Convert symbols in supported fonts to corresponding Unicode characters

1 files changed, 20 insertions, 0 deletions

diff --git a/lib/docx/body-reader.js b/lib/docx/body-reader.js
index 2616dde..90c2b39 100644
--- a/lib/docx/body-reader.js
+++ b/lib/docx/body-reader.js
@@ -1,6 +1,7 @@
 exports.createBodyReader = createBodyReader;
 exports._readNumberingProperties = readNumberingProperties;
 
+var dingbatToUnicode = require("dingbat-to-unicode");
 var _ = require("underscore");
 
 var documents = require("../documents");
@@ -167,6 +168,24 @@ function BodyReader(options) {
         return emptyResult();
     }
 
+    function readSymbol(element) {
+        // See 17.3.3.30 sym (Symbol Character) of ECMA-376 4th edition Part 1
+        var font = element.attributes["w:font"];
+        var char = element.attributes["w:char"];
+        var unicodeCharacter = dingbatToUnicode.hex(font, char);
+        if (unicodeCharacter == null && /^F0..$/.test(char)) {
+            unicodeCharacter = dingbatToUnicode.hex(font, char.substring(2));
+        }
+
+        if (unicodeCharacter == null) {
+            return emptyResultWithMessages([warning(
+                "A w:sym element with an unsupported character was ignored: char " +  char + " in font " + font
+            )]);
+        } else {
+            return elementResult(new documents.Text(unicodeCharacter.string));
+        }
+    }
+
     function noteReferenceReader(noteType) {
         return function(element) {
             var noteId = element.attributes["w:id"];
@@ -240,6 +259,7 @@ function BodyReader(options) {
         "w:softHyphen": function(element) {
             return elementResult(new documents.Text("\u00AD"));
         },
+        "w:sym": readSymbol,
         "w:hyperlink": function(element) {
             var relationshipId = element.attributes["r:id"];
             var anchor = element.attributes["w:anchor"];