using System.Text.Json;
using System.Text.RegularExpressions;
using GerbilManagerWebAPI.Models;
using Microsoft.EntityFrameworkCore;
namespace GerbilManagerWebAPI.Import
{
///
/// FEAT-8c import loader. Consumes tools/import/output (animals.json + litters.json
/// produced by extract.py) and loads conflict-free data into the database:
/// litters first (Wurfchronik = authoritative), then animals matched onto them.
///
/// Load policy:
/// - Litters: all created (idempotent by ExternalRef = source litter id).
/// - Animals: created when they have a DOB and are NOT in conflict. The birth-litter
/// link is set only for HIGH-confidence matches (litterRef.confidence == "hoch");
/// date-only/ambiguous links are quarantined (animal loads with LitterId = null).
/// - QUARANTINED (never loaded): conflicts + stubs (no DOB) — await the wife's review.
/// Idempotent: re-running matches on ExternalRef and skips existing rows.
/// Execute is gated by the endpoint; this service only acts when asked.
///
public sealed class ImportService
{
private static readonly string[] LocusOrder = { "A", "C", "D", "E", "G", "P", "Sp", "Re" };
private static readonly JsonSerializerOptions Json = new()
{
PropertyNameCaseInsensitive = true,
};
private const string ImportSourceTag = "FEAT-8 Stammbaum/Wurfchronik";
private readonly ApplicationContext _db;
private readonly string _sourceDir;
private readonly string _photoRoot;
/// Test-friendly constructor with explicit paths.
public ImportService(ApplicationContext db, string sourceDir, string photoRoot)
{
_db = db;
_sourceDir = sourceDir;
_photoRoot = photoRoot;
}
public ImportService(ApplicationContext db, IConfiguration config, IWebHostEnvironment env)
: this(db,
config["Import:SourcePath"]
?? Path.GetFullPath(Path.Combine(env.ContentRootPath, "..", "tools", "import", "output")),
config["Photos:RootPath"] ?? Path.Combine(env.ContentRootPath, "photo-storage"))
{
}
public async Task RunAsync(bool execute)
{
var notes = new List();
var samples = new List();
var animals = Load>("animals.json") ?? new();
var litters = Load>("litters.json") ?? new();
if (animals.Count == 0 && litters.Count == 0)
notes.Add($"Keine Quelldaten gefunden in {_sourceDir} (animals.json/litters.json). extract.py zuerst ausführen.");
// ---- categorise animals ----
var loadable = new List();
int conflicts = 0, stubs = 0, dateOnly = 0, ambiguous = 0;
foreach (var a in animals)
{
if (a.Conflict) { conflicts++; continue; }
if (string.IsNullOrEmpty(a.Dob)) { stubs++; continue; }
loadable.Add(a);
var conf = a.LitterRef?.Confidence;
if (a.LitterRef?.Candidates is { Count: > 0 }) ambiguous++;
else if (conf == "niedrig") dateOnly++;
}
// existing rows (idempotency). Gerbils carry ExternalRef; Litters have no such
// column, so we key litter idempotency on the stable (Name + Date) pair instead.
