first commit

apps/api/src/classifier/classifier.service.ts

@@ -0,0 +1,77 @@
+import { Injectable } from '@nestjs/common';
+import { InjectRepository } from '@nestjs/typeorm';
+import { Repository } from 'typeorm';
+import { CatalogItemEntity } from '../entities/catalog-item.entity';
+import { ClassificationTagEntity } from '../entities/classification-tag.entity';
+
+export type ClassificationResult = {
+  publisher?: string;
+  format?: string[];
+  territory?: string[];
+  topics?: string[];
+};
+
+@Injectable()
+export class ClassifierService {
+  constructor(
+    @InjectRepository(ClassificationTagEntity)
+    private readonly tags: Repository<ClassificationTagEntity>,
+  ) {}
+
+  async classify(item: CatalogItemEntity): Promise<ClassificationResult> {
+    const res: ClassificationResult = {
+      publisher: item.publisher || undefined,
+      format: this.splitFormats(item.format),
+    };
+
+    const text = `${item.title || ''} ${item.description || ''}`.toLowerCase();
+
+    const territory: string[] = [];
+    const topics: string[] = [];
+
+    if (text.includes('madrid')) territory.push('Madrid');
+    if (text.includes('andaluc')) territory.push('Andalucía');
+    if (text.includes('catalu')) territory.push('Cataluña');
+
+    if (text.includes('subvenc')) topics.push('Subvenciones');
+    if (text.includes('licit')) topics.push('Licitaciones');
+    if (text.includes('contrat')) topics.push('Contratación');
+    if (text.includes('presupuesto')) topics.push('Presupuestos');
+
+    res.territory = territory.length ? territory : undefined;
+    res.topics = topics.length ? topics : undefined;
+
+    return res;
+  }
+
+  async applyTags(item: CatalogItemEntity, classification: ClassificationResult) {
+    await this.tags.delete({ itemId: item.id });
+    const next: ClassificationTagEntity[] = [];
+
+    if (classification.publisher) {
+      next.push(this.tags.create({ itemId: item.id, type: 'publisher', value: classification.publisher }));
+    }
+
+    for (const format of classification.format || []) {
+      next.push(this.tags.create({ itemId: item.id, type: 'format', value: format }));
+    }
+
+    for (const topic of classification.topics || []) {
+      next.push(this.tags.create({ itemId: item.id, type: 'topic', value: topic }));
+    }
+
+    for (const territory of classification.territory || []) {
+      next.push(this.tags.create({ itemId: item.id, type: 'territory', value: territory }));
+    }
+
+    if (next.length) await this.tags.save(next);
+  }
+
+  private splitFormats(value?: string): string[] {
+    if (!value) return [];
+    return value
+      .split(',')
+      .map((v) => v.trim())
+      .filter(Boolean);
+  }
+}