publication-ids/dist/index.js

Javascript / Typescript validator and parse for publication ids; DOI, PMID, PMCID, ISBN, and ISSN

// src/doi/regex.ts
var REGEX_DOI = /(10.\d{4,9}\/[-._;()/:A-Z0-9]+)/gi;
var regex_default = REGEX_DOI;

// src/doi/validate.ts
var validate_default = (identifier) => {
  if (!identifier) {
    return false;
  }
  const doiRegex = new RegExp(`^${regex_default.source}$`, "gi");
  return doiRegex.test(identifier);
};

// src/isbn/convert.ts
var calculateIsbn10Checksum = (isbn) => {
  let sum = 0;
  for (let i = 0;i < 9; i++) {
    sum += (i + 1) * parseInt(isbn[i], 10);
  }
  const remainder = sum % 11;
  return remainder === 10 ? "X" : remainder.toString();
};
var calculateIsbn13Checksum = (isbn) => {
  let sum = 0;
  for (let i = 0;i < 12; i++) {
    sum += (i % 2 === 0 ? 1 : 3) * parseInt(isbn[i], 10);
  }
  const remainder = sum % 10;
  return ((10 - remainder) % 10).toString();
};
var convertIsbn13ToIsbn10 = (isbn13) => {
  const isbn10Base = isbn13.substring(3, 12);
  const checksum = calculateIsbn10Checksum(isbn10Base);
  return isbn10Base + checksum;
};
var convertIsbn10ToIsbn13 = (isbn10) => {
  const isbn13Base = "978" + isbn10.substring(0, 9);
  const checksum = calculateIsbn13Checksum(isbn13Base);
  return isbn13Base + checksum;
};

// src/isbn/regex.ts
var REGEX_ISBN = /^(?:ISBN(?:-1[03])?:? )?(?=[0-9X]{10}$|(?=(?:[0-9]+[- ]){3})[- 0-9X]{13}$|97[89][0-9]{10}$|(?=(?:[0-9]+[- ]){4})[- 0-9]{17}$)(?:97[89][- ]?)?[0-9]{1,5}[- ]?[0-9]+[- ]?[0-9]+[- ]?[0-9X]$/;
var regex_default2 = REGEX_ISBN;

// src/isbn/sanitize.ts
var isbnRegex = new RegExp(regex_default2);
var sanitize_default = (input) => {
  if (!input) {
    return [];
  }
  const inputArray = Array.isArray(input) ? input : [...input.split(" "), input];
  const cleanIsbns = inputArray.reduce((acc, inputString) => {
    const cleanInput = inputString.replace(/-/g, "");
    const matches = cleanInput.trim().match(isbnRegex)?.filter(Boolean);
    if (matches) {
      matches.forEach((candidate) => {
        acc.push(candidate.replace(/[-\s]|^ISBN(?:-1[03])?:?/g, ""));
      });
    }
    return acc;
  }, []);
  return [...new Set(cleanIsbns)];
};

// src/isbn/validate.ts
var validate_default2 = (identifier) => {
  if (!identifier) {
    return false;
  }
  const sanitized = sanitize_default(identifier);
  if (!sanitized.length) {
    return false;
  }
  const chars = sanitized[0].split("");
  const last = chars.pop();
  let sum = 0;
  let check;
  if (chars.length === 9) {
    for (let i = 0;i < chars.length; i++) {
      sum += (i + 1) * parseInt(chars[i], 10);
    }
    check = sum % 11;
    check = check === 10 ? "X" : check.toString();
  } else {
    for (let i = 0;i < chars.length; i++) {
      sum += (i % 2 === 0 ? 1 : 3) * parseInt(chars[i], 10);
    }
    check = (10 - sum % 10) % 10;
    check = check.toString();
  }
  return check === last;
};

// src/isbn/parse.ts
var parse_default = (source) => {
  const sanitizedIsbns = sanitize_default(source);
  if (!sanitizedIsbns.length) {
    return [{
      source,
      isValid: false
    }];
  }
  return sanitizedIsbns.map((sanitizedIsbn) => {
    const isValid = validate_default2(sanitizedIsbn);
    return {
      source,
      isValid,
      ...isValid && {
        isbn10: sanitizedIsbn.length === 13 ? convertIsbn13ToIsbn10(sanitizedIsbn) : sanitizedIsbn,
        isbn13: sanitizedIsbn.length === 10 ? convertIsbn10ToIsbn13(sanitizedIsbn) : sanitizedIsbn
      }
    };
  });
};