var existingGerbilExtRefs = await _db.Gerbils
.Where(g => g.ExternalRef != null)
.Select(g => g.ExternalRef!).ToListAsync();
var existingGerbilSet = existingGerbilExtRefs.ToHashSet();
var existingLitterKeys = await _db.Litters
.Select(l => new { l.Name, l.Date }).ToListAsync();
var existingLitterKeySet = existingLitterKeys
.Select(x => $"{x.Name}|{x.Date:yyyy-MM-dd}").ToHashSet();
// colour-variety name -> id (case-insensitive)
var varieties = await _db.ColorVarieties.Select(v => new { v.Id, v.Name }).ToListAsync();
var varietyByName = varieties
.GroupBy(v => v.Name.Trim().ToLowerInvariant())
.ToDictionary(g => g.Key, g => g.First().Id);
// gender inference from litter roles (sire -> male, dam -> female; both -> unknown)
var sireNames = litters.Select(l => Normalize(StripZucht(l.SireName))).Where(s => s.Length > 0).ToHashSet();
var damNames = litters.Select(l => Normalize(StripZucht(l.DamName))).Where(s => s.Length > 0).ToHashSet();
// ---- litters: create map source.id -> Litter (for high-confidence animal links) ----
int littersCreated = 0, littersExisting = 0;
var litterIdMap = new Dictionary(); // source litter id -> Litter.Id
foreach (var sl in litters)
{
var date = ParseDate(sl.Date);
var name = $"Wurf {sl.LitterId}".Trim();
var key = $"{name}|{date:yyyy-MM-dd}";
if (existingLitterKeySet.Contains(key)) { littersExisting++; continue; }
var id = Guid.NewGuid();
litterIdMap[sl.Id] = id;
littersCreated++;
if (execute && date is DateOnly d)
{
_db.Litters.Add(new Litter
{
Id = id,
Name = name,
Date = d,
TotalBorn = sl.TotalBorn,
Notes = string.IsNullOrWhiteSpace(sl.Note) ? null : sl.Note,
PairingCode = string.IsNullOrWhiteSpace(sl.Zuchtnummer) ? null : sl.Zuchtnummer,
});
}
if (samples.Count < 8 && date is not null)
samples.Add($"Wurf: {name} ({sl.Date}) — {sl.DamName} × {sl.SireName}");
}
if (execute) await _db.SaveChangesAsync();
// ---- animals ----
int animalsCreated = 0, linked = 0, fbMatched = 0, fbUnmatched = 0, animalsExisting = 0;
int photosAttached = 0, photosMissing = 0;
var createdAnimalByName = new Dictionary(); // normalized name -> gerbil id (for litter back-link)
// name+DOB -> gid index, across EXISTING rows AND this run's planned animals, so that
// chart-position parentRefs (PEDIGREE-LINK) can resolve a parent to a real gerbil id.
var existingRows = await _db.Gerbils
.Select(g => new { g.Id, g.Name, g.DateOfBirth, g.ExternalRef, g.LitterId }).ToListAsync();
var gidByNameDob = new Dictionary();
foreach (var g in existingRows)
gidByNameDob[NameDobKey(g.Name, g.DateOfBirth)] = g.Id;
var existingLitterByExtRef = existingRows.Where(g => g.ExternalRef != null)
.ToDictionary(g => g.ExternalRef!, g => g.LitterId);
// PASS 1: assign ids + resolve fb/gender/Wurfchronik link (no writes yet).
var plan = new List();
foreach (var a in loadable)
{
bool exists = existingGerbilSet.Contains(a.Id);
var gid = exists ? gidByNameDob[NameDobKey(a.Name, ParseDate(a.Dob))] : Guid.NewGuid();
Guid? wurfLitterId = null;
if (a.LitterRef?.Confidence == "hoch" && a.LitterRef.Candidates is not { Count: > 0 }
&& litterIdMap.TryGetValue(a.LitterRef.LitterId, out var lid))
wurfLitterId = lid;
Guid? colorVarietyId = null;
var fbCandidates = new[] { a.Farbschlag }.Concat(a.FarbschlagVariants)
.Where(s => !string.IsNullOrWhiteSpace(s));
foreach (var fb in fbCandidates)
if (varietyByName.TryGetValue(fb.Trim().ToLowerInvariant(), out var vid))
{ colorVarietyId = vid; break; }
var gender = InferGender(a, sireNames, damNames);
var norm = Normalize(StripZucht(a.Name));
if (norm.Length > 0) createdAnimalByName.TryAdd(norm, gid);
if (!exists) gidByNameDob.TryAdd(NameDobKey(a.Name, ParseDate(a.Dob)), gid);
var currentLitter = exists && existingLitterByExtRef.TryGetValue(a.Id, out var el) ? el : null;
plan.Add(new AnimalPlan(a, gid, exists, wurfLitterId, currentLitter, colorVarietyId, gender));
}
// PASS 1.5: PEDIGREE-LINK — synthesize/reuse a litter from chart-position parentRefs for
// any animal that has no Wurfchronik litter and isn't already litter-linked. Siblings
// (same father+mother+dob) share one derived litter. Computed for dry-run counts too.
