matthiasm@8: %[ANR,SNR,STR]	=  ASORT(INP,'OPT',...);
matthiasm@8: % S		=  ASORT(INP,'OPT',...);
matthiasm@8: %		   to sort alphanumeric strings numerically if
matthiasm@8: %		   they contain one properly formatted number
matthiasm@8: %		   otherwise, ascii dictionary sorting is applied
matthiasm@8: %
matthiasm@8: % INP	unsorted input:
matthiasm@8: %	- a char array
matthiasm@8: %	- a cell array of strings
matthiasm@8: % OPT	options
matthiasm@8: %  -s	- sorting option
matthiasm@8: %	  '-s','ascend'					[def]
matthiasm@8: %	  '-s','descend'
matthiasm@8: %  -st	- force output form S				[def: nargout dependent]
matthiasm@8: %  -t	- replace matching template(s) with one space
matthiasm@8: %	  prior to sorting
matthiasm@8: %	  '-t','template'
matthiasm@8: %	  '-t',{'template1','template2',...}
matthiasm@8: %  -w	- remove space(s) prior to sorting
matthiasm@8: %
matthiasm@8: %	  NOTE	-t/-w options are processed in the
matthiasm@8: %		      order that they appear in
matthiasm@8: %		      the command line
matthiasm@8: %
matthiasm@8: %  -v	- verbose output				[def: quiet]
matthiasm@8: %  -d	- debug mode
matthiasm@8: %	  save additional output in S
matthiasm@8: %	  .c:	lex parser input
matthiasm@8: %	  .t:	lex parser table
matthiasm@8: %	  .n:	lex parser output
matthiasm@8: %	  .d:	numbers read from .n
matthiasm@8: %
matthiasm@8: % ANR	numerically sorted alphanumeric strings		[eg, 'f.-1.5e+2x.x']
matthiasm@8: %	- contain one number that can be read by
matthiasm@8: %	  <strread> | <sscanf>
matthiasm@8: % SNR	ascii dict  sorted alphanumeric strings
matthiasm@8: % http://www.mathworks.com/matlabcentral/fileexchange/loadFile.do?objectId=7212#
matthiasm@8: %
matthiasm@8: %	- contain more than one number			[eg, 'f.-1.5e +2.x']
matthiasm@8: %	- contain incomplete|ambiguous numbers		[eg, 'f.-1.5e+2.x']
matthiasm@8: % STR	ascii dict  sorted strings
matthiasm@8: %	- contain no numbers				[eg, 'a test']
matthiasm@8: %
matthiasm@8: % S	structure with fields
matthiasm@8: %	.anr
matthiasm@8: %	.srn
matthiasm@8: %	.str
matthiasm@8: 
matthiasm@8: % created:
matthiasm@8: %	us	03-Mar-2002
matthiasm@8: % modified:
matthiasm@8: %	us	30-Mar-2005 11:57:07 	/ TMW R14.sp2
matthiasm@8: 
matthiasm@8: %--------------------------------------------------------------------------------
matthiasm@8: function	varargout=asort(inp,varargin)
matthiasm@8: 
matthiasm@8: varargout(1:nargout)={[]};
matthiasm@8: if	~nargin
matthiasm@8: 	help(mfilename);
matthiasm@8: 	return;
matthiasm@8: end
matthiasm@8: 
matthiasm@8: % - common parameters/options
matthiasm@8: n=[];
matthiasm@8: ds=[];
matthiasm@8: anr={};
matthiasm@8: snr={};
matthiasm@8: str={};
matthiasm@8: smod='ascend';	% sorting option
matthiasm@8: tmpl={};	% template(s)
matthiasm@8: sflg=false;	% output  mode: structure
matthiasm@8: tflg=false;	% remove  template(s)
matthiasm@8: dflg=false;	% debug   mode
matthiasm@8: vflg=false;	% verbose output
matthiasm@8: wflg=false;	% remove  spaces
matthiasm@8: 
matthiasm@8: if	nargin > 1
matthiasm@8: 	ix=find(strcmp('-s',varargin));
matthiasm@8: 	if	~isempty(ix) && nargin > ix(end)+1
