mummert/meilisearch-bundle
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Bugfix

Browse Source
This commit is contained in:
Jürgen Mummert
2025-12-28 11:51:53 +01:00
parent 0fd119f873
commit 2a22253f18
2 changed files with 73 additions and 72 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/Service/OfficeIndexService.php
+28 -4
View File
@@ -88,7 +88,15 @@ class OfficeIndexService
$decoded = html_entity_decode($url); $decoded = html_entity_decode($url);
$parts = parse_url($decoded); $parts = parse_url($decoded);
// direkter /files/-Pfad // 1) files/... (ohne führenden Slash)
if (!empty($parts['path']) && str_starts_with($parts['path'], 'files/')) {
$ext = strtolower(pathinfo($parts['path'], PATHINFO_EXTENSION));
if (in_array($ext, ['docx', 'xlsx', 'pptx'], true)) {
return ['/' . $parts['path'], $ext];
}
}
// 2) /files/...
if (!empty($parts['path']) && str_starts_with($parts['path'], '/files/')) { if (!empty($parts['path']) && str_starts_with($parts['path'], '/files/')) {
$ext = strtolower(pathinfo($parts['path'], PATHINFO_EXTENSION)); $ext = strtolower(pathinfo($parts['path'], PATHINFO_EXTENSION));
if (in_array($ext, ['docx', 'xlsx', 'pptx'], true)) { if (in_array($ext, ['docx', 'xlsx', 'pptx'], true)) {
@@ -96,10 +104,27 @@ class OfficeIndexService
} }
} }
// Contao-Download-Link mit ?p= if (empty($parts['query'])) {
if (!empty($parts['query'])) { return null;
}
parse_str($parts['query'], $query); parse_str($parts['query'], $query);
// 3) Contao 4: ?file=files/...
if (!empty($query['file'])) {
$file = urldecode((string) $query['file']);
$file = ltrim($file, '/');
$ext = strtolower(pathinfo($file, PATHINFO_EXTENSION));
if (
str_starts_with($file, 'files/')
&& in_array($ext, ['docx', 'xlsx', 'pptx'], true)
) {
return ['/' . $file, $ext];
}
}
// 4) Contao 5: ?p=...
if (!empty($query['p'])) { if (!empty($query['p'])) {
$p = urldecode((string) $query['p']); $p = urldecode((string) $query['p']);
$ext = strtolower(pathinfo($p, PATHINFO_EXTENSION)); $ext = strtolower(pathinfo($p, PATHINFO_EXTENSION));
@@ -108,7 +133,6 @@ class OfficeIndexService
return ['/files/' . ltrim($p, '/'), $ext]; return ['/files/' . ltrim($p, '/'), $ext];
} }
} }
}
return null; return null;
} }
src/Service/PdfIndexService.php
+39 -62
View File
@@ -16,17 +16,26 @@ class PdfIndexService
public function __construct(ParameterBagInterface $params) public function __construct(ParameterBagInterface $params)
{ {
$this->projectDir = rtrim((string) $params->get('kernel.project_dir'), '/'); $this->projectDir = rtrim((string) $params->get('kernel.project_dir'), '/');
fwrite(STDERR, "[Meili PDF DEBUG] projectDir={$this->projectDir}\n"); $this->debug("projectDir={$this->projectDir}");
}
private function debug(string $message): void
{
$stream = \defined('STDERR')
? STDERR
: fopen('php://stderr', 'wb');
fwrite($stream, "[Meili PDF DEBUG] {$message}\n");
} }
public function resetTableOnce(): void public function resetTableOnce(): void
{ {
if ($this->didReset) { if ($this->didReset) {
fwrite(STDERR, "[Meili PDF DEBUG] resetTableOnce(): already reset\n"); $this->debug('resetTableOnce(): already reset');
return; return;
} }
fwrite(STDERR, "[Meili PDF DEBUG] resetTableOnce(): TRUNCATE tl_search_pdf\n"); $this->debug('resetTableOnce(): TRUNCATE tl_search_pdf');
$this->didReset = true; $this->didReset = true;
$this->seenThisCrawl = []; $this->seenThisCrawl = [];
@@ -34,89 +43,70 @@ class PdfIndexService
try { try {
Database::getInstance()->execute('TRUNCATE tl_search_pdf'); Database::getInstance()->execute('TRUNCATE tl_search_pdf');
} catch (\Throwable $e) { } catch (\Throwable $e) {