int parentLinksAdded = 0, derivedLitters = 0;
var synthLitterForGid = new Dictionary(); // offspring gid -> synth litter id
var synthLitters = new Dictionary(); // parents+date key -> synth litter
foreach (var p in plan)
{
if (p.WurfLitterId is not null || p.CurrentLitterId is not null) continue;
if (p.A.ParentRefs is not { Count: > 0 }) continue;
var father = ResolveParentGid(p.A, "father", gidByNameDob);
var mother = ResolveParentGid(p.A, "mother", gidByNameDob);
if (father is null && mother is null) continue; // nothing resolvable to link
var dob = ParseDate(p.A.Dob);
var conf = p.A.ParentRefs.FirstOrDefault()?.Confidence ?? "medium";
var key = $"{father}|{mother}|{dob:yyyy-MM-dd}";
if (!synthLitters.TryGetValue(key, out var sl))
{
sl = new SynthLitter(Guid.NewGuid(), father, mother, dob, conf);
synthLitters[key] = sl;
derivedLitters++;
}
synthLitterForGid[p.Gid] = sl.Id;
parentLinksAdded++;
}
// FK-INTEGRITY (PEDIGREE-LINK bug fix): a litter's Father/MotherId must resolve to a
// gerbil that is created-or-existing, or Postgres throws FK_Litters_Gerbils_*. Compute
// the persisted set (existing DB rows + this run's loadable animals) and drop any parent
// FK that isn't in it; SKIP a derived litter whose BOTH parents are unresolvable (its
// offspring then load with LitterId=null — still better than 'unbekannt' won't regress).
// This runs in dry-run too, so a green dry-run GUARANTEES /import/execute won't FK-fault.
var persisted = new HashSet(existingRows.Select(r => r.Id));
foreach (var p in plan) persisted.Add(p.Gid);
int litterParentFksDropped = 0, derivedLittersSkipped = 0;
foreach (var key in synthLitters.Keys.ToList())
{
var sl = synthLitters[key];
var f = sl.Father is Guid gf && persisted.Contains(gf) ? sl.Father : null;
var m = sl.Mother is Guid gm && persisted.Contains(gm) ? sl.Mother : null;
if (sl.Father is not null && f is null) litterParentFksDropped++;
if (sl.Mother is not null && m is null) litterParentFksDropped++;
if (f is null && m is null)
{
derivedLittersSkipped++; derivedLitters--;
foreach (var gid in synthLitterForGid.Where(kv => kv.Value == sl.Id).Select(kv => kv.Key).ToList())
synthLitterForGid.Remove(gid);
synthLitters.Remove(key);
continue;
}
synthLitters[key] = sl with { Father = f, Mother = m };
}
parentLinksAdded = synthLitterForGid.Count;
// litter id -> (father, mother) gids, across synthesized + Wurfchronik (by name) litters,
// each FK guarded by the persisted set. Used by residency rule (b) below; augmented with
// existing DB litters under execute.
var litterParents = new Dictionary();
foreach (var sl in synthLitters.Values)
litterParents[sl.Id] = (sl.Father, sl.Mother);
foreach (var sl in litters)
if (litterIdMap.TryGetValue(sl.Id, out var lid))
{
Guid? f = createdAnimalByName.TryGetValue(Normalize(StripZucht(sl.SireName)), out var fid) && persisted.Contains(fid) ? fid : null;
Guid? m = createdAnimalByName.TryGetValue(Normalize(StripZucht(sl.DamName)), out var mid) && persisted.Contains(mid) ? mid : null;
litterParents[lid] = (f, m);
}
// PASS 2: stage synthesized litters (parents already guarded above). DO NOT save them
// before the animals — the offspring AND the parent gerbils are created in the loop
// below, so a single SaveChanges at the end lets EF order parents→litters→offspring
// (all FKs are nullable). Saving litters first is exactly what caused the FK fault.
if (execute)
{
// reuse an existing litter with the same parents+date instead of duplicating.
var existingLitterRows = await _db.Litters
.Select(l => new { l.Id, l.FatherId, l.MotherId, l.Date }).ToListAsync();
var litterByParentsDate = new Dictionary();
foreach (var l in existingLitterRows)
{
litterByParentsDate[$"{l.FatherId}|{l.MotherId}|{l.Date:yyyy-MM-dd}"] = l.Id;
litterParents[l.Id] = (l.FatherId, l.MotherId);
}
foreach (var sl in synthLitters.Values.ToList())
{
var reuseKey = $"{sl.Father}|{sl.Mother}|{sl.Date:yyyy-MM-dd}";
if (litterByParentsDate.TryGetValue(reuseKey, out var existingId))
{
// remap offspring to the existing litter; don't create a duplicate.
