|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Kasifi\PdfParserBundle\Command; |
|
4
|
|
|
|
|
5
|
|
|
use Kasifi\PdfParserBundle\PdfParser; |
|
6
|
|
|
use Kasifi\PdfParserBundle\Util\ParseHelper; |
|
7
|
|
|
use Symfony\Bundle\FrameworkBundle\Command\ContainerAwareCommand; |
|
8
|
|
|
use Symfony\Component\Console\Helper\Table; |
|
9
|
|
|
use Symfony\Component\Console\Input\InputArgument; |
|
10
|
|
|
use Symfony\Component\Console\Input\InputInterface; |
|
11
|
|
|
use Symfony\Component\Console\Input\InputOption; |
|
12
|
|
|
use Symfony\Component\Console\Output\OutputInterface; |
|
13
|
|
|
use Symfony\Component\Console\Question\ChoiceQuestion; |
|
14
|
|
|
use Symfony\Component\Finder\Finder; |
|
15
|
|
|
use Symfony\Component\Finder\SplFileInfo; |
|
16
|
|
|
use Symfony\Component\Yaml\Dumper; |
|
17
|
|
|
|
|
18
|
|
|
/** |
|
19
|
|
|
* Class ParserCommand. |
|
20
|
|
|
*/ |
|
21
|
|
|
class ParserCommand extends ContainerAwareCommand |
|
22
|
|
|
{ |
|
23
|
|
|
/** |
|
24
|
|
|
* @var OutputInterface |
|
25
|
|
|
*/ |
|
26
|
|
|
private $output; |
|
27
|
|
|
|
|
28
|
|
|
/** |
|
29
|
|
|
* Configure the command. |
|
30
|
|
|
*/ |
|
31
|
|
|
protected function configure() |
|
32
|
|
|
{ |
|
33
|
|
|
$this |
|
34
|
|
|
->setName('pdf-parser:parse') |
|
35
|
|
|
->setDescription('Parse document of many types.') |
|
36
|
|
|
->addArgument('processor', InputArgument::OPTIONAL, 'The id of the processor') |
|
37
|
|
|
->addArgument('file_path', InputArgument::OPTIONAL, 'The absolute path to the PDF file to parse.') |
|
38
|
|
|
->addOption('format', null, InputOption::VALUE_REQUIRED, 'The output format (console, json, yml)', 'console'); |
|
39
|
|
|
} |
|
40
|
|
|
|
|
41
|
|
|
/** |
|
42
|
|
|
* @param InputInterface $input |
|
43
|
|
|
* @param OutputInterface $output |
|
44
|
|
|
* |
|
45
|
|
|
* @return void |
|
46
|
|
|
*/ |
|
47
|
|
|
protected function execute(InputInterface $input, OutputInterface $output) |
|
48
|
|
|
{ |
|
49
|
|
|
$this->output = $output; |
|
50
|
|
|
$container = $this->getContainer(); |
|
51
|
|
|
$pdfParser = $container->get('kasifi_pdfparser.pdf_parser'); |
|
52
|
|
|
$pdfDirectoryPath = $container->getParameter('kasifi_pdfparser.pdf_directory_path'); |
|
53
|
|
|
|
|
54
|
|
|
// Select processor |
|
55
|
|
|
$this->selectProcessor($input, $pdfParser); |
|
56
|
|
|
|
|
57
|
|
|
// Select file |
|
58
|
|
|
$filePath = $this->selectFile($input, $pdfDirectoryPath); |
|
59
|
|
|
|
|
60
|
|
|
// Parse |
|
61
|
|
|
$rows = $pdfParser->parse($filePath); |
|
62
|
|
|
$data = ParseHelper::inlineDates($rows->toArray()); |
|
63
|
|
|
|
|
64
|
|
|
// Write output |
|
65
|
|
|
$this->writeOutput($input->getOption('format'), $data); |
