organic: Enhance external apis

www/organic/external/bioc/operators.php

@@ -1,11 +1,15 @@
 <?php
 
 header('Access-Control-Allow-Origin: *');
+header('Content-Type: application/json; charset=UTF-8');
+
+$base = 'https://' . $_SERVER['SERVER_NAME'] . explode('?', $_SERVER['REQUEST_URI'])[0];
 
 if ($_SERVER['REQUEST_METHOD'] !== 'GET' && $_SERVER['REQUEST_METHOD'] !== 'HEAD') {
     header('Status: 405');
-    header('Content-Length: 0');
+    header('Content-Length: 56');
     header('Allow: GET, HEAD');
+    echo "{\"error\":\"method not allowed\",\"allow\":[\"GET\",\"HEAD\"]}\n";
     exit;
 }
 
@@ -16,17 +20,84 @@ function jenc($data): string {
 $info = $_SERVER['PATH_INFO'];
 if ($info !== '') {
     $id = substr($info, 1);
+
+    $url = "https://www.bioc.info/search/producerdetail?producer=" . urlencode($id);
+    $s = curl_init($url);
+    curl_setopt($s, CURLOPT_RETURNTRANSFER, true);
+    if (($html = curl_exec($s)) === false) {
+        header('Status: 500');
+        header('Content-Length: 34');
+        echo "{\"error\":\"internal server error\"}\n";
+        exit;
+    }
+    $p1 = strpos($html, '<main>');
+    $p2 = strpos($html, '</main>');
+    if ($p1 === false || $p2 === false) {
         header('Status: 404');
-    header('Content-Length: 0');
+        header('Content-Length: 22');
+        echo "{\"error\":\"not found\"}\n";
+        exit;
+    }
+    $html = substr($html, $p1 + 6, $p2 - $p1 - 6);
+
+    $data = [];
+    preg_match_all('@<div class="row">\s*<div class="col[^>]*?>\s*(.*?)\s*</div>\s*<div class="col[^>]*?>\s*(.*?)\s*</div>\s*</div>@s', $html, $matches, PREG_SET_ORDER);
+    foreach ($matches as $m) {
+        $data[$m[1]] = trim(preg_replace('/\s*\n\s*/', "\n", html_entity_decode(preg_replace('/<.*?>/', '', str_replace('<br/>', "\n", preg_replace('/\s+/', ' ', trim($m[2])))))));
+    }
+
+    $certRows = [];
+    preg_match_all('@<tr>\s*<td.*?href="([^"]*)">\s*(.*?)\s*</td>\s*<td>\s*(.*?)\s*</td>\s*<td>\s*(.*?)\s*</td>\s*<td>\s*(.*?)\s*</td>\s*</tr>@s', $html, $matches, PREG_SET_ORDER);
+    foreach ($matches as $m) {
+        $cert = [];
+        foreach (array_splice($m, 1) as $s) {
+            $cert[] = trim(preg_replace('/\s*\n\s*/', "\n", preg_replace('/<.*?>/', '', str_replace('<br/>', "\n", preg_replace('/\s+/', ' ', trim(html_entity_decode($s)))))));
+        }
+        $certRows[] = $cert;
+    }
+
+    $lines = explode("\n", $data['Address']);
+    [$cc, $_, $postalCode, $city] = explode(' ', $lines[2], 4);
+
+    $certs = [];
+    foreach ($certRows as $row) {
+        [$nr, $lang] = explode('(', $row[1], 2);
+        [$from, $to] = explode(' - ', $row[3], 2);
+        $certs[] = [
+            'nr' => trim($nr),
+            'language' => substr($lang, 0, 2),
+            'pdfUrl' => $row[0],
+            'type' => $row[2],
+            'validFrom' => date('Y-m-d', strtotime($from)),
+            'validUntil' => date('Y-m-d', strtotime($to)),
+        ];
+    }
+
+    [$authCode, $authName] = explode(' ', $data['Certification body'], 2);
+    echo jenc([
+        'id' => $id,
+        'idNr' => $data['Operator ID'] ?? null,
+        'name' => $lines[0],
+        'address' => $lines[1],
+        'postalCode' => $postalCode,
+        'city' => $city,
+        'countryCode' => $cc,
+        'authority' => ['code' => $authCode, 'name' => $authName],
+        'url' => $base,
+        'originUrl' => $url,
+        'certificates' => $certs,
+    ]);
     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'],
     ['',
-        '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'
@@ -43,38 +114,44 @@ $idNr = $_GET['idNr'] ?? null;
 
