BUG: read_excel failed with empty rows after MultiIndex header #40649
Conversation
There was a problem hiding this comment.
Thanks @ahawryluk, can you update the title to a more meaningful one?
ahawryluk
changed the title
|
@dsaxton thanks for catching those |
| @@ -554,7 +554,9 @@ def parse( | |||
| header_name, _ = pop_header_name(data[row], index_col) | |||
| header_names.append(header_name) | |||
|
|
|||
| has_index_names = is_list_like(header) and len(header) > 1 | |||
| has_index_names = ( | |||
There was a problem hiding this comment.
can you add a comment here on what the inference rules are
| @@ -702,7 +702,7 @@ cdef class TextReader: | |||
| ic = (len(self.index_col) if self.index_col | |||
| is not None else 0) | |||
|
|
|||
| if lc != unnamed_count and lc - ic > unnamed_count: | |||
| if (lc != unnamed_count and lc - ic > unnamed_count) or ic == 0: | |||
There was a problem hiding this comment.
can you add a similar comment to below (e.g. for future readers what are the inference rules)
| columns = MultiIndex.from_tuples([("a", "A"), ("b", "B")]) | ||
| expected = DataFrame(data, columns=columns) | ||
| data = "a,b\nA,B\n,\n1,2\n3,4" | ||
| result = parser.read_csv(StringIO(data), header=[0, 1]) |
There was a problem hiding this comment.
what happens with index_col not None (do we already tests this)
There was a problem hiding this comment.
Yes, these two tests both have MultiIndex headers and index_col not None:
pandas/pandas/tests/io/parser/test_header.py
Line 126 in 526d52f
pandas/pandas/tests/io/parser/test_header.py
Line 219 in 526d52f
|
@jreback I've added the new comments; let me know if anything else is needed. Thanks |
|
thanks @ahawryluk very nice |
Prior to this fix, a blank data row after a MultiIndex header was a interpreted as containing a blank index name, but that only works if the user has specified an index column. If index_col is None all subsequent rows should be treated as data, even if the first one is empty.