Re: [CHANGESET] updated print.m

[Benjamin Lindner]

Ben Abbott wrote:
> On May 22, 2009, at 9:43 AM, Benjamin Lindner wrote:
>
> > > I've pushed this changeset. If anyone has trouble printing let me 
> > > know (particularly on Windows).
> >
> > I did a check using print.m on mingw32 (not printing to hardware) 
> > using this simple test sequence:
> >
> > plot(0:0.1:10, sin(0:0.1:10), "@-;sin;", 0:0.1:10, cos(0:0.1:10), 
> > "@-;cos;");
> > print -depsc2 test.eps
> > print -dpsc2  test.ps
> > print -dpng   test.png
> > print -demf   test.emf
> > print -dpdf   test.pdf
> >
> > and all work except the pdf output.
> >
> > I suggest the attached two changesets for the windows users:
> > 1) ensure that gnuplot_binary is quoted when used, because it may 
> > contain spaces (this sadly happens quite frequently under windows).
> >
> > 2.a) in print.m, for the same reason, ensure that ghostscript_binary 
> > is always quoted when used.
> > 2.b) use double quotes, since cmd.exe does not seem to understand 
> > single quotes.
> > 2.c) the windows binary of ghostscript uses the environment variable 
> > GSC as location for the console ghostscript binary. I added code in 
> > print.m to test for %GSC% and use it, and default to "gswin32c" 
> > otherwise.
> > 2.d) single-quote the output file name in gnuplot_drawnow.m to allow 
> > for backslash file separators. Otherwise 
> > print("-demf","t:\\temp\\test.emf") will not work, while 
> > print("-demf","t:/temp/test.emf") works.
> >
> > With this I get working output for all 5 file formats.
> >
> > Somethig else I realized: gnuplot's pdfcairo terminal is not 
> > recognized, so even while gnuplot has a direct pdf output terminal, 
> > print.m uses the indirect way ps->pdf via ghostscript.
>
> Ben,
>
> I tried to import your changesets using "hg import ..." and "patch -p1 < 
> your.patch", but they do not apply for me.
>
> I assume there are some EOL problems ... which might  be introduced by 
> my email client.
>
> Can you resend them as attachements, and if you create them using "hg 
> import ..." please include the hg header as well. Also John has 
> committed a patch since you last pulled. That needs to be corrected as 
> well.

Sorry, I mixed up changesets with mq patches.

I updated my 3.1.x repos and corrected my changes.
I have now attached the exported changesets.

benjamin

# HG changeset patch
# User Benjamin Lindner <address@hidden>
# Date 1243152559 -7200
# Node ID 692621f2fd099bca03e5cbf574085730b91b7408
# Parent  ec08d8d25076ff80241dd66ea0aac5edf17eed4d
ghostscript handling for windows in print.m

diff -r ec08d8d25076 -r 692621f2fd09 scripts/ChangeLog
--- a/scripts/ChangeLog Sun May 24 10:08:35 2009 +0200
+++ b/scripts/ChangeLog Sun May 24 10:09:19 2009 +0200
@@ -1,3 +1,10 @@
+2009-05-24 Benjamin Lindner <address@hidden>
+
+       * plot/gnuplot_drawnow.m: single-quote output name to allow backslash
+       characters as filesep under windows
+       * plot/print.m: Support ps->pdf using ghostscript under windows, check
+       for %GSC% environment variable.
+
 2009-05-24 Benjamin Lindner <address@hidden>
 
        * plot/__gnuplot_version__.m: quote gnuplot_binary to allow spaces
diff -r ec08d8d25076 -r 692621f2fd09 scripts/plot/gnuplot_drawnow.m
--- a/scripts/plot/gnuplot_drawnow.m    Sun May 24 10:08:35 2009 +0200
+++ b/scripts/plot/gnuplot_drawnow.m    Sun May 24 10:09:19 2009 +0200
@@ -278,7 +278,7 @@
 
   if (nargin == 5)
     if (! isempty (file))
-      fprintf (plot_stream, "set output \"%s\";\n", file);
+      fprintf (plot_stream, "set output '%s';\n", file);
     endif
   endif
 
