feat: complete phase 2 UX improvements

JobTrackerApi/Controllers/ProfileCvController.cs

@@ -72,8 +72,9 @@ public sealed class ProfileCvController : ControllerBase
     private readonly ICvProcessingQueue _cvProcessingQueue;
     private readonly IAppEmailSender _emailSender;
     private readonly ICareerProfileService _careerProfileService;
+    private readonly ICvProfileDiffService _cvProfileDiffService;
 
-    public ProfileCvController(UserManager<ApplicationUser> users, ISummarizerService aiService, JobTrackerContext db, AppPaths paths, ILogger<ProfileCvController>? logger = null, ICvAiClassifier? cvAiClassifier = null, ICvAiNormalizer? cvAiNormalizer = null, ICvTemplateRenderer? cvTemplateRenderer = null, ICvPdfExporter? cvPdfExporter = null, ICvProcessingQueue? cvProcessingQueue = null, IAppEmailSender? emailSender = null, ICareerProfileService? careerProfileService = null)
+    public ProfileCvController(UserManager<ApplicationUser> users, ISummarizerService aiService, JobTrackerContext db, AppPaths paths, ILogger<ProfileCvController>? logger = null, ICvAiClassifier? cvAiClassifier = null, ICvAiNormalizer? cvAiNormalizer = null, ICvTemplateRenderer? cvTemplateRenderer = null, ICvPdfExporter? cvPdfExporter = null, ICvProcessingQueue? cvProcessingQueue = null, IAppEmailSender? emailSender = null, ICareerProfileService? careerProfileService = null, ICvProfileDiffService? cvProfileDiffService = null)
     {
         _users = users;
         _aiService = aiService;
@@ -87,6 +88,7 @@ public sealed class ProfileCvController : ControllerBase
         _cvProcessingQueue = cvProcessingQueue ?? NoOpCvProcessingQueue.Instance;
         _emailSender = emailSender ?? NoOpEmailSender.Instance;
         _careerProfileService = careerProfileService ?? new CareerProfileService(db);
+        _cvProfileDiffService = cvProfileDiffService ?? new CvProfileDiffService();
     }
 
     private sealed class NoOpEmailSender : IAppEmailSender
@@ -120,6 +122,7 @@ public sealed class ProfileCvController : ControllerBase
     // in-file copies are dropped here to avoid duplicate definitions. The LayoutFamily/AtsRating
     // fields the branch added to CvTemplateDescriptor belong to the deferred ATS-badge work
     // (Phase 4), not this foundation integration.
+    public sealed record AcceptCvRunRequest(List<string>? AcceptedLowConfidenceIds);
     private sealed record ExtractionPipelineResult(string RawText, string NormalizedText, StructuredCvProfile StructuredCv);
     private sealed record ClassifiedCvBlock(int Index, string OriginalBlock, string SectionName, string Content, CvBlockClassificationResult? Classification);
 
@@ -159,45 +162,21 @@ public sealed class ProfileCvController : ControllerBase
         try
         {
             var result = await ExtractStructuredCvFromFileAsync(file, extension, HttpContext.RequestAborted);
-            result.StructuredCv.Metadata.ProfileVersion = (user.CurrentCvProfileVersion ?? 0) + 1;
-            result.StructuredCv.Metadata.AppliedExtractionRunId = run.Id;
-            result.StructuredCv.Metadata.UpdatedAtUtc = DateTimeOffset.UtcNow;
-            await _careerProfileService.SaveVersionAsync(user.Id, result.StructuredCv, "upload", HttpContext.RequestAborted);
-            var structuredJson = StructuredCvProfileJson.Serialize(result.StructuredCv);
-
             run.RawExtractedText = result.RawText;
             run.NormalizedText = result.NormalizedText;
-            run.StructuredProfileJson = structuredJson;
-            run.Status = "applied";
+            run.StructuredProfileJson = StructuredCvProfileJson.Serialize(result.StructuredCv);
+            run.Status = "pending_review";
             run.CompletedAtUtc = DateTimeOffset.UtcNow;
-            run.AppliedAtUtc = run.CompletedAtUtc;
-
-            user.ProfileCvText = result.NormalizedText;
-            user.ProfileCvStructureJson = structuredJson;
-            user.CurrentCvUploadArtifactId = artifact.Id;
-            user.CurrentCvExtractionRunId = run.Id;
-            user.CurrentCvProfileVersion = result.StructuredCv.Metadata.ProfileVersion;
-
-            var update = await _users.UpdateAsync(user);
-            if (!update.Succeeded)
-            {
-                run.Status = "failed";
-                run.ErrorMessage = string.Join("; ", update.Errors.Select(e => e.Description));
-                await _db.SaveChangesAsync(HttpContext.RequestAborted);
-                return BadRequest(run.ErrorMessage);
-            }
-
             await _db.SaveChangesAsync(HttpContext.RequestAborted);
 