 if ($country === null) {
     header('Status: 400');
-    header('Content-Length: 0');
+    header('Content-Length: 88');
+    echo "{\"error\":\"bad request\",\"message\":\"The 'country' request URL query parameter is required\"}\n";
     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') {
+$m = curl_multi_init();
-    $s = curl_init($url . implode('&sources[]=', $source_groups[0]));
+$requests = [];
+foreach ($source_groups as $group) {
+    $s = curl_init($url . implode('&sources[]=', $group));
+    $requests[] = $s;
     curl_setopt($s, CURLOPT_RETURNTRANSFER, true);
-    if (($json = curl_exec($s)) === false) {
+    curl_multi_add_handle($m, $s);
-        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[]=', $source_groups[1]));
+$running = 1;
-curl_setopt($s, CURLOPT_RETURNTRANSFER, true);
+do {
-if (($json = curl_exec($s)) === false) {
+    curl_multi_exec($m, $running);
+} while ($running);
+foreach ($requests as $r) {
+    curl_multi_remove_handle($m, $r);
+}
+curl_multi_close($m);
+
+foreach ($requests as $s) {
+    if (($json = curl_multi_getcontent($s)) === false) {
         header('Status: 500');
-    header('Content-Length: 0');
+        header('Content-Length: 34');
+        echo "{\"error\":\"internal server error\"}\n";
         exit;
     }
     $res = json_decode($json, true)['results'];
     foreach ($res as $r) {
         $data[$r['id']] = $r['row'];
     }
+}
 
 header('Content-Type: application/json; charset=UTF-8');
 
@@ -90,8 +167,8 @@ foreach ($data as $id => $row) {
         'name' => substr($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],
+        'authority' => ['code' => $auth[0], 'name' => $auth[1]],
-        'authorityName' => $auth[1],
+        'url' => "$base/$id",
     ]);
     $first = false;
 }

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

@@ -18,10 +18,16 @@ def main() -> None:
     args = parser.parse_args()
 
     s = requests.Session()
+    while True:
+        try:
             r = s.get(f'{URL}?menu_sid=5002')
             uri = ACTION_RE.findall(r.text)[0]
+            break
+        except IndexError:
+            pass
     hidden = {m[1]: m[2] for m in HIDDEN_RE.finditer(r.text)}
 
+    while True:
         r = s.post(f'{BASE_URL}{uri}', data={
             'PartnerCertSearchForm:pcs_seqidall': args.cert_nr,
             'PartnerCertSearchForm:button_search': 'Suche starten...',
@@ -29,12 +35,17 @@ def main() -> None:
             'javax.faces.ViewState': hidden['javax.faces.ViewState'],
         })
 
+        if 'class="msg_error"' in r.text:
+            print(f'404', file=sys.stderr)
+            exit(1)
+
         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)
+            continue
         id = r.text[p2 + 4:p3]
+        break
 
     r = s.post(f'{BASE_URL}{uri}', data={
         'PartnerCertSearchForm:_idcl': id,

