feat: deterministic CV-to-job match score endpoint

New JobCvMatchService: a pure, AI-free keyword-coverage scorer that returns a stable, reproducible 0-100 match score plus matched/missing keyword lists and per-CV-section coverage. Unlike candidate-fit (AI narrative), it makes no model calls, so results are instant and identical for identical inputs - the Jobscan-style differentiator. - GET /api/jobapplications/{id}/match-score - keywords = curated SkillTagger tags (high weight) + salient posting terms (title terms boosted); word-boundary matching avoids false hits - section coverage shows where CV evidence is concentrated - fix(SkillTagger): punctuation-tolerant C#/.NET patterns; the old \b boundaries silently missed 'C#,' and '.NET,' everywhere they are used - 7 unit tests on the pure scorer; full backend suite green (104) Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-07-03 03:16:32 +02:00
parent 83e6430a24
commit 3fad43a9e2
5 changed files with 404 additions and 3 deletions
@@ -0,0 +1,105 @@
+using JobTrackerApi.Services;
+using Xunit;
+
+namespace JobTrackerApi.Tests;
+
+public sealed class JobCvMatchServiceTests
+{
+    private readonly JobCvMatchService _service = new();
+
+    private static Dictionary<string, string> Sections(params (string Name, string Text)[] items)
+        => items.ToDictionary(i => i.Name, i => i.Text, StringComparer.OrdinalIgnoreCase);
+
+    [Fact]
+    public void Strong_overlap_scores_high_and_lists_matched_keywords()
+    {
+        var result = _service.Evaluate(
+            jobTitle: "Senior C# Backend Developer",
+            jobText: "We need a backend engineer with strong C#, .NET, SQL and Docker experience building REST APIs.",
+            cvSections: Sections(
+                ("Skills", "C# .NET SQL Docker Kubernetes"),
+                ("Experience", "Built REST APIs in C# and .NET with SQL Server and Docker.")));
+
+        Assert.True(result.Score >= 75, $"expected strong score, got {result.Score}");
+        Assert.Equal("Strong", result.Band);
+        Assert.Contains("C#", result.MatchedKeywords);
+        Assert.Contains(".NET", result.MatchedKeywords);
+        Assert.True(result.HasEnoughSignal);
+    }
+
+    [Fact]
+    public void No_overlap_scores_low_and_surfaces_missing_keywords()
+    {
+        var result = _service.Evaluate(
+            jobTitle: "Kubernetes Platform Engineer",
+            jobText: "Deep Kubernetes, AWS, and Docker platform experience required. Terraform and CI/CD pipelines.",
+            cvSections: Sections(
+                ("Skills", "Graphic design, Adobe Photoshop, Illustrator, copywriting"),
+                ("Experience", "Ran marketing campaigns and brand design work.")));
+
+        Assert.True(result.Score < 50, $"expected low score, got {result.Score}");
+        Assert.Equal("Low", result.Band);
+        Assert.Contains("Kubernetes", result.MissingKeywords);
+        Assert.Contains("AWS", result.MissingKeywords);
+    }
+
+    [Fact]
+    public void Is_deterministic_for_identical_inputs()
+    {
+        var a = _service.Evaluate("Data Engineer", "Python, SQL, Spark, ETL pipelines and AWS.", Sections(("Skills", "Python SQL AWS")));
+        var b = _service.Evaluate("Data Engineer", "Python, SQL, Spark, ETL pipelines and AWS.", Sections(("Skills", "Python SQL AWS")));
+
+        Assert.Equal(a.Score, b.Score);
+        Assert.Equal(a.MatchedKeywords, b.MatchedKeywords);
+        Assert.Equal(a.MissingKeywords, b.MissingKeywords);
+    }
+
+    [Fact]
+    public void Word_boundary_prevents_false_substring_matches()
+    {
+        // "go" (the language) must not match inside "goals"/"ago".
+        var result = _service.Evaluate(
+            jobTitle: "Go Developer",
+            jobText: "Go programming language, goroutines, concurrency.",
+            cvSections: Sections(("Experience", "Achieved company goals two years ago in a great environment.")));
+
+        Assert.DoesNotContain("go", result.MatchedKeywords, StringComparer.OrdinalIgnoreCase);
+    }
+
+    [Fact]
+    public void Section_coverage_reports_where_matches_are_concentrated()
+    {
+        var result = _service.Evaluate(
+            jobTitle: "React Frontend Engineer",
+            jobText: "Build UIs with React, TypeScript and JavaScript. Strong testing culture.",
+            cvSections: Sections(
+                ("Skills", "React TypeScript JavaScript"),
+                ("Experience", "Wrote documentation and managed budgets.")));
+
+        var skills = Assert.Single(result.SectionCoverage, s => s.Section == "Skills");
+        var experience = Assert.Single(result.SectionCoverage, s => s.Section == "Experience");
+        Assert.True(skills.Matched > experience.Matched);
+    }
+
+    [Fact]
+    public void Empty_cv_reports_no_signal()
+    {
+        var result = _service.Evaluate("Anything", "Some role text with several words here.", Sections());
+        Assert.False(result.HasEnoughSignal);
+        Assert.Equal("Unknown", result.Band);
+        Assert.Equal(0, result.MatchedCount);
+    }
+
+    [Fact]
+    public void Title_keywords_are_weighted_and_missing_ones_rank_first()
+    {
+        // The title term "kubernetes" is absent from the CV; it should lead the missing list
+        // because title terms carry the title bonus weight.
+        var result = _service.Evaluate(
+            jobTitle: "Kubernetes Specialist",
+            jobText: "Kubernetes orchestration. Some familiarity with logging and monitoring dashboards.",
+            cvSections: Sections(("Skills", "logging monitoring dashboards")));
+
+        Assert.Equal("Kubernetes", result.MissingKeywords.First());
+    }
+}