[SOLVED]index based on unknown field
lazarus / 16-Databases

----------------------------------------------------------------------
micoudic · 2013-12-25 19:14

Hello,
I just installed a debian 7.3.0 amd-64 and install lazarus.
In a new project, I put a TDatasource and TDbf.
I connect the datasource to dbf. I setup an existing dbf table. I do
nothing more. When I want to make the active table, I get this
message:
  Code: [Select]index based on unknown field "+"
Thanks

----------------------------------------------------------------------
howardpc · 2013-12-25 20:33

This error indicates that your existing dbf file has an expression
index that cannot be parsed successfully.
My guess is that you may have an index that tries to concatenate two
or more field names and the syntax for the expression is not quite
right (at least for the Lazarus TDBF implementation - the syntax may
work in dbase or other dbf-aware database tools), so when the
expression is parsed the "+" in the expression is considered a field
name rather than a concatenation operator.

You could try making a backup, then deleting the index and indexdefs
to see if you get access to the data content. If so, you could then
recreate the index(es) you want using Lazarus TDBF routines.

----------------------------------------------------------------------
micoudic · 2013-12-27 08:46

Hello,
I'll try what you suggest
But curiously, this file was created with Borland tools with Delphi.
It should therefore be consistent with the standard?
Moreover, with an older version of Lazarus, and TDbf third party
component, it worked well ...
Thanks

----------------------------------------------------------------------
BigChimp · 2013-12-27 09:00

"Consistent with the standard"? What standard? Clipper? DBaseIV?
FoxPro? Some Borland BDE flavour perhaps?

As you don't specify what version of Lazarus and FPC you are using
(and which one used to work), what DBF format/tableleve, index file
format (.mdx? something else) you are using, your complaining is not
very useful, TBH.

BTW: Lazarus/FPC use the "TDBF third party component".... so yes,
there may be regressions in the implementation... or... you may
remember things incorrectly.

I'm happy to improve the FPC TDBF code in FPC trunk if we find the
problem, but I'll need some solid information rather than vague
descriptions, sorry.
(Hint: why not attach the dbf and a sample project (code only), and
provide the info)

----------------------------------------------------------------------
micoudic · 2013-12-28 07:48

- laraus V0.9.30.4
- fpc 2.6.0.9
I am under debian 7.3.0 amd64

The third party component is here http://tdbf.sourceforge.net/

I have nothing other than this:

- Ask a TDatasource
- Ask a TDbf
- Connect the TDataSource to TDbf
- Ask a TDBGrid and TDBNavigator and connect to TDatasource
- Define an existing dbf table
- Try to activate the current table, and it is at this time that the
message comes.

----------------------------------------------------------------------
howardpc · 2013-12-28 10:21

Mixing use of two different classes with the same name - a third party
TDBF, apparently last updated March 2007, and the Lazarus TDBF
(currently maintained) - is basically a flawed decision.
Never mind that the classes appear to have similar functionality and
probably nearly identical sets of identically-named methods and
properties. Whatever their history, they are now independent, and
bound to give all sorts of problems if you attempt to use them
together.
Just one of the sources of incompatibility is that the Lazarus TDBF
uses FCL db units which have been designed to be cross-platform. The
third party TDBF is based on Windows-only code.

----------------------------------------------------------------------
BigChimp · 2013-12-28 11:42

What howardpc said. I'd suggest using the FPC-supplied TDBF.
Additionally, FPC 2.6.2 has several fixes in TDBF; they may fix the
situation (or not). FPC trunk has still more fixes.

Quote from: bigchimpwhat DBF format/tableleve, index file format
(.mdx? something else) you are using
Quote from: bigchimpwhy not attach the dbf

----------------------------------------------------------------------
micoudic · 2014-01-04 20:25

Hello,
I just installed a new distribution on a blank partition. Here is my
setup:

My PC
moi[~]$ uname -a
Linux rbs457 3.2.29-smp #2 SMP Mon Sep 17 13:16:43 CDT 2012 i686 AMD
Athlon(tm) 64 Processor 3200+ AuthenticAMD GNU/Linux

moi[~]$ cat /etc/slackware-version
Slackware 14.0 (32 bits)

