Dec 232017
An great extended sort module for OPDIR in Object Professional v1.0. Object Professional is an excellent toolkit by Turbo Power. | |||
---|---|---|---|
File Name | File Size | Zip Size | Zip Type |
EXTSORT.TXT | 9527 | 3716 | deflated |
OPDIRX.PAS | 2062 | 588 | deflated |
Download File OPDIRX.ZIP Here
Contents of the EXTSORT.TXT file
EXTENSION SORT FOR OPDIR.TPU
or
THE SORT METHOD TURBOPOWER SORT OF FORGOT
Copyright 1990
by Gil Yoder [73237,3103]
Portions Copyright 1989 by TurboPower Software
Permission granted to copy and distribute
if such is done without profit
Thank you TurboPower for a great new product! Object
Professional is great. Although at first glance the packaged
seemed unwieldy (my wife had to get a neighbor to help her carry
it into the house ;->), after reading about a third of the
printed documentation I am coming to the realization that
implementing the features of this tool box will be much easier
than its predecessor, Turbo Professional.
There are, however, just a few things one could wish for; not
much, but a few. For example OPDIR needs at least one new sort
procedure for directory lists. You include sorts based on name
and extension; directory, name and extension; size, name and
extension; and date, name and extension; but where are the
procedures for sorting by directory, extension and name? For
picking filenames out of large directories this extra method of
sorting is essential.
Really, I am not complaining! In fact you said, "If you need to
write a sort procedure that isn't already covered, study the
sort routines in OPDIR.IN1" (volume 1, page 4-246). That is
what I did. The code that you have written is so well
documented and easy to understand, I was able to prepare two new
sort methods within a matter of about 30 minutes. Below are the
changes I have made to OPDIR.IN1 and OPDIR.PAS to implement
sorting by extension and name, and sorting by directory,
extension and name.
These four lines were added to OPDIR.PAS in the INTERFACE
section to make the procedures accessible to the outside. Add
them along with the other Sort procedures (search for
"SortName").
procedure SortExt(DirPtr : DirListPtr);
{-Sort alphabetically by extension, then by name}
procedure SortDirExt(DirPtr : DirListPtr);
{-Sort directories first, then alphabetically by extension and name}
These two procedures are added along with two local functions to
OPDIR.IN1. If you keep the order given here, you can put them
just about anywhere in the file, but it would be wise to search
for the sort routines and place these routines with those.
(Again search for "SortName.")
function LessExt(var X, Y : DirRec) : Boolean;
{-Sort ordering -- alphabetically by extension then name}
var
Xdrive : Boolean;
Ydrive : Boolean;
begin
Xdrive := (X.Attr = diDriveAttr);
Ydrive := (Y.Attr = diDriveAttr);
if Xdrive = Ydrive then
if ByteFlagIsSet(X.Attr, Directory) then
LessExt := (X.Name < Y.Name)
else
LessExt := (JustExtension(X.Name)+X.Name <
JustExtension(Y.Name)+Y.Name)
else
LessExt := Xdrive;
end;
procedure SortExt(DirPtr : DirListPtr);
{-Sort alphabetically by name, then by extension}
begin
with DirPtr^ do begin
diLess := LessExt;
diQuickSort(1, pkItems);
end;
end;
function LessDirExt(var X, Y : DirRec) : Boolean;
{-Sort directories first, then alphabetically by extension and name}
var
Xdir : Boolean;
Ydir : Boolean;
Xdrive : Boolean;
Ydrive : Boolean;
begin
Xdrive := (X.Attr = diDriveAttr);
Ydrive := (Y.Attr = diDriveAttr);
if Xdrive = Ydrive then begin
Xdir := ByteFlagIsSet(X.Attr, Directory);
Ydir := ByteFlagIsSet(Y.Attr, Directory);
if Xdir = YDir then
LessDirExt := LessExt(X, Y)
else
LessDirExt := Xdir;
end else
LessDirExt := Xdrive;
end;
procedure SortDirExt(DirPtr : DirListPtr);
{-Sort directories first, then alphabetically by extension and name}
begin
with DirPtr^ do begin
diLess := LessDirExt;
diQuickSort(1, pkItems);
end;
end;
[I can't take credit for these routines, because they are only
slight variations of the routines already supplied with OPRO.]
