duplicate cells returned from row.cells

[daveoloughlin]

The following code iterates through rows in the attached docx. Multiple cell instances have the the same text values somehow.
pu.docx

import docx
import os

doc = docx.Document('pu.docx')
for table in doc.tables:
    rowNo = 0
    for row in table.rows:
        cellNo = 0
        for cell in row.cells:
            print 'rowidx = {0} cellidx = {1} text ={2} '.format(rowNo, cellNo, cell.text)
            cellNo += 1
        rowNo += 1

[scanny]

This is the expected behavior when there are merged cells. If you want to get just the "visual" cells, you need to get the merge information for the cells and skip the merged width number of cells.

While this may seem less than an optimum design for the API, I remember there was a lot of deliberation that went into it and it was the best of the possible options. It would compromise other operations to make it "automatically" jump to the start of the next visual cell.

[daveoloughlin]

Thanks very much for the prompt response. That makes sense now.
I would like to try and find only "visual" cells, if possible. How would I determine the merged width number of cells to skip? I think I know the number of column cells (table.columns) and row cells (table.rows) in the "layout grid" but I'm not sure I can determine the width (or height) of a cell?
For my current test, summing the unique widths of each cell in a row will get me the table width (snippet below). Is this a good approach?
snippet.pdf

[scanny]

Apologies @daveoloughlin, Github seems to have eaten my response from last week.

I think you want something like this:

def iter_visual_cells(row):
    prior_tc = None:
    for cell in row.cells:
        this_tc = cell._tc
        if this_tc is prior_tc:  # skip cells pointing to same `<w:tc>` element
            continue
        yield cell
        prior_tc = this_tc

[daveoloughlin]

Thanks very much. Much simpler ;-)