cv-manager / apps / server / src / modules / data-import / sources / file-import-source.ts

import {
  CVParserService as CVParser,
  type ExistingUserContext,
  type ParsedCVData,
} from "@cv/ai-parser";
import type { User } from "@cv/auth";
import { PrismaService } from "@cv/system";
import { TextExtractorRegistry, TEXT_EXTRACTOR_REGISTRY } from "@cv/file-upload";
import { Inject, Injectable, Logger } from "@nestjs/common";
import { EducationService } from "@/modules/education/education.service";
import { AIProviderResolverService } from "@/modules/cv-parser/ai-provider-resolver.service";
import { UserJobExperienceService } from "@/modules/job-experience/employment/user-job-experience.service";
import { ProfileService } from "@/modules/profile/profile.service";
import type { DataImportSource } from "../data-import-source.interface";

/**
 * Build ExistingUserContext from domain models for AI prompt enrichment.
 */
const buildExistingUserContext = (
  profile: { fullName?: string | null; headline?: string | null; city?: string | null; country?: string | null } | null,
  jobs: Array<{ company: { name: string }; role: { name: string }; startDate: Date; endDate?: Date }>,
  educations: Array<{ institution: { name: string }; degree: string; startDate: Date; endDate: Date | null }>,
): ExistingUserContext | undefined => {
  const context: ExistingUserContext = {};

  if (profile?.fullName) context.name = profile.fullName;
  if (profile?.headline) context.headline = profile.headline;
  if (profile?.city) context.city = profile.city;
  if (profile?.country) context.country = profile.country;

  if (jobs.length > 0) {
    context.jobs = jobs.map((j) => {
      const entry: { company: string; role: string; startDate: string; endDate?: string } = {
        company: j.company.name,
        role: j.role.name,
        startDate: j.startDate.toISOString().slice(0, 10),
      };
      if (j.endDate) entry.endDate = j.endDate.toISOString().slice(0, 10);
      return entry;
    });
  }

  if (educations.length > 0) {
    context.education = educations.map((e) => {
      const entry: { institution: string; degree: string; startDate: string; endDate?: string } = {
        institution: e.institution.name,
        degree: e.degree,
        startDate: e.startDate.toISOString().slice(0, 10),
      };
      if (e.endDate) entry.endDate = e.endDate.toISOString().slice(0, 10);
      return entry;
    });
  }

  const hasData = Object.keys(context).length > 0;
  return hasData ? context : undefined;
};

/**
 * Wraps the existing file extraction + AI parsing pipeline
 * as a DataImportSource.
 */
@Injectable()
export class FileImportSource implements DataImportSource {
  readonly name = "file-upload";
  private readonly logger = new Logger(FileImportSource.name);

  constructor(
    @Inject(TEXT_EXTRACTOR_REGISTRY)
    private readonly textExtractorRegistry: TextExtractorRegistry,
    private readonly providerResolver: AIProviderResolverService,
    private readonly profileService: ProfileService,
    private readonly jobExperienceService: UserJobExperienceService,
    private readonly educationService: EducationService,
    private readonly prisma: PrismaService,
  ) {}

  async execute(
    user: User,
    params: Record<string, unknown>,
    onStatus: (message: string) => Promise<void>,
  ): Promise<ParsedCVData> {
    const buffer = params["buffer"] as Buffer;
    const mimeType = params["mimeType"] as string;

    await onStatus("Extracting text from file");

    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");
    }

    this.logger.log(`Extracted ${extraction.text.length} chars from ${mimeType}`);
    this.logger.debug(`Extracted text preview: ${extraction.text.slice(0, 500)}`);

    await onStatus("Gathering existing profile data");

    const existingContext = await this.gatherUserContext(user);

    await onStatus("Analyzing with AI");

    const provider = await this.providerResolver.resolveForUser(user);
    this.logger.log(`AI provider: ${provider.constructor.name}`);

    const parser = new CVParser(provider);
    const result = await parser.parseCVText(extraction.text, existingContext);

    return result;
  }

  private async gatherUserContext(user: User): Promise<ExistingUserContext | undefined> {
    try {
      const firstProfile = await this.prisma.profile.findFirst({
        where: { userId: user.id },
        select: { id: true },
      });

      if (!firstProfile) return undefined;

      const [profile, jobs, educationResult] = await Promise.all([
        this.profileService.findByIdOrFail(firstProfile.id),
        this.jobExperienceService.findForProfile(firstProfile.id),
        this.educationService.findManyForProfile(firstProfile.id),
      ]);

      const context = buildExistingUserContext(
        profile,
        jobs,
        educationResult.edges.map((e) => e.node),
      );

      if (context) {
        this.logger.debug(`User context: ${JSON.stringify(context)}`);
      }

      return context;
    } catch (err) {
      this.logger.warn(
        `Failed to gather user context (non-fatal): ${err instanceof Error ? err.message : err}`,
      );
      return undefined;
    }
  }
}