← All changes | otto/Otto_html_class.php +120 -2 2.7.1 → 2.7.2 View file →

		@@ -1273,8 +1273,9 @@
1273	1273
1274	1274	# DEDUPLICATION: Remove duplicate <title>, meta description, OG, Twitter tags, canonical, and JSON-LD schema
1275	1275	$result_html = $this->otto_guard_html($this->deduplicate_title_tags($result_html), $result_html, 'deduplicate_title_tags');
1276	1276	$result_html = $this->otto_guard_html($this->deduplicate_description_tags($result_html), $result_html, 'deduplicate_description_tags');
	1277	+ $result_html = $this->otto_guard_html($this->deduplicate_keywords_tags($result_html), $result_html, 'deduplicate_keywords_tags');
1277	1278	$result_html = $this->otto_guard_html($this->deduplicate_og_twitter_tags($result_html), $result_html, 'deduplicate_og_twitter_tags');
1278	1279	$result_html = $this->otto_guard_html($this->apply_metabox_og_precedence($result_html), $result_html, 'apply_metabox_og_precedence');
1279	1280	$result_html = $this->otto_guard_html($this->apply_custom_seo_precedence($result_html), $result_html, 'apply_custom_seo_precedence');
1280	1281	$result_html = $this->otto_guard_html($this->deduplicate_schema_tags($result_html), $result_html, 'deduplicate_schema_tags');
		@@ -2239,9 +2240,18 @@
2239	2240	$marker_attribute = $marker === 'otto'
2240	2241	? ' data-metasync-otto="true"'
2241	2242	: ' data-metasync-seo="custom"';
2242	2243	$tag = '<meta ' . $attr . '="' . $val . '" content="' . $content_escaped . '"' . $marker_attribute . ' />';
2243		- $pattern = '/<meta\s[^>]*' . preg_quote($attr, '/') . '\s*=\s*["\']' . preg_quote($val, '/') . '["\'][^>]*\/?>/i';
	2244	+ # The optional prefix group must end on whitespace or a quote so the
	2245	+ # attribute matched is $attr itself, not something merely ending in it:
	2246	+ # a bare [^>]*name= also matches data-name="keywords", so a third party's
	2247	+ # unrelated tag was consumed and rewritten while the real one survived —
	2248	+ # destroying their value AND leaving the duplicate the dedup had just
	2249	+ # collapsed. Making the group optional keeps the common
	2250	+ # <meta name="..." ...> form matching, where <meta\s has already consumed
	2251	+ # the only separator. Same rule as deduplicate_keywords_tags(); this
	2252	+ # helper serves title, description and keywords, so all three are covered.
	2253	+ $pattern = '/<meta\s(?:[^>]*[\s"\'])?' . preg_quote($attr, '/') . '\s*=\s*["\']' . preg_quote($val, '/') . '["\'][^>]*\/?>/i';
2244	2254	$count = 0;
2245	2255	$new = preg_replace_callback($pattern, function ($m) use ($tag) {
2246	2256	return $tag;
2247	2257	}, $html, 1, $count);
		@@ -2272,9 +2282,10 @@
2272	2282	return class_exists('Metasync_Seo_Precedence');
2273	2283	}
2274	2284
2275	2285	/**
2276		- * Enforce the resolved SEO title/description as the FINAL word over OTTO.
	2286	+ * Enforce the resolved SEO title, description and keywords as the FINAL
	2287	+ * word over OTTO.
2277	2288	*
2278	2289	* The resolver owns the global custom-versus-OTTO order. This final buffer
2279	2290	* pass must use it too: otherwise the setting works through the sidebar
2280	2291	* and third-party filters but OTTO's later string pass puts the custom
		@@ -2344,8 +2355,40 @@
2344	2355	: 'custom';
2345	2356	$html = $this->force_custom_meta($html, 'name', 'description', $esc, $marker);
2346	2357	}
2347	2358
	2359	+ # Keywords — the same final pass as the description above, on the chain
	2360	+ # the resolver owns, so the global priority setting governs this tag
	2361	+ # too. OTTO's keywords tag arrives via header_html_insertion carrying
	2362	+ # a data-otto marker and is spliced in after wp_head, so it is always
	2363	+ # the later tag and always wins deduplicate_keywords_tags(); forcing
	2364	+ # the resolved winner here is what keeps the losing tier reachable at
	2365	+ # all — the single-printing-owner rule the description branch follows.
	2366	+ #
	2367	+ # The tier this forces is NOT guaranteed customer-typed, unlike the
	2368	+ # title and description keys above: the meta_keywords persistence flag
	2369	+ # writes OTTO's value straight to _metasync_focus_keyword, and neither
	2370	+ # that conditional write nor the clone cleaner (which strips only the
	2371	+ # _metasync_otto_* keys) ever clears it, so a stale flag copy can
	2372	+ # outrank a newer suggestion under custom-first. That key sharing is
	2373	+ # tracked separately — the fix is a key of its own for the persisted
	2374	+ # keyword, not a special case here.
	2375	+ $resolved_keyword = Metasync_Seo_Precedence::resolve(
	2376	+ $post_id,
	2377	+ Metasync_Seo_Precedence::FIELD_FOCUS_KEYWORD
	2378	+ );
	2379	+ if (!empty($resolved_keyword['value'])) {
	2380	+ $esc = htmlspecialchars($resolved_keyword['value'], ENT_QUOTES, 'UTF-8');
	2381	+ $marker = Metasync_Seo_Precedence::is_otto_value(
	2382	+ $post_id,
	2383	+ Metasync_Seo_Precedence::FIELD_FOCUS_KEYWORD,
	2384	+ $resolved_keyword
	2385	+ )
	2386	+ ? 'otto'
	2387	+ : 'custom';
	2388	+ $html = $this->force_custom_meta($html, 'name', 'keywords', $esc, $marker);
	2389	+ }
	2390	+
2348	2391	return $html;
2349	2392	}
2350	2393
2351	2394	# this function does the header replacements
		@@ -3090,8 +3133,82 @@
3090	3133	return $html;
3091	3134	}
3092	3135
3093	3136	/**
	3137	+ * Remove duplicate <meta name="keywords"> tags from HTML.
