Category : Utilities for DOS and Windows Machines
Archive   : NJSORT11.ZIP
Filename : NJSORT.DOC

Output of file : NJSORT.DOC contained in archive : NJSORT11.ZIP

Nifty James' Famous Sort Tool
Version 1.00 of 08 May 1988
Version 1.05 of 19 May 1988
Version 1.10 of 03 July 1988
Version 1.11 of 30 October 1988

(C) Copyright 1988 by Mike "Nifty James" Blaszczak


It is very often necessary to sort information. It makes the
retrevial of information more rapid, and gives the data a more
humanized look. It also facilitates maintenance of lists.

Thus, having a powerful sorting tool is essential to good
productivity. This program is capable of very large lists, and is
limited only by the availability of storage space. It runs very
quickly, using optimized C code and hand-written assembly language
for the more crucial routines.

Using NJSORT is quite easy; it is run from the command line using a
traditional syntax:

NJSORT [] []

is the name of the input file -- the file to be sorted.
It must be present. If it is not, NJSORT will show its usage and
syntax and terminate.

specifies the name of the output file. If this name is
not present, NJSORT will write the sorted output to the same disk and
directory as the input fille, but will change the extention to
".BAK". For this reason, you must specify an alternate output file
name if you would like to use NJSORT to sort a file with the
extention of ".BAK".

is a list of options. These are the valid options:

/B Strip blank lines from the input. If a line is found to
have nothing on it, or only tabs and spaces on it, NJSORT
will remove it from the output file if this option is

/C Case Insensitive sort. This will tell NJSORT not to pay
attention to the case of the text it is sorting. NJSORT
will consider "lowercase" equal to "LOWERCASE", for example.

/D Remove duplicate lines. This will cause NJSORT to remove
identical lines from the output. If /C is specified, the
comparison for duplicate lines will not be case sensitive.
If /I is specified, the check for duplicate lines will not
pay attention to leading whitespace.

/I Ignore leading whitespace. This will cause NJSORT to ignore
the tabs or spaces at the beginning of a line.

/R Sort in reverse order. Normally, NJSORT sorts starting with
the lowest ASCII code and going up. Thus, 'A' comes before
'Z'. If this option is specified, NJSORT will sort starting
with the highest ASCII code and going down: 'Z' will come
before 'A'.

/V Display sort statistics. This verbose mode will display
information about the sorting process after it has completed.

Normally, NJSORT sorts ASCII files by looking at each line as a
separate item. Often times, there may be more than one sort field on
each line. It may also be desirable to sort the text based on the
content of a certain colum or field of data.

To make use of this feature, these options may be used:

/F# Specify the first key field number
/S# Specify the second key field number
/T# Specify the third key field number
/Q# Specify the fourth key field number
/P# Specify the fifth key field number
/M, Specify field delimiter
/N Sort by columns and not fields

These options define the fields to be used, and their respective
"weight" in the sort. When no fields are specified, NJSORT uses the
entire line starting at the beginning as the sort key.

When a key field is specified, NJSORT checks them in order. If the
first key field of two items are equal, then the second key field is
checked. If the second key field is equal, then the third is
compared, and so on. If all the specified key fields are equal, or
there are no more key fields on the line, the items are assumed to be
equal and their final order is arbitrary.

Since I maintain a database of the records in my collection, I have
an ASCII file with this format:


A typical entry might look like this:

Pink Floyd,Dark Side of The Moon; The,Money,6:23,0201,73

I can sort this database using the command

NJSORT MUSIC.DAT /F1 /S2 /T3 /Q5 /F6 /M, /C

This uses the /C option to make the sort case insensitive; thus,
"PINK FLOYD" will be equal to "Pink Floyd". The program is told
which keys to order the sort with, and it is told that the comma
spearates the felds of data.

Thus, when NJSORT sees the line in the file, it looks at it like

Pink Floyd,Dark Side of The Moon; The,Money,6:23,0201,73
Field 1 | Field 2 | 3 | | 4 | 5

When it sorts the file, the list will be ordered by the name of the
artist. "Pink Floyd" will come before "ZZ Top". If NJSORT finds
that the first field of two items are equal, it will check the second
field. "Dark Side of The Moon; The" comes before "Wall; The", even
thoug they are both Pink Floyd albums.

If the program finds that the album names in the second field are
equal, it will check the third field, which is the song title. In
the unlikely event they are also equal, the program will also go on
to check the side and cut of the song on the album, and finally the
copyright date of the song.

