[API Proposal]: IndexOfAnyValues<T>.Contains(T)

[MihaZupan]

Background and motivation

Augmenting the recently-added IndexOfAnyValues APIs: #68328 (comment)

When updating existing code to use the new IndexOfAnyValues API, it is very common to replace the existing needle constant with the IndexOfAnyValues instance.

This is fine if the only operations using said needle were {Last}IndexOfAny{Except} calls, but it is not uncommon to also use that needle for single-value checks Needle.Contains(c). In those cases, you are forced into keeping the original needle around.

As IndexOfAnyValues is also already a representation of values optimized for searching, it can be more efficient than a full linear scan of the needle.

API Proposal

namespace System.Buffers;

public class IndexOfAnyValues<T> where T : IEquatable<T>?
{
    public virtual bool Contains(T value);
}

API Usage

Instead of

private const string Needle = ",[]&*+";
private static readonly IndexOfAnyValues<char> s_needle = IndexOfAnyValues.Create(Needle);

if (Needle.Contains(c))
{
    // ...
}

you can do

private static readonly IndexOfAnyValues<char> s_needle = IndexOfAnyValues.Create(",[]&*+");

if (s_needle.Contains(c))
{
    // ...
}

Alternative Designs

No response

Risks

No response

[ghost]

Tagging subscribers to this area: @dotnet/area-system-memory
See info in area-owners.md if you want to be subscribed.

Issue Details

---

Background and motivation

When updating existing code to use the new IndexOfAnyValues API, it is very common to replace the existing needle constant with the IndexOfAnyValues instance.
This is fine if the only operations using said needle were {Last}IndexOfAny{Except} calls, but it is not uncommon to also use that needle for single-value checks Needle.Contains(c). In those cases, you are forced into keeping the original needle around.

private const string Needle = ",[]&*+";
private static readonly IndexOfAnyValues<char> s_needle = IndexOfAnyValues.Create(Needle);

if (Needle.Contains(c))
{
    // ...
}

As IndexOfAnyValues is also already a representation of values optimized for searching, it can be more efficient than a full linear scan of the needle.

private static readonly IndexOfAnyValues<char> s_needle = IndexOfAnyValues.Create(",[]&*+");

if (s_needle.Contains(c))
{
    // ...
}

API Proposal

namespace System.Buffers;

public class IndexOfAnyValues<T> where T : IEquatable<T>?
{
    public bool Contains(T value);
}

API Usage

Instead of

private const string Needle = ",[]&*+";
private static readonly IndexOfAnyValues<char> s_needle = IndexOfAnyValues.Create(Needle);

if (Needle.Contains(c))
{
    // ...
}

you can do

private static readonly IndexOfAnyValues<char> s_needle = IndexOfAnyValues.Create(",[]&*+");

if (s_needle.Contains(c))
{
    // ...
}

Alternative Designs

No response

Risks

No response

Author:	MihaZupan
Assignees:	-
Labels:	api-suggestion, area-System.Memory, tenet-performance
Milestone:	-

[MihaZupan]

cc: @stephentoub

[jkotas]

Alternative Designs

Should we have a different type for this? Searching for a single match in a fixed set seems to be very different operation from searching for a fixed set match in a sequence.

[stephentoub]

We could, though the few examples I've seen so far have been related enough that I think it makes sense to have this on the same type (plus, the instance has also likely already precomputed the state to do the single check quickly due to needing it for scalar IndexOfAny operations). Examples have been of the form:

int pos = data.IndexOfAny(s_values);
if (pos < 0)
    return data;

data = data.Slice(pos);
foreach (char c in data)
{
    if (s_values.Contains(c))
        Process(c);
}

so basically using the values for both a fast path and a slow path.

[MihaZupan]

Another example where this API would be really useful: improving Uri.Escape helpers MihaZupan@a5027e1

A few other cases:

• runtime/src/libraries/Common/src/System/IO/Archiving.Utils.Windows.cs

  Line 10 in c18e3d9

  internal static string SanitizeEntryFilePath(string entryPath)

• runtime/src/libraries/System.Security.Cryptography/src/System/Security/Cryptography/PemEncoding.cs

  Line 200 in 62e87b4

  if (IsLabelChar(c))

[terrajobst]

Video

• Looks good but we'd prefer if the public method weren't virtual, i.e. the virtuality would be hidden.

namespace System.Buffers;

public class IndexOfAnyValues<T> where T : IEquatable<T>?
{
    public bool Contains(T value);
}