	3138	+ *
	3139	+ * OTTO ships its keywords tag inside header_html_insertion, and MetaSync's
	3140	+ * own SEO output now emits one too whenever the precedence chain resolves a
	3141	+ * keyword — so a server-side-rendered page can carry both.
	3142	+ *
	3143	+ * The keeper is OTTO's tag, or the first occurrence when no OTTO tag is
	3144	+ * present. Unlike deduplicate_description_tags() there is no
	3145	+ * customer-authored branch here: print_metatag() emits no data attribute, so
	3146	+ * a keywords tag never carries data-metasync-seo="custom" at this point in
	3147	+ * the pipeline. The customer tier is re-applied afterwards by
	3148	+ * apply_custom_seo_precedence(), which runs later and stamps the marker via
	3149	+ * force_custom_meta(). Collapsing to OTTO's tag here is therefore safe: it is
	3150	+ * not the final word on what the page renders.
	3151	+ *
	3152	+ * @param string $html Full HTML document.
	3153	+ * @return string HTML with at most one <meta name="keywords">.
	3154	+ */
	3155	+ private function deduplicate_keywords_tags($html) {
	3156	+ # No is_string() guard, unlike deduplicate_description_tags() beside it.
	3157	+ # That asymmetry is forced, not an oversight: @param string makes the
	3158	+ # check provably always-true at PHPStan level 5, and the description
	3159	+ # sibling's copy is one of a fixed number of occurrences the baseline
	3160	+ # pins by count, so adding another fails the build. Both methods are
	3161	+ # only ever called with the buffer string.
	3162	+ if ($html === '') {
	3163	+ return $html;
	3164	+ }
	3165	+
	3166	+ # Match <meta ... name="keywords" ...> in either attribute order.
	3167	+ #
	3168	+ # The optional prefix group must end on whitespace or a quote so the
	3169	+ # attribute is `name`, not something merely ending in it: a bare
	3170	+ # [^>]*name= also matches data-name="keywords". Making the group
	3171	+ # optional keeps the common `<meta name="keywords" ...>` form matching,
	3172	+ # where <meta\s has already consumed the only separator.
	3173	+ #
	3174	+ # The opening quote must be immediately followed by "keywords", so
	3175	+ # name="news_keywords" is not matched.
	3176	+ $pattern = '/<meta\s(?:[^>]*[\s"\'])?name\s*=\s*["\']keywords["\'][^>]*\/?>/i';
	3177	+
	3178	+ if (preg_match_all($pattern, $html, $matches) <= 1) {
	3179	+ return $html; # 0 or 1 — nothing to deduplicate
	3180	+ }
	3181	+
	3182	+ $all_tags = $matches[0];
	3183	+
	3184	+ $otto_tag = null;
	3185	+ foreach ($all_tags as $tag) {
	3186	+ if (stripos($tag, 'data-otto-pixel') !== false ||
	3187	+ stripos($tag, 'data-metasync-otto') !== false ||
	3188	+ stripos($tag, 'data-otto') !== false
	3189	+ ) {
	3190	+ $otto_tag = $tag;
	3191	+ break;
	3192	+ }
	3193	+ }
	3194	+ $keeper = $otto_tag ?: $all_tags[0];
	3195	+
	3196	+ # Callback form avoids backreference injection when the keyword list
	3197	+ # contains $ followed by digits.
	3198	+ $first_replaced = false;
	3199	+ $html = preg_replace_callback($pattern, function ($m) use ($keeper, &$first_replaced) {
	3200	+ if (!$first_replaced) {
	3201	+ $first_replaced = true;
	3202	+ return $keeper;
	3203	+ }
	3204	+ return ''; # Remove subsequent duplicates
	3205	+ }, $html);
	3206	+
	3207	+ return $html;
	3208	+ }
	3209	+
	3210	+ /**
3094	3211	* Remove duplicate <link rel="canonical"> tags from HTML.
3095	3212	*
3096	3213	* When OTTO injects a canonical via header_html_insertion and MetaSync's
3097	3214	* SEO output (or WordPress core) has already emitted one, keep only the
		@@ -4037,8 +4154,9 @@
4037	4154
4038	4155	# DEDUPLICATION: Remove duplicate <title>, meta description, OG, Twitter tags, canonical, and JSON-LD schema
4039	4156	$result_html = $this->otto_guard_html($this->deduplicate_title_tags($result_html), $result_html, 'deduplicate_title_tags');
4040	4157	$result_html = $this->otto_guard_html($this->deduplicate_description_tags($result_html), $result_html, 'deduplicate_description_tags');
	4158	+ $result_html = $this->otto_guard_html($this->deduplicate_keywords_tags($result_html), $result_html, 'deduplicate_keywords_tags');
4041	4159	$result_html = $this->otto_guard_html($this->deduplicate_og_twitter_tags($result_html), $result_html, 'deduplicate_og_twitter_tags');
4042	4160	$result_html = $this->otto_guard_html($this->apply_metabox_og_precedence($result_html), $result_html, 'apply_metabox_og_precedence');
4043	4161	$result_html = $this->otto_guard_html($this->apply_custom_seo_precedence($result_html), $result_html, 'apply_custom_seo_precedence');
4044	4162	$result_html = $this->otto_guard_html($this->deduplicate_schema_tags($result_html), $result_html, 'deduplicate_schema_tags');