diff --git a/bin/envs/cli-setup.sh b/bin/envs/cli-setup.sh
index e1a98de0c5..4d1c342cd2 100755
--- a/bin/envs/cli-setup.sh
+++ b/bin/envs/cli-setup.sh
@@ -21,6 +21,13 @@ init_environment(){
echo "Installing Neve theme from $NEVE_LOCATION"
wp --allow-root theme install --activate $NEVE_LOCATION
wp --allow-root option update fresh_site 0
+
+ # Activating Neve remaps widgets by sidebar id, and 'blog-sidebar' does not
+ # match the previous theme's, so the default widgets can be left unassigned.
+ if [ "$(wp --allow-root widget list blog-sidebar --format=count)" = "0" ]; then
+ echo "Populating blog-sidebar"
+ wp --allow-root widget add block blog-sidebar --content=''
+ fi
echo "Installing Theme API Plugin"
wp --allow-root plugin install https://github.com/codeinwp/wp-thememods-api/archive/refs/heads/main.zip --force --activate
}
diff --git a/e2e-tests/specs/customizer/layout/blog-archive-settings.spec.ts b/e2e-tests/specs/customizer/layout/blog-archive-settings.spec.ts
index 17df82fc77..1cbcf2c52e 100755
--- a/e2e-tests/specs/customizer/layout/blog-archive-settings.spec.ts
+++ b/e2e-tests/specs/customizer/layout/blog-archive-settings.spec.ts
@@ -2,6 +2,22 @@ import { test, expect } from '@playwright/test';
import { setCustomizeSettings } from '../../../utils';
import data from '../../../fixtures/customizer/layout/blog-archive-setting-setup.json';
+const SEARCH_TERM = 'nvexcerpthtmlfixture';
+const POST_SLUG = 'nv-excerpt-html-fixture';
+const LINK_HREF = 'https://example.com/neve-excerpt-link';
+const EXCERPT = `Neve keeps this excerpt link visible while trimming the rest sentinelword away.`;
+
+const excerptHtmlSettings = {
+ neve_blog_archive_layout: 'default',
+ neve_post_excerpt_length: 8,
+ neve_post_content_ordering: '["title-meta","excerpt"]',
+};
+
+const excerptHtmlCutSettings = {
+ ...excerptHtmlSettings,
+ neve_post_excerpt_length: 4,
+};
+
test.describe('Blog/Archive 1 / Default Layout', () => {
test.beforeAll(async ({ request, baseURL }) => {
await setCustomizeSettings('defaultLayout', data.archive1, {
@@ -283,3 +299,97 @@ test.describe('Blog/Archive 4 / Default Layout', () => {
).toEqual(0);
});
});
+
+test.describe('Blog/Archive / Excerpt markup', () => {
+ test.describe.configure({ mode: 'serial' });
+
+ let postId: number;
+
+ test.beforeAll(async ({ request, baseURL }) => {
+ await setCustomizeSettings('excerptHtml', excerptHtmlSettings, {
+ request,
+ baseURL,
+ });
+ await setCustomizeSettings('excerptHtmlCut', excerptHtmlCutSettings, {
+ request,
+ baseURL,
+ });
+
+ // A run killed before afterAll leaves the post behind; drop it by slug so
+ // the search page has exactly one result either way.
+ const stale = await request.get(
+ baseURL + `/wp-json/wp/v2/posts?slug=${POST_SLUG}&status=any`
+ );
+ expect(stale.ok()).toBeTruthy();
+ for (const post of await stale.json()) {
+ const staleDeleteResponse = await request.delete(
+ baseURL + `/wp-json/wp/v2/posts/${post.id}?force=true`
+ );
+ expect(staleDeleteResponse.ok()).toBeTruthy();
+ }
+
+ const response = await request.post(baseURL + '/wp-json/wp/v2/posts', {
+ data: {
+ title: `Excerpt markup ${SEARCH_TERM}`,
+ slug: POST_SLUG,
+ content: `Body copy for ${SEARCH_TERM}.`,
+ excerpt: EXCERPT,
+ status: 'publish',
+ // Dated far back so the post lands on the last archive page and
+ // leaves the other archive specs alone.
