commit 920f6f44ff8c64a8456f6ee47f73ef782f311a40
Author: mxmehl <mail@mehl.mx>
Date:   Fri May 5 12:39:03 2017 +0200

    initial commit

diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..2d20446
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1,2 @@
+config.php
+angebote.json*
diff --git a/config.php.sample b/config.php.sample
new file mode 100644
index 0000000..41ddbff
--- /dev/null
+++ b/config.php.sample
@@ -0,0 +1,15 @@
+;<?php
+;die(); // For further security
+;/*
+
+[general]
+# "url" has to contain page indicator (/P-x/). This is not existent on the first search result page but can be added manually
+url = "https://www.immobilienscout24.de/Suche/S-2/P-1/Wohnung-Miete/Nordrhein-Westfalen/Koeln/70_6_118_39_20_101_2_50_11_56_120/2,00-3,00/-80,00/EURO--900,00/-/-/false/-/-/true/-/-/true"
+db = "angebote.json"
+
+[mail]
+to = "to@example.com"
+from = "from@example.com"
+
+;*/
+;?>
diff --git a/index.php b/index.php
new file mode 100644
index 0000000..d2dfdb0
--- /dev/null
+++ b/index.php
@@ -0,0 +1,386 @@
+<?php
+
+$config = parse_ini_file("config.php", true);
+$db = $config['general']['db'];
+$queryurl = $config['general']['url'];
+
+// make backup of database
+$db_bak = $db . ".bak_" . date('Y-m-d');
+if (! file_exists($db_bak)) {
+  copy($db, $db_bak);
+}
+
+// Called via CLI or web?
+if (defined('STDIN')) {
+  $do = $argv[1];
+} else {
+  if ($_SERVER['REQUEST_METHOD'] === 'POST') {
+    $do = isset($_POST['do']) ? $_POST['do'] : false;
+  } else {
+    $do = isset($_GET['do']) ? $_GET['do'] : false;
+  }
+}
+
+// define action via "do" parameter
+if ($do === "download") {
+  immo_download($db);
+} else if ($do === "update") {
+  immo_update($db);
+} else if ($do === "show" || $do == "") {
+  immo_show($db);
+} else {
+  echo "No valid do action defined";
+}
+
+/////////////////////
+/// do = DOWNLOAD ///
+/////////////////////
+function immo_download($db) {
+  global $data;
+  global $queryurl;
+  global $config;
+  read_db($db);
+  
+  $site = xsite($queryurl); // load and transform URL (page 1) to queryable $site
+  
+  // get amount of available pages
+  $searchPages = $site->query('//div[@id="pageSelection"]/select');
+  $pages = $searchPages->item(0)->childNodes->length;
+  // create a dynamic url in which the current page number can be set in
+  $queryurldyn = str_replace("/P-1/", "/P-%page%/", $queryurl);
+  
+  // loop through available pages
+  for ($page = 1; $page <= $pages; $page++) {
+    $queryurlcur = str_replace("%page%", $page, $queryurldyn);
+    echo "Current search page: " . $queryurlcur . "<br />\n";
+    
+    $site = xsite($queryurlcur);  // load and transform URL of current page
+    // get all links to expose pages
+    $searchResult = $site->query('//a[@class="result-list-entry__brand-title-container"]/@href');
+
+    // loop through web search results
+    foreach($searchResult as $result){
+      global $data;
+      $new = "y";  // is the search entry new?
+      // extract ID of link
+      preg_match("/\d+$/",$result->textContent, $matches);
+      $id = $matches[0];
+      $id = (int)$id; // convert to int value
+      
+      $total = count($data);
+      for ($row = 0; $row < $total; $row++) {
+        if ($id === $data[$row]['id']) {
+          $new = "n";   // entry isn't new anymore
+        }
+      }
+      
+      if ($new === "y") { // entry is new
+        
+        // load and transform expose URL
+        $site = xsite("https://www.immobilienscout24.de/expose/" . $id);
+        
+        // description
+        $searchResult = $site->query('//h1[@id="expose-title"]');
+        $desc = trim($searchResult->item(0)->nodeValue);
+        $desc = mb_convert_encoding($desc, 'UTF-8', 'UTF-8');   // remove/replace invalid characters
+        $descs = substr($desc, 0, 15);
+        $descs = mb_convert_encoding($descs, 'UTF-8', 'UTF-8');
+        
+        // flat available from
+        $searchResult = $site->query('//dd[@class="is24qa-bezugsfrei-ab grid-item three-fifths"]');
+        $bezug = trim($searchResult->item(0)->nodeValue);
+        
+        // warm rent
+        $searchResult = $site->query('//dd[@class="is24qa-gesamtmiete grid-item three-fifths font-bold"]');
+        $mietew = trim($searchResult->item(0)->nodeValue);
+        
+        // cold rent
+        $searchResult = $site->query('//dd[@class="is24qa-kaltmiete grid-item three-fifths"]');
+        $mietek = trim($searchResult->item(0)->nodeValue);
+        
+        // rooms
+        $searchResult = $site->query('//dd[@class="is24qa-zimmer grid-item three-fifths"]');
+        $zimmer = trim($searchResult->item(0)->nodeValue);
+        
+        // size
+        $searchResult = $site->query('//dd[@class="is24qa-wohnflaeche-ca grid-item three-fifths"]');
+        $qm = trim($searchResult->item(0)->nodeValue);
+        
+        // location
+        $searchResult = $site->query('//div[@class="address-block"]');
+        $ort = trim($searchResult->item(0)->nodeValue);
+        $ort = str_replace("(zur Karte)  ", "", $ort);
+        $ort = str_replace("Die vollständige Adresse der Immobilie erhalten Sie vom Anbieter.", "", $ort);
+        
+        // append new array entry
+        $data[] = array("id" => $id,
+                        "date" => date('d.m. H:i'),
+                        "desc" => $desc,
+                        "descs" => $descs,
+                        "bezug" => $bezug,
+                        "mietew" => $mietew,
+                        "mietek" => $mietek,
+                        "ort" => $ort,
+                        "zimmer" => $zimmer,
+                        "qm" => $qm,
+                        "rating" => 0,
+                        "status" => "NEU",
+                        "note" => "");
+        
+        // send mail for new item
+        $mailcontent =  "*" . $desc . "* \r\n\r\n" .
