404-solution / includes / ngram / TermNGramCoveragePolicy.php

TermNGramCoveragePolicy.php in 404 Solution trunk, at includes/ngram/TermNGramCoveragePolicy.php

118 lines 4.1 KB

Raw Download Zip

1	<?php
2
3	if (!defined('ABSPATH')) {
4	exit;
5	}
6
7	/**
8	* Decides whether the per-type (category/tag) n-gram prefilter is trustworthy
9	* enough to use on a given 404 request, or whether the caller should fall back
10	* to the full getPublishedCategories()/getPublishedTags() scan.
11	*
12	* Mirrors the post-side gates in SpellNGramPrefilter: a minimum absolute number
13	* of cache entries for the type AND a minimum coverage ratio (cached term
14	* n-grams / published terms of that type). Both must hold. This is intentionally
15	* a separate policy from NGramCoveragePolicy, which measures post coverage
16	* against the permalink cache; the term denominator is the published-taxonomy
17	* count, a different source.
18	*
19	* Unit-testable in isolation: construct with the ngram filter (count source)
20	* and a published-content repository exposing getPublishedCategoryCount() /
21	* getPublishedTagCount(), then call isReady($type).
22	*/
23	class ABJ_404_Solution_TermNGramCoveragePolicy {
24
25	/** Minimum cached term n-grams of a type before the prefilter is trusted. */
26	const TERM_MIN_CACHE_ENTRIES = 50;
27
28	/** Minimum (cached / published) coverage ratio for a type. */
29	const TERM_MIN_COVERAGE_RATIO = 0.8;
30
31	/** N-gram cache type for category archives. */
32	const TYPE_CATEGORY = 'category';
33
34	/** N-gram cache type for tag archives. */
35	const TYPE_TAG = 'tag';
36
37	/** @var ABJ_404_Solution_NGramFilter */
38	private $ngramFilter;
39
40	/** @var mixed Object exposing getPublishedCategoryCount() and getPublishedTagCount(). */
41	private $contentRepo;
42
43	/** @var ABJ_404_Solution_Logging */
44	private $logger;
45
46	/**
47	* @param ABJ_404_Solution_NGramFilter $ngramFilter Source of getCacheCountForType().
48	* @param mixed $contentRepo Object exposing getPublishedCategoryCount() / getPublishedTagCount().
49	* @param ABJ_404_Solution_Logging $logger
50	*/
51	public function __construct($ngramFilter, $contentRepo, $logger) {
52	$this->ngramFilter = $ngramFilter;
53	$this->contentRepo = $contentRepo;
54	$this->logger = $logger;
55	}
56
57	/**
58	* Is the prefilter trustworthy for this term type right now?
59	*
60	* @param string $type 'category' or 'tag'.
61	* @return bool True when the cache has enough entries AND covers enough of
62	* the published terms; false on cold/insufficient coverage.
63	*/
64	public function isReady(string $type): bool {
65	if ($type !== self::TYPE_CATEGORY && $type !== self::TYPE_TAG) {
66	return false;
67	}
68
69	$cacheCount = $this->ngramFilter->getCacheCountForType($type);
70	if ($cacheCount < self::TERM_MIN_CACHE_ENTRIES) {
71	$this->logger->debugMessage(sprintf(
72	"Term n-gram prefilter not ready (min entries): type=%s count=%d (need %d)",
73	$type,
74	$cacheCount,
75	self::TERM_MIN_CACHE_ENTRIES
76	));
77	return false;
78	}
79
80	$publishedCount = $this->publishedCountFor($type);
81	if ($publishedCount <= 0) {
82	// No published terms of this type: nothing to prefilter against.
83	return false;
84	}
85
86	$ratio = $cacheCount / $publishedCount;
87	if ($ratio < self::TERM_MIN_COVERAGE_RATIO) {
88	$this->logger->debugMessage(sprintf(
89	"Term n-gram prefilter not ready (low coverage): type=%s ratio=%.2f (need %.2f)",
90	$type,
91	$ratio,
92	self::TERM_MIN_COVERAGE_RATIO
93	));
94	return false;
95	}
96
97	return true;
98	}
99
100	/**
101	* @param string $type
102	* @return int Published term count for the type, or 0 on unknown type.
103	*/
104	private function publishedCountFor(string $type): int {
105	$repo = $this->contentRepo;
106	if (!is_object($repo)) {
107	return 0;
108	}
109	if ($type === self::TYPE_CATEGORY && method_exists($repo, 'getPublishedCategoryCount')) {
110	return (int)$repo->getPublishedCategoryCount();
111	}
112	if ($type === self::TYPE_TAG && method_exists($repo, 'getPublishedTagCount')) {
113	return (int)$repo->getPublishedTagCount();
114	}
115	return 0;
116	}
117	}
118