+ date: '2001-01-01T00:00:00',
+ },
+ });
+ expect(response.ok()).toBeTruthy();
+ postId = (await response.json()).id;
+ });
+
+ test.afterAll(async ({ request, baseURL }) => {
+ if (postId) {
+ const deleteResponse = await request.delete(
+ baseURL + `/wp-json/wp/v2/posts/${postId}?force=true`
+ );
+ expect(deleteResponse.ok()).toBeTruthy();
+ }
+ });
+
+ test('Trimming the excerpt keeps its HTML', async ({ page }) => {
+ await page.goto(`/?s=${SEARCH_TERM}&test_name=excerptHtml`);
+
+ const excerpt = page.locator('article.post .excerpt-wrap');
+ await expect(excerpt).toHaveCount(1);
+
+ const link = excerpt.locator(`a[href="${LINK_HREF}"]`);
+ await expect(link).toBeVisible();
+ await expect(link).toHaveText('excerpt link');
+ await expect(excerpt.locator('strong')).toHaveText('keeps');
+
+ const text = await excerpt.innerText();
+ // The trim still happens: the last kept word is in, the next one is out.
+ expect(text).toContain('trimming');
+ expect(text).not.toContain('sentinelword');
+ // Markup is rendered, not printed as escaped text.
+ expect(text).not.toContain(' {
+ await page.goto(`/?s=${SEARCH_TERM}&test_name=excerptHtmlCut`);
+
+ const excerpt = page.locator('article.post .excerpt-wrap');
+ await expect(excerpt).toHaveCount(1);
+
+ // Word 4 is the first half of the link text and word 5 is past the cut, so
+ // only an anchor closed by the trim can render as a link at all.
+ const link = excerpt.locator(`a[href="${LINK_HREF}"]`);
+ await expect(link).toBeVisible();
+ await expect(link).toHaveText('excerpt');
+ // The read more marker follows the link instead of being swallowed by it.
+ await expect(link).not.toContainText('…');
+ await expect(excerpt.locator('strong')).toHaveText('keeps');
+
+ const text = await excerpt.innerText();
+ expect(text).toContain('…');
+ expect(text).not.toContain('visible');
+ expect(text).not.toContain('trim_words_keep_html( get_the_excerpt( $post_id ), $length, $excerpt_more );
return apply_filters( 'the_excerpt', $content );
}
@@ -89,6 +89,147 @@ private function get_excerpt( $length = 25, $post_id = null ) {
return apply_filters( 'the_excerpt', $content );
}
+ /**
+ * Trim words while preserving HTML markup.
+ *
+ * Similar to wp_trim_words(), but preserves HTML tags.
+ *
+ * @param string $text HTML content to trim.
+ * @param int $num_words Maximum number of words.
+ * @param string $more String to append when the content is trimmed.
+ *
+ * @return string
+ */
+ private function trim_words_keep_html( $text, $num_words, $more = '...' ) {
+ $num_words = (int) $num_words;
+ $trimmed = $this->trim_markup( $text, $num_words, $more );
+
+ return apply_filters( 'wp_trim_words', $trimmed, $num_words, $more, $text );
+ }
+
+ /**
+ * Trim a text to a number of words, leaving the markup around them in place.
+ *
+ * @param string $text HTML content to trim.
+ * @param int $num_words Maximum number of words.
+ * @param string $more String to append when the content is trimmed.
+ *
+ * @return string
+ */
+ private function trim_markup( $text, $num_words, $more ) {
+ if ( $num_words <= 0 || '' === $text ) {
+ return '';
+ }
+
+ // `wp_get_word_count_type()` is WP 6.2+; older installs count words.
+ $count_type = function_exists( 'wp_get_word_count_type' ) ? wp_get_word_count_type() : 'words';
+
+ // Some locales budget characters rather than words, as `wp_trim_words()` does.
+ $count_chars = 0 === strpos( $count_type, 'characters' )
+ && 1 === preg_match( '/^utf\-?8$/i', get_option( 'blog_charset' ) );
+
+ $tokens = preg_split(
+ '/(<[^>]*>)/',
+ $text,
+ -1,
+ PREG_SPLIT_DELIM_CAPTURE | PREG_SPLIT_NO_EMPTY
+ );
+
+ if ( ! is_array( $tokens ) ) {
+ return wp_trim_words( $text, $num_words, $more );
+ }
+
+ $output = '';
+ $remaining = $num_words;
+ $cut = false;
+
+ // Markup and whitespace are held back until a kept word follows them, so a
+ // tag opened right at the cut does not leave an empty element behind.