foreach (var g in synthLitterForGid.Where(kv => kv.Value == sl.Id).Select(kv => kv.Key).ToList())
synthLitterForGid[g] = existingId;
derivedLitters--;
continue;
}
_db.Litters.Add(new Litter
{
Id = sl.Id,
Name = $"Wurf (aus Diagramm) {sl.Date:yyyy-MM-dd}".Trim(),
Date = sl.Date ?? default,
FatherId = sl.Father,
MotherId = sl.Mother,
Notes = $"aus Stammbaum-Diagramm abgeleitet (Konfidenz: {sl.Confidence})",
});
}
// NOTE: no SaveChanges here — staged with the gerbils below.
}
// OWNERSHIP/RESIDENCY (runs AFTER litter links exist): (a) Zuchtname matches the Clan
// kennel (zuchtCanon contains 'kleinechaote'); (b) parent of a Clan offspring, even if
// the parent's own Zuchtname is foreign. See hive/agents/god/OWNERSHIP-residency.md.
static bool ClanCanon(string? zc) =>
(zc ?? "").Contains("kleinechaote", StringComparison.OrdinalIgnoreCase);
var resident = new HashSet();
foreach (var p in plan) if (ClanCanon(p.A.ZuchtCanon)) resident.Add(p.Gid); // (a)
int residentByA = resident.Count, flippedByParentRule = 0;
foreach (var p in plan)
{
if (!ClanCanon(p.A.ZuchtCanon)) continue;
var litId = p.WurfLitterId ?? (synthLitterForGid.TryGetValue(p.Gid, out var s) ? s : (Guid?)null);
if (litId is null || !litterParents.TryGetValue(litId.Value, out var par)) continue;
if (par.F is Guid gf && resident.Add(gf)) flippedByParentRule++; // (b)
if (par.M is Guid gm && resident.Add(gm)) flippedByParentRule++;
}
int residentTotal = plan.Count(p => resident.Contains(p.Gid));
int externalTotal = plan.Count - residentTotal;
foreach (var p in plan)
{
Guid? litterId = p.WurfLitterId
?? (synthLitterForGid.TryGetValue(p.Gid, out var slid) ? slid : (Guid?)null);
if (litterId is not null) linked++;
bool isResident = resident.Contains(p.Gid);
if (p.ColorVarietyId is null) fbUnmatched++; else fbMatched++;
if (samples.Count < 16)
samples.Add($"Tier: {p.A.Name} (*{p.A.Dob}), Genotyp {ComposeGenotype(p.A.Genotype)}"
+ (litterId is not null ? (p.WurfLitterId is not null ? ", Wurf-verknüpft" : ", Eltern aus Diagramm") : "")
+ (p.ColorVarietyId is not null ? $", Farbschlag „{p.A.Farbschlag}\"" : ""));
if (p.Exists)
{
animalsExisting++;
// sweep idempotency: re-link a now-linkable animal + refresh its residency.
if (execute)
{
var row = await _db.Gerbils.FirstOrDefaultAsync(g => g.Id == p.Gid);
if (row is not null)
{
if (p.CurrentLitterId is null && litterId is not null) row.LitterId = litterId;
row.IsResident = isResident;
}
}
continue;
}
animalsCreated++;
if (execute)
{
_db.Gerbils.Add(new Gerbil
{
Id = p.Gid,
Name = p.A.Name,
Gender = p.Gender,
Status = GerbilStatus.Active,
DateOfBirth = ParseDate(p.A.Dob),
DateOfDeath = ParseDate(p.A.Death),
LitterId = litterId,
ColorVarietyId = p.ColorVarietyId,
Genotype = ComposeGenotype(p.A.Genotype),
IsDeaf = p.A.Deaf,
IsResident = isResident,
ImportSource = ImportSourceTag,
ExternalRef = p.A.Id,
OriginBreeder = string.IsNullOrWhiteSpace(p.A.Zucht) ? null : p.A.Zucht.Trim(),
RawImportData = JsonSerializer.Serialize(new
{
p.A.Genotype.RawGenotype,
p.A.Genotype.UnmappedTokens,
// GEN-3b: Sls (2nd spotting locus) preserved here until Kevin's GEN-3a
// parser adopts it into the compact Genotype contract; tags + deaf too.
