Statement Columns as Graphics

[flywire]

The Australian Citibank cheque account uses graphics rather than text for statement columns (ie can't swipe it like the transactions) so pdf_statement_reader can't detect the start of the columns. It makes some attempt at the first two columns but it would be better if it used CLOSING BALANCE to detect the end of the transactions rather than picking up broken parts of bank notices.

The general layout is similar to CBA except dates are dd Mmm yyyy.

[flywire]

Looks like no headers need a change to config.json and a bit more code - citi.cheque.pdf

>>> import tabula
>>> dfs = tabula.read_pdf("citi.cheque.pdf", pages="all", area=(530, 47, 800, 558), options="--columns 105,275,340,444,558", pandas_options={'header': None})
>>> for df in dfs: df.columns = ['Date','Desc','DB','CR','Balance']
...
>>> dfs
[          Date                              Desc       DB        CR   Balance
0   1 Feb 2021                   Opening Balance      NaN       NaN  1,000.00
1   1 Feb 2021       Lorem ipsum dolor sit amet,  -197.60       NaN    802.40
2          NaN  consectetur adipiscing elit, sed      NaN       NaN       NaN
3          NaN      do eiusmod tempor incididunt      NaN       NaN       NaN
4   3 Feb 2021    labore et dolore magna aliqua.      NaN    651.12  1,453.52
5          NaN  Elit pellentesque habitant morbi      NaN       NaN       NaN
6   5 Feb 2021       tristique senectus. Quisque      NaN    666.06  2,119.58
7  28 Feb 2021                   Closing Balance      NaN       NaN  2,119.58
8          NaN                             TOTAL  -197.60  1,317.18       NaN
9          NaN                               NaN   Page 1       NaN       NaN]