uri_string

uri_string

uri_string
URI processing functions.
Module uri_string was introduced in OTP 21.0.

This module contains functions for parsing and handling URIs (RFC 3986) and form-urlencoded query strings (HTML 5.2).

Parsing and serializing non-UTF-8 form-urlencoded query strings are also supported (HTML 5.0).

A URI is an identifier consisting of a sequence of characters matching the syntax rule named URI in RFC 3986.

The generic URI syntax consists of a hierarchical sequence of components referred to as the scheme, authority, path, query, and fragment:

    URI         = scheme ":" hier-part [ "?" query ] [ "#" fragment ]
    hier-part   = "//" authority path-abempty
                   / path-absolute
                   / path-rootless
                   / path-empty
    scheme      = ALPHA *( ALPHA / DIGIT / "+" / "-" / "." )
    authority   = [ userinfo "@" ] host [ ":" port ]
    userinfo    = *( unreserved / pct-encoded / sub-delims / ":" )

    reserved    = gen-delims / sub-delims
    gen-delims  = ":" / "/" / "?" / "#" / "[" / "]" / "@"
    sub-delims  = "!" / "$" / "&" / "'" / "(" / ")"
                / "*" / "+" / "," / ";" / "="

    unreserved  = ALPHA / DIGIT / "-" / "." / "_" / "~"
    

The interpretation of a URI depends only on the characters used and not on how those characters are represented in a network protocol.

The functions implemented by this module cover the following use cases:

There are four different encodings present during the handling of URIs:

  • Inbound binary encoding in binaries
  • Inbound percent-encoding in lists and binaries
  • Outbound binary encoding in binaries
  • Outbound percent-encoding in lists and binaries

Functions with uri_string() argument accept lists, binaries and mixed lists (lists with binary elements) as input type. All of the functions but transcode/2 expects input as lists of unicode codepoints, UTF-8 encoded binaries and UTF-8 percent-encoded URI parts ("%C3%B6" corresponds to the unicode character "ö").

Unless otherwise specified the return value type and encoding are the same as the input type and encoding. That is, binary input returns binary output, list input returns a list output but mixed input returns list output.

In case of lists there is only percent-encoding. In binaries, however, both binary encoding and percent-encoding shall be considered. transcode/2 provides the means to convert between the supported encodings, it takes a uri_string() and a list of options specifying inbound and outbound encodings.

RFC 3986 does not mandate any specific character encoding and it is usually defined by the protocol or surrounding text. This library takes the same assumption, binary and percent-encoding are handled as one configuration unit, they cannot be set to different values.

Quoting functions are intended to be used by URI producing application during component preparation or retrieval phase to avoid conflicts between data and characters used in URI syntax. Quoting functions use percent encoding, but with different rules than for example during execution of recompose/1. It is user responsibility to provide quoting functions with application data only and using their output to combine an URI component.
Quoting functions can for instance be used for constructing a path component with a segment containing '/' character which should not collide with '/' used as general delimiter in path component.

Error tuple indicating the type of error. Possible values of the second component:

  • invalid_character
  • invalid_encoding
  • invalid_input
  • invalid_map
  • invalid_percent_encoding
  • invalid_scheme
  • invalid_uri
  • invalid_utf8
  • missing_value

The third component is a term providing additional information about the cause of the error.

List of unicode codepoints, a UTF-8 encoded binary, or a mix of the two, representing an RFC 3986 compliant URI (percent-encoded form). A URI is a sequence of characters from a very limited set: the letters of the basic Latin alphabet, digits, and a few special characters.

This is a utility function meant to be used in the shell for printing the allowed characters in each major URI component, and also in the most important characters sets. Please note that this function does not replace the ABNF rules defined by the standards, these character sets are derived directly from those aformentioned rules. For more information see the Uniform Resource Identifiers chapter in stdlib's Users Guide.

Types

QueryList = [{unicode:chardata(), unicode:chardata() | true}]
QueryString = uri_string() | error()

Composes a form-urlencoded QueryString based on a QueryList, a list of non-percent-encoded key-value pairs. Form-urlencoding is defined in section 4.10.21.6 of the HTML 5.2 specification and in section 4.10.22.6 of the HTML 5.0 specification for non-UTF-8 encodings.

See also the opposite operation dissect_query/1.

Example:

1> uri_string:compose_query([{"foo bar","1"},{"city","örebro"}]).
"foo+bar=1&city=%C3%B6rebro"
2> uri_string:compose_query([{<<"foo bar">>,<<"1">>},
2> {<<"city">>,<<"örebro"/utf8>>}]).
<<"foo+bar=1&city=%C3%B6rebro">>
	

Types

QueryList = [{unicode:chardata(), unicode:chardata() | true}]
Options = [{encoding, atom()}]
QueryString = uri_string() | error()

Same as compose_query/1 but with an additional Options parameter, that controls the encoding ("charset") used by the encoding algorithm. There are two supported encodings: utf8 (or unicode) and latin1.

