namespace Friendica\Model;
use Friendica\Contact\LocalRelationship\Entity\LocalRelationship;
+use Friendica\Content\Image;
use Friendica\Content\Post\Collection\PostMedias;
use Friendica\Content\Post\Entity\PostMedia;
use Friendica\Content\Text\BBCode;
// Check for hashtags in the body and repair or add hashtag links
$item['body'] = self::setHashtags($item['body']);
- $item['language'] = self::getLanguage($item);
-
$notify_type = Delivery::POST;
// Filling item related side tables
}
}
+ $item['language'] = self::getLanguage($item);
+
$inserted = Post::insert($item['uri-id'], $item);
if ($item['gravity'] == self::GRAVITY_PARENT) {
return '';
}
- $languages = self::getLanguageArray(trim($item['title'] . "\n" . $item['body']), 3, $item['uri-id'], $item['author-id']);
+ $languages = self::getLanguageArray($item['title'] . ' ' . ($item['content-warning'] ?? '') . ' ' . $item['body'], 3, $item['uri-id'], $item['author-id']);
if (empty($languages)) {
return '';
}
*/
public static function getLanguageArray(string $body, int $count, int $uri_id = 0, int $author_id = 0): array
{
- $naked_body = BBCode::toSearchText($body, $uri_id);
+ $searchtext = BBCode::toSearchText($body, $uri_id);
- if ((count(explode(' ', $naked_body)) < 10) && (mb_strlen($naked_body) < 30) && $author_id) {
+ if ((count(explode(' ', $searchtext)) < 10) && (mb_strlen($searchtext) < 30) && $author_id) {
$author = Contact::selectFirst(['about'], ['id' => $author_id]);
if (!empty($author['about'])) {
$about = BBCode::toSearchText($author['about'], 0);
- $about = self::getDominantLanguage($about);
- Logger::debug('About field added', ['author' => $author_id, 'body' => $naked_body, 'about' => $about]);
- $naked_body .= ' ' . $about;
+ Logger::debug('About field added', ['author' => $author_id, 'body' => $searchtext, 'about' => $about]);
+ $searchtext .= ' ' . $about;
}
}
- if (empty($naked_body)) {
+ if (empty($searchtext)) {
return [];
}
- $naked_body = self::getDominantLanguage($naked_body);
-
$availableLanguages = DI::l10n()->getAvailableLanguages(true);
$availableLanguages = DI::l10n()->convertForLanguageDetection($availableLanguages);
$ld = new Language(array_keys($availableLanguages));
- $languages = $ld->detect($naked_body)->limit(0, $count)->close() ?: [];
- $data = [
- 'text' => $naked_body,
- 'detected' => $languages,
- 'uri-id' => $uri_id,
- ];
+ $result = [];
- Hook::callAll('detect_languages', $data);
- $languages = $data['detected'];
+ foreach (self::splitByBlocks($searchtext) as $block) {
+ $languages = $ld->detect($block)->limit(0, $count)->close() ?: [];
- return $languages;
+ $data = [
+ 'text' => $block,
+ 'detected' => $languages,
+ 'uri-id' => $uri_id,
+ 'author-id' => $author_id,
+ ];
+ Hook::callAll('detect_languages', $data);
+
+ foreach ($data['detected'] as $language => $quality) {
+ $result[$language] = max($result[$language] ?? 0, $quality * (strlen($block) / strlen($searchtext)));
+ }
+ }
+
+ arsort($result);
+ $result = array_slice($result, 0, $count);
+
+ return $result;
}
/**
- * Check if latin or non latin are dominant in the body and only return the dominant one
+ * Split a string into different unicode blocks
+ * Currently the text is split into the latin and the non latin part.
*
* @param string $body
- * @return string
+ * @return array
*/
- private static function getDominantLanguage(string $body): string
+ private static function splitByBlocks(string $body): array
{
- $latin = '';
- $non_latin = '';
+ if (!class_exists('IntlChar')) {
+ return [$body];
+ }
+
+ $blocks = [];
+ $previous_block = 0;
+
for ($i = 0; $i < mb_strlen($body); $i++) {
$character = mb_substr($body, $i, 1);
- $ord = mb_ord($character);
-
- // We add the most common characters to both strings.
