Text not in the expected cell

Here is my document:

I want to extract tuples like("户名", "张三"),("户号", "1102"), but I found I cannot locate values correctly using the following code:

from docx import Document

file = '/paht/to/my_doc.docx'

doc = Document(file)
tables = doc.tables
table = tables[0]  # now, I get the table object successfully
# Intuitively, "户    名" is in the first cell of second row. When I try to find it:
row_num = 1
cell_num = 0
row = table.rows[row_num]
cell = row.cells[cell_num]
print(cell.text)  # will print "张三"
# Actually,"户    名"  is in the last cell of first row:
row_num = 0
row = table.rows[row_num]
cell = row.cells[-1]
print(cell.text)  # will print "户    名"

Could anybody please tell me what’s wrong with it:)

[scanny]

Okay, in the XML you sent me, w:tbl/w:tblGrid contains 32 w:gridCol entries. This means the table layout grid is 32 columns wide.

The first row (the long merged heading) only accounts for 31 layout grid columns. I think that’s the source of the problem and python-docx is interpreting the first cell of the second row as the 32nd cell in the first column.

Basically, the table is just too complex for python-docx to navigate it. There are rules that Word has for rendering tables that go beyond the simple-ish uniform layout-grid concept and that are not implemented in python-docx. So pursuing the “row-by-row” “apparent-cell” approach we mentioned above is probably the best course.

[scanny]

@zifeiYv ok, that all looks in order, I suspect that the table is not consistent with the spec and its underlying layout-grid is corrupted, which means the python-docx cell addressing won’t work reliably. There is a lot of merging of cells in that table and it can happen that Word can still render the table but the layout-grid addressing of python-docx can’t figure out which cell is which.

Can you dump the overall table to an XML file and attach it? (print(table._element.xml) it will be too long to paste into a comment-box here)

But whatever we find out there, I think the best available path is for you to resort to “physical” indexing:

from docx.table import _Cell

def iter_visible_cells(row):
    tr = row._tr
    for tc in tr.tc_lst:
        yield _Cell(tc, row.parent.table)

row = table.rows[1]
row_cells = list(iter_visible_cells(row))
print(row_cells[0].text)

The cells accessed this way should work just as well as any other cell and will be much easier to address in a complex layout like this.