<?php

function get_address($array, $from, $to): array {
    $address = [];
    $postalCode = [];
    $city = [];
    for ($i = $to; $i >= $from; $i--) {
        $el = $array[$i];
        if (sizeof($postalCode) > 0) {
            if (sizeof($address) === 0) $el = rtrim($el, ", \n\r\t\v\0");
            if (strlen($el) === 0) continue;
            array_unshift($address, $el);
        } else if (preg_match("/^[A-Z0-9.\-]{3,},?$/", $el)) {
            array_unshift($postalCode, trim($el, ", \n\r\t\v\0"));
        } else {
            array_unshift($city, $el);
        }
    }
    return [implode(' ', $address), implode(' ', $postalCode), implode(' ', $city)];
}

function parse_pdf(string $filename, bool $loadUrl = false): array | null {
    if (!file_exists($filename))
        return null;
    if (exec("pdftotext -raw " . escapeshellarg($filename) . " -", $text) === false)
        return null;

    $text = implode("\n", $text);
    exec("pdfsig " . escapeshellarg($filename), $sig);
    $sig = implode("\n", $sig);

    if ($loadUrl) {
        $prefix = "/tmp/pdfimages-" . uniqid();
        exec("zbarimg -q --raw $(pdfimages -print-filenames " . escapeshellarg($filename) . " $prefix) | uniq; rm -rf $prefix-*.*", $qrCodes);
    } else {
        $qrCodes = null;
    }

    $r = preg_match('@([a-z]{2}) (https://webgate\.ec\.europa\.eu/tracesnt/directory/publication/organic-operator/(.*?)\.pdf) (\d+) / (\d+)@', $text, $matches);
    if ($r === 1) {
        // TRACES certificate

        $data = [];
        $parts = preg_split('@\n(I+\.\d+) ([^\n]*)@', $text, -1, PREG_SPLIT_DELIM_CAPTURE);
        $status = str_replace("\n", '', $parts[0]);
        for ($i = 3; $i < sizeof($parts); $i += 3) {
            $data[$parts[$i - 2]] = trim($parts[$i]);
        }

        $lang = $matches[1];
        $splitAddr = [
            'de' => 'Adresse',
            'en' => 'Address',
        ][$lang];
        $splitCountry = [
            'de' => 'Land',
            'en' => 'Country',
        ][$lang];
        $statusMap = [
            'de' => [
                'AUSGESTELLT' => 'issued',
            ],
            'en' => [
                'ISSUED' => 'issued',
            ]
        ][$lang];
        $activityMap = [
            'de' => [
                'Aufbereitung' => 'preparation',
                'Ausfuhr' => 'export',
                'Einfuhr' => 'import',
                'Lagerung' => 'storing',
                'Produktion' => 'production',
                'Vertrieb' => 'distribution',
                'Vertrieb/Inverkehrbringen' => 'distribution_placing_on_the_market',
            ],
            'en' => [
                'Distribution' => 'distribution',
                'Distribution/Placing on the market' => 'distribution_placing_on_the_market',
                'Export' => 'export',
                'Import' => 'import',
                'Preparation' => 'preparation',
                'Production' => 'production',
                'Storing' => 'storing',
            ],
        ][$lang];

        $certUrl = $matches[2];
        $certId = $matches[3];
        $authorityId = explode('.', $certId)[0];
        $operatorId = explode('.', $certId)[1];
        $operator = preg_split('@\s+@', trim($data['I.3']));
        $p1 = array_search($splitAddr, $operator);
        $p2 = array_search($splitCountry, $operator);
        $operatorName = trim(implode(' ', array_filter($operator, fn($k,$i) => $i > 0 && $i < $p1, ARRAY_FILTER_USE_BOTH)), ', ');
        [$opAddr, $opPostal, $opCity] = get_address($operator, $p1 + 1, $p2 - 1);

        $authority = preg_split('@\s+@', trim($data['I.4']));
        $until = array_search("($authorityId)", $authority);
        $p1 = array_search($splitAddr, $authority);
        $p2 = array_search($splitCountry, $authority);
        $authorityName = implode(' ', array_filter($authority, fn($k,$i) => $i > 0 && $i < $p1 - 1 && ($i !== $p1 - 2 || !str_starts_with($k, '(')), ARRAY_FILTER_USE_BOTH));
        [$aAddr, $aPostal, $aCity] = get_address($authority, $p1 + 1, $p2 - 1);

        $activities = [];
        foreach (explode("\n", $data['I.5']) as $a) {
            $activities[] = $activityMap[trim($a, '• ')];
        }

        preg_match_all('/\([a-g]\)/', $data['I.6'], $matches, PREG_SET_ORDER);
        $products = [];
        foreach ($matches as $m) {
            $products[] = $m[0];
        }

        preg_match_all('@\d+/\d+/\d+@', $data['I.8'], $matches, PREG_SET_ORDER);
        $valid1 = implode('-', array_reverse(explode('/', $matches[0][0])));
        $valid2 = implode('-', array_reverse(explode('/', $matches[1][0])));