- if (($ord <= 64) || ($ord >= 91 && $ord <= 96) || ($ord >= 123 && $ord <= 191) || in_array($ord, [215, 247]) || ($ord >= 697 && $ord <= 735) || ($ord > 65535)) {
- $latin .= $character;
- $non_latin .= $character;
- } elseif ($ord < 768) {
- $latin .= $character;
+ $previous = ($i > 0) ? mb_substr($body, $i - 1, 1) : '';
+ $next = ($i < mb_strlen($body)) ? mb_substr($body, $i + 1, 1) : '';
+
+ if (!\IntlChar::isalpha($character)) {
+ if (($previous != '') && (\IntlChar::isalpha($previous))) {
+ $previous_block = self::getBlockCode($previous);
+ }
+
+ $block = (($next != '') && \IntlChar::isalpha($next)) ? self::getBlockCode($next) : $previous_block;
+ $blocks[$block] = ($blocks[$block] ?? '') . $character;
} else {
- $non_latin .= $character;
+ $block = self::getBlockCode($character);
+ $blocks[$block] = ($blocks[$block] ?? '') . $character;
+ }
+ }
+
+ foreach (array_keys($blocks) as $key) {
+ $blocks[$key] = trim($blocks[$key]);
+ if (empty($blocks[$key])) {
+ unset($blocks[$key]);
}
}
- return (mb_strlen($latin) > mb_strlen($non_latin)) ? $latin : $non_latin;
+
+ return array_values($blocks);
+ }
+
+ /**
+ * returns the block code for the given character
+ *
+ * @param string $character
+ * @return integer 0 = no alpha character (blank, signs, emojis, ...), 1 = latin character, 2 = character in every other language
+ */
+ private static function getBlockCode(string $character): int
+ {
+ if (!\IntlChar::isalpha($character)) {
+ return 0;
+ }
+ return self::isLatin($character) ? 1 : 2;
+ }
+
+ /**
+ * Checks if the given character is in one of the latin code blocks
+ *
+ * @param string $character
+ * @return boolean
+ */
+ private static function isLatin(string $character): bool
+ {
+ return in_array(\IntlChar::getBlockCode($character), [
+ \IntlChar::BLOCK_CODE_BASIC_LATIN, \IntlChar::BLOCK_CODE_LATIN_1_SUPPLEMENT,
+ \IntlChar::BLOCK_CODE_LATIN_EXTENDED_A, \IntlChar::BLOCK_CODE_LATIN_EXTENDED_B,
+ \IntlChar::BLOCK_CODE_LATIN_EXTENDED_C, \IntlChar::BLOCK_CODE_LATIN_EXTENDED_D,
+ \IntlChar::BLOCK_CODE_LATIN_EXTENDED_E, \IntlChar::BLOCK_CODE_LATIN_EXTENDED_ADDITIONAL
+ ]);
}
public static function getLanguageMessage(array $item): string
$used_languages = '';
foreach (json_decode($item['language'], true) as $language => $reliability) {
- $used_languages .= $iso639->languageByCode1($language) . ' (' . $language . "): " . number_format($reliability, 5) . '\n';
+ $used_languages .= $iso639->nativeByCode1(substr($language, 0, 2)) . ' (' . $iso639->languageByCode1(substr($language, 0, 2)) . ' - ' . $language . "): " . number_format($reliability, 5) . '\n';
}
$used_languages = DI::l10n()->t('Detected languages in this post:\n%s', $used_languages);
return $used_languages;
}
if (!empty($sharedSplitAttachments)) {
- $s = self::addGallery($s, $sharedSplitAttachments['visual'], $item['uri-id']);
+ $s = self::addGallery($s, $sharedSplitAttachments['visual']);
$s = self::addVisualAttachments($sharedSplitAttachments['visual'], $shared_item, $s, true);
$s = self::addLinkAttachment($shared_uri_id ?: $item['uri-id'], $sharedSplitAttachments, $body, $s, true, $quote_shared_links);
$s = self::addNonVisualAttachments($sharedSplitAttachments['additional'], $item, $s, true);
$s = substr($s, 0, $pos);
}
- $s = self::addGallery($s, $itemSplitAttachments['visual'], $item['uri-id']);
+ $s = self::addGallery($s, $itemSplitAttachments['visual']);
$s = self::addVisualAttachments($itemSplitAttachments['visual'], $item, $s, false);