+ $pending = '';
+ $pending_cost = 0;
+
+ foreach ( $tokens as $token ) {
+ // Tokens are split on this same pattern, so a match is one of the tags.
+ if ( preg_match( '#^<[^>]*>$#', $token ) ) {
+ $pending .= $token;
+
+ continue;
+ }
+
+ $parts = preg_split(
+ '/(\s+)/u',
+ $token,
+ -1,
+ PREG_SPLIT_DELIM_CAPTURE | PREG_SPLIT_NO_EMPTY
+ );
+
+ if ( ! is_array( $parts ) ) {
+ $output .= $pending . $token;
+ $pending = '';
+ $pending_cost = 0;
+
+ continue;
+ }
+
+ foreach ( $parts as $part ) {
+ if ( '' === trim( $part ) ) {
+ // Core collapses each run of whitespace to a single space.
+ $pending .= $count_chars ? ' ' : $part;
+ $pending_cost += $count_chars ? 1 : 0;
+
+ continue;
+ }
+
+ $remaining -= $pending_cost;
+
+ if ( $remaining <= 0 ) {
+ $cut = true;
+ break;
+ }
+
+ $chars = $count_chars ? $this->split_characters( $part ) : array( $part );
+
+ // Character locales can cut part way through a run of text.
+ if ( count( $chars ) > $remaining ) {
+ $output .= $pending . implode( '', array_slice( $chars, 0, $remaining ) );
+ $cut = true;
+ break;
+ }
+
+ $output .= $pending . $part;
+ $pending = '';
+ $pending_cost = 0;
+ $remaining -= count( $chars );
+ }
+
+ if ( $cut ) {
+ break;
+ }
+ }
+
+ if ( ! $cut ) {
+ return $text;
+ }
+
+ // Close any tags that are still open so the resulting HTML remains valid.
+ return force_balance_tags( $output ) . $more;
+ }
+
+ /**
+ * Split a string into its characters.
+ *
+ * @param string $text Text to split.
+ *
+ * @return string[]
+ */
+ private function split_characters( $text ) {
+ if ( ! preg_match_all( '/./u', $text, $matches ) ) {
+ return array();
+ }
+
+ return $matches[0];
+ }
+
/**
* Get the excerpt length option casted as `int`.
*
diff --git a/tests/test-neve-excerpt-html.php b/tests/test-neve-excerpt-html.php
new file mode 100644
index 0000000000..4e9e6872bd
--- /dev/null
+++ b/tests/test-neve-excerpt-html.php
@@ -0,0 +1,342 @@
+setAccessible( true );
+
+ return $method->invoke( $excerpt, $text, $num_words, $this->more );
+ }
+
+ /**
+ * Trim a text with the locale counting characters, as CJK locales do.
+ *
+ * @param string $text Text to trim.
+ * @param int $limit Number of characters to keep.
+ *
+ * @return string
+ */
+ private function trim_by_characters( $text, $limit ) {
+ global $wp_locale;
+
+ $previous = $wp_locale->word_count_type;
+ $wp_locale->word_count_type = 'characters_including_spaces';
+
+ $output = $this->trim( $text, $limit );
+
+ $wp_locale->word_count_type = $previous;
+
+ return $output;
+ }
+
+ /**
+ * A link before the cut keeps both its markup and its text.
+ */
+ public function test_keeps_link_when_cut_falls_after_it() {
+ $text = 'Read the full announcement and then a long tail of words';
+
+ $this->assertSame(
+ 'Read the full announcement and then a long' . $this->more,
+ $this->trim( $text, 8 )
+ );
+ }
+
+ /**
+ * A cut inside a link closes it and leaves the marker outside.
+ */
+ public function test_closes_inline_tag_left_open_by_the_cut() {
+ $text = 'Read the full release announcement tail words';
+
+ $this->assertSame(
+ 'Read the full release' . $this->more,
+ $this->trim( $text, 4 )
+ );
+ }
+
+ /**
+ * A text shorter than the limit is returned untouched, marker included.