             return Ok(new
             {
-                imported = true,
+                imported = false,
+                pendingReview = true,
                 characters = result.NormalizedText.Length,
-                structuredCv = result.StructuredCv,
-                sections = result.StructuredCv.Sections,
                 artifactId = artifact.Id,
                 extractionRunId = run.Id,
-                profileVersion = result.StructuredCv.Metadata.ProfileVersion,
+                status = run.Status,
             });
         }
         catch (Exception ex)
@@ -238,6 +217,71 @@ public sealed class ProfileCvController : ControllerBase
         return Ok(runs);
     }
 
+    [HttpGet("runs/{id:int}/diff")]
+    public async Task<IActionResult> GetRunDiff([FromRoute] int id)
+    {
+        var user = await _users.GetUserAsync(User);
+        if (user is null) return Unauthorized();
+
+        var run = await _db.CvExtractionRuns.AsNoTracking().FirstOrDefaultAsync(x => x.Id == id && x.OwnerUserId == user.Id, HttpContext.RequestAborted);
+        if (run is null) return NotFound();
+        if (string.IsNullOrWhiteSpace(run.StructuredProfileJson)) return Conflict("This extraction run has no reviewable profile.");
+
+        var current = await _careerProfileService.LoadStructuredAsync(user.Id, HttpContext.RequestAborted);
+        var extracted = StructuredCvProfileJson.Deserialize(run.StructuredProfileJson);
+        return Ok(new { runId = run.Id, run.Status, diff = _cvProfileDiffService.Diff(current, extracted) });
+    }
+
+    [HttpPost("runs/{id:int}/accept")]
+    public async Task<IActionResult> AcceptRun([FromRoute] int id, [FromBody] AcceptCvRunRequest? request = null)
+    {
+        var user = await _users.GetUserAsync(User);
+        if (user is null) return Unauthorized();
+
+        var run = await _db.CvExtractionRuns.FirstOrDefaultAsync(x => x.Id == id && x.OwnerUserId == user.Id, HttpContext.RequestAborted);
+        if (run is null) return NotFound();
+        if (run.Status != "pending_review") return Conflict("This extraction run is not awaiting review.");
+        if (string.IsNullOrWhiteSpace(run.StructuredProfileJson)) return Conflict("This extraction run has no reviewable profile.");
+
+        var current = await _careerProfileService.LoadStructuredAsync(user.Id, HttpContext.RequestAborted);
+        var extracted = StructuredCvProfileJson.Deserialize(run.StructuredProfileJson);
+        var diff = _cvProfileDiffService.Diff(current, extracted);
+        var acceptedLowConfidenceIds = (request?.AcceptedLowConfidenceIds ?? new List<string>()).ToHashSet(StringComparer.Ordinal);
+        var merged = _cvProfileDiffService.Merge(current, extracted, acceptedLowConfidenceIds);
+        merged.Metadata.ProfileVersion = (user.CurrentCvProfileVersion ?? 0) + 1;
+        merged.Metadata.AppliedExtractionRunId = run.Id;
+        merged.Metadata.UpdatedAtUtc = DateTimeOffset.UtcNow;
+        await _careerProfileService.SaveVersionAsync(user.Id, merged, $"{run.Trigger}:accepted", HttpContext.RequestAborted);
+
+        user.ProfileCvStructureJson = StructuredCvProfileJson.Serialize(merged);
+        if (string.IsNullOrWhiteSpace(user.ProfileCvText)) user.ProfileCvText = run.NormalizedText;
+        user.CurrentCvExtractionRunId = run.Id;
+        user.CurrentCvProfileVersion = merged.Metadata.ProfileVersion;
+        if (run.ArtifactId.HasValue) user.CurrentCvUploadArtifactId = run.ArtifactId.Value;
+
+        run.Status = "applied";
+        run.AppliedAtUtc = DateTimeOffset.UtcNow;
+        var update = await _users.UpdateAsync(user);
+        if (!update.Succeeded) return BadRequest(string.Join("; ", update.Errors.Select(e => e.Description)));
+        await _db.SaveChangesAsync(HttpContext.RequestAborted);
+        return Ok(new { runId = run.Id, run.Status, diff, structuredCv = merged });
+    }
+
+    [HttpPost("runs/{id:int}/discard")]
+    public async Task<IActionResult> DiscardRun([FromRoute] int id)
+    {
+        var user = await _users.GetUserAsync(User);
+        if (user is null) return Unauthorized();
+
+        var run = await _db.CvExtractionRuns.FirstOrDefaultAsync(x => x.Id == id && x.OwnerUserId == user.Id, HttpContext.RequestAborted);
+        if (run is null) return NotFound();
+        if (run.Status != "pending_review") return Conflict("This extraction run is not awaiting review.");
+
+        run.Status = "discarded";
+        await _db.SaveChangesAsync(HttpContext.RequestAborted);
+        return NoContent();
+    }
+
     [HttpPost("reprocess")]
     public async Task<IActionResult> Reprocess()
     {
@@ -807,6 +851,7 @@ public sealed class ProfileCvController : ControllerBase
             throw new InvalidOperationException("The uploaded CV file could not be read or was empty.");
         }
 
