WebSVN – DevTools – Diff – /EA_DocGen/trunk/EA_DocGen/TextParser.cs

 using System;
 using System.Collections;
 using System.Text;
+using System.Text.RegularExpressions;
-using Word;
+using Microsoft.Office.Interop.Word;
 using Microsoft.Office.Core;
 // TODO
 //
          STYLE_ID_LIST_NUMBER_4,
          STYLE_ID_LIST_ALPHA_0,
          STYLE_ID_LIST_ALPHA_1,
          STYLE_ID_LIST_ALPHA_2,
          STYLE_ID_LIST_ALPHA_3,
-         STYLE_ID_LIST_ALPHA_4
+         STYLE_ID_LIST_ALPHA_4,
          // do not put anything below here
+         STYLE_ID_FONT_COLOR
       };
       // A token is a block of text associated with a tag (style) name and type
       private struct token_type
+      {
       // appended to the word document. For this formatting, we need to remember the word
       // ranges that denote the text and the style name to apply. So we do it in a list of
       // the following item.
       private struct postFormat_type
+      {
-         public postFormat_type(Word.Range wr, token_type tk)
+         public postFormat_type(Range wr, token_type tk)
+         {
             m_wr = wr;
             m_tk = tk;
+         }
-         public Word.Range m_wr;
+         public Range m_wr;
          public token_type m_tk;
       };
       // Use a hash table for recording the allowed tags and their attributes, facilitating rapid
       // lookup during parsing.
          // MS-Word formatting tags
          styleDefs.Add( "b", formStyleDef( style_type_e.STYLE_MS_WORD, style_id_e.STYLE_ID_BOLD, EA_Constants.styleName_Bold      , 0, style_handling_e.STYLE_POST_FORMAT) );
          styleDefs.Add( "i", formStyleDef( style_type_e.STYLE_MS_WORD, style_id_e.STYLE_ID_ITALIC, EA_Constants.styleName_Italic    , 0, style_handling_e.STYLE_POST_FORMAT) );
          styleDefs.Add( "u", formStyleDef( style_type_e.STYLE_MS_WORD, style_id_e.STYLE_ID_UNDERLINE, EA_Constants.styleName_Underline , 0, style_handling_e.STYLE_POST_FORMAT) );
+         styleDefs.Add("font", formStyleDef(style_type_e.STYLE_MS_WORD, style_id_e.STYLE_ID_FONT_COLOR, EA_Constants.styleName_Font_Color, 0, style_handling_e.STYLE_POST_FORMAT));
          styleDefs.Add( "lb0", formStyleDef( style_type_e.STYLE_MS_WORD, style_id_e.STYLE_ID_LIST_BULLET_0, EA_Constants.styleName_ListBullet0,1, style_handling_e.STYLE_POST_FORMAT ) );
          styleDefs.Add( "lb1", formStyleDef( style_type_e.STYLE_MS_WORD, style_id_e.STYLE_ID_LIST_BULLET_1, EA_Constants.styleName_ListBullet1,2, style_handling_e.STYLE_POST_FORMAT ) );
          styleDefs.Add( "lb2", formStyleDef( style_type_e.STYLE_MS_WORD, style_id_e.STYLE_ID_LIST_BULLET_2, EA_Constants.styleName_ListBullet2,3, style_handling_e.STYLE_POST_FORMAT ) );
          styleDefs.Add( "lb3", formStyleDef( style_type_e.STYLE_MS_WORD, style_id_e.STYLE_ID_LIST_BULLET_3, EA_Constants.styleName_ListBullet3,4, style_handling_e.STYLE_POST_FORMAT ) );
          int pos_tagName;
          int pos_ContentStart;
          int pos_ContentEnd;
          bool lookingForTagEnd;
          bool foundError = false;
-         Word.Range wr_body;
+         Range wr_body;
          // Convert EA7.1 embedded RTF controls into EA_DocGen controls where possible
          s = convert_EA7_1_RTF_ListTag(s, "<ol>\r\n", "</ol>", "<ln0>", "</ln0>");
          s = convert_EA7_1_RTF_ListTag(s, "<ul>\r\n", "</ul>", "<lb0>", "</lb0>");
          // human users and so EAv7.1 converts < and > chars into a form such that they do not look like RTF tags.
