organic: Add external BioC api

www/organic/external/bioc/operators.php

@@ -0,0 +1,99 @@
+<?php
+
+header('Access-Control-Allow-Origin: *');
+
+if ($_SERVER['REQUEST_METHOD'] !== 'GET' && $_SERVER['REQUEST_METHOD'] !== 'HEAD') {
+    header('Status: 405');
+    header('Content-Length: 0');
+    header('Allow: GET, HEAD');
+    exit;
+}
+
+function jenc($data): string {
+    return json_encode($data, JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE);
+}
+
+$info = $_SERVER['PATH_INFO'];
+if ($info !== '') {
+    $id = substr($info, 1);
+    header('Status: 404');
+    header('Content-Length: 0');
+    exit;
+}
+
+$source_groups = [
+    [ '',
+        '0892fdd7-6649-cb3b-75ef-a60649cfebfd', 'fde4ccef-0f36-4db9-9f2f-9bb6009c872a',
+        '8e6d20d3-6bbe-c7ae-7a77-fb7fa80ee291', '830b3bd7-f1da-e20e-300d-e561bfb8205b',
+        'ed025d59-056f-ffd0-79e0-05348fe88d59', '95d536c8-c869-9f72-1bce-4256a2a4ce49'
+    ], [ '',
+        'bef6b49a-3186-ba7f-fffb-281bfb4ca2d2', 'fc7a0cc3-c4b4-a4fb-5b7e-5cf9b68ed198',
+        '9aecd829-8f3c-6829-449f-9cb511e0c1ff', 'e7a7089b-0fd7-60b9-48af-c0988a90ec37',
+        '63d570d8-0ab2-b793-fe11-62e0777e4236'
+    ],
+];
+
+$country = [
+    'AT' => 'a1e51f85-27b5-a1e3-65f7-561569884d38',
+    'DE' => '1be4d468-ba47-08f1-f048-f45869af856f',
+][$_GET['country'] ?? null] ?? null;
+$postalCode = $_GET['postalCode'] ?? null;
+$name = $_GET['name'] ?? null;
+$idNr = $_GET['idNr'] ?? null;
+
+if ($country === null) {
+    header('Status: 400');
+    header('Content-Length: 0');
+    exit;
+}
+
+$data = [];
+
+$url = "https://www.bioc.info/search/producerSearchQuery?search[name]=" . urlencode($name) . "&search[citycode]=" . urlencode($postalCode) . "&search[operatorId]=" . urlencode($idNr) . "&search[country]=$country";
+if ($_GET['allDBs'] === 'true') {
+    $s = curl_init($url . implode('&sources[]=', $source_groups[0]));
+    curl_setopt($s, CURLOPT_RETURNTRANSFER, true);
+    if (($json = curl_exec($s)) === false) {
+        header('Status: 500');
+        header('Content-Length: 0');
+        exit;
+    }
+    $res = json_decode($json, true)['results'];
+    foreach ($res as $r) {
+        $data[$r['id']] = $r['row'];
+    }
+}
+
+$s = curl_init($url . implode('&sources[]=', $source_groups[1]));
+curl_setopt($s, CURLOPT_RETURNTRANSFER, true);
+if (($json = curl_exec($s)) === false) {
+    header('Status: 500');
+    header('Content-Length: 0');
+    exit;
+}
+$res = json_decode($json, true)['results'];
+foreach ($res as $r) {
+    $data[$r['id']] = $r['row'];
+}
+
+header('Content-Type: application/json; charset=UTF-8');
+
+$first = true;
+echo "{\"data\":[\n";
+foreach ($data as $id => $row) {
+    if (!$first) echo ",\n";
+    $row = explode("</", $row);
+    $auth = explode(' ', substr($row[3], strrpos($row[3], '>') + 1), 2);
+    echo "  ";
+    echo jenc([
+        'id' => $id,
+        'name' => substr($row[0], strrpos($row[0], '>') + 1),
+        'postalCode' => substr($row[1], strrpos($row[1], '>') + 1),
+        'city' => substr($row[2], strrpos($row[2], '>') + 1),
+        'authorityCode' => $auth[0],
+        'authorityName' => $auth[1],
+    ]);
+    $first = false;
+}
+echo "\n]}\n";
+

