Add extractRawText function

1 files changed, 25 insertions, 0 deletions

diff --git a/lib/index.js b/lib/index.js
index 35bcbaf..879327f 100644
--- a/lib/index.js
+++ b/lib/index.js
@@ -4,6 +4,7 @@ var readStyle = require("./style-reader").readStyle;
 var readOptions = require("./options-reader").readOptions;
 
 exports.convertToHtml = convertToHtml;
+exports.extractRawText = extractRawText;
 
 function convertToHtml(input, options) {
     var fullOptions = readOptions(options);
@@ -26,6 +27,30 @@ function convertDocumentToHtml(documentResult, options) {
     });
 }
 
+
+function extractRawText(input) {
+    return docxReader.read(input)
+        .then(function(documentResult) {
+            return documentResult.map(convertElementToRawText);
+        });
+}
+
+function convertElementToRawText(element) {
+    if (element.type === "text") {
+        return element.value;
+    } else {
+        var tail = element.type === "paragraph" ? "\n\n" : "";
+        return (element.children || []).map(convertElementToRawText).join("") + tail;
+    }
+}
+
+function convertElements(elements, html, messages, callback) {
+    async.eachSeries(elements, function(element, callback) {
+        return elementToHtml(element, html, messages, callback);
+    }, callback);
+}
+
+
 exports.styleMapping = function() {
     throw new Error('Use a raw string instead of mammoth.styleMapping e.g. "p[style-naming=\'Title\'] => h1" instead of mammoth.styleMapping("p[style-naming=\'Title\'] => h1")');
 };