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

@@ -33,8 +33,13 @@ def main() -> None:
     query = {'PartnerCertSearchForm:pcs_' + q.split('=', 1)[0]: urllib.parse.unquote(q.split('=', 1)[-1]) for q in args.query.split('&')}
 
     s = requests.Session()
+    while True:
+        try:
             r = s.get(f'{URL}?menu_sid=5002')
             uri = ACTION_RE.findall(r.text)[0]
+            break
+        except IndexError:
+            pass
     hidden = {m[1]: m[2] for m in HIDDEN_RE.finditer(r.text)}
 
     r = s.post(f'{BASE_URL}{uri}', data={
@@ -69,7 +74,7 @@ def main() -> None:
                 '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])}',
+                'pdfUrl': f'https://elwig.at/organic/external/bioqs/attachments/{urllib.parse.quote(srow[0])}',
             })
         if not first:
             print(',', flush=True)
@@ -80,10 +85,10 @@ def main() -> None:
             'postalCode': row[3],
             'city': row[4],
             'address': row[5],
-            'autorityName': meta['Kontrollstelle'],
+            'authority': {'name': meta['Kontrollstelle']},
             'productGroups': meta['Bereiche'],
             'certificates': certificates,
-        }, ensure_ascii=False), end='')
+        }, ensure_ascii=False, separators=(',',':')), end='')
         first = False
     print('\n]')
 

www/organic/external/bioqs/attachments.php

@@ -3,30 +3,38 @@
 header('Access-Control-Allow-Origin: *');
 
 if ($_SERVER['REQUEST_METHOD'] !== 'GET' && $_SERVER['REQUEST_METHOD'] !== 'HEAD') {
+    header('Content-Type: application/json; charset=UTF-8');
     header('Status: 405');
-    header('Content-Length: 0');
+    header('Content-Length: 56');
     header('Allow: GET, HEAD');
+    echo "{\"error\":\"method not allowed\",\"allow\":[\"GET\",\"HEAD\"]}\n";
     exit;
 }
 
 $info = $_SERVER['PATH_INFO'];
 if ($info === '') {
+    header('Content-Type: application/json; charset=UTF-8');
     header('Status: 404');
-    header('Content-Length: 0');
+    header('Content-Length: 22');
+    echo "{\"error\":\"not found\"}\n";
     exit;
 }
 $certId = substr($info, 1);
 
 $file = tmpfile();
 if (!$file) {
+    header('Content-Type: application/json; charset=UTF-8');
     header('Status: 500');
-    header('Content-Length: 0');
+    header('Content-Length: 34');
+    echo "{\"error\":\"internal server error\"}\n";
     exit;
 }
 $filename = stream_get_meta_data($file)['uri'];
-if (($pdfName = exec("python3 .attachment.py " . escapeshellarg($certId) . " 2>&1 > $filename ")) === false) {
+if (($pdfName = exec("python3 .attachment.py " . escapeshellarg($certId) . " 2>&1 > $filename ")) === "404") {
-    header('Status: 500');
+    header('Content-Type: application/json; charset=UTF-8');
-    header('Content-Length: 0');
+    header('Status: 404');
+    header('Content-Length: 22');
+    echo "{\"error\":\"not found\"}\n";
     exit;
 }
 

www/organic/external/bioqs/operators.php

@@ -1,18 +1,21 @@
 <?php
 
 header('Access-Control-Allow-Origin: *');
+header('Content-Type: application/json; charset=UTF-8');
 
 if ($_SERVER['REQUEST_METHOD'] !== 'GET' && $_SERVER['REQUEST_METHOD'] !== 'HEAD') {
     header('Status: 405');
-    header('Content-Length: 0');
+    header('Content-Length: 56');
     header('Allow: GET, HEAD');
+    echo "{\"error\":\"method not allowed\",\"allow\":[\"GET\",\"HEAD\"]}\n";
     exit;
 }
 
 $info = $_SERVER['PATH_INFO'];
 if ($info !== '') {
     header('Status: 404');
-    header('Content-Length: 0');
+    header('Content-Length: 22');
+    echo "{\"error\":\"not found\"}\n";
     exit;
 }
 