www/organic/external/bioqs/.attachment.py

@@ -0,0 +1,54 @@
+#!/bin/env python3
+
+import re
+import argparse
+import requests
+import sys
+
+
+BASE_URL = 'https://www.bioqs.at'
+URL = f'{BASE_URL}/ACM/faces/form/cms/portal/index.jsp'
+ACTION_RE = re.compile(r'action="([^"]*)"')
+HIDDEN_RE = re.compile(r'<input type="hidden" name="([^"]*)" .*?value="([^"]*)"')
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser()
+    parser.add_argument('cert_nr', type=str)
+    args = parser.parse_args()
+
+    s = requests.Session()
+    r = s.get(f'{URL}?menu_sid=5002')
+    uri = ACTION_RE.findall(r.text)[0]
+    hidden = {m[1]: m[2] for m in HIDDEN_RE.finditer(r.text)}
+
+    r = s.post(f'{BASE_URL}{uri}', data={
+        'PartnerCertSearchForm:pcs_seqidall': args.cert_nr,
+        'PartnerCertSearchForm:button_search': 'Suche starten...',
+        'PartnerCertSearchForm_SUBMIT': '1',
+        'javax.faces.ViewState': hidden['javax.faces.ViewState'],
+    })
+
+    p1 = r.text.find(f'>{args.cert_nr}<')
+    p2 = r.text.find('id="', p1)
+    p3 = r.text.find('"', p2 + 4)
+    if p1 == -1 or p2 == -1 or p3 == -1:
+        exit(1)
+    id = r.text[p2 + 4:p3]
+
+    r = s.post(f'{BASE_URL}{uri}', data={
+        'PartnerCertSearchForm:_idcl': id,
+        'PartnerCertSearchForm_SUBMIT': '1',
+        'javax.faces.ViewState': hidden['javax.faces.ViewState'],
+    })
+
+    if 'Content-Disposition' in r.headers:
+        dispo = r.headers['Content-Disposition']
+        if 'filename="' in dispo:
+            filename = dispo[dispo.find('filename="') + 10:dispo.rfind('"')]
+            print(filename, file=sys.stderr)
+    sys.stdout.buffer.write(r.content)
+
+
+if __name__ == '__main__':
+    main()