diff -r ec08d8d25076 -r 692621f2fd09 scripts/plot/print.m
--- a/scripts/plot/print.m      Sun May 24 10:08:35 2009 +0200
+++ b/scripts/plot/print.m      Sun May 24 10:09:19 2009 +0200
@@ -186,7 +186,11 @@
   if (isunix ())
     persistent ghostscript_binary = "gs";
   elseif (ispc ())
-    persistent ghostscript_binary = "gswin32c";
+    if (~isempty (getenv ("GSC")))
+      persistent ghostscript_binary = getenv ("GSC");
+    else
+      persistent ghostscript_binary = "gswin32c";
+    endif
   endif
 
   old_fig = get (0, "currentfigure");
@@ -253,7 +257,8 @@
       [status, output] = system (sprintf ("which %s 2>&1", 
ghostscript_binary));
       have_ghostscript = (status == 0);
     elseif (ispc ())
-      have_ghostscript = true;
+      [status, output] = system (sprintf ("if exist \"%s\" ( exit /B 1 ) else 
( exit /B 0 )", ghostscript_binary));
+      have_ghostscript = (status ~= 0);
     endif
 
     doprint = isempty (name);
@@ -571,12 +576,12 @@
       endif
       ghostscript_options = sprintf ("%s -sDEVICE=%s", ghostscript_options,
                                      ghostscript_device);
-      command = sprintf ("%s %s -sOutputFile='%s' '%s' 2>&1", 
ghostscript_binary,
+      command = sprintf ("\"%s\" %s -sOutputFile=\"%s\" \"%s\" 2>&1", 
ghostscript_binary,
                           ghostscript_options, ghostscript_output, name);
       [errcode, output] = system (command);
       unlink (name);
       if (errcode)
-        error ("print: Conversion failed, %s -> %s.", name, 
ghostscript_output);
+        error ("print: Conversion failed, %s -> %s.\nError was:\n%s\n", name, 
ghostscript_output, output);
       endif
     elseif (is_eps_file && tight_flag && ! doprint)
       ## If the saved output file is an eps file, use ghostscript to set a 
tight bbox.
@@ -625,7 +630,7 @@
   box_string = "%%BoundingBox:";
 
   ghostscript_options = "-q -dBATCH -dSAFER -dNOPAUSE -dTextAlphaBits=4 
-sDEVICE=bbox";
-  cmd = sprintf ("%s %s '%s' 2>&1", ghostscript_binary, ghostscript_options, 
eps_file_name);
+  cmd = sprintf ("\"%s\" %s \"%s\" 2>&1", ghostscript_binary, 
ghostscript_options, eps_file_name);
   [status, output] = system (cmd);
 
   if (status == 0)
@@ -672,7 +677,7 @@
     end_unwind_protect
   elseif (warn_on_no_ghostscript)
     warn_on_no_ghostscript = false;
-    warning ("print.m: Ghostscript could not be used to adjust bounding box.")
+    warning ("print.m: Ghostscript could not be used to adjust bounding 
box.\nError was:\n%s\n", output)
   endif
 
 endfunction

# HG changeset patch
# User Benjamin Lindner <address@hidden>
# Date 1243152330 -7200
# Node ID fbbef32dc1eab043c56840b4874551fe5e455933
# Parent  7b94eb96bec7b6d19e84e14c2e506d364fa54d18
ensure quoted gnuplot binary

diff -r 7b94eb96bec7 -r fbbef32dc1ea scripts/ChangeLog
--- a/scripts/ChangeLog Sun May 24 10:04:44 2009 +0200
+++ b/scripts/ChangeLog Sun May 24 10:05:30 2009 +0200
@@ -1,3 +1,8 @@
+2009-05-24 Benjamin Lindner <address@hidden>
+
+       * plot/__gnuplot_version__.m: quote gnuplot_binary to allow spaces
+       in file name
+
 2009-05-22  John W. Eaton  <address@hidden>
 
        * general/interp3.m: Don't require interpolation grid to have same
diff -r 7b94eb96bec7 -r fbbef32dc1ea scripts/plot/__gnuplot_version__.m
--- a/scripts/plot/__gnuplot_version__.m        Sun May 24 10:04:44 2009 +0200
+++ b/scripts/plot/__gnuplot_version__.m        Sun May 24 10:05:30 2009 +0200
@@ -30,7 +30,7 @@
   persistent __version__ = "";
 
   if (isempty (__version__))
-    [status, output] = system (sprintf ("%s --version", gnuplot_binary ()));
+    [status, output] = system (sprintf ("\"%s\" --version", gnuplot_binary 
()));
     if (status != 0)
       ## This message ends in a newline so that the traceback messages
       ## are skipped and people might actually see the message, read it,