69 lines
2.6 KiB
PHP
69 lines
2.6 KiB
PHP
|
<?php
|
||
|
|
namespace AfFulltext;
|
||
|
|
|
||
|
|
/**
|
||
|
|
* A parsed FiveFilters site-config rule.
|
||
|
|
*
|
||
|
|
* Field names track the ftr-site-config directive names so a rule file and this
|
||
|
|
* object read the same way. Directives that may repeat accumulate into arrays,
|
||
|
|
* exactly as Full-Text RSS treats them.
|
||
|
|
*/
|
||
|
|
final class Rule {
|
||
|
|
/** @var string[] */ public array $titles = [];
|
||
|
|
/** @var string[] */ public array $bodies = [];
|
||
|
|
/** @var string[] */ public array $authors = [];
|
||
|
|
/** @var string[] */ public array $dates = [];
|
||
|
|
/** @var string[] */ public array $strips = [];
|
||
|
|
/** @var string[] */ public array $strip_id_or_class = [];
|
||
|
|
/** @var string[] */ public array $strip_image_src = [];
|
||
|
|
/** @var string[] */ public array $dissolve = [];
|
||
|
|
/** @var string[] */ public array $single_page_links = [];
|
||
|
|
/** @var string[] */ public array $next_page_links = [];
|
||
|
|
/** @var string[] */ public array $find_strings = [];
|
||
|
|
/** @var string[] */ public array $replace_strings = [];
|
||
|
|
/** @var string[] */ public array $test_urls = [];
|
||
|
|
/** @var array<string,string> */ public array $http_headers = [];
|
||
|
|
/** @var array<string,string> map of xpath => wrapper spec e.g. "div.foo" */
|
||
|
|
public array $wrap_in = [];
|
||
|
|
|
||
|
|
public bool $prune = true;
|
||
|
|
public bool $tidy = true;
|
||
|
|
public bool $autodetect_on_failure = true;
|
||
|
|
|
||
|
|
/**
|
||
|
|
* Files this rule was assembled from, nearest-match first. Surfaced in the
|
||
|
|
* UI so a stale rule is visible rather than silently falling through to
|
||
|
|
* Readability -- the exact failure mode that hid a broken comic rule for
|
||
|
|
* years.
|
||
|
|
*
|
||
|
|
* @var string[]
|
||
|
|
*/
|
||
|
|
public array $sources = [];
|
||
|
|
|
||
|
|
public function is_empty(): bool {
|
||
|
|
return !$this->bodies && !$this->strips && !$this->strip_id_or_class
|
||
|
|
&& !$this->strip_image_src && !$this->titles && !$this->dissolve;
|
||
|
|
}
|
||
|
|
|
||
|
|
/** Merge $other UNDER $this: existing scalars win, list directives concatenate. */
|
||
|
|
public function merge_under(self $other): self {
|
||
|
|
$m = clone $this;
|
||
|
|
|
||
|
|
foreach (['titles', 'bodies', 'authors', 'dates', 'single_page_links', 'next_page_links'] as $f)
|
||
|
|
if (!$m->$f) $m->$f = $other->$f;
|
||
|
|
|
||
|
|
foreach (['strips', 'strip_id_or_class', 'strip_image_src', 'dissolve', 'test_urls'] as $f)
|
||
|
|
$m->$f = array_values(array_unique([...$m->$f, ...$other->$f]));
|
||
|
|
|
||
|
|
// find/replace are positional pairs -- appending keeps them aligned.
|
||
|
|
$m->find_strings = [...$m->find_strings, ...$other->find_strings];
|
||
|
|
$m->replace_strings = [...$m->replace_strings, ...$other->replace_strings];
|
||
|
|
|
||
|
|
$m->http_headers = $m->http_headers + $other->http_headers;
|
||
|
|
$m->wrap_in = $m->wrap_in + $other->wrap_in;
|
||
|
|
$m->sources = [...$m->sources, ...$other->sources];
|
||
|
|
|
||
|
|
return $m;
|
||
|
|
}
|
||
|
|
}
|