README

Interfaces for interop with hershel-theodore-layton/lecof-router

RequestInfo exposes URL-decoded path segments. Split the URI path on raw / first, then decode each segment once. Remove one leading slash; the empty path and / have no segments. Keep other empty and trailing segments, and exclude query and fragment text before splitting.

For a custom RequestInfo, this copyable constructor follows that contract ($path is the URI path component):

private vec<string> $segments;

public function __construct(string $path)[] {
  $this->segments = \HH\Lib\Str\strip_prefix($path, '/')
    |> \HH\Lib\Str\split($$, '/')
    |> $$ === vec[''] ? vec[] : $$
    |> \HH\Lib\Vec\map($$, \urldecode<>);
}

Return these same decoded segments from lookup/slicing methods. Never split a segment again: a decoded slash belongs inside its segment. literal('a/b', ...) can match /a%2Fb; nested literals 'a', 'b' and slashed_literal('a/b', ...) match /a/b instead. Variable parsers receive the decoded segment too, and getRawValue() retains that supplied text. Do not URL-decode it again.

getPathSegmentsAsString() joins segments with / for convenience. Its result cannot distinguish vec['a/b'] from vec['a', 'b']; filters must preserve the segment vector when matching. For the empty path and /, use done() directly; slashed filters require a segment to be present.

Upstream URL normalization can change the path before this adapter sees it. If an encoded slash must remain inside one segment, configure proxies to preserve its encoded spelling until splitting.