cv-manager / apps / server / src / modules / cv-parser / cv-parser.service.ts

import { CVParserService as CVParser, type ParsedCVData } from "@cv/ai-parser";
import type { User } from "@cv/auth";
import {
  TEXT_EXTRACTOR_REGISTRY,
  type TextExtractorRegistry,
  validateFile,
} from "@cv/file-upload";
import { Inject, Injectable } from "@nestjs/common";
import { AIProviderResolverService } from "./ai-provider-resolver.service";
import {
  EntityResolverService,
  type ResolvedEducation,
  type ResolvedJobExperience,
} from "./entity-resolver.service";

/**
 * Parsed CV data with resolved entities
 */
export interface ParsedCVDataWithResolution {
  jobExperiences: ResolvedJobExperience[];
  education: ResolvedEducation[];
}

@Injectable()
export class CVParserService {
  constructor(
    @Inject(TEXT_EXTRACTOR_REGISTRY)
    private readonly textExtractorRegistry: TextExtractorRegistry,
    private readonly entityResolver: EntityResolverService,
    private readonly providerResolver: AIProviderResolverService,
  ) {}

  /** Create a per-request CVParser using the user's resolved provider */
  private async createParser(user: User): Promise<CVParser> {
    const provider = await this.providerResolver.resolveForUser(user);
    return new CVParser(provider);
  }

  async parseFile(
    user: User,
    buffer: Buffer,
    mimeType: string,
    originalName: string,
  ): Promise<ParsedCVData> {
    const validation = validateFile({
      buffer,
      mimeType,
      originalName,
      sizeBytes: buffer.length,
    });

    if (!validation.valid) {
      throw new Error(`File validation failed: ${validation.error}`);
    }

    const extraction = await this.textExtractorRegistry.extract(
      buffer,
      mimeType,
    );

    if (!extraction.success) {
      throw new Error(`Text extraction failed: ${extraction.error}`);
    }

    if (!extraction.text || extraction.text.trim().length === 0) {
      throw new Error("Could not extract any text from the file");
    }

    const parser = await this.createParser(user);
    return parser.parseCVText(extraction.text);
  }

  async parseStory(user: User, storyText: string): Promise<ParsedCVData> {
    if (!storyText || storyText.trim().length === 0) {
      throw new Error("Story text cannot be empty");
    }

    if (storyText.trim().length > 50000) {
      throw new Error("Story text is too long (max 50,000 characters)");
    }

    const parser = await this.createParser(user);
    return parser.parseCVText(storyText);
  }

  /**
   * Parse story text and resolve entities to existing database records
   * Returns draft data with entity IDs where matches were found
   */
  async parseStoryWithResolution(
    user: User,
    storyText: string,
  ): Promise<ParsedCVDataWithResolution> {
    const parsed = await this.parseStory(user, storyText);
    return this.resolveEntities(parsed);
  }

  /**
   * Parse file and resolve entities to existing database records
   */
  async parseFileWithResolution(
    user: User,
    buffer: Buffer,
    mimeType: string,
    originalName: string,
  ): Promise<ParsedCVDataWithResolution> {
    const parsed = await this.parseFile(user, buffer, mimeType, originalName);
    return this.resolveEntities(parsed);
  }

  /**
   * Resolve entity names to existing database records
   */
  private async resolveEntities(
    parsed: ParsedCVData,
  ): Promise<ParsedCVDataWithResolution> {
    const [jobExperiences, education] = await Promise.all([
      this.resolveJobExperiences(parsed.jobExperiences),
      this.resolveEducation(parsed.education),
    ]);

    return { jobExperiences, education };
  }

  /**
   * Resolve job experience entities
   */
  private async resolveJobExperiences(
    jobs: ParsedCVData["jobExperiences"],
  ): Promise<ResolvedJobExperience[]> {
    return Promise.all(
      jobs.map(async (job) => {
        const [company, role, level, skills] = await Promise.all([
          this.entityResolver.resolveCompany(job.companyName),
          this.entityResolver.resolveRole(job.roleName),
          this.entityResolver.resolveLevel(job.levelName),
          this.entityResolver.resolveSkills(job.skills ?? []),
        ]);

        return {
          company,
          role,
          level,
          skills,
          startDate: new Date(job.startDate),
          endDate: job.endDate ? new Date(job.endDate) : null,
          description: job.description ?? null,
        };
      }),
    );
  }

  /**
   * Resolve education entities
   */
  private async resolveEducation(
    education: ParsedCVData["education"],
  ): Promise<ResolvedEducation[]> {
    return Promise.all(
      education.map(async (edu) => {
        const [institution, skills] = await Promise.all([
          this.entityResolver.resolveInstitution(edu.institutionName),
          this.entityResolver.resolveSkills(edu.skills ?? []),
        ]);

        return {
          institution,
          degree: edu.degree,
          fieldOfStudy: edu.fieldOfStudy ?? null,
          skills,
          startDate: new Date(edu.startDate),
          endDate: edu.endDate ? new Date(edu.endDate) : null,
          description: edu.description ?? null,
        };
      }),
    );
  }
}