|
66
|
|
|
|
|
67
|
|
|
return; |
|
68
|
|
|
} |
|
69
|
|
|
|
|
70
|
|
|
/** |
|
71
|
|
|
* @param $format |
|
72
|
|
|
* @param $data |
|
73
|
|
|
*/ |
|
74
|
|
|
protected function writeOutput($format, $data) |
|
75
|
|
|
{ |
|
76
|
|
|
switch ($format) { |
|
77
|
|
|
case 'json': |
|
78
|
|
|
$outputData = json_encode($data); |
|
79
|
|
|
$this->output->write($outputData); |
|
80
|
|
|
break; |
|
81
|
|
|
case 'yml': |
|
82
|
|
|
$dumper = new Dumper(); |
|
83
|
|
|
$outputData = $dumper->dump($data, 1); |
|
84
|
|
|
$this->output->write($outputData); |
|
85
|
|
|
break; |
|
86
|
|
|
case 'console': |
|
87
|
|
|
if (count($data)) { |
|
88
|
|
|
$table = new Table($this->output); |
|
89
|
|
|
$table |
|
90
|
|
|
->setHeaders([array_keys($data[0])]) |
|
91
|
|
|
->setRows($data); |
|
92
|
|
|
$table->render(); |
|
93
|
|
|
} |
|
94
|
|
|
break; |
|
95
|
|
|
} |
|
96
|
|
|
} |
|
97
|
|
|
|
|
98
|
|
|
/** |
|
99
|
|
|
* @param InputInterface $input |
|
100
|
|
|
* @param $pdfDirectoryPath |
|
101
|
|
|
* |
|
102
|
|
|
* @return mixed |
|
103
|
|
|
*/ |
|
104
|
|
|
protected function selectFile(InputInterface $input, $pdfDirectoryPath) |
|
105
|
|
|
{ |
|
106
|
|
|
$filePath = $input->getArgument('file_path'); |
|
107
|
|
|
if (!$filePath) { |
|
108
|
|
|
$helper = $this->getHelper('question'); |
|
109
|
|
|
$finder = new Finder(); |
|
110
|
|
|
$finder->files()->in($pdfDirectoryPath); |
|
111
|
|
|
$files = []; |
|
112
|
|
|
foreach ($finder as $key => $file) { |
|
113
|
|
|
/* @var SplFileInfo $file */ |
|
114
|
|
|
$files[] = $file->getRealPath(); |
|
115
|
|
|
} |
|
116
|
|
|
|
|
117
|
|
|
$question = new ChoiceQuestion('Which file? Enter the key.', $files); |
|
118
|
|
|
$filePath = $helper->ask($input, $this->output, $question); |
|
119
|
|
|
|
|
120
|
|
|
return $filePath; |
|
121
|
|
|
} |
|
122
|
|
|
|
|
123
|
|
|
return $filePath; |
|
124
|
|
|
} |
|
125
|
|
|
|
|
126
|
|
|
/** |
|
127
|
|
|
* @param InputInterface $input |
|
128
|
|
|
* @param $pdfParser |
|
129
|
|
|
*/ |
|
130
|
|
|
protected function selectProcessor(InputInterface $input, PdfParser $pdfParser) |
|
131
|
|
|
{ |
|
132
|
|
|
$processors = $pdfParser->getAvailableProcessors(); |
|
133
|
|
|
$processorId = $input->getArgument('processor'); |
|
134
|
|
|
if (!$processorId) { |
|
135
|
|
|
$helper = $this->getHelper('question'); |
|
136
|
|
|
$question = new ChoiceQuestion('Which processor to use?', $processors); |
|
137
|
|
|
$processorId = $helper->ask($input, $this->output, $question); |
|
138
|
|
|
} |
|
139
|
|
|
$pdfParser->setProcessor($processors[$processorId]); |
|
140
|
|
|
} |
|
141
|
|
|
} |
|
142
|
|
|
|
lucascherifi /
PdfParserBundle