$s = self::addLinkAttachment($item['uri-id'], $itemSplitAttachments, $body, $s, false, $shared_links);
$s = self::addNonVisualAttachments($itemSplitAttachments['additional'], $item, $s, false);
return $hook_data['html'];
}
- /**
- * @param array $images
- * @return string
- * @throws \Friendica\Network\HTTPException\ServiceUnavailableException
- */
- private static function makeImageGrid(array $images): string
- {
- // Image for first column (fc) and second column (sc)
- $images_fc = [];
- $images_sc = [];
-
- for ($i = 0; $i < count($images); $i++) {
- ($i % 2 == 0) ? ($images_fc[] = $images[$i]) : ($images_sc[] = $images[$i]);
- }
-
- return Renderer::replaceMacros(Renderer::getMarkupTemplate('content/image_grid.tpl'), [
- 'columns' => [
- 'fc' => $images_fc,
- 'sc' => $images_sc,
- ],
- ]);
- }
-
/**
* Modify links to pictures to links for the "Fancybox" gallery
*
* @param string $s
* @param PostMedias $PostMedias
- * @param int $uri_id
* @return string
*/
- private static function addGallery(string $s, PostMedias $PostMedias, int $uri_id): string
+ private static function addGallery(string $s, PostMedias $PostMedias): string
{
foreach ($PostMedias as $PostMedia) {
if (!$PostMedia->preview || ($PostMedia->type !== Post\Media::IMAGE)) {
continue;
}
- $s = str_replace('<a href="' . $PostMedia->url . '"', '<a data-fancybox="' . $uri_id . '" href="' . $PostMedia->url . '"', $s);
+ if ($PostMedia->hasDimensions()) {
+ $pattern = '#<a href="' . preg_quote($PostMedia->url) . '">(.*?)"></a>#';
+
+ $s = preg_replace_callback($pattern, function () use ($PostMedia) {
+ return Renderer::replaceMacros(Renderer::getMarkupTemplate('content/image/single_with_height_allocation.tpl'), [
+ '$image' => $PostMedia,
+ '$allocated_height' => $PostMedia->getAllocatedHeight(),
+ '$allocated_max_width' => ($PostMedia->previewWidth ?? $PostMedia->width) . 'px',
+ ]);
+ }, $s);
+ } else {
+ $s = str_replace('<a href="' . $PostMedia->url . '"', '<a data-fancybox="uri-id-' . $PostMedia->uriId . '" href="' . $PostMedia->url . '"', $s);
+ }
}
return $s;
DI::profiler()->startRecording('rendering');
$leading = '';
$trailing = '';
- $images = [];
+ $images = new PostMedias();
// @todo In the future we should make a single for the template engine with all media in it. This allows more flexibilty.
foreach ($PostMedias as $PostMedia) {
continue;
}
- if ($PostMedia->mimetype->type == 'image') {
- $preview_url = DI::baseUrl() . $PostMedia->getPreviewPath($PostMedia->width > $PostMedia->height ? Proxy::SIZE_MEDIUM : Proxy::SIZE_LARGE);
- } elseif ($PostMedia->preview) {
- $preview_url = DI::baseUrl() . $PostMedia->getPreviewPath(Proxy::SIZE_LARGE);
+ if ($PostMedia->mimetype->type == 'image' || $PostMedia->preview) {
+ $preview_size = Proxy::SIZE_MEDIUM;
+ $preview_url = DI::baseUrl() . $PostMedia->getPreviewPath($preview_size);
} else {
+ $preview_size = 0;
$preview_url = '';
}
continue;
}
- $images[] = [
- 'src' => $src_url,
- 'preview' => $preview_url,
- 'attachment' => $PostMedia,
- ];
+ $images[] = $PostMedia->withUrl(new Uri($src_url))->withPreview(new Uri($preview_url), $preview_size);
}
}
- $media = '';
- if (count($images) > 1) {
- $media = self::makeImageGrid($images);
- } elseif (count($images) == 1) {
- $media = Renderer::replaceMacros(Renderer::getMarkupTemplate('content/image.tpl'), [
- '$image' => $images[0],
- ]);
- }
+ $media = Image::getBodyAttachHtml($images);
// On Diaspora posts the attached pictures are leading
if ($item['network'] == Protocol::DIASPORA) {