        $sigs = [];
        foreach (array_slice(explode("\nSignature #", $sig), 1) as $s) {
            $sData = [];
            $sData2 = [];
            preg_match_all('/\n {2}- (([^:\n]*): )?([^\n]*)/', $s, $matches, PREG_SET_ORDER);
            foreach ($matches as $m) {
                if (strlen($m[2]) === 0) {
                    $sData2[] = $m[3];
                } else {
                    $sData[$m[2]] = $m[3];
                }
            }
            $sigs[] = [
                'signerCommonName' => $sData['Signer Certificate Common Name'],
                'valid' => $sData['Signature Validation'] === 'Signature is Valid.',
                'trusted' => $sData['Certificate Validation'] === 'Certificate is Trusted.',
                'totalDocument' => in_array('Total document signed', $sData2),
                'timestamp' => gmdate('Y-m-d\TH:i:s\Z', strtotime($sData['Signing Time'])),
                'type' => $sData['Signature Type'],
                'hashAlgorithm' => $sData['Signing Hash Algorithm'],
                'signerDistinguishedName' => $sData['Signer full Distinguished Name'],
                'fieldName' => $sData['Signature Field Name'],
            ];
        }

        return [
            'type' => 'traces',
            'lang' => $lang,
            'id' => $certId,
            'status' => $statusMap[$status],
            'operator' => [
                'id' => $operatorId,
                'groupOfOperators' => !str_starts_with($data['I.2'], '☑'),
                'name' => $operatorName,
                'address' => $opAddr,
                'postalCode' => $opPostal,
                'city' => $opCity,
                'countryCode' => $operator[sizeof($operator) - 1],
            ],
            'authority' => [
                'id' => $authorityId,
                'name' => $authorityName,
                'address' => $aAddr,
                'postalCode' => $aPostal,
                'city' => $aCity,
                'countryCode' => $authority[sizeof($authority) - 1],
            ],
            'activities' => $activities,
            'productCategories' => $products,
            'validFrom' => $valid1,
            'validTo' => $valid2,
            'url' => $certUrl,
            'digitalSignatures' => $sigs,
            'qrCodes' => $qrCodes,
        ];
    }

    $isLacon = str_contains($text, "\nLACON GmbH\n") && (str_contains($text, "\nAnlage zum Zertifikat\n"));
    if (preg_match('/AT-BIO-[0-9]{3}/', $text, $matches) === 1 || $isLacon) {
        $authorityId = $matches[0] ?? ($isLacon ? 'AT-BIO-402' : null);
        $certId = null;
        $certNr = null;
        if (preg_match("/$authorityId\.040-[0-9]{7}\.[0-9]{4}\.[0-9]{3}/", $text, $matches) === 1) {
            $certId = $matches[0];
        }
        if (preg_match_all("/\b[0-9]+([._-])[0-9]+\g{-1}[0-9]+\b/", $text, $matches, PREG_SET_ORDER) !== false) {
            foreach ($matches as $m) {
                if (strlen($m[0]) > 10 && !str_ends_with($certId, $m[0]))
                    $certNr = $m[0];
            }
        }

        if ($certId === null && $qrCodes !== null && sizeof($qrCodes) > 0 && str_starts_with($qrCodes[0], 'www.easy-cert.com/')) {
            $res = exec("curl -L " . escapeshellarg($qrCodes[0]) .  " | grep -A 100 '<table' | grep -B 100 '</table>' | sed 's/<[^>]*>//g;s/\\s\\+/ /g;s/^ //g;s/ $//g;s/\\n//m' | grep 'Certificate number' -A 1");
            if ($res !== false && preg_match("/$authorityId\.040-[0-9]{7}\.[0-9]{4}\.[0-9]{3}/", $res, $matches) === 1) {
                $certId = $matches[0];
            }
        }

        $operator = ['lfbisNr' => null];
        if ($authorityId === 'AT-BIO-301') {
            // TODO
        } else if ($authorityId === 'AT-BIO-302') {
            if (preg_match("/^lfbis[^0-9]*([0-9]{6,7})$/im", $text, $matches) === 1)
                $operator['lfbisNr'] = str_pad($matches[1], 7, '0', STR_PAD_LEFT);
        } else if ($authorityId === 'AT-BIO-401') {
            // TODO
        } else if ($authorityId === 'AT-BIO-402') {
            if (preg_match("/^Betriebsnummer ([0-9]{6,7})$/im", $text, $matches) === 1)
                $operator['lfbisNr'] = str_pad($matches[1], 7, '0', STR_PAD_LEFT);
        } else if ($authorityId === 'AT-BIO-902') {
            if (preg_match("/ lfbis:([0-9]{6,7})$/im", $text, $matches) === 1)
                $operator['lfbisNr'] = str_pad($matches[1], 7, '0', STR_PAD_LEFT);
        } else if ($authorityId === 'AT-BIO-903') {
            // TODO
        }

        return [
            'type' => $authorityId,
            'lang' => 'de',
            'id' => $certId,
            'nr' => $certNr,
            'operator' => $operator,
            'authority' => ['id' => $authorityId],
            'qrCodes' => $qrCodes,
        ];
    }

    return ['type' => 'unknown'];
}