initial commit

.gitignore

@@ -0,0 +1 @@
+.*.sw*

README.md

@@ -0,0 +1 @@
+TODO: block using https://drupal.stackexchange.com/a/245759

related_ror.info.yml

@@ -0,0 +1,9 @@
+name: Podcast per ROR
+type: module
+description: 'Podcast specifici per radio ondarossa'
+core: 8.x
+version: 0.1
+package: Ondarossa
+dependencies:
+        - field
+        - link

related_ror.routing.yml

@@ -0,0 +1,27 @@
+related_ror.time:
+    path: '/related/time.json'
+    defaults:
+        _controller: '\Drupal\related_ror\Controller\RelatedRorController::relatedTime'
+        _title: 'Get near-in-time articles'
+    requirements:
+        _access: 'TRUE'
+    options:
+        no_cache: 'TRUE'
+related_ror.trx:
+    path: '/related/trx.json'
+    defaults:
+        _controller: '\Drupal\related_ror\Controller\RelatedRorController::relatedTrx'
+        _title: 'Post from the same show'
+    requirements:
+        _access: 'TRUE'
+    options:
+        no_cache: 'TRUE'
+related_ror.topic:
+    path: '/related/topic.json'
+    defaults:
+        _controller: '\Drupal\related_ror\Controller\RelatedRorController::relatedTopic'
+        _title: 'Posts with similar topic'
+    requirements:
+        _access: 'TRUE'
+    options:
+        no_cache: 'TRUE'