matthiasm@8: 		smod=varargin{ix(end)+1};
matthiasm@8: 	end
matthiasm@8: 	ix=find(strcmp('-t',varargin));
matthiasm@8: 	if	~isempty(ix) && nargin > ix(end)+1
matthiasm@8: 		tflg=ix(end);
matthiasm@8: 		tmpl=varargin{ix(end)+1};
matthiasm@8: 	end
matthiasm@8: 	if	find(strcmp('-d',varargin));
matthiasm@8: 		dflg=true;
matthiasm@8: 	end
matthiasm@8: 	if	find(strcmp('-st',varargin));
matthiasm@8: 		sflg=true;
matthiasm@8: 	end
matthiasm@8: 	if	find(strcmp('-v',varargin));
matthiasm@8: 		vflg=true;
matthiasm@8: 	end
matthiasm@8: 	ix=find(strcmp('-w',varargin));
matthiasm@8: 	if	~isempty(ix)
matthiasm@8: 		wflg=ix(end);
matthiasm@8: 	end
matthiasm@8: end
matthiasm@8: %   spec numbers
matthiasm@8: ntmpl={
matthiasm@8: 	' inf '
matthiasm@8: 	'+inf '
matthiasm@8: 	'-inf '
matthiasm@8: 	' nan '
matthiasm@8: 	'+nan '
matthiasm@8: 	'-nan '
matthiasm@8: 	};
matthiasm@8: %   spec chars
matthiasm@8: ctmpl={
matthiasm@8: 	'.'	% decimal point
matthiasm@8: 	'd'	% exponent
matthiasm@8: 	'e'	% exponent
matthiasm@8: 	};
matthiasm@8: 
matthiasm@8: if	nargout <= 3
matthiasm@8: 	varargout{1}=inp;
matthiasm@8: else
matthiasm@8: 	disp(sprintf('ASORT> too many output args [%-1d/%-1d]\n',nargout,3));
matthiasm@8: 	help(mfilename);
matthiasm@8: 	return;
matthiasm@8: end
matthiasm@8: if	isempty(inp)
matthiasm@8: 	disp(sprintf('ASORT> input is empty'));
matthiasm@8: 	return;
matthiasm@8: end
matthiasm@8: 
matthiasm@8: ti=clock;
matthiasm@8: winp=whos('inp');
matthiasm@8: switch	winp.class
matthiasm@8: 	case	'cell'
matthiasm@8: 		if	~iscellstr(inp)
matthiasm@8: 			disp(sprintf('ASORT> cell is not an array of strings'));
matthiasm@8: 			return;
matthiasm@8: 		end
matthiasm@8: 		inp=inp(:);
matthiasm@8: 		[ins,inx]=sort(inp);
matthiasm@8: 	case	'char'
matthiasm@8: 		%		[ins,inx]=sortrows(inp);
matthiasm@8: 		inp=cstr(inp);
matthiasm@8: 	otherwise
matthiasm@8: 		disp(sprintf('ASORT> does not sort input of class <%s>',winp.class));
matthiasm@8: 		return;
matthiasm@8: end
matthiasm@8: 
matthiasm@8: inp=inp(:);
matthiasm@8: inp=setinp(inp,tmpl,[tflg wflg]);
matthiasm@8: [ins,inx]=sort(inp);
matthiasm@8: if	strcmp(smod,'descend')
matthiasm@8: 	ins=ins(end:-1:1,:);
matthiasm@8: 	inx=inx(end:-1:1);
matthiasm@8: end
matthiasm@8: ins=inp(inx);
matthiasm@8: c=lower(char(ins));
matthiasm@8: wins=whos('c');
matthiasm@8: [cr,cc]=size(c);
matthiasm@8: 
matthiasm@8: % - LEXICAL PARSER
matthiasm@8: %--------------------------------------------------------------------------------
matthiasm@8: % - extend input on either side for search
matthiasm@8: c=[' '*ones(cr,2) c ' '*ones(cr,2)];
matthiasm@8: 
matthiasm@8: % - search for valid alphanumeric items in strings
matthiasm@8: %   numbers/signs
matthiasm@8: t=(c>='0'&c<='9');
matthiasm@8: t=t|c=='-';
matthiasm@8: t=t|c=='+';
matthiasm@8: [tr,tc]=size(t);
matthiasm@8: %   decimal points
matthiasm@8: %   note: valid numbers with dec points must follow these templates
matthiasm@8: %         nr.nr
matthiasm@8: %	  sign.nr
matthiasm@8: %         nr.<SPACE>
matthiasm@8: %         <SPACE>.nr
matthiasm@8: ix1=	 t(:,1:end-2) & ...