Lazarus : 1.0.14
date 2013-11-16
FPC 2.6.2
Revision : SVN:43447
i386-linux-gtk 2

I tried a re-testing of TDbf component, with the same result. I looked
on the side of the index file and made the following tests: I replaced
the character "+" by another to see the result. I did not update
components.
I'll let you see what happens.
http://imageshack.us/a/img845/9145/c9tl.jpg
http://imageshack.us/a/img560/8222/4tmt.jpg
http://imageshack.us/a/img543/9787/t0yw.jpg
http://imageshack.us/a/img849/6422/qima.jpg

My dbf file has a composite key ("APAYS" and "A_AKLE")

NB: I prepared a dbf file and mdx index, but I do not know how to tie
them.
I do not know how to display my images ( [img] .... [/ img] does not
work )

----------------------------------------------------------------------
cdbc · 2014-01-04 21:01

Hi
That looks like clipper compound indexes
Just my 2 cents...
Regards Benny

----------------------------------------------------------------------
micoudic · 2014-01-08 13:00

Hi,
I made a copy of my file using "dbd32.exe" Borland ( under Windows ).
I used this new file with Lazarus, and I no longer have the error when
rendering the active table.
But trying to display my data in a different order, via a different
key, nothing happens (the display remains in the same state).
After verification, the index were not copied to the new file. So I
re-created multiple indexes, always with the dbd32.exe program
Borland. Now Lazarus shows me the same error when I try to activate
the table.

I think a table generated with Borland product must conform to the
standard dBase?

----------------------------------------------------------------------
BigChimp · 2014-01-08 13:12

Il n'y a pas "un" standard dBase. Il y a plusieurs.
Voyez par exemple
http://sourceforge.net/projects/tdbf/
"TDBF is a native dBASE III+, dBase IV and dBase 2k data access
component for Delphi, BCB, Kylix, FreePascal"
Pas de Clipper la.

Si vous voulez, vous pouvez essayer d'ajouter support pour le format
Clipper et ses indices. S'il vous plait, faites un patch contre la
version de developpement actuelle de FPC (2.7.1/trunk).

(Excusez-moi pour mon pauvre francais)

edit: @benny: thanks!

----------------------------------------------------------------------
BigChimp · 2014-01-08 13:15

Quote from: micoudic on January 04, 2014, 09:25:51 pmNB: I prepared a
dbf file and mdx index, but I do not know how to tie them.
Do you mean you don't know how to upload them? Zip them and add them
as an attachment to a post. See Attachments and other options below
your message window when writing a new message.

----------------------------------------------------------------------
micoudic · 2014-01-10 21:23

Hi,
A big thank you your efforts to write to me in French  (Maybe because
my English is very bad ?  )

My file was not created in Clipper, but with the BDE Borland.
I downloaded TDBf and I'll install it to test.

I attach the dbase file with its index file

----------------------------------------------------------------------
BigChimp · 2014-01-11 01:17

Well, I think your English is better than my French

Quote from: micoudic on January 10, 2014, 10:23:39 pmMy file was not
created in Clipper, but with the BDE Borland.
Yes, YES, YES, but which dbase format did BDE create? DBase4,
dbaseIII, Clipper, Foxpro, something else? Apparently it's clipper.

Quote from: micoudic on January 10, 2014, 10:23:39 pmI downloaded TDBf
and I'll install it to test.
Don't download Tdbf if you're using Lazarus. It's built in. As
Howardpc said, don't mix the various versions.
If you're talking about using Tdbf with Delphi, sure, go ahead.

----------------------------------------------------------------------
micoudic · 2014-01-11 20:13

Quote from: BigChimp on January 11, 2014, 02:17:02 amWell, I think
your English is better than my French Quote
Google

Quote from: micoudic on January 10, 2014, 10:23:39 pmMy file was not
created in Clipper, but with the BDE Borland.
Quote from: BigChimp on January 11, 2014, 02:17:02 amYes, YES, YES,
but which dbase format did BDE create? DBase4, dbaseIII, Clipper,
Foxpro, something else? Apparently it's clipper.
Good question;) I created this file with Delphi5 (there a few years
ago ...) but I do not know what format?
I searched the net for information on the structure of dBase files,
but definitions vary across sites found: (
I attach the header of my file. It seems that this is dBase IV with
Memo (X8B Code). My file actually contains Memos.

