Replace the "n/a" retuned values with empty string

nemethviktor · Mar 17, 2024 · 28e6151 · 28e6151
1 parent 8766c18
commit 28e6151
Show file tree

Hide file tree

Showing 7 changed files with 249 additions and 223 deletions.
diff --git a/FrmMainApp.cs b/FrmMainApp.cs
@@ -476,7 +476,7 @@ private static async Task GetHtmlAsync(string url, FrmMainApp formInstance, Canc
                     GBP_Market_capitalisation = marketCap * gbpEqivalent,
                     Sector = sector,
                     ETF_Type = etfType,
-                    Top10_Components = TagsToModelValueTransformations.T2M_Top10_Components(pageText: pageText),
+                    Top10_Exposures = TagsToModelValueTransformations.T2M_Top10_Exposures(pageText: pageText),
                     Exchange = TagsToModelValueTransformations.T2M_Exchange(companyPageText: companyPageText),
                     Country = TagsToModelValueTransformations.T2M_Country(companyPageText: companyPageText),
                     Indices = TagsToModelValueTransformations.T2M_Indices(companyPageText: companyPageText)

diff --git a/HLWebScraper.Net.csproj b/HLWebScraper.Net.csproj
@@ -9,6 +9,7 @@
 	<ForceDesignerDPIUnaware>true</ForceDesignerDPIUnaware>
 	<ApplicationIcon>Resources\AppIcon.ico</ApplicationIcon>
 	<SupportedOSPlatformVersion>10.0.17763.0</SupportedOSPlatformVersion>
+	 <RuntimeIdentifier>win-x64</RuntimeIdentifier>    
   <GenerateAssemblyInfo>false</GenerateAssemblyInfo>
   <ApplicationManifest>app.manifest</ApplicationManifest>
 
@@ -34,7 +35,6 @@
   <ItemGroup>
     <PackageReference Include="CsvHelper" Version="31.0.2" />
     <PackageReference Include="HtmlAgilityPack" Version="1.11.59" />
-    <PackageReference Include="Microsoft.Data.Sqlite" Version="9.0.0-preview.2.24128.4" />
     <PackageReference Include="Newtonsoft.Json" Version="13.0.3" />
     <PackageReference Include="System.Data.SQLite" Version="1.0.118" />
   </ItemGroup>
@@ -67,13 +67,16 @@
   </ItemGroup>
 
   <ItemGroup>
-    <None Update="Output\example_HLWebScraper_Output_20240315_192630.csv">
+    <None Update="changelog.md">
       <CopyToOutputDirectory>Always</CopyToOutputDirectory>
     </None>
-    <None Update="Properties\Settings.settings">
-      <Generator>SettingsSingleFileGenerator</Generator>
-      <LastGenOutput>Settings.Designer.cs</LastGenOutput>
+    <None Update="Output\example_HLWebScraper_Output_20240317_112225.csv">
+      <CopyToOutputDirectory>Always</CopyToOutputDirectory>
+    </None>
+    <None Update="readme.md">
+      <CopyToOutputDirectory>Always</CopyToOutputDirectory>
     </None>
+
   </ItemGroup>
 
 </Project>
diff --git a/Model/SEDOL.cs b/Model/SEDOL.cs
@@ -9,7 +9,7 @@ internal class SEDOL
     public string SEDOL_ID { get; set; }
     public string Sector { get; set; }
     public string ETF_Type { get; set; }
-    public string Top10_Components { get; set; }
+    public string Top10_Exposures { get; set; }
     public string Exchange { get; set; }
     public string Country { get; set; }
     public string Indices { get; set; }

diff --git a/Model/TagsToModelValueTransformations.cs b/Model/TagsToModelValueTransformations.cs
@@ -180,28 +180,50 @@ public static string T2M_ETF_Type(string name)
         return "Not classified";
     }
 
+
+    /// <summary>
+    ///     Reads the Exchange list (where available)
+    /// </summary>
+    /// <param name="companyPageText"></param>
+    /// <returns></returns>
     public static string T2M_Exchange(string companyPageText)
     {
-        return HelperStringUtils.ClearUTFChars(input: HelperStringUtils.FindTextBetween(
+        string likelyExchange = HelperStringUtils.ClearUTFChars(input: HelperStringUtils.FindTextBetween(
             pageText: companyPageText,
             textStart: "Exchange:<dd>",
             textEnd: "</dd>"));
+
+        return likelyExchange == "-" || likelyExchange == "n/a" ? string.Empty : likelyExchange;
     }
 
+    /// <summary>
+    ///     Reads the Country (where available)
+    /// </summary>
+    /// <param name="companyPageText"></param>
+    /// <returns></returns>
     public static string T2M_Country(string companyPageText)
     {
-        return HelperStringUtils.ClearUTFChars(input: HelperStringUtils.FindTextBetween(
+        string likelyCountry = HelperStringUtils.ClearUTFChars(input: HelperStringUtils.FindTextBetween(
             pageText: companyPageText,
             textStart: "Country:<dd>",
             textEnd: "</dd>"));
+
+        return likelyCountry == "-" || likelyCountry == "n/a" ? string.Empty : likelyCountry;
     }
 
+    /// <summary>
+    ///     Reads the Indices [really, index] (where available)
+    /// </summary>
+    /// <param name="companyPageText"></param>
+    /// <returns></returns>
     public static string T2M_Indices(string companyPageText)
     {
-        return HelperStringUtils.ClearUTFChars(input: HelperStringUtils.FindTextBetween(
+        string likelyIndex = HelperStringUtils.ClearUTFChars(input: HelperStringUtils.FindTextBetween(
             pageText: companyPageText,
             textStart: "Indices:<dd>",
             textEnd: "</dd>"));
+
+        return likelyIndex == "-" || likelyIndex == "n/a" ? string.Empty : likelyIndex;
     }
 
     /// <summary>
@@ -497,24 +519,24 @@ public static double T2M_Volume(string pageText)
     /// </summary>
     /// <param name="pageText"></param>
     /// <returns></returns>
-    public static string T2M_Top10_Components(string pageText)
+    public static string T2M_Top10_Exposures(string pageText)
     {
-        string top10components = string.Empty;
-        string likelyComponents = HelperStringUtils.FindTextBetween(
+        string top10Exposures = string.Empty;
+        string likelyExposures = HelperStringUtils.FindTextBetween(
             pageText: pageText,
             textStart: "<div id=\"top_10_exposures_data\">",
             textEnd: "</div>");
-        if (likelyComponents.Contains(value: "No top ten information is available at this stage"))
-            return top10components;
+        if (likelyExposures.Contains(value: "No top ten information is available at this stage"))
+            return top10Exposures;
 
         try
         {
-            likelyComponents = HelperStringUtils.FindTextBetween(
-                pageText: likelyComponents,
+            likelyExposures = HelperStringUtils.FindTextBetween(
+                pageText: likelyExposures,
                 textStart: "<tbody>",
                 textEnd: "</tbody>");
 
-            string[] lines = likelyComponents.Split(separator: new[] { '\r', '\n' },
+            string[] lines = likelyExposures.Split(separator: new[] { '\r', '\n' },
                 options: StringSplitOptions.RemoveEmptyEntries);
 
 
@@ -525,15 +547,15 @@ public static string T2M_Top10_Components(string pageText)
                     // Extract text from the row (remove HTML tags)
                     string rowText = RemoveHtmlTags(line: line);
 
-                    top10components += rowText;
+                    top10Exposures += rowText;
                 }
         }
         catch
         {
             // nothing
         }
 
-        return top10components;
+        return top10Exposures;
 
         string RemoveHtmlTags(string line)
         {