www/organic/external/bioqs/.operators.py

@@ -0,0 +1,92 @@
+#!/bin/env python3
+
+import re
+import argparse
+import requests
+import html
+import json
+import urllib.parse
+
+
+BASE_URL = 'https://www.bioqs.at'
+URL = f'{BASE_URL}/ACM/faces/form/cms/portal/index.jsp'
+ACTION_RE = re.compile(r'action="([^"]*)"')
+HIDDEN_RE = re.compile(r'<input type="hidden" name="([^"]*)" .*?value="([^"]*)"')
+
+ROW_RE = re.compile(r'<tr[^>]*>\s*(.*?)\s*</tr>', re.DOTALL)
+UNCOLLAPSED_ROW_RE = re.compile(r'<tr style="">(\s*<td>\s*(.*?)\s*</td>\s*){7}</tr>', re.DOTALL)
+COLLAPSED_ROW_RE = re.compile(r'<table width=[^>]*>\s*(.*?)\s*</table>', re.DOTALL)
+TD_RE = re.compile(r'<td[^>]*>\s*(.*?)\s*</td>', re.DOTALL)
+TAG_RE = re.compile(r'<[^>]*>')
+SPACE_RE = re.compile(r'\s+')
+ATTACHMENT_RE = re.compile(r"\[\['cert_attachment_sid','([^']*)'\]\]")
+
+
+def remove_tags(text: str) -> str:
+    return SPACE_RE.sub(' ', html.unescape(TAG_RE.sub(' ', text))).strip()
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser()
+    parser.add_argument('query', type=str)
+    args = parser.parse_args()
+    query = {'PartnerCertSearchForm:pcs_' + q.split('=', 1)[0]: urllib.parse.unquote(q.split('=', 1)[-1]) for q in args.query.split('&')}
+
+    s = requests.Session()
+    r = s.get(f'{URL}?menu_sid=5002')
+    uri = ACTION_RE.findall(r.text)[0]
+    hidden = {m[1]: m[2] for m in HIDDEN_RE.finditer(r.text)}
+
+    r = s.post(f'{BASE_URL}{uri}', data={
+        **query,
+        'PartnerCertSearchForm:button_search': 'Suche starten...',
+        'PartnerCertSearchForm_SUBMIT': '1',
+        'javax.faces.ViewState': hidden['javax.faces.ViewState'],
+    })
+
+    result_table = r.text[r.text.find('<table'):r.text.rfind('</table>') + 8]
+    uncollapsed_rows = [tuple(remove_tags(m[1])
+                              for m in TD_RE.finditer(row[0]))
+                        for row in UNCOLLAPSED_ROW_RE.finditer(result_table)]
+    collapsed_rows = [[tuple(remove_tags((ATTACHMENT_RE.search(m[1]) or m)[1]) for m in TD_RE.finditer(row[1]))
+                       for row in ROW_RE.finditer(tbl[0])]
+                      for tbl in COLLAPSED_ROW_RE.finditer(result_table)]
+    print('[')
+    first = True
+    for row, tbl in zip(uncollapsed_rows, collapsed_rows):
+        meta = {}
+        certificates = []
+        for srow in tbl:
+            if len(srow) == 1:
+                [k,v] = srow[0].split(':', 1)
+                meta[k.strip()] = v.strip()
+                continue
+            if len(srow) == 0:
+                continue
+            certificates.append({
+                'nr': srow[0],
+                'validFrom': '-'.join(reversed(srow[1].split('-'))),
+                'validTo': '-'.join(reversed(srow[2].split('-'))),
+                'type': srow[3],
+                'attachmentSid': srow[4],
+                'url': f'https://elwig.at/organic/external/bioqs/attachments/{urllib.parse.quote(srow[0])}',
+            })
+        if not first:
+            print(',', flush=True)
+        print('  ', json.dumps({
+            'idNr': row[0],
+            'lfbisNr': row[1] or None,
+            'name': row[2],
+            'postalCode': row[3],
+            'city': row[4],
+            'address': row[5],
+            'autorityName': meta['Kontrollstelle'],
+            'productGroups': meta['Bereiche'],
+            'certificates': certificates,
+        }, ensure_ascii=False), end='')
+        first = False
+    print('\n]')
+
+
+if __name__ == '__main__':
+    main()

www/organic/external/bioqs/attachments.php

@@ -0,0 +1,36 @@
+<?php
+
+header('Access-Control-Allow-Origin: *');
+
+if ($_SERVER['REQUEST_METHOD'] !== 'GET' && $_SERVER['REQUEST_METHOD'] !== 'HEAD') {
+    header('Status: 405');
+    header('Content-Length: 0');
+    header('Allow: GET, HEAD');
+    exit;
+}
+
+$info = $_SERVER['PATH_INFO'];
+if ($info === '') {
+    header('Status: 404');
+    header('Content-Length: 0');
+    exit;
+}
+$certId = substr($info, 1);
+
+$file = tmpfile();
+if (!$file) {
+    header('Status: 500');
+    header('Content-Length: 0');
+    exit;
+}
+$filename = stream_get_meta_data($file)['uri'];
+if (($pdfName = exec("python3 .attachment.py " . escapeshellarg($certId) . " 2>&1 > $filename ")) === false) {
+    header('Status: 500');
+    header('Content-Length: 0');
+    exit;
+}
+
+header('Content-Type: application/pdf');
+header('Content-Length: ' . filesize($filename));
+header('Content-Disposition: inline; filename="' . $pdfName . '"');
+readfile($filename);