Quote from: micoudic on January 10, 2014, 10:23:39 pmI downloaded TDBf
and I'll install it to test.

Quote from: BigChimp on January 11, 2014, 02:17:02 amDon't download
Tdbf if you're using Lazarus. It's built in. As Howardpc said, don't
mix the various versions.
If you're talking about using Tdbf with Delphi, sure, go ahead.

OK, thanks

----------------------------------------------------------------------
BigChimp · 2014-01-11 20:42

1. Just tried ${fpcdir}\packages\fcl-db\examples\dbftool in
trunk/2.7.1/development version of fpc. It cannot deal with the index
file: Error reading file 0: Index based on unknown field "+".
However, removing the mdx allowed for reading the dbf:
Code: [Select]Database tablelevel: 4
Database codepage:   1252
tablelevel 4=>tdbf thinks this is a DBase IV file

Code: [Select]*** Opening: D:\Cop\t\Membres.DBF
Database tablelevel: 4
Database codepage:   1252
Record 1
Field A_AKLE has value: 10
Field ANOM has value: GILETTE Frank
Field AADR1 has value: 42 rue Caban
Field AADR2 is          ***NULL***
Field ACPOST has value: 45000
Field AVILLE has value: ORLEANS
Field AEMAIL has value: monnom.prenom@monentreprise.fr
Field AWEB is          ***NULL***
Field ASTATUS is          ***NULL***
Field ADATRAD is          ***NULL***
Field ANOTES is          ***NULL***
Field APAYS has value: FRANCE
Field ADATEMAJ has value: 04/01/2014
Field ADATECREAT has value: 04/01/2014

Record 2
Field A_AKLE has value: 20
Field ANOM has value: LAMACHOIRE Roland
Field AADR1 has value: 8 rue des sapins
Field AADR2 is          ***NULL***
Field ACPOST has value: 33000
Field AVILLE has value: BORDEAUX
Field AEMAIL is          ***NULL***
Field AWEB is          ***NULL***
Field ASTATUS is          ***NULL***
Field ADATRAD is          ***NULL***
Field ANOTES is          ***NULL***
Field APAYS has value: ALLEMAGNE
Field ADATEMAJ is          ***NULL***
Field ADATECREAT has value: 04/01/2014

Record 3
Field A_AKLE has value: 30
Field ANOM has value: LEMARCHAND Serge
Field AADR1 has value: 16 route departementale
Field AADR2 is          ***NULL***
Field ACPOST has value: 03000
Field AVILLE has value: MOULINS
Field AEMAIL is          ***NULL***
Field AWEB is          ***NULL***
Field ASTATUS is          ***NULL***
Field ADATRAD is          ***NULL***
Field ANOTES is          ***NULL***
Field APAYS has value: ITALIE
Field ADATEMAJ is          ***NULL***
Field ADATECREAT has value: 05/01/2014

Record 4
Field A_AKLE has value: 40
Field ANOM has value: INTOSH Mac
Field AADR1 has value: 6 street Elisabeth II
Field AADR2 is          ***NULL***
Field ACPOST has value: 53800
Field AVILLE has value: LONDON
Field AEMAIL is          ***NULL***
Field AWEB is          ***NULL***
Field ASTATUS is          ***NULL***
Field ADATRAD is          ***NULL***
Field ANOTES is          ***NULL***
Field APAYS has value: ANGLETERRE
Field ADATEMAJ is          ***NULL***
Field ADATECREAT has value: 06/01/2014
You could remove the index file and recreate the index using tdbf; see
here for details:
http://wiki.lazarus.freepascal.org/Lazarus_Tdbf_Tutorial#How_to_add_in
dexes_to_a_table
... then you can use Lazarus with the indexes from then on (don't
remember by heart what format it uses - but it is quite probably
compatible with the BDE in case you want that).

