DotCov 1.0.0

DotCov

Read Cobertura XML from your own code — parse, merge, threshold, and diff coverage reports. Zero package references, Native-AOT clean, and streaming rather than DOM-loading.

Getting started

dotnet add package DotCov

using DotCov;
using DotCov.Formatters;

var inputs = ReportResolver.Resolve("TestResults/");      // file or directory
var report = CoberturaParser.Parse(inputs)
                            .Exclude(ExclusionRules.WellKnown);

Console.WriteLine(TableFormatter.Format(report));

var gate = report.Evaluate(minLinePercent: 80, minBranchPercent: 60);
if (!gate.IsPass)
{
    Console.Error.WriteLine(gate);   // "NODATA: line n/a (min 80%) - …"
    return 1;
}

ReportResolver.Resolve takes a file or a directory; a directory is searched for **/*cobertura*.xml (pass a ReportPattern to change that) and Parse merges every match, so a sharded test matrix needs no merge step. An existing directory with no match yields no inputs and therefore an empty report; a missing path throws.

Two things the API insists on

Rates are double?. null means unanswerable — not 0.0, not 1.0. An empty report has no rate, and treating that as 0% would fail a build that simply measured nothing.

A threshold check has four outcomes, not two. Evaluate returns a GateResult whose Outcome is Pass, Fail, NoData, or Disabled; IsPass covers Pass alone. Branch on the structured flags (LineBelowThreshold, BranchBelowThreshold, IsInconclusive) rather than parsing Reason.

What else is in the box

// Compare two reports — added / removed / modified, per-file deltas, line-level flips
var diff = CoverageDiff.Compare(
    CoberturaParser.Parse(ReportInput.FromFile("before.xml")),
    CoberturaParser.Parse(ReportInput.FromFile("after.xml")));

foreach (var r in diff.Regressions)
    Console.WriteLine($"{r.Path}: {r.Before:P1} → {r.After:P1}");

// Async streaming, cancellable
await using var stream = File.OpenRead("coverage.cobertura.xml");
var report = await CoberturaParser.ParseAsync(stream, ct: cancellationToken);

// Per-method detail — complexity and line hits, the input behind the CRAP gate
var methods = CoberturaParser.ParseMethods(ReportResolver.Resolve("TestResults/"));
// methods.Methods, methods.Warnings, methods.SourceRoots

TableFormatter, MarkdownFormatter, and JsonFormatter render a report for a terminal, a PR comment, or a pipeline. Every numeric rendering is invariant-formatted: 62.0% on every host, never 62,0%. CoverageSnapshot wraps a report with commit, branch, project, timestamp, and a SHA-256 of the source file.

Parsing and safety

CoberturaParser walks the document with XmlReader — no XDocument.Load, no full-DOM allocation, bounded memory. DTDs are prohibited and XmlResolver is null, so there is no XXE or entity-expansion surface, and each file is capped at 50,000,000 characters by default. Raise or remove the cap with the maxChars overloads (0 disables it).

The package has no PackageReferences at all and builds with the trim/AOT analyzers on and warnings as errors, so AOT-cleanliness is enforced by the compiler rather than by convention.

Every public type carries XML docs — read them in IntelliSense or in the source.

Also in this family

DotCov.Tool — the dotcov CLI · DotCov.Fallout — Fallout build component

Feedback

Documentation · Issues · MIT