@@ -36,8 +39,6 @@ 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

@@ -1,11 +1,15 @@
 <?php
 
 header('Access-Control-Allow-Origin: *');
+header("Content-Type: application/json; charset=UTF-8");
+
+$base = 'https://' . $_SERVER['SERVER_NAME'] . explode('?', $_SERVER['REQUEST_URI'])[0];
 
 if ($_SERVER['REQUEST_METHOD'] !== 'GET' && $_SERVER['REQUEST_METHOD'] !== 'HEAD') {
     header('Status: 405');
-    header('Content-Length: 0');
+    header('Content-Length: 56');
     header('Allow: GET, HEAD');
+    echo "{\"error\":\"method not allowed\",\"allow\":[\"GET\",\"HEAD\"]}\n";
     exit;
 }
 
@@ -22,13 +26,15 @@ if ($info !== '') {
         curl_setopt($s, CURLOPT_RETURNTRANSFER, true);
         if (($json = curl_exec($s)) === false) {
             header('Status: 500');
-            header('Content-Length: 0');
+            header('Content-Length: 34');
+            echo "{\"error\":\"internal server error\"}\n";
             exit;
         }
         $json = json_decode($json);
         if ($json->{'totalCount'} !== 1) {
             header('Status: 404');
-            header('Content-Length: 0');
+            header('Content-Length: 22');
+            echo "{\"error\":\"not found\"}\n";
             exit;
         }
         header('Status: 303');
@@ -36,7 +42,6 @@ if ($info !== '') {
         exit;
     }
 
-
     $refDate = $_GET['referenceDate'] ?? null;
     if ($refDate !== null) {
         $refDate = explode('-', $refDate);
@@ -47,7 +52,8 @@ if ($info !== '') {
     curl_setopt($s, CURLOPT_RETURNTRANSFER, true);
     if (($html = curl_exec($s)) === false) {
         header('Status: 500');
-        header('Content-Length: 0');
+        header('Content-Length: 34');
+        echo "{\"error\":\"internal server error\"}\n";
         exit;
     }
 
@@ -73,7 +79,8 @@ if ($info !== '') {
 
     if ($data['Name'] === null) {
         header('Status: 404');
-        header('Content-Length: 0');
+        header('Content-Length: 22');
+        echo "{\"error\":\"not found\"}\n";
         exit;
     }
 
@@ -107,21 +114,21 @@ if ($info !== '') {
         ];
     }
 
-    header("Content-Type: application/json; charset=UTF-8");
+    echo jenc([
-
+        'db' => $parts[0],
-    echo '{"db":' . jenc($parts[0]) .
+        'id' => $parts[1],
-        ',"id":' . jenc($parts[1]) .
+        'idNr' => $data['ID-Nummer'] ?? null,
-        ",\n  \"idNr\":" . jenc($data['ID-Nummer'] ?? null) .
+        'name' => $data['Name'],
-        ',"name":' . jenc($data['Name']) .
+        'address' => $data['Strasse'] ?? null,
-        ',"address":' . jenc($data['Strasse'] ?? null) .
+        'postalCode' => $postalCode,
-        ',"postalCode":' . jenc($postalCode) .
+        'city' => $city,
-        ',"city":' . jenc($city) .
+        'countryCode' => $data['Land'] ?? null,
-        ',"countryCode":' . jenc($data['Land'] ?? null) .
+        'url' => $base,
-        ",\n  \"referenceDate\":" . jenc($refDate) .
+        'originUrl' => $url,
-        ",\n  \"certificates\":" . jenc($certs) .
+        'referenceDate' => $refDate,
-        ",\n  \"privateStandardApprovals\":" . jenc($labels) .
+        'certificates' => $certs,
-        '}';
+        'privateStandardApprovals' => $labels,
-
+    ]);
     exit;
 }
 
@@ -207,12 +214,12 @@ if ($url === null) {
 }
 
 if ($url === null) {
-    header("Status: 500");
+    header('Status: 500');
-    header("Content-Length: 0");
+    header('Content-Length: 34');
+    echo "{\"error\":\"internal server error\"}\n";
     exit;
 }
 
-
 $sed = [
     's/^.*"searchresults":\[//',
     's/],.*$//',
@@ -235,8 +242,6 @@ $replace = [
 ];
 $replaceSed = array_map(fn($v, $k): string => "s/$k/$v/", $replace, array_keys($replace));
 
