From 4e3875c0ce7f3b17e3d358dc5ecb1f8bed64546b Mon Sep 17 00:00:00 2001 From: ArthurHoaro Date: Mon, 12 Oct 2020 11:35:55 +0200 Subject: Feature: highlight fulltext search results How it works: 1. when a fulltext search is made, Shaarli looks for the first occurence position of every term matching the search. No change here, but we store these positions in an array, in Bookmark's additionalContent. 2. when formatting bookmarks (through BookmarkFormatter implementation): 1. first we insert specific tokens at every search result positions 2. we format the content (escape HTML, apply markdown, etc.) 3. as a last step, we replace our token with displayable span elements Cons: this tightens coupling between search filters and formatters Pros: it was absolutely necessary not to perform the search twice. this solution has close to no impact on performances. Fixes #205 --- application/bookmark/Bookmark.php | 46 +++++++ application/bookmark/BookmarkFilter.php | 111 ++++++++++++++--- application/formatter/BookmarkDefaultFormatter.php | 132 ++++++++++++++++++++- application/formatter/BookmarkFormatter.php | 79 ++++++++++-- .../formatter/BookmarkMarkdownFormatter.php | 6 +- 5 files changed, 342 insertions(+), 32 deletions(-) (limited to 'application') diff --git a/application/bookmark/Bookmark.php b/application/bookmark/Bookmark.php index fa45d2fc..ea565d1f 100644 --- a/application/bookmark/Bookmark.php +++ b/application/bookmark/Bookmark.php @@ -54,6 +54,9 @@ class Bookmark /** @var bool True if the bookmark can only be seen while logged in */ protected $private; + /** @var mixed[] Available to store any additional content for a bookmark. Currently used for search highlight. */ + protected $additionalContent = []; + /** * Initialize a link from array data. Especially useful to create a Bookmark from former link storage format. * @@ -95,6 +98,8 @@ class Bookmark * - the URL with the permalink * - the title with the URL * + * Also make sure that we do not save search highlights in the datastore. + * * @throws InvalidBookmarkException */ public function validate(): void @@ -112,6 +117,9 @@ class Bookmark if (empty($this->title)) { $this->title = $this->url; } + if (array_key_exists('search_highlight', $this->additionalContent)) { + unset($this->additionalContent['search_highlight']); + } } /** @@ -435,6 +443,44 @@ class Bookmark return $this; } + /** + * Get entire additionalContent array. + * + * @return mixed[] + */ + public function getAdditionalContent(): array + { + return $this->additionalContent; + } + + /** + * Set a single entry in additionalContent, by key. + * + * @param string $key + * @param mixed|null $value Any type of value can be set. + * + * @return $this + */ + public function addAdditionalContentEntry(string $key, $value): self + { + $this->additionalContent[$key] = $value; + + return $this; + } + + /** + * Get a single entry in additionalContent, by key. + * + * @param string $key + * @param mixed|null $default + * + * @return mixed|null can be any type or even null. + */ + public function getAdditionalContentEntry(string $key, $default = null) + { + return array_key_exists($key, $this->additionalContent) ? $this->additionalContent[$key] : $default; + } + /** * Rename a tag in tags list. * diff --git a/application/bookmark/BookmarkFilter.php b/application/bookmark/BookmarkFilter.php index 4232f114..c79386ea 100644 --- a/application/bookmark/BookmarkFilter.php +++ b/application/bookmark/BookmarkFilter.php @@ -201,7 +201,7 @@ class BookmarkFilter 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. @@ -213,8 +213,8 @@ class BookmarkFilter $explodedSearchAnd = array_values(array_filter($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); @@ -234,33 +234,38 @@ class BookmarkFilter } } - // 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; } } @@ -477,4 +482,74 @@ class BookmarkFilter 