About
Tool No. 59 of Pikkola
An address is read by more software than any other string a person types: a browser, a proxy, a log, a redirect chain, an analytics script. This page reads it the way the browser does, then lets you change one part at a time and hands back the whole.
What it does
- Splits an address into scheme, host, port, path segments, query pairs and fragment, and puts it back together as the fields change.
- Edits the query as a table: add, rename, reorder, sort, switch off or delete a pair, with duplicates kept apart.
- Builds a campaign address from five fields without doubling tags that are already on the base address.
- Removes the parameters advertising platforms append, and names every one it removed.
- Runs the same text through encodeURI, encodeURIComponent and a form body, and shows where they disagree.
- Moves a domain between its readable form and punycode, in both directions.
- Speaks English, Korean and Japanese, follows the light and dark theme, and copies every result.
How an address is read
The six parts come from the browser's own URL parser. That is the point: a host is lower-cased, dot segments in a path are resolved, a default port is dropped and a space is percent-encoded, because that is what the browser would send. A tool with its own parser would be describing a request nobody makes.
The query is read from the raw text instead of through URLSearchParams, for two reasons: that API turns a literal plus into a space, and it merges nothing but still re-serialises the string. The table shows each value percent-decoded once, so a name and a value can be read and edited as text, and the address is rebuilt from them.
What the query table keeps
Order, duplicates and empty values. Each row is one pair, so page=2 and page=3 can sit next to each other and both are sent. Sorting by name keeps two pairs with the same name together rather than merging them. A row can be switched off without being deleted, which is how a parameter is tested.
When the address is put back together, every name and value is encoded once. A value that arrived as %EA%B0%80 is shown as 가 and leaves as %EA%B0%80, not %25EA%25B0%2580. A literal percent sign in a value, which is not a valid escape, is encoded, so it survives the round trip.
What the campaign bench adds
Five fields in the order analytics tools expect: source, medium, campaign, term, content. Any utm_ parameter already on the base address is removed before the new ones are written, so a campaign run twice does not produce utm_source=newsletter&utm_source=newsletter. Other parameters, the path and the fragment stay where they were.
What the cleaner removes
Only parameters that a platform appended, and only from the groups that are switched on. A parameter the site itself uses is in no group and stays. Empty values can be dropped as well, which is useful after a spreadsheet has filled a campaign with blanks.
Two things it cannot know. It cannot tell a tracking parameter from a parameter that happens to share a name, and it cannot know whether a site signs its links. If a link carries a signature over the whole query string, removing anything will invalidate it, and the site will say so more clearly than this page can.
How the three encoders differ
encodeURI leaves the characters that separate the parts of a URL alone, so it is the one for a whole address. encodeURIComponent encodes those separators, so it is the one for a single value that has to sit inside a query. A form body is the third: it encodes like a value, and it writes a space as a plus sign, because that is what application/x-www-form-urlencoded means.
The bench lists the characters where the three disagree for the current input, and a small table shows the usual suspects: space, plus, ampersand, equals, question mark, hash, slash, percent, and one Korean character.
Where the limits are
- Encoding a domain to punycode uses the browser, so it matches what the browser would request. Decoding is done here. The full UTS 46 mapping table is not applied, so an unusual label can read differently to a registrar.
- The tool does not check that a host exists, that a path is served, or that a cleaned link still opens. It builds the address, and the site decides the rest.
- A very long address is rendered and copied, but a browser address bar and some mail clients will still refuse it.
- There is no history. A reload empties the benches, and nothing is written to the server.
Why it is free
The page is a set of static files. There is no account, no quota and no paid tier. It carries the publisher meta tag the site uses for its AdSense verification, and no advertising script is loaded: the content policy allows scripts from this domain only, and the network policy allows no connection at all.
Contact
A question, or an address that this page reads differently from your browser: [email protected].