Mercurial > hg > octave-lyh

comparison scripts/strings/strchr.m @ 13320:9da8fbd17b92

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
strchr.m: Tune switching between algorithms. Add error messages to input validation. * strchr.m: Switch to mask creation via indexing above 4 CHARS. Tell user what went wrong when input validation fails.
author Rik <octave@nomad.inbox5.com>
date Tue, 11 Oct 2011 13:18:40 -0700
parents 26d3164fd58d
children 72c96de7a403
comparison
equal deleted inserted replaced
13319:eb0ce6ffefb0 13320:9da8fbd17b92
18 18
19 ## -*- texinfo -*- 19 ## -*- texinfo -*-
20 ## @deftypefn {Function File} {@var{idx} =} strchr (@var{str}, @var{chars}) 20 ## @deftypefn {Function File} {@var{idx} =} strchr (@var{str}, @var{chars})
21 ## @deftypefnx {Function File} {@var{idx} =} strchr (@var{str}, @var{chars}, @var{n}) 21 ## @deftypefnx {Function File} {@var{idx} =} strchr (@var{str}, @var{chars}, @var{n})
22 ## @deftypefnx {Function File} {@var{idx} =} strchr (@var{str}, @var{chars}, @var{n}, @var{direction}) 22 ## @deftypefnx {Function File} {@var{idx} =} strchr (@var{str}, @var{chars}, @var{n}, @var{direction})
23 ## @deftypefnx {Function File} {[@var{i}, @var{j}] =} strchr (@dots{})
23 ## Search for the string @var{str} for occurrences of characters from 24 ## Search for the string @var{str} for occurrences of characters from
24 ## the set @var{chars}. The return value, as well as the @var{n} and 25 ## the set @var{chars}. The return value(s), as well as the @var{n} and
25 ## @var{direction} arguments behave identically as in @code{find}. 26 ## @var{direction} arguments behave identically as in @code{find}.
26 ## 27 ##
27 ## This will be faster than using regexp in most cases. 28 ## This will be faster than using regexp in most cases.
28 ## 29 ##
29 ## @seealso{find} 30 ## @seealso{find}
30 ## @end deftypefn 31 ## @end deftypefn
31 32
32 function varargout = strchr (str, chars, varargin) 33 function varargout = strchr (str, chars, varargin)
33 if (nargin < 2 || ! ischar (str) || ! ischar (chars)) 34
35 if (nargin < 2)
34 print_usage (); 36 print_usage ();
37 elseif (! ischar (str))
38 error ("strchr: STR argument must be a string or string array");
39 elseif (! ischar (chars))
40 error ("strchr: CHARS argument must be a string");
35 endif 41 endif
42
36 if (isempty (chars)) 43 if (isempty (chars))
37 mask = false (size (str)); 44 mask = false (size (str));
38 elseif (length (chars) <= 6) 45 elseif (length (chars) <= 4)
39 ## With a few characters, it pays off to build the mask incrementally. 46 ## With a few characters, it pays off to build the mask incrementally.
40 ## We do it via a for loop to save memory. 47 ## We do it via a for loop to save memory.
41 mask = str == chars(1); 48 mask = str == chars(1);
42 for i = 2:length (chars) 49 for i = 2:length (chars)
43 mask |= str == chars(i); 50 mask |= str == chars(i);
44 endfor 51 endfor
45 else 52 else
46 ## Index the str into a mask of valid values. This is slower than 53 ## Index the str into a mask of valid values.
47 ## it could be because of the +1 issue. 54 ## This is slower than it could be because of the +1 issue.
48 f = false (1, 256); 55 f = false (256, 1);
49 f(uint8(chars)+1) = true; 56 f(uint8(chars)+1) = true;
50 ## Default goes via double -- unnecessarily long. 57 ## Default goes via double -- unnecessarily long.
51 si = uint32 (str); 58 si = uint32 (str);
52 ## in-place 59 ## in-place is faster than str+1
53 ++si; 60 ++si;
54 mask = reshape (f(si), size (str)); 61 mask = reshape (f(si), size (str));
55 endif 62 endif
63
56 varargout = cell (1, nargout); 64 varargout = cell (1, nargout);
57 varargout{1} = []; 65 varargout{1} = [];
58 [varargout{:}] = find (mask, varargin{:}); 66 [varargout{:}] = find (mask, varargin{:});
67
59 endfunction 68 endfunction
60 69
61 %!assert(strchr("Octave is the best software",""),zeros(1,0))
62 %!assert(strchr("Octave is the best software","best"),[3, 6, 9, 11, 13, 15, 16, 17, 18, 20, 23, 27])
63 %!assert(strchr("Octave is the best software","software"),[3, 4, 6, 9, 11, 13, 16, 17, 18, 20, 21, 22, 23, 24, 25, 26, 27])
64 70
71 %!assert (strchr ("Octave is the best software", ""), zeros (1,0))
72 %!assert (strchr ("Octave is the best software", "best"), [3, 6, 9, 11, 13, 15, 16, 17, 18, 20, 23, 27])
73 %!assert (strchr ("Octave is the best software", "software"), [3, 4, 6, 9, 11, 13, 16, 17, 18, 20, 21, 22, 23, 24, 25, 26, 27])
74
75 %% Test input validation
76 %!error strchr ()
77 %!error strchr (1)
78 %!error <STR argument must be a string> strchr (1, "aeiou")
79 %!error <CHARS argument must be a string> strchr ("aeiou", 1)
80