-header("Content-Type: application/json; charset=UTF-8");
-
 echo '{"searchUrl":' . jenc($search_url) .
     ',"queryId":' . jenc($query_id) .
     ',"rawFileUrl":' . jenc($url) .
@@ -262,6 +267,7 @@ $process = proc_open(
         "tail -n +$offset | " .                           // apply offset
         ($limit !== null ? " head -n $limit | " : "") .   // optionally apply limit
         "sed '" . implode(';', $replaceSed) . "' | " .  // replace strings in json
+        "sed 's|{\\(\"db\":\"\\([^\"]*\\)\".*\"id\":\"\\([^\"]*\\)\".*\\)}|{\\1,\"url\":\"$base/\\2:\\3\"}|' | " .  // add "url"
         "sed '\$s/.$//'"],                                // remove last comma of last line
     $fd_spec,
     $pipes

www/organic/external/index.php

@@ -0,0 +1,45 @@
+<?php
+
+$APIS = [
+    'bioc' => ['id' => 'bioc', 'endpoints' => ['https://elwig.at/organic/external/bioc/operators']],
+    'bioqs' => ['id' => 'bioqs', 'endpoints' => ['https://elwig.at/organic/external/bioqs/operators', 'https://elwig.at/organic/external/bioqs/attachments']],
+    'easy-cert' => ['id' => 'easy-cert', 'endpoints' => ['https://elwig.at/organic/external/easy-cert/operators']],
+    'lkv' => ['id' => 'lkv', 'endpoints' => ['https://elwig.at/organic/external/lkv/operators']],
+];
+
+function jenc($data): string {
+    return json_encode($data, JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE);
+}
+
+header('Access-Control-Allow-Origin: *');
+header('Content-Type: application/json; charset=UTF-8');
+
+if ($_SERVER['REQUEST_METHOD'] !== 'GET' && $_SERVER['REQUEST_METHOD'] !== 'HEAD') {
+    header('Status: 405');
+    header('Content-Length: 56');
+    header('Allow: GET, HEAD');
+    echo "{\"error\":\"method not allowed\",\"allow\":[\"GET\",\"HEAD\"]}\n";
+    exit;
+}
+
+if ($_SERVER['PATH_INFO'] !== '') {
+    $code = rtrim(substr($_SERVER['PATH_INFO'], 1), '/');
+    header('Content-Type: application/json; charset=UTF-8');
+    if (array_key_exists($code, $APIS)) {
+        echo jenc($APIS[$code]);
+        echo "\n";
+        exit;
+    }
+    header('Status: 404');
+    header('Content-Length: 22');
+    echo "{\"error\":\"not found\"}\n";
+}
+
+echo "{\"apis\":[\n";
+$first = true;
+foreach ($APIS as $api) {
+    if (!$first) echo ",\n";
+    echo "  " . jenc($api);
+    $first = false;
+}
+echo "\n]}\n";

www/organic/external/lkv/operators.php

@@ -1,11 +1,15 @@
 <?php
 
+$url = 'https://lkv.at/at/zertifizierung/themen/BIO/zertifizierte-BIO-Betriebe.php';
+
 header('Access-Control-Allow-Origin: *');
+header('Content-Type: application/json; charset=UTF-8');
 
 if ($_SERVER['REQUEST_METHOD'] !== 'GET' && $_SERVER['REQUEST_METHOD'] !== 'HEAD') {
     header('Status: 405');
-    header('Content-Length: 0');
+    header('Content-Length: 56');
     header('Allow: GET, HEAD');
+    echo "{\"error\":\"method not allowed\",\"allow\":[\"GET\",\"HEAD\"]}\n";
     exit;
 }
 
@@ -16,18 +20,39 @@ function jenc($data): string {
 $info = $_SERVER['PATH_INFO'];
 if ($info !== '') {
     header('Status: 404');
-    header('Content-Length: 0');
+    header('Content-Length: 22');
+    echo "{\"error\":\"not found\"}\n";
     exit;
 }
 
 header('Content-Type: application/json; charset=UTF-8');
 