+ */
+ public function test_returns_short_text_unchanged() {
+ $text = 'Short bold excerpt';
+
+ $this->assertSame( $text, $this->trim( $text, 25 ) );
+ $this->assertSame( $text, $this->trim( $text, 3 ) );
+ }
+
+ /**
+ * A tag opened right at the cut must not leave an empty element behind.
+ */
+ public function test_drops_element_opened_at_the_cut() {
+ $this->assertSame(
+ 'one
two
' . $this->more,
+ $this->trim( 'one
two
three
', 2 )
+ );
+ }
+
+ /**
+ * The same, for a list: an empty `li` would render as a stray bullet.
+ */
+ public function test_drops_list_item_opened_at_the_cut() {
+ $this->assertSame(
+ '' . $this->more,
+ $this->trim( '', 2 )
+ );
+ }
+
+ /**
+ * Nested tags are closed at the cut.
+ */
+ public function test_closes_nested_tags() {
+ $this->assertSame(
+ 'One two three' . $this->more,
+ $this->trim( 'One two three four five six', 3 )
+ );
+ }
+
+ /**
+ * A tag nested in itself is unwound rather than left doubled.
+ */
+ public function test_unwinds_a_tag_nested_in_itself() {
+ $this->assertSame(
+ 'Nested one two' . $this->more,
+ $this->trim( 'Nested one two three four five', 3 )
+ );
+ }
+
+ /**
+ * Markup the author never closed is closed at the cut anyway.
+ */
+ public function test_closes_unbalanced_author_markup() {
+ $this->assertSame(
+ 'Unclosed link never closed' . $this->more,
+ $this->trim( 'Unclosed link never closed and more words', 4 )
+ );
+ }
+
+ /**
+ * A closing tag with nothing to close is dropped.
+ */
+ public function test_drops_orphan_closing_tag() {
+ $this->assertSame(
+ 'Stray closing tag here' . $this->more,
+ $this->trim( 'Stray closing tag here and more', 4 )
+ );
+ }
+
+ /**
+ * Void elements are neither counted as words nor closed.
+ */
+ public function test_leaves_void_elements_alone() {
+ $this->assertSame(
+ 'Image 
after
text words' . $this->more,
+ $this->trim( 'Image after
text words more', 4 )
+ );
+ }
+
+ /**
+ * A hyphenated custom element is a tag, not three words of text.
+ */
+ public function test_treats_custom_elements_as_markup() {
+ $this->assertSame(
+ 'Custom element text' . $this->more,
+ $this->trim( 'Custom element text here more words', 3 )
+ );
+
+ $this->assertSame(
+ 'Cut inside' . $this->more,
+ $this->trim( 'Cut inside element tail', 2 )
+ );
+ }
+
+ /**
+ * A tag name carrying a colon is markup too, and is closed at the cut.
+ */
+ public function test_treats_namespaced_elements_as_markup() {
+ $this->assertSame(
+ 'Vector icon label' . $this->more,
+ $this->trim( 'Vector icon label here more words', 3 )
+ );
+
+ $this->assertSame(
+ 'Cut inside' . $this->more,
+ $this->trim( 'Cut inside icon tail', 2 )
+ );
+ }
+
+ /**
+ * Comments carry no words, so their contents must not spend the budget.
+ */
+ public function test_does_not_count_comments_as_words() {
+ $this->assertSame(
+ 'bold text more' . $this->more,
+ $this->trim( 'bold text more words here', 3 )
+ );
+ }
+
+ /**
+ * Whitespace between kept words is preserved rather than normalised.
+ */
+ public function test_preserves_whitespace_between_kept_words() {
+ $this->assertSame(
+ "Line\nbreaks and\ttabs" . $this->more,
+ $this->trim( "Line\nbreaks and\ttabs here plus more", 4 )
+ );
+ }
+
+ /**
+ * A zero length keeps nothing.
+ */
+ public function test_zero_length_returns_empty_string() {
+ $this->assertSame( '', $this->trim( 'a b c', 0 ) );
+ $this->assertSame( '', $this->trim( '', 8 ) );
+ }
+
+ /**
+ * Text without markup is trimmed exactly as `wp_trim_words()` would.
