BBS: Inland Empire Archive
Date: 03-10-92 (21:01) Number: 159
From: TOM HAMMOND Refer#: NONE
To: TRENT SHIRLEY Recvd: NO
Subj: Text search routine Conf: (2) Quik_Bas
TS>file for any given word or text string. I need the search
TS>to be Case indifferent and was hoping that someone out
TS>there had a routine, or shareware/freeware library routine...
Someone sent a message a while back which described a pretty easy (and
what should be VERY FAST) method of doing such a search... I think it
may have been Matt Hart... lemme see if I still have it... ah, yes, here
it is... let's see if I can possibly format it into a bit more easily
readable form... it's not a source code listing, but it's certainly a
good set of pseudocode from which you can work...
Hope this helps.
good Luck from Tom Hammond (N0SS)
Date: 02-04-92 (10:09) Number: 5286
From: MATT HART Refer#: NONE
To: FRANK DIDIK Recvd: NO
Subj: HASH FILES Conf: (13) QuickBasic
QB can open a sequential file as binary and do a search using the
largest available buffer size, then find the beginning an ending line
very quickly.
Once the seq file is created, open as binary, get records into a string
of say 10K, do an INSTR(Strg$,SearchStrg$).
If not found, do a SEEK from the current position back the length of the
search string (so as not to miss it if is was in between the 10K
boundaries), then continue the search.
When found, SEEK backwards 2 bytes at a time until a CHR$(13)+CHR$(10)
is found, then GET the record, set
FinalStrg$ = LEFT$(Record$,INSTR(Record$,CHR$(13)+CHR$(10)))
and you'll have that line as it was PRINTed to a sequential file, but
VERY fast.
--- WM v2.00/91-0231
* Origin: The Modem Zone BBS (314) 893-5106 (1:289/2)
