A simple fully tested Glob to Regex conversion library for .NET with no external dependencies targeting netstadard2.0
Glob is a pattern that defines a set of file system entries using wildcards. GlobRegex allows you to convert a glob into a regular expression pattern splitting the glob into a base directory from which a recursive traversal can be started, and a stem that can be used to match a relative path.
var glob = GlobConvert.ToRegexPattern("../usr/doc?/**/*.txt", GlobRegexOptions.MatchFullString);
var regex = new Regex(glob.StemRegexPattern, RegexOptions.Compiled | RegexOptions.CultureInvariant | RegexOptions.IgnoreCase);
Directory.EnumerateFiles(glob.BasePath, "*", new EnumerationOptions {RecurseSubdirectories = true})
.Select(path => Path.GetRelativePath(glob.BasePath, path))
.Where(path => regex.IsMatch(path))
.ForEach(Console.WriteLine);The following patterns are supported:
| Pattern | Description | Example |
|---|---|---|
? |
Any single character except the directory separator | 202305??.log |
* |
Zero or more of any characters except the directory separator | src/*.md |
**/ |
Zero or more path segments ending with the directory separator | **/bin/ |
** |
Equivalent of **/*, only valid at the beginning of the segment, if it occurs not at the beginning of the segment it is interpreted as consecutive * |
**, **.cs |
*. |
Trailing dot changes the behavior of the wildcards (?, *, **) of the last segment of the path to not include a dot, i.e. allows to match file system entries without extension. If there are no wildcards in the last segment, the trailing dot is treated as a regular character |
test/*., split/x?a. |
Both / and \ are treated as directory separator characters. So **/* is the same as **\*.
Consecutive * after ** (and * if not at the beginning of the segment) are ignored. So ***/abc*** is the same as **/abc*.
Escaping of wildcard characters is not supported.
The passed glob is not normalized, in particular ./ and ../, consecutive directory separators like // or \\, etc. processed literally. ./ and ../ in normal use are expected to appear only in the BasePath and can be normalized afterwards. In the usage example above Path.GetRelativePath method resolve paths by calling the GetFullPath normalization method before calculating the difference.
GlobRegex nuget package is built using ReproducibleBuilds with the SourceLink. Reproducible builds give confidence by allowing to validate that the package has actually been built using public sources. To be able to reproduce a build, you need the source files, the referenced DLLs, the compiler version, and the compiler options (language version, defines, nullables, etc.). All this information is available using Nuget Package Explorer. Or you can use dotnet-validate tool to validate the package.
Copyright (c) Dmitriy dscheg Titarenko 2023
GlobRegex is distributed under BSD 3-Clause License