matthiasm@8: 	~isletter(c(:,1:end-2)) & ...
matthiasm@8: 	c(:,2:end-1)=='.';
matthiasm@8: t(:,2:end-1)=t(:,2:end-1)|ix1;
matthiasm@8: ix1=	(t(:,3:end) & ...
matthiasm@8: 	(~isletter(c(:,3:end)) & ...
matthiasm@8: 	~isletter(c(:,1:end-2))) | ...
matthiasm@8: 	(c(:,3:end)=='e' | ...
matthiasm@8: 	c(:,3:end)=='d')) & ...
matthiasm@8: 	c(:,2:end-1)=='.';
matthiasm@8: t(:,2:end-1)=t(:,2:end-1)|ix1;
matthiasm@8: %		t(:,3:end)=t(:,3:end)|ix1;
matthiasm@8: %   signs
matthiasm@8: t(c=='-')=false;
matthiasm@8: t(c=='+')=false;
matthiasm@8: ix1=	 t(:,3:end) & ...
matthiasm@8: 	(c(:,2:end-1)=='-' | ...
matthiasm@8: 	c(:,2:end-1)=='+');
matthiasm@8: t(:,2:end-1)=t(:,2:end-1)|ix1;
matthiasm@8: %   exponents
matthiasm@8: ix1=	 t(:,1:end-2) & ...
matthiasm@8: 	(c(:,2:end-1)=='e' | ...
matthiasm@8: 	c(:,2:end-1)=='d');
matthiasm@8: t(:,2:end-1)=t(:,2:end-1)|ix1;
matthiasm@8: %   spec numbers
matthiasm@8: c=reshape(c.',1,[]);
matthiasm@8: t=t';
matthiasm@8: ic=[];
matthiasm@8: for	j=1:numel(ntmpl)
matthiasm@8: 	ic=[ic,strfind(c,ntmpl{j})];
matthiasm@8: end
matthiasm@8: ic=sort(ic);
matthiasm@8: for	i=1:numel(ic)
matthiasm@8: 	ix=ic(i)+0:ic(i)+4;
matthiasm@8: 	t(ix)=true;
matthiasm@8: end
matthiasm@8: t=t';
matthiasm@8: c=reshape(c.',[tc,tr]).';
matthiasm@8: t(c==' ')=false;
matthiasm@8: %--------------------------------------------------------------------------------
matthiasm@8: 
matthiasm@8: % - only allow one number per string
matthiasm@8: il=~any(t,2);
matthiasm@8: ib=strfind(reshape(t.',1,[]),[0 1]);
matthiasm@8: if	~isempty(ib)
matthiasm@8: 	ixe=cell(3,1);
matthiasm@8: 	n=reshape(char(t.*c).',1,[]);
matthiasm@8: 	for	i=1:numel(ctmpl)
matthiasm@8: 		id=strfind(n,ctmpl{i});
matthiasm@8: 		if	~isempty(id)
matthiasm@8: 			[dum,dum,ixu{i},ixe{i}]=dupinx(id,tc);
matthiasm@8: 		end
matthiasm@8: 	end
matthiasm@8: 	in=false(tr,1);
matthiasm@8: 	im=in;
matthiasm@8: 	%   must check for anomalous cases like <'.d'>
matthiasm@8: 	id=sort(...
