Last record in a sequential file

[Michael Mattias]

One little thing to watch out for...

Sometimes 'text' files END with a CRLF, sometimes they don't.

[Mike Doty]

If a file is very large this could prevent having to read it all.

#COMPILE EXE
#DIM ALL
DECLARE FUNCTION GetLastRecordSequential(sFile AS STRING) AS STRING
FUNCTION PBMAIN AS LONG
  'Create a sequential file and get last record
  LOCAL sFile AS STRING
  LOCAL h     AS LONG
  LOCAL s     AS STRING
  LOCAL x     AS LONG
  sFile =    "c:\test.txt"
  h = FREEFILE
  OPEN sFile FOR OUTPUT AS #h
  PRINT #h, "Record 1"
  PRINT #h, "Record 2 here"
  PRINT #h, "Record 3 is the last record in the file"
  CLOSE #h
  ? $DQ + GetLastRecordSequential(sFile) + $DQ
  SLEEP 3000
END FUNCTION
FUNCTION GetLastRecordSequential(sFile AS STRING) AS STRING
  LOCAL x AS LONG
  LOCAL s AS STRING
  LOCAL h AS LONG
  h = FREEFILE
  OPEN sFile FOR BINARY AS #h
  IF LOF(h) = 0 THEN
     CLOSE #h
     BEEP
     EXIT FUNCTION
  END IF
  s = " "
  FOR x = LOF(h)-1 TO 1 STEP -1
    GET #h, x, s
    IF ASC(s) = 10 THEN            'search for previous line feed
      s = SPACE$(LOF(h)-x-2)
      GET #h, x+1,s
      EXIT FOR
    END IF
  NEXT
  IF x = 0 THEN                    'no line feed found, entire file is last record
    s = SPACE$(LOF(h)-2)
    GET #h, 1, s
  END IF
  CLOSE #h
  FUNCTION = s
END FUNCTION

[Cliff Nichols]

Much faster to read it all in and then look for the record you want anyways.


Less read, writes, and case in point, try reading it all from a floppy vs on your hard-drive....you will see phenomenal speed differences

[Michael Mattias]

> think it will have to be filescan and read the file

If all you want is the last record, it does not have to be that, as pointed out above

But if you don't feel like doing that little bit, there's no need to FILESCAN:

 hin = FREEFILE
 OPEN "thefile" for INPUT AS hIN
 WHILE NOT EOF(hIn)
   LINE INPUT#hIn, S 
 WEND
 ' when you get here, 'S' is the last sequential record.
 CLOSE hin

MCM

[Kerry Farmer]

Thanks

I think it will have to be filescan and read the file

[Marco Pontello]

Or, for a text file, you can open it in binary mode, and start reading backward blocks of bytes (for example, 16KB), until you find a $CRLF in it.

Bye!

[Kev Peel]

Do you mean for RANDOM file access? If so, a simple calculation can be used to get the record count:

rec_count = LOF(nFile) \ rec_size

If opening as INPUT with variable-length data then no, although FILESCAN is pretty quick and may help. If either of these methods is not practical then it might be better for you to maintain a separate index file with the record positions.

[Kerry Farmer]

Is there anyway of accessing the last record in a sequential file without reading the whole file?