Contao 5.3 compatibility

[CMSworker]

Context
Contao version: 5.3.31
Bundle version: 3.0.1
PHP version: 8.2+

Description

TL_ROOT was removed in Contao 5

In the current tag version 3.0.1 of contao-search-bundle this constant is still used here:

contao-search-bundle/src/Indexer/PdfSearchIndexer.php

Line 73 in 1e0ed10

if (!file_exists(TL_ROOT . '/' . $strFile)) {

Because of this the indexing of PDF files doesn't work in Contao 5 currently. When rebuilding the search index this error message comes in the crawl_debug_log.csv:

"Forwarded to the search indexer. Did not index because of the following reason: Could not add a search index entry: Undefined constant ""HeimrichHannot\SearchBundle\Indexer\TL_ROOT"""

It can be solved by replacing TL_ROOT with System::getContainer()->getParameter('kernel.project_dir')

But I guess you've prepared another update already, because the source code in the current master branch looks different to the latest stable tag release?

contao-search-bundle/src/Indexer/PdfSearchIndexer.php

Line 72 in f6d3ff9

if (!file_exists($this->parameterBag->get('kernel.project_dir') . '/' . $strFile)) {

[koertho]

Hi, your're right, the CI tools already fixed the issue, I will tag it :)

[CMSworker]

Thank you. I noticed two more things while testing it in Contao 5.3:

(1) The PDF is missing in the search result if the search term is in the file name only (not within the PDF file content).

I think maybe this could be fixed by adding the file name at the beginning of the $strContent variable? For example here:

contao-search-bundle/src/Indexer/PdfSearchIndexer.php

Line 159 in 05fdad5

$strContent = trim((string) preg_replace('/ +/', ' ', StringUtil::decodeEntities($strContent)));

by adding the title:

$strContent = $arrMeta['title'].' '.trim((string) preg_replace('/ +/', ' ', StringUtil::decodeEntities($strContent)));

(2) Some PDF files are skipped by the crawler.

In the crawl_debug_log.csv (available after rebuilding the search index in the backend) there are error messages like this for PDF files:

Do not request because when the crawl URI was found, the "type" attribute was present and did not contain "text/html"

I guess this comes from here:

https://github.com/contao/contao/blob/c1dc859a3cefd71c6346cc2591a642b6954f5fc3/core-bundle/src/Crawl/Escargot/Subscriber/SearchIndexSubscriber.php#L97

Probably the type application/pdf have to be added, but I don't know how this could be done in the contao-search-bundle

[koertho]

This bundle is not really tested with contao 5, as we're just starting migrating to contao 5. I think there is some potential when we use the potential of the new search indexer, but the implementation uses still the old one before escargot. Can't promise a soon bigger refactoring, we have bundles with higher priority :/