This allows you to sort with multiple keys on fields with variable
length. If the file, however, is listed in columns, for example:

Pink Floyd Dark Side of The Moon; The Money
Pink Floyd Dark Side of The Moon; The Eclipse

NJSORT should be given the option /N, which tells the program to use
the numbers provided ot the key field options as column numbers. So,
to sort the above file, one would use


NJSORT will reguard the field starting in the first colum as the
first sort key. The second key will start in column 16, and the
third key starts in column 45. To have NJSORT regard these numbers
as column numbers, the /N option must be specified.

With Version 1.10, NJSORT also has the ability to sort binary files
that contain fixed-length records. You can specify the offset of
the sort keys in each record using the /F, /S, /T, /Q, and /P
options, as normal. You must also specify the /W option. The /W
option must be immediately followed by the number of bytes in each
record of the input file.

NJSORT will sort the records, and will write out a binary file of the
same format -- just the records will be rearranged. If NJSORT can't
read a full record, it will stop and print an error message.

Using NJSORT is as simple as that, even for all its power and

NJSORT will read the datafile when it is invoked, and will then
attempt to sort all the information in memory. If the program cannot
load the entire input file in one attempt, it will sort the contents
of memory and write it to a temporary disk file. It will then clear
the memory, and read again. Once all the input has been read, the
program will then merge the temporary files into an output file.

NJSORT will look for the environment variables NJTEMP, TEMP, and TMP
to see if a path or drive for the temporary files was specified. The
first variable found will be used for the temporary file area.
Specifying no variable will result in the temporary files being
stored on the current drive in the current directory. Using a large
RAM disk, such as NJRAMD, for the temporary files will greatly
increase the speed with which the merge takes place. The area
specified for the temporary area, however, must be large enough to
hold the entire file.

This "sort and merge" method allows NJSORT to sort files that are
much larger than memory. However, in the worst case, NJSORT will
need twice the lenght of the file available as free space to sort
the file. If, for example, you have a 100000 character file to
sort, you must have at least 200000 bytes of free space on the
diskette you wish to sort on.

If you specify an alternate temporary directory, NJSORT only needs
enough space to store the output file. The disk will contain the
100000 byte file and the 100000 byte output file, and the TEMP
directory will contain the 100000 bytes of temporary files.

On my 640k system, with 580k after DOS and all TSRs are loaded,
NJSORT can typically sort files between 450 and 500 kilobytes in
length without having to use temporary disk files.

The complete sourcecode to the program is included in NJSORT.C and
COMPARES.ASM. The makefile used to build the program is called
NJSORT. I used BREIF to edit all the files here. Since I am a fool
for good formatting, I set BRIEF's tab stops at every three
characters for .C files and at the

This program is a work of Shareware. I am requesting a $10
registration if you like the program. If you do not like, or do not
use, the program, please pass it on to a friend or another BBS.

Currently, I have not found any bugs in NJSORT. I have run billions
of bytes of data through it in constructing and testing the program.
I have not found any problems with even the largest files using many
keys and fields.

One note, however. Even if the input file contains an end of file
marker, NJSORT will NOT write the marker to the output file. For
this reason, the output file may be one byte shorter than the input

Thank you very much!

Mike "Nifty James" Blaszczak
35 Ginger Lane #229
East Hartford, Connecticut

Version 1.10 Changes ---

Version 1.10 implemented the /W# option to allow sorting of
fixed-length fields. This option may be specified to cause NJSORT to
treat the input file as a file of records that are the of the
specified length. No terminating character will be looked for in the
records; they will simply be read as blocks of bytes.

This option does not work with the /I option to ignore whitespace.
It also will not work with the /C option to fix case, or the /D
option to blast duplicate lines. Of course, the /N and /M options
are not allowed, either, as they support different reading methods.

The sort key options may be used. They will specify the byte offsets
into the records at which keys should be considered. /F0 is the
first byte of the record, for example.

If the file to be sorted contains incomplete records, NJSORT will not
complete the sort. Besides swapping the records, no other action is
taken upon the binary data.

The new version contains a small change that allows more items to be
sorted in memory. Previously, NJSORT would create a temporary file
if more than 8000 items need to be sorted. Now, NJSORT will wait
until there are 13000 items in memory before creating a
temporary file. This improves efficiency if the items in memory are
all rather short.

Version 1.10 corrected some bugs that came about when sorting files
larger than memory. There were a handful of other minor bugs
corrected throughout the program.

In Version 1.10, the COMPARES.ASM module was changed for the
Microsoft Macro Assembler Version 5.10. This will allow
compatibility for a possible later release of NJSORT which will use
the newer Microsoft Languages to provide compatibility with NJSORT
and versions of OS/2.