After adding these routines to OPDIR, I modified DESKPOP to take
advantage of the new sort methods. Two of DESKPOP's source
files need changing. DESKPOP.ICD must be changed on one line.
Change line 11 from this...
DirSortType = (dstName, dstDate, dstSize, dstDos);
... to this...
DirSortType = (dstName, dstExt, dstDate, dstSize, dstDos);
If you want to make Extension sorting the default, change line
57 from this...
CurDirSortType : DirSortType = dstName; {sort method}
... to this...
CurDirSortType : DirSortType = dstExt; {sort method}
DESKMAIN.IN1 needs to be changed in a couple of places. Change
line 353 from this...
'Name ', 'Date ', 'Size ', 'MS-DOS');
... to this...
'Name ', 'Ext ', 'Date ', 'Size ', 'MS-DOS');
... and then add this at line 568
dstExt : SetSortOrder(SortDirExt);
Make these few changes to the files mentioned and compile with
TPC. You will need about 520k memory, and should specify /L to
link to disk. Otherwise all your changes will be for nought.
What else could a programmer want? Well so far this is my wish
list:
1. A multiple choice pick list that returns choices in a linked
list or array rather than a BitSet. BitSets are efficient
for small lists, but the larger the pick list the less
efficient BitSets become. I have an application requiring
multiple choices from a list of over 12,000 items! That
would require a structure containing over 1,500 bytes. That's
a lot of data to reference 1 to 20 items.
2. A quick incremental search for large pick lists (same
project). This will probably be easy to program, given
objects and the excellent design of OPRO. If you can assume
(or determine) that a list is in alphabetical order, for a normal
incremental search there is way too much overhead in
comparing every item with the search string. A much faster
approach is to compare a mid point with the search string.
If there is no match there, the list can be divided in half
and the half that could contain the string could be searched
in the same way. Eventually the string would be found, or
else no other portion of the list could be divided. If the
string is found, then each string one less than that could be
compared with the string, until no match was found. The new
item would be the last string found that matched the search
string. I implemented this method with TurboPower, and
found that I could do an incremental search in an
alphabetical list of 12,500 words and update the bar cursor
as fast as anyone could type.
3. A larger index. Six pages for three volumes of material as
rich as you have is just a little conservative.
All the same Object Professional is a great package; it's hard
to imagine how you produced it in the short time that you did
(documentation problems notwithstanding). Thank you, and keep
up the good work.
NOTE: Except for small changes the code listed in this article
is taken from code found in the Object Professional package.
The copyright (c) 1989 to the code found here belongs to TurboPower
Software, P.O. Box 66747, Scotts Valley, CA 95066. CompuServe
76004,2611. The people at TurboPower have graciously granted
permission to place this file at the disposal of their customers
by means of CompuServe or any other avenue.
POST SCRIPT
Because of the similarity between the procedures found here and
those found in OPDIR, before uploading this file I sent it to
TurboPower to ask for their permission to make it available to
others. Kim Kokkonen replied...
>> Gil - nice job on your "article." I don't have any problem
>> with your posting it on PCVENB or elsewhere. One
>> suggestion: your new sorts for OPDIR can just as easily be
>> implemented without changing any of our source files. I'd
>> suggest that you just write a standalone unit that
>> includes your new functions for sorting. The DESKPOP
>> changes, of course, would be slightly modified.
>>
>> Regarding your suggestions: 1) it would be easy to
>> post-process the OPPICK selected bitset to turn it into a
>> linked list and then dispose of the bitset. For
>> performance reasons, OPPICK must have indexed access to
>> the elements of the bitset while it is processing, so a
>> linked list is not a good choice internally. 2) the
>> searching functions for the pick list were designed to be
>> extensible. I think that you could write a separate unit
>> using the same kind of methodology you did for OPDIR
>> sorting that would provide the kind of incremental search
>> you want. 3) I agree that the index could be expanded.
>> Someday when we have time we'll go through and beef it up
>> and create a text file that people can use.
