Skip to content

Arabic spelling search test

Does your site search find a word however it is spelled: ة or ه, أ or ا, ى or ي?

  • Checks your page
  • Reads:Site search
  • 1 rule
  • Last updated

Reads: Site search

Free, no sign-up.

What we keep

We keep the address exactly as you send it, including anything after a “?”, and its report shows it. Anyone who has the report’s link can open it, so leave out tokens, keys and personal data.

We send the site’s own search at most 12 requests, one at a time with a pause, as ArablyzerBot, after reading its robots.txt. They show in the site’s logs and may count in its search statistics.

The result appears here

Enter your page’s address and press “Check the page”. This is what the tool runs on it:

  • Site search loses Arabic spelling variantsnot started

What it checks

  • That your search finds a word spelled in the other correct way: ة for ه, أ إ آ for ا, ى for ي, with a tatweel or a diacritic added, with Arabic-Indic digits for Latin ones.
  • The result is «your search loses N of M variants», with the words, the variants asked and what each found, against what the word's own spelling found.
  • It asks at most 12 requests, one at a time, as ArablyzerBot, after reading robots.txt: if robots.txt keeps the bot from your search pages, it asks nothing and says so.

Example

WrongHTML
<!-- WHERE title LIKE '%word%': the exact letters only -->
RightHTML
<!-- WHERE normalize(title) LIKE normalize('%word%'): one spelling for all -->

How to fix

Normalize the text you index and the query you receive, with the same function, folding أ إ آ to ا, ى to ي, ة to ه, dropping tatweel and diacritics, and writing digits in one script:

  • Elasticsearch and OpenSearch: the arabic_normalization filter in an arabic analyzer. Meilisearch and Typesense fold most of these by default.
  • WordPress, Salla, Zid and Shopify: look for a search app or plugin that normalizes Arabic; the stores' default search often matches letters as typed.
JavaScript
const normalize = (text) =>
  text
    .replace(/[ً-ٰٟـ]/g, '')
    .replace(/[أإآٱ]/g, 'ا')
    .replace(/ى/g, 'ي')
    .replace(/ة/g, 'ه')
    .replace(/[٠-٩]/g, (d) => String('٠١٢٣٤٥٦٧٨٩'.indexOf(d)))

FAQ

At most 12, one at a time with a pause, from ArablyzerBot, which says so in its user agent. It reads your robots.txt first, and asks nothing if it keeps the bot from your search pages. It never submits any other form.

From the page: a form with a GET method marked as a search (role="search") or with an input[type=search], on your own site. If the page has none, from the platform's own search address, when the platform is one we know (WordPress, Shopify, Salla, Zid). Those patterns are not something your page said, and the result says which way it was found.

Because no search is expected to know how someone spells Arabic in Latin letters («mktba»). It is shown, since many visitors type it, but it is not part of the loss.

A word whose own spelling finds nothing tells nothing of its variants, so it is left out of the count.

How we measure

Last updated Methodology and scoring

We ask a query that means nothing first, to learn the links every answer holds (menu, footer, sidebar). Then we ask up to four words from your page (headings first) and for each its variants. We count, for each answer, the links it holds beyond those, and the first. A variant is lost when it finds none, or fewer than half of what the word's own spelling found. The count is what the first page of the answer shows, so it is a floor. See the rule's page for the details.