Assessor_2.0/src/domain/import-intermediate.ts

import type { JsonValue } from './canonical-model';

export type ImportFormat = 'CSV' | 'HTML' | 'Excel' | 'JSON' | 'XML';
export type ImportStatus = 'complete' | 'partial' | 'failed';
export type ImportDiagnosticSeverity = 'info' | 'warning' | 'error';
export type ImportValueType = 'empty' | 'string' | 'number' | 'boolean' | 'date' | 'json';
export type ImportProcessingStage = 'format-decoding' | 'structure-decoding' | 'extraction' | 'mapping' | 'normalization' | 'persistence';
export type ImportProcessingStatus = 'pending' | 'running' | 'complete' | 'partial' | 'failed' | 'skipped';
export type ImportProcessorKind = 'format-adapter' | 'source-extractor' | 'mapping-profile' | 'normalizer' | 'artifact-parser' | 'repository';
export type ImportFormatEvidenceSignal = 'file-signature' | 'container-signature' | 'content-structure' | 'extension' | 'media-type';

export interface ImportFormatEvidence {
  signal: ImportFormatEvidenceSignal;
  format?: ImportFormat;
  contribution: number;
  message: string;
}

export interface ImportFormatCandidate {
  format: ImportFormat;
  score: number;
  evidence: ImportFormatEvidence[];
}

export interface ImportFormatDetectionResult {
  version: 1;
  selected: ImportFormat;
  score: number;
  candidates: ImportFormatCandidate[];
  evidence: ImportFormatEvidence[];
  conflicts: ImportFormatEvidence[];
  structurallyValidated: boolean;
  requiresParserValidation: boolean;
  binary: boolean;
  containerKind?: 'zip' | 'excel-zip' | 'ole';
}

export interface ImportProcessorReference {
  kind: ImportProcessorKind;
  id: string;
  version: string;
}

export interface SourceLocator {
  fileName: string;
  relativePath?: string;
  containerEntryPath?: string;
  tableIndex?: number;
  rowIndex?: number;
  columnIndex?: number;
  cellReference?: string;
  objectPath?: string;
  elementSelector?: string;
}

export interface ImportTransformationRecord {
  id: string;
  version: string;
  kind: 'decode' | 'sanitize' | 'flatten-for-compatibility' | 'extract' | 'map' | 'normalize';
  lossless: boolean;
}

export interface ImportProvenance {
  version: 1;
  source: SourceLocator;
  processor: ImportProcessorReference;
  transformations: ImportTransformationRecord[];
}

export interface ImportProcessingState {
  stage: ImportProcessingStage;
  status: ImportProcessingStatus;
  processor: ImportProcessorReference;
  diagnosticCodes: string[];
}

export interface ImportSourcePosition extends Omit<SourceLocator, 'fileName' | 'tableIndex'> {
  fileName?: string;
  tableIndex: number;
}

export interface ImportFileMetadata {
  name: string;
  sizeBytes: number;
  lastModified: number;
  extension: string;
  mediaType?: string;
  encoding?: string;
}

export interface ImportCell {
  rawValue: JsonValue;
  decodedValue: string;
  detectedType: ImportValueType;
  source: ImportSourcePosition;
}

export interface ImportRow {
  index: number;
  cells: ImportCell[];
  source: ImportSourcePosition;
  /** Der Bericht markiert diese Zeile selbst (z. B. CIS-CAT: `<tr class="evaluated">`). */
  highlighted?: boolean;
}

export interface HeaderCandidate {
  rowIndex: number;
  headers: string[];
  confidence: number;
  reason: string;
}

export interface ImportColumn {
  index: number;
  headerCandidates: Array<{
    rowIndex: number;
    value: string;
    confidence: number;
  }>;
}

export interface ImportTable {
  id: string;
  name: string;
  index: number;
  columns: ImportColumn[];
  rows: ImportRow[];
  rawText: string;
  headerCandidates: HeaderCandidate[];
  attributes: Record<string, JsonValue>;
}

export interface ImportDiagnostic {
  severity: ImportDiagnosticSeverity;
  code: string;
  message: string;
  stage?: ImportProcessingStage;
  source?: SourceLocator;
  provenance?: ImportProvenance;
  details?: JsonValue;
}

export interface DiscardedImportContent {
  kind: string;
  reason: string;
  count?: number;
  source?: SourceLocator;
  details?: JsonValue;
}

export interface DecodedViewBase {
  id: string;
  name: string;
  source: SourceLocator;
  provenance: ImportProvenance;
  attributes: Record<string, JsonValue>;
}

export interface DecodedTableView extends DecodedViewBase {
  kind: 'table';
  tableId: string;
}

export interface DecodedObjectView extends DecodedViewBase {
  kind: 'object';
  rootKind: 'object' | 'array' | 'primitive';
  value: JsonValue;
}

export interface DecodedHierarchyNode {
  path: string;
  name: string;
  value?: JsonValue;
  attributes: Record<string, JsonValue>;
  source: SourceLocator;
}

export interface DecodedHierarchyView extends DecodedViewBase {
  kind: 'hierarchy';
  nodes: DecodedHierarchyNode[];
}

export interface DecodedTextView extends DecodedViewBase {
  kind: 'text';
  text: string;
}

export interface DecodedContainerEntry {
  path: string;
  sizeBytes?: number;
  mediaType?: string;
  encoding?: string;
  source: SourceLocator;
}

export interface DecodedContainerView extends DecodedViewBase {
  kind: 'container';
  entries: DecodedContainerEntry[];
}

export type DecodedSourceView = DecodedTableView | DecodedObjectView | DecodedHierarchyView | DecodedTextView | DecodedContainerView;

export interface DecodedSourceDocument {
  version: 1;
  status: ImportStatus;
  format: ImportFormat;
  file: ImportFileMetadata;
  rawText: string;
  tables: ImportTable[];
  diagnostics: ImportDiagnostic[];
  discardedContent: DiscardedImportContent[];
  views: DecodedSourceView[];
  processing: ImportProcessingState[];
  formatDetection?: ImportFormatDetectionResult;
}

/** Compatibility name used by the current UI and recognition pipeline. */
export type ImportDocument = DecodedSourceDocument;