From 32efd3010417f3b10b04d4548e5080cf71234d44 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Stephan=20B=C3=B6sch-Plepelits?= <skunk@xover.mud.at>
Date: Thu, 28 May 2020 14:00:39 +0200
Subject: [PATCH] tag2link: convert tag2link info into dist/tag2link.json

---
 bin/download_dependencies |  1 +
 bin/tag2link-converter    | 39 +++++++++++++++++++++++++++++++++++++++
 2 files changed, 40 insertions(+)
 create mode 100755 bin/tag2link-converter

diff --git a/bin/download_dependencies b/bin/download_dependencies
index 24d42ef9..3fb150ba 100755
--- a/bin/download_dependencies
+++ b/bin/download_dependencies
@@ -2,6 +2,7 @@
 
 curl -H "Accept: application/json" -H "Content-Type: application/sparql-query" -H "User-Agent: OpenStreetBrowser" -XPOST -d @'lib/tag2link-wikidata.qry' https://query.wikidata.org/sparql > data/tag2link-wikidata.json
 curl -H "Accept: application/json" -H "Content-Type: application/sparql-query" -H "User-Agent: OpenStreetBrowser" -XPOST -d @'lib/tag2link-sophox.qry' https://sophox.org/sparql > data/tag2link-sophox.json
+bin/tag2link-converter
 
 mkdir -p data/GeoIP
 cd data/GeoIP
diff --git a/bin/tag2link-converter b/bin/tag2link-converter
new file mode 100755
index 00000000..401c56d8
--- /dev/null
+++ b/bin/tag2link-converter
@@ -0,0 +1,39 @@
+#!/usr/bin/php
+<?php
+$tag2link = array();
+
+$files = array('data/tag2link-wikidata.json', 'data/tag2link-sophox.json');
+foreach ($files as $file) {
+  $data = json_decode(file_get_contents($file), true);
+
+  foreach ($data['results']['bindings'] as $entry) {
+    $key = substr($entry['OSM_key']['value'], 4);
+    $link = $entry['formatter_URL']['value'];
+
+    if (array_key_exists($key, $tag2link)) {
+      // avoid duplicates
+      if (sizeof(array_filter($tag2link[$key]['formatter'], function ($e) use ($link) {
+        return $e['link'] === $link;
+      }))) {
+        continue;
+      }
+    }
+    else {
+      $tag2link[$key] = array(
+        'formatter' => array(),
+      );
+    }
+    
+    $formatter = array(
+      'link' => $link,
+    );
+
+    if (preg_match("/^https?:\/\/([^\/]*)(\/.*|)$/", $link, $m)) {
+      $formatter['operator'] = $m[1];
+    }
+
+    $tag2link[$key]['formatter'][] = $formatter;
+  }
+}
+
+file_put_contents('dist/tag2link.json', json_encode($tag2link, JSON_PRETTY_PRINT|JSON_UNESCAPED_SLASHES|JSON_UNESCAPED_UNICODE));