www/organic/external/bioqs/operators.php

@@ -0,0 +1,43 @@
+<?php
+
+header('Access-Control-Allow-Origin: *');
+
+if ($_SERVER['REQUEST_METHOD'] !== 'GET' && $_SERVER['REQUEST_METHOD'] !== 'HEAD') {
+    header('Status: 405');
+    header('Content-Length: 0');
+    header('Allow: GET, HEAD');
+    exit;
+}
+
+$info = $_SERVER['PATH_INFO'];
+if ($info !== '') {
+    header('Status: 404');
+    header('Content-Length: 0');
+    exit;
+}
+
+$query = [];
+if (isset($_GET['country'])) {
+    $query[] = 'country=' . [
+        'AT' => '1',
+        'DE' => '2',
+    ][$_GET['country']] ?? '';
+}
+if (isset($_GET['postalCode'])) {
+    $query[] = 'zipcode=' . urlencode($_GET['postalCode']);
+}
+if (isset($_GET['name'])) {
+    $query[] = 'aname=' . urlencode($_GET['name']);
+}
+if (isset($_GET['idNr'])) {
+    $query[] = 'clientcode=' . urlencode($_GET['idNr']);
+}
+if (isset($_GET['lfbisNr'])) {
+    $query[] = 'lfbis=' . urlencode($_GET['lfbisNr']);
+}
+
+header('Content-Type: application/json; charset=UTF-8');
+
+echo "{\"data\":";
+passthru("python3 .operators.py " . escapeshellarg(implode('&', $query)));
+echo "}\n";

www/organic/external/easy-cert/operators.php

@@ -1,12 +1,5 @@
 <?php
 
-$info = $_SERVER['PATH_INFO'];
-if ($info !== '') {
-    header('Status: 404');
-    header('Content-Length: 0');
-    exit;
-}
-
 header('Access-Control-Allow-Origin: *');
 
 if ($_SERVER['REQUEST_METHOD'] !== 'GET' && $_SERVER['REQUEST_METHOD'] !== 'HEAD') {
@@ -16,6 +9,122 @@ if ($_SERVER['REQUEST_METHOD'] !== 'GET' && $_SERVER['REQUEST_METHOD'] !== 'HEAD
     exit;
 }
 