+        text = RepairKnownMojibake(text);
         var normalizedText = (await MaybeReconstructStructuredCvAsync(text, cancellationToken)).Trim();
         var structuredCv = await BuildStructuredCvAsync(normalizedText, cancellationToken);
         return new ExtractionPipelineResult(text, normalizedText, structuredCv);
@@ -916,7 +961,7 @@ public sealed class ProfileCvController : ControllerBase
 
                     var normalizedText = rebuilt.Trim();
                     var structuredCv = await BuildStructuredCvAsync(normalizedText, cancellationToken);
-                    await ApplyQueuedRunResultAsync(run, user, normalizedText, normalizedText, structuredCv, run.ArtifactId, cancellationToken);
+                    await CompleteQueuedRunForReviewAsync(run, normalizedText, normalizedText, structuredCv, cancellationToken);
                     break;
                 }
                 case "improve":
@@ -931,7 +976,7 @@ public sealed class ProfileCvController : ControllerBase
 
                     var normalizedText = improved.Trim();
                     var structuredCv = await BuildStructuredCvAsync(normalizedText, cancellationToken);
-                    await ApplyQueuedRunResultAsync(run, user, normalizedText, normalizedText, structuredCv, run.ArtifactId, cancellationToken);
+                    await CompleteQueuedRunForReviewAsync(run, normalizedText, normalizedText, structuredCv, cancellationToken);
                     break;
                 }
                 case "reprocess":
@@ -951,7 +996,7 @@ public sealed class ProfileCvController : ControllerBase
                     };
                     var extension = Path.GetExtension(artifact.OriginalFileName ?? string.Empty);
                     var result = await ExtractStructuredCvFromFileAsync(file, extension, cancellationToken);
-                    await ApplyQueuedRunResultAsync(run, user, result.RawText, result.NormalizedText, result.StructuredCv, artifact.Id, cancellationToken);
+                    await CompleteQueuedRunForReviewAsync(run, result.RawText, result.NormalizedText, result.StructuredCv, cancellationToken);
                     break;
                 }
                 default:
@@ -971,39 +1016,13 @@ public sealed class ProfileCvController : ControllerBase
         }
     }
 
