Split out docx reading into separate module

2 files changed, 79 insertions, 65 deletions

diff --git a/lib/docx-reader.js b/lib/docx-reader.js
new file mode 100644
index 0000000..3078d2f
--- /dev/null
+++ b/lib/docx-reader.js
@@ -0,0 +1,74 @@
+exports.read = read;
+
+var q = require("q");
+var _ = require("underscore");
+var ZipFile = require("zipfile").ZipFile;
+
+var xmlreader = require("./xmlreader");
+var documents = require("./documents");
+
+
+function read(docxFile) {
+    if (_.isString(docxFile)) {
+        var zipFile = new ZipFile(docxFile);
+        docxFile = {
+            read: function(path) {
+                return q.ninvoke(zipFile, "readFile", path, "utf-8")
+                    .then(function(value) {
+                        return value.toString();
+                    });
+            }
+        };
+    }
+    
+    return docxFile.read("word/document.xml")
+        .then(xmlreader.read)
+        .then(convertXmlToDocument);
+}
+
+function convertXmlToDocument(documentXml) {
+    var body = documentXml.root.first("w:body");
+    
+    var children = readXmlElements(body.children);
+    return new documents.Document(children);
+}
+
+function readXmlElements(elements) {
+    return elements.map(readXmlElement).filter(notNull);
+}
+
+function readXmlElement(element) {
+    if (element.type === "element") {
+        var handler = xmlElementReaders[element.name];
+        if (handler) {
+            return handler(element);
+        } else {
+            return null;
+        }
+    } else {
+        return null;
+    }
+}
+
+var xmlElementReaders = {
+    "w:p": function(element) {
+        var children = readXmlElements(element.children);
+        return new documents.Paragraph(children);
+    },
+    "w:r": function(element) {
+        var children = readXmlElements(element.children);
+        return new documents.Run(children);
+    },
+    "w:t": function(element) {
+        return new documents.Text(element.children[0].value);
+    }
+};
+
+function parseXml(string) {
+    return xmlreader.read(string);
+}
+
+function notNull(value) {
+    return value !== null;
+}
+
diff --git a/lib/index.js b/lib/index.js
index 5db51a8..b124b81 100644
--- a/lib/index.js
+++ b/lib/index.js
@@ -2,7 +2,7 @@ var q = require("q");
 var _ = require("underscore");
 var ZipFile = require("zipfile").ZipFile;
 
-var xmlreader = require("./xmlreader");
+var docxReader = require("./docx-reader");
 var documents = require("./documents");
 var styles = require("./styles");
 
@@ -21,69 +21,9 @@ function Converter(options) {
 }
 
 Converter.prototype.convertToHtml = function(docxFile) {
-    if (_.isString(docxFile)) {
-        var zipFile = new ZipFile(docxFile);
-        docxFile = {
-            read: function(path) {
-                return q.ninvoke(zipFile, "readFile", path, "utf-8")
-                    .then(function(value) {
-                        return value.toString();
-                    });
-            }
-        };
-    }
-    
     var documentConverter = this._documentConverter;
-    return docxFile.read("word/document.xml").then(function(documentXmlString) {
-        return parseXml(documentXmlString);
-    }).then(function(documentXml) {
-        var document = convertXmlToDocument(documentXml);
-        return documentConverter.convertToHtml(document);
-    });
-}
-
-function convertXmlToDocument(documentXml) {
-    var body = documentXml.root.first("w:body");
-    
-    var children = readXmlElements(body.children);
-    return new documents.Document(children);
-}
-
-function readXmlElements(elements) {
-    return elements.map(readXmlElement).filter(notNull);
-}
-
-function readXmlElement(element) {
-    if (element.type === "element") {
-        var handler = xmlElementReaders[element.name];
-        if (handler) {
-            return handler(element);
-        } else {
-            return null;
-        }
-    } else {
-        return null;
-    }
-}
-
-var xmlElementReaders = {
-    "w:p": function(element) {
-        var children = readXmlElements(element.children);
-        return new documents.Paragraph(children);
-    },
-    "w:r": function(element) {
-        var children = readXmlElements(element.children);
-        return new documents.Run(children);
-    },
-    "w:t": function(element) {
-        return new documents.Text(element.children[0].value);
-    }
-};
-
-function parseXml(string) {
-    return xmlreader.read(string);
-}
-
-function notNull(value) {
-    return value !== null;
+    return docxReader.read(docxFile)
+        .then(function(document) {
+            return documentConverter.convertToHtml(document);
+        });
 }