Each character in the entry's name and value that cannot be expressed using the selected character encoding, is replaced by a string consisting of a U+0026 AMPERSAND character (&), a "#" (U+0023) character, one or more ASCII digits representing the Unicode code point of the character in base ten, and finally a ";" (U+003B) character.

Bytes that are out of the range 0x2A, 0x2D, 0x2E, 0x30 to 0x39, 0x41 to 0x5A, 0x5F, 0x61 to 0x7A, are percent-encoded (U+0025 PERCENT SIGN character (%) followed by uppercase ASCII hex digits representing the hexadecimal value of the byte).

See also the opposite operation dissect_query/1.

Example:

1> uri_string:compose_query([{"foo bar","1"},{"city","örebro"}],
1> [{encoding, latin1}]).
"foo+bar=1&city=%F6rebro"
2> uri_string:compose_query([{<<"foo bar">>,<<"1">>},
2> {<<"city">>,<<"東京"/utf8>>}], [{encoding, latin1}]).
<<"foo+bar=1&city=%26%2326481%3B%26%2320140%3B">>
	

Types

QueryString = uri_string()
QueryList =
    [{unicode:chardata(), unicode:chardata() | true}] | error()

Dissects an urlencoded QueryString and returns a QueryList, a list of non-percent-encoded key-value pairs. Form-urlencoding is defined in section 4.10.21.6 of the HTML 5.2 specification and in section 4.10.22.6 of the HTML 5.0 specification for non-UTF-8 encodings.

See also the opposite operation compose_query/1.

Example:

1> uri_string:dissect_query("foo+bar=1&city=%C3%B6rebro").
[{"foo bar","1"},{"city","örebro"}]
2> uri_string:dissect_query(<<"foo+bar=1&city=%26%2326481%3B%26%2320140%3B">>).
[{<<"foo bar">>,<<"1">>},
 {<<"city">>,<<230,157,177,228,186,172>>}] 
	

Types

NormalizedURI = uri_string() | error()

Transforms an URI into a normalized form using Syntax-Based Normalization as defined by RFC 3986.

This function implements case normalization, percent-encoding normalization, path segment normalization and scheme based normalization for HTTP(S) with basic support for FTP, SSH, SFTP and TFTP.

Example:

1> uri_string:normalize("/a/b/c/./../../g").
"/a/g"
2> uri_string:normalize(<<"mid/content=5/../6">>).
<<"mid/6">>
3> uri_string:normalize("http://localhost:80").
"http://localhost/"
4> uri_string:normalize(#{scheme => "http",port => 80,path => "/a/b/c/./../../g",
4> host => "localhost-örebro"}).
"http://localhost-%C3%B6rebro/a/g"
	

Types

Options = [return_map]
NormalizedURI = uri_string() | uri_map() | error()

Same as normalize/1 but with an additional Options parameter, that controls whether the normalized URI shall be returned as an uri_map(). There is one supported option: return_map.

Example:

1> uri_string:normalize("/a/b/c/./../../g", [return_map]).
#{path => "/a/g"}
2> uri_string:normalize(<<"mid/content=5/../6">>, [return_map]).
#{path => <<"mid/6">>}
3> uri_string:normalize("http://localhost:80", [return_map]).
#{scheme => "http",path => "/",host => "localhost"}
4> uri_string:normalize(#{scheme => "http",port => 80,path => "/a/b/c/./../../g",
4> host => "localhost-örebro"}, [return_map]).
#{scheme => "http",path => "/a/g",host => "localhost-örebro"}
	

Types

URIString = uri_string()
URIMap = uri_map() | error()

Parses an RFC 3986 compliant uri_string() into a uri_map(), that holds the parsed components of the URI. If parsing fails, an error tuple is returned.

See also the opposite operation recompose/1.

Example:

1> uri_string:parse("foo://user@example.com:8042/over/there?name=ferret#nose").
#{fragment => "nose",host => "example.com",
  path => "/over/there",port => 8042,query => "name=ferret",
  scheme => foo,userinfo => "user"}
2> uri_string:parse(<<"foo://user@example.com:8042/over/there?name=ferret">>).
#{host => <<"example.com">>,path => <<"/over/there">>,
  port => 8042,query => <<"name=ferret">>,scheme => <<"foo">>,
  userinfo => <<"user">>}
	

Types

Result =
    uri_string() |
    uri_map() |
    {error, {invalid, {atom(), {term(), term()}}}}

Decodes all percent-encoded triplets in the input that can be both a uri_string() and a uri_map(). Note, that this function performs raw decoding and it shall be used on already parsed URI components. Applying this function directly on a standard URI can effectively change it.

If the input encoding is not UTF-8, an error tuple is returned.

