www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.learn - Fuzzy string matching?

reply Andrej Mitrovic <andrej.mitrovich gmail.com> writes:
Is there any such method in Phobos?

I have to rename some files based on a string array of known names
which need to be fuzzy-matched to file names and then rename the files
to the matches.

E.g.:

string[] strings = ["food", "lamborghini", "architecture"]

files on system:
.\foo.ext
.\lmbrghinione.ext
.\archtwo.ext

and if there's a fuzzy match then the matched files would be renamed to:
.\food.ext
.\lamborghini.ext
.\architecture.ext

Perhaps there's a C library I can use for this?
Jul 15 2011
next sibling parent reply dsmith <ds nomail.com> writes:
Until recently, you could easily use std.regexp.search(target_string,
find_string), but regexp is apparently no longer in phobos.  I seek a simple
substitute.  std.algorithm.canFind might work, as
it is bool.

Maybe try something like:

foreach(str; strings)
    foreach(fls; system_files)
        if(std.algorithm.canFind(fls, str))          // usage needs verification
            str ~= ".ext";


== Repost the article of Jonathan M Davis (jmdavisProg gmx.com)
== Posted at 2011/07/15 22:03 to digitalmars.D.learn

On Saturday 16 July 2011 01:17:36 Andrej Mitrovic wrote:

  parent  Jonathan M Davis <jmdavisProg gmx.com>  writes:


On Saturday 16 July 2011 05:07:38 dsmith wrote:

 Until recently, you could easily use std.regexp.search(target_string,
 find_string), but regexp is apparently no longer in phobos.  I seek a
 simple substitute.  std.algorithm.canFind might work, as it is bool.
 
 Maybe try something like:
 
 foreach(str; strings)
     foreach(fls; system_files)
         if(std.algorithm.canFind(fls, str))          // usage needs
 verification str ~= ".ext";


std.regex is std.regexp's replacement.

- Jonathan M davis



 Jul 15 2011



prev sibling next sibling parent reply dsmith <ds nomail.com>  writes:


Could you demonstrate how to use std.regex for pattern matching, preferably
with a bool method?

My usage of std.regex.match yields this error:
core.exception.AssertError /usr/include/d/dmd/phobos/std/regex.d(1796):
4294967295 .. 4294967295 vs. 5

My usage is:     auto m = match(long_string, regex(str));
                        writeln(m.hit);


== Repost the article of Jonathan M Davis (jmdavisProg gmx.com)
== Posted at 2011/07/16 01:08 to digitalmars.D.learn


 next sibling parent  Jonathan M Davis <jmdavisProg gmx.com>  writes:


On Saturday 16 July 2011 06:17:56 dsmith wrote:

 Could you demonstrate how to use std.regex for pattern matching, preferably
 with a bool method?
 
 My usage of std.regex.match yields this error:
 core.exception.AssertError /usr/include/d/dmd/phobos/std/regex.d(1796):
 4294967295 .. 4294967295 vs. 5
 
 My usage is:     auto m = match(long_string, regex(str));
                         writeln(m.hit);


I'm afraid that I've never used either std.regexp or std.regex, so I'm not 
familiar with the usage of either one. There's every chance that this is a bug 
rather than misuse on your part. I'd advise posting a question about it 
separately (so that people are more likely to see it) with an appropriate 
subject, and there's a decent chance that someone who's actually familiar with 
std.regex will answer your question.

- Jonathan M Davis



 Jul 15 2011



prev sibling  parent  Dmitry Olshansky <dmitry.olsh gmail.com>  writes:


On 16.07.2011 10:17, dsmith wrote:

 Could you demonstrate how to use std.regex for pattern matching, preferably
with a bool method?

 My usage of std.regex.match yields this error:
core.exception.AssertError /usr/include/d/dmd/phobos/std/regex.d(1796):
4294967295 .. 4294967295 vs. 5

 My usage is:     auto m = match(long_string, regex(str));
                          writeln(m.hit);


Your and apparently an awful lot of people hit this, the thing is that 
.hit method is returning _matched slice_ of string if there is a match 
and asserts otherwise. (there is also issue of this assert having 
message is of a _very_ poor quality)
As it stands now regex works like ranges: you need to check if it was 
empty then use it, so if all you want to do is a test:

auto m = match(long_string, regex(str));
writeln(!m.empty); // substitute for "there was match"


Thinking more about this, it should be in synopsis part of std.regex in 
docs on d-p-l.org. Along with something like:

foreach(m; match("abc", regex("\w", "g")) //uses range syntax to iterate 
over all matches (so empty is checked)
     writeln(m.hit); // here m.hit is guaranteed to hold something (and 
not asserting)


 == Repost the article of Jonathan M Davis (jmdavisProg gmx.com)
 == Posted at 2011/07/16 01:08 to digitalmars.D.learn

 On�Saturday�16�July�2011�05:07:38�dsmith�wrote:

 �Until�recently,�you�could�easily�use�std.regexp.search(target_string,
 �find_string),�but�regexp�is�apparently�no�longer�in�phobos.��I�seek�a
 �simple�substitute.��std.algorithm.canFind�might�work,�as�it�is�bool.
 �Maybe�try�something�like:

 �foreach(str;�strings)
 �����foreach(fls;�system_files)
 ���������if(std.algorithm.canFind(fls,�str))���//�usage�needs�verification
              str�~=�".ext";


 std.regex�is�std.regexp's�replacement.

 -�Jonathan�M�davis



-- 
Dmitry Olshansky



 Jul 16 2011



prev sibling  parent  Mike Wey <mike-wey example.com>  writes:


On 07/16/2011 01:17 AM, Andrej Mitrovic wrote:

 Is there any such method in Phobos?

 I have to rename some files based on a string array of known names
 which need to be fuzzy-matched to file names and then rename the files
 to the matches.

 E.g.:

 string[] strings = ["food", "lamborghini", "architecture"]

 files on system:
 .\foo.ext
 .\lmbrghinione.ext
 .\archtwo.ext

 and if there's a fuzzy match then the matched files would be renamed to:
 .\food.ext
 .\lamborghini.ext
 .\architecture.ext

 Perhaps there's a C library I can use for this?


You could try stc.algorithm.levenshteinDistance to check how much the 
two names differ.

-- 
Mike Wey



 Jul 16 2011