+                        "Bezug: " . $bezug . "\r\n" . 
+                        "Miete: " . $mietew . "/" . $mietek . "\r\n" . 
+                        "Größe: " . $qm . "\r\n" .
+                        "Zimmer: " . $zimmer . "\r\n" .
+                        "Ort: " . $ort . "\r\n" . 
+                        "Link: https://www.immobilienscout24.de/expose/" . $id . "\r\n\r\n" . 
+                        "Übersicht: https://" . $_SERVER["SERVER_NAME"] . $_SERVER["PHP_SELF"];
+        mail($config['mail']['to'], "Neues Wohnungsinserat: " . $descs . "...", $mailcontent, "From: " . $config['mail']['from'] . "\r\nMIME-Version: 1.0\r\nContent-Type: text/plain; charset=UTF-8");
+        
+        echo $id . " is new and has been downloaded.<br />\n";
+        
+      } else {  // entry is old
+        echo $id . " already exists.<br />\n";
+      }
+      
+    } // END foreach &searchResult
+  } // END for loop through pages
+
+  
+  echo "<p><a href='.'>Back to overview</a></p>\n";
+  
+  write_db($db, $data);
+}
+
+/////////////////
+/// do = SHOW ///
+/////////////////
+function immo_show($db) {
+  global $data;
+  global $queryurl;
+  read_db($db);
+  
+  $hide = isset($_GET['hide']) ? $_GET['hide'] : "yes";
+  ?>
+
+<!DOCTYPE html>
+<html lang="de-DE">
+<head>
+  <title>Immobilienscout Search Helper</title>
+  <meta charset="UTF-8" />
+  <style type="text/css">
+  table {
+    border-collapse: collapse;
+  }
+  table, th, td {
+     border: 1px solid orange;
+  }
+  th, td {
+    padding: 3px 2px;
+  }
+  tr:nth-child(even) {
+    background-color: #f2f2f2
+  }
+  tr:hover {background-color: #FFECBA}
+  textarea, input, select {
+    vertical-align: middle;
+    height: 3em;
+  }
+  input[type="submit"] {
+    border: 1px solid #888;
+    height: 20px;
+    padding: 0;
+    width: 2em;
+  }
+  a.btn {
+    background-color: #98b879;
+    color: #fff;
+    font-weight: 700;
+    padding: 10px;
+    text-decoration: none;
+    text-transform: uppercase;
+  }
+  #message {
+    background-color: #FFE990;
+    margin-bottom: 8px;
+    padding-left: 20px;
+  }
+  </style>
+</head>
+<body>
+  <div id="message">
+    <?php 
+    session_start(); 
+    if ( ! empty($_SESSION['message'])) {
+      echo $_SESSION['message'];
+      $_SESSION['message'] = NULL;
+    }
+    ?>
+  </div>
+  <table>
+    <tr>
+      <th>Datum</th>
+      <th>Beschreibung</th>
+      <th>Status</th>
+      <th>Bezug</th>
+      <th>Miete warm (kalt)</th>
+      <th>Zi</th>
+      <th>Größe</th>
+      <th>Ort</th>
+      <th style="min-width:140px">Notiz</th>
+      <th style="min-width:140px">Status ändern</th>
+    </tr>
+    <?php
+    
+    // Sort for date descending when displaying
+    foreach ($data as $key => $row) {
+      $date[$key]  = $row['date'];
+      $id[$key]  = $row['id'];
+    }
+    array_multisort($date, SORT_DESC, $id, SORT_DESC, $data);
+    
+    $total = count($data);
+    for ($row = 0; $row < $total; $row++) {
+      $id = $data[$row]['id'];
+      $date = $data[$row]['date'];
+      $desc = $data[$row]['desc'];
+      $descs = $data[$row]['descs'];
+      $bezug = $data[$row]['bezug'];
+      $mietew = $data[$row]['mietew'];
+      $mietek = $data[$row]['mietek'];
+      $zimmer = $data[$row]['zimmer'];
+      $qm = $data[$row]['qm'];
+      $ort = $data[$row]['ort'];
+      $status = $data[$row]['status'];
+      $note = $data[$row]['note'];
+      $link = "https://www.immobilienscout24.de/expose/" . $id;
+      
+      if ($status !== "del" || $hide === "no") {
+        echo "<tr>";
+          echo "<td>" . $date . "</td>";
+          echo "<td><span title='" . $desc . "'><em><a target='_blank' href='" . $link . "'>" . $descs . "...</a></em></span></td>";
+          echo "<td>" . $status . "</td>";
+          echo "<td>" . $bezug . "</td>";
+          echo "<td>" . $mietew . " (<em>" . $mietek . "</em>)" . "</td>";
