Last modified on $Date$ To account for pages or eventually ink usage, you must plug your accounting system somewhere into the printing system you use, be it either CUPS or LPRng. One way to do this without having to modify the printing system itself, is by using a filter. A filter is a computer program which takes data in one format as its input, and outputs the same data but transformed into another format. CUPS already contains many filters. For example there's one filter named pstops which accepts PostScript data as its input, and, as its name implies, outputs PostScript data too, but after having eventually rearranged the pages to fit several pages on a single sheet of paper, or other manipulations like that. The pstops filter described above is also in charge of doing basic page accounting, but PyKota currently doesn't use this facility since it may prove to be unreliable depending on the drivers used or if a paper jam occur for example. LPRng can also use filters, one often used in combination with LPRng is magicfilter which can convert different input formats to native printers languages like PostScript or ESC/P2. So to do its own accounting, PyKota has its own filter, named pykota, which you have to plug into the different set of filters used by your printing system of choice. The procedure to install the pykota filter is described in the chapter. Currently with a CUPS backend, the pykota filter is used at the very last stage of the print mechanism, just before the final data is sent to the printer, but it may eventually be used earlier with some modifications to CUPS's filtering configuration. You can find in the mailing list archives of July 2003 an explanation on how to do this. This may allow you to use PyKota with a not-PostScript printer or with a PostScript printer which really needs to have a specific filter like magicfilter or cupsomatic. When used with an LPRng backend, the accounting filter is not defined the same way input filters are, so the problem described above doesn't exist, and PyKota may work with non-postscript printers, provided they can report their page counter and you know how to do this. When you submit a print job, pykota is automatically launched by your printing system, so it has to detect which system you are using (CUPS or LPRng) and behave like if it was specifically designed for your printing system. Fortunately there's not a lot of differences, the more important one is the exit codes used to tell the printing environment if a job has to be accepted or rejected. When using the querying accounting method, pykota then asks the printer for its internal page counter, read from the Quota DataBase the internal page counter for this printer when the previous job was launched, computes the difference, and report it as the previous job's size in the Quota DataBase. It then updates the last user's print quota and account balance, and warn him if he is over quota or his account balance below 0. Finally it checks if the user who launched the current job is below or above his print quota, and either allow or deny the job's datas to pass to the underlying layer (the printer itself). When using the external accounting method, and if the user is still allowed to print, the command you specified is launched with the job's data on its standard input. Your command must print the job's size in number of pages on a single line on its standard output. This number is then read by PyKota and used to update the current user's quota information. If a problem occurs, it is logged either to the filter's standard output or to the system logger, depending on your preferences in PyKota's configuration file. Also if a print quota is reached you may choose if the administrator, the user, both or no-one will receive an email message explaining the situation and proposing a solution.