Fast HTTP User-Agent parsing for .NET.

Repository: https://github.com/mycsharp/HttpUserAgentParser

dotnet add package MyCSharp.HttpUserAgentParser

string userAgent = "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/90.0.4430.212 Safari/537.36";
HttpUserAgentInformation info = HttpUserAgentParser.Parse(userAgent);
// or: HttpUserAgentInformation.Parse(userAgent)

If you want to inject a parser (e.g., in ASP.NET Core), use IHttpUserAgentParserProvider.

services
	.AddHttpUserAgentParser();

services
	.AddHttpUserAgentCachedParser();
// or: .AddHttpUserAgentParser<HttpUserAgentParserCachedProvider>();

Telemetry is:

• Opt-in: disabled by default (keeps hot path overhead-free)
• Low overhead: counters are only written when a listener is attached

services
	.AddHttpUserAgentParser()
	.WithTelemetry();

EventSource: MyCSharp.HttpUserAgentParser (constant: HttpUserAgentParserEventSource.EventSourceName)

• parse.requests (incrementing)
• parse.duration (ms, event counter)
• cache.hit (incrementing)
• cache.miss (incrementing)
• cache.size (polling)

dotnet-counters monitor --process-id <pid> MyCSharp.HttpUserAgentParser

In addition to EventCounters, this package can emit native System.Diagnostics.Metrics instruments.

Telemetry is:

• Opt-in: disabled by default (keeps hot path overhead-free)
• Low overhead: measurements are only recorded when enabled

services
	.AddHttpUserAgentParser()
	.WithMeterTelemetry();

Meter: MyCSharp.HttpUserAgentParser (constant: HttpUserAgentParserMeters.MeterName)

• parse.requests (counter)
• parse.duration (histogram, ms)
• cache.hit (counter)
• cache.miss (counter)
• cache.size (observable gauge)

You can collect these EventCounters via OpenTelemetry metrics and export them (OTLP, Prometheus, Azure Monitor, …).

Packages you typically need:

• OpenTelemetry
• OpenTelemetry.Exporter.OpenTelemetryProtocol (or another exporter)
• OpenTelemetry.Instrumentation.EventCounters

Example (minimal):

using OpenTelemetry.Metrics;
using MyCSharp.HttpUserAgentParser.Telemetry;

builder.Services.AddOpenTelemetry()
	.WithMetrics(metrics =>
	{
		metrics
			.AddEventCountersInstrumentation(options =>
			{
				options.AddEventSources(HttpUserAgentParserEventSource.EventSourceName);
			})
			.AddOtlpExporter();
	});

If you also use the MemoryCache/AspNetCore packages, add their EventSource names too.

If you enabled native meters (see above), collect them via AddMeter(...):

using OpenTelemetry.Metrics;
using MyCSharp.HttpUserAgentParser.Telemetry;

builder.Services.AddOpenTelemetry()
	.WithMetrics(metrics =>
	{
		metrics
			.AddMeter(HttpUserAgentParserMeters.MeterName)
			.AddOtlpExporter();
	});

There are two common approaches:

Collect with OpenTelemetry (see above) and export to Azure Monitor / Application Insights using an Azure Monitor exporter. This keeps your pipeline consistent and avoids custom listeners.

Typical packages (names may differ by version):

• OpenTelemetry
• OpenTelemetry.Instrumentation.EventCounters
• Azure.Monitor.OpenTelemetry.Exporter

If you prefer a direct listener, you can attach an EventListener and forward values as custom metrics.

High-level idea:

• Enable the EventSource
• Parse the EventCounters payload
• Track as Application Insights metrics

Notes:

• This is best-effort telemetry (caches can race)
• Keep aggregation intervals reasonable (e.g. 10s)