// src/doi/sanitize.ts
var sanitize_default2 = (input) => {
  if (!input) {
    return [];
  }
  const doiRegex = new RegExp(regex_default, "gi");
  const inputString = Array.isArray(input) ? input.join(" ") : input;
  const decoded = decodeURIComponent(inputString);
  const matches = decoded.match(doiRegex)?.filter(Boolean);
  if (!matches) {
    return [];
  }
  const badEndings = [
    /v.*\.pdf/,
    /\/asset\/.*/,
    /\.$/,
    ".full.pdf",
    ".full.html",
    ".full.htm",
    ".full.txt",
    ".pdf",
    ".html",
    ".htm",
    ".txt",
    ".full",
    "/full",
    "/html",
    "/abstract",
    "/full/html",
    "/html/full",
    "/pdf",
    "/endnote",
    "/reference",
    "/epub",
    "/text",
    "/bibtext"
  ];
  const cleanDois = matches.map((candidate) => {
    let cleanDoi = candidate.toLowerCase().trim();
    badEndings.forEach((ending) => {
      if (ending instanceof RegExp) {
        cleanDoi = cleanDoi.replace(ending, "");
      } else if (cleanDoi.endsWith(ending)) {
        cleanDoi = cleanDoi.slice(0, -ending.length);
      }
    });
    return cleanDoi;
  });
  return [...new Set(cleanDois)];
};

// src/doi/parse.ts
var ISBN13_REGEX = /97[89][0-9]{10}/;
var parse_default2 = (source) => {
  const sanitizedDois = sanitize_default2(source);
  if (!sanitizedDois.length) {
    return [{
      source,
      doi: undefined,
      isValid: false
    }];
  }
  return sanitizedDois.map((sanitizedDoi) => {
    const isValid = validate_default(sanitizedDoi);
    const potentialIsbnFragment = sanitizedDoi.replace(/-/g, "");
    const potentialIsbn = ISBN13_REGEX.exec(potentialIsbnFragment)?.[0] ?? "";
    const isbn = potentialIsbn ? parse_default(potentialIsbn)[0] : { isValid: false };
    if (isbn.isValid) {
      const [doiWithoutChapter, chapter] = sanitizedDoi.split(/[-_]/).reduce((acc, part, index, array) => {
        if (index === array.length - 1 && /^\d+$/.test(part) && /10\.\d{4,9}\/[-._;()/:A-Z0-9]+$/.test(acc[0])) {
          acc[1] = part;
        } else {
          acc[0] += (acc[0] ? "-" : "") + part;
        }
        return acc;
      }, ["", undefined]);
      const resolve2 = isValid ? `https://doi.org/${doiWithoutChapter}` : "";
      return {
        source,
        doi: doiWithoutChapter,
        isValid,
        resolve: resolve2,
        isbn: {
          ...isbn,
          chapter
        }
      };
    }
    const resolve = isValid ? `https://doi.org/${sanitizedDoi}` : "";
    return {
      source,
      doi: sanitizedDoi,
      isValid,
      resolve,
      isbn
    };
  });
};

// src/issn/regex.ts
var REGEX_ISSN = /[\d]{4}[-\s]?[\d]{3}[\dX]/;
var regex_default3 = REGEX_ISSN;

// src/issn/sanitize.ts
var issnRegex = new RegExp(regex_default3, "gi");
var sanitize_default3 = (input) => {
  if (!input) {
    return [];
  }
  const inputArray = Array.isArray(input) ? input : [...input.split(" "), input];
  const cleanIssns = inputArray.reduce((acc, inputString) => {
    const matches = inputString.trim().match(issnRegex)?.filter(Boolean);
    if (matches) {
      matches.forEach((candidate) => {
        acc.push(candidate.replace(/[-\s]|^ISSN:?/g, ""));
      });
    }
    return acc;
  }, []);
  return [...new Set(cleanIssns)];
};

// src/issn/validate.ts
var validate_default3 = (identifier) => {
  if (!identifier) {
    return false;
  }
  if (!regex_default3.test(identifier)) {
    return false;
  }
  const chars = sanitize_default3(identifier)[0].split("");
  const last = chars.pop();
  let sum = 0;
  for (let i = 0;i < chars.length; i++) {
    sum += parseInt(chars[i], 10) * (8 - i);
  }
  const remainder = sum % 11;
  const checkDigit = remainder === 0 ? "0" : 11 - remainder === 10 ? "X" : (11 - remainder).toString();
  return checkDigit === last;
};

