"Steve" wrote in message
...
I distributed some workbook apps in 2001 that are
currently being used
for budgeting and forecasting work. A main feature of
the apps is the
ability to pull current-year data from the
proprietary accounting software
that the businesses use. The users were able to run a
general ledger
trial balance, save it in plain text format, and my
apps would parse the
data and import it into the workbooks.

Well, now their accounting software has been updated
to export the
reports into html format, not plain text.

A lot of accounting software these days will export to
XML
format. See if maybe that is an option. If so you can
use
MSXML to easily parse the XML.

If not, and you are stuck with HTML, the first thing I
would
do is find the <Table...</Table section and strip it
out.
Then I would scan through the remainder, and count the
number of rows, at the same time I would count the max
number of columns. This is easily done by counting the
<TR
tags for rows, and each time you see a <TR count the
number of <TD or <TH tags that show up before the
next </TR. Only keep track of the max number of
columns.

Then redim an array as myArray(0 to numCols - 1, 0 to
numRows - 1)
Now read throught the data again, and start at row 0,
col 0. Each time
you hit a new </TD or </TH tag increment the col
pointer.
Each time you hit a </TR tag increment the row
pointer. Read the
data between each tag in the corresponding
myArray(curCol, curRow).

There are probably easier ways to do this. But this is
just
the method that first came to mind.

Andrew Faust