Changeset 69 for trunk/Modules/Meet/Import

Timestamp:

Nov 15, 2023, 12:40:49 AM (13 months ago)

Author:

chronos

Message:

• Modified: Updated parsing of web tanecniskola.cz.

File:

• trunk/Modules/Meet/Import/TanecniSkola.php (modified) (6 diffs)

trunk/Modules/Meet/Import/TanecniSkola.php

@@ -28 +28 @@
     $Content = file_get_contents($this->URL);
 
-    $BlockStart = '<div><!--[if IE]><input type=IEbug disabled style="display:none"><![endif]--></div>';
-    $BlockEnd = '</article>';
+    $BlockStart = 'data-content="inzeraty">';
+    $BlockEnd = '</section>';
     $Content = GetTextBetween($Content, $BlockStart, $BlockEnd);
     if ($Content == '')
@@ -37 +37 @@
     }
 
-    $ItemStart = '<div>';
-    $ItemEnd = '</div>';
+    $ItemStart = '<div class="item item-dating">';
+    $ItemEnd = 'Odpovědět na inzerát</a></p>';
     $PreviousTime = null;
     while (strpos($Content, $ItemStart) !== false)
@@ -46 +46 @@
       {
         $MeetItem = new MeetItem();
-        $MeetItem->Time = HumanDateToTime(trim(GetTextBetween($Item, '<h3>', '</h3>')));
+        $Time = trim(GetTextBetween($Item, '<p class="date">', '</p>'));
+        if ($Time == 'Včera') $Time = HumanDate(time() - 60*60*24);
+        else if ($Time == 'Dnes') $Time = HumanDate(time());
+        else {
+          $Time = str_replace('ledna', '1.', $Time);
+          $Time = str_replace('února', '2.', $Time);
+          $Time = str_replace('března', '3.', $Time);
+          $Time = str_replace('dubna', '4.', $Time);
+          $Time = str_replace('května', '5.', $Time);
+          $Time = str_replace('června', '6.', $Time);
+          $Time = str_replace('července', '7.', $Time);
+          $Time = str_replace('srpna', '8.', $Time);
+          $Time = str_replace('září', '9.', $Time);
+          $Time = str_replace('října', '10.', $Time);
+          $Time = str_replace('listopadu', '11.', $Time);
+          $Time = str_replace('prosince', '1.', $Time);
+        }
+        $MeetItem->Time = HumanDateToTime($Time);
         if ($MeetItem->Time != null)
         {
@@ -52 +69 @@
         } else $MeetItem->Time = $PreviousTime;
 
-        $Part = GetTextBetween($Item, '</strong>', '<strong>');
-        $Part = explode(',', $Part);
-        if (count($Part) > 0) $MeetItem->Name = trim($Part[0]);
-        if (count($Part) > 1) $MeetItem->Height = trim($Part[1]);
-        if (strpos($MeetItem->Height, '(') !== false)
+        $Part = GetTextBetween($Item, '<h2>', '</h2>');
+        if (strpos($Part, '(') !== false)
         {
-          $MeetItem->Age = substr($MeetItem->Height, strpos($MeetItem->Height, '(') + 1);
+          $MeetItem->Name = substr($Part, 0, strpos($Part, '('));
+          $MeetItem->Age = substr($Part, strpos($Part, '(') + 1);
           $MeetItem->Age = substr($MeetItem->Age, 0, strpos($MeetItem->Age, ')'));
           $MeetItem->Age = trim(str_replace('let', '', $MeetItem->Age));
-          $MeetItem->Height = trim(substr($MeetItem->Height, 0, strpos($MeetItem->Height, '(')));
-        } else $MeetItem->Age = '';
+        } else 
+        {
+          $MeetItem->Name = $Part;
+          $MeetItem->Age = '';
+        }
+
+        $Gender = trim(GetTextBetween($Item, '</em>', '</p></div>'));
+        if ($Gender == 'Hledám partnera') $MeetItem->Gender = Gender::Female;
+          else if ($Gender == 'Hledám partnerku') $MeetItem->Gender = Gender::Male;
+          else $MeetItem->Gender = Gender::Undefined;
+
+        $MeetItem->Height = trim(GetTextBetween($Item, '<div class="item py-5 px-10"><p><em class="fa fa-ruler"></em>', '</p></div>'));
         while (strpos($MeetItem->Height, '  ') !== false)
         {
@@ -69 +94 @@
         if (array_key_exists($MeetItem->Height, $HeightRange))
           $MeetItem->Height = $HeightRange[$MeetItem->Height];
-        $MeetItem->Email = trim(GetTextBetween($Item, '">', '</a>'));
-        $MeetItem->Phone = trim(GetTextBetween($Item, 'tel.:', '<br />'));
-        $Gender = trim(GetTextBetween($Item, '</strong>', '<strong>'));
-        if ($Gender == 'partnera,') $MeetItem->Gender = Gender::Female;
-          else if ($Gender == 'partnerku,') $MeetItem->Gender = Gender::Male;
-          else $MeetItem->Gender = Gender::Undefined;
-        $MeetItem->Message = trim(GetTextBetween($Item, '<p class="message">', '</p>'));
+
+        $MeetItem->Message = trim(GetTextBetween($Item, '<p>', '</p>'));
+
+        $MeetItem->Email = GetEmailFromText($MeetItem->Message);
         $MeetItem->Weight = GetWeightFromText($MeetItem->Message);
+        
         // Update age only if it is not set already
         if ($MeetItem->Age == '')
@@ -83 +106 @@
           if ($MessageAge != '') $MeetItem->Age = $MessageAge;
         }
+        
         // Update height from message text even if height is already set
         // because it can be more precise
         $MessageHeight = GetHeightFromText($MeetItem->Message);
         if ($MessageHeight != '') $MeetItem->Height = $MessageHeight;
+
         $MeetItem->Location = 'Praha';
         $MeetItem->Database = $this->Database;
         $MeetItem->Source = $this->Id;
+        //print_r($MeetItem);
         $this->MeetItems[] = $MeetItem;
       }