[Octave-cvsupdate] SF.net SVN: octave:[7931] trunk/octave-forge/extra/dataframe/inst/@ dataframe

[<cde...@us...>]

Revision: 7931
          http://octave.svn.sourceforge.net/octave/?rev=7931&view=rev
Author:   cdemills
Date:     2010-11-17 17:07:37 +0000 (Wed, 17 Nov 2010)

Log Message:
-----------
Try to detect more cases of badly formatted regexpes

Modified Paths:
--------------
    trunk/octave-forge/extra/dataframe/inst/@dataframe/dataframe.m
    trunk/octave-forge/extra/dataframe/inst/@dataframe/private/df_name2idx.m
    trunk/octave-forge/extra/dataframe/inst/@dataframe/subsref.m

Modified: trunk/octave-forge/extra/dataframe/inst/@dataframe/dataframe.m
===================================================================
--- trunk/octave-forge/extra/dataframe/inst/@dataframe/dataframe.m	2010-11-17 15:12:16 UTC (rev 7930)
+++ trunk/octave-forge/extra/dataframe/inst/@dataframe/dataframe.m	2010-11-17 17:07:37 UTC (rev 7931)
@@ -160,7 +160,10 @@
 	unwind_protect_cleanup
 	  fclose(fid);
 	end_unwind_protect
-	eol = '(\r\n|\n|\r|\f|\x85)';
+	%# explicit list taken from 'man pcrepattern' -- we enclose all
+	%# vertical separators in case the underlying regexp engine
+	%# doesn't have them all.
+	eol = '(\r\n|\n|\v|\f|\r|\x85)';
 	%# cut into lines -- include the EOL to have a one-to-one
 	%# matching between line numbers. Use a non-greedy match.
 	lines = regexp(in, ['.*?' eol], 'match');

Modified: trunk/octave-forge/extra/dataframe/inst/@dataframe/private/df_name2idx.m
===================================================================
--- trunk/octave-forge/extra/dataframe/inst/@dataframe/private/df_name2idx.m	2010-11-17 15:12:16 UTC (rev 7930)
+++ trunk/octave-forge/extra/dataframe/inst/@dataframe/private/df_name2idx.m	2010-11-17 17:07:37 UTC (rev 7931)
@@ -61,15 +61,16 @@
     %# translate list of variables to list of indices
     for indi= 1:size(subs, 1),
       %# regexp doesn't like empty patterns
-      if isempty(subs{indi}), continue, endif 
-      %# convert '*' from standard pattern to regexp pattern
-      subs{indi} = regexprep(subs{indi}, '([^\.]|^)\*', "$1.*");
-      %# quote '|[]()?$', otherwise the regexp will stall forever/fail
+      if isempty(subs{indi}), continue, endif
+      %# convert  from standard pattern to regexp pattern
+      subs{indi} = regexprep(subs{indi}, '([^\.\\])(\*|\?)', "$1.$2");
+      %# quote repetition ops at begining of line, otherwise the regexp
+      %# will stall forever/fail
       subs{indi} = regexprep(subs{indi}, \
-			     '([^\\])([\|\[\(\)\]\?\$])', "$1\\$2");
-      %# work at start of line too
-      subs{indi} = regexprep(subs{indi}, \
-			     '^([\|\[\(\)\]\?\$])', "\\$1");
+			     '^([\*\+\?\{\}\|])', "\\$1");
+      %# detect | followed by EOL 
+      subs{indi} = regexprep(subs{indi}, '([^\\])\|$', "$1\\|");
+    
       if 0 == index(subs{indi}, ':'),
 	for indj = 1:min(length(names), count), %# sanity check
 	  if ~isempty(regexp(names{indj}, subs{indi})),

Modified: trunk/octave-forge/extra/dataframe/inst/@dataframe/subsref.m
===================================================================
--- trunk/octave-forge/extra/dataframe/inst/@dataframe/subsref.m	2010-11-17 15:12:16 UTC (rev 7930)
+++ trunk/octave-forge/extra/dataframe/inst/@dataframe/subsref.m	2010-11-17 17:07:37 UTC (rev 7931)
@@ -254,7 +254,7 @@
 	if ~strcmp(output_type, df._type{indc(indi)}),
 	  if dummy, continue; endif
 	  %# unmixable args -- falls back to type of parent container 
-	  error("Selected columns not compatible with cat() -- use 'cell' as output format");
+	  error("Selected columns %s not compatible with cat() -- use 'cell' as output format", mat2str(indc));
 	  %# dead code -- suppress previous line for switching automagically the output format to df
 	  output_type = class(df); 
 	  break;


This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.