2. You may look into ADS; perhaps that can read the existing index
file if you want to keep it:
http://wiki.lazarus.freepascal.org/Databases#Advantage_TDataSet_Descen
dant
(No idea if it can, have never used it myself)

----------------------------------------------------------------------
micoudic · 2014-01-12 10:20

Quote from: BigChimp on January 11, 2014, 09:42:34 pm1. Just tried
${fpcdir}\packages\fcl-db\examples\dbftool in trunk/2.7.1/development
version of fpc. It cannot deal with the index file: Error reading file
0: Index based on unknown field "+".his file is a file for testing
(reduced to a single index or memo, just for testing)
Quote from: BigChimp on January 11, 2014, 09:42:34 pmHowever, removing
the mdx allowed for reading the dbf:Yes, I had noticed. But in this
case I can't display my data in a different order.
I'll do as you say: delete indexes, and then re - create with lazarus.
I found a wiki in French
http://wiki.freepascal.org/Lazarus_Tdbf_Tutorial/fr.

many tanks

----------------------------------------------------------------------
BigChimp · 2014-01-12 10:28

Quote from: micoudic on January 12, 2014, 11:20:57 amQuote from:
BigChimp on January 11, 2014, 09:42:34 pm1. Just tried
${fpcdir}\packages\fcl-db\examples\dbftool in trunk/2.7.1/development
version of fpc. It cannot deal with the index file: Error reading file
0: Index based on unknown field "+".his file is a file for testing
(reduced to a single index or memo, just for testing)

That's why I decided to test it for you with FPC trunk in case you had
not tested it with fpc trunk
(I did understand it's a testing file; those email addresses and
street names, as well as surnames, would be very unusual to see in a
production database  )

Quote from: micoudic on January 12, 2014, 11:20:57 amYes, I had
noticed. But in this case I can't display my data in a different
order.
I'll do as you say: delete indexes, and then re - create with lazarus.
I found a wiki in French
http://wiki.freepascal.org/Lazarus_Tdbf_Tutorial/fr.
many tanks

Bonne chance

----------------------------------------------------------------------
micoudic · 2014-01-12 16:18

Is the copy of the file real, but empty of all his datas and I re-
created a few registration only.
I have not filled all the fields, only a few.

After extensive research and testing, I finally found the syntax to
create a multiple index.
  Code: [Select]Dbf1.AddIndex('KNOMKLE', 'ANOM' + '+AADR1' ,
[ixCaseInsensitive]);When I visualise the mdx file, I see no
difference in syntax.

This works because both fields are of the same type. But I want to
create an index with the fields "Name" and "A_AKLE" (String and
SmallInt). I even can't do the translation of the key A_AKLE in a
String field. StrToInt is rejected, and I found nothing else

----------------------------------------------------------------------
BigChimp · 2014-01-13 07:59

1. IIRC, there was a post recently about compound indexes with a
solution. Nobody added anything to the wiki as usual, so I did it just
now  => http://wiki.lazarus.freepascal.org/Lazarus_Tdbf_Tutorial#Index
_on_multiple_fields
2. Regarding inttostr, have you looked at the STR function (http://wik
i.lazarus.freepascal.org)?/Lazarus_Tdbf_Tutorial#Index.2Ffilter_operat
ors.2Ffunctions)
3. I would very much appreciate to see your additions/updates to the
French dbf tutorial (and/or the English one).. I'll be happy to
translate if needed (to English, that is  )

----------------------------------------------------------------------
micoudic · 2014-01-13 20:04

-I could nothing do with the STR function.
-If I complete the name of the "A_AKLE" field with .asString
(A_AKLE.AsString), I get an index name "ANOM10", 10 being the contents
of the field A_AKLE of the 1st record in my file "Membres.dbf"
-If I normally codifies the index, jobtiens the error cited in the
title of this thread (Index based on unknown field "+")
I'll leave it at that and return to my Delphi. Too bad.

NB
Code: [Select]Dbf1.AddIndex('KNOMKLE', 'ANOM+' + 'AADR1' ,
[ixCaseInsensitive]); is identical to
Code: [Select]Dbf1.AddIndex('KNOMKLE', 'ANOM' + '+AADR1' ,
[ixCaseInsensitive]); I tested

----------------------------------------------------------------------
BigChimp · 2014-01-14 06:58

Hope it works for you with Delphi. Don't understand what your last
remark about concatenated strings is supposed to mean...

