Fix broken XLSX parsing. Closes #285.

csvkit/convert/xlsx.py

@@ -21,6 +21,19 @@ def normalize_datetime(dt):
 
     return dt
 
+def has_time_elements(cell):
+    """
+    Try to use formatting to determine if a cell contains only time info.
+
+    See: http://office.microsoft.com/en-us/excel-help/number-format-codes-HP005198679.aspx
+    """
+    if 'h' in cell.number_format or \
+        'H' in cell.number_format or \
+        'hh' in cell.number_format:
+        return True
+
+    return False
+
 def xlsx2csv(f, output=None, **kwargs):
     """
     Convert an Excel .xlsx file to csv.
@@ -36,23 +49,28 @@ def xlsx2csv(f, output=None, **kwargs):
     writer = CSVKitWriter(output)
 
     book = load_workbook(f, use_iterators=True, data_only=True)
+
     if 'sheet' in kwargs:
         sheet = book.get_sheet_by_name(kwargs['sheet'])
     else:
         sheet = book.get_active_sheet()
 
     for i, row in enumerate(sheet.iter_rows()):
         if i == 0:
-            writer.writerow([c.internal_value for c in row]) 
+            writer.writerow([c.value for c in row]) 
             continue
 
         out_row = []
 
         for c in row:
-            value = c.internal_value
+            value = c.value
 
             if value.__class__ is datetime.datetime:
-                if value.time() != NULL_TIME:
+                if value.time() != NULL_TIME or has_time_elements(c):
+                    # Handle default XLSX date as 00:00 time 
+                    if value.date() == datetime.date(1904, 1, 1):
+                        value = value.time() 
+
                     value = normalize_datetime(value)
                 else:
                     value = value.date()

examples/sheetsxlsx_converted.csv

@@ -1,5 +1,5 @@
 text,date,integer,boolean,float,time,datetime,empty_column,
-Chicago Tribune,1920-01-01,164,False,41800000.01,00:00:00,1920-01-01,,
+Chicago Tribune,1920-01-01,164,False,41800000.01,00:00:00,1920-01-01T00:00:00,,
 Chicago Sun-Times,1948-01-01,63,True,1.27,14:57:13,1948-01-01T14:57:13,,Extra data past headers will be trimmed
 Chicago Reader,1971-01-01,40,True,1,04:14:00,1971-01-01T04:14:00,,
 This row has blanks,,,,,,,,

examples/testxlsx_converted.csv

@@ -1,6 +1,6 @@
 text,date,integer,boolean,float,time,datetime,empty_column,
 Chicago Reader,1971-01-01,40,True,1,04:14:00,1971-01-01T04:14:00,,
 Chicago Sun-Times,1948-01-01,63,True,1.27,14:57:13,1948-01-01T14:57:13,,Extra data beyond headers will be trimmed
-Chicago Tribune,1920-01-01,164,False,41800000.01,00:00:00,1920-01-01,,
+Chicago Tribune,1920-01-01,164,False,41800000.01,00:00:00,1920-01-01T00:00:00,,
 This row has blanks,,,,,,,,
 Unicode! Σ,,,,,,,,

requirements.txt

@@ -5,7 +5,7 @@ python-dateutil>=1.5
 SQLAlchemy>=0.9.3
 sphinx>=1.0.7
 coverage>=3.5.1b1
-openpyxl>=1.8.4
+openpyxl>=2.0.3
 tox>=1.3
 dbf==0.95.004
 unittest2==0.5.1

setup.py

@@ -7,7 +7,7 @@
     'xlrd>=0.7.1',
     'python-dateutil>=1.5',
     'sqlalchemy>=0.6.6',
-    'openpyxl>=1.5.7',
+    'openpyxl>=2.0.3',
     'dbf==0.94.003']
 
 if sys.version_info < (2, 7):