|
HP OpenVMS Systemsask the wizard |
|
The Question is: Is there a way to edit a file that is in a relative organized file? The Answer is :
There is no standard OpenVMS utility that can accomplish the task
of (directly) editing a relative-format file.
There are data manipulation products (DATATRIEVE, ACCELR) for sale
and potentially freeware that may help.
The OpenVMS Wizard would encourage you to write a simple program in
the language of your choice (either a compiled language, DCL, or
otherwise) to perform the required edits while maintaining the
file integrety from the perspective of the file contents. Alternatively,
write a tool that can export the data into a (sequential) file that can
be edited, and a tool that then imports the data into the appropriate
format.
For INDEXED files containing (only) text data, you can potentially edit
the file with a regular text editor and reconverting the output back to
indexed using an FDL description obtained from the original. There are
caveats with this: order of duplicates, and you must be exceedingly
careful to avoid altering other data or other columns in the file.
For RELATIVE files, a sequential view of the file as seen by a regular
text editor or 'TYPE' commands will drop the cell-level information.
In other words, the file will be packed and all records will appear
densely populated (in adjacent cells) while the relative file may
actually have been sparely populated. As long as the input file was
tightly packed, with no deleted records or empty record cells, then
you can reconvert the edited contents, but such simple cases are rare
applications.
For occasional edits or data recovery purposes on relative-format and
indexed-format files, the most common tool is often a simple (and
custom-written) DCL command procedure. Using the WRITE/UPDATE command,
you can update a previously read record IN PLACE. Using READ/DELETE you
can remove record, but you can not add new records (no WRITE/KEY)
To read the target record, either use sequential reads (READ with no
argument other than handle and buffer) until the right count or
contents is seen. Or... use KEYed reads, but those are tricky at best
due to the BINARY nature of relative file record numbers.
Some sample manipulations are included below. The output is based on
a file loaded with record 1,5 and 9 using the included simple BASIC
load program. Use DUMP/RECORD[=(star:nn,count:mm)] to verify the CELL
number in the RFA.
$ create/fdl=sys$input tmp.rel
file; organization relative
record; size 80
$ run load
$ open/read/write x tmp.rel
$ read x rec
$ show sym rec
$ read x rec
$ show sym rec
$ write /update x "hello there"
$ close x
$ typ tmp.tmp
aap test
hello there
mies test
$ dump/rec tmp.tmp
Record number 1 (00000001), 80 (0050) bytes, RFA(0001,0000,0000)
20202020 20202020 20202020 20706161 aap 000000
20202020 20202020 74736574 20202020 test 000010
20202020 20202020 20202020 20202020 000020
20202020 20202020 20202020 20202020 000030
20202020 20202020 20202020 20202020 000040
Record number 2 (00000002), 11 (000B) bytes, RFA(0005,0000,0000)
657265 6874206F 6C6C6568 hello there..... 000000
Record number 3 (00000003), 80 (0050) bytes, RFA(0009,0000,0000)
20202020 20202020 20202020 7365696D mies 000000
20202020 20202020 74736574 20202020 test 000010
20202020 20202020 20202020 20202020 000020
20202020 20202020 20202020 20202020 000030
20202020 20202020 20202020 20202020 000040
$ open/read/write x tmp.tmp
$ key="1234"
$ key[0,32]=5
$ read/key=&key x rec
$ show symb rec
REC = "hello there"
$ key[0,32]=6
$ read/key=&key x rec
%RMS-E-RNF, record not found
$ write/update x "Hmmmm...."
%RMS-F-CUR, no current record (operation not preceded by $GET/$FIND)
$ type load.bas
open "tmp.rel" as file #1, recordtype any, ORGANIZATION RELATIVE, map x
map (x) string x=20, y=20
y = "test"
input "record number"; i
while i
input "record data"; x
put #1, record i
input "next record"; i
next
close #1
|