parsing als parse-rki (Robert Koch Institut)

parse-rki.php

@@ -0,0 +1,47 @@
+#! /usr/bin/php
+<?php
+include "config.php";
+
+$sql = "SELECT * FROM scans where site=1";
+$stmt = $pdo->query($sql);
+$scans = $stmt->fetchAll(PDO::FETCH_ASSOC);
+foreach ($scans as $scan){
+    $html = $scan['content'];
+    $parts = explode('<h1 class="firstInSlot"><abbr title="Schweres Akutes Respiratorisches Syndrom Coronavirus 2">SARS-CoV-2</abbr>: Fallzahlen in Deutschland, China und weltweit</h1>', $html);
+    $page = new DOMDocument();
+    $page->loadHTML($html);
+    $document = array();
+    for ($i = 0; $i < $page->getElementsByTagName("table")->length; $i++){
+        $table = $page->getElementsByTagName("table")[$i];
+        for ($j = 0; $j < $table->childNodes->length; $j++){
+            $sub = $table->childNodes[$j];
+            for ($k = 0; $k < $sub->childNodes->length; $k++){
+                $zeile = array();
+                $tr = $sub->childNodes[$k];
+                for ($l = 0; $l < $tr->childNodes->length; $l++){
+                    if($tr->childNodes[$l]->textContent != ''){
+                        if ($i == 0 and $l == 0){
+                            $zeile[] = 'D-'.str_replace(".", "", $tr->childNodes[$l]->textContent);
+                        }else{
+                            $zeile[] = str_replace(".", "", $tr->childNodes[$l]->textContent);
+                        }
+                    }
+                }
+                $document[] = $zeile;
+            }
+        }
+    }
+
+    $scanid = $scan['id'];
+
+    foreach ($document as $entry){
+        $sql = "INSERT INTO `data`(`scan`, `place`, `amount`) VALUES (?, ?, ?)";
+        $stmt = $pdo->prepare($sql);
+        $stmt->bindParam(1, $scanid);
+        $stmt->bindParam(2, $entry[0]);
+        $stmt->bindParam(3, $entry[1]);
+        $stmt->execute();
+        if($stmt->errorInfo()[0] != '00000' and $stmt->errorInfo()[0] != '23000' and $stmt->errorInfo()[0] != '22007')
+        var_dump($stmt->errorInfo());
+    }
+}