-    private async Task ApplyQueuedRunResultAsync(CvExtractionRun run, ApplicationUser user, string rawText, string normalizedText, StructuredCvProfile structuredCv, int? artifactId, CancellationToken cancellationToken)
+    private async Task CompleteQueuedRunForReviewAsync(CvExtractionRun run, string rawText, string normalizedText, StructuredCvProfile structuredCv, CancellationToken cancellationToken)
     {
-        structuredCv.Metadata.ProfileVersion = (user.CurrentCvProfileVersion ?? 0) + 1;
-        structuredCv.Metadata.AppliedExtractionRunId = run.Id;
-        structuredCv.Metadata.UpdatedAtUtc = DateTimeOffset.UtcNow;
-        await _careerProfileService.SaveVersionAsync(user.Id, structuredCv, run.Trigger, cancellationToken);
-        var structuredJson = StructuredCvProfileJson.Serialize(structuredCv);
-
         run.RawExtractedText = rawText;
         run.NormalizedText = normalizedText;
-        run.StructuredProfileJson = structuredJson;
-        run.Status = "applied";
+        run.StructuredProfileJson = StructuredCvProfileJson.Serialize(structuredCv);
+        run.Status = "pending_review";
         run.CompletedAtUtc = DateTimeOffset.UtcNow;
-        run.AppliedAtUtc = run.CompletedAtUtc;
-
-        user.ProfileCvText = normalizedText;
-        user.ProfileCvStructureJson = structuredJson;
-        user.CurrentCvExtractionRunId = run.Id;
-        user.CurrentCvProfileVersion = structuredCv.Metadata.ProfileVersion;
-        if (artifactId.HasValue)
-        {
-            user.CurrentCvUploadArtifactId = artifactId.Value;
-        }
-
-        var update = await _users.UpdateAsync(user);
-        if (!update.Succeeded)
-        {
-            run.Status = "failed";
-            run.ErrorMessage = string.Join("; ", update.Errors.Select(e => e.Description));
-            await _db.SaveChangesAsync(cancellationToken);
-            throw new InvalidOperationException(run.ErrorMessage);
-        }
-
         await _db.SaveChangesAsync(cancellationToken);
     }
 
@@ -1606,6 +1625,15 @@ public sealed class ProfileCvController : ControllerBase
     {
         var normalized = content.Replace("\r\n", "\n").Trim();
         var structured = StructuredCvProfileJson.FromSections(new[] { new StructuredCvSection { Name = "Work Experience", Content = normalized } }).Jobs;
+        var earlierRoles = ParseEarlierRoles(normalized);
+        foreach (var role in earlierRoles)
+        {
+            if (!structured.Any(job => string.Equals(job.Title, role.Title, StringComparison.OrdinalIgnoreCase)
+                && string.Equals(job.Company, role.Company, StringComparison.OrdinalIgnoreCase)))
+            {
+                structured.Add(role);
+            }
+        }
         if (ArePlausibleJobs(structured, null))
         {
             return structured;
@@ -1711,6 +1739,36 @@ public sealed class ProfileCvController : ControllerBase
         return jobs;
     }
 
+    private static List<StructuredCvJob> ParseEarlierRoles(string content)
+    {
+        var heading = Regex.Match(content, @"(?im)^\s*(?:[-*]\s*)?Earlier roles(?:\s*\(part[- ]?time\))?\s*:?\s*$");
+        if (!heading.Success) return new List<StructuredCvJob>();
+
+        var roles = new List<StructuredCvJob>();
+        foreach (var rawLine in content[(heading.Index + heading.Length)..].Split('\n', StringSplitOptions.RemoveEmptyEntries | StringSplitOptions.TrimEntries))
+        {
+            if (rawLine.StartsWith('#')) break;
+            var line = rawLine.Trim().TrimStart('-', '*', '•', ' ');
+            var dates = Regex.Match(line, @"(?<start>\d{4})\s*[-–—]\s*(?<end>\d{4}|Present|Current)", RegexOptions.IgnoreCase);
+            if (!dates.Success) continue;
+
+            var identity = Regex.Replace(line, @"\s*[|,(]?\s*\d{4}\s*[-–—]\s*(?:\d{4}|Present|Current)\s*\)?\s*$", string.Empty, RegexOptions.IgnoreCase).Trim();
+            var parts = Regex.Split(identity, @"\s+(?:—|–|\||at)\s+", RegexOptions.IgnoreCase);
+            if (parts.Length != 2 || parts.Any(string.IsNullOrWhiteSpace)) continue;
+
+            roles.Add(new StructuredCvJob
+            {
+                Title = parts[0].Trim(),
+                Company = parts[1].Trim(),
+                Start = dates.Groups["start"].Value,
+                End = dates.Groups["end"].Value,
+                IsCurrent = dates.Groups["end"].Value.Equals("Present", StringComparison.OrdinalIgnoreCase)
+                    || dates.Groups["end"].Value.Equals("Current", StringComparison.OrdinalIgnoreCase),
+            });
+        }
+        return roles;
+    }
+
     private static string? TitleCasePreservingAcronyms(string? value)
     {
         if (string.IsNullOrWhiteSpace(value)) return null;
@@ -2129,6 +2187,7 @@ public sealed class ProfileCvController : ControllerBase
 