          // We have to convert EA's escaping mechanism back into ascii text chars otherwise the parser wont work.
          // There may be issues to resolve with this parser if EAv7.1 users have used any of EA's RTF editing features
          // because if they do, the < and > chars there will not be escaped and the parser will find them and try to
          // decode them. Most will probably not be decodable.
-         s = s.Replace("&lt;","<");
-         s = s.Replace("&gt;",">");
+         s = TextualContent.HtmlDecode(s);
          // Begin to construct a range that will eventually encompass ALL of the text we will serialize during the execution
          // of this function. This is needed later only if the caller has specified a requirement style that needs to have
          // global strikethrough or italicising applied to (all of) the text
          object startLocation;
             // look for the closing bracket of the tag
             pos = s.IndexOf(">", pos+1);
             if (pos >= 0)
+            {
+               string sTag = s.Substring(pos_tagName, pos - pos_tagName);
+               if (!isEnding)
+               {
+                   // check for font start tag with color attribute
+                   Regex re = new Regex("font.*?color=\"#(?<color>[0-9A-F]*)\"",RegexOptions.IgnoreCase);
+                   if (re.IsMatch(sTag))
+                     sTag = "font";
+               }
                // use hash table to identify the tag
-               found = styleDefs.Contains(s.Substring(pos_tagName, pos - pos_tagName));
+               found = styleDefs.Contains(sTag);
+            }
             else
+            {
                // Cannot find any '>' so we should just exit the loop
                break;
+               }
                else
+               {
                   // If there is content prior to now that has not been consumed, tokenise it now
                   if ((pos_LeftBracket - pos_ContentStart) > 0)
+                  {
                      token.txt = s.Substring(pos_ContentStart, pos_LeftBracket - pos_ContentStart);
                      tokens.Add(token);
+                  }
                   else
+                  {
                      // update the token variable with this tags atributes from the hash table lookup. This overwrites
                      // the default values assigned when parsing began, or when we resumed parsing after dealing with
                      // the previous token found.
-                     token_type lookupToken = ((token_type)styleDefs[s.Substring(pos_tagName, pos_RightBracket-pos_tagName)]);
+                     string df = s.Substring(pos_tagName, pos_RightBracket-pos_tagName);
+                     // check for font tag with color attribute
+                     int iFontColor = -1;
+                     Regex re = new Regex("font.*?color=\"#(?<color>[0-9A-F]*)\"", RegexOptions.IgnoreCase);
+                     if (re.IsMatch(df))
+                     {
+                         // get color attribute value
+                         string sColor = re.Matches(df)[0].Groups["color"].Value;
+                         iFontColor = Convert.ToInt32(sColor,16);
+                         df = "font";
+                     }
+                     token_type lookupToken = ((token_type)styleDefs[df]);
                      token.styleId   = lookupToken.styleId;
                      token.styleType = lookupToken.styleType;
                      token.styleName = lookupToken.styleName;
+                     if (token.styleId == style_id_e.STYLE_ID_FONT_COLOR)
+                         token.level = iFontColor;
+                     else
-                     token.level     = lookupToken.level;
+                         token.level = lookupToken.level;
                      token.styleHandling = lookupToken.styleHandling;
                      token.txt = null; // we dont know what the text content will be yet. This is obtained when we encounter the end tag
                      pos_ContentStart = pos_RightBracket + 1;
                      last_list_level = 0;
                      break;
                   case style_id_e.STYLE_ID_UNDERLINE:
                      pf.m_wr.Select();
-                     createWordDoc.WordApp.Selection.Range.Underline = Word.WdUnderline.wdUnderlineSingle;
+                     createWordDoc.WordApp.Selection.Range.Underline = WdUnderline.wdUnderlineSingle;
                      last_list_level = 0;
                      break;
+                   case style_id_e.STYLE_ID_FONT_COLOR:
+                     pf.m_wr.Select();
+                     createWordDoc.WordApp.Selection.Range.Font.Color = (WdColor) ReverseRGB((UInt32)pf.m_tk.level);
+                     last_list_level = 0;
+                    break;
                   default:
                      // Handle bullets/lists
                      if (pf.m_tk.styleId >= style_id_e.STYLE_ID_LIST_BULLET_0)
+                     {
                         style = pf.m_tk.styleName;
+                           {
                               // To restart numbering, the only way to do it is to (re-)apply the list template to the
                               // selection with a "continue previous list" setting of false, and an "apply to"
                               // setting of "whole list".
