[Share] Validate Metadata

[matebitte]

Description

Sometimes Zotero won't grab things perfectly; sometimes there are weird pages, inconsistencies, etc. Occasionally you import stuff from a messy colleague. Easy to miss stuff like:

• missing author for print stuff
• malformatted DOI
• a URL that's just a DOI with https://doi.org/ prepended
• a URL not starting with http(s)://
• unplausible dates
• empty dates
• non numeric pages or weird page ranges
• PDFs without OCR

This script will not change these things, but only add a tag like "⚠️ check pages" so you can filter the stuff you want to review. The prepended emoji is also visible in the cite dialogue and thereby prevents most fuckups.

This script will change m/n-dashes in page ranges to hyphens because I think everything else is stupid.

Event

None

Operation

Script

Data

/**
 * A description of this script.
 * @author matebitte & livia
 * @usage 
 * @link https://github.com/windingwind/zotero-actions-tags/discussions/638
 * @see https://github.com/windingwind/zotero-actions-tags/discussions/
 */

// Only regular (metadata) items: skips notes, standalone PDFs/attachments, annotations
if (!item || !item.isRegularItem()) return;

var errors = [];
var tagsBefore = item.getTags().length;

function check(label, fn) {
  try { fn(); } catch (e) { errors.push(label + ": " + e.message); }
}

check("Creators", () => {
  // I only check this for typical print stuff, your preference may vary
  var PRINT_TYPES = ["book", "bookSection", "journalArticle", "magazineArticle", "newspaperArticle", "conferencePaper", "report", "thesis"];
  
  var type = Zotero.ItemTypes.getName(item.itemTypeID);
  if (PRINT_TYPES.includes(type) && item.getCreators().length === 0) {
    item.addTag("⚠️ check author");
  }
});

check("DOI", () => {
  var doi = item.getField("DOI");
  if (doi && !/^10\.\d{4,9}\/\S+$/.test(doi)) item.addTag("⚠️ check DOI");
});

check("URL", () => {
  var url = item.getField("url");
  if (!url) return;
  if (/doi\.org\//i.test(url)) item.addTag("⚠️ check URL");
  if (!/^https?:\/\//i.test(url)) item.addTag("⚠️ check URL");
});

check("Date", () => {
  // non-print stuff I work with is usually newer than 1950s
  // HOWEVER, im not a historian. If you are, you shall adjust "MIN_YEAR"!
  var NON_PRINT = [
    "webpage", "blogPost", "forumPost", "podcast", "instantMessage", "email",
    "computerProgram","dataset",
	"videoRecording", "audioRecording", "radioBroadcast", "film", "tvBroadcast"
  ];
  var MIN_YEAR = 1950;
  
  var date = item.getField("date");
  if (!date) {
    item.addTag("⚠️ check date");
    return;
  }
  var type = Zotero.ItemTypes.getName(item.itemTypeID);
  if (NON_PRINT.includes(type)) {
    var year = parseInt(Zotero.Date.strToDate(date).year, 10);
    if (!year || year < MIN_YEAR) item.addTag("⚠️ check date");
  }
});

check("Pages", () => {
  var pages = item.getField("pages").trim();
  if (!pages) return;

  // Article-number range, e.g. "210:1-210:27" (ACM-style)
  var a = pages.match(/^(\d+):(\d+)\s*[-–—]\s*(\d+):(\d+)$/);
  if (a) {
    var norm = a[1] + ":" + a[2] + "-" + a[3] + ":" + a[4];
    if (norm !== pages) { item.setField("pages", norm); pagesDirty = true; }
    // flag if article numbers differ or end page < start page
    if (a[1] !== a[3] || parseInt(a[4], 10) < parseInt(a[2], 10)) {
      item.addTag("⚠️ check pages");
    }
    return;
  }

  // Plain numeric range, e.g. "210-227"
  var m = pages.match(/^(\d+)\s*[-–—]\s*(\d+)$/);
  if (m) {
    var norm2 = m[1] + "-" + m[2];
    if (norm2 !== pages) { item.setField("pages", norm2); pagesDirty = true; }
    if (parseInt(m[2], 10) < parseInt(m[1], 10)) item.addTag("⚠️ check pages");
  } else if (!/^\d+$/.test(pages)) {
    item.addTag("⚠️ check pages");
  }
});

// OCR check (async, so not inside check())
try {
  for (let att of Zotero.Items.get(item.getAttachments())) {
    if (!att.isPDFAttachment()) continue;
    let text = await att.attachmentText;
    if (!text || text.trim().length < 50) {
      item.addTag("⚠️ check OCR");
      break;
    }
  }
} catch (e) {
  errors.push("OCR: " + e.message);
}

// Persist tag changes
if (item.getTags().length !== tagsBefore) {
  await item.saveTx();
}

if (errors.length) {
  return item.getField("title") + ": " + errors.join("; ");
}

Anything else

Heads up for historians and ENIAC fans: I rarely work with item types like "film" or "computerProgram" dated before 1950. You can change the variables "NON_PRINT" and "MIN_YEAR" if you do (note that some sites like https://media.ccc.de/ notoriously mess up the date, have some kind of check in place if thats your source). If your type of OCD wants a creator for blog posts, you can expand the media in "PRINT_TYPES" accordingly.