is there Ada package similar to Matlab textscan?

is there Ada package similar to Matlab textscan?

[Nasser M. Abbasi]

Matlab textscan() build-in function is very useful, I use it
all the time to read textual data from text files.

One bascially tells it the format of each field, and will load
all the data to Matlab data structure (called cells). Then
one can simply loop over the data and process it. The
format used is that of printf() standard format.

It is described here

http://www.mathworks.com/help/matlab/ref/textscan.html

One can also tell it how many lines to skip before
starting to read (this is in order to  bypass any
file headers) and tell it what is the field delimiter
between the columns of the data.

For example, given this text file

----- foo.txt---------------------
Student_ID  | Test1  | Test2  | Test3
    1,           91.5,     89.2,     77.3
    2,           88.0,     67.8,     91.0
    3,           76.3,     78.1,     92.5
--------------------------------

It is read as

C = textscan(fileID,'%d %f %f %f','HeaderLines',1,'Delimiter',',')

That is all. Now C will contain all the data in, in
what is called cell columns. C{1}, C{2}, C{3}, C{4}.

Is there a similar way to read data using Ada or must
one parse each field one by one themselves after
reading the file line by line?  I searched for an
Ada package that does something similar to textscan, but
so far did not find one.

thanks,
--Nasser

re: is there Ada package similar to Matlab textscan?

[tmoran]

That's a CSV file.  I'm sure there are lots of folks' implementations.
The important part of mine is:

  function Field_Name(Csv_File: Csv_Files;
                      Index   : Field_Counts) return String;
  -- raises Constraint_Error if Index not in 1 .. Header's Field_Count

  function Field_Index(Csv_File:   Csv_Files;
                       Name: String) return Field_Counts;
  -- Ignoring case, find first field whose prefix = Name
  -- returns zero if not found

  procedure Read_A_Record(Csv_File: in out Csv_Files);
  -- May End_Error.
  -- If a record has fewer fields than the header indicated, it's
  -- padded out with null strings.

  function Field_Content(Csv_File: Csv_Files;
                         Index   : Field_Counts) return String;
  -- Raises Use_Error if called before any Read_A_Record.

Re: is there Ada package similar to Matlab textscan?

[Dmitry A. Kazakov]

John Woodruff's Numeric_IO packages:

   http://www.dmitry-kazakov.de/ada/Numeric-Name-IO.htm

Download link:

   http://www.dmitry-kazakov.de/ada/Name_IO_Distribution-Nov10.zip

-- 
Regards,
Dmitry A. Kazakov
http://www.dmitry-kazakov.de

Re: is there Ada package similar to Matlab textscan?

[Björn Lundin]

On 2015-07-17 05:52, Nasser M. Abbasi wrote:


> one parse each field one by one themselves after
> reading the file line by line? 


Gnat.Awk is line based but parsing is simple:


with GNAT; use GNAT;
with GNAT.AWK;
with Text_io;
procedure Test is

  Computer_File : AWK.Session_Type;

begin
  AWK.Set_Current (Computer_File);
  AWK.Open (Separators => ",",
            Filename   => "path/to/file.dat");

  while not AWK.End_Of_File loop
    AWK.Get_Line;
    if AWK.Field(4)  = "GB" and then
       AWK.Field (18) = "PE"  then

       Text_io.Put_Line(AWK.Field (5) & " | " &
                        AWK.Field (8) & " | " &
                        AWK.Field (2) & " | " &
                        AWK.Field (11));
    end if;
  end loop;
  AWK.Close (Computer_File);

  see
<https://www2.adacore.com/gap-static/GNAT_Book/html/rts/g-awk__ads.htm>

for more examples

-- 
--
Björn