<?php

function pdf2Json($pdfFilePath) {
    $output = \shell_exec('pdf2json -noframes -q -enc UTF-8 '.\escapeshellarg($pdfFilePath).' /dev/stdout');

    if(empty($output))
        throw new \RuntimeException('Failed to run pdf2json binary on pdf file at '.$pdfFilePath."\n".\implode("\n", $output));

    return json_decode($output, true)[0]['text'];
}

function buildPhrases($items) {
    $phrases = [];

    foreach($items as $item) {
        $item['right'] = $item['left'] + $item['width'];

        $found = false;
        foreach($phrases as &$p) {
            if($p['top'] == $item['top'] &&
                ($p['right'] >= $item['left'] - 1 && $p['right'] <= $item['left'] + 1)) {
                $p['data'] .= ($item['data']);
                $p['right'] = $item['right'];
                $p['width'] = $item['right'] - $p['left'];
                $found = true;
            }
        }

        if($found)
            continue;

        $phrases[] = $item;
    }

    return $phrases;
}

function toMatrix($items) {
    $out = [];

    foreach($items as $item) {
        $out[$item['left']][$item['top']] = $item;
    }

    return $out;
}

function dataFromPdf($filePath) {
    $itemsRaw = pdf2Json($filePath);
    $items = buildPhrases($itemsRaw);
    $itemsRaw = toMatrix($itemsRaw);
    $items = toMatrix($items);

    return [
        'proprietario'  => $items[66][100]['data'],
        'orgao_atuador' => $items[66][130]['data'],
        'local'         => $items[66][161]['data'],
        'infracao'      => $items[66][191]['data'],
        'natureza'      => $items[66][252]['data'],

        'velocidade_permitida'      => $items[66][222]['data'],
        'velocidade_aferida'      => $items[232][222]['data'],
        'velocidade_considerada'      => $items[398][222]['data'],
        'instrumento'      => $items[563][222]['data'],
        'placa'      => $items[66][283]['data'],
        'marca_modelo'      => $items[398][283]['data'],
        'condutor'      => @$items[66][313]['data'],

        'auto_infracao' => $items[232][130]['data'],

        'agente_atuador' => $items[398][130]['data'],

        'data' => $items[563][130]['data'],

        'emissao' => $items[716][65]['data'],
        'renainf' => $items[716][100]['data'],
        'hora'    => $items[716][130]['data'],
        'pontos'  => $items[716][252]['data'],
        'data_apresentacao' => $itemsRaw[751][356]['data'],
    ];
}

$data = dataFromPdf('pdf/grm-3.pdf');
print_r($data);