Hello all,

I have been given what seems to me a slightly hairy task. I have a report with fixed-width fields similar to this:

Line containing uninteresting information

Another line containing uninteresting information

More uninteresting information

Transaction   Employee    ---
Date          Name        Amount
 ------------ ------------------
111/222/ABC

23-JAN-06     Baker, Abel 111.11

24-JAN-06     Baker, Abel 222.22

25-JAN-06     Baker, Abel 333.33

26-JAN-06     Baker, Abel 444.44

27-JAN-06     Baker, Abel 555.55

Line containing uninteresting information

Another line containing uninteresting information

Transaction   Employee    ---
Date          Name        Amount
 ------------ ------------------
333/444/DEF

23-JAN-06     Dog,        111.11
              Charlie

24-JAN-06     Dog,        222.22
              Charlie

25-JAN-06     Dog,        333.33
              Charlie
[download]

Note the spurious ^L characters, which appear as boxes in my browser, at the beginning of some lines and the over-wide fields running into the following line.

This needs to be converted to Excel-readable format, e.g. CSV, containing the information from the table-like parts with information from the line after the headers in additional colums, like this:

Code1|Code2|Transaction Date|Employee Name|Amount
111|ABC|23-JAN-06|Baker, Abel|111.11
111|ABC|24-JAN-06|Baker, Abel|222.22
111|ABC|25-JAN-06|Baker, Abel|333.33
111|ABC|26-JAN-06|Baker, Abel|444.44
111|ABC|27-JAN-06|Baker, Abel|555.55
333|DEF|23-JAN-06|Dog, Charlie|111.11
333|DEF|24-JAN-06|Dog, Charlie|222.22
333|DEF|25-JAN-06|Dog, Charlie|333.33
[download]

I would approach this in the following way

• slurp the file
• for each line, remove any spurious characters
• then, check whether any of a number of regexps match
• then, do the appropriate thing for the identified line type

Does any one have any less naive ideas? I wondered whether there is there something like format for reading files?

Thanks,

loris