Version 1.11 Changes ---

A user in Hawaii provided most of the feedback that brought about
this new version. A problem in NJSORT's merge caused records to be
dropped in some larger-than-memory sorts. This problem has been
corrected, and NJSORT should no longer lose information while sorting
larger files.

An additional option has also been added to NJSORT; /O#. This option
will allow the user to have NJSORT ignore the first records of the
file to be sorted. As an example, /O5 would cause NJSORT to read
the first five records or lines from the input file and write them to
the ouput file without changing their order.

NJSORT underwent a few other changes to improve its speed. The
program should prove to be slightly more efficient than the previous

Previously, NJSORT would incorrectly report the number of records or
lines sorted. This has been fixed in the new version.

NJSORT Error Messages

These are the error messages that NJSORT can generate, and the cause
for each:

NJSORT: Can't Sort a file with extension .BAK
A file with the extension of .BAK was supplied to NJSORT as
the filename to sort. Since NJSORT names the sorted output
with the filename extension of .BAK, the program will not
sort input files with the extension .BAK

NJSORT: Not enough memory
There was not enough memory to set up the basic housekeeping
data areas that NJSORT needs. Generally, NSORT will require
no more ten kilobytes of memory for its own internal data and
buffer space. You should never see this error -- if you do,
please make a note of the circumstances (computer, main
memory, TSRs loaded, file size to be sorted, command line
options, and DOS version) and send me the information.

NJSORT: Can't write record number # in file "Filename"
NJSORT: Can't write line number # in file "Filename"
These errors can occur when NJSORT can not write to a file.
If NJSORT runs out of space while writing a temporary file,
or if the program runs out of space while writing the ouput
file, this error will result. Try to create more free space
on the device, or use another device that has more room.

NJSORT: Record # was incompletely read, only # of # bytes
In a Binary record sort, if NJSORT tries to read a record of
the specified length and actually reads less information than
that, this error will result. The most common cause of this
error is specifying a record length that is not divisible by
the size of the file to be sorted. For example, if the file
is 1050 bytes long and a record length of 100 was specified,
NJSORT would only find ten 100 byte records; the remaining 50
bytes would not constitute a full, sortable record.

NJSORT: Input line # exceeds 2047 characters
NJSORT cannot handle records longer than 2047 characters when
doing an ASCII sort. The input file is simply too wide to
sort. This error also comes about when NJSORT gets a binary
file and wasn't given the /W parameter.

NJSORT: there weren't enough input records to skip
You told NJSORT to skip more records than were actually found
in the input file. Lessen the number of records to skip.

NJSORT: can't have both and
When parsing the options specified, NJSORT found two
conflicting options. Some options are incompatible; for
example, you can not specify both /W and /M.

NJSORT: key offset greater than recwidth
In a Binary Sort, NJSORT realized a record with was specified
and one of the keys was given as having an offset greater than
that record width. Check the parameters that you gave.

NJSORT: must have at least one field with /M or /N
The /M and /N options imply the use of sorting fields, and no
field size or offset was specified.

NJSORT: was too big
One of the specified key widths was larger than 2047, the
maximum record size for NJSORT.

NJSORT: Can't open for input
The specified file was not found or was not available to
NJSORT for read access.

NJSORT: Can't open for output
The requested or default output file was not valid or could
not be created by NJSORT.

NJSORT: Too many temporary files! Could not sort.
NJSORT needed to use more than eighteen temporary files, and
could not do so. This error should only occur when sorting
files in the range of four to ten megabytes -- if it happens
in less demanding circumstances, inform the author.

NJSORT: Couldn't setup temporary buffer
A problem occurred with NJSORT while trying to manipulate a
temporary file. This error should rarely occur. If it
happens twice, make a note of the environment and notify the

NJSORT: Couldn't open file for temporary use
There was a problem in accessing a temporary file. This
should not occurr, unless the temporary file area ran out of
space just as NJSORT was opening the file. Also, in a
multiuser environment, if another program attempts to
manipulate NJSORT's temporary file, this error may result.

  3 Responses to “Category : Utilities for DOS and Windows Machines
Archive   : NJSORT11.ZIP
Filename : NJSORT.DOC

  1. Very nice! Thank you for this wonderful archive. I wonder why I found it only now. Long live the BBS file archives!

  2. This is so awesome! 😀 I’d be cool if you could download an entire archive of this at once, though.

  3. But one thing that puzzles me is the “mtswslnkmcjklsdlsbdmMICROSOFT” string. There is an article about it here. It is definitely worth a read: