Avoid repeated iterations through the zip file

2 files changed, 28 insertions, 31 deletions

diff --git a/lib/docx-reader.js b/lib/docx-reader.js
index 0027a0a..35f1627 100644
--- a/lib/docx-reader.js
+++ b/lib/docx-reader.js
@@ -5,6 +5,7 @@ var _ = require("underscore");
 var unzip = require("unzip");
 var path = require("path");
 var fs = require("fs");
+var Buffers = require("buffers");
 
 var xmlreader = require("./xmlreader");
 var documents = require("./documents");
@@ -33,12 +34,13 @@ function read(docxFile) {
 function openZip(zipPath) {
     var deferred = q.defer();
     var names = [];
+    var files = {};
     
     fs.createReadStream(zipPath)
         .pipe(unzip.Parse())
         .on("entry", function(entry) {
             names.push(entry.path);
-            entry.autodrain();
+            files[entry.path] = readEntry(entry);
         })
         .on("close", function() {
             deferred.resolve(zipFile);
@@ -50,40 +52,34 @@ function openZip(zipPath) {
     var exists = function(path) {
         return names.indexOf(path) !== -1;
     }
+            
+    function readEntry(entry) {
+        var readDeferred = q.defer();
+        var buffers = new Buffers();
+        
+        entry
+            .on("data", function(data) {
+                buffers.push(data);
+            })
+            .on("end", function() {
+                readDeferred.resolve(buffers);
+            })
+            .on("error", readDeferred.reject.bind(readDeferred))
+            
+        return readDeferred.promise;
+    }
     var zipFile = {
         read: function(path, encoding) {
             if (!exists(path)) {
                 return q.reject(new Error("No file in zip: " + path));
             }
-            var readDeferred = q.defer();
-            
-            var readStream = fs.createReadStream(zipPath)
-                .pipe(unzip.Parse())
-                .on("entry", function(entry) {
-                    if (entry.path === path) {
-                        readEntry(entry);
-                    } else {
-                        entry.autodrain();
-                    }
-                })
-                .on("error", readDeferred.reject.bind(readDeferred));
-            
-            function readEntry(entry) {
-                entry.setEncoding(encoding);
-                
-                var fragments = [];
-                
-                entry
-                    .on("data", function(data) {
-                        fragments.push(data);
-                    })
-                    .on("end", function() {
-                        readDeferred.resolve(fragments.join(""));
-                    })
-                    .on("error", readDeferred.reject.bind(readDeferred))
-            }
-            
-            return readDeferred.promise;
+            return files[path].then(function(buffer) {
+                if (encoding) {
+                    return buffer.toString(encoding);
+                } else {
+                    return buffer;
+                }
+            });
         },
         exists: exists
     };
diff --git a/package.json b/package.json
index cb9a44b..e67bbf8 100644
--- a/package.json
+++ b/package.json
@@ -17,7 +17,8 @@
         "optimist": "~0.3.5",
         "connect": "~2.7.3",
         "dustjs-linkedin": "~1.2.1",
-        "async": "0.2.6"
+        "async": "~0.2.6",
+        "buffers": "~0.1.1"
     },
     "devDependencies": {
         "mocha": "~1.8"