-echo "{\"data\":[\n";
+$postalCode = '';
+if (isset($_GET['postalCode']) && ctype_digit($_GET['postalCode'])) {
+    $postalCode = $_GET['postalCode'];
+}
+$lfbisNr = '';
+if (isset($_GET['lfbisNr'])) {
+    if (strlen($_GET['lfbisNr']) === 7) {
+        $lfbisNr = '"' . $_GET['lfbisNr'];
+    } else {
+        $lfbisNr = 'invalid';
+    }
+}
+
+echo "{\"sourceUrl\":\"$url\",\"data\":[\n";
 passthru(<<<EOF
-curl -s 'https://lkv.at/at/zertifizierung/themen/BIO/zertifizierte-BIO-Betriebe.php' | grep '<tr' -A 3 \
+curl -s '$url' | grep -A3 '<tr' \
-| 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@<tr>@  {@;s@</tr>@  },@;s@<td><a .*href="\([^"]*\)".*>\(.*\)</a></td>@    "certUrl":"\\1","certNr":"\\2",@;s@<td>\(.*\)<br/>\(.*\)<br/>\([0-9]*\) \([^<]*\)\(<br/>\)\?</td>@    "name":"\\1","address":"\\2","postalCode":"\\3","city":"\\4"@;s@<td>\(.*\)<br/>\(.*\)</td>@    "name":"\\1","type":"\\2"@' \
-| sed 's@"certUrl":"\(/[^"]*\)"@"certUrl":"https://lkv.at\\1"@' \
+| sed 's@"certUrl":"\(/[^"]*\)"@"certUrl":"https://lkv.at\\1"@;s@"certNr":"[^0-9]*0\?\([0-9]\{7\}\|[0-9]\{1,4\}\|[0-9]\{7,\}\)_[^"]*"@\\0,"id":"\\1"@;s@"id":"\([0-9]\{7\}\)"@\\0,"lfbisNr":"\\1"@;s@"id":"\([^"]\{,6\}\|[^"]\{8,\}\)"@\\0,"lfbisNr":null@' \
+| sed 's@"certUrl":"\([^"]*\)","certNr":"\([^"]*\)",@"certificates":[{"pdfUrl":"\\1","nr":"\\2"}],\\n    @' \
+| perl -ne 'BEGIN { our \$N = 8; our @buf; } push @buf, \$_; if (@buf > \$N) { print shift @buf; } my \$chunk = join "", @buf;
+    if (\$chunk =~ /(}],\n(?:.*\n){6}\s*"name":"[^"]*","type":)/) {
+        \$chunk =~ s/(}],\n((?:.*\n){2})(?:.*\n){2}\s*"certificates":\[\{"pdfUrl":"([^"]*)","nr":"([^"]*)"\}\],\n.*\n\s*"name":"[^"]*","type":"([^"]*)")/},{"pdfUrl":"\$3","nr":"\$4","type":"\$5"}],\n\$2/;
+        @buf = map { "\$_\n" } split /\n/, \$chunk;
+    } END { print for @buf; }' \
+| grep -B3 -A1 --no-group-separator '"postalCode":"$postalCode' | grep -B2 -A2 --no-group-separator '"lfbisNr":$lfbisNr' \
+| sed '\$s/.$//'
 | sed 's@\s*",@",@g;s@:"\s*@:"@g'
 EOF);
-
 echo "]}\n";