----------------------------------------------------------------------
micoudic · 2014-01-14 08:41

Hi,
Quote from: BigChimp on January 14, 2014, 07:58:42 amHope it works for
you with Delphi.
It works very well (application developed with Delphi5, and subsequent
versions, hence my surprise that I can not do operate these indexes
with Lazarus.)
Quote from: BigChimp on January 14, 2014, 07:58:42 amDon't understand
what your last remark about concatenated strings is supposed to
mean...
It was in connection with my previous thread and for information. I
had so much trouble finding this syntax.

----------------------------------------------------------------------
BigChimp · 2014-01-14 16:59

Quote from: micoudic on January 14, 2014, 09:41:02 amIt works very
well (application developed with Delphi5, and subsequent versions,
hence my surprise that I can not do operate these indexes with
Lazarus.)
Hmm, I think there is a bug. Need to verify again but put it on the
bugtracker as a reminder:
http://bugs.freepascal.org/view.php?id=25533

----------------------------------------------------------------------
micoudic · 2014-01-15 21:05

hello

I looked at the sites for which you have provided me the links and I
did other tests.


Reminder: the "A_AKLE" field is of type SmallInt in the file
"Membres.dbf"

With this syntax,
Code: [Select]DBf1.AddIndex('KNOMKLE', ('ANOM+' +
IntToStr(Dbf1A_AKLE)),[]);the compiler sends to me following message:
Code: [Select]/home/moi/Desktop/Lazarus_dbf/Test_Tris/Bis/utestdbf.pas
(109,58) Error: Incompatible type for arg no. 1: Got "TSmallintField",
expected "QWord"
and with this syntax
Code: [Select]DBf1.AddIndex('KNOMKLE', ('ANOM+' + Dbf1A_AKLE),[]);I
get this message:
Code: [Select]/home/moi/Desktop/Lazarus_dbf/Test_Tris/Bis/utestdbf.pas
(109,37) Error: Operator is not overloaded: "Constant String" +
"TSmallintField"
===

Since I can't create the index with a field of type smallint and the
"IntToStr" function, I redid tests, by creating the table and indexes
by program, but by choosing the type integer for the A_AKLE field, in
order to test the IntToSTR function.

With this syntax (A_AKLE is defined "integer")
Code: [Select]AddIndex('KNOMKLE', 'ANOM', + '+A_AKLE', []); I get this
message
Code: [Select]unit1.pas(122,43) Error: Operator is not overloaded: +
"Constant String"
With this syntax
Code: [Select]AddIndex('KNOMKLE', ('ANOM+' + IntToStr(A_AKLE)), []); I
get this message
Code: [Select]unit1.pas(122,50) Error: Identifier not found "A_AKLE"
If I codifies "Dbf1A_AKLE", I get the same message:
Code: [Select]unit1.pas(122,56) Error: Identifier not found
"Dbf1A_AKLE"
Lazarus 1.10.1
Date 2013/11/16
fpc 2.6.2
Revision : SVN 43447
i386-linux gtk2

Salix 14.0 (32bits)

@ BigChimp
I did not understand if I have to watch this link or if I have to ask
the question again. In the latter case, I don't know how to do this

----------------------------------------------------------------------
BigChimp · 2014-01-16 09:54

Micoudic, no you do not have to ask your question again. I raised the
bug because the solution I was looking was did not work. See the
example code attached to the bug report.

Perhaps I have trouble understanding but this:
Code: [Select]DBf1.AddIndex('KNOMKLE', ('ANOM+' +
IntToStr(Dbf1A_AKLE)),[]);
you are trying to generate a string that contains
the string 'ANOM+' andthe string representation of whatever fpc thinks
Dbf1a_AKLE is
You then pass this as the expression for index 'KNOMKLE' to the TDBF
expression parser using the DBf1.AddIndex call.

In other words: the inttostr part is more static: it is an Object
Pascal function whose outcome is determined by the program at run
time, not by the TDBF filtering engine when looking at the data.

Fine if you want to do that, but I don't think this is your goal.

Have a look at this example on the English wiki article:
http://wiki.lazarus.freepascal.org/Lazarus_Tdbf_Tutorial#Index.2Ffilte
r_on_TDateTimes
where you use the TDBF expression parser DTOS() function to create an
index expression.

I think here you might be inserting a comma too mach after the 'ANOM'
part...
Code: [Select]AddIndex('KNOMKLE', 'ANOM', + '+A_AKLE', []);

Haven't looked at the other examples.

