curl / trurl / Manpage

trurl.1 the manpage

GitHub repo


trurl - transpose URLs


trurl [options / URLs]


trurl parses, manipulates and outputs URLs and parts of URLs.

It uses the RFC 3986 definition of URLs and it uses libcurl's URL parser to do so, which includes a few "extensions". The URL support is limited to "hierarchical" URLs, the ones that use "://" separators after the scheme.

Typically you pass in one or more URLs and decide what of that you want output. Possibly modifying the URL as well.

trurl knows URLs and every URL consists of up to ten separate and independent "components". These components can be extracted, removed and updated with trurl and they are referred to by their respective names: scheme, user, password, options, host, port, path, query, fragment and zoneid.


Options start with one or two dashes. Many of the options require an additional value next to them.

Any other argument is interpreted as a URL argument, and is treated as if it was following a --url option.

The first argument that is exactly two dashes ("--"), marks the end of options; any argument after the end of options is interpreted as a URL argument even if it starts with a dash.

Long options can be provided either as "--flag argument" or as "--flag=argument".

-a, --append [component]=[data]

Append data to a component. This can only append data to the path and the query components.

For path, this URL encodes and appends the new segment to the path, separated with a slash.

For query, this URL encodes and appends the new segment to the query, separated with an ampersand (&). If the appended segment contains an equal sign ('=') that one is kept verbatim and both sides of the first occurrence are URL encoded separately.


When set, trurl tries to accept spaces as part of the URL and instead URL encode such occurrences accordingly.

According to RFC 3986, a space cannot legally be part of a URL. This option provides a best-effort to convert the provided string into a valid URL.


When set, trurl uses the scheme's default port number for URLs with a known scheme, and without an explicit port number.

Note that trurl only knows default port numbers for URL schemes that are supported by libcurl.

Since, by default, trurl removes default port numbers from URLs with a known scheme, this option is pretty much ignored unless one of --get, --json, and --keep-port is not also specified.

-f, --url-file [file name]

Read URLs to work on from the given file. Use the file name "-" (a single minus) to tell trurl to read the URLs from stdin.

Each line needs to be a single valid URL. trurl removes one carriage return character at the end of the line if present, trims off all the trailing space and tab characters, and skips all empty (after trimming) lines.

The maximum line length supported in a file like this is 4094 bytes. Lines that exceed that length are skipped, and a warning is printed to stderr when they are encountered.

-g, --get [format]

Output text and URL data according to the provided format string. Components from the URL can be output when specified as {component} or [component], with the name of the part show within curly braces or brackets. You can not mix braces and brackets for this purpose in the same command line.

The following component names are available (case sensitive): url, scheme, user, password, options, host, port, path, query, fragment and zoneid.

{component} expands to nothing if the given component does not have a value.

Components are shown URL decoded by default. If you instead write the component prefixed with "url:" like "{url:path}", it gets output URL encoded. As a shortcut, "url:" also works written as a single colon: "{:path}".

URL decoding a component may cause problems to display it. Such problems make a warning get displayed unless --quiet is used. URL decode problems can optionally be turned into errors by prefixing the comopnent name with "strict:", as in "{strict:path}". In this stricter mode, a URL decode problem makes trurl stop what it is doing and return with exit code 10.

You may also prefix components with default: and/or puny: or idn:, in any order.

If default: is specified, like "{default:url}" or "{default:port}", and the port is not explicitly specified in the URL, the scheme's default port is output if it is known.

If puny: is specified, like "{puny:url}" or "{puny:host}", the "punycoded" version of the host name is used in the output. This option is mutually exclusive with idn:.

If idn: is specified like "{idn:url}" or "{idn:host}", the International Domain Name version of the host name is used in the output if it is provided as a correctly encoded punycode version. This option is mutually exclusive with puny:.

If --default-port is specified, all formats are expanded as if they used default:; and if --punycode is used, all formats are expanded as if they used puny:. Also note that "{url}" is affected by the --keep-port option.

Hosts provided as IPv6 numerical addresses are provided within square brackets. Like "[fe80::20c:29ff:fe9c:409b]".

Hosts provided as IPv4 numerical addresses are "normalized" and provided as four dot-separated decimal numbers when output.

You can access specific keys in the query string using the format {query:key}. Then the value of the first matching key is output using a case sensitive match. When extracting a URL decoded query key that contains %00, such octet is replaced with a single period '.' in the output.

You can access specific keys in the query string and out all values using the format {query-all:key}. This looks for 'key' case sensitively and outputs all values for that key space-separated.

The "format" string supports the following backslash sequences:

\ - backslash

\t - tab

\n - newline

\r - carriage return

\{ - an open curly brace that does not start a variable

\[ - an open bracket that does not start a variable

All other text in the format string is shown as-is.

-h, --help

Show the help output.

--iterate [component]=[item1 item2 ...]

Set the component to multiple values and output the result once for each iteration. Several combined iterations are allowed to generate combinations, but only one --iterate option per component. The listed items to iterate over should be separated by single spaces.


Outputs all set components of the URLs as JSON objects. All components of the URL that have data get populated in the parts object using their component names. See below for details on the format.


By default, trurl removes default port numbers from URLs with a known scheme even if they are explicitly specified in the input URL. This options, makes trurl not remove them.


Disables libcurl's scheme guessing feature. URLs that do not contain a scheme are treated as invalid URLs.


Uses the "punycoded" version of the host name, which is how International Domain Names are converted into plain ASCII. If the host name is not using IDN, the regular ASCII name is used.


Converts a "punycoded" ASCII host name to its original International Domain Name in Unicode. If the host name is not using punycode then the original host name is used.

