From 8072e27565cd6b00dc0f115ad3ee4655b671f45d Mon Sep 17 00:00:00 2001
From: asajedi <asajedi@gwdg.de>
Date: Wed, 1 Dec 2021 03:17:06 +0100
Subject: [PATCH] Import and index sample TEI document

Resolves #28
---
 config/services.yaml    |  6 ++++--
 src/Import/Importer.php | 28 +++++++++++++++++++++++++---
 src/Import/Indexer.php  | 10 +++++++++-
 3 files changed, 38 insertions(+), 6 deletions(-)

diff --git a/config/services.yaml b/config/services.yaml
index 51723be..878c285 100755
--- a/config/services.yaml
+++ b/config/services.yaml
@@ -23,8 +23,8 @@ parameters:
         SMB_ZA: 'Zentralarchiv Staatliche Museen zu Berlin – Preußischer Kulturbesitz'
         UAHW: 'Archiv der Martin-Luther-Universität Halle-Wittenberg, Halle / S.'
         SUB_HSD: 'SUB HSD Niedersächsische Staats- und Universitätsbibliothek Göttingen, Abteilung Handschriften und Seltene Drucke'
-    tei_dir: '%kernel.project_dir%/data/gitlab/'
-    tei_sample_dir: '%kernel.project_dir%/teis/sampletei/'
+    tei_dir: '%kernel.project_dir%/data/teis/'
+    tei_sample_dir: '%kernel.project_dir%/data/sampletei/'
 
 services:
     # default configuration for services in *this* file
@@ -93,10 +93,12 @@ services:
             - method: 'setConfigs'
               arguments:
                   - '%tei_dir%'
+                  - '%tei_sample_dir%'
                   - '%env(GITLAB_REPO_TOKEN)%'
                   - '%env(GITLAB_REPO_TREE_URL)%'
                   - '%env(GITLAB_PROCESSED_TEI_REPO_URL)%'
                   - '%env(INVALIDE_TEI_LIST_FILE)%'
+                  - '%env(SAMPLE_TEI_DOCUMENT_URL)%'
 
     App\Import\Indexer:
         calls:
diff --git a/src/Import/Importer.php b/src/Import/Importer.php
index 4ab6889..433c4d1 100644
--- a/src/Import/Importer.php
+++ b/src/Import/Importer.php
@@ -4,19 +4,23 @@ declare(strict_types=1);
 
 namespace App\Import;
 
+use App\Import\ImporterInterface;
 use Symfony\Component\Filesystem\Filesystem;
 use Symfony\Component\HttpFoundation\File\Exception\FileException;
 
 class Importer implements ImporterInterface
 {
+    private ?string $teiDir = null;
+    private ?string  $teiSampleDir = null;
     private ?string $gitlabProcessedTeiRepoUrl;
     private ?string $gitlabRepoToken;
     private ?string $gitlabRepoTreeUrl;
     private ?string $invalidTeiListFile;
-    private ?string $teiDir = null;
+    private ?string $sampleTEIDocumentUrl;
 
     public function import(): void
     {
+        $this->importSampleTEIDocument();
         $filesystem = new Filesystem();
         if (!$filesystem->exists($this->teiDir)) {
             $filesystem->mkdir($this->teiDir);
@@ -58,23 +62,41 @@ class Importer implements ImporterInterface
         }
     }
 
-    public function setConfigs(string $teiDir, string $gitlabRepoToken, string $gitlabRepoTreeUrl, string $gitlabProcessedTeiRepoUrl, string $invalidTeiListFile): void
+    public function setConfigs(string $teiDir, string $teiSampleDir, string $gitlabRepoToken, string $gitlabRepoTreeUrl, string $gitlabProcessedTeiRepoUrl, string $invalidTeiListFile, string $sampleTEIDocumentUrl): void
     {
         $this->teiDir = $teiDir;
+        $this->teiSampleDir = $teiSampleDir;
         $this->gitlabRepoToken = $gitlabRepoToken;
         $this->gitlabRepoTreeUrl = $gitlabRepoTreeUrl;
         $this->gitlabProcessedTeiRepoUrl = $gitlabProcessedTeiRepoUrl;
         $this->invalidTeiListFile = $invalidTeiListFile;
+        $this->sampleTEIDocumentUrl = $sampleTEIDocumentUrl;
     }
 
     private function getInvalidTeiList(): array
     {
         $invalidTeiList = [];
         $file_headers = @get_headers($this->invalidTeiListFile);
-        if($file_headers[0] !== 'HTTP/1.1 404 Not Found') {
+        if($file_headers[0] === 'HTTP/1.1 200 OK') {
             $invalidTeiList = json_decode(file_get_contents($this->invalidTeiListFile), true);
         }
 
         return $invalidTeiList;
     }
+
+    private function importSampleTEIDocument(): void
+    {
+        $sampleTEIDocumentUrl = $this->sampleTEIDocumentUrl.'&access_token='.$this->gitlabRepoToken;
+        $file_headers = @get_headers($sampleTEIDocumentUrl);
+        if($file_headers[0] === 'HTTP/1.1 200 OK') {
+            $sampleDocumentData = json_decode(file_get_contents($sampleTEIDocumentUrl), true);
+            $sampleTEIDocument = base64_decode($sampleDocumentData['content']);
+            $filesystem = new Filesystem();
+            if (!$filesystem->exists($this->teiSampleDir)) {
+                $filesystem->mkdir($this->teiSampleDir);
+            }
+
+            $filesystem->dumpFile($this->teiSampleDir.$sampleDocumentData['file_name'], $sampleTEIDocument);
+        }
+    }
 }
diff --git a/src/Import/Indexer.php b/src/Import/Indexer.php
index ce420ac..a24dddb 100644
--- a/src/Import/Indexer.php
+++ b/src/Import/Indexer.php
@@ -14,6 +14,7 @@ use DOMElement;
 use DOMXPath;
 use League\Flysystem\Exception;
 use Solarium\Client;
+use Symfony\Component\Filesystem\Exception\IOExceptionInterface;
 use Symfony\Component\Filesystem\Filesystem;
 use Symfony\Component\Finder\Finder;
 
@@ -126,7 +127,7 @@ class Indexer implements IndexerInterface
 
     public function tei2solr(): void
     {
-        $this->client->getEndpoint()->setOptions(['timeout' => 60]);
+        $this->client->getEndpoint()->setOptions(['timeout' => 60, 'index_timeout' => 60]);
         $finder = new Finder();
         $finder->files()->in($this->teiDir);
 
@@ -183,6 +184,13 @@ class Indexer implements IndexerInterface
                 libxml_clear_errors();
             }
         }
+
+        try {
+            $filesystem = new Filesystem();
+            $filesystem->remove($this->teiDir);
+        } catch (IOExceptionInterface $exception) {
+            echo "Error deleting directory at". $exception->getPath();
+        }
     }
 
     private function convertSoftHyphenToHyphen(string $text): string
-- 
GitLab