Percent encode the URI template literals as per RFC6570 section 3.1

Template::expandAll swapped every expression notation in the raw template
string with str_replace, so the literal text surrounding the expressions was
copied verbatim and never percent encoded. The expression detection pattern is
now used to split the template into an ordered list of parts: literals are
encoded once at construction, expressions stay Expression objects, and the
expansion simply concatenates them. Keeping the parts apart is what prevents
the already encoded expansion results from being encoded a second time.

Fixes the "Additional Examples 8: Literal Encoding" case of the vendored
uritemplate-test compliance suite.
This commit is contained in:
dualfroz committed 2026-09-04 08:43:47 +02:00
1 parent 432904515e
commit 2f4f2d71c4
3 files changed
+143 -22

No files matched your search

+25
View File
@@ -85,6 +85,31 @@ echo $uriTemplate->expandOrFail($params);
<p class="message-notice"><code>expandToUriOrFail()</code> and <code>expandToUrlOrFail()</code> are available since <p class="message-notice"><code>expandToUriOrFail()</code> and <code>expandToUrlOrFail()</code> are available since
version <code>7.6.0</code></p> version <code>7.6.0</code></p>
### Literal Expansion
The characters located outside of the template expressions are literals. As per
[RFC6570 section 3.1](https://www.rfc-editor.org/rfc/rfc6570#section-3.1) a literal character is
copied as is when it is allowed anywhere in the URI syntax, in other words when it belongs to the
`unreserved / reserved / pct-encoded` set. Any other character is percent encoded using its UTF-8
octets.
~~~php
$uriTemplate = new UriTemplate('https://example.com/café/{hotel}');
echo $uriTemplate->expand(['hotel' => 'Rest & Relax']), PHP_EOL;
// display https://example.com/caf%C3%A9/Rest%20%26%20Relax
~~~
Percent encoded triplets already present in the template are left untouched, they are never
encoded a second time.
~~~php
$uriTemplate = new UriTemplate('https://example.com/caf%C3%A9/{hotel}');
echo $uriTemplate->expand(['hotel' => 'Rest & Relax']), PHP_EOL;
// display https://example.com/caf%C3%A9/Rest%20%26%20Relax
~~~
## Template Variables ## Template Variables
<p class="message-notice">For maximum interoperability you should make sure your variables are <p class="message-notice">For maximum interoperability you should make sure your variables are
+61 -22
View File
@@ -20,14 +20,18 @@ use Stringable;
use function array_filter; use function array_filter;
use function array_map; use function array_map;
use function array_reduce;
use function array_unique; use function array_unique;
use function preg_match_all; use function array_values;
use function implode;
use function preg_replace; use function preg_replace;
use function str_replace; use function preg_replace_callback;
use function preg_split;
use function rawurlencode;
use function str_starts_with;
use function strpbrk; use function strpbrk;
use const PREG_SET_ORDER; use const PREG_SPLIT_DELIM_CAPTURE;
use const PREG_SPLIT_NO_EMPTY;
/** /**
* @internal The class exposes the internal representation of a Template and its usage * @internal The class exposes the internal representation of a Template and its usage
@@ -39,19 +43,37 @@ final class Template implements Stringable
*/ */
private const REGEXP_EXPRESSION_DETECTOR = '/(?<expression>\{[^}]*})/x'; private const REGEXP_EXPRESSION_DETECTOR = '/(?<expression>\{[^}]*})/x';
/** @var array<Expression> */ /**
private readonly array $expressions; * Literal characters which are not allowed to appear as-is in the expanded template.
*
* Every character outside of the "unreserved / reserved / pct-encoded" set has to be
* percent encoded, a "%" which starts no percent encoded triplet included.
*
* @link https://www.rfc-editor.org/rfc/rfc6570#section-3.1
*/
private const REGEXP_LITERAL_TO_ENCODE = '/[^A-Za-z\d\-._~:\/?#\[\]@!$&\'()*+,;=%]+|%(?![A-Fa-f\d]{2})/';
/** @var array<string|Expression> */
private readonly array $parts;
/** @var array<string> */ /** @var array<string> */
public readonly array $variableNames; public readonly array $variableNames;
private function __construct(public readonly string $value, Expression ...$expressions) private function __construct(public readonly string $value, string|Expression ...$parts)
{ {
$this->expressions = $expressions; $this->parts = $parts;
$expressions = [];
foreach ($parts as $part) {
if ($part instanceof Expression) {