+function jenc($data): string {
+    return json_encode($data, JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE);
+}
+
+$info = $_SERVER['PATH_INFO'];
+if ($info !== '') {
+    $parts = explode(':', substr($info, 1), 2);
+    if (sizeof($parts) !== 2) {
+        $url = "https://elwig.at/organic/external/easy-cert/operators?idNr=" . substr($info, 1);
+        $s = curl_init($url);
+        curl_setopt($s, CURLOPT_RETURNTRANSFER, true);
+        if (($json = curl_exec($s)) === false) {
+            header('Status: 500');
+            header('Content-Length: 0');
+            exit;
+        }
+        $json = json_decode($json);
+        if ($json->{'totalCount'} !== 1) {
+            header('Status: 404');
+            header('Content-Length: 0');
+            exit;
+        }
+        header('Status: 303');
+        header('Location: ./' . $json->{'data'}[0]->{'db'} . ':' . $json->{'data'}[0]->{'id'});
+        exit;
+    }
+
+
+    $refDate = $_GET['referenceDate'] ?? null;
+    if ($refDate !== null) {
+        $refDate = explode('-', $refDate);
+        $refDate = $refDate[2] . '.' . $refDate[1] . '.' . $refDate[0];
+    }
+    $url = "https://www.easy-cert.com/htm/suchresultat-detail.htm?sprache=de&db=" . urlencode($parts[0]) . "&id=" . urlencode(str_replace('-', '_', $parts[1])) . "&historyDate=" . urlencode($refDate);
+    $s = curl_init($url);
+    curl_setopt($s, CURLOPT_RETURNTRANSFER, true);
+    if (($html = curl_exec($s)) === false) {
+        header('Status: 500');
+        header('Content-Length: 0');
+        exit;
+    }
+
+    $html = preg_replace('/<!--.*?-->/s', ' ', $html);
+    preg_match_all('@<tr>\s*<th[^>]*>([^<]*)</th>\s*<td[^>]*>([^<]*)</td>\s*</tr>@', $html, $matches, PREG_SET_ORDER);
+
+    $data = [];
+    foreach ($matches as $m) {
+        $data[$m[1]] = trim(html_entity_decode($m[2]));
+        if ($data[$m[1]] === '') {
+            $data[$m[1]] = null;
+        }
+    }
+    $plzOrt = $data['PLZ / Ort'] ?? null;
+    if ($plzOrt === null) {
+        $postalCode = null;
+        $city = null;
+    } else {
+        $p = explode(' ', $plzOrt, 2);
+        $postalCode = trim($p[0]);
+        $city = trim($p[1]);
+    }
+
+    if ($data['Name'] === null) {
+        header('Status: 404');
+        header('Content-Length: 0');
+        exit;
+    }
+
+    if (preg_match('@name="historydate_input" value="([^"]*)"@', $html, $matches) === 1) {
+        $refDate = $matches[1];
+        $refDate = explode('.', $refDate);
+        $refDate = $refDate[2] . '-' . $refDate[1] . '-' . $refDate[0];
+    } else {
+        $refDate = null;
+    }
+
+    $certs = [];
+    preg_match_all('@<tr>\s*<td nowrap>([^<]*)</td>\s*<td[^>]*>([^<]*)</td>\s*<td[^>]*>\s*([^<]*)\s*.*?</td>\s*<td [^\']*\'(https?://[^\']*)\'[^>]*>[^<]*</td>\s*</tr>@', $html, $matches, PREG_SET_ORDER);
+    foreach ($matches as $m) {
+        $certs[] = [
+            'nr' => $m[1],
+            'validUntil' => implode('-', array_reverse(explode('.', $m[2]))),
+            'type' => $m[3] === '' ? null : $m[3],
+            'pdfUrl' => str_replace('http://', 'https://', $m[4]),
+        ];
+    }
+
+    $labels = [];
+    preg_match_all('@<tr>\s*<td>([^<]*)</td>\s*<td[^>]*>([^<]*)</td>\s*<td[^>]*>\s*([^<]*)\s*.*?</td>\s*<td [^\']*\'(https?://[^\']*)\'[^>]*>[^<]*</td>\s*</tr>@', $html, $matches, PREG_SET_ORDER);
+    foreach ($matches as $m) {
+        $labels[] = [
+            'nr' => $m[1],
+            'validUntil' => implode('-', array_reverse(explode('.', $m[2]))),
+            'type' => $m[3] === '' ? null : $m[3],
+            'pdfUrl' => str_replace('http://', 'https://', $m[4]),
+        ];
+    }
+
+    header("Content-Type: application/json; charset=UTF-8");
+
+    echo '{"db":' . jenc($parts[0]) .
+        ',"id":' . jenc($parts[1]) .
+        ",\n  \"idNr\":" . jenc($data['ID-Nummer'] ?? null) .
+        ',"name":' . jenc($data['Name']) .
+        ',"address":' . jenc($data['Strasse'] ?? null) .
+        ',"postalCode":' . jenc($postalCode) .
+        ',"city":' . jenc($city) .
+        ',"countryCode":' . jenc($data['Land'] ?? null) .
+        ",\n  \"referenceDate\":" . jenc($refDate) .
+        ",\n  \"certificates\":" . jenc($certs) .
+        ",\n  \"privateStandardApprovals\":" . jenc($labels) .
+        '}';
+
+    exit;
+}
+
 $search_url = null;
 $url = null;
 $query_id = null;
@@ -43,9 +152,10 @@ if (isset($_GET['queryId'])) {
     $country = $_GET['country'] ?? null;
     $postalCode = $_GET['postalCode'] ?? null;
     $name = $_GET['name'] ?? null;
+    $idNr = $_GET['idNr'] ?? null;
     $renew = ($_GET['renew'] ?? 'false') === 'true';
 
-    $search_url = "https://www.easy-cert.com/htm/suchergebnis.htm?suchtyp=einfach&CountryCode=$country&PostalCode=$postalCode&Name=$name";
+    $search_url = "https://www.easy-cert.com/htm/suchergebnis.htm?suchtyp=einfach&CountryCode=$country&PostalCode=$postalCode&Name=$name&CustomerNumber=$idNr";
 
     if (!$renew && $cache = fopen('.cache.csv', 'r')) {
         while (($line = fgets($cache)) !== false) {
@@ -111,15 +221,28 @@ $sed = [
     's/^/  /',
     's/}$/},/',
 ];
+$replace = [
+    '"DB":' => '"db":',
+    '"Name":' => '"name":',
+    '"PostalCode":' => '"postalCode":',
+    '"Town":' => '"city":',
+    '"CustomerNumber":' => '"idNr":',
+    '"ID":' => '"id":',
+    '"CountryCode":' => '"countryCode":',
+    '"xx"' => 'null',
+    '"XX"' => 'null',
+    ':""' => ':null',
+];
+$replaceSed = array_map(fn($v, $k): string => "s/$k/$v/", $replace, array_keys($replace));
 
 header("Content-Type: application/json; charset=UTF-8");
 
