Added markdown-writer which will write directly from docx to Markdown.

3 files changed, 142 insertions, 6 deletions

diff --git a/lib/document-to-html.js b/lib/document-to-html.js
index b012ea9..695b34c 100644
--- a/lib/document-to-html.js
+++ b/lib/document-to-html.js
@@ -43,7 +43,7 @@ function DocumentConversion(options) {
     
     function convertToHtml(document) {
         var messages = [];
-        var html = new HtmlGenerator({prettyPrint: options.prettyPrint});
+        var html = new HtmlGenerator({prettyPrint: options.prettyPrint, markdown: options.markdown});
         
         return promises.nfcall(elementToHtml, document, html, messages).then(function() {
             html.closeAll();
@@ -88,7 +88,7 @@ function DocumentConversion(options) {
     }
     
     function convertRun(run, html, messages, callback) {
-        var runHtml = new HtmlGenerator();
+        var runHtml = new HtmlGenerator({ markdown: options.markdown });
         if (run.styleId) {
             var style = findStyle(run);
             if (style) {
@@ -160,7 +160,7 @@ function DocumentConversion(options) {
         html.open(htmlPaths.element("td"), {alwaysWrite: true});
         
         async.map(element.children, function(child, callback) {
-            var cellHtml = new HtmlGenerator();
+            var cellHtml = new HtmlGenerator({ markdown: options.markdown });
             elementToHtml(child, cellHtml, messages, function(error) {
                 if (error) {
                     callback(error);
@@ -251,7 +251,7 @@ function DocumentConversion(options) {
         },
         "note": function(element, html, messages, callback) {
             html.open(htmlPaths.element("li", {id: noteHtmlId(element)}));
-            var footnoteHtml = new HtmlGenerator();
+            var footnoteHtml = new HtmlGenerator({ markdown: options.markdown });
             convertElements(element.body, footnoteHtml, messages, function(err) {
                 if (err) {
                     callback(err);
diff --git a/lib/html-writer.js b/lib/html-writer.js
index f1761c9..33441cb 100644
--- a/lib/html-writer.js
+++ b/lib/html-writer.js
@@ -5,9 +5,15 @@ exports.writer = writer;
 
 function writer(options) {
     options = options || {};
-    if (options.prettyPrint) {
+    if(options.prettyPrint) {
         return prettyWriter();
-    } else {
+    }
+    else if(options.markdown) {
+        var markdown = require("./markdown-writer");
+
+        return markdown.writer();
+    }
+    else {
         return simpleWriter();
     }
 }
diff --git a/lib/markdown-writer.js b/lib/markdown-writer.js
new file mode 100644
index 0000000..1bfa942
--- /dev/null
+++ b/lib/markdown-writer.js
@@ -0,0 +1,130 @@
+
+function mdElement(start, end) {
+    if (end === undefined)
+        end = start;
+
+    this.open = function(attributes) {
+        return start;
+    };
+
+    this.close = function() {
+        return end;
+    };
+}
+function mdLink(link, text) {
+    var attrs = {};
+
+    this.open = function(attributes) {
+        attrs = attributes;
+        return "[";
+    };
+
+    this.close = function() {
+        return (attrs[text] || "") + "](" + (attrs[link] || "") + ")";
+    };
+}
+function mdListItem() {
+    this.open = function(attributes) {
+        var output = "";
+
+        for(var i = 0; i < attributes["indent"]; ++i)
+            output += "\t";
+
+        return output + (attributes["type"] === "ul" ? "-" : attributes["count"] + ".") + " ";
+    };
+
+    this.close = function() {
+        return "\n";
+    };
+}
+
+// map HTML tags to Markdown syntax
+var htmlmd = {
+    "h1": new mdElement("# ", "\n\n"),
+    "h2": new mdElement("## ", "\n\n"),
+    "h3": new mdElement("### ", "\n\n"),
+    "h4": new mdElement("#### ", "\n\n"),
+    "h5": new mdElement("##### ", "\n\n"),
+    "h6": new mdElement("###### ", "\n\n"),
+    "p": new mdElement("", "\n\n"),
+    "br": new mdElement("", "  \n"),
+    "ul": new mdElement("\n", ""),
+    "ol": new mdElement("\n", ""),
+    "li": new mdListItem(),
+    "strong": new mdElement("**"),
+    "em": new mdElement("*"),
+    "a": new mdLink("href"),
+    "img": new mdLink("src", "alt"),
+};
+
+function mdWriter() {
+    var fragments = [];
+    var list = {
+        indent: -1,
+        type: "ul",
+        count: 0
+    };
+    
+    function isList(tagName) {
+        return tagName == "ul" || tagName == "ol";
+    }
+
+    function open(tagName, attributes) {
+        if(htmlmd[tagName]) {
+            if(tagName == "li") {
+                attributes = attributes || {};
+                attributes["indent"] = list.indent;
+                attributes["type"] = list.type;
+                attributes["count"] = ++list.count;
+            }
+
+            fragments.push(htmlmd[tagName].open(attributes));
+
+            if(isList(tagName)) {
+                list.type = tagName;
+                list.indent++;
+                list.count = 0;
+            }
+        }
+    }
+    
+    function close(tagName) {
+        if(htmlmd[tagName]) {
+            fragments.push(htmlmd[tagName].close());
+
+            if(isList(tagName)) {
+                list.indent--;
+            }
+        }
+    }
+    
+    function selfClosing(tagName, attributes) {
+        if(htmlmd[tagName]) {
+            fragments.push(htmlmd[tagName].open(attributes));
+            fragments.push(htmlmd[tagName].close());
+        }
+    }
+    
+    function text(value) {
+        fragments.push(value);
+    }
+    
+    function append(html) {
+        fragments.push(html);
+    }
+    
+    function asString() {
+        return fragments.join("").replace(/\n{2,}/, "\n\n");
+    }
+
+    return {
+        asString: asString,
+        open: open,
+        close: close,
+        text: text,
+        selfClosing: selfClosing,
+        append: append
+    };
+}
+
+exports.writer = mdWriter;
\ No newline at end of file