--query-separator [what]

Specify the single letter used for separating query pairs. The default is "&" but at least in the past sometimes semicolons ";" or even colons ":" have been used for this purpose. If your URL uses something other than the default letter, setting the right one makes sure trurl can do its query operations properly.

--redirect [URL]

Redirect the URL to this new location. The redirection is performed on the base URL, so, if no base URL is specified, no redirection is performed.

--replace [data]

Replaces a URL query.

data can either take the form of a single value, or as a key/value pair in the shape foo=bar. If replace is called on an item that isn't in the list of queries trurl ignores that item.

--force-replace [data]

Works the same as --replace, but trurl appends a missing query string if it is not in the query list already.

-s, --set [component][:]=[data]

Set this URL component. Setting blank string ("") clears the component from the URL.

The following components can be set: url, scheme, user, password, options, host, port, path, query, fragment and zoneid.

If a simple "="-assignment is used, the data is URL encoded when applied. If ":=" is used, the data is assumed to already be URL encoded and stored as-is.

If "?=" is used, the set is only performed if the component is not already set. It avoids overwriting any already set data.

You can also combine : and ? into "?:=" if desired.

If no URL or --url-file argument is provided, trurl tries to create a URL using the components provided by the --set options. If not enough components are specified, this fails.


The "variable=content" tuplets in the query component are sorted in a case insensitive alphabetical order. This helps making URLs identical that otherwise only had their query pairs in different orders.

--url [URL]

Set the input URL to work with. The URL may be provided without a scheme, which then typically is not actually a legal URL but trurl tries to figure out what is meant and guess what scheme to use (unless --no-guess-scheme is used).

Providing multiple URLs makes trurl act on all URLs in a serial fashion.

If the URL cannot be parsed for whatever reason, trurl simply moves on to the next provided URL - unless --verify is used.


Outputs URL encoded version of components by default when using --get or --json.

--trim [component]=[what]

Trims data off a component. Currently this can only trim a query component.

"what" is specified as a full word or as a word prefix (using a single trailing asterisk ('*')) which makes trurl remove the tuples from the query string that match the instruction.

To match a literal trailing asterisk instead of using a wildcard, escape it with a backslash in front of it. Like "\*".

-v, --version

Show version information and exit.


When a URL is provided, return error immediately if it does not parse as a valid URL. In normal cases, trurl can forgive a bad URL input.


Suppress (some) notes and warnings.

Json output format

The --json option outputs a JSON array with one or more objects. One for each URL.

Each URL JSON object contains a number of properties, a series of key/value pairs. The exact set depends on the given URL.


This key exists in every object. It is the complete URL. Affected by --default-port, --keep-port, and --punycode.


This key exists in every object, and contains an object with a key for each of the settable URL components. If a component is missing, it means it is not present in the URL. The parts are URL decoded unless --urlencode is used.

scheme The URL scheme.

user The user name.

password The password.

options The options. Note that only a few URL schemes support the "options" component.

host The and normalized host name. It might be a UTF-8 name if an IDN name was used. It can also be a normalized IPv4 or IPv6 address. An IPv6 address always starts with a bracket ([) - and no other host names can contain such a symbol. If --punycode is used, the punycode version of the host is outputted instead.

port The provided port number as a string. If the port number was not provided in the URL, but the scheme is a known one, and --default-port is in use, the default port for that scheme is provided here.

path The path. Including the leading slash.

query The full query, excluding the question mark separator.

fragment The fragment, excluding the pound sign separator.

zoneid The zone id, which can only be present in an IPv6 address. When this key is present, then host is an IPv6 numerical address.


This key contains an array of query key/value objects. Each such pair is listed with "key" and "value" and their respective contents in the output.

The key/values are extracted from the query where they are separated by ampersands (&) - or the user sets with --query-separator.

The query pairs are listed in the order of appearance in a left-to-right order, but can be made alpha-sorted with --sort-query.

It is only present if the URL has a query.


Replace the host name of a URL

$ trurl --url --set

Create a URL by setting components

 $ trurl --set --set scheme=ftp

Redirect a URL

$ trurl --url --redirect here.html

Change port number

This also shows how trurl removes dot-dot sequences

$ trurl --url --set port=8080

Extract the path from a URL

$ trurl --url --get '{path}'

Extract the port from a URL

This gets the default port based on the scheme if the port is not set in the URL.

$ trurl --url --get '{default:port}'

Append a path segment to a URL

$ trurl --url --append path=you

Append a query segment to a URL

$ trurl --url "" --append query=search=string

Read URLs from stdin

$ cat urllist.txt | trurl --url-file -

Output JSON

$ trurl "" --json
    "url": "",
    "parts": [
        "scheme": "https",
        "host": "",
        "path": "/search",
        "query": "q=answers&user=me"
        "fragment": "frag",
    "params": [
        "key": "q",
        "value": "answers"
        "key": "user",
        "value": "me"

Remove tracking tuples from query

$ trurl "" --trim query="utm_*"

Show a specific query key value

$ trurl "" -g '{query:a}'

Sort the key/value pairs in the query component

$ trurl "" --sort-query

Work with a query that uses a semicolon separator

$ trurl ";page=5" --trim query="search" --query-separator ";"

Accept spaces in the URL path

$ trurl " has space/index.html" --accept-space

Create multiple variations of a URL with different schemes

$ trurl "" --iterate "scheme=http ftp sftp"

Exit codes

trurl returns a non-zero exit code to indicate problems.


A problem with --url-file


A problem with --append


A command line option misses an argument


A command line option mistake or an illegal option combination.


A problem with --set


Out of memory


Could not output a valid URL


A problem with --trim


If --verify is set and the input URL cannot parse.


A problem with --get


A problem with --iterate


A problem with --replace or --force-replace


See also

curl_url_set(3), curl_url_get(3)

This HTML page was made with roffit.