From 340e2c4c51d4b933269cff7d085c09ceb145631e Mon Sep 17 00:00:00 2001
From: Jedidiah Barber <contact@jedbarber.id.au>
Date: Thu, 4 Nov 2021 18:15:08 +1300
Subject: Fixed HTML character encoding, question/answer format field parsing

---
 src/fmd.adb | 64 ++++++++++++++++++++++++++++++++++++++++++++-----------------
 1 file changed, 46 insertions(+), 18 deletions(-)

(limited to 'src/fmd.adb')

diff --git a/src/fmd.adb b/src/fmd.adb
index ae11660..8d8fcde 100644
--- a/src/fmd.adb
+++ b/src/fmd.adb
@@ -48,8 +48,7 @@ package body FMD is
     begin
         Put_Line (File_Handle, (4 * ' ') & "<fields>");
         for FID of Field_IDs loop
-            Text := To_Unbounded_String (FID);
-            Standard_Escapes (Text);
+            Text := Prep (To_Unbounded_String (FID));
             Put_Line (File_Handle, (8 * ' ') & "<field>" & (-Text) & "</field>");
         end loop;
         Put_Line (File_Handle, (4 * ' ') & "</fields>");
@@ -73,18 +72,15 @@ package body FMD is
         Text : SU.Unbounded_String;
     begin
         Put_Line (File_Handle, (8 * ' ') & "<pack>");
-        Text := To_Unbounded_String (Q_Data.First_Element);
-        Standard_Escapes (Text);
+        Text := Prep (To_Unbounded_String (Q_Data.First_Element));
         Put_Line (File_Handle, (12 * ' ') & "<qst>" & (-Text) & "</qst>");
         for I in Vector_Index range Vector_Index'Succ (Q_Data.First_Index) .. Q_Data.Last_Index loop
-            Text := To_Unbounded_String (Q_Data.Element (I));
-            Standard_Escapes (Text);
+            Text := Prep (To_Unbounded_String (Q_Data.Element (I)));
             --  Fresh Memory unfortunately cannot cope with multiple question fields
             Put_Line (File_Handle, (12 * ' ') & "<!--" & "<qst>" & (-Text) & "</qst>" & "-->");
         end loop;
         for FID of A_Data loop
-            Text := To_Unbounded_String (FID);
-            Standard_Escapes (Text);
+            Text := Prep (To_Unbounded_String (FID));
             Put_Line (File_Handle, (12 * ' ') & "<ans>" & (-Text) & "</ans>");
         end loop;
         Put_Line (File_Handle, (8 * ' ') & "</pack>");
@@ -119,8 +115,7 @@ package body FMD is
         Put_Line (File_Handle, (8 * ' ') & "<e>");
         while Counter <= Quantity loop
             if Position <= Data.Last_Index then
-                Text := To_Unbounded_String (Data.Element (Position));
-                Standard_Escapes (Text);
+                Text := Prep (To_Unbounded_String (Data.Element (Position)));
                 Put_Line (File_Handle, (12 * ' ') & "<f>" & (-Text) & "</f>");
             else
                 Put_Line (File_Handle, (12 * ' ') & "<f></f>");
@@ -150,7 +145,7 @@ package body FMD is
 
 
 
-    procedure Escape
+    procedure Replace_All
            (Text : in out SU.Unbounded_String;
             Char : in     Character;
             Sub  : in     String)
@@ -162,16 +157,49 @@ package body FMD is
             exit when Position = 0;
             SU.Replace_Slice (Text, Position, Position, Sub);
         end loop;
-    end Escape;
+    end Replace_All;
 
 
-    procedure Standard_Escapes
-           (Text : in out SU.Unbounded_String) is
+    procedure Replace_All
+           (Text : in out SU.Unbounded_String;
+            Item : in     String;
+            Sub  : in     String)
+    is
+        Position : Natural;
+    begin
+        loop
+            Position := SU.Index (Text, Item);
+            exit when Position = 0;
+            SU.Replace_Slice (Text, Position, Position + Sub'Length - 1, Sub);
+        end loop;
+    end Replace_All;
+
+
+    function Prep
+           (Text : in SU.Unbounded_String)
+        return SU.Unbounded_String
+    is
+        Result : SU.Unbounded_String := Text;
     begin
-        Escape (Text, '<', "&lt;");
-        Escape (Text, '>', "&gt;");
-        Escape (Text, '"', "&quot;");
-    end Standard_Escapes;
+        --  Fresh Memory needs these character codes
+        Replace_All (Result, '<', "&lt;");
+        Replace_All (Result, '>', "&gt;");
+        Replace_All (Result, '"', "&quot;");
+
+        --  Not sure why these cause Fresh Memory to not load the dict
+        --  Replace_All (Result, Latin.CR & Latin.LF, "<br>");
+        --  Replace_All (Result, Latin.LF, "<br>");
+
+        --  Fresh memory doesn't recognise these character codes so get rid of them
+        --  Note that this list may not be complete
+        Replace_All (Result, "&ndash;", "–");
+        Replace_All (Result, "&rsquo;", "’");
+        Replace_All (Result, "&lsquo;", "‘");
+        Replace_All (Result, "&rdquo;", "”");
+        Replace_All (Result, "&ldquo;", "“");
+
+        return Result;
+    end Prep;
 
 
 end FMD;
-- 
cgit