-echo '{"searchUrl":' . json_encode($search_url) .
-    ',"queryId":' . json_encode($query_id) .
-    ',"rawFileUrl":' . json_encode($url) .
-    ',"timestamp":' . json_encode(gmdate('Y-m-d\TH:i:s\Z', $timestamp)) .
-    ',"limit":' . json_encode($limit) .
-    ',"offset":' . json_encode($offset) . ",\"data\":[\n";
+echo '{"searchUrl":' . jenc($search_url) .
+    ',"queryId":' . jenc($query_id) .
+    ',"rawFileUrl":' . jenc($url) .
+    ',"timestamp":' . jenc(gmdate('Y-m-d\TH:i:s\Z', $timestamp)) .
+    ',"limit":' . jenc($limit) .
+    ',"offset":' . jenc($offset) . ",\"data\":[\n";
 
 $fd_spec = [
     0 => ["pipe", "r"],  // stdin
@@ -138,14 +261,13 @@ $process = proc_open(
         "tee >(wc -l 1>&3) | " .                          // copy stdout into wc and write result into fd 3
         "tail -n +$offset | " .                           // apply offset
         ($limit !== null ? " head -n $limit | " : "") .   // optionally apply limit
+        "sed '" . implode(';', $replaceSed) . "' | " .  // replace strings in json
         "sed '\$s/.$//'"],                                // remove last comma of last line
     $fd_spec,
     $pipes
 );
 fclose($pipes[0]);
-while (($line = fgets($pipes[1])) !== false) {
-    echo $line;
-}
+fpassthru($pipes[1]);
 fclose($pipes[1]);
 $stderr = stream_get_contents($pipes[2]);
 fclose($pipes[2]);
@@ -153,4 +275,4 @@ $count = intval(trim(stream_get_contents($pipes[3])));
 fclose($pipes[3]);
 $return_value = proc_close($process);
 
-echo '],"totalCount":' . json_encode($count) .  "}\n";
+echo '],"totalCount":' . jenc($count) .  "}\n";

www/organic/external/lkv/operators.php

@@ -0,0 +1,33 @@
+<?php
+
+header('Access-Control-Allow-Origin: *');
+
+if ($_SERVER['REQUEST_METHOD'] !== 'GET' && $_SERVER['REQUEST_METHOD'] !== 'HEAD') {
+    header('Status: 405');
+    header('Content-Length: 0');
+    header('Allow: GET, HEAD');
+    exit;
+}
+
+function jenc($data): string {
+    return json_encode($data, JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE);
+}
+
+$info = $_SERVER['PATH_INFO'];
+if ($info !== '') {
+    header('Status: 404');
+    header('Content-Length: 0');
+    exit;
+}
+
+header('Content-Type: application/json; charset=UTF-8');
+
+echo "{\"data\":[\n";
+passthru(<<<EOF
+curl -s 'https://lkv.at/at/zertifizierung/themen/BIO/zertifizierte-BIO-Betriebe.php' | grep '<tr' -A 3 \
+| sed 's@<tr>@  {@;s@</tr>@  },@;s@<td><a .*href="\([^"]*\)".*>\(.*\)</a></td>@    "certUrl":"\\1","certNr":"\\2",@;s@<td>\(.*\)<br/>\(.*\)<br/>\([0-9]*\) \(.*\)</td>@    "name":"\\1","address":"\\2","postalCode":"\\3","city":"\\4"@;s@<td>\(.*\)<br/>\(.*\)</td>@    "name":"\\1","type":"\\2"@;\$s/.$//' \
+| sed 's@"certUrl":"\(/[^"]*\)"@"certUrl":"https://lkv.at\\1"@' \
+| sed 's@\s*",@",@g;s@:"\s*@:"@g'
+EOF);
+
+echo "]}\n";