// src/issn/parse.ts
var parse_default3 = (source) => {
  const sanitizedIssns = sanitize_default3(source);
  if (!sanitizedIssns.length) {
    return [{
      source,
      isValid: false
    }];
  }
  return sanitizedIssns.map((sanitizedIssn) => {
    const isValid = validate_default3(sanitizedIssn);
    return {
      source,
      isValid,
      ...isValid && {
        issn: sanitizedIssn
      }
    };
  });
};

// src/pmid/regex.ts
var REGEX_PMID = /\b(?:PMID:)?\s?\d{1,8}/;
var REGEX_PMCID = /(?:PMC)\s?\d{1,8}/;

// src/pmid/sanitize.ts
var sanitize_default4 = (input) => {
  if (!input) {
    return [];
  }
  const inputString = Array.isArray(input) ? input.join(" ") : input;
  const decoded = decodeURIComponent(inputString);
  const pmidMatches = decoded.match(new RegExp(REGEX_PMID, "gi")) ?? [];
  const pmcidMatches = decoded.match(new RegExp(REGEX_PMCID, "gi")) ?? [];
  const matches = [...pmidMatches, ...pmcidMatches].filter(Boolean);
  if (!matches.length) {
    return [];
  }
  const cleanPMIDs = matches.map((candidate) => candidate.toString().toUpperCase().trim());
  return [...new Set(cleanPMIDs)];
};

// src/pmid/validate.ts
var validate_default4 = (identifier) => {
  if (!identifier) {
    return false;
  }
  const pmidRegex = new RegExp(`^${REGEX_PMID.source}$`, "gi");
  const pmcidRegex = new RegExp(`^${REGEX_PMCID.source}$`, "gi");
  return pmidRegex.test(identifier) || pmcidRegex.test(identifier);
};

// src/pmid/parse.ts
var parse_default4 = (source) => {
  const sanitizedPMIDs = sanitize_default4(source);
  if (!sanitizedPMIDs.length) {
    return [{
      source,
      pmid: undefined,
      pmcid: undefined,
      isValid: false
    }];
  }
  return sanitizedPMIDs.map((sanitizedId) => {
    const isValid = validate_default4(sanitizedId);
    const isPMCID = sanitizedId.startsWith("PMC");
    return {
      source,
      isValid,
      ...isPMCID ? {
        pmcid: sanitizedId,
        pmid: undefined,
        resolve: `https://www.ncbi.nlm.nih.gov/pmc/articles/${sanitizedId}/`
      } : {
        pmid: sanitizedId,
        pmcid: undefined,
        resolve: `https://pubmed.ncbi.nlm.nih.gov/${sanitizedId.replace("PMID: ", "")}/`
      }
    };
  });
};

// src/global/parse.ts
var parse_default5 = (input) => {
  const inputArray = Array.isArray(input) ? input : input.split(" ");
  return inputArray.map((id) => {
    const results = [];
    if (validate_default(id)) {
      results.push(...parse_default2(id));
    }
    if (validate_default2(id)) {
      results.push(...parse_default(id));
    }
    if (validate_default3(id)) {
      results.push(...parse_default3(id));
    }
    if (validate_default4(id)) {
      results.push(...parse_default4(id));
    }
    if (results.length === 0) {
      results.push({
        source: id,
        isValid: false
      });
    }
    return results;
  }).flat();
};

// src/index.ts
var src_default = {
  parseDoi: parse_default2,
  sanitizeDoi: sanitize_default2,
  validateDoi: validate_default,
  parseIsbn: parse_default,
  sanitizeIsbn: sanitize_default,
  validateIsbn: validate_default2,
  parseIssn: parse_default3,
  sanitizeIssn: sanitize_default3,
  validateIssn: validate_default3,
  parsePmid: parse_default4,
  sanitizePmid: sanitize_default4,
  validatePmid: validate_default4,
  parse: parse_default5
};
export {
  validate_default4 as validatePmid,
  validate_default3 as validateIssn,
  validate_default2 as validateIsbn,
  validate_default as validateDoi,
  sanitize_default4 as sanitizePmid,
  sanitize_default3 as sanitizeIssn,
  sanitize_default as sanitizeIsbn,
  sanitize_default2 as sanitizeDoi,
  parse_default4 as parsePmid,
  parse_default3 as parseIssn,
  parse_default as parseIsbn,
  parse_default2 as parseDoi,
  parse_default5 as parse,
  src_default as default
};

export { parse_default5 as parse_default };