                               object continuePreviousList = false;
-                              object applyTo = Word.WdListApplyTo.wdListApplyToWholeList;
+                              object applyTo = WdListApplyTo.wdListApplyToWholeList;
-                              object defListBehavour = Word.WdDefaultListBehavior.wdWord10ListBehavior;
+                              object defListBehavour = WdDefaultListBehavior.wdWord10ListBehavior;
-                              Word.ListTemplate lt = createWordDoc.WordApp.Selection.Range.ListFormat.ListTemplate;
+                              ListTemplate lt = createWordDoc.WordApp.Selection.Range.ListFormat.ListTemplate;
                               createWordDoc.WordApp.Selection.Range.ListFormat.ApplyListTemplate(
                                  lt, ref continuePreviousList, ref applyTo, ref defListBehavour);
+                           }
+                        }
          if (callerStyle.Equals(EA_Constants.styleName_ReqPropBody))
+         {
             // Complete construction of a range that will encompass ALL of the text we will serialize during the execution
             // of this function
             endLocation = createWordDoc.WordDocument.Content.End;
-            Word.Range wr_total = createWordDoc.WordDocument.Range(ref startLocation, ref endLocation);
+            Range wr_total = createWordDoc.WordDocument.Range(ref startLocation, ref endLocation);
             while (wr_total.Characters.Last.Text.Equals("\r") || wr_total.Characters.Last.Text.Equals("\n"))
                wr_total.End = wr_total.End - 1;  // don't format the \r\n char at the end - doing so causes wierd ms-word exceptions later on
             // italicise
             wr_total.Font.Italic = (int)MsoTriState.msoTrue;
          else if (callerStyle.Equals(EA_Constants.styleName_ReqRejBody))
+         {
             // Complete construction of a range that will encompass ALL of the text we will serialize during the execution
             // of this function
             endLocation = createWordDoc.WordDocument.Content.End;
-            Word.Range wr_total = createWordDoc.WordDocument.Range(ref startLocation, ref endLocation);
+            Range wr_total = createWordDoc.WordDocument.Range(ref startLocation, ref endLocation);
             while (wr_total.Characters.Last.Text.Equals("\r") || wr_total.Characters.Last.Text.Equals("\n"))
                wr_total.End = wr_total.End - 1;  // don't format the \r\n char at the end - doing so causes wierd ms-word exceptions later on
             // strikethrough - use msoCTrue since msoTrue simply toggles strikethough attribute, or so it seems
             wr_total.Font.StrikeThrough = (int)MsoTriState.msoCTrue;
+         }
          return true;
+      }
+      public static UInt32 ReverseRGB(UInt32 value)
+      {
+          UInt32 i = (value & 0x000000FFU) << 16 | (value & 0x0000FF00U) |
+                 (value & 0x00FF0000U) >> 16;
+          return i;
+      }
+   }
+}

Subversion Repositories DevTools

(root)/EA_DocGen/trunk/EA_DocGen/TextParser.cs – Rev 2134 → 2136