<?php
+declare(strict_types=1);
+
namespace Shaarli\Bookmark;
use Exception;
*
* @throws BookmarkNotFoundException
*/
- public function filter($type, $request, $casesensitive = false, $visibility = 'all', $untaggedonly = false)
- {
+ public function filter(
+ string $type,
+ $request,
+ bool $casesensitive = false,
+ string $visibility = 'all',
+ bool $untaggedonly = false
+ ) {
if (!in_array($visibility, ['all', 'public', 'private'])) {
$visibility = 'all';
}
*
* @return Bookmark[] filtered bookmarks.
*/
- private function noFilter($visibility = 'all')
+ private function noFilter(string $visibility = 'all')
{
if ($visibility === 'all') {
return $this->bookmarks;
*
* @param string $smallHash permalink hash.
*
- * @return array $filtered array containing permalink data.
+ * @return Bookmark[] $filtered array containing permalink data.
*
- * @throws \Shaarli\Bookmark\Exception\BookmarkNotFoundException if the smallhash doesn't match any link.
+ * @throws BookmarkNotFoundException if the smallhash doesn't match any link.
*/
- private function filterSmallHash($smallHash)
+ private function filterSmallHash(string $smallHash)
{
foreach ($this->bookmarks as $key => $l) {
if ($smallHash == $l->getShortUrl()) {
* @param string $searchterms search query.
* @param string $visibility Optional: return only all/private/public bookmarks.
*
- * @return array search results.
+ * @return Bookmark[] search results.
*/
- private function filterFulltext($searchterms, $visibility = 'all')
+ private function filterFulltext(string $searchterms, string $visibility = 'all')
{
if (empty($searchterms)) {
return $this->noFilter($visibility);
}
- $filtered = array();
+ $filtered = [];
$search = mb_convert_case(html_entity_decode($searchterms), MB_CASE_LOWER, 'UTF-8');
$exactRegex = '/"([^"]+)"/';
// Retrieve exact search terms.
$explodedSearchAnd = array_values(array_filter($explodedSearchAnd));
// Filter excluding terms and update andSearch.
- $excludeSearch = array();
- $andSearch = array();
+ $excludeSearch = [];
+ $andSearch = [];
foreach ($explodedSearchAnd as $needle) {
if ($needle[0] == '-' && strlen($needle) > 1) {
$excludeSearch[] = substr($needle, 1);
}
}
- // Concatenate link fields to search across fields.
- // Adds a '\' separator for exact search terms.
- $content = mb_convert_case($link->getTitle(), MB_CASE_LOWER, 'UTF-8') .'\\';
- $content .= mb_convert_case($link->getDescription(), MB_CASE_LOWER, 'UTF-8') .'\\';
- $content .= mb_convert_case($link->getUrl(), MB_CASE_LOWER, 'UTF-8') .'\\';
- $content .= mb_convert_case($link->getTagsString(), MB_CASE_LOWER, 'UTF-8') .'\\';
+ $lengths = [];
+ $content = $this->buildFullTextSearchableLink($link, $lengths);
// Be optimistic
$found = true;
+ $foundPositions = [];
// First, we look for exact term search
- for ($i = 0; $i < count($exactSearch) && $found; $i++) {
- $found = strpos($content, $exactSearch[$i]) !== false;
- }
-
- // Iterate over keywords, if keyword is not found,
+ // Then iterate over keywords, if keyword is not found,
// no need to check for the others. We want all or nothing.
- for ($i = 0; $i < count($andSearch) && $found; $i++) {
- $found = strpos($content, $andSearch[$i]) !== false;
+ foreach ([$exactSearch, $andSearch] as $search) {
+ for ($i = 0; $i < count($search) && $found !== false; $i++) {
+ $found = mb_strpos($content, $search[$i]);
+ if ($found === false) {
+ break;
+ }
+
+ $foundPositions[] = ['start' => $found, 'end' => $found + mb_strlen($search[$i])];
+ }
}
// Exclude terms.
- for ($i = 0; $i < count($excludeSearch) && $found; $i++) {
+ for ($i = 0; $i < count($excludeSearch) && $found !== false; $i++) {
$found = strpos($content, $excludeSearch[$i]) === false;
}
- if ($found) {
+ if ($found !== false) {
+ $link->addAdditionalContentEntry(
+ 'search_highlight',
+ $this->postProcessFoundPositions($lengths, $foundPositions)
+ );
+
$filtered[$id] = $link;
}
}
*
* @return string generated regex fragment
*/
- private static function tag2regex($tag)
+ private static function tag2regex(string $tag): string
{
$len = strlen($tag);
if (!$len || $tag === "-" || $tag === "*") {
* You can specify one or more tags, separated by space or a comma, e.g.
* print_r($mydb->filterTags('linux programming'));
*
- * @param string $tags list of tags separated by commas or blank spaces.
- * @param bool $casesensitive ignore case if false.
- * @param string $visibility Optional: return only all/private/public bookmarks.
+ * @param string|array $tags list of tags, separated by commas or blank spaces if passed as string.