www/organic/index.php

@@ -51,8 +51,20 @@ if (str_starts_with($info, '/certificates/')) {
 
     exit;
 } else if ($info === '/authorities') {
-    header('Content-Length: 0');
-    header('Status: 501');
+    header('Content-Type: application/json; charset=UTF-8');
+    echo <<<EOF
+{"data":[
+  {"id":"AT-BIO-301","countryCode":"AT","handle":"ABG","name":"Austria Bio Garantie GmbH","website":"https://www.bio-garantie.at/","apis":["easy-cert"]},
+  {"id":"AT-BIO-302","countryCode":"AT","handle":"ABG-LW","name":"Austria Bio Garantie – Landwirtschaft GmbH","website":"https://www.bio-garantie.at/","apis":["easy-cert"]},
+  {"id":"AT-BIO-401","countryCode":"AT","handle":"BIOS","name":"BIOS – Biokontrollservice Österreich GmbH","website":"https://www.bios-kontrolle.at/","apis":["bioqs"]},
+  {"id":"AT-BIO-402","countryCode":"AT","handle":"LACON","name":"LACON GmbH ","website":"https://www.lacon-institut.com/","apis":["easy-cert"]},
+  {"id":"AT-BIO-501","countryCode":"AT","handle":"SLK","name":"SLK GesmbH","website":"https://slk.at/","apis":["bioc"]},
+  {"id":"AT-BIO-901","countryCode":"AT","handle":"LVA","name":"LVA GmbH","website":"https://www.lva.at/","apis":[]},
+  {"id":"AT-BIO-902","countryCode":"AT","handle":"SGS","name":"SGS Austria Controll-Co. Ges.m.b.H.","website":"https://www.sgs.com/de-at ","apis":["bioc"]},
+  {"id":"AT-BIO-903","countryCode":"AT","handle":"LKV","name":"LKV Austria Gemeinnützige GmbH","website":"https://www.lkv.at/","apis":["bioc","lkv"]}
+]}
+
+EOF;
     exit;
 } else if (str_starts_with($info, '/authorities/')) {
     $code = $parts[2];

