DataLoader changes during development

PROMS/DataLoader/TextConvert.cs

@@ -88,6 +88,8 @@ namespace DataLoader
 			dicChar[8734] = 8857;
 			dicChar[7] = 9679;
 			dicChar[8976] = 9830;
+			dicChar[9632] = 9604;
+			//dicChar[236] = 38914;
 			char[] creg = new char[dicChar.Count];
 			int i = 0;
 			foreach (int ic in dicChar.Keys)
@@ -111,9 +113,86 @@ namespace DataLoader
 			}
 			return new string(cs);
 		}
+		public static string ConvertText(string s1, bool DoCaret)
+		{
+			string s2 = s1;
+			if (DoCaret) s2 = s2.Replace("^", @"\u916");
+			return ConvertText(s2);
+		}
+		public static string ReplaceUnicode(string s2)
+		{
+			char[] tmp;
+			tmp = s2.ToCharArray();
+			s2 = s2.Replace("`", "\'b0");			// convert backquote to degree - left over from DOS days.
+			s2 = s2.Replace("\'a0",@"\'a0");		// hardspace
+			s2 = s2.Replace("\xb0", @"\'b0");		// degree
+			s2 = s2.Replace("\x7f", @"\u916?");		// delta 
+			s2 = s2.Replace("\x2265",@"\u8805?");	// greater than or equal
+			s2 = s2.Replace("\x2264",@"\u8804?");	// less than or equal
+			s2 = s2.Replace("\xB1",@"\'b1");		// plus minus
+			s2 = s2.Replace("\x3A3",@"\u931?");		// sigma
+			s2 = s2.Replace("\x3C4",@"\u947?");		// gamma
+			s2 = s2.Replace("\xBD",@"\'bd");		// half
+			s2 = s2.Replace("\x25A0",@"\u9604?");	// accum 2584
+			s2 = s2.Replace("\x7",@"\u9679?");		// bullet 25CF
+			s2 = s2.Replace("\x2248",@"\u8776?");	// approx eq
+			s2 = s2.Replace("\x2261",@"\u8773?");	// similar eq 2245
+			s2 = s2.Replace("\xF7",@"\'f7");		// division
+			s2 = s2.Replace("\x221A",@"\u8730?");	// square root
+			s2 = s2.Replace("\x393",@"\u961?");		// rho 3C1
+			s2 = s2.Replace("\x3C0",@"\u960?");		// pi
+			s2 = s2.Replace("\xb5", @"\u956?");		// micro 3BC  (try e6, if not work try 109)
+			s2 = s2.Replace("\x3B4", @"\u948?");	// lower case delta
+			s2 = s2.Replace("\x3C3", @"\u963?");	// lower case sigma
+			s2 = s2.Replace("\xBC", @"\'bc");		// quarter
+			s2 = s2.Replace("\x3C6", @"\'d8");		// dist zero, D8
+			s2 = s2.Replace("\xC9", @"\u274?");		// energy, 112
+			s2 = s2.Replace("\xEC", @"\'ec");		// grave
+			s2 = s2.Replace("\x2502", @"\u9474?");	// bar 
+			s2 = s2.Replace("\x3B5", @"\u949?");	// epsilon	
+			s2 = s2.Replace("\x398", @"\u952?");	// theta, 3B8
+			s2 = s2.Replace("\x221E", @"\u8857?");	// dot in oval, 2299
+			s2 = s2.Replace("\xBF", @"\u964?");		// tau, 3C4
+			s2 = s2.Replace("\x2310", @"\u9830?");	// diamond, 2666
+			s2 = s2.Replace("\x2192", @"\u8594?");
+			s2 = s2.Replace("\x2190", @"\u8592?");
+			s2 = s2.Replace("\x2191", @"\u8593?");
+			s2 = s2.Replace("\x2193", @"\u8595?");
+			s2 = s2.Replace("\x2207", @"\u8711?");
+			
+			return s2;
+		}
 		public static string ConvertText(string s1)
 		{
-			string s2 = Reg2.Replace(s1, new MatchEvaluator(ReplaceChars));
+            string s2 = s1;
+			
+			s2 = ReplaceUnicode(s2);
+            // now replace underline on/off (AE,AF), super on/off (C6,C7)
+            // bold on/off (D5, D6), subscript on/off (D1 A6), and
+            // italics on/off (B2, DD)
+            s2 = s2.Replace("\xAB", "\\ul ");
+            s2 = s2.Replace("\xBB", "\\ul0 ");
+            s2 = s2.Replace("\x255E", "\\super ");
+            s2 = s2.Replace("\x255F", "\\nosupersub ");
+            s2 = s2.Replace("\x2552", "\\b ");
+            s2 = s2.Replace("\x2553", "\\b0 ");
+            s2 = s2.Replace("\x2564", "\\sub ");
+            s2 = s2.Replace("\xAA", "\\nosupersub ");
+            s2 = s2.Replace("\x2593", "\\i ");
+            s2 = s2.Replace("\x258C", "\\i0 ");
+            s2 = s2.Replace("\x2559", "\\ul\\b ");
+            s2 = s2.Replace("\x2558", "\\b0\\ul0 ");
+
+            // underline next word is 0x17
+            // superscript next is 0x18
+            // subscript next is 0x19
+            // bold next is 0x13
+            s2 = Regex.Replace(s2, @"\x17([^\x17 ]*?)(?:[\x17]|(?= )|\Z)(.*?)", @"\ul $1\ul0 $2");
+            s2 = Regex.Replace(s2, @"\x18([^\x18 ]*?)(?:[\x18]|(?= )|\Z)(.*?)", @"\super $1\nosupersub $2");
+            s2 = Regex.Replace(s2, @"\x19([^\x19 ]*?)(?:[\x19]|(?= )|\Z)(.*?)", @"\sub $1\nosupersub $2");
+            s2 = Regex.Replace(s2, @"\x13([^\x13 ]*?)(?:[\x13]|(?= )|\Z)(.*?)", @"\b $1\b0 $2");
+
+            s2 = Reg2.Replace(s2, new MatchEvaluator(ReplaceChars));
 			return s2;
 		}
 	}