src/Controller/RelatedRorController.php

@@ -0,0 +1,176 @@
+<?php
+namespace Drupal\related_ror\Controller;
+
+use Drupal\Core\Controller\ControllerBase;
+use Symfony\Component\HttpFoundation\Response;
+use Symfony\Component\HttpFoundation\JsonResponse;
+use Drupal\Core\Cache\CacheableJsonResponse;
+use Drupal\Core\Cache\CacheableMetadata;
+use Symfony\Component\DependencyInjection\ContainerInterface;
+use Drupal\Core\Template\TwigEnvironment;
+use Drupal\node\NodeInterface;
+
+class RelatedRorController extends ControllerBase {
+    protected $state;
+    protected $twig;
+    public function __construct($state, TwigEnvironment $twig) {
+        $this->state = $state;
+        $this->twig = $twig;
+    }
+
+    public static function create(ContainerInterface $container) {
+        return new static(
+            $container->get('state'),
+            $container->get('twig')
+        );
+    }
+    private function nodeToLinkdata($node_entity) {
+        $arr = $node_entity->toArray();
+        return array(
+            'nid'   => intval($arr['nid'] [0] ['value']),
+            'title' => $arr['title'][0]['value'],
+            //'body'  => $arr['body'][0] ['value'],
+            'summary' => htmlspecialchars(substr(
+                html_entity_decode(strip_tags($arr['body'][0]['value'])), 0, 3500), ENT_XML1, 'UTF-8'),
+            'url'   => $arr['path'][0] ['alias']
+        );
+    }
+
+    private function queryNearInTime(int $time, int $days = 7, int $limit = 0): array {
+        $arg = array('ror_news', 'redazionali', 'news_trasmissioni');
+        $query = \Drupal::entityQuery('node');
+        $query = \Drupal::database()->select('node', 'n');
+        $query->addJoin('INNER', 'node__field_tx_date', 'dt', 'n.nid=dt.entity_id');
+        $query->addJoin('INNER', 'node_field_data', 'field', 'n.nid=field.nid');
+        $query ->addField('n', 'nid');
+        $query->addExpression("CAST(UNIX_TIMESTAMP(field_tx_date_value) as UNSIGNED)", 'ts');
+        $query->addExpression("ABS(cast(UNIX_TIMESTAMP(field_tx_date_value) as signed) - $time)", 'dist');
+        $query->condition('field.status', '1');
+        $query->condition('n.type', $arg, 'IN');
+        $query->where('CAST(UNIX_TIMESTAMP(dt.field_tx_date_value) as UNSIGNED) > :from', array('from' => $time - 3600*24*$days));
+        $query->where('CAST(UNIX_TIMESTAMP(dt.field_tx_date_value) as UNSIGNED) < :to', array('to' => $time + 3600*24*$days));
+        $query->orderBy('dist', 'ASC');
+        if($limit > 0) {
+            $query->range(0, $limit);
+        }
+        if($query->preExecute() !== TRUE) {
+            return null;
+        }
+        $nids = $query->execute()->fetchCol(0);
+        return $nids;
+    }
+
+    private function presentNids(array $nids) {
+        $nodes_e = \Drupal\node\Entity\Node::loadMultiple($nids);
+        $data = ['nodes' => []];
+        foreach($nodes_e as $nid => $node) {
+            $nodedata = $this->nodeToLinkdata($node);
+            array_push($data['nodes'], $nodedata);
+        }
+
+        return $data;
+    }
+
+    public function relatedTime() {
+        $nid = \Drupal::request()->query->get('nid');
+        if($nid == null) {
+            return new Response("Must supply a NID", 400, array('Content-Type' => 'text/plain'));
+        }
+        if(!is_numeric($nid)) {
+            return new Response("NID must be integer, not `$nid`", 400, array('Content-Type' => 'text/plain'));
+        }
+        $nid = intval($nid);
+        $node = \Drupal\node\Entity\Node::load($nid);
+        if($node == null) {
+            return new Response("Node not found", 400, array('Content-Type' => 'text/plain'));
+        }
+        $time = intval($node->getCreatedTime());
+        $content .= "\ntime={$time}";
+        $nearnodes = $this->presentNids(array_filter(
+            $this->queryNearInTime($time, $days=4, $limit=20),
+            function ($n) use ($nid): bool { return intval($n) != $nid; }));
+        $resp = $this->cachedJsonResp(array('time' => $nearnodes));
+        return $resp;
+    }
+
+    public function relatedTrx() {
+        // TODO: get node time
+        // TODO: queryNearInTime($time)
+        $nid = \Drupal::request()->query->get('nid');
+        if($nid == null) {
+            return new Response("Must supply a NID", 400, array('Content-Type' => 'text/plain'));
+        }
+        $content = "asd $nid";
+        $resp = new Response($content, 500, array( 'Content-Type' => 'text/plain'));
+        return $resp;
+    }
+
+    private function getTermWeight(int $termid): int {
+        $term = \Drupal\node\Entity\Term::load($termid);
+        $query = \Drupal::database()->select('taxonomy_index', 'ti');
+        $query->fields('ti', ['nid']);
+        $query->condition('ti.tid', $term_id);
+        $cnt = $query->execute()->rowCount();
+        if($cnt < 100) {
+            return 2;
+        }
+        return 1;
+    }
+
+    private function getSimilarity(array $orig, array $other): int {
+        $orig_tags = array_map(function($t) { return $t['target_id']; }, $orig['field_tags']);
+        $other_tags = array_map(function($t) { return $t['target_id']; }, $other['field_tags']);
+        $both = array_intersect($orig_tags, $other_tags);
+        $weighted = array_map($this->getTermWeight, $both);
+        return array_sum($weighted);
+    }
+
+    private function cachedJsonResp(array $data) {
+        $data['#cache'] = [
+            'max-age' => 600, 
+            'contexts' => [
+                'url',
+            ]];
+        $resp = new CacheableJsonResponse($data);
+        $resp->addCacheableDependency(CacheableMetadata::createFromRenderArray($data));
+        return $resp;
+    }
+
+    public function relatedTopic() {
+        // TODO: get node time
+        // TODO: queryNearInTime($time)
+        $nid = \Drupal::request()->query->get('nid');
+        if($nid == null) {
+            return new Response("Must supply a NID", 400, array('Content-Type' => 'text/plain'));
+        }
+        if(!is_numeric($nid)) {
+            return new Response("NID must be integer, not `$nid`", 400, array('Content-Type' => 'text/plain'));
+        }
+        $nid = intval($nid);
+        $node = \Drupal\node\Entity\Node::load($nid);
+        if($node == null) {
+            return new Response("Node not found", 400, array('Content-Type' => 'text/plain'));
+        }
+        $orig_arr = $node->toArray();
+        $time = intval($node->getCreatedTime());
+
+        $goodnids = [];
+        $scores = [];
+        $other_e = \Drupal\node\Entity\Node::loadMultiple($this->queryNearInTime($time, $days=30, $limit=100));
+        foreach($other_e as $other_nid => $other_node) {
+            if(intval($other_nid) === $nid) {
+                continue;
+            }
+            $score = $this->getSimilarity($orig_arr, $other_node->toArray());
+            if($score > 0) {
+                $scores[$other_nid] =  $score;
+            }
+        }
+        arsort($scores);
+        foreach($scores as $nid => $score) {
+            array_push($goodnids, $nid);
+        }
+        $resp = $this->cachedJsonResp(['topic' => $this->presentNids($goodnids)]);
+        return $resp;
+    }
+}