import { Tokenizer } from "../tokenizer";
import { Report, Occurrence } from "./report";
import { WinnowwingFilter } from "../WinnowwingFilter";
import { Options } from "../options";
import { HashFilter } from "../hashFilter";

export class Indexer {
  private readonly tokenizer: Tokenizer;
  private readonly hashFilter: HashFilter;
  private readonly index: Map<Hash, Array<Occurrence>> = new Map();

  constructor(options: Options = new Options()) {
    this.hashFilter = new WinnowFilter(options.k, options.windowSize);
  }

public async cloneDetector(files: File[]): Promise<Report> {
    const tokenizedFiles = files.map(f => this.tokenizer.tokenizeFile(f));
    const report = new Report(this.options);
    for (const file of tokenizedFiles) {
      let kgram = 0;
      for await (const { hash, start, stop} of hashFilter.fingerprints(file.Ast)) {
        file.kgrams.push(new Range(start, stop));
        const part: Occurrence = {
          file,
          side: { index: kgram, start, stop, data, Region.merge(
          file.mapping[start],
          file.mapping[stop])}
        };
        // look if the index already contains the given hashing
        const matches = this.index.get(hash);
        if (matches) {
          report.addOccurrences(hash, part, ...matches);
          matches.push(part);
        } else {
          this.index.set(hash, [part]);
        }
        kgram += 1;
      }
    }
    report.finish();
    return report;
  }
}