$expressions[$part->value] = $part;
}
}
$this->variableNames = array_unique( $this->variableNames = array_unique(
array_merge( array_merge(
...array_map( ...array_map(
static fn (Expression $expression): array => $expression->variableNames, static fn (Expression $expression): array => $expression->variableNames,
$expressions array_values($expressions)
) )
) )
); );
@@ -72,17 +94,35 @@ final class Template implements Stringable
$remainder = preg_replace(self::REGEXP_EXPRESSION_DETECTOR, '', $template); $remainder = preg_replace(self::REGEXP_EXPRESSION_DETECTOR, '', $template);
false === strpbrk($remainder, '{}') || throw new SyntaxError('The template "'.$template.'" contains invalid expressions.'); false === strpbrk($remainder, '{}') || throw new SyntaxError('The template "'.$template.'" contains invalid expressions.');
preg_match_all(self::REGEXP_EXPRESSION_DETECTOR, $template, $founds, PREG_SET_ORDER); /** @var array<string> $segments */
$segments = preg_split(self::REGEXP_EXPRESSION_DETECTOR, $template, -1, PREG_SPLIT_DELIM_CAPTURE | PREG_SPLIT_NO_EMPTY);
return new self($template, ...array_values( $parts = [];
array_reduce($founds, function (array $carry, array $found): array { $expressions = [];
if (!isset($carry[$found['expression']])) { foreach ($segments as $segment) {
$carry[$found['expression']] = Expression::new($found['expression']); if (!str_starts_with($segment, '{')) {
$parts[] = self::encodeLiteral($segment);
continue;
} }
return $carry; $parts[] = $expressions[$segment] ??= Expression::new($segment);
}, []) }
));
return new self($template, ...$parts);
}
/**
* Percent encodes the template characters which are not allowed to be copied as-is into a URI.
*
* @link https://www.rfc-editor.org/rfc/rfc6570#section-3.1
*/
private static function encodeLiteral(string $literal): string
{
return (string) preg_replace_callback(
self::REGEXP_LITERAL_TO_ENCODE,
static fn (array $matches): string => rawurlencode($matches[0]),
$literal
);
} }
/** /**
@@ -116,11 +156,10 @@ final class Template implements Stringable
private function expandAll(VariableBag $variables): string private function expandAll(VariableBag $variables): string
{ {
return array_reduce( return implode('', array_map(
$this->expressions, static fn (string|Expression $part): string => $part instanceof Expression ? $part->expand($variables) : $part,
fn (string $uri, Expression $expr): string => str_replace($expr->value, $expr->expand($variables), $uri), $this->parts
$this->value ));
);
} }
public function __toString(): string public function __toString(): string
+57
View File
@@ -174,6 +174,63 @@ final class TemplateTest extends TestCase
]; ];
} }
#[DataProvider('providesLiteralEncoding')]
#[Test]
public function testItEncodesTheTemplateLiterals(string $notation, array $variables, string $expected): void
{
self::assertSame($expected, Template::new($notation)->expand($variables));
}
/**
* @see https://www.rfc-editor.org/rfc/rfc6570#section-3.1
*/
public static function providesLiteralEncoding(): iterable
{
$variables = [
'var' => 'value',
'path' => '/foo/bar',
'hello' => 'Hello World!',
];
return [
'a character disallowed in a URI is encoded as its UTF-8 octets' => [
'notation' => 'café/{var}',
'variables' => $variables,
'expected' => 'caf%C3%A9/value',
],
'a percent encoded triplet is copied as is' => [
'notation' => 'x%20y/{var}',
'variables' => $variables,
'expected' => 'x%20y/value',
],
'a percent encoded triplet is copied as is on both sides of an expression' => [
'notation' => 'x%20y{var}z%20w',
'variables' => $variables,
'expected' => 'x%20yvaluez%20w',
],
'a space is encoded' => [
'notation' => 'a b/{var}',
'variables' => $variables,
'expected' => 'a%20b/value',
],
'a percent sign which starts no triplet is encoded' => [
'notation' => '100%/{var}',
'variables' => $variables,
'expected' => '100%25/value',
],
'reserved characters are copied as is' => [
'notation' => "/a[0]:b@c!d\$e&f'g(h)i*j+k,l;m=n?o#{var}",
'variables' => $variables,
'expected' => "/a[0]:b@c!d\$e&f'g(h)i*j+k,l;m=n?o#value",
],
'the expanded expressions are never encoded twice' => [
'notation' => 'café{+path}{#hello}',
'variables' => $variables,
'expected' => 'caf%C3%A9/foo/bar#Hello%20World!',
],
];
}
public function testExpandOrFailIfAtLeastOneVariableIsMissing(): void public function testExpandOrFailIfAtLeastOneVariableIsMissing(): void
{ {
$this->expectException(TemplateCanNotBeExpanded::class); $this->expectException(TemplateCanNotBeExpanded::class);