Add CV extraction review surfaces

2026-03-28 23:51:03 +01:00
parent 107c181506
commit 2392b135c2
7 changed files with 390 additions and 34 deletions
@@ -76,6 +76,18 @@ public sealed class ProfileCvController : ControllerBase
    public sealed record ParseCvRequest(string? Text);

    private sealed record ExtractionPipelineResult(string RawText, string NormalizedText, StructuredCvProfile StructuredCv);
+    public sealed record CvExtractionRunListItem(
+        int Id,
+        string Trigger,
+        string Status,
+        string? ArtifactFileName,
+        DateTimeOffset StartedAtUtc,
+        DateTimeOffset? CompletedAtUtc,
+        DateTimeOffset? AppliedAtUtc,
+        string ParserVersion,
+        string NormalizerVersion,
+        string LlmPromptVersion,
+        string? ErrorMessage);

    [HttpPost("upload")]
    [RequestSizeLimit(MaxFileSizeBytes)]
@@ -163,6 +175,34 @@ public sealed class ProfileCvController : ControllerBase
        }
    }

+    [HttpGet("runs")]
+    public async Task<ActionResult<IEnumerable<CvExtractionRunListItem>>> GetRuns()
+    {
+        var user = await _users.GetUserAsync(User);
+        if (user is null) return Unauthorized();
+
+        var runs = await _db.CvExtractionRuns
+            .AsNoTracking()
+            .Where(x => x.OwnerUserId == user.Id)
+            .OrderByDescending(x => x.StartedAtUtc)
+            .Take(10)
+            .Select(x => new CvExtractionRunListItem(
+                x.Id,
+                x.Trigger,
+                x.Status,
+                x.Artifact != null ? x.Artifact.OriginalFileName : null,
+                x.StartedAtUtc,
+                x.CompletedAtUtc,
+                x.AppliedAtUtc,
+                x.ParserVersion,
+                x.NormalizerVersion,
+                x.LlmPromptVersion,
+                x.ErrorMessage))
+            .ToListAsync(HttpContext.RequestAborted);
+
+        return Ok(runs);
+    }
+
    [HttpPost("reprocess")]
    public async Task<IActionResult> Reprocess()
    {
@@ -308,7 +348,9 @@ public sealed class ProfileCvController : ControllerBase
            .ToList();

        var sectionFallback = StructuredCvProfileJson.FromSections(fallbackSections);
+        AnnotateStructuredCv(sectionFallback, "repair", 0.56);
        var heuristicFallback = BuildHeuristicStructuredCv(parseSource, text);
+        AnnotateStructuredCv(heuristicFallback, "deterministic", 0.68);
        heuristicFallback.Sections = new List<StructuredCvSection>();
        var fallback = StructuredCvProfileJson.Merge(heuristicFallback, sectionFallback);
        fallback.Contact.FullName ??= GuessFullName(text) ?? GuessFullNameFromEmail(fallback.Contact.Email);
@@ -433,6 +475,40 @@ public sealed class ProfileCvController : ControllerBase
        await _db.SaveChangesAsync(cancellationToken);
    }

+    private static void AnnotateStructuredCv(StructuredCvProfile profile, string method, double confidence)
+    {
+        var now = DateTimeOffset.UtcNow;
+        profile.Metadata ??= new StructuredCvMetadata();
+        profile.Metadata.Fields ??= new Dictionary<string, StructuredCvFieldMetadata>();
+
+        void SetIf(string key, string? value)
+        {
+            if (string.IsNullOrWhiteSpace(value)) return;
+            profile.Metadata.Fields[key] = new StructuredCvFieldMetadata
+            {
+                Confidence = confidence,
+                Method = method,
+                SourceSnippet = value.Length > 180 ? value[..180] : value,
+                ReviewState = "suggested",
+                LastUpdatedAtUtc = now,
+            };
+        }
+
+        SetIf("contact.fullName", profile.Contact.FullName);
+        SetIf("contact.headline", profile.Contact.Headline);
+        SetIf("contact.email", profile.Contact.Email);
+        SetIf("contact.phone", profile.Contact.Phone);
+        SetIf("contact.location", profile.Contact.Location);
+        SetIf("contact.website", profile.Contact.Website);
+        SetIf("contact.linkedIn", profile.Contact.LinkedIn);
+        SetIf("summary", profile.Summary.FirstOrDefault());
+        SetIf("skills", profile.Skills.FirstOrDefault());
+        SetIf("languages", profile.Languages.FirstOrDefault()?.Name);
+        SetIf("interests", profile.Interests.FirstOrDefault());
+        SetIf("jobs", profile.Jobs.FirstOrDefault()?.Title ?? profile.Jobs.FirstOrDefault()?.Company);
+        SetIf("education", profile.Education.FirstOrDefault()?.Qualification ?? profile.Education.FirstOrDefault()?.Institution);
+    }
+
    private async Task<StructuredCvProfile?> TryExtractStructuredCvAsync(string text, CancellationToken cancellationToken)
    {
        var structuredJson = await _aiService.SummarizeSectionAsync(
@@ -446,7 +522,10 @@ public sealed class ProfileCvController : ControllerBase
        if (string.IsNullOrWhiteSpace(extracted)) return null;

        var parsed = StructuredCvProfileJson.Deserialize(extracted);
-        return IsMeaningfullyStructured(parsed) ? parsed : null;
+        if (!IsMeaningfullyStructured(parsed)) return null;
+
+        AnnotateStructuredCv(parsed, "llm", 0.82);
+        return parsed;
    }

    private static bool IsMeaningfullyStructured(StructuredCvProfile profile)