I haven't had my coffee yet so I hope this is clear...

----------------------------------------------------------------------
BigChimp · 2014-01-16 12:35

The "identifier not found" errors are because you are trying to use
DBF field/column names, apply an FPC function to it and pass the
result to the expression parser.
FPC has no way of knowing what e.g. A_AKLE is... it may know about
something like yourdbf.fields.fieldbyname('A_AKLE').asinteger but that
will give only the value for the record you currently are in (if the
dbf is even open) and will not help you.

You should not be passing these field names to FPC, you should be
passing them to the expression parser by including them inside the
expression index string - e.g.
Code: [Select]AddIndex('KNOMKLE', ('ANOM+' + IntToStr(A_AKLE)), []);
would become
Code: [Select]AddIndex('KNOMKLE', '(ANOM)+(IntToStr(A_AKLE))', []);
... if the dbf expression parser has an inttostr function, which it
hasn't.

I hope I have clarified instead of confused things...

----------------------------------------------------------------------
BigChimp · 2014-01-16 13:12

The demo program I used in my bug report was... buggy. Sorry.

This code should demonstrate creating an index exoression built up out
of:
1. a string field
2. an integer field
... and should give you an idea of how to adapt your code:
... I'm guessing something like:
Code: [Select]NewDBF.AddIndex('IDXCOMPO','(A_PAYS)+(STR(A_AKLE))',[ixC
aseInsensitive]);

Please don't forget to
1. check you actually apply the index (MyDBF.IndexName:= in the code
below)
2. and move to the first record before using it (see printrecords
below).

Code: [Select]program compoundindex2;

{
Demonstrates creating DBF tables, filling it with data,
and setting multifield index, including numeric field

Please delete any customer.dbf and customer.mdx if you have them in
your directory
so this program can recreate it.
}

{$mode objfpc}{$H+}

uses {$IFDEF UNIX} {$IFDEF UseCThreads}
  cthreads, {$ENDIF} {$ENDIF}
  Classes,
  SysUtils,
  CustApp,
  DB,
  dbf,
  dbf_fields,
  dbf_common,
  dateutils;

type

  { TDBFTool }

  TDBFTool = class(TCustomApplication)
  protected
    procedure DoRun; override;
  public
    constructor Create(TheOwner: TComponent); override;
    destructor Destroy; override;
  end;

  procedure CreateDemoDBFs(Directory: string);
  // Creates dbase4 demonstration DBF in Directory
  var
    NewDBF: TDBF;
    i: integer;
  begin

    NewDBF := TDBF.Create(nil);
    try
      if Directory = '' then
        NewDBF.FilePath := '' { application directory}
      else
        NewDBF.FilePathFull := ExpandFileName(Directory) {full
absolute path};
      NewDBF.TableLevel := 4; //default to DBase IV

      NewDBF.TableName := 'CUSTOMER.DBF';
      writeln('Creating ', NewDBF.TableName, ' with table level ',
NewDBF.TableLevel);
      NewDBF.FieldDefs.Add('CUST_NO', ftInteger);
      NewDBF.FieldDefs.Add('CUSTOMER', ftString, 25);
      NewDBF.FieldDefs.Add('CITY', ftString, 25);
      NewDBF.FieldDefs.Add('COUNTRY', ftString, 15);
      NewDBF.CreateTable;
      NewDBF.Exclusive:=true;
      NewDBF.Open;
      // Index on field customer (a string field) followed by the
string value of integer field cust_no
      NewDBF.AddIndex('IDXCOMPO','(CUSTOMER)+(STR(CUST_NO))',[ixCaseIn
sensitive]);

      // Another example: reverse sort on customer number:
      //NewDBF.AddIndex('IDXCOMPO','-CUST_NO',[ixCaseInsensitive]);

      // Compound index on the customer name followed by the customer
number sorted backward.
      // this doesn't seem to sort correctly on the customer number
part though
      //NewDBF.AddIndex('IDXCOMPO','CUSTOMER+(STR(CUST_NO*-
1))',[ixCaseInsensitive]);