+ * @param bool $casesensitive ignore case if false.
+ * @param string $visibility Optional: return only all/private/public bookmarks.
*
- * @return array filtered bookmarks.
+ * @return Bookmark[] filtered bookmarks.
*/
- public function filterTags($tags, $casesensitive = false, $visibility = 'all')
+ public function filterTags($tags, bool $casesensitive = false, string $visibility = 'all')
{
// get single tags (we may get passed an array, even though the docs say different)
$inputTags = $tags;
*
* @param string $visibility return only all/private/public bookmarks.
*
- * @return array filtered bookmarks.
+ * @return Bookmark[] filtered bookmarks.
*/
- public function filterUntagged($visibility)
+ public function filterUntagged(string $visibility)
{
$filtered = [];
foreach ($this->bookmarks as $key => $link) {
* @param string $day day to filter.
* @param string $visibility return only all/private/public bookmarks.
- * @return array all link matching given day.
+ * @return Bookmark[] all link matching given day.
*
* @throws Exception if date format is invalid.
*/
- public function filterDay($day, $visibility)
+ public function filterDay(string $day, string $visibility)
{
if (!checkDateFormat('Ymd', $day)) {
throw new Exception('Invalid date format');
* @param string $tags string containing a list of tags.
* @param bool $casesensitive will convert everything to lowercase if false.
*
- * @return array filtered tags string.
+ * @return string[] filtered tags string.
*/
- public static function tagsStrToArray($tags, $casesensitive)
+ public static function tagsStrToArray(string $tags, bool $casesensitive): array
{
// We use UTF-8 conversion to handle various graphemes (i.e. cyrillic, or greek)
$tagsOut = $casesensitive ? $tags : mb_convert_case($tags, MB_CASE_LOWER, 'UTF-8');
return preg_split('/\s+/', $tagsOut, -1, PREG_SPLIT_NO_EMPTY);
}
+
+ /**
+ * This method finalize the content of the foundPositions array,
+ * by associated all search results to their associated bookmark field,
+ * making sure that there is no overlapping results, etc.
+ *
+ * @param array $fieldLengths Start and end positions of every bookmark fields in the aggregated bookmark content.
+ * @param array $foundPositions Positions where the search results were found in the aggregated content.
+ *
+ * @return array Updated $foundPositions, by bookmark field.
+ */
+ protected function postProcessFoundPositions(array $fieldLengths, array $foundPositions): array
+ {
+ // Sort results by starting position ASC.
+ usort($foundPositions, function (array $entryA, array $entryB): int {
+ return $entryA['start'] > $entryB['start'] ? 1 : -1;
+ });
+
+ $out = [];
+ $currentMax = -1;
+ foreach ($foundPositions as $foundPosition) {
+ // we do not allow overlapping highlights
+ if ($foundPosition['start'] < $currentMax) {
+ continue;
+ }
+
+ $currentMax = $foundPosition['end'];
+ foreach ($fieldLengths as $part => $length) {
+ if ($foundPosition['start'] < $length['start'] || $foundPosition['start'] > $length['end']) {
+ continue;
+ }
+
+ $out[$part][] = [
+ 'start' => $foundPosition['start'] - $length['start'],
+ 'end' => $foundPosition['end'] - $length['start'],
+ ];
+ break;
+ }
+ }
+
+ return $out;
+ }
+
+ /**
+ * Concatenate link fields to search across fields. Adds a '\' separator for exact search terms.
+ * Also populate $length array with starting and ending positions of every bookmark field
+ * inside concatenated content.
+ *
+ * @param Bookmark $link
+ * @param array $lengths (by reference)
+ *
+ * @return string Lowercase concatenated fields content.
+ */
+ protected function buildFullTextSearchableLink(Bookmark $link, array &$lengths): string
+ {
+ $content = mb_convert_case($link->getTitle(), MB_CASE_LOWER, 'UTF-8') .'\\';
+ $content .= mb_convert_case($link->getDescription(), MB_CASE_LOWER, 'UTF-8') .'\\';
+ $content .= mb_convert_case($link->getUrl(), MB_CASE_LOWER, 'UTF-8') .'\\';
+ $content .= mb_convert_case($link->getTagsString(), MB_CASE_LOWER, 'UTF-8') .'\\';
+
+ $lengths['title'] = ['start' => 0, 'end' => mb_strlen($link->getTitle())];
+ $nextField = $lengths['title']['end'] + 1;
+ $lengths['description'] = ['start' => $nextField, 'end' => $nextField + mb_strlen($link->getDescription())];
+ $nextField = $lengths['description']['end'] + 1;
+ $lengths['url'] = ['start' => $nextField, 'end' => $nextField + mb_strlen($link->getUrl())];
+ $nextField = $lengths['url']['end'] + 1;
+ $lengths['tags'] = ['start' => $nextField, 'end' => $nextField + mb_strlen($link->getTagsString())];
+
+ return $content;
+ }
}
projects / github / shaarli / Shaarli.git / blobdiff