Comments: The suggestion to place the procedures and functions
in a standalone unit is a good one. As far as possible it is
best not to modify TurboPower's routines unless official changes
are released. OpDirX.pas implements both of the preceeding
sorts without changing OpDir. The changes detailed above for
DESKPOP.ICD and DESKMAIN.IN1 are still necessary. In addition
DESKMAIN.PAS needs to add OPDIRX in its "uses" statement.
Enjoy.
or
THE SORT METHOD TURBOPOWER SORT OF FORGOT
Copyright 1990
by Gil Yoder [73237,3103]
Portions Copyright 1989 by TurboPower Software
Permission granted to copy and distribute
if such is done without profit
Thank you TurboPower for a great new product! Object
Professional is great. Although at first glance the packaged
seemed unwieldy (my wife had to get a neighbor to help her carry
it into the house ;->), after reading about a third of the
printed documentation I am coming to the realization that
implementing the features of this tool box will be much easier
than its predecessor, Turbo Professional.
There are, however, just a few things one could wish for; not
much, but a few. For example OPDIR needs at least one new sort
procedure for directory lists. You include sorts based on name
and extension; directory, name and extension; size, name and
extension; and date, name and extension; but where are the
procedures for sorting by directory, extension and name? For
picking filenames out of large directories this extra method of
sorting is essential.
Really, I am not complaining! In fact you said, "If you need to
write a sort procedure that isn't already covered, study the
sort routines in OPDIR.IN1" (volume 1, page 4-246). That is
what I did. The code that you have written is so well
documented and easy to understand, I was able to prepare two new
sort methods within a matter of about 30 minutes. Below are the
changes I have made to OPDIR.IN1 and OPDIR.PAS to implement
sorting by extension and name, and sorting by directory,
extension and name.
These four lines were added to OPDIR.PAS in the INTERFACE
section to make the procedures accessible to the outside. Add
them along with the other Sort procedures (search for
"SortName").
procedure SortExt(DirPtr : DirListPtr);
{-Sort alphabetically by extension, then by name}
procedure SortDirExt(DirPtr : DirListPtr);
{-Sort directories first, then alphabetically by extension and name}
These two procedures are added along with two local functions to
OPDIR.IN1. If you keep the order given here, you can put them
just about anywhere in the file, but it would be wise to search
for the sort routines and place these routines with those.
(Again search for "SortName.")
function LessExt(var X, Y : DirRec) : Boolean;
{-Sort ordering -- alphabetically by extension then name}
var
Xdrive : Boolean;
Ydrive : Boolean;
begin
Xdrive := (X.Attr = diDriveAttr);
Ydrive := (Y.Attr = diDriveAttr);
if Xdrive = Ydrive then
if ByteFlagIsSet(X.Attr, Directory) then
LessExt := (X.Name < Y.Name)
else
LessExt := (JustExtension(X.Name)+X.Name <
JustExtension(Y.Name)+Y.Name)
else
LessExt := Xdrive;
end;
procedure SortExt(DirPtr : DirListPtr);
{-Sort alphabetically by name, then by extension}
begin
with DirPtr^ do begin
diLess := LessExt;
diQuickSort(1, pkItems);
end;
end;
function LessDirExt(var X, Y : DirRec) : Boolean;
{-Sort directories first, then alphabetically by extension and name}
var
Xdir : Boolean;
Ydir : Boolean;
Xdrive : Boolean;
Ydrive : Boolean;
begin
Xdrive := (X.Attr = diDriveAttr);
Ydrive := (Y.Attr = diDriveAttr);
if Xdrive = Ydrive then begin
Xdir := ByteFlagIsSet(X.Attr, Directory);
Ydir := ByteFlagIsSet(Y.Attr, Directory);
if Xdir = YDir then
LessDirExt := LessExt(X, Y)
else
LessDirExt := Xdir;
end else
LessDirExt := Xdrive;
end;
procedure SortDirExt(DirPtr : DirListPtr);
{-Sort directories first, then alphabetically by extension and name}
begin
with DirPtr^ do begin
diLess := LessDirExt;
diQuickSort(1, pkItems);
end;
end;
[I can't take credit for these routines, because they are only
slight variations of the routines already supplied with OPRO.]
After adding these routines to OPDIR, I modified DESKPOP to take
advantage of the new sort methods. Two of DESKPOP's source
files need changing. DESKPOP.ICD must be changed on one line.
Change line 11 from this...
