🐛 fix #18, convert the non-seekable io instance to a seekable one because python's ZipFile would rewind itself at some point

Author: chfw

pyexcel_xlsx/xlsxr.py

@@ -8,10 +8,11 @@
     :license: New BSD License
 """
 import openpyxl
+from io import UnsupportedOperation
 
 from pyexcel_io.book import BookReader
 from pyexcel_io.sheet import SheetReader
-from pyexcel_io._compact import OrderedDict
+from pyexcel_io._compact import OrderedDict, BytesIO
 
 
 class XLSXSheet(SheetReader):
@@ -49,6 +50,16 @@ def open(self, file_name, skip_hidden_sheets=True, **keywords):
         self._load_the_excel_file(file_name)
 
     def open_stream(self, file_stream, skip_hidden_sheets=True, **keywords):
+        if not hasattr(file_stream, 'seek'):
+            # python 2
+            # Hei zipfile in odfpy would do a seek
+            # but stream from urlib cannot do seek
+            file_stream = BytesIO(file_stream.read())
+        try:
+            file_stream.seek(0)
+        except UnsupportedOperation:
+            # python 3
+            file_stream = BytesIO(file_stream.read())
         BookReader.open_stream(self, file_stream, **keywords)
         self.skip_hidden_sheets = skip_hidden_sheets
         self._load_the_excel_file(file_stream)

tests/test_bug_fixes.py

@@ -9,6 +9,9 @@
 from textwrap import dedent
 import pyexcel as pe
 from nose.tools import eq_
+from nose import SkipTest
+
+IN_TRAVIS = 'TRAVIS' in os.environ
 
 
 PY36_ABOVE = sys.version_info[0] == 3 and sys.version_info[1] >= 6
@@ -97,5 +100,11 @@ def test_issue_8_hidden_sheet_2():
     eq_(book_dict['hidden'], [['a', 'b']])
 
 
+def test_issue_20():
+    #if not IN_TRAVIS:
+    #    raise SkipTest()
+    pe.get_book(url="https://github.com/pyexcel/pyexcel-xlsx/raw/master/tests/fixtures/file_with_an_empty_sheet.xlsx");  # flake8: noqa
+
+
 def get_fixtures(file_name):
     return os.path.join("tests", "fixtures", file_name)