Sls = p.A.Genotype.Mapped8locus.TryGetValue("Sls", out var sls) ? sls : null,
p.A.Tags,
p.A.Deaf,
p.A.Zucht,
p.A.SourceFiles,
FarbschlagRaw = p.A.Farbschlag,
}),
});
}
// photos
foreach (var rel in p.A.Photos)
{
var src = Path.Combine(_sourceDir, rel.Replace('/', Path.DirectorySeparatorChar));
if (!File.Exists(src)) { photosMissing++; continue; }
photosAttached++;
if (execute)
{
Directory.CreateDirectory(_photoRoot);
var fileName = $"{Guid.NewGuid():N}{Path.GetExtension(src)}";
File.Copy(src, Path.Combine(_photoRoot, fileName), overwrite: true);
_db.GerbilPhotos.Add(new GerbilPhoto
{
Id = Guid.NewGuid(),
GerbilId = p.Gid,
FileName = fileName,
SortOrder = 0,
CreatedAt = DateTimeOffset.UtcNow,
});
}
}
}
if (execute) await _db.SaveChangesAsync();
// ---- back-link Wurfchronik litter parents by name (best effort) ----
if (execute)
{
foreach (var sl in litters)
{
if (!litterIdMap.TryGetValue(sl.Id, out var lid)) continue;
var litter = await _db.Litters.FirstOrDefaultAsync(l => l.Id == lid);
if (litter is null) continue;
// guard: only link parents that are actually persisted (avoid an orphan FK).
if (createdAnimalByName.TryGetValue(Normalize(StripZucht(sl.SireName)), out var fId) && persisted.Contains(fId))
litter.FatherId = fId;
if (createdAnimalByName.TryGetValue(Normalize(StripZucht(sl.DamName)), out var mId) && persisted.Contains(mId))
litter.MotherId = mId;
}
await _db.SaveChangesAsync();
}
notes.Add("Quarantäne (kein Import): Konflikte + Stubs ohne Geburtsdatum + unsichere Wurf-Zuordnungen — warten auf die Prüfung durch die Züchterin.");
if (parentLinksAdded > 0)
notes.Add($"Stammbaum-Diagramm: {parentLinksAdded} Tiere über Eltern-Verknüpfung einem (abgeleiteten) Wurf zugeordnet ({derivedLitters} abgeleitete Würfe).");
notes.Add($"FK-Integrität: {litterParentFksDropped} Eltern-Verknüpfung(en) verworfen (Elternteil nicht ladbar), {derivedLittersSkipped} abgeleitete Würfe übersprungen (kein ladbares Elternteil). Bei 0/0 ist /import/execute FK-sicher.");
notes.Add($"Bestand/Herkunft: {residentTotal} im Bestand (Clan Kleine Chaoten), {externalTotal} externe Ahnen ({flippedByParentRule} davon über die Eltern-Regel als Bestand erkannt).");
int conflictsResolvedByDecision = loadable.Count(a => a.ResolvedByDecision);
if (conflictsResolvedByDecision > 0)
notes.Add($"Konfliktauflösungen: {conflictsResolvedByDecision} Tier(e) anhand von conflict-decisions.json un-quarantänet (Genotyp/Farbschlag der Züchterin ist maßgeblich).");
if (!execute) notes.Add("DRY-RUN: nichts gespeichert. /import/execute lädt die konfliktfreien Daten.");
return new ImportReport(
Executed: execute,
Litters: new LitterSummary(litters.Count, littersCreated, littersExisting, derivedLitters, derivedLittersSkipped, litterParentFksDropped),
Animals: new AnimalSummary(
animals.Count, animalsCreated, linked, fbMatched, fbUnmatched, animalsExisting,
new QuarantineSummary(conflicts, stubs, dateOnly, ambiguous, conflicts + stubs),
parentLinksAdded, conflictsResolvedByDecision),
Photos: new PhotoSummary(photosAttached, photosMissing),
Samples: samples,
Notes: notes,
Residency: new ResidencySummary(residentTotal, externalTotal, flippedByParentRule));
}
private T? Load(string file)
{
var path = Path.Combine(_sourceDir, file);
if (!File.Exists(path)) return default;
using var fs = File.OpenRead(path);
return JsonSerializer.Deserialize(fs, Json);
}
public static string ComposeGenotype(SourceGenotype g)
{
var tokens = LocusOrder.Select(locus =>
{
if (g.Mapped8locus.TryGetValue(locus, out var pair) && pair.Count == 2)
return StripCaret(pair[0]) + StripCaret(pair[1]);
return "??";
}).ToList();
// GEN-3a contract (Kevin): Sls is appended LAST and ONLY for carriers — the
// wild-type sl/sl is omitted so existing 8-locus strings stay unchanged. WP het
// renders as the trailing token "Slsl"; S(l)S(l) is lethal so never appears.
