Setting an Excel Range with an Array using Python and comtypes?

问题

Using comtypes to drive Python, it seems some magic is happening behind the scenes that is not converting tuples and lists to VARIANT types:

# RANGE(“C14:D21”) has values
# Setting the Value on the Range with a Variant should work, but
# list or tuple is not getting converted properly it seems

>>>from comtypes.client import CreateObject
>>>xl = CreateObject("Excel.application")
>>>xl.Workbooks.Open(r'C:\temp\my_file.xlsx')
>>>xl.Visible = True
>>>vals=tuple([(x,y) for x,y in zip('abcdefgh',xrange(8))])
# creates: 
#(('a', 0), ('b', 1), ('c', 2), ('d', 3), ('e', 4), ('f', 5), ('g', 6), ('h', 7))
>>>sheet = xl.Workbooks[1].Sheets["Sheet1"]
>>>sheet.Range["C14","D21"].Value()
(('foo',1),('foo',2),('foo',3),('foo',4),('foo',6),('foo',6),('foo',7),('foo',8))
>>>sheet.Range["C14","D21"].Value[()] = vals
# no error, this blanks out the cells in the Range

According to the comtypes docs:

When you pass simple sequences (lists or tuples) as VARIANT parameters, the COM server will receive a VARIANT containing a SAFEARRAY of VARIANTs with the typecode VT_ARRAY | VT_VARIANT.

This seems to be inline with what MSDN says about passing an array to a Range's Value. I also found this page showing something similar in C#. Can anybody tell me what I'm doing wrong?

EDIT

I've come up with a simpler example that performs the same way (in that, it does not work):

>>>from comtypes.client import CreateObject
>>>xl = CreateObject("Excel.application")
>>>xl.Workbooks.Add()
>>>sheet = xl.Workbooks[1].Sheets["Sheet1"]
# at this point, I manually typed into the range A1:B3
>>> sheet.Range("A1","B3").Value()
((u'AAA', 1.0), (u'BBB', 2.0), (u'CCC', 3.0))
>>>sheet.Range("A1","B3").Value[()] = [(x,y) for x,y in zip('xyz',xrange(3))]
# Using a generator expression, per @Mike's comment
# However, this still blanks out my range :(

回答1:

I've spent a lot of time trying to figure out a solution to this problem to be able to fully substitute python for matlab, reading on various different forums with no real direct answer.

Here is my robust solution that works very well. I have to write daily/weekly/monthly/quarterly reports that write to xlsx a lot, this function works much better than some of the information out their about writing to xlsx using python & com.

from numpy import *
from win32com.client import DispatchEx
# DispatchEx opens up an independent instance of Excel so writing to a document won't interfere with any other instances you have running
def xlsxwrite(filename, sheet, data, cellstr, screenupdating = False, direction = 'h', visible = 0):
'''
Write to an excel document by setting ranges equal to arrays.
'''
xl = DispatchEx("Excel.Application")
xl.ScreenUpdating = screenupdating
xl.Visible = visible
try:
  excel_type = get_exceltype(filename)
  # Check to see if workbook exists, if it doesn't create workbook
  try:
      xlBook = xl.Workbooks.Open(filename)
  except:
      print '\nFile Doesnt Exist, Writing File...\n\n\n'
      xlBook = xl.Workbooks.Add()
      try:
          xlBook.SaveAs(filename, excel_type)
      except:
          xl.Quit()
          raise NameError('Error writing file: %s, check to make sure path exists' % filename)
  # Get wksht names
  wksht_names = [xlBook.Sheets(i).Name for i in range(1,xlBook.Sheets.Count+1)]
  # If 'sheet' variable is an integer, get sheet by index number, else get it by name, or add new one
  try:
      int(sheet)
      try:
          xlSheet = xlBook.Sheets(int(sheet))
      except:
          raise NameError('Error, referencing an invalid sheet')
  except:
      # If sheet input not in wksht names, add it
      if sheet not in wksht_names:
          print 'Worksheet, "%s", not found, Adding Worksheet' % sheet
          xlBook.Sheets.Add(After=xlBook.Sheets(xlBook.Sheets.Count)).Name = sheet
      xlSheet = xlBook.Sheets(sheet)
  # Convert Excel Range to Python Range
  row,col = getcell(cellstr)
  # Write out data
  output_dict, shp = data_export_cleaner(data, direction)
  a,b = shp
  start_cells = [(row,col+i) for i in range(b)]
  end_cells = [(row + a -1,col+i) for i in range(b)]
  for i in output_dict.keys():
      cell_range = eval('xlSheet.Range(xlSheet.Cells%s,xlSheet.Cells%s)' %   (start_cells[i],end_cells[i]))
      cell_range.Value = output_dict[i]
  # Save and close document, Quit Excel App
  xlBook.Close(True)
  xl.Quit()
  return
  except:
    xlBook.Close(False)
    xl.Quit()
    raise NameError('Error occurred while trying to write file')