DirSortType = (dstName, dstDate, dstSize, dstDos);
... to this...
DirSortType = (dstName, dstExt, dstDate, dstSize, dstDos);
If you want to make Extension sorting the default, change line
57 from this...
CurDirSortType : DirSortType = dstName; {sort method}
... to this...
CurDirSortType : DirSortType = dstExt; {sort method}
DESKMAIN.IN1 needs to be changed in a couple of places. Change
line 353 from this...
'Name ', 'Date ', 'Size ', 'MS-DOS');
... to this...
'Name ', 'Ext ', 'Date ', 'Size ', 'MS-DOS');
... and then add this at line 568
dstExt : SetSortOrder(SortDirExt);
Make these few changes to the files mentioned and compile with
TPC. You will need about 520k memory, and should specify /L to
link to disk. Otherwise all your changes will be for nought.
What else could a programmer want? Well so far this is my wish
list:
1. A multiple choice pick list that returns choices in a linked
list or array rather than a BitSet. BitSets are efficient
for small lists, but the larger the pick list the less
efficient BitSets become. I have an application requiring
multiple choices from a list of over 12,000 items! That
would require a structure containing over 1,500 bytes. That's
a lot of data to reference 1 to 20 items.
2. A quick incremental search for large pick lists (same
project). This will probably be easy to program, given
objects and the excellent design of OPRO. If you can assume
(or determine) that a list is in alphabetical order, for a normal
incremental search there is way too much overhead in
comparing every item with the search string. A much faster
approach is to compare a mid point with the search string.
If there is no match there, the list can be divided in half
and the half that could contain the string could be searched
in the same way. Eventually the string would be found, or
else no other portion of the list could be divided. If the
string is found, then each string one less than that could be
compared with the string, until no match was found. The new
item would be the last string found that matched the search
string. I implemented this method with TurboPower, and
found that I could do an incremental search in an
alphabetical list of 12,500 words and update the bar cursor
as fast as anyone could type.
3. A larger index. Six pages for three volumes of material as
rich as you have is just a little conservative.
All the same Object Professional is a great package; it's hard
to imagine how you produced it in the short time that you did
(documentation problems notwithstanding). Thank you, and keep
up the good work.
NOTE: Except for small changes the code listed in this article
is taken from code found in the Object Professional package.
The copyright (c) 1989 to the code found here belongs to TurboPower
Software, P.O. Box 66747, Scotts Valley, CA 95066. CompuServe
76004,2611. The people at TurboPower have graciously granted
permission to place this file at the disposal of their customers
by means of CompuServe or any other avenue.
POST SCRIPT
Because of the similarity between the procedures found here and
those found in OPDIR, before uploading this file I sent it to
TurboPower to ask for their permission to make it available to
others. Kim Kokkonen replied...
>> Gil - nice job on your "article." I don't have any problem
>> with your posting it on PCVENB or elsewhere. One
>> suggestion: your new sorts for OPDIR can just as easily be
>> implemented without changing any of our source files. I'd
>> suggest that you just write a standalone unit that
>> includes your new functions for sorting. The DESKPOP
>> changes, of course, would be slightly modified.
>>
>> Regarding your suggestions: 1) it would be easy to
>> post-process the OPPICK selected bitset to turn it into a
>> linked list and then dispose of the bitset. For
>> performance reasons, OPPICK must have indexed access to
>> the elements of the bitset while it is processing, so a
>> linked list is not a good choice internally. 2) the
>> searching functions for the pick list were designed to be
>> extensible. I think that you could write a separate unit
>> using the same kind of methodology you did for OPDIR
>> sorting that would provide the kind of incremental search
>> you want. 3) I agree that the index could be expanded.
>> Someday when we have time we'll go through and beef it up
>> and create a text file that people can use.
Comments: The suggestion to place the procedures and functions
in a standalone unit is a good one. As far as possible it is
best not to modify TurboPower's routines unless official changes
are released. OpDirX.pas implements both of the preceeding
sorts without changing OpDir. The changes detailed above for
DESKPOP.ICD and DESKMAIN.IN1 are still necessary. In addition
DESKMAIN.PAS needs to add OPDIRX in its "uses" statement.
Enjoy.
December 23, 2017
Add comments