[scilab-Users] weird problem with fscanfMat

[Antoine Monmayrant]

Le 09/02/2011 16:12, Stéphane Mottelet a écrit :
> Le 09/02/2011 11:41, Antoine Monmayrant a écrit :
>> Le 09/02/2011 10:29, Stéphane Mottelet a écrit :
>>> Le 09/02/2011 10:07, Antoine Monmayrant a écrit :
>>>> Le 09/02/2011 09:26, Stéphane Mottelet a écrit :
>>>>> Le 09/02/2011 09:24, Antoine Monmayrant a écrit :
>>>>>> Le 08/02/2011 22:04, Stephane Mottelet a écrit :
>>>>>>> Hi all,
>>>>>>>
>>>>>>> I cannot manage to read the attached file with
>>>>>>> fscanfMat, although it has a compatible structure,
>>>>>>> i.e. one line one text then many lines of the same
>>>>>>> number of columns.
>>>>>>>
>>>>>>> I have even tried to read the first line with mgetl,
>>>>>>> but then eof is reached although I only get the first line
>>>>>>> (composed of text).
>>>>>>>
>>>>>>> I know each line is only finished with a cr and not
>>>>>>> an eol. Is it a problem ?
>>>>>>> Thanks for help of anybody willing to solve this
>>>>>>> weird issue !!!
>>>>>>>
>>>>>>> S.
>>>>>> When I open the file you attached to your email, I don't see any 
>>>>>> separator to mark the end of line (no cr, no \n, no nothing).
>>>>>> It might have been mangled by my mail program but I doubt it.
>>>>>> I managed to open it in scilab with read_csv and it also looks 
>>>>>> like there is no separator at the end of line: the last item of a 
>>>>>> line is touching the first item on next line:
>>>>>>
>>>>>> -->a_string=read_csv("20101122_VOLUME.txt","\t");
>>>>>>
>>>>>> -->a_string(16)
>>>>>> ans  =
>>>>>>
>>>>>>  ch15 (L)0.000
>>>>>>
>>>>>> -->a_string(31)
>>>>>>  ans  =
>>>>>>
>>>>>>  0.0000.001
>>>>>>
>>>>>> -->a_string(46)
>>>>>>  ans  =
>>>>>>
>>>>>>  0.0000.002
>>>>>>
>>>>>>
>>>>>> Hope it helps,
>>>>>>
>>>>>> Antoine
>>>>>>
>>>>> Well, when I open it with vi, I see some ^M at the end of each line.
>>>>>
>>>>> S.
>>>> I don't, so it has been mangled in the process.
>>>> With :set list, I can see the column separator (^I), but nothing at 
>>>> the end of line, the two adjacent items being merged...
>>>>
>>>> Antoine
>>>> PS: I join your file back, can you check if it has been modified or 
>>>> not?
>>> The ^M are missing, you are right. I join the original
>>> file (still unreadable with fscanfMat), but zipped in order
>>> to preserve it.
>>>
>>> S.
>> Got it!
>> There is a problem with your file format.
>> It might be a bug in fscanfMat, but I don't think as it is only 
>> supposed to open files created with fprintfMat.
>> As such, it does not handle the slightest deviation from the expect 
>> format.
>
> well, are files with a ^M instead of ^L  non-standard ?
>
> S.
I won't say that.
What I know is that fscanfMat is only supposed to open files made using 
fprintfMat.
I learned it the hard way as I was using it to load other kind of files 
and things changed at some point (5.2 or 5.3).
My point is that it is not really a bug if fscanfMat cannot open a file 
that it is not designed to open...
I think I ended up using read_csv to avoid this problem.

Antoinz
>
>> Here is a way to correct things under linux:
>>
>> cat 20101122_VOLUME.txt | tr '\r' '\n' > corr_20101122_VOLUME.txt
>>
>> then in scilab you can load it:
>>
>> -->dat=fscanfMat("corr_20101122_VOLUME.txt");
>>
>> -->size(dat)
>>  ans  =
>>
>>     291.    16.
>>
>> Antoine
>


-- 
+++++++++++++++++++++++++++++++++++++++++++++++++++++++

  Antoine Monmayrant LAAS - CNRS
  7 avenue du Colonel Roche
  31077 TOULOUSE
  Cedex 4 FRANCE

  Tel:+33 5 61 33 64 59

  email : antoine.monmayrant at laas.fr
  permanent email : antoine.monmayrant at polytechnique.org

+++++++++++++++++++++++++++++++++++++++++++++++++++++++