def data_export_cleaner(data,direction):
  """
  Summary: Return data in a format that works with Excel Com (Numpy int32 for some reason was causing an error, have to turn it into a string, doesn't affect any formatting possibilities).
  Defaults: Going to set the default for writing data with len(shape(array(data))) == 1, such as a list, to horizontal, if you want to write it vertically, specify 'v', only applicable for lists.
  """
  darray = array(data)
  shp = shape(darray)
  if len(shp) == 0:
      darray = array([data])
      darray = darray.reshape(1,1)
  if len(shp) == 1:
      darray = array([data])
      if direction.lower() == 'v':
          darray = darray.transpose()
  shp = shape(darray)
  tempdict = dict()
  for i in range(shp[1]):
      tempdict[i] = [(str(darray[j,i]),) for j in range(shp[0])]
  return tempdict, shp


def get_exceltype(filename):
  format_dict = {'xlsx':51,'xlsm':52,'xlsb':50,'xls':56}
  temp = character_count(filename)
  if (temp['.'] > 1 or temp['.'] == 0):
      raise NameError('Error: Incorrect File Path Name, multiple or no periods')
  f_type = filename.split('.')
  f_type = f_type[len(f_type)-1]
  if f_type not in format_dict.keys():
      raise NameError('Error: Incorrect File Path, No excel file specified')
  else:
      return format_dict[f_type]


def character_count(a_string):
  temp = dict()
  for c in a_string:
      temp[c] = temp.get(c,0) + 1
  return temp



def getcell(cell):
  '''Take a cell such as 'A1' and return the corresponding numerical row and column in excel'''
  a = len(cell)
  temp_column = []
  row = []
  temp_row = []
  if a < 2:
      raise NameError('Error, the cell you entered is not valid')
  for i in range(a):
      if str.isdigit(cell[i])==False:
          temp_column.append(cell[i])
      else:
          temp_row.append(cell[i])
  row.append(string.join(temp_row,''))
  row = int(row[0])
  column = getnumericalcolumn(temp_column)
  return row, column


def getnumericalcolumn(column):
  '''Take an excel column specification such as 'A' and return its numerical equivalent in excel'''
  alpha = str(string.ascii_uppercase)
  alphadict = dict(zip(alpha,range(1,len(alpha)+1)))
  if len(column) == 1:
      numcol = alphadict[column[0]]
  elif len(column) == 2:
      numcol = alphadict[column[0]]*26 + alphadict[column[1]]
  elif len(column) == 3:
      numcol = 26**2 + alphadict[column[1]]*26 + alphadict[column[2]]
  return numcol

Notes: I use Numpy a lot because it is very useful for creating the tables in the format I want to write them, so that is a required library for the below functions to work. I know all these functions could combined to create a class, but because this function is called in a script and their isn't really a significant benefit to creating it as a class, I haven't done so.

回答2:

Try sheet.Range("C14", "D21").Value = vals. I'm not quite sure how the API is structured, but it works for me.

(Also, tuple([(x,y) for x,y in zip('abcdefgh',xrange(8))]) can be replaced with a generator expression tuple((x, y) for x, y in zip('abcdefgh', xrange(8))), which is seems a bit cleaner habit to have. In this particular case, just the list comprehension [(x, y) for x, y in zip('abcdefgh', xrange(8))] would also do.)

来源：https://stackoverflow.com/questions/2150017/setting-an-excel-range-with-an-array-using-python-and-comtypes