path matching problem

[Charles Hixson]

On 11/27/2012 01:31 PM, Joshua Niehus wrote:
> On Tuesday, 27 November 2012 at 19:40:56 UTC, Charles Hixson wrote:
>> Is there a better way to do this? (I want to find files that match any
>> of some extensions and don't match any of several other strings, or
>> are not in some directories.):
>>
>> import std.file;
>>
>> ...
>>
>> string exts = "*.{txt,utf8,utf-8,TXT,UTF8,UTF-8}";
>> string[] exclude = ["/template/", "biblio.txt", "categories.txt",
>> "subjects.txt", "/toCDROM/"]
>>
>> int limit = 1
>> // Iterate a directory in depth
>> foreach (string name; dirEntries(sDir, exts, SpanMode.depth))
>> { bool excl = false;
>> foreach (string part; exclude)
>> { if (part in name)
>> { excl = true;
>> break;
>> }
>> }
>> if (excl) break;
>> etc.
>
> maybe this:?
>
> import std.algorithm, std.array, std.regex;
> import std.stdio, std.file;
> void main()
> {
> enum string[] exts = [`".txt"`, `".utf8"`, `".utf-8"`, `".TXT"`,
> `".UTF8"`, `".UTF-8"`];
> enum string exclude =
> `r"/template/|biblio\.txt|categories\.txt|subjects\.txt|/toCDROM/"`;
>
> auto x = dirEntries("/path", SpanMode.depth)
> .filter!(`endsWith(a.name,` ~ exts.join(",") ~ `)`)
> .filter!(`std.regex.match(a.name,` ~ exclude ~ `).empty`);;
>
> writeln(x);
> }

That's a good approach, except that I want to step through the matching 
paths rather than accumulate them in a vector...though ... the filter 
documentation could mean that it would return an iterator.  So I could 
replace
writeln (x);
by
foreach (string name; x)
{
	...
}
and x wouldn't have to hold all the matching strings at the same time.

But why the chained filters, rather than using the option provided by 
dirEntries for one of them?  Is it faster?  Just the way you usually do 
things? (Which I accept as a legitimate answer.  I can see that that 
approach would be more flexible.)