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 $content; + } } diff --git a/application/formatter/BookmarkDefaultFormatter.php b/application/formatter/BookmarkDefaultFormatter.php index 9d4a0fa0..d58a5e39 100644 --- a/application/formatter/BookmarkDefaultFormatter.php +++ b/application/formatter/BookmarkDefaultFormatter.php @@ -12,10 +12,13 @@ namespace Shaarli\Formatter; */ class BookmarkDefaultFormatter extends BookmarkFormatter { + const SEARCH_HIGHLIGHT_OPEN = '|@@HIGHLIGHT'; + const SEARCH_HIGHLIGHT_CLOSE = 'HIGHLIGHT@@|'; + /** * @inheritdoc */ - public function formatTitle($bookmark) + protected function formatTitle($bookmark) { return escape($bookmark->getTitle()); } @@ -23,10 +26,28 @@ class BookmarkDefaultFormatter extends BookmarkFormatter /** * @inheritdoc */ - public function formatDescription($bookmark) + protected function formatTitleHtml($bookmark) + { + $title = $this->tokenizeSearchHighlightField( + $bookmark->getTitle() ?? '', + $bookmark->getAdditionalContentEntry('search_highlight')['title'] ?? [] + ); + + return $this->replaceTokens(escape($title)); + } + + /** + * @inheritdoc + */ + protected function formatDescription($bookmark) { $indexUrl = ! empty($this->contextData['index_url']) ? $this->contextData['index_url'] : ''; - return format_description(escape($bookmark->getDescription()), $indexUrl); + $description = $this->tokenizeSearchHighlightField( + $bookmark->getDescription() ?? '', + $bookmark->getAdditionalContentEntry('search_highlight')['description'] ?? [] + ); + + return $this->replaceTokens(format_description(escape($description), $indexUrl)); } /** @@ -40,7 +61,27 @@ class BookmarkDefaultFormatter extends BookmarkFormatter /** * @inheritdoc */ - public function formatTagString($bookmark) + protected function formatTagListHtml($bookmark) + { + if (empty($bookmark->getAdditionalContentEntry('search_highlight')['tags'])) { + return $this->formatTagList($bookmark); + } + + $tags = $this->tokenizeSearchHighlightField( + $bookmark->getTagsString(), + $bookmark->getAdditionalContentEntry('search_highlight')['tags'] + ); + $tags = $this->filterTagList(explode(' ', $tags)); + $tags = escape($tags); + $tags = $this->replaceTokensArray($tags); + + return $tags; + } + + /** + * @inheritdoc + */ + protected function formatTagString($bookmark) { return implode(' ', $this->formatTagList($bookmark)); } @@ -48,7 +89,7 @@ class BookmarkDefaultFormatter extends BookmarkFormatter /** * @inheritdoc */ - public function formatUrl($bookmark) + protected function formatUrl($bookmark) { if ($bookmark->isNote() && isset($this->contextData['index_url'])) { return rtrim($this->contextData['index_url'], '/') . '/' . escape(ltrim($bookmark->getUrl(), '/')); @@ -77,6 +118,19 @@ class BookmarkDefaultFormatter extends BookmarkFormatter return escape($bookmark->getUrl()); } + /** + * @inheritdoc + */ + protected function formatUrlHtml($bookmark) + { + $url = $this->tokenizeSearchHighlightField( + $bookmark->getUrl() ?? '', + $bookmark->getAdditionalContentEntry('search_highlight')['url'] ?? [] + ); + + return $this->replaceTokens(escape($url)); + } + /** * @inheritdoc */ @@ -84,4 +138,72 @@ class BookmarkDefaultFormatter extends BookmarkFormatter { return escape($bookmark->getThumbnail()); } + + /** + * Insert search highlight token in provided field content based on a list of search result positions + * + * @param string $fieldContent + * @param array|null $positions List of of search results with 'start' and 'end' positions. + * + * @return string Updated $fieldContent. + */ + protected function tokenizeSearchHighlightField(string $fieldContent, ?array $positions): string + { + if (empty($positions)) { + return $fieldContent; + } + + $insertedTokens = 0; + $tokenLength = strlen(static::SEARCH_HIGHLIGHT_OPEN); + foreach ($positions as $position) { + $position = [ + 'start' => $position['start'] + ($insertedTokens * $tokenLength), + 'end' => $position['end'] + ($insertedTokens * $tokenLength), + ]; + + $content = mb_substr($fieldContent, 0, $position['start']); + $content .