webapp/bookstack
1
0
Fork 0
mirror of https://github.com/BookStackApp/BookStack.git synced 2025-11-03 02:13:16 +03:00
Code Activity

Make building of search results work for multi-byte encoded characters

Browse Source
This commit is contained in:
Kristian Krastev
2021-12-15 16:29:43 +02:00
parent aa9dafec85
commit d0fd1b7f5c
2 changed files with 20 additions and 20 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
app/Entities/Tools/PageContent.php
View File

@@ -80,7 +80,7 @@ class PageContent
*/ */
protected function extractBase64ImagesFromHtml(string $htmlText): string protected function extractBase64ImagesFromHtml(string $htmlText): string
{ {
if (empty($htmlText) || strpos($htmlText, 'data:image') === false) { if (empty($htmlText) || mb_strpos($htmlText, 'data:image') === false) {
return $htmlText; return $htmlText;
} }

38
app/Entities/Tools/SearchResultsFormatter.php
View File

@@ -57,17 +57,17 @@ class SearchResultsFormatter
protected function highlightTagsContainingTerms(array $tags, array $terms): void protected function highlightTagsContainingTerms(array $tags, array $terms): void
{ {
foreach ($tags as $tag) { foreach ($tags as $tag) {
$tagName = strtolower($tag->name); $tagName = mb_strtolower($tag->name);
$tagValue = strtolower($tag->value); $tagValue = mb_strtolower($tag->value);
foreach ($terms as $term) { foreach ($terms as $term) {
$termLower = strtolower($term); $termLower = mb_strtolower($term);
if (strpos($tagName, $termLower) !== false) { if (mb_strpos($tagName, $termLower) !== false) {
$tag->setAttribute('highlight_name', true); $tag->setAttribute('highlight_name', true);
} }
if (strpos($tagValue, $termLower) !== false) { if (mb_strpos($tagValue, $termLower) !== false) {
$tag->setAttribute('highlight_value', true); $tag->setAttribute('highlight_value', true);
} }
} }
@@ -84,17 +84,17 @@ class SearchResultsFormatter
protected function getMatchPositions(string $text, array $terms): array protected function getMatchPositions(string $text, array $terms): array
{ {
$matchRefs = []; $matchRefs = [];
$text = strtolower($text); $text = mb_strtolower($text);
foreach ($terms as $term) { foreach ($terms as $term) {
$offset = 0; $offset = 0;
$term = strtolower($term); $term = mb_strtolower($term);
$pos = strpos($text, $term, $offset); $pos = mb_strpos($text, $term, $offset);
while ($pos !== false) { while ($pos !== false) {
$end = $pos + strlen($term); $end = $pos + mb_strlen($term);
$matchRefs[$pos] = $end; $matchRefs[$pos] = $end;
$offset = $end; $offset = $end;
$pos = strpos($text, $term, $offset); $pos = mb_strpos($text, $term, $offset);
} }
} }
@@ -141,7 +141,7 @@ class SearchResultsFormatter
*/ */
protected function formatTextUsingMatchPositions(array $matchPositions, string $originalText, int $targetLength): string protected function formatTextUsingMatchPositions(array $matchPositions, string $originalText, int $targetLength): string
{ {
$maxEnd = strlen($originalText); $maxEnd = mb_strlen($originalText);
$fetchAll = ($targetLength === 0); $fetchAll = ($targetLength === 0);
$contextLength = ($fetchAll ? 0 : 32); $contextLength = ($fetchAll ? 0 : 32);
@@ -165,7 +165,7 @@ class SearchResultsFormatter
$contextStart = $start; $contextStart = $start;
// Trims off '$startDiff' number of characters to bring it back to the start // Trims off '$startDiff' number of characters to bring it back to the start
// if this current match zone. // if this current match zone.
$content = substr($content, 0, strlen($content) + $startDiff); $content = mb_substr($content, 0, mb_strlen($content) + $startDiff);
$contentTextLength += $startDiff; $contentTextLength += $startDiff;
} }
@@ -176,16 +176,16 @@ class SearchResultsFormatter
} elseif ($fetchAll) { } elseif ($fetchAll) {
// Or fill in gap since the previous match // Or fill in gap since the previous match
$fillLength = $contextStart - $lastEnd; $fillLength = $contextStart - $lastEnd;
$content .= e(substr($originalText, $lastEnd, $fillLength)); $content .= e(mb_substr($originalText, $lastEnd, $fillLength));
$contentTextLength += $fillLength; $contentTextLength += $fillLength;
} }
// Add our content including the bolded matching text // Add our content including the bolded matching text
$content .= e(substr($originalText, $contextStart, $start - $contextStart)); $content .= e(mb_substr($originalText, $contextStart, $start - $contextStart));
$contentTextLength += $start - $contextStart; $contentTextLength += $start - $contextStart;
$content .= '<strong>' . e(substr($originalText, $start, $end - $start)) . '</strong>'; $content .= '<strong>' . e(mb_substr($originalText, $start, $end - $start)) . '</strong>';
$contentTextLength += $end - $start; $contentTextLength += $end - $start;
$content .= e(substr($originalText, $end, $contextEnd - $end)); $content .= e(mb_substr($originalText, $end, $contextEnd - $end));
$contentTextLength += $contextEnd - $end; $contentTextLength += $contextEnd - $end;
// Update our last end position // Update our last end position
@@ -204,7 +204,7 @@ class SearchResultsFormatter
// Just copy out the content if we haven't moved along anywhere. // Just copy out the content if we haven't moved along anywhere.
if ($lastEnd === 0) { if ($lastEnd === 0) {
$content = e(substr($originalText, 0, $targetLength)); $content = e(mb_substr($originalText, 0, $targetLength));
$contentTextLength = $targetLength; $contentTextLength = $targetLength;
$lastEnd = $targetLength; $lastEnd = $targetLength;
} }
@@ -213,7 +213,7 @@ class SearchResultsFormatter
$remainder = $targetLength - $contentTextLength; $remainder = $targetLength - $contentTextLength;
if ($remainder > 10) { if ($remainder > 10) {
$padEndLength = min($maxEnd - $lastEnd, $remainder); $padEndLength = min($maxEnd - $lastEnd, $remainder);
$content .= e(substr($originalText, $lastEnd, $padEndLength)); $content .= e(mb_substr($originalText, $lastEnd, $padEndLength));
$lastEnd += $padEndLength; $lastEnd += $padEndLength;
$contentTextLength += $padEndLength; $contentTextLength += $padEndLength;
} }
@@ -223,7 +223,7 @@ class SearchResultsFormatter
$firstStart = $firstStart ?: 0; $firstStart = $firstStart ?: 0;
if (!$fetchAll && $remainder > 10 && $firstStart !== 0) { if (!$fetchAll && $remainder > 10 && $firstStart !== 0) {
$padStart = max(0, $firstStart - $remainder); $padStart = max(0, $firstStart - $remainder);
$content = ($padStart === 0 ? '' : '...') . e(substr($originalText, $padStart, $firstStart - $padStart)) . substr($content, 4); $content = ($padStart === 0 ? '' : '...') . e(mb_substr($originalText, $padStart, $firstStart - $padStart)) . mb_substr($content, 4);
} }
// Add ellipsis if we're not at the end // Add ellipsis if we're not at the end