standardebooks/web
1
0
Fork 0
mirror of https://github.com/standardebooks/web.git synced 2025-07-16 11:26:37 -04:00
Code Issues Projects Releases Packages Wiki Activity Actions

Keep words with apostrophes intact when searching

Browse source 
The code will convert a search for:

`Whistler's Mother`

into:

`\b(whistler's|mother)\b`

Before this fix, it was:

`\b(whistler|s|mother)\b`
This commit is contained in:
Mike Colagrosso 2024-01-21 14:41:23 -07:00 • committed by Alex Cabal
parent d8360b28ca
commit 8a8692619c
1 changed files with 7 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

10
lib/Library.php
View file

@ -213,10 +213,14 @@ class Library{
$orderBy = 'art.CompletedYear desc';
}
// Remove diacritics and non-alphanumeric characters
$query = trim(preg_replace('|[^a-zA-Z0-9 ]|ius', ' ', Formatter::RemoveDiacritics($query ?? '')));
// Remove diacritics and non-alphanumeric characters, but preserve apostrophes
$query = trim(preg_replace('|[^a-zA-Z0-9\' ]|ius', ' ', Formatter::RemoveDiacritics($query ?? '')));
$tokenizedQuery = '\b(' . implode('|', preg_split('/\b|\s+/', $query, -1, PREG_SPLIT_NO_EMPTY)) . ')\b';
// Split the query on word boundaries followed by spaces. This keeps words with apostrophes intact.
$tokenArray = preg_split('/\b\s+/', $query, -1, PREG_SPLIT_NO_EMPTY);
// Join the tokens with '|' to search on any token, but add word boundaries to force the full token to match
$tokenizedQuery = '\b(' . implode('|', $tokenArray) . ')\b';
$params[] = $tokenizedQuery; // art.Name
$params[] = $tokenizedQuery; // art.EbookWwwFilesystemPath