      NewDBF.Exclusive:=False;
      NewDBF.Close;
      NewDBF.Open;

      for i := 1 to 5 do //keep size manageable until we have working
files
      begin
        NewDBF.Append;
        if (NewDBF.FieldDefs.Find('CUST_NO').DataType <> ftAutoInc)
then
          NewDBF.FieldByName('CUST_NO').AsInteger := i;
        case i of
          5:
          begin
            NewDBF.FieldByName('CUSTOMER').AsString := 'Michael
Design';
            NewDBF.FieldByName('CITY').AsString := 'San Diego';
            NewDBF.FieldByName('COUNTRY').AsString := 'USA';
          end;
          4:
          begin
            NewDBF.FieldByName('CUSTOMER').AsString := 'Johnson';
            NewDBF.FieldByName('CITY').AsString := 'San Ysidro';
            NewDBF.FieldByName('COUNTRY').AsString := 'Mexico';
          end;
          3:
          begin
            NewDBF.FieldByName('CUSTOMER').AsString := 'Michael
Design';
            NewDBF.FieldByName('CITY').AsString := 'Dallas';
            NewDBF.FieldByName('COUNTRY').AsString := 'USA';
          end;
          2:
          begin
            NewDBF.FieldByName('CUSTOMER').AsString := 'Johnson';
//same name as previous one, useful for sorting
            NewDBF.FieldByName('CITY').AsString := 'Boston';
            NewDBF.FieldByName('COUNTRY').AsString := 'USA';
          end;
          1:
          begin
            NewDBF.FieldByName('CUSTOMER').AsString := 'Zack
Zucchini';
            NewDBF.FieldByName('CITY').AsString := 'Manchester';
            NewDBF.FieldByName('COUNTRY').AsString := 'England';
          end;
        end;
        NewDBF.Post;
      end;
      NewDBF.Close;
    finally
      NewDBF.Free;
    end;

  end;

  procedure GetDBFList(Results: TStringList);
  // Gets list of all .dbf files in a directory and its
subdirectories.
  var
    r: TSearchRec;
  begin
    results.Clear;
    if FindFirst('*.dbf', faAnyFile -
{$WARNINGS OFF}
      faVolumeID - faSymLink
{$WARNINGS ON}
      , r) = 0 then
    begin
      repeat
        if (r.Attr and faDirectory) <> faDirectory then
        begin
          results.add(expandfilename(r.Name));
        end;
      until (FindNext(r) <> 0);
      findclose(r);
    end;
  end;

  procedure PrintRecords(DBf: TDBf);
  // Prints contents of available records to screen
  var
    i: integer;
    RecordCount: integer;
  begin
    RecordCount:=0;
    DBF.First;
    while not (Dbf.EOF) do
    begin
      RecordCount := RecordCount + 1;
      writeln('Record ' + IntToStr(RecordCount));
      for i := 0 to DBf.Fields.Count - 1 do
      begin
        if DBF.fields[i].IsNull then
          writeln('Field ', DBf.Fields[i].FieldName, ' is         
***NULL***')
        else
        if DBF.Fields[i].DataType in [ftVarBytes, ftBytes] then
          writeln('Field ', DBF.Fields[i].FieldName, ' is         
***binary***.')
        else
          writeln('Field ', DBf.Fields[i].FieldName, ' has value: ' +
DBf.fields[i].AsString);
      end;
      DBF.Next;
      writeln('');
    end;
  end;

  { TDBFTool }
  procedure TDBFTool.DoRun;
  var
    DBFs: TStringList;
    ErrorMsg: string;
    FileNo: integer;
    MyDbf: TDbf;
  begin
    DBFs := TStringList.Create;
    try
      try
        CreateDemoDBFs('');
      except
        on E: Exception do
        begin
          writeln('*** Error creating demo databases: ' + E.Message);
          Terminate;
          Exit;
        end;
      end;