+          echo "<td>" . $zimmer . "</td>";
+          echo "<td>" . $qm . "</td>";
+          echo "<td><a target='_blank' href='https://www.openstreetmap.org/search?query=" . $ort . "'>" . $ort . "</a></td>";
+          
+          echo "<td>";
+          ?>
+          <form action="" method="POST">
+            <input name="do" value="update" type="hidden" />
+            <input name="id" value="<?php echo $id; ?>" type="hidden" />
+            <textarea name="note" rows="1" cols="14"><?php echo $note; ?></textarea>
+            <input type="submit" value="OK">
+          </form>
+          <?php
+          echo "</td>";
+          
+          echo "<td>";
+          ?>
+          <form action="" method="POST">
+            <input name="do" value="update" type="hidden" />
+            <input name="id" value="<?php echo $id; ?>" type="hidden" />
+            <select name="status">
+              <option value="del">löschen</option>
+              <option value="alt">alt</option>
+              <option value="NEU">NEU</option>
+              <option value="kontaktiert">kontaktiert</option>
+              <option value="abgelehnt">abgelehnt</option>
+              <option value="termin">termin</option>
+            </select>
+            <input type="submit" value="OK">
+          </form>
+          <?php
+          echo "</td>";
+          
+        echo "</tr>\n";
+      }
+    }
+    
+    ?>
+  </table>
+  
+  <p>Letztes Update: <?php echo date('d.m.Y H:i:s', filemtime($db)); ?></p>
+  <p>Gelöschte Inserate <a href="?hide=no">einblenden</a> / <a href=".">ausblenden</a></p>
+  <p>Suchlink: <a target="_blank" href="<?php echo $queryurl; ?>">Klick</a></p>
+  <p><a class="btn" href="?do=download">Update</a></p>
+</body>
+</html>
+
+  <?php
+}
+
+///////////////////
+/// do = UPDATE ///
+///////////////////
+function immo_update($db) {
+  global $data;
+  read_db($db);
+  
+  $id = isset($_POST['id']) ? $_POST['id'] : false;
+  $status = isset($_POST['status']) ? $_POST['status'] : "NONE";
+  $note = isset($_POST['note']) ? $_POST['note'] : "NONE";
+
+  $key = array_search($id, array_column($data, 'id'));
+  
+  if (strlen($key) === 0) {
+    echo $id . " does not exist.<br />";
+    echo "<br />";
+    echo "<a href='.'>Back to overview</a>\n";
+  } else {
+    session_start();  // start Sessions
+    if ($status != "NONE") {
+      $data[$key]['status'] = $status;
+      save_return_page("Changed status of " . $id . " to \"". $status . "\".");
+    }
+    if ($note != "NONE") {
+      $data[$key]['note'] = $note;
+      save_return_page("Changed note of " . $id . " to \"". $note . "\".");
+    }
+  }
+}
+
+////////////////////////
+/// HELPER FUNCTIONS ///
+////////////////////////
+// load database from file and decode it to array $data
+function read_db($db) {
+  global $data;   // declare $data a global variable to access it outside this function
+  if (! file_exists($db)) {
+    touch($db);
+  }
+  $file = file_get_contents($db, true);
+  $data = json_decode($file, true);
+  unset($file);
+}
+
+// Encode $data array and write database to file
+function write_db($db, $data) {
+  // encode and write file
+  $encoded = json_encode($data, JSON_PRETTY_PRINT);
+  file_put_contents($db, $encoded, LOCK_EX);
+}
+
+// load URL and transform it to a XPath-searchable document
+function xsite($url) {
+  libxml_use_internal_errors(true); // suppress errors
+  $dl = file_get_contents($url);
+  $site = new DOMDocument();
+  $site->loadHTML($dl);
+  $xpathvar = new Domxpath($site);
+  return $xpathvar;
+}
+
+// save database, put message to session, and load refering page which shows the message
+function save_return_page($output) {
+  global $db;
+  global $data;
+  write_db($db, $data);
+  $_SESSION['message'] = "Update: " . $output;
+  header('Location: ' . $_SERVER['HTTP_REFERER']);
+  exit;
+}
+?>