib.import_set(csv_text, format="csv")
+        self.assertEqual(
+            str(dataset),
+            'H1|H2|H3\n'
+            '--|--|--\n'
+            'A |B |  \n'
+            'C |D |E \n'
+            'F |  |  '
+        )
+
     def test_csv_export(self):
         """Verify exporting dataset object as CSV."""