     private static StructuredCvProfile BuildStructuredCvFromNormalizedMarkdown(string text)
     {
+        text = SeparateGluedDateAndTitle(text);
         var sections = ParseSections(text)
             .Select(section => new StructuredCvSection
             {
@@ -2140,6 +2199,16 @@ public sealed class ProfileCvController : ControllerBase
 
         var profile = StructuredCvProfileJson.FromSections(sections);
         profile.Sections = sections;
+        var workExperience = sections.FirstOrDefault(section => section.Name == "Work Experience")?.Content;
+        if (!string.IsNullOrWhiteSpace(workExperience))
+        {
+            profile.Jobs.RemoveAll(job => (job.Title ?? string.Empty).StartsWith("Earlier roles", StringComparison.OrdinalIgnoreCase));
+            foreach (var role in ParseEarlierRoles(workExperience))
+            {
+                if (!profile.Jobs.Any(job => string.Equals(job.Title, role.Title, StringComparison.OrdinalIgnoreCase)
+                    && string.Equals(job.Company, role.Company, StringComparison.OrdinalIgnoreCase))) profile.Jobs.Add(role);
+            }
+        }
 
         if (string.IsNullOrWhiteSpace(profile.Contact.FullName))
         {
@@ -2175,11 +2244,33 @@ public sealed class ProfileCvController : ControllerBase
     private static List<string> OrderSkills(List<string> skills)
     {
         return skills
+            .Select(CleanSkillGroupPrefix)
+            .Where(skill => skill.Length > 0)
             .Distinct(StringComparer.OrdinalIgnoreCase)
             .OrderBy(skill => skill, StringComparer.OrdinalIgnoreCase)
             .ToList();
     }
 
+    private static string CleanSkillGroupPrefix(string skill)
+        => Regex.Replace(skill.Trim(), @"^(?:Development|DevOps(?:\s*&\s*Infrastructure)?|Infrastructure|Practices|Tools|Technologies|Technical Skills)\s*:\s*", string.Empty, RegexOptions.IgnoreCase).Trim();
+
+    private static string SeparateGluedDateAndTitle(string text)
+        => Regex.Replace(text, @"(?<date>\b\d{4}\s*[-–—]\s*(?:\d{4}|Present|Current))(?<title>[\p{L}][^\r\n]*)", "${title}\n${date}", RegexOptions.IgnoreCase);
+
+    private static string RepairKnownMojibake(string text)
+        => text
+            .Replace("ø", "ø", StringComparison.Ordinal)
+            .Replace("Ø", "Ø", StringComparison.Ordinal)
+            .Replace("æ", "æ", StringComparison.Ordinal)
+            .Replace("Æ", "Æ", StringComparison.Ordinal)
+            .Replace("Ã¥", "å", StringComparison.Ordinal)
+            .Replace("Ã…", "Å", StringComparison.Ordinal)
+            .Replace("–", "–", StringComparison.Ordinal)
+            .Replace("—", "—", StringComparison.Ordinal)
+            .Replace("’", "’", StringComparison.Ordinal)
+            .Replace("“", "“", StringComparison.Ordinal)
+            .Replace("”", "”", StringComparison.Ordinal);
+
     private static List<string> CleanInterestItems(List<string> interests)
     {
         return interests
@@ -2229,7 +2320,7 @@ public sealed class ProfileCvController : ControllerBase
 
         if (string.Equals(extension, ".pdf", StringComparison.OrdinalIgnoreCase))
         {
-            var raw = Encoding.UTF8.GetString(bytes);
+            var raw = Encoding.Latin1.GetString(bytes);
             var textMatches = Regex.Matches(raw, @"\((.*?)\)Tj", RegexOptions.Singleline)
                 .Select(match => match.Groups[1].Value)
                 .Concat(Regex.Matches(raw, @"\[(.*?)\]TJ", RegexOptions.Singleline)