ReadaBook/apps/api/src/metadata/use-cases/extract-identifiers.ts

import { readFileSync } from "node:fs";
import { basename, extname } from "node:path";
import AdmZip from "adm-zip";
import { XMLParser } from "fast-xml-parser";

export type ExtractedIdentifiers = {
  isbn10: string | null;
  isbn13: string | null;
  candidates: string[];
};

const xmlParser = new XMLParser({
  ignoreAttributes: false,
  attributeNamePrefix: "@_",
  textNodeName: "#text"
});

export class ExtractIdentifiers {
  fromMetadataAndFile(metadata: { isbn?: string | null }, filePath: string): ExtractedIdentifiers {
    const candidates = new Set<string>();
    for (const value of [metadata.isbn, basename(filePath, extname(filePath))]) {
      for (const isbn of findIsbns(String(value ?? ""))) candidates.add(isbn);
    }

    const extension = extname(filePath).toLowerCase();
    if (extension === ".epub") {
      for (const isbn of findIsbns(readLimitedEpubText(filePath))) candidates.add(isbn);
    }
    if (extension === ".pdf") {
      const buffer = readFileSync(filePath);
      const head = buffer.subarray(0, Math.min(buffer.length, 256 * 1024)).toString("latin1");
      for (const isbn of findIsbns(head)) candidates.add(isbn);
    }

    return normalizeCandidates([...candidates]);
  }
}

export function normalizeIsbn(value: string): string | null {
  const compact = value.replace(/[^0-9X]/gi, "").toUpperCase();
  if (compact.length === 10 && isValidIsbn10(compact)) return compact;
  if (compact.length === 13 && isValidIsbn13(compact)) return compact;
  return null;
}

export function toIsbn13(value: string): string | null {
  const isbn = normalizeIsbn(value);
  if (!isbn) return null;
  if (isbn.length === 13) return isbn;
  const stem = `978${isbn.slice(0, 9)}`;
  let sum = 0;
  for (let index = 0; index < stem.length; index += 1) {
    sum += Number(stem[index]) * (index % 2 === 0 ? 1 : 3);
  }
  return `${stem}${(10 - (sum % 10)) % 10}`;
}

function normalizeCandidates(values: string[]): ExtractedIdentifiers {
  const normalized = [...new Set(values.map(normalizeIsbn).filter((value): value is string => Boolean(value)))];
  const isbn13 = normalized.map(toIsbn13).find((value): value is string => Boolean(value)) ?? null;
  const isbn10 = normalized.find((value) => value.length === 10) ?? null;
  return { isbn10, isbn13, candidates: normalized };
}

function findIsbns(text: string): string[] {
  const matches = text.match(/(?:ISBN(?:-1[03])?:?\s*)?(?:97[89][-\s]?)?(?:\d[-\s]?){9,12}[\dX]/gi) ?? [];
  return matches.map((match) => match.replace(/^ISBN(?:-1[03])?:?\s*/i, ""));
}

function isValidIsbn10(value: string): boolean {
  let sum = 0;
  for (let index = 0; index < 10; index += 1) {
    const char = value[index];
    const digit = char === "X" && index === 9 ? 10 : Number(char);
    if (!Number.isInteger(digit)) return false;
    sum += digit * (10 - index);
  }
  return sum % 11 === 0;
}

function isValidIsbn13(value: string): boolean {
  let sum = 0;
  for (let index = 0; index < 13; index += 1) {
    const digit = Number(value[index]);
    if (!Number.isInteger(digit)) return false;
    sum += digit * (index % 2 === 0 ? 1 : 3);
  }
  return sum % 10 === 0;
}

function readLimitedEpubText(filePath: string): string {
  try {
    const zip = new AdmZip(filePath);
    const fragments: string[] = [zip.readAsText("META-INF/container.xml")];
    for (const entry of zip.getEntries()) {
      if (fragments.join("").length > 256 * 1024) break;
      if (!entry.isDirectory && /\.(opf|xhtml|html|htm|xml)$/i.test(entry.entryName)) {
        fragments.push(stripXml(zip.readAsText(entry)));
      }
    }
    return fragments.join("\n");
  } catch {
    return "";
  }
}

function stripXml(value: string): string {
  try {
    const parsed = xmlParser.parse(value);
    return JSON.stringify(parsed).slice(0, 256 * 1024);
  } catch {
    return value.slice(0, 256 * 1024);
  }
}