Browse Prior Art Database

Indexing Input Line Data to AFPDS/MODCA Output

IP.com Disclosure Number: IPCOM000111378D
Original Publication Date: 1994-Feb-01
Included in the Prior Art Database: 2005-Mar-26
Document File: 2 page(s) / 96K

Publishing Venue

IBM

Related People

Black, BA: AUTHOR [+5]

Abstract

A program is disclosed that indexes individual pages or groups of pages of AFPDS output based on the location and content of the corresponding input data. A way to identify output pages based on the contents of the original input was needed to enable archiving and retrieval of the formatted print output. Considering the size of the potential documents, the ability to save the indices and their locations in the output file external to the document was necessary for performance reasons.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 52% of the total text.

Indexing Input Line Data to AFPDS/MODCA Output

      A program is disclosed that indexes individual pages or groups
of pages of AFPDS output based on the location and content of the
corresponding input data.  A way to identify output pages based on
the contents of the original input was needed to enable archiving and
retrieval of the formatted print output.  Considering the size of the
potential documents, the ability to save the indices and their
locations in the output file external to the document was necessary
for performance reasons.

      To create indices for output pages, a method of associating an
output page with the input records that were used in creating it was
necessary.  When using ANSI or machine carriage controls, with
pagedef processing the end of a page is not necessarily known until
after the input record that caused it has already been processed.  In
order to implement the indexing of AFPDS output pages from its input,
a dual buffering scheme was used (Figure).  Input record and output
page buffers external to the AFPDS transform were created.  In order
to determine how to  identify where "new" segments of input data
occur, the concept of "triggers" (conceptually similar to
eyecatchers) and index value fields on the input records was
introduced.  For flexi- bility it was decided that the index values
could occur before their corresponding triggers.

      Control of the input record processing is based on a pair of
simple state machines, one for the input buffer and one for the
index.  The input buffer has 3 states: 'empty', 'partial', or 'full'.
The index has 4 possible states: 'searching', 'possible match',
'false match' or 'complete match'.  Records are read into the input
FIFO buffer until the input buffer state is 'full' or the index state
is 'possible match'.  Until one of these 2 conditions is met, every
input record is scanned for the first trigger field.  The size of the
FIFO buffer is determined by the span of records needed to hold all
the index value and trigger fields.

      When the input buffer state is 'full', the oldest (first)
record is released to the conversion machine to be converted to AFPDS
page data.

      When the index state is 'possible match', all records prior to
those containing the index values or triggers are...