www/organic/pdf.php

@@ -0,0 +1,269 @@
+<?php
+
+if ($_SERVER['REQUEST_METHOD'] !== 'GET' && $_SERVER['REQUEST_METHOD'] !== 'POST' && $_SERVER['REQUEST_METHOD'] !== 'HEAD') {
+    header('Status: 405');
+    header('Content-Length: 0');
+    header('Allow: GET, POST, HEAD');
+    exit;
+}
+
+$info = $_SERVER['PATH_INFO'];
+if ($info !== '') {
+    header('Status: 404');
+    header('Content-Length: 0');
+    exit;
+}
+
+$url = isset($_GET['url']) ? str_replace(' ', '+', $_GET['url']) : null;
+$format = $_GET['format'] ?? 'json';
+
+function get_address($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(' ', $city)];
+}
+
+function jenc($data): string {
+    return json_encode($data, JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE);
+}
+
+$file = tmpfile();
+$headerfile = tmpfile();
+if (!$file || !$headerfile) {
+    header('Status: 500');
+    header('Content-Length: 0');
+    exit;
+}
+$filename = stream_get_meta_data($file)['uri'];
+$headerfilename = stream_get_meta_data($headerfile)['uri'];
+if ($_SERVER['REQUEST_METHOD'] === 'POST') {
+    $stdin = fopen("php://input", "rb");
+    while (!feof($stdin)) {
+        if (($data = fread($stdin, 8192)) === false)
+            break;
+        fwrite($file, $data);
+    }
+    fclose($stdin);
+} else {
+    if (exec("curl -s -D " . escapeshellarg($headerfilename) . " -o " . escapeshellarg($filename) . "  " . escapeshellarg($url)) === false) {
+        header('Status: 500');
+        header('Content-Length: 0');
+        exit;
+    }
+}
+
+if ($format === 'text') {
+    header('Content-Type: text/plain; charset=UTF-8');
+    passthru("pdftotext -raw " . escapeshellarg($filename) . " -");
+} else if ($format === 'sig') {
+    header('Content-Type: text/plain; charset=UTF-8');
+    passthru("pdfsig " . escapeshellarg($filename));
+} else if ($format === 'json') {
+    header('Content-Type: application/json; charset=UTF-8');
+    if (exec("pdftotext -raw " . escapeshellarg($filename) . " -", $text) === false) {
+        header('Status: 500');
+        header('Content-Length: 0');
+        exit;
+    }
+    $text = implode("\n", $text);
+    exec("pdfsig " . escapeshellarg($filename), $sig);
+    $sig = implode("\n", $sig);
+
+    $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' => $sData['Signature Field Name'],
+            ];
+        }
+
+        echo "{\"type\":\"traces\",\"lang\":\"$lang\",\"id\":\"$certId\",\"status\":\"$statusMap[$status]\"";
+        echo ",\n  \"operator\":{\"id\":" . jenc($operatorId).
+            ',"groupOfOperators":' . jenc(!str_starts_with($data['I.2'], '☑')) .
+            ',"name":' . jenc($operatorName) .
+            ',"address":' . jenc($opAddr) .
+            ',"postalCode":' . jenc($opPostal) .
+            ',"city":' . jenc($opCity) .
+            ',"countryCode":' . jenc($operator[sizeof($operator) - 1]) .
+            "},\n  \"authority\":{\"id\":" . jenc($authorityId) .
+            ',"name":' . jenc($authorityName) .
+            ',"address":' . jenc($aAddr) .
+            ',"postalCode":' . jenc($aPostal) .
+            ',"city":' . jenc($aCity) .
+            ',"countryCode":' . jenc($authority[sizeof($authority) - 1]) .
+            "},\n  \"activities\":" . jenc($activities) .
+            ",\n  \"productCategories\":" . jenc($products) .
+            ",\n  \"validFrom\":" . jenc($valid1) .
+            ',"validTo":' . jenc($valid2) .
+            ",\n  \"url\":\"$certUrl\"" .
+            ",\n  \"digitalSignatures\":" . jenc($sigs) .
+            "\n}\n";
+        exit;
+    }
+
+    if (preg_match('/AT-BIO-[0-9]{3}/', $text, $matches) === 1) {
+        $authorityId = $matches[0];
+        $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 = $m[0];
+            }
+        }
+        echo "{\"type\":\"$authorityId\",\"lang\":\"de\",\"id\":" . jenc($certId) . ",\"nr\":" . jenc($certNr);
+        echo ",\n  \"operator\":{},\n  \"authority\":{\"id\":" . jenc($authorityId) . "}}\n";
+        exit;
+    }
+
+    echo "{\"type\":\"unknown\"}\n";
+} else {
+    $headers = [];
+    $status_code = null;
+    foreach (explode("\n", file_get_contents($headerfilename)) as $line) {
+        if (trim($line) === '') break;
+        if ($status_code === null) {
+            $status_code = intval(explode(' ', $line)[1]);
+            if ($status_code !== 200)
+                break;
+            continue;
+        }
+        [$k,$v] = explode(':', $line, 2);
+        $k = strtolower(trim($k));
+        $v = trim($v);
+        $headers[$k] = $v;
+    }
+
+    if (str_starts_with($headers['content-type'], "application/pdf")) {
+        header('Content-Type: application/pdf');
+        $content_length = null;
+        if (isset($headers['content-length'])) {
+            $content_length = intval($headers['content-length']);
+            header('Content-Length: ' . $headers['content-length']);
+        }
+        $parts = explode('/', $url);
+        $realFilename = $parts[sizeof($parts) - 1];
+        if (isset($headers['content-disposition'])) {
+            preg_match('@filename="(.*?)"@', $headers['content-disposition'], $matches);
+            $realFilename = $matches[1];
+        }
+        header('Content-Disposition: inline; filename="' . $realFilename . '"');
+        fpassthru($file);
+    } else {
+        header('Status: 500');
+        header('Content-Length: 0');
+    }
+}