textread

[Michael Grossbach]

There seems to be a bug in textread (Octave 3.0.1 binary from
sourceforge, Windows XP) related to the 'headerlines' parameter.
Assume the following in a file test.txt (without the start and end tags):
--start
1stHeaderLine

1    1stDataLine
2    2ndDataLine
--end

This, according to the help text can be read by issuing:
 [a, b] = textread ('E:\test.txt', '%d %s', 'headerlines',2)
a =  1
b =

{
  [1,1] = 1stDataLine
}

Hence, the last line is is not read. Changing the file test.txt to
--start
1stHeaderLine
1    1stDataLine
2    2ndDataLine
--end

, so by omitting the empty second header line, textread behaves as expected:
[a, b] = textread ('E:\test.txt', '%d %s', 'headerlines',1)
a =

   1
   2

b =

{
  [1,1] = 1stDataLine
  [2,1] = 2ndDataLine
}

Cheers, Michael

_______________________________________________
Bug-octave mailing list
Bug-octave@...
https://www.cae.wisc.edu/mailman/listinfo/bug-octave

[David Bateman-3]

Michael Grossbach wrote:
> There seems to be a bug in textread (Octave 3.0.1 binary from
> sourceforge, Windows XP) related to the 'headerlines' parameter.
> Assume the following in a file test.txt (without the start and end tags):

As this is currently an octave-forge function the bug should be sent to
octave-dev@.... That being said it, textread probably
should be a core function, so hey..

The issue is that textread ignores the blank lines in the header in the
TextRead::lines method and then effective does it against when it
subtracts the number of headerlines from nr_rows. Patch attached.

D.

Index: textread.cc
===================================================================
--- textread.cc (revision 5037)
+++ textread.cc (working copy)
@@ -80,7 +80,7 @@
     }
 
     long unsigned int
-    lines()
+    lines(int headerlines)
     {
  if (_lines == 0) {
 
@@ -93,7 +93,7 @@
     char buf[BUFFER_SIZE];
     while (!tmpdata.eof()) {
  tmpdata.getline(buf, BUFFER_SIZE);
- if (std::string(buf).length() != 0) {
+ if (_lines < headerlines || std::string(buf).length() != 0) {
     _lines++;
  }
     }
@@ -252,8 +252,6 @@
  return retval;
     }
 
-
-
     TextFile input(filename.c_str());
     if (!input.is_valid()) {
  return retval;
@@ -268,7 +266,7 @@
     input.set_format(repeated_fmt);    
     input.ignore_whitespace();
 
-    long unsigned int nr_rows = input.lines();
+    long unsigned int nr_rows = input.lines(headerlines);
 
     if (nr_rows == 0) {
         return octave_value (Matrix ());

_______________________________________________
Bug-octave mailing list
Bug-octave@...
https://www.cae.wisc.edu/mailman/listinfo/bug-octave

[John W. Eaton]

On 13-May-2008, David Bateman wrote:

| Michael Grossbach wrote:
| > There seems to be a bug in textread (Octave 3.0.1 binary from
| > sourceforge, Windows XP) related to the 'headerlines' parameter.
| > Assume the following in a file test.txt (without the start and end tags):
|
| As this is currently an octave-forge function the bug should be sent to
| octave-dev@.... That being said it, textread probably
| should be a core function, so hey..

I'd consider it if somone would update it to match the Octave coding
style and submit it as a patch.

Thanks,

jwe
_______________________________________________
Bug-octave mailing list
Bug-octave@...
https://www.cae.wisc.edu/mailman/listinfo/bug-octave