      // Process all dbfs if no files specified
      if DBFs.Count = 0 then
        GetDBFList(DBFs);

      if DBFs.Count = 0 then
        writeln('Could not find any dbf files');

      for FileNo := 0 to DBFs.Count - 1 do
      begin
        if not (fileexists(DBFs[FileNo])) then
        begin
          // for some reason, fpc trunk suddenly returns the directory
as well...
          //writeln('Sorry, file ',DBFs[FileNo],' does not exist.');
          break;
        end;
        MyDbf := TDbf.Create(nil);
        try
          try
            MyDbf.FilePath := ExtractFilePath(DBFs[FileNo]);
            MyDbf.TableName := ExtractFileName(DBFs[FileNo]);
            MyDbf.ReadOnly := true;
            writeln('*** Opening: ' + DBFs[FileNo]);
            MyDbf.Open;
            writeln('Database tablelevel: ' +
IntToStr(MyDbf.TableLevel));
            writeln('Database codepage:   ' +
IntToStr(MyDBF.CodePage));
            MyDBF.IndexName:='IDXCOMPO';

            PrintRecords(MyDBF);

            MyDbf.Close;
          except
            on E: Exception do
            begin
              writeln('*** Error reading file ', FileNo, ': ',
E.Message);
            end;
          end;
        finally
          MyDbf.Free;
        end;
      end;
    finally
      DBFs.Free;
    end;

    // stop program loop
    Terminate;
  end;

  constructor TDBFTool.Create(TheOwner: TComponent);
  begin
    inherited Create(TheOwner);
    StopOnException := true;
  end;

  destructor TDBFTool.Destroy;
  begin
    inherited Destroy;
  end;

var
  Application: TDBFTool;
begin
  Application := TDBFTool.Create(nil);
  Application.Title := 'DBFTool';
  Application.Run;
  Application.Free;
end.

----------------------------------------------------------------------
micoudic · 2014-01-18 19:04

Quote from: BigChimp on January 16, 2014, 02:12:44 pmPerhaps I have
trouble understanding but this:
Code: [Select]DBf1.AddIndex('KNOMKLE', ('ANOM+' +
IntToStr(Dbf1A_AKLE)),[]);
you are trying to generate a string that contains

    the string 'ANOM+' and
    the string representation of whatever fpc thinks Dbf1a_AKLE is I
have found that TDC accepts indexes multiples only if the fields are
of the same type. That is why I thought this function "IntToStr" ... 
(1)


Quote from: BigChimp on January 16, 2014, 02:12:44 pmI think here you
might be inserting a comma too mach after the 'ANOM' part...
Code: [Select]

AddIndex('KNOMKLE', 'ANOM', + '+A_AKLE', []);The missing comma is due
to a bad transcript (2). She was well in my source, and the message is
good.

Quote from: BigChimp on January 16, 2014, 02:12:44 pmThis code should
demonstrate creating an index exoression built up out of:
1. a string field
2. an integer field
... and should give you an idea of how to adapt your code:
... I'm guessing something like:
Code: [Select]NewDBF.AddIndex('IDXCOMPO','(A_PAYS)+(STR(A_AKLE))',[ixC
aseInsensitive]);

YES, impeccable. I almost give up to work with Lazarus. A big thank
you

(1) I am not a professional, but an autodidact. I know that I
certainly commit errors
(2) I prepared my answers in French, then I translated them with G...
but the syntax is sometimes modified. I then corrects directly in my
reply on the forum, which explains this typo..

----------------------------------------------------------------------
BigChimp · 2014-01-19 11:24

Excellent. Glad it works!

----------------------------------------------------------------------
micoudic · 2014-01-20 18:38

My second key field is numeric. He became alpha in the index. Thus,
the key "15" and larger than the "135" key (framing to left, as the
alpha)

To recap:

-you have an application that uses the file dBase (xxxxxxxx.dbf)
-you have created these files with the Borland Database (dbd32.exe)
module
-you have the index multiples with different type fields(char and num)

This works well with Borland's Delphi, but not with Lazarus

Solution: You keep your .dbf files, but we must re-create the index
programmatically in Lazarus.

----------------------------------------------------------------------
BigChimp · 2014-03-10 08:49

Quote from: BigChimp on January 14, 2014, 05:59:02 pmHmm, I think
there is a bug. Need to verify again but put it on the bugtracker as a
reminder:
http://bugs.freepascal.org/view.php?id=25533
Blush... totally irrationally, I had expected sorting to work
numerically and even partially (i.e. BILL-10 is higher than BILL-11)
while TDBF does a regular alphanumerical sort.
Thanks for helping me, as usual, Lacak2

With current FPC trunk, the test code in the bug report works fine,
just sorts in a way that idealists like me may not want