Example:

1> uri_string:percent_decode(#{host => "localhost-%C3%B6rebro",path => [],
1> scheme => "http"}).
#{host => "localhost-örebro",path => [],scheme => "http"}
2> uri_string:percent_decode(<<"%C3%B6rebro">>).
<<"örebro"/utf8>>
	
Warning

Using uri_string:percent_decode/1 directly on a URI is not safe. This example shows, that after each consecutive application of the function the resulting URI will be changed. None of these URIs refer to the same resource.

3> uri_string:percent_decode(<<"http://local%252Fhost/path">>).
<<"http://local%2Fhost/path">>
4> uri_string:percent_decode(<<"http://local%2Fhost/path">>).
<<"http://local/host/path">>
        

Types

Data = QuotedData = unicode:chardata()

Replaces characters out of unreserved set with their percent encoded equivalents.

Unreserved characters defined in RFC 3986 are not quoted.

Example:

1> uri_string:quote("SomeId/04").
"SomeId%2F04"
2> uri_string:quote(<<"SomeId/04">>).
<<"SomeId%2F04">>
	
Warning

Function is not aware about any URI component context and should not be used on whole URI. If applied more than once on the same data, might produce unexpected results.

Types

Safe = string()
QuotedData = unicode:chardata()

Same as quote/1, but Safe allows user to provide a list of characters to be protected from encoding.

Example:

1> uri_string:quote("SomeId/04", "/").
"SomeId/04"
2> uri_string:quote(<<"SomeId/04">>, "/").
<<"SomeId/04">>
	
Warning

Function is not aware about any URI component context and should not be used on whole URI. If applied more than once on the same data, might produce unexpected results.

Types

URIMap = uri_map()
URIString = uri_string() | error()

Creates an RFC 3986 compliant URIString (percent-encoded), based on the components of URIMap. If the URIMap is invalid, an error tuple is returned.

See also the opposite operation parse/1.

Example:

1> URIMap = #{fragment => "nose", host => "example.com", path => "/over/there",
1> port => 8042, query => "name=ferret", scheme => "foo", userinfo => "user"}.
#{fragment => "nose",host => "example.com",
  path => "/over/there",port => 8042,query => "name=ferret",
  scheme => "foo",userinfo => "user"}

2> uri_string:recompose(URIMap).
"foo://example.com:8042/over/there?name=ferret#nose"

Types

RefURI = BaseURI = uri_string() | uri_map()
TargetURI = uri_string() | error()

Convert a RefURI reference that might be relative to a given base URI into the parsed components of the reference's target, which can then be recomposed to form the target URI.

Example:

1> uri_string:resolve("/abs/ol/ute", "http://localhost/a/b/c?q").
"http://localhost/abs/ol/ute"
2> uri_string:resolve("../relative", "http://localhost/a/b/c?q").
"http://localhost/a/relative"
3> uri_string:resolve("http://localhost/full", "http://localhost/a/b/c?q").
"http://localhost/full"
4> uri_string:resolve(#{path => "path", query => "xyz"}, "http://localhost/a/b/c?q").
"http://localhost/a/b/path?xyz"
	

Types

RefURI = BaseURI = uri_string() | uri_map()
Options = [return_map]

Same as resolve/2 but with an additional Options parameter, that controls whether the target URI shall be returned as an uri_map(). There is one supported option: return_map.

Example:

1> uri_string:resolve("/abs/ol/ute", "http://localhost/a/b/c?q", [return_map]).
#{host => "localhost",path => "/abs/ol/ute",scheme => "http"}
2> uri_string:resolve(#{path => "/abs/ol/ute"}, #{scheme => "http",
2> host => "localhost", path => "/a/b/c?q"}, [return_map]).
#{host => "localhost",path => "/abs/ol/ute",scheme => "http"}
	

Types

URIString = uri_string()
Options =
    [{in_encoding, unicode:encoding()} |
     {out_encoding, unicode:encoding()}]

Transcodes an RFC 3986 compliant URIString, where Options is a list of tagged tuples, specifying the inbound (in_encoding) and outbound (out_encoding) encodings. in_encoding and out_encoding specifies both binary encoding and percent-encoding for the input and output data. Mixed encoding, where binary encoding is not the same as percent-encoding, is not supported. If an argument is invalid, an error tuple is returned.

Example:

1> uri_string:transcode(<<"foo%00%00%00%F6bar"/utf32>>,
1> [{in_encoding, utf32},{out_encoding, utf8}]).
<<"foo%C3%B6bar"/utf8>>
2> uri_string:transcode("foo%F6bar", [{in_encoding, latin1},
2> {out_encoding, utf8}]).
"foo%C3%B6bar"
	

Types

QuotedData = Data = unicode:chardata()

Percent decode characters.

Example:

1> uri_string:unquote("SomeId%2F04").
"SomeId/04"
2> uri_string:unquote(<<"SomeId%2F04">>).
<<"SomeId/04">>
	
Warning

Function is not aware about any URI component context and should not be used on whole URI. If applied more than once on the same data, might produce unexpected results.