matthiasm@8: 		[find(n>='0' & n<='9'),...
matthiasm@8: 		strfind(n,'inf'),...
matthiasm@8: 		strfind(n,'nan')]);
matthiasm@8: 	%		[ibu,ibd,ixbu,ixe{i+1}]=dupinx(id,tc);
matthiasm@8: 	[ibu,ibd,ixbu,ixbd]=dupinx(id,tc);
matthiasm@8: 	in(ixbu)=true;
matthiasm@8: 	in(ixbd)=true;
matthiasm@8: 	[ibu,ibd,ixbu,ixbd]=dupinx(ib,tc);
matthiasm@8: 	im(ixbu)=true;
matthiasm@8: 	in=in&im;
matthiasm@8: 	in([ixe{:}])=false;
matthiasm@8: 	il=~any(t,2);
matthiasm@8: 	ia=~(in|il);
matthiasm@8: 
matthiasm@8: 	% - read valid strings
matthiasm@8: 	n=t(in,:).*c(in,:);
matthiasm@8: 	n(n==0)=' ';
matthiasm@8: 	n=char(n);
matthiasm@8: 	dn=strread(n.','%n');
matthiasm@8: 	if	numel(dn) ~= numel(find(in))
matthiasm@8: 		%disp(sprintf('ASORT> unexpected fatal error reading input!'));
matthiasm@8: 		if	nargout
matthiasm@8: 			s.c=c;
matthiasm@8: 			s.t=t;
matthiasm@8: 			s.n=n;
matthiasm@8: 			s.d=dn;
matthiasm@8: 			varargout{1}=s;
matthiasm@8: 		end
matthiasm@8: 		return;
matthiasm@8: 	end
matthiasm@8: 
matthiasm@8: 	% - sort numbers
matthiasm@8: 	[ds,dx]=sort(dn,1,smod);
matthiasm@8: 	in=find(in);
matthiasm@8: 	anr=ins(in(dx));
matthiasm@8: 	snr=ins(ia);
matthiasm@8: end
matthiasm@8: str=ins(il);
matthiasm@8: to=clock;
matthiasm@8: 
matthiasm@8: % - prepare output
matthiasm@8: if	nargout < 3 || sflg
matthiasm@8: 	s.magic='ASORT';
matthiasm@8: 	s.ver='30-Mar-2005 11:57:07';
matthiasm@8: 	s.time=datestr(clock);
matthiasm@8: 	s.runtime=etime(to,ti);
matthiasm@8: 	s.input_class=winp.class;
matthiasm@8: 	s.input_msize=winp.size;
matthiasm@8: 	s.input_bytes=winp.bytes;
matthiasm@8: 	s.strng_class=wins.class;
matthiasm@8: 	s.strng_msize=wins.size;
matthiasm@8: 	s.strng_bytes=wins.bytes;
matthiasm@8: 	s.anr=anr;
matthiasm@8: 	s.snr=snr;
matthiasm@8: 	s.str=str;
matthiasm@8: 	if	dflg
matthiasm@8: 		s.c=c;
matthiasm@8: 		s.t=t;
matthiasm@8: 		s.n=n;
matthiasm@8: 		s.d=ds;
matthiasm@8: 	end
matthiasm@8: 	varargout{1}=s;
matthiasm@8: else
matthiasm@8: 	s={anr,snr,str};
matthiasm@8: 	for	i=1:nargout
matthiasm@8: 		varargout{i}=s{i};
matthiasm@8: 	end
matthiasm@8: end
matthiasm@8: 
matthiasm@8: if	vflg
matthiasm@8: 	inp=cstr(inp);
matthiasm@8: 	an=[{'--- NUMERICAL'};		anr];
matthiasm@8: 	as=[{'--- ASCII NUMBERS'};	snr];
matthiasm@8: 	at=[{'--- ASCII STRINGS'};	str];
matthiasm@8: 	nn=[{'--- NUMBERS'};		num2cell(ds)];