www/organic/index.php

@@ -1,16 +1,58 @@
 <?php
 
+$AUTHORITIES = [
+    'AT-BIO-301' => [
+        'id' => 'AT-BIO-301', 'countryCode' => 'AT', 'handle' => 'ABG',
+        'name' => 'Austria Bio Garantie GmbH', 'website' => 'https://www.bio-garantie.at/', 'apis' => ['easy-cert']
+    ],
+    'AT-BIO-302' => [
+        'id' => 'AT-BIO-302', 'countryCode' => 'AT', 'handle' => 'ABG-LW',
+        'name' => 'Austria Bio Garantie – Landwirtschaft GmbH', 'website' => 'https://www.bio-garantie.at/', 'apis' => ['easy-cert']
+    ],
+    'AT-BIO-401' => [
+        'id' => 'AT-BIO-401', 'countryCode' => 'AT', 'handle' => 'BIOS',
+        'name' => 'BIOS – Biokontrollservice Österreich GmbH', 'website' => 'https://www.bios-kontrolle.at/', 'apis' => ['bioqs']
+    ],
+    'AT-BIO-402' => [
+        'id' => 'AT-BIO-402', 'countryCode' => 'AT', 'handle' => 'LACON',
+        'name' => 'LACON GmbH ', 'website' => 'https://www.lacon-institut.com/', 'apis' => ['easy-cert']
+    ],
+    'AT-BIO-501' => [
+        'id' => 'AT-BIO-501', 'countryCode' => 'AT', 'handle' => 'SLK',
+        'name' => 'SLK GesmbH', 'website' => 'https://slk.at/', 'apis' => ['bioc']
+    ],
+    'AT-BIO-901' => [
+        'id' => 'AT-BIO-901', 'countryCode' => 'AT', 'handle' => 'LVA',
+        'name' => 'LVA GmbH', 'website' => 'https://www.lva.at/', 'apis' => []
+    ],
+    'AT-BIO-902' => [
+        'id' => 'AT-BIO-902', 'countryCode' => 'AT', 'handle' => 'SGS',
+        'name' => 'SGS Austria Controll-Co. Ges.m.b.H.', 'website' => 'https://www.sgs.com/de-at ', 'apis' => ['bioc']
+    ],
+    'AT-BIO-903' => [
+        'id' => 'AT-BIO-903', 'countryCode' => 'AT', 'handle' => 'LKV',
+        'name' => 'LKV Austria Gemeinnützige GmbH', 'website' => 'https://www.lkv.at/', 'apis' => ['bioc', 'lkv']
+    ],
+];
+
+header('Access-Control-Allow-Origin: *');
+
+function jenc($data): string {
+    return json_encode($data, JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE);
+}
+
 if ($_SERVER['REQUEST_METHOD'] !== 'GET' && $_SERVER['REQUEST_METHOD'] !== 'HEAD') {
+    header('Content-Type: application/json; charset=UTF-8');
     header('Status: 405');
-    header('Content-Length: 0');
+    header('Content-Length: 56');
     header('Allow: GET, HEAD');
+    echo "{\"error\":\"method not allowed\",\"allow\":[\"GET\",\"HEAD\"]}\n";
     exit;
 }
 