= static::SEARCH_HIGHLIGHT_OPEN; + $content .= mb_substr($fieldContent, $position['start'], $position['end'] - $position['start']); + $content .= static::SEARCH_HIGHLIGHT_CLOSE; + $content .= mb_substr($fieldContent, $position['end']); + + $fieldContent = $content; + + $insertedTokens += 2; + } + + return $fieldContent; + } + + /** + * Replace search highlight tokens with HTML highlighted span. + * + * @param string $fieldContent + * + * @return string updated content. + */ + protected function replaceTokens(string $fieldContent): string + { + return str_replace( + [static::SEARCH_HIGHLIGHT_OPEN, static::SEARCH_HIGHLIGHT_CLOSE], + ['', ''], + $fieldContent + ); + } + + /** + * Apply replaceTokens to an array of content strings. + * + * @param string[] $fieldContents + * + * @return array + */ + protected function replaceTokensArray(array $fieldContents): array + { + foreach ($fieldContents as &$entry) { + $entry = $this->replaceTokens($entry); + } + + return $fieldContents; + } } diff --git a/application/formatter/BookmarkFormatter.php b/application/formatter/BookmarkFormatter.php index 0042dafe..e1b7f705 100644 --- a/application/formatter/BookmarkFormatter.php +++ b/application/formatter/BookmarkFormatter.php @@ -2,7 +2,7 @@ namespace Shaarli\Formatter; -use DateTime; +use DateTimeInterface; use Shaarli\Bookmark\Bookmark; use Shaarli\Config\ConfigManager; @@ -11,6 +11,29 @@ use Shaarli\Config\ConfigManager; * * Abstract class processing all bookmark attributes through methods designed to be overridden. * + * List of available formatted fields: + * - id ID + * - shorturl Unique identifier, used in permalinks + * - url URL, can be altered in some way, e.g. passing through an HTTP reverse proxy + * - real_url (legacy) same as `url` + * - url_html URL to be displayed in HTML content (it can contain HTML tags) + * - title Title + * - title_html Title to be displayed in HTML content (it can contain HTML tags) + * - description Description content. It most likely contains HTML tags + * - thumbnail Thumbnail: path to local cache file, false if there is none, null if hasn't been retrieved + * - taglist List of tags (array) + * - taglist_urlencoded List of tags (array) URL encoded: it must be used to create a link to a URL containing a tag + * - taglist_html List of tags (array) to be displayed in HTML content (it can contain HTML tags) + * - tags Tags separated by a single whitespace + * - tags_urlencoded Tags separated by a single whitespace, URL encoded: must be used to create a link + * - sticky Is sticky (bool) + * - private Is private (bool) + * - class Additional CSS class + * - created Creation DateTime + * - updated Last edit DateTime + * - timestamp Creation timestamp + * - updated_timestamp Last edit timestamp + * * @package Shaarli\Formatter */ abstract class BookmarkFormatter @@ -55,13 +78,16 @@ abstract class BookmarkFormatter $out['shorturl'] = $this->formatShortUrl($bookmark); $out['url'] = $this->formatUrl($bookmark); $out['real_url'] = $this->formatRealUrl($bookmark); + $out['url_html'] = $this->formatUrlHtml($bookmark); $out['title'] = $this->formatTitle($bookmark); + $out['title_html'] = $this->formatTitleHtml($bookmark); $out['description'] = $this->formatDescription($bookmark); $out['thumbnail'] = $this->formatThumbnail($bookmark); - $out['urlencoded_taglist'] = $this->formatUrlEncodedTagList($bookmark); $out['taglist'] = $this->formatTagList($bookmark); - $out['urlencoded_tags'] = $this->formatUrlEncodedTagString($bookmark); + $out['taglist_urlencoded'] = $this->formatTagListUrlEncoded($bookmark); + $out['taglist_html'] = $this->formatTagListHtml($bookmark); $out['tags'] = $this->formatTagString($bookmark); + $out['tags_urlencoded'] = $this->formatTagStringUrlEncoded($bookmark); $out['sticky'] = $bookmark->isSticky(); $out['private'] = $bookmark->isPrivate(); $out['class'] = $this->formatClass($bookmark); @@ -69,6 +95,7 @@ abstract class BookmarkFormatter $out['updated'] = $this->formatUpdated($bookmark); $out['timestamp'] = $this->formatCreatedTimestamp($bookmark); $out['updated_timestamp'] = $this->formatUpdatedTimestamp($bookmark); + return $out; } @@ -135,6 +162,18 @@ abstract class BookmarkFormatter return $this->formatUrl($bookmark); } + /** + * Format Url Html: to be displayed in HTML content, it can contains HTML tags. + * + * @param Bookmark $bookmark instance + * + * @return string formatted Url HTML + */ + protected function formatUrlHtml($bookmark) + { + return $this->formatUrl($bookmark); + } + /** * Format Title * @@ -147,6 +186,18 @@ abstract class BookmarkFormatter return $bookmark->getTitle(); } + /** + * Format Title HTML: to be displayed in HTML content, it can contains HTML tags. + * + * @param Bookmark $bookmark instance + * + * @return string formatted Title + */ + protected function formatTitleHtml($bookmark) + { + return $bookmark->getTitle(); + } + /** * Format Description * @@ -190,11 +241,23 @@ abstract class BookmarkFormatter * * @return array formatted Tags */ - protected function formatUrlEncodedTagList($bookmark) + protected function formatTagListUrlEncoded($bookmark) { return array_map('urlencode', $this->filterTagList($bookmark->getTags())); } + /** + * Format Tags HTML: to be displayed in HTML content, it can contains HTML tags. + * + * @param Bookmark $bookmark instance + * + * @return array formatted Tags + */ + protected function formatTagListHtml($bookmark) + { + return $this->formatTagList($bookmark); + } + /** * Format TagString * @@ -214,9 +277,9 @@ abstract class BookmarkFormatter * * @return string formatted TagString */ - protected function formatUrlEncodedTagString($bookmark) + protected function formatTagStringUrlEncoded($bookmark) { - return implode(' ', $this->formatUrlEncodedTagList($bookmark)); + return implode(' ', $this->formatTagListUrlEncoded($bookmark)); } /** @@ -237,7 +300,7 @@ abstract class BookmarkFormatter * * @param Bookmark $bookmark instance * - * @return DateTime instance + * @return DateTimeInterface instance */ protected function formatCreated(Bookmark $bookmark) { @@ -249,7 +312,7 @@ abstract class BookmarkFormatter * * @param Bookmark $bookmark instance * - * @return DateTime instance + * @return DateTimeInterface instance */ protected function formatUpdated(Bookmark $bookmark) { diff --git a/application/formatter/BookmarkMarkdownFormatter.php b/application/formatter/BookmarkMarkdownFormatter.php index 5d244d4c..f7714be9 100644 --- a/application/formatter/BookmarkMarkdownFormatter.php +++ b/application/formatter/BookmarkMarkdownFormatter.php @@ -56,7 +56,10 @@ class BookmarkMarkdownFormatter extends BookmarkDefaultFormatter return parent::formatDescription($bookmark); } - $processedDescription = $bookmark->getDescription(); + $processedDescription = $this->tokenizeSearchHighlightField( + $bookmark->getDescription() ?? '', + $bookmark->getAdditionalContentEntry('search_highlight')['description'] ?? [] + ); $processedDescription = $this->filterProtocols($processedDescription); $processedDescription = $this->formatHashTags($processedDescription); $processedDescription = $this->reverseEscapedHtml($processedDescription); @@ -65,6 +68,7 @@ class BookmarkMarkdownFormatter extends BookmarkDefaultFormatter ->setBreaksEnabled(true) ->text($processedDescription); $processedDescription = $this->sanitizeHtml($processedDescription); + $processedDescription = $this->replaceTokens($processedDescription); if (!empty($processedDescription)) { $processedDescription = '
'. $processedDescription . '
'; -- cgit v1.2.3