Skip to content

Commit 99d7096

Browse files
zembrowskizembrowski
authored
Merge pull request #9 from rechtlogisch/feat-extend-output-and-tests
feat: extend output and tests
1 parent b2944e5 commit 99d7096

16 files changed

+606
-84
lines changed

‎composer.json‎

Lines changed: 5 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -44,17 +44,19 @@
4444
"scripts": {
4545
"analyse": "vendor/bin/phpstan analyse",
4646
"arch": "vendor/bin/pest --group=arch",
47-
"coverage": "vendor/bin/pest --coverage --min=90",
47+
"coverage": "vendor/bin/pest --coverage --min=99",
48+
"external": "vendor/bin/pest --group=external",
4849
"format": "vendor/bin/pint",
4950
"quality": [
5051
"@arch",
5152
"@format",
5253
"@test",
5354
"@analyse",
5455
"@types",
55-
"@coverage"
56+
"@coverage",
57+
"@external"
5658
],
57-
"test": "vendor/bin/pest --exclude-group=arch",
59+
"test": "vendor/bin/pest --exclude-group=arch --exclude-group=external",
5860
"types": "vendor/bin/pest --type-coverage --min=100"
5961
},
6062
"bin": [

‎src/Retrieve.php‎

Lines changed: 58 additions & 44 deletions
Original file line numberDiff line numberDiff line change
@@ -4,6 +4,7 @@
44

55
namespace Rechtlogisch\TseId;
66

7+
use RuntimeException;
78
use Symfony\Component\BrowserKit\HttpBrowser;
89
use Symfony\Component\DomCrawler\Crawler;
910
use Symfony\Component\HttpClient\HttpClient;
@@ -13,15 +14,18 @@ class Retrieve
1314
{
1415
private int $countPages = 1;
1516

16-
private const URL = 'https://www.bsi.bund.de/EN/Themen/Unternehmen-und-Organisationen/Standards-und-Zertifizierung/Zertifizierung-und-Anerkennung/Listen/Zertifizierte-Produkte-nach-TR/Technische_Sicherheitseinrichtungen/TSE_node.html?gts=913608_list%253Dtitle_text_sort%252Bdesc&gtp=913608_list%253D';
17+
public const URL = 'https://www.bsi.bund.de/EN/Themen/Unternehmen-und-Organisationen/Standards-und-Zertifizierung/Zertifizierung-und-Anerkennung/Listen/Zertifizierte-Produkte-nach-TR/Technische_Sicherheitseinrichtungen/TSE_node.html?gts=913608_list%253Dtitle_text_sort%252Bdesc&gtp=913608_list%253D';
1718

1819
/**
1920
* @var array<string, array<string, string>>
2021
*/
2122
private array $retrieved = [];
2223

23-
public function __construct()
24+
private HttpBrowser $browser;
25+
26+
public function __construct(?HttpBrowser $browser = null)
2427
{
28+
$this->browser = $browser ?? new HttpBrowser(HttpClient::create());
2529
$this->run();
2630
}
2731

@@ -41,48 +45,53 @@ public function page(int $no = 1): void
4145
{
4246
$url = self::URL.$no;
4347

44-
$browser = new HttpBrowser(HttpClient::create());
45-
$crawler = $browser->request('GET', $url);
48+
try {
49+
$crawler = $this->browser->request('GET', $url);
4650

47-
if ($no === 1) {
48-
$paginationText = $crawler->filter('#content nav.c-pagination p')->text();
49-
preg_match('/Search results (\d+) to (\d+) from a total of (\d+)/', $paginationText, $matches);
50-
$max = $matches[3] ?? '0';
51+
if ($no === 1) {
52+
$paginationText = $crawler->filter('#content nav.c-pagination p')->text();
53+
preg_match('/Search results (\d+) to (\d+) from a total of (\d+)/', $paginationText, $matches);
54+
$max = $matches[3] ?? '0';
5155

52-
$this->countPages = (int) ceil((int) $max / 10);
53-
}
56+
$this->countPages = (int) ceil((int) $max / 10);
57+
}
5458

55-
$crawler->filter('#content div.wrapperTable table.textualData tbody tr')->each(function (Crawler $row) {
56-
$rowData = [];
57-
$tseId = null;
58-
59-
$row->filter('td')->each(function (Crawler $cell, int $index) use (&$rowData, &$tseId) {
60-
$header = '';
61-
62-
switch ($index) {
63-
case 0: $header = 'tse_id';
64-
break;
65-
case 1: $header = 'content';
66-
break;
67-
case 2: $header = 'manufacturer';
68-
break;
69-
case 3: $header = 'date_issuance';
70-
break;
71-
}
72-
73-
if ($header === 'tse_id') {
74-
$fullIdText = $cell->text();
75-
$tseId = str_replace('BSI-K-TR-', '', $fullIdText);
76-
[$id, $year] = explode('-', $tseId);
77-
$rowData['id'] = $id;
78-
$rowData['year'] = $year;
79-
} else {
80-
$rowData[$header] = trim($cell->text());
81-
}
59+
$crawler->filter('#content div.wrapperTable table.textualData tbody tr')->each(function (Crawler $row) {
60+
$rowData = [];
61+
$tseId = null;
62+
63+
$row->filter('td')->each(function (Crawler $cell, int $index) use (&$rowData, &$tseId) {
64+
$header = '';
65+
66+
switch ($index) {
67+
case 0: $header = 'tse_id';
68+
break;
69+
case 1: $header = 'content';
70+
break;
71+
case 2: $header = 'manufacturer';
72+
break;
73+
case 3: $header = 'date_issuance';
74+
break;
75+
}
76+
77+
if ($header === 'tse_id') {
78+
$fullIdText = $cell->text();
79+
$tseId = str_replace('BSI-K-TR-', '', $fullIdText);
80+
[$id, $year] = explode('-', $tseId);
81+
$rowData['id'] = $id;
82+
$rowData['year'] = $year;
83+
} else {
84+
$rowData[$header] = trim($cell->text());
85+
}
86+
});
87+
88+
$this->retrieved[$tseId] = $rowData;
8289
});
83-
84-
$this->retrieved[$tseId] = $rowData;
85-
});
90+
} catch (Throwable $e) {
91+
$html = $this->browser->getResponse()->getContent();
92+
throw (new RetrieveException($e->getMessage(), $e->getCode(), $e->getPrevious()))
93+
->addContext($url, $html);
94+
}
8695
}
8796

8897
/**
@@ -112,13 +121,13 @@ public function json(?string $tseId = null, bool $pretty = true): string
112121
/**
113122
* @return array<string, string>
114123
*/
115-
public function save(mixed $path = '.'): array
124+
public function save(string $path = '.'): array
116125
{
117126
$files = [];
118127

119128
$prefix = date('Y-m-d');
120129

121-
$files['json'] = $this->saveJson($path.DIRECTORY_SEPARATOR.$prefix); // @phpstan-ignore-line
130+
$files['json'] = $this->saveJson($path.DIRECTORY_SEPARATOR.$prefix);
122131

123132
return $files;
124133
}
@@ -127,16 +136,21 @@ private function saveJson(string $path): string
127136
{
128137
$content = $this->json();
129138
$pathWithExtension = $path.'.json';
139+
$dir = dirname($pathWithExtension);
140+
if (! is_dir($dir) || ! is_writable($dir)) {
141+
return '';
142+
}
143+
130144
$result = file_put_contents($pathWithExtension, $content);
131145

132146
if ($result === false) {
133-
return '';
147+
throw new RuntimeException('Could not write file: '.$pathWithExtension);
134148
}
135149

136150
$path = realpath($pathWithExtension);
137151

138152
if ($path === false) {
139-
return '';
153+
throw new RuntimeException('Could not determine real path: '.$pathWithExtension);
140154
}
141155

142156
return $path;

‎src/RetrieveCommand.php‎

Lines changed: 27 additions & 8 deletions
Original file line numberDiff line numberDiff line change
@@ -4,6 +4,7 @@
44

55
namespace Rechtlogisch\TseId;
66

7+
use Symfony\Component\BrowserKit\HttpBrowser;
78
use Symfony\Component\Console\Attribute\AsCommand;
89
use Symfony\Component\Console\Command\Command;
910
use Symfony\Component\Console\Input\InputArgument;
@@ -13,6 +14,14 @@
1314
#[AsCommand(name: 'tse-id:retrieve', description: 'Retrieve the TSE list from BSI')]
1415
class RetrieveCommand extends Command
1516
{
17+
private ?HttpBrowser $browser;
18+
19+
public function __construct(?HttpBrowser $browser = null)
20+
{
21+
parent::__construct();
22+
$this->browser = $browser;
23+
}
24+
1625
protected function configure(): void
1726
{
1827
$this->addArgument('targetPath', InputArgument::OPTIONAL, 'The path to save output. Can be relative or absolute. Must not end with a DIRECTORY_SEPARATOR.');
@@ -21,17 +30,27 @@ protected function configure(): void
2130
protected function execute(InputInterface $input, OutputInterface $output): int
2231
{
2332
$output->writeln('Retrieving TSE list from BSI...');
24-
$targetPath = $input->getArgument('targetPath');
25-
26-
if (is_string($targetPath)) {
27-
$targetPath = trim($targetPath);
28-
}
29-
30-
if (empty($targetPath)) {
33+
$arg = $input->getArgument('targetPath');
34+
$targetPath = is_string($arg) ? trim($arg) : '';
35+
if ($targetPath === '') {
3136
$targetPath = '.';
3237
}
3338

34-
$retrieve = new Retrieve;
39+
try {
40+
$retrieve = new Retrieve($this->browser);
41+
} catch (RetrieveException $e) {
42+
$output->writeln('Error while processing URL: '.$e->getUrl());
43+
$output->writeln('Exception message: '.$e->getMessage());
44+
45+
$html = $e->getHtml();
46+
if (! empty($html)) {
47+
$output->writeln('Response HTML begin >>>');
48+
$output->writeln($html);
49+
$output->writeln('<<< [Retrieve] Response HTML end');
50+
}
51+
52+
return Command::FAILURE;
53+
}
3554
$result = $retrieve->save($targetPath);
3655

3756
if (empty($result) || empty($result[array_key_first($result)])) {

‎src/RetrieveException.php‎

Lines changed: 38 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,38 @@
1+
<?php
2+
3+
declare(strict_types=1);
4+
5+
namespace Rechtlogisch\TseId;
6+
7+
use RuntimeException;
8+
use Throwable;
9+
10+
class RetrieveException extends RuntimeException
11+
{
12+
private string $url;
13+
14+
private ?string $html = null;
15+
16+
public function __construct(string $message = '', int $code = 0, ?Throwable $previous = null)
17+
{
18+
parent::__construct($message, $code, $previous);
19+
}
20+
21+
public function addContext(string $url, ?string $html = null): self
22+
{
23+
$this->html = $html;
24+
$this->url = $url;
25+
26+
return $this;
27+
}
28+
29+
public function getUrl(): string
30+
{
31+
return $this->url;
32+
}
33+
34+
public function getHtml(): ?string
35+
{
36+
return $this->html;
37+
}
38+
}

0 commit comments

Comments
 (0)