Implemented the TheaterScheduleHTMLParser.

bofirial · Jun 11, 2017 · 2e533a6 · 2e533a6
1 parent 7d9442b
commit 2e533a6
Showing 1 changed file with 67 additions and 2 deletions.
diff --git a/Melody49Notifier/DataAbstraction/TheaterScheduleHTMLParser.cs b/Melody49Notifier/DataAbstraction/TheaterScheduleHTMLParser.cs
@@ -25,11 +25,76 @@ public TheaterSchedule ParseTheaterScheduleHTML(string html)
 
             TheaterSchedule theaterSchedule = new TheaterSchedule();
 
-            //theaterSchedule.ScheduleDescription = htmlDocument.DocumentNode.Descendants("div").Where(x => x?.Attributes?["id"]?.Value == "playweek").First().ChildNodes.First().InnerText;
+            theaterSchedule.ScheduleDescription = GetScheduleDescription(htmlDocument);
 
-            theaterSchedule.ScheduleDescription = htmlDocument.DocumentNode.SelectNodes("//div[@id=\"playweek\"]/p").First().InnerText;
+            HtmlNodeCollection tableRows = GetTheaterShowingsTableRows(htmlDocument);
+
+            string currentScreen = string.Empty;
+            TheaterShowing currentTheaterShowing = null;
+            theaterSchedule.Showings = new List<TheaterShowing>();
+
+            foreach (HtmlNode tableRow in tableRows)
+            {
+                string column1Text = GetColumnText(tableRow, 0);
+                string column2Text = GetColumnText(tableRow, 2);
+
+                if (IsScreenRow(column1Text, column2Text))
+                {
+                    currentScreen = column1Text.Replace(":", "").Trim();
+                }
+                else if (!IsEmptyString(column1Text) && !IsEmptyString(column2Text))
+                {
+                    if (IsShowingScheduleRow(column2Text))
+                    {
+                        currentTheaterShowing.ShowingScheduleDescription = $"{currentTheaterShowing.ShowingScheduleDescription}{column1Text} {column2Text}\r\n";
+                    }
+                    else
+                    {
+                        currentTheaterShowing = new TheaterShowing()
+                        {
+                            Screen = currentScreen,
+                            MovieDescription = column1Text,
+                            ActorDescription = column2Text
+                        };
+
+                        theaterSchedule.Showings.Add(currentTheaterShowing);
+                    }
+                }
+            }
 
             return theaterSchedule;
         }
+
+        private static HtmlNodeCollection GetTheaterShowingsTableRows(HtmlDocument htmlDocument)
+        {
+            return htmlDocument.DocumentNode.SelectNodes("(//body/div/table)[2]//tr");
+        }
+
+        private static string GetScheduleDescription(HtmlDocument htmlDocument)
+        {
+            return htmlDocument.DocumentNode.SelectNodes("//div[@id=\"playweek\"]/p").First().InnerText;
+        }
+
+        private static bool IsShowingScheduleRow(string column2Text)
+        {
+            return TimeSpan.TryParse(column2Text.Trim(), out TimeSpan time);
+        }
+
+        private static bool IsScreenRow(string column1Text, string column2Text)
+        {
+            return column1Text.ToLower().Contains("screen") && IsEmptyString(column2Text);
+        }
+
+        private static string GetColumnText(HtmlNode tableRow, int index)
+        {
+            return tableRow.ChildNodes.Where(x => x.Name == "td").ElementAt(index).InnerText;
+        }
+
+        private static bool IsEmptyString(string columnText)
+        {
+            string trimmedLoweredColumnText = columnText?.Trim()?.ToLower();
+
+            return string.IsNullOrEmpty(trimmedLoweredColumnText) || trimmedLoweredColumnText == "&nbsp;";
+        }
     }
 }