 $info = $_SERVER['PATH_INFO'];
 if ($info !== '/' && str_ends_with($info, '/')) {
     header('Status: 303');
-    header('Content-Length: 0');
     header('Location: /organic/' . substr($info, 0, -1));
     exit;
 }
@@ -22,8 +64,10 @@ if (str_starts_with($info, '/certificates/')) {
         $id = substr($id, 0, -4);
     }
     if (str_contains($id, '/') || !file_exists("certificates/$id.pdf")) {
-        header('Content-Length: 0');
+        header('Content-Type: application/json; charset=UTF-8');
         header('Status: 404');
+        header('Content-Length: 22');
+        echo "{\"error\":\"not found\"}\n";
         exit;
     }
     if (str_ends_with($parts[2], '.txt')) {
@@ -52,28 +96,32 @@ if (str_starts_with($info, '/certificates/')) {
     exit;
 } else if ($info === '/authorities') {
     header('Content-Type: application/json; charset=UTF-8');
-    echo <<<EOF
+    echo "{\"data\":[\n";
-{"data":[
+    $first = true;
-  {"id":"AT-BIO-301","countryCode":"AT","handle":"ABG","name":"Austria Bio Garantie GmbH","website":"https://www.bio-garantie.at/","apis":["easy-cert"]},
+    foreach ($AUTHORITIES as $auth) {
-  {"id":"AT-BIO-302","countryCode":"AT","handle":"ABG-LW","name":"Austria Bio Garantie – Landwirtschaft GmbH","website":"https://www.bio-garantie.at/","apis":["easy-cert"]},
+        if (!$first) echo ",\n";
-  {"id":"AT-BIO-401","countryCode":"AT","handle":"BIOS","name":"BIOS – Biokontrollservice Österreich GmbH","website":"https://www.bios-kontrolle.at/","apis":["bioqs"]},
+        echo "  " . jenc($auth);
-  {"id":"AT-BIO-402","countryCode":"AT","handle":"LACON","name":"LACON GmbH ","website":"https://www.lacon-institut.com/","apis":["easy-cert"]},
+        $first = false;
-  {"id":"AT-BIO-501","countryCode":"AT","handle":"SLK","name":"SLK GesmbH","website":"https://slk.at/","apis":["bioc"]},
+    }
-  {"id":"AT-BIO-901","countryCode":"AT","handle":"LVA","name":"LVA GmbH","website":"https://www.lva.at/","apis":[]},
+    echo "\n]}\n";
-  {"id":"AT-BIO-902","countryCode":"AT","handle":"SGS","name":"SGS Austria Controll-Co. Ges.m.b.H.","website":"https://www.sgs.com/de-at ","apis":["bioc"]},
-  {"id":"AT-BIO-903","countryCode":"AT","handle":"LKV","name":"LKV Austria Gemeinnützige GmbH","website":"https://www.lkv.at/","apis":["bioc","lkv"]}
-]}
-
-EOF;
     exit;
 } else if (str_starts_with($info, '/authorities/')) {
     $code = $parts[2];
-    header('Content-Type: text/plain; charset=UTF-8');
+    header('Content-Type: application/json; charset=UTF-8');
-    echo "Control Authority Code: $code\n";
+    if (array_key_exists($code, $AUTHORITIES)) {
+        echo jenc($AUTHORITIES[$code]);
+        echo "\n";
+    } else {
+        header('Status: 404');
+        header('Content-Length: 22');
+        echo "{\"error\":\"not found\"}\n";
+    }
     exit;
 } else if ($info === '/operators') {
+    header('Content-Type: application/json; charset=UTF-8');
     header('Status: 501');
-
+    header('Content-Length: 27');
+    echo "{\"error\":\"not implemented\"}\n";
     exit;
 } else if (str_starts_with($info, '/operators/')) {
     $ooid = $parts[2];
@@ -82,5 +130,6 @@ EOF;
     exit;
 }
 
-header('Content-Length: 0');
 header('Status: 404');
+header('Content-Length: 22');
+echo "{\"error\":\"not found\"}\n";

www/organic/pdf.php

@@ -224,8 +224,21 @@ if ($format === 'text') {
                     $certNr = $m[0];
             }
         }
+        $operator = ['lfbisNr' => null];
+        if ($authorityId === 'AT-BIO-301') {
+            // TODO
+        } else if ($authorityId === 'AT-BIO-302') {
+            if (preg_match("/^lfbis[^0-9]*([0-9]{7})$/im", $text, $matches) === 1)
+                $operator['lfbisNr'] = $matches[1];
+        } else if ($authorityId === 'AT-BIO-401') {
+            // TODO
+        } else if ($authorityId === 'AT-BIO-402') {
+            // TODO
+        } else if ($authorityId === 'AT-BIO-903') {
+            // TODO
+        }
         echo "{\"type\":\"$authorityId\",\"lang\":\"de\",\"id\":" . jenc($certId) . ",\"nr\":" . jenc($certNr);
-        echo ",\n  \"operator\":{},\n  \"authority\":{\"id\":" . jenc($authorityId) . "}}\n";
+        echo ",\n  \"operator\":" . jenc($operator) . ",\n  \"authority\":{\"id\":" . jenc($authorityId) . "}}\n";
         exit;
     }