matthiasm@8: 	ag={' ';' ';' '};
matthiasm@8: 	u=[{'INPUT'};			inp;ag];
matthiasm@8: 	v=[{'ASCII SORT'};		ins;ag];
matthiasm@8: 	w=[{'NUM SORT'};		an;as;at];
matthiasm@8: 	x=[{'NUM READ'};		nn;as;at];
matthiasm@8: 	w=[u,v,w,x];
matthiasm@8: 	disp(w);
matthiasm@8: end
matthiasm@8: 
matthiasm@8: return;
matthiasm@8: %--------------------------------------------------------------------------------
matthiasm@8: function	c=cstr(s)
matthiasm@8: % - bottleneck waiting for a good <cellstr> replacement
matthiasm@8: %   it consumes ~75% of <asort>'s processing time!
matthiasm@8: 
matthiasm@8: c=s;
matthiasm@8: if	ischar(s)
matthiasm@8: 	sr=size(s,1);
matthiasm@8: 	c=cell(sr,1);
matthiasm@8: 	for	i=1:sr
matthiasm@8: 		c{i}=s(i,:);	% no deblanking!
matthiasm@8: 	end
matthiasm@8: end
matthiasm@8: return;
matthiasm@8: %--------------------------------------------------------------------------------
matthiasm@8: function	[idu,idd,ixu,ixd]=dupinx(ix,nc)
matthiasm@8: % - check for more than one entry/row in a matrix of column size <nc>
matthiasm@8: %   unique    indices:	idu / ixu
matthiasm@8: %   duplicate indices:	idd / ixd
matthiasm@8: 
matthiasm@8: if	isempty(ix)
matthiasm@8: 	idu=[];
matthiasm@8: 	idd=[];
matthiasm@8: 	ixu=[];
matthiasm@8: 	ixd=[];
matthiasm@8: 	return;
matthiasm@8: end
matthiasm@8: id=fix(ix/nc)+1;
matthiasm@8: idi=diff(id)~=0;
matthiasm@8: ide=[true idi];
matthiasm@8: idb=[idi true];
matthiasm@8: idu=idb & ide;
matthiasm@8: idd=idb==1 & ide==0;
matthiasm@8: ixu=id(idu);
matthiasm@8: ixd=id(idd);
matthiasm@8: return;
matthiasm@8: %--------------------------------------------------------------------------------
matthiasm@8: function	inp=setinp(inp,tmpl,flg)
matthiasm@8: % - remove space(s) and/or templates
matthiasm@8: 
matthiasm@8: if	isempty(inp) || ~any(flg)
matthiasm@8: 	return;
matthiasm@8: end
matthiasm@8: 
matthiasm@8: for	i=sort(flg)
matthiasm@8: 	switch	i
matthiasm@8: 		case	flg(1)
matthiasm@8: 			if	ischar(tmpl)
matthiasm@8: 				tmpl={tmpl};
matthiasm@8: 			end
matthiasm@8: 			for	i=1:numel(tmpl)
matthiasm@8: 				inp=strrep(inp,tmpl{i},' ');
matthiasm@8: 			end
matthiasm@8: 		case	flg(2)
matthiasm@8: 			inp=strrep(inp,' ','');
matthiasm@8: 	end
matthiasm@8: end
matthiasm@8: return;
matthiasm@8: %--------------------------------------------------------------------------------