if (g.Mapped8locus.TryGetValue("Sls", out var sls) && sls.Count == 2
&& !(sls[0] == "sl" && sls[1] == "sl"))
{
tokens.Add(StripCaret(sls[0]) + StripCaret(sls[1]));
}
return string.Join(' ', tokens);
}
private static string StripCaret(string allele) => allele.Replace("^", "");
private static Gender InferGender(SourceAnimal a, HashSet sires, HashSet dams)
{
// Box colour (blue=male, white=female) is the authoritative breeder signal — prefer it
// over sire/dam name inference (PEDIGREE-LINK, Julian 2026-06-06).
if (string.Equals(a.Gender, "male", StringComparison.OrdinalIgnoreCase)) return Gender.male;
if (string.Equals(a.Gender, "female", StringComparison.OrdinalIgnoreCase)) return Gender.female;
var n = Normalize(StripZucht(a.Name));
bool isSire = sires.Contains(n), isDam = dams.Contains(n);
if (isSire && !isDam) return Gender.male;
if (isDam && !isSire) return Gender.female;
return Gender.unknown;
}
public static DateOnly? ParseDate(string s)
{
if (string.IsNullOrWhiteSpace(s)) return null;
var m = Regex.Match(s, @"(\d{1,2})\.(\d{1,2})\.(\d{2,4})");
if (!m.Success) return null;
int d = int.Parse(m.Groups[1].Value), mo = int.Parse(m.Groups[2].Value);
int y = int.Parse(m.Groups[3].Value);
if (y < 100) y += 2000;
try { return new DateOnly(y, mo, d); }
catch { return null; }
}
private static string StripZucht(string name)
{
var n = Regex.Replace(name ?? "", @"\[.*?\]", " ");
n = Regex.Replace(n, @"\s+(?:of|von\s+den|von\s+der|v\.\s?d\.|von)\s+.*$", "", RegexOptions.IgnoreCase);
return n.Trim();
}
private static string Normalize(string name)
{
var n = (name ?? "").ToLowerInvariant();
n = Regex.Replace(n, @"\bgen\.\b", " ");
n = Regex.Replace(n, @"[^a-z0-9äöüß]", "");
return n;
}
/// Dedup identity for parent resolution: normalized call-name + DOB.
private static string NameDobKey(string name, DateOnly? dob) =>
$"{Normalize(StripZucht(name))}|{dob:yyyy-MM-dd}";
/// Resolve a chart-position parentRef (by role) to a known gerbil id, or null.
private static Guid? ResolveParentGid(SourceAnimal a, string role, Dictionary gidByNameDob)
{
var pr = a.ParentRefs.FirstOrDefault(p =>
string.Equals(p.RoleGuess, role, StringComparison.OrdinalIgnoreCase));
if (pr is null || string.IsNullOrWhiteSpace(pr.Name)) return null;
return gidByNameDob.TryGetValue(NameDobKey(pr.Name, ParseDate(pr.Dob)), out var id) ? id : null;
}
/// Per-animal plan computed before any write so synthesis can run in dry-run too.
private sealed record AnimalPlan(
SourceAnimal A, Guid Gid, bool Exists, Guid? WurfLitterId,
Guid? CurrentLitterId, Guid? ColorVarietyId, Gender Gender);
/// A litter synthesized from chart-position parentRefs (PEDIGREE-LINK).
private sealed record SynthLitter(Guid Id, Guid? Father, Guid? Mother, DateOnly? Date, string Confidence);
}
}