+ */
+ public function test_matches_core_for_text_without_markup() {
+ $text = 'one two three four five six';
+
+ foreach ( array( 1, 3, 5, 6, 25 ) as $limit ) {
+ $this->assertSame(
+ wp_trim_words( $text, $limit, $this->more ),
+ $this->trim( $text, $limit ),
+ 'Diverged from wp_trim_words() at length ' . $limit
+ );
+ }
+ }
+
+ /**
+ * Character locales keep their markup rather than falling back to core.
+ */
+ public function test_keeps_link_in_character_counting_locales() {
+ $text = '你好世界链接文字更多内容';
+
+ $this->assertSame(
+ '你好世界链接' . $this->more,
+ $this->trim_by_characters( $text, 6 )
+ );
+ }
+
+ /**
+ * Character locales spend the budget per character, whitespace included.
+ */
+ public function test_matches_core_in_character_counting_locales() {
+ $cases = array(
+ array( 'ab cd efgh', 5 ),
+ array( 'aa bb cc dd', 4 ),
+ array( 'a b c d e f', 3 ),
+ array( '你好世界链接文字', 6 ),
+ );
+
+ foreach ( $cases as list( $text, $limit ) ) {
+ global $wp_locale;
+
+ $previous = $wp_locale->word_count_type;
+ $wp_locale->word_count_type = 'characters_including_spaces';
+ $core = wp_trim_words( $text, $limit, $this->more );
+ $wp_locale->word_count_type = $previous;
+
+ $this->assertSame(
+ $core,
+ $this->trim_by_characters( $text, $limit ),
+ 'Diverged from wp_trim_words() for "' . $text . '" at length ' . $limit
+ );
+ }
+ }
+
+ /**
+ * A text within the character budget is returned untouched.
+ */
+ public function test_character_locales_return_short_text_unchanged() {
+ $text = '短文字';
+
+ $this->assertSame( $text, $this->trim_by_characters( $text, 25 ) );
+ }
+
+ /**
+ * The `wp_trim_words` filter runs on the result, as it does in core.
+ */
+ public function test_applies_the_wp_trim_words_filter() {
+ $calls = array();
+
+ $capture = function ( $trimmed, $num_words, $more, $original ) use ( &$calls ) {
+ $calls[] = compact( 'num_words', 'more', 'original' );
+
+ return strtoupper( $trimmed );
+ };
+
+ add_filter( 'wp_trim_words', $capture, 10, 4 );
+
+ $text = 'one two three four five';
+
+ // Both the trimmed and the untrimmed return paths go through the filter.
+ $this->assertSame( 'ONE TWO THREE' . strtoupper( $this->more ), $this->trim( $text, 3 ) );
+ $this->assertSame( strtoupper( $text ), $this->trim( $text, 25 ) );
+
+ remove_filter( 'wp_trim_words', $capture, 10 );
+
+ $this->assertCount( 2, $calls );
+ $this->assertSame( 3, $calls[0]['num_words'] );
+ $this->assertSame( $this->more, $calls[0]['more'] );
+ $this->assertSame( $text, $calls[0]['original'] );
+ }
+
+ /**
+ * The rendered archive excerpt keeps the link of a hand-written excerpt.
+ */
+ public function test_rendered_excerpt_keeps_the_link() {
+ $post_id = self::factory()->post->create(
+ array(
+ 'post_content' => 'Body content, not used by this assertion.',
+ 'post_excerpt' => 'Read the full announcement and then a long tail of words to drop',
+ )
+ );
+
+ $GLOBALS['post'] = get_post( $post_id );
+ setup_postdata( $GLOBALS['post'] );
+ set_theme_mod( 'neve_post_excerpt_length', 8 );
+
+ $partial = new \Neve\Views\Partials\Excerpt();
+
+ ob_start();
+ $partial->render_post_excerpt( 'index', $post_id );
+ $output = (string) ob_get_clean();
+
+ remove_theme_mod( 'neve_post_excerpt_length' );
+ wp_reset_postdata();
+
+ $this->assertStringContainsString( 'full announcement', $output );
+ $this->assertStringNotContainsString( 'to drop', $output );
+ }
+}