Slug Generator

What a slug is, and why it matters

A slug is the URL-safe version of a title — the part after the last slash in /blog/my-post-title/. It matters for two separate audiences at once: search engines weight the words in a URL, and a human reading a link in a browser tab or a shared message can tell what the page is about before clicking. A slug generator's whole job is turning arbitrary human text — with capital letters, punctuation, accented characters, sometimes an entirely different script — into something that survives being a URL path segment without losing the words that made the title meaningful.

Why "Straße" becomes "stra-e" on most other tools

The standard recipe every slugify tutorial teaches is: Unicode-normalize to NFD, strip combining marks, then drop anything outside [a-z0-9]. Measured directly: "Straße".normalize("NFD") with marks stripped gives "strae" after the specials filter runs, and the visible result on most competing tools is the broken stra-e. The reason is that ß has no Unicode decomposition at all — it isn't a base letter plus a combining accent, so mark-stripping has nothing to strip and the character just falls through the ASCII filter and disappears. The same recipe reduces "Привет мир" and "你好世界" to nothing but a bare separator, because Cyrillic and CJK characters don't decompose into Latin letters at all. This tool's Universal pack carries a small named exception map for exactly the characters that break the naive recipe (ß, ø, đ, ł, æ, œ, þ and about three dozen more), so the common case works by default, and named packs for Cyrillic and Greek handle those scripts with a cited transliteration standard instead of silence.

Transliteration is a choice, not a fact

There is no single correct answer to "what does ä become in a URL". German software convention expands it to ae; Scandinavian library-catalog convention drops the dot and keeps a; a strict transliteration standard like ISO 9 for Cyrillic maps a character to exactly one Latin character with a diacritic (ж → ž) so the mapping is reversible, while the more familiar "zh" is not. Every one of these is defensible and none of them is universal — two teams using two different libraries will get two different slugs from the same title, and neither is a bug. This tool names the convention it used next to the result instead of presenting a silent default, and its named packs are documented in the file that implements them so the choice is traceable.

NFC vs NFD: two identical-looking inputs, two different slugs

A visually identical accented character can arrive as one composed code point (NFC — what most browsers and text editors produce) or as a base letter plus a separate combining mark (NFD — what macOS's filesystem and some input methods produce). They render the same on screen and compare unequal with ===. A slugifier that doesn't normalize first can turn the same-looking title into two different slugs depending on which form arrived, which is exactly the kind of silent inconsistency non-negotiable #1 forbids. This tool always normalizes to NFC before anything else touches the string, and states so.

Why removing stop words is off by default

Dropping "the", "of", "a" and similar words from a slug is a real, common option — it makes slugs shorter and reads naturally to search engines. It also manufactures collisions no naive implementation warns about: "The Best of Times" and "Best Times" both reduce to best-times, and if your CMS doesn't handle that, the second post silently overwrites the first post's URL or fails to save. This tool defaults the option off, and bulk mode's duplicate detection — automatic -2, -3 suffixing — is what makes turning it on safe for a real batch of titles.

RFC 3986 is necessary but not sufficient

RFC 3986's unreserved character set (A-Za-z0-9-._~) is what a URL path segment is allowed to contain without percent-encoding, but a slug that technically satisfies it can still be a bad slug: an all-digit string like 2024 collides with a numeric ID route such as /posts/123; a bare . or .. is a reserved relative-path segment in every URL and filesystem; CON, PRN and the other Windows device names can break a filesystem-backed host trying to create a matching file. This tool checks all of them and labels each one individually, rather than folding them into one pass/fail verdict that doesn't say which rule failed.

A Unicode slug is legal — and becomes something else downstream

RFC 3987's IRI spec permits non-ASCII characters in a URL, so /blog/größe is a valid, working link that renders correctly in a modern browser's address bar. But it doesn't stay that way everywhere: server access logs, many analytics tools and older HTTP clients record and transmit it as its percent-encoded byte form, /blog/gr%C3%B6%C3%9Fe. Choosing "Keep Unicode" without seeing that encoded form would hide the actual consequence of the choice, so this tool shows both side by side.