fwrite(STDERR, "[Meili PDF DEBUG] TRUNCATE failed: {$e->getMessage()}\n"); $this->debug('TRUNCATE failed: ' . $e->getMessage());
} }
} }
public function handlePdfLinks(array $pdfLinks): void public function handlePdfLinks(array $pdfLinks): void
{ {
fwrite( $this->debug('handlePdfLinks(): count=' . count($pdfLinks));
STDERR,
"[Meili PDF DEBUG] handlePdfLinks(): count=" . count($pdfLinks) . "\n"
);
foreach ($pdfLinks as $row) { foreach ($pdfLinks as $row) {
$url = (string) ($row['url'] ?? ''); $url = (string) ($row['url'] ?? '');
$linkText = $row['linkText'] ?? null; $linkText = $row['linkText'] ?? null;
fwrite(STDERR, "\n[Meili PDF DEBUG] URL={$url}\n"); $this->debug("URL={$url}");
if ($url === '') { if ($url === '') {
fwrite(STDERR, "[Meili PDF DEBUG] → empty URL, skip\n"); $this->debug('→ empty URL, skip');
continue; continue;
} }
$seenKey = md5($url); $seenKey = md5($url);
if (isset($this->seenThisCrawl[$seenKey])) { if (isset($this->seenThisCrawl[$seenKey])) {
fwrite(STDERR, "[Meili PDF DEBUG] → already processed, skip\n"); $this->debug('→ already processed, skip');
continue; continue;
} }
$this->seenThisCrawl[$seenKey] = true; $this->seenThisCrawl[$seenKey] = true;
$normalizedPath = $this->normalizePdfUrl($url); $normalizedPath = $this->normalizePdfUrl($url);
fwrite( $this->debug('normalizePdfUrl() → ' . ($normalizedPath ?? 'NULL'));
STDERR,
"[Meili PDF DEBUG] normalizePdfUrl() → "
. ($normalizedPath ?? 'NULL')
. "\n"
);
if ($normalizedPath === null) { if ($normalizedPath === null) {
fwrite(STDERR, "[Meili PDF DEBUG] → normalization failed, skip\n"); $this->debug('→ normalization failed, skip');
continue; continue;
} }
$absolutePath = $this->getAbsolutePath($normalizedPath); $absolutePath = $this->getAbsolutePath($normalizedPath);
fwrite(STDERR, "[Meili PDF DEBUG] absolutePath={$absolutePath}\n"); $this->debug("absolutePath={$absolutePath}");
if (!is_file($absolutePath)) { if (!is_file($absolutePath)) {
fwrite(STDERR, "[Meili PDF DEBUG] → file does NOT exist\n"); $this->debug('→ file does NOT exist');
continue; continue;
} }
fwrite(STDERR, "[Meili PDF DEBUG] → file exists\n"); $this->debug('→ file exists');
$mtime = (int) (filemtime($absolutePath) ?: 0); $mtime = (int) (filemtime($absolutePath) ?: 0);
$checksum = md5($normalizedPath . '|' . $mtime); $checksum = md5($normalizedPath . '|' . $mtime);
fwrite( $this->debug("mtime={$mtime} checksum={$checksum}");
STDERR,
"[Meili PDF DEBUG] mtime={$mtime} checksum={$checksum}\n"
);
$pdfMetaTitle = $this->readPdfMetaTitle($absolutePath); $pdfMetaTitle = $this->readPdfMetaTitle($absolutePath);
fwrite( $this->debug('metaTitle=' . ($pdfMetaTitle ?: 'NULL'));
STDERR,
"[Meili PDF DEBUG] metaTitle="
. ($pdfMetaTitle ?: 'NULL')
. "\n"
);
$title = $linkText ?: ($pdfMetaTitle ?: basename($absolutePath)); $title = $linkText ?: ($pdfMetaTitle ?: basename($absolutePath));
fwrite(STDERR, "[Meili PDF DEBUG] final title={$title}\n"); $this->debug("final title={$title}");
$text = $this->parsePdf($absolutePath); $text = $this->parsePdf($absolutePath);
fwrite( $this->debug('parsed text length=' . strlen($text));
STDERR,
"[Meili PDF DEBUG] parsed text length=" . strlen($text) . "\n"
);
if ($text === '') { if ($text === '') {
fwrite(STDERR, "[Meili PDF DEBUG] → empty text, skip\n"); $this->debug('→ empty text, skip');
continue; continue;
} }
fwrite(STDERR, "[Meili PDF DEBUG] → writing to DB\n"); $this->debug('→ writing to DB');
$this->upsertPdf( $this->upsertPdf(
$normalizedPath, $normalizedPath,
@@ -130,52 +120,48 @@ class PdfIndexService
private function normalizePdfUrl(string $url): ?string private function normalizePdfUrl(string $url): ?string
{ {
fwrite(STDERR, "[Meili PDF DEBUG] normalizePdfUrl(): {$url}\n"); $this->debug("normalizePdfUrl(): {$url}");
$decoded = html_entity_decode($url); $decoded = html_entity_decode($url);
$parts = parse_url($decoded); $parts = parse_url($decoded);
// 1) files/...pdf (ohne führenden Slash)
if (!empty($parts['path']) && str_starts_with($parts['path'], 'files/') && str_ends_with(strtolower($parts['path']), '.pdf')) { if (!empty($parts['path']) && str_starts_with($parts['path'], 'files/') && str_ends_with(strtolower($parts['path']), '.pdf')) {
$r = '/' . $parts['path']; $r = '/' . $parts['path'];
fwrite(STDERR, "[Meili PDF DEBUG] → relative files path {$r}\n"); $this->debug("→ relative files path {$r}");
return $r; return $r;
} }
// 2) /files/...pdf
if (!empty($parts['path']) && str_starts_with($parts['path'], '/files/') && str_ends_with(strtolower($parts['path']), '.pdf')) { if (!empty($parts['path']) && str_starts_with($parts['path'], '/files/') && str_ends_with(strtolower($parts['path']), '.pdf')) {
fwrite(STDERR, "[Meili PDF DEBUG] → absolute files path {$parts['path']}\n"); $this->debug("→ absolute files path {$parts['path']}");
return $parts['path']; return $parts['path'];
} }
if (empty($parts['query'])) { if (empty($parts['query'])) {
fwrite(STDERR, "[Meili PDF DEBUG] → no query\n"); $this->debug('→ no query');
return null; return null;
} }
parse_str($parts['query'], $query); parse_str($parts['query'], $query);
// 3) Contao 4: ?file=files/...
if (!empty($query['file'])) { if (!empty($query['file'])) {
$file = urldecode((string) $query['file']); $file = urldecode((string) $query['file']);
$file = ltrim($file, '/'); $file = ltrim($file, '/');
if (str_starts_with($file, 'files/') && str_ends_with(strtolower($file), '.pdf')) { if (str_starts_with($file, 'files/') && str_ends_with(strtolower($file), '.pdf')) {
$r = '/' . $file; $r = '/' . $file;
fwrite(STDERR, "[Meili PDF DEBUG] → file= normalized {$r}\n"); $this->debug("→ file= normalized {$r}");
return $r; return $r;
} }
} }
// 4) Contao 5: ?p=...
if (!empty($query['p'])) { if (!empty($query['p'])) {
$p = urldecode((string) $query['p']); $p = urldecode((string) $query['p']);
$r = '/files/' . ltrim($p, '/'); $r = '/files/' . ltrim($p, '/');
fwrite(STDERR, "[Meili PDF DEBUG] → p= normalized {$r}\n"); $this->debug("→ p= normalized {$r}");
return $r; return $r;
} }
fwrite(STDERR, "[Meili PDF DEBUG] → no usable parameter\n"); $this->debug('→ no usable parameter');
return null; return null;
} }
@@ -214,12 +200,9 @@ class PdfIndexService
$mtime $mtime
); );
fwrite(STDERR, "[Meili PDF DEBUG] → DB write OK\n"); $this->debug('→ DB write OK');
} catch (\Throwable $e) { } catch (\Throwable $e) {
fwrite( $this->debug('DB write failed: ' . $e->getMessage());
STDERR,
"[Meili PDF DEBUG] DB write failed: {$e->getMessage()}\n"
);
} }
} }
@@ -231,10 +214,7 @@ class PdfIndexService
$text = $this->cleanPdfContent($pdf->getText()); $text = $this->cleanPdfContent($pdf->getText());
return mb_substr($text, 0, 20000); return mb_substr($text, 0, 20000);
} catch (\Throwable $e) { } catch (\Throwable $e) {
fwrite( $this->debug('parsePdf failed: ' . $e->getMessage());
STDERR,
"[Meili PDF DEBUG] parsePdf failed: {$e->getMessage()}\n"
);
return ''; return '';
} }
} }
@@ -252,10 +232,7 @@ class PdfIndexService
} }
} }
} catch (\Throwable $e) { } catch (\Throwable $e) {
fwrite( $this->debug('readPdfMetaTitle failed: ' . $e->getMessage());
STDERR,
"[Meili PDF DEBUG